Support: introduce public API annotation support

[tstellar]

Summary

This provides a model for annotating the "public" interfaces of a module
in LLVM. This is motivated by the desire to have the ability to build
LLVM libraries in the shared mode to enable extraction of the shared
content between the tools.

Unlike Unix platforms, Windows has a more stringent requirement of
annotating the "dll-interface" when building a shared library. All of
the public APIs should be marked by the appropriate *_API macro. The
annotation indicates what APIs are public but are expected to live in
the dynamic library (DLL/dylib/so) that the macro is associated with.

The longer term desire here is to evolve this to support creating a
small set of libraries. Because there is a cost associated with the
call across the image boundary (GOT+PLT/IAT), the intent is to minimise
the number of libraries. The separation of the libraries should prefer
to split along the following disjoint sets:

• minimal set of interfaces required for llvm-ar (archiver/librarian)
• minimal set of interfaces required for lld (linker)
• minimal set of interfaces required for clang (compiler)

The reasoning behind this split set is that these tools are executed the
most, and minimising the penalty for these three sets is crucial for
keeping the penalty to dynamic linking (and LTO) low. Keeping the
number of libraries low reduces the complexity in distribution and
should also help reduce the overall size of the toolchain distribution.

[compnerd]

This is a preliminary patch which is incomplete. I wanted to put this up earlier rather than later because I suspect that this is going to require a few iterations while we figure out the exact details of the style that we want to use (e.g. *_API vs *_ABI vs *_EXPORT), location of the annotation, and the style of the annotation (e.g. LLVMSupport_API inline vs prefix vs LLVMSupport_API(return type)). Additionally, the API surface to audit is fairly large, so snapshotting is helpful.

[tschuett]

It is completely up to you, but for me Support is kind of redundant or the wrong kind of information. I would expect something like LLVM_PUBLIC_API.

[compnerd]

In D109192#2980913, @tschuett wrote:

It is completely up to you, but for me Support is kind of redundant or the wrong kind of information. I would expect something like LLVM_PUBLIC_API.

LLVM_PUBLIC_API is not feasible IMO as there will be a macro per dynamic library that we end up with and each must have a separate spelling [1]. I suppose I can do a bit of macro expansion magic to do something like:

LLVM_PUBLIC_API(LLVMSupport)
LLVM_PUBLIC_API(LLVM)
LLVM_PUBLIC_API(LLVMObject)

where the optional parameter is the name of the dynamic object that serves as the home (assuming that we end up with LLVMSupport.dll, libLLVM.dylib and libLLVMObject.so in the fictional world).

[1] The separate spelling is required for cases where multiple libraries are in an address space. Consider the following:

Library A: A_API
Library B: B_API
The executable see:

int A_API f(void);
int B_API g(void);
int main() {
  return f() + g();
}

What happens if B also depends on A and we use the same macro? It will see int LLVM_PUBLIC_API f(void); but it is defining the "public API" so it will incorrectly mark f as being local.

[fodinabor]

Thanks for working on this!
In general looks good to me, I'd opt for all CAPS macros, though, so e.g. LLVM_SUPPORT_API. The return variant LLVM_SUPPORT_API(retTy) is only beneficial when needing something like extern "C" retTy __stdcall iirc and is less ideal for marking classes as exported.
A parameterized version LLVM_API(LLVMSupport) would be fine for me as well, although I don't see any benefit.

[compnerd]

In D109192#2981682, @fodinabor wrote:

Thanks for working on this!
In general looks good to me, I'd opt for all CAPS macros, though, so e.g. LLVM_SUPPORT_API. The return variant LLVM_SUPPORT_API(retTy) is only beneficial when needing something like extern "C" retTy __stdcall iirc and is less ideal for marking classes as exported.

Sure, I can switch it to LLVM_SUPPORT_API just as well. That is the easy part of the change. The difficult part is the auditing of the interfaces.

Correct, the return variant is useful for the extra decoration. While not particularly useful for the classes, it does allow writing the decorations more naturally when expanded:

class __declspec(dllexport) exported { }; on Windows
__attribute__((__visibility__("default"))) class exported { }; on other platforms

The places where it is however useful is is on variable and function declarations. Exported functions really should be decorated with extern "C", but that is something that I don't think we should conflate with the work here. The goal here is to minimally attribute enough of LLVM with dll interface/visibility and switch to hidden visibility on non-Windows platforms while enabling a build where the core of LLVM can be a shared library.

A parameterized version LLVM_API(LLVMSupport) would be fine for me as well, although I don't see any benefit.

I agree, it isn't particularly useful (and it is more complicated). That was a suggestion to @tschuett's comment that we aim for a single macro rather than a library specific macro.

[compnerd]

One question that remains is the spelling of the macro. Do we want to go with API or ABI? I think that there is an argument to be made that ABI is more precise - this marks how you call into the library not just how you use it. When a function is moved beyond the module boundary, while the API will not change, the ABI of the module would. This of course pedantic, but these items are important to nail down correctly early as these are fairly large scale invasive changes.

[hans]

In D109192#2982429, @compnerd wrote:

One question that remains is the spelling of the macro. Do we want to go with API or ABI? I think that there is an argument to be made that ABI is more precise - this marks how you call into the library not just how you use it. When a function is moved beyond the module boundary, while the API will not change, the ABI of the module would. This of course pedantic, but these items are important to nail down correctly early as these are fairly large scale invasive changes.

I've seen *API macros used for this in other libraries, but I don't think I've ever seen the *ABI variant.

[tschuett]

Again up to you, but my understanding of LLVM is to get something working and then iterate on it. Most of your work went into SupportMacros.h. The decision for API vs. ABI has less entropy.

[compnerd]

• use LLVM_SUPPORT_ABI as per feedback
• remove changes to the implementation to make it easier to iterate
• change tactics to target llvm-tblgen first

This gets close to being able to build llvm-tblgen (8 unresolved externals).

[compnerd]

Some additional questions:

• should we be encouraging of exporting entire classes or should we be more judicious in the visibility (e.g. attribute members)? I'm not 100% certain that this works for non-dll-interface targets.

I'm more convinced now that we should introduce an option to introduce this behaviour. The new option can opt into building the shared library components. Perhaps something like LLVM_EXPERIMENTAL_SHARED_LIBRARY_COMPONENTS is a good option name?

[rnk]

In D109192#2982723, @compnerd wrote:

Some additional questions:

• should we be encouraging of exporting entire classes or should we be more judicious in the visibility (e.g. attribute members)? I'm not 100% certain that this works for non-dll-interface targets.

I think you can give entire classes default visibility, and that will apply to all members. I think we should annotate classes because it will reduce the total number of annotations, and that will reduce the churn. For large, key classes that have too many private methods (Sema, ASTContext?) we can go back later and export only the public APIs, or something like that.

I'm more convinced now that we should introduce an option to introduce this behaviour. The new option can opt into building the shared library components. Perhaps something like LLVM_EXPERIMENTAL_SHARED_LIBRARY_COMPONENTS is a good option name?

I agree this new thing shouldn't become the default build config. How close is this to BUILD_SHARED_LIBS vs LLVM_ENABLE_DYLIB? Can we repurpose one of those?

---

Going in a completely different direction, why even start by splitting out Support as its own shared library? We already have the single-DSO build config, why not improve that by adding annotations and using -fvisibility=hidden?

llvm/lib/Support/AArch64TargetParser.cpp

27 ↗

(On Diff #370395)

Is this necessary? In general, you shouldn't have to annotate the definitions so long as the previous declaration was annotated. I don't see any compilers that warn about this:
https://gcc.godbolt.org/z/E7T1c9z81

llvm/lib/Support/CMakeLists.txt

104

As I understand it, this makes the support library always shared. I don't think that gives people using the traditional static build a pathway to transition. What I had in mind was making this conditional on LLVM_BUILD_DYLIB or whatever we're calling it, the one that builds libLLVM.so, the one all the Linux distros use.

I think our long term goal should be to align the default LLVM cmake build configuration with the one that distributors are using. The fact that most devs don't use the most commonly distributed LLVM build configuration leads to surprises like the ones from earlier this year:
https://maskray.me/blog/2021-05-16-elf-interposition-and-bsymbolic
https://lore.kernel.org/lkml/CAHk-=whs8QZf3YnifdLv57+FhBi5_WeNTG1B-suOES=[e-mail]/

[compnerd]

In D109192#2982999, @rnk wrote:

In D109192#2982723, @compnerd wrote:

Some additional questions:

• should we be encouraging of exporting entire classes or should we be more judicious in the visibility (e.g. attribute members)? I'm not 100% certain that this works for non-dll-interface targets.

I think you can give entire classes default visibility, and that will apply to all members. I think we should annotate classes because it will reduce the total number of annotations, and that will reduce the churn. For large, key classes that have too many private methods (Sema, ASTContext?) we can go back later and export only the public APIs, or something like that.

This is sort of the direction that I had in mind. The problem is, there are cases where I cannot get the desired behaviour correct. Doing those where it doesn't just work seems like an easier path and one that could be refined later. Its a bit unfortunate, but seems like a decent trade off.

I'm more convinced now that we should introduce an option to introduce this behaviour. The new option can opt into building the shared library components. Perhaps something like LLVM_EXPERIMENTAL_SHARED_LIBRARY_COMPONENTS is a good option name?

I agree this new thing shouldn't become the default build config. How close is this to BUILD_SHARED_LIBS vs LLVM_ENABLE_DYLIB? Can we repurpose one of those?

I think that BUILD_SHARED_LIBS is desirable, and LLVM_ENABLE_DYLIB should go away. That said, I think that the ideal thing here is to have a different configuration that is less granular than BUILD_SHARED_LIBS. That would correspond to LLVMTools, LLVMCompiler and depending on the delta, a potential third library. These three would give us a nice hybrid approach to dynamic and static linking. It would be dynamic for buckets of use (bintuils, clang, and possibly lld if it makes LLVMTools too large) but statically linking large components of LLVM to enable us to take advantage of LTO and DCE amongst that set. It would try to minimise the cost of the GOT/PLT and IAT so that we don't introduce an extremely heavy penalty for these libraries.

---

Going in a completely different direction, why even start by splitting out Support as its own shared library? We already have the single-DSO build config, why not improve that by adding annotations and using -fvisibility=hidden?

I don't really care if we keep LLVMSupport as a separate library or not in the end. The point here is to work through a representative set to setup such a configuration. This gives us a good pattern to apply and tries to iron out the issues that this brings up in a smaller context. The -fvisbility=hidden (and -fvisibility-inlines-hidden) is a requirement for this so we do not continually regress. One of the hacks in this patch is the unconditional application of that to LLVMSupport (which I will remove as this patch gets closer to a final form). I don't think that the single DSO configuration is ideal either, we would still have an unnecessarily large overhead for tooling. I think that the ideal split would be a library sufficient for the binutils, a second library that has everything else that is needed for clang.

[aganea]

In D109192#2983041, @compnerd wrote:

In D109192#2982999, @rnk wrote:

In D109192#2982723, @compnerd wrote:

Some additional questions:

• should we be encouraging of exporting entire classes or should we be more judicious in the visibility (e.g. attribute members)? I'm not 100% certain that this works for non-dll-interface targets.

I think you can give entire classes default visibility, and that will apply to all members. I think we should annotate classes because it will reduce the total number of annotations, and that will reduce the churn. For large, key classes that have too many private methods (Sema, ASTContext?) we can go back later and export only the public APIs, or something like that.

This is sort of the direction that I had in mind. The problem is, there are cases where I cannot get the desired behaviour correct. Doing those where it doesn't just work seems like an easier path and one that could be refined later. Its a bit unfortunate, but seems like a decent trade off.

There's a 2^16 symbol limit per DLL on Windows, so exporting entire classes would potentially reach that limit much faster. From past experience, we started our DLL targets with class annotations, only to reach that limit, and fall back to function-only annotation later. It's a bit more tedious to mark functions independently, but also avoids the performance issues mentionned below. Exporting an entire class could introduce, by inadvertance, indirections at the caller site.
There's a also build time cost for exporting (too) many symbols, but maybe that less of a concern here (because it would only affect LLVM developers using this specific configuration).

Regarding the API boundaries, or how to wisely chose those boundaries: I am wondering if there's a way to make use of the whole program analysis, to find the performance-critical paths. Any function in that path shouldn't be exported, to avoid the DLL thunk indirection I was mentionning. Ideally we should only export "high level" functions and keep performance-critical code internal.

I am also wondering how does this fit with the llvm-busybox proposal? That goes in a completly different direction, by coalescing all of llvm/clang/lld in a single executable. We could simply go along those lines: a single DLL per "project" (clang, llvm, lld) with the existing bintools as executable wrappers on the top of that. Is that the single-DSO config that @rnk mentions? Doing that seems almost a no-brainer in terms of defining the API boundaries.

[mstorsjo]

In D109192#2983146, @aganea wrote:

There's a 2^16 symbol limit per DLL on Windows, so exporting entire classes would potentially reach that limit much faster.

FWIW, in mingw configurations, the existing dylib setup works (as mingw toolchains fall back to exporting all symbols if there's no dllexports). libLLVM is cutting it quite close though, it currently ends up at 58000 symbols, so not many to spare (and this with only X86, ARM and AArch64 targets enabled, I think it ends up too many if all targets are enabled, and if building with asserts enabled, there's a couple more too).

[compnerd]

In D109192#2983153, @mstorsjo wrote:

In D109192#2983146, @aganea wrote:

There's a 2^16 symbol limit per DLL on Windows, so exporting entire classes would potentially reach that limit much faster.

FWIW, in mingw configurations, the existing dylib setup works (as mingw toolchains fall back to exporting all symbols if there's no dllexports). libLLVM is cutting it quite close though, it currently ends up at 58000 symbols, so not many to spare (and this with only X86, ARM and AArch64 targets enabled, I think it ends up too many if all targets are enabled, and if building with asserts enabled, there's a couple more too).

That is both good and bad. Its good to know that we are able to get a build with everything exported, though I think that we will want to be able to include all the targets, and might be too close to that limit. However, I suspect that splitting up LLVM.dll into LLVM.dll and LLVMSupport.dll we might be able to scrape through.

[compnerd]

In D109192#2983146, @aganea wrote:

In D109192#2983041, @compnerd wrote:

In D109192#2982999, @rnk wrote:

In D109192#2982723, @compnerd wrote:

Some additional questions:

• should we be encouraging of exporting entire classes or should we be more judicious in the visibility (e.g. attribute members)? I'm not 100% certain that this works for non-dll-interface targets.

I think you can give entire classes default visibility, and that will apply to all members. I think we should annotate classes because it will reduce the total number of annotations, and that will reduce the churn. For large, key classes that have too many private methods (Sema, ASTContext?) we can go back later and export only the public APIs, or something like that.

This is sort of the direction that I had in mind. The problem is, there are cases where I cannot get the desired behaviour correct. Doing those where it doesn't just work seems like an easier path and one that could be refined later. Its a bit unfortunate, but seems like a decent trade off.

There's a 2^16 symbol limit per DLL on Windows, so exporting entire classes would potentially reach that limit much faster. From past experience, we started our DLL targets with class annotations, only to reach that limit, and fall back to function-only annotation later. It's a bit more tedious to mark functions independently, but also avoids the performance issues mentionned below. Exporting an entire class could introduce, by inadvertance, indirections at the caller site.

Yes, though, my reasoning for not wanting to annotate the non-inline functions only is less the churn and more the worry about the difference wrt vftable emission. Although, now that I think about it, IIRC, MS ABI will always universally emit the vftable and so this point is moot - we should fallback to the anchors for the vtable emission for itanium targets. Although, I think we might want to consider creating a macro for this now and eliding the anchor for MS ABI - that would save us a potentially large set of unnecessary exported symbols that do nothing on Windows.

There's a also build time cost for exporting (too) many symbols, but maybe that less of a concern here (because it would only affect LLVM developers using this specific configuration).

I think that there is another aspect beyond build time costs - it inhibits potential optimizations with LTO. Anything that can be fully internalized is a win in terms of potential LTO and DCE options.

Regarding the API boundaries, or how to wisely chose those boundaries: I am wondering if there's a way to make use of the whole program analysis, to find the performance-critical paths. Any function in that path shouldn't be exported, to avoid the DLL thunk indirection I was mentionning. Ideally we should only export "high level" functions and keep performance-critical code internal.

That would be nice to do as a follow up I think. I don't think that I could take that on, but would love to see it happen.

I am also wondering how does this fit with the llvm-busybox proposal? That goes in a completly different direction, by coalescing all of llvm/clang/lld in a single executable. We could simply go along those lines: a single DLL per "project" (clang, llvm, lld) with the existing bintools as executable wrappers on the top of that. Is that the single-DSO config that @rnk mentions? Doing that seems almost a no-brainer in terms of defining the API boundaries.

I don't think that the multicall binary is particularly great, but, yes, this will solve the problem of the size at least - all of the tools should be effectively in a single DLL. The frontends would be tiny. I am not hardset on the original idea of 2/3 DLLs for LLVM and 1/2 DLLs for clang. I could be convinced that 2 DLLs for LLVM (Support and LLVM) and 1 for clang is a decent enough split. My intuition is that 3 for LLVM and 1/2 for Clang would be a pretty good compromise (LLVMSupport, LLVMTooling, LLVMCompiler for LLVM and potentially just Clang for clang).

[compnerd]

Further progress; gets through most of the build of clang.

[compnerd]

llvm/lib/Support/CMakeLists.txt

104

You are correct, and this hunk is not something we can check in. However, it is the fastest way to burn through the annotations. Once we have the annotations correct, tweaking the build to match the correct behaviour is a simpler task.

[compnerd]

Rebase to the current main branch; further extend the annotations for building the unittests.

[MaskRay]

Thanks for working on this! I agree that LLVM_*_ABI is a better name because it is related to what interfaces other libraries can use.

In D109192#2983041, @compnerd wrote:

In D109192#2982999, @rnk wrote:

In D109192#2982723, @compnerd wrote:

Some additional questions:

• should we be encouraging of exporting entire classes or should we be more judicious in the visibility (e.g. attribute members)? I'm not 100% certain that this works for non-dll-interface targets.

I think you can give entire classes default visibility, and that will apply to all members. I think we should annotate classes because it will reduce the total number of annotations, and that will reduce the churn. For large, key classes that have too many private methods (Sema, ASTContext?) we can go back later and export only the public APIs, or something like that.

This is sort of the direction that I had in mind. The problem is, there are cases where I cannot get the desired behaviour correct. Doing those where it doesn't just work seems like an easier path and one that could be refined later. Its a bit unfortunate, but seems like a decent trade off.

I'm more convinced now that we should introduce an option to introduce this behaviour. The new option can opt into building the shared library components. Perhaps something like LLVM_EXPERIMENTAL_SHARED_LIBRARY_COMPONENTS is a good option name?

I agree this new thing shouldn't become the default build config. How close is this to BUILD_SHARED_LIBS vs LLVM_ENABLE_DYLIB? Can we repurpose one of those?

I think that BUILD_SHARED_LIBS is desirable, and LLVM_ENABLE_DYLIB should go away. That said, I think that the ideal thing here is to have a different configuration that is less granular than BUILD_SHARED_LIBS. That would correspond to LLVMTools, LLVMCompiler and depending on the delta, a potential third library. These three would give us a nice hybrid approach to dynamic and static linking. It would be dynamic for buckets of use (bintuils, clang, and possibly lld if it makes LLVMTools too large) but statically linking large components of LLVM to enable us to take advantage of LTO and DCE amongst that set. It would try to minimise the cost of the GOT/PLT and IAT so that we don't introduce an extremely heavy penalty for these libraries.

Why should LLVM_ENABLE_DYLIB go away? To reduce the number of configurations when the few-dylibs scheme is adopted more?
Other than that, I don't see any downside.

---

Going in a completely different direction, why even start by splitting out Support as its own shared library? We already have the single-DSO build config, why not improve that by adding annotations and using -fvisibility=hidden?

I don't really care if we keep LLVMSupport as a separate library or not in the end. The point here is to work through a representative set to setup such a configuration. This gives us a good pattern to apply and tries to iron out the issues that this brings up in a smaller context. The -fvisbility=hidden (and -fvisibility-inlines-hidden) is a requirement for this so we do not continually regress. One of the hacks in this patch is the unconditional application of that to LLVMSupport (which I will remove as this patch gets closer to a final form). I don't think that the single DSO configuration is ideal either, we would still have an unnecessarily large overhead for tooling. I think that the ideal split would be a library sufficient for the binutils, a second library that has everything else that is needed for clang.

llvm/include/llvm/Support/LLVMSupportExports.h

12

For flexibility, maybe give user a choice to redefine LLVM_*_ABI. Some users may want to customize this to hidden to enable better internalization (LTO,DCE) and linker GC.

llvm/lib/Support/CMakeLists.txt

289

-fvisibility=hidden implies -fvisibility-inlines-hidden

[fodinabor]

The missing symbols on Linux are: DebugFlag, isCurrentDebugType, Error::fatalUncheckedError and vtables of cl::opt, ErrorList and cl::OptionValue<std::string>.

llvm/include/llvm/ADT/APInt.h

72

Linux needs LLVM_SUPPORT_ABI LLVM_NODISCARD

Same with APSInt

llvm/include/llvm/ADT/SmallVector.h

92

missing class after template
two lines below as well

As the external functions are already marked LLVM_SUPPORT_ABI I'm not sure we need the annotation here anyways?
Otherwise, the LLVM_SUPPORT_ABI here should only be added for Linux, and the annotation has to be added to the explicit instantiation in SmallVector.cpp as well (for both Win / Linux)

llvm/include/llvm/Support/LLVMSupportExports.h

31

typo

llvm/include/llvm/Support/YAMLTraits.h

21

U typo, in MD5.h as well.

llvm/lib/Support/Windows/Threading.inc

27

same changes required for Unix/Threading.inc to silence the warnings there

[compnerd]

• builds clang, clang-tools-extra, lld
• cleans up C4910
• TODO clean up CMakeLists.txt hack
• C4251 remains to be sorted out (/wd4251 recommended on Windows)

Statistics:

• ~4K exports from LLVMSupport
• 32 anchors exported (unnecessary

I'd like to remove the anchors globally in a subsequent change, introduce a DECLARE_LLVM_KEY_FUNCTION and DECLARE_LLVM_KEY_FUNCTION(type) macro which would remove the anchors on Windows

• 364 symbols are being over-exported (private CXXMethodDecls/Variables)

I'd like to fix this subsequently as there is a cost associated with this on Windows - IAT relocation will be paid on them but will never be used and we cannot selectively mark them for late binding

[compnerd]

llvm/include/llvm/Support/LLVMSupportExports.h

12

Wait, we want to give the user the flexibility to *not* define that? I intend to make -fvisibility=hidden the only option once the annotations are complete. I want to simply globally set:

set(CMAKE_C_VISIBILITY_PRESET hidden)
set(CMAKE_CXX_VISIBILITY_PRESET hidden)
set(CMAKE_C_VISIBILITY_INLINES_HIDDEN YES)
set(CMAKE_CXX_VISIBILITY_INLINES_HIDDEN YES)

Everything in LLVM should be internalized unless you are doing a shared library build, where the annotations will expose the necessary interfaces.

Is there a use case that I am not considering?

[compnerd]

In D109192#2983824, @MaskRay wrote:

I think that BUILD_SHARED_LIBS is desirable, and LLVM_ENABLE_DYLIB should go away. That said, I think that the ideal thing here is to have a different configuration that is less granular than BUILD_SHARED_LIBS. That would correspond to LLVMTools, LLVMCompiler and depending on the delta, a potential third library. These three would give us a nice hybrid approach to dynamic and static linking. It would be dynamic for buckets of use (bintuils, clang, and possibly lld if it makes LLVMTools too large) but statically linking large components of LLVM to enable us to take advantage of LTO and DCE amongst that set. It would try to minimise the cost of the GOT/PLT and IAT so that we don't introduce an extremely heavy penalty for these libraries.

Why should LLVM_ENABLE_DYLIB go away? To reduce the number of configurations when the few-dylibs scheme is adopted more?
Other than that, I don't see any downside.

Because the LLVM build system is far too complex. Once we have the annotations, I don't see why we should continue to use custom build system logic versus the CMake standard BUILD_SHARED_LIBS. That would allow the few dylib scheme to be adopted and reduce the complexity in the build system and homogenise the distribution and development builds hopefully.

[jrtc27]

In D109192#2983869, @compnerd wrote:

In D109192#2983824, @MaskRay wrote:

I think that BUILD_SHARED_LIBS is desirable, and LLVM_ENABLE_DYLIB should go away. That said, I think that the ideal thing here is to have a different configuration that is less granular than BUILD_SHARED_LIBS. That would correspond to LLVMTools, LLVMCompiler and depending on the delta, a potential third library. These three would give us a nice hybrid approach to dynamic and static linking. It would be dynamic for buckets of use (bintuils, clang, and possibly lld if it makes LLVMTools too large) but statically linking large components of LLVM to enable us to take advantage of LTO and DCE amongst that set. It would try to minimise the cost of the GOT/PLT and IAT so that we don't introduce an extremely heavy penalty for these libraries.

Why should LLVM_ENABLE_DYLIB go away? To reduce the number of configurations when the few-dylibs scheme is adopted more?
Other than that, I don't see any downside.

Because the LLVM build system is far too complex. Once we have the annotations, I don't see why we should continue to use custom build system logic versus the CMake standard BUILD_SHARED_LIBS. That would allow the few dylib scheme to be adopted and reduce the complexity in the build system and homogenise the distribution and development builds hopefully.

The whole point of the option though is to make incremental builds faster which is really helpful for development. I'd be against removing that as all my development builds use it.

[compnerd]

• correct order of LLVM_NODISCARD and LLVM_SUPPORT_ABI
• annotate kInvalidFile

[compnerd]

• force a vtable instantiation to create an anchor for the for opt specializations
• more tweaks for Windows path (clang-test and llvm-test now build)

[compnerd]

• correct declaration for cl::opt extern template instantiations
• adjust ABI breaking check handling

[compnerd]

correct typo in CommandLine.cpp

[compnerd]

annotate some types for MLIR from flang

[compnerd]

check-mlir should build now

[jrtc27]

As before: what's the strategy for downstreams that modify these libraries? Has this been done entirely by hand, or is there some kind of script they can use to automate annotating their own additions?

[compnerd]

In D109192#2984222, @jrtc27 wrote:

As before: what's the strategy for downstreams that modify these libraries? Has this been done entirely by hand, or is there some kind of script they can use to automate annotating their own additions?

Before?

Yes, this has been done by hand. I'm not sure if I can get a perfect result through automation (https://github.com/compnerd/libtool), but I'm looking into that as an option. At least for the standalone functions, its relatively easy to get the desired behaviour correct. But, in general, for downstream repositories, this will likely require the usual porting and an additional audit of their changes.

[compnerd]

I should mention that this doesn't matter to the downstream projects at the end of the day - they aren't compelled to enable the shared library mode - static linking will be unimpacted.

[jrtc27]

In D109192#2984249, @compnerd wrote:

I should mention that this doesn't matter to the downstream projects at the end of the day - they aren't compelled to enable the shared library mode - static linking will be unimpacted.

We have significant downstream changes, and we regularly build with shared libraries for development, only static linking in our release builds. This will affect us and having to chase down a bunch of functions with LLVM_SUPPORT_ABI does not fill me with joy. Yes, static linking is great for releases, but it sucks for development, full per-component shared libraries make things much faster, please don't just dismiss it.

[compnerd]

• annotate Errno.h functions
• change StringRef to decorate explicit functions, fixing flang builds

[compnerd]

In D109192#2984252, @jrtc27 wrote:

In D109192#2984249, @compnerd wrote:

I should mention that this doesn't matter to the downstream projects at the end of the day - they aren't compelled to enable the shared library mode - static linking will be unimpacted.

We have significant downstream changes, and we regularly build with shared libraries for development, only static linking in our release builds. This will affect us and having to chase down a bunch of functions with LLVM_SUPPORT_ABI does not fill me with joy. Yes, static linking is great for releases, but it sucks for development, full per-component shared libraries make things much faster, please don't just dismiss it.

Well, per-component would be significantly worse - LLVM_SUPPORT_ABI marks what just LLVMSupport exports. You would need a separate macro per component if you do a shared library build with a per-component shared library. Doing a set 2-3 libraries means that there are only 2-3 macros that need to be contended with and its easier to identify which module the function resides in. That also means that it is more viable to find LTO/DCE optimizations.

[jrtc27]

In D109192#2984255, @compnerd wrote:

In D109192#2984252, @jrtc27 wrote:

In D109192#2984249, @compnerd wrote:

I should mention that this doesn't matter to the downstream projects at the end of the day - they aren't compelled to enable the shared library mode - static linking will be unimpacted.

We have significant downstream changes, and we regularly build with shared libraries for development, only static linking in our release builds. This will affect us and having to chase down a bunch of functions with LLVM_SUPPORT_ABI does not fill me with joy. Yes, static linking is great for releases, but it sucks for development, full per-component shared libraries make things much faster, please don't just dismiss it.

Well, per-component would be significantly worse - LLVM_SUPPORT_ABI marks what just LLVMSupport exports. You would need a separate macro per component if you do a shared library build with a per-component shared library. Doing a set 2-3 libraries means that there are only 2-3 macros that need to be contended with and its easier to identify which module the function resides in. That also means that it is more viable to find LTO/DCE optimizations.

I don't care about efficiency downstream, I just don't want it to be broken. If you break per-component shared libraries you force me to change my workflow and have slower incremental builds. If I have to go chasing down build errors because our additions need exporting in order for the monolithic shared build to work then I get grumpy at the busywork. If there's strong technical justification for the latter, fine, but please do it in as least disruptive a way as possible (i.e. with as much automation as possible, and in one go rather than spread across 10s of commits). I see no technical justification for the former, that would just be breaking workflows that you personally don't use.

[compnerd]

In D109192#2984273, @jrtc27 wrote:

In D109192#2984255, @compnerd wrote:

In D109192#2984252, @jrtc27 wrote:

In D109192#2984249, @compnerd wrote:

I should mention that this doesn't matter to the downstream projects at the end of the day - they aren't compelled to enable the shared library mode - static linking will be unimpacted.

We have significant downstream changes, and we regularly build with shared libraries for development, only static linking in our release builds. This will affect us and having to chase down a bunch of functions with LLVM_SUPPORT_ABI does not fill me with joy. Yes, static linking is great for releases, but it sucks for development, full per-component shared libraries make things much faster, please don't just dismiss it.

Well, per-component would be significantly worse - LLVM_SUPPORT_ABI marks what just LLVMSupport exports. You would need a separate macro per component if you do a shared library build with a per-component shared library. Doing a set 2-3 libraries means that there are only 2-3 macros that need to be contended with and its easier to identify which module the function resides in. That also means that it is more viable to find LTO/DCE optimizations.

I don't care about efficiency downstream, I just don't want it to be broken. If you break per-component shared libraries you force me to change my workflow and have slower incremental builds. If I have to go chasing down build errors because our additions need exporting in order for the monolithic shared build to work then I get grumpy at the busywork. If there's strong technical justification for the latter, fine, but please do it in as least disruptive a way as possible (i.e. with as much automation as possible, and in one go rather than spread across 10s of commits). I see no technical justification for the former, that would just be breaking workflows that you personally don't use.

Well, it may be that your downstream project doesn't care about efficiency, but every project has a differing set of needs and makes different tradeoffs. This is an issue for LLVM, and impacting performance of the compiler to make a particular workflow possible is not valid justification. There are technical grounds for preferring the semi-monolithic library for the tooling and the compiler.

[compnerd]

Expose LLVMParseCommandLineOptions

[jrtc27]

In D109192#2984294, @compnerd wrote:

In D109192#2984273, @jrtc27 wrote:

In D109192#2984255, @compnerd wrote:

In D109192#2984252, @jrtc27 wrote:

In D109192#2984249, @compnerd wrote:

I should mention that this doesn't matter to the downstream projects at the end of the day - they aren't compelled to enable the shared library mode - static linking will be unimpacted.

We have significant downstream changes, and we regularly build with shared libraries for development, only static linking in our release builds. This will affect us and having to chase down a bunch of functions with LLVM_SUPPORT_ABI does not fill me with joy. Yes, static linking is great for releases, but it sucks for development, full per-component shared libraries make things much faster, please don't just dismiss it.

Well, per-component would be significantly worse - LLVM_SUPPORT_ABI marks what just LLVMSupport exports. You would need a separate macro per component if you do a shared library build with a per-component shared library. Doing a set 2-3 libraries means that there are only 2-3 macros that need to be contended with and its easier to identify which module the function resides in. That also means that it is more viable to find LTO/DCE optimizations.

I don't care about efficiency downstream, I just don't want it to be broken. If you break per-component shared libraries you force me to change my workflow and have slower incremental builds. If I have to go chasing down build errors because our additions need exporting in order for the monolithic shared build to work then I get grumpy at the busywork. If there's strong technical justification for the latter, fine, but please do it in as least disruptive a way as possible (i.e. with as much automation as possible, and in one go rather than spread across 10s of commits). I see no technical justification for the former, that would just be breaking workflows that you personally don't use.

Well, it may be that your downstream project doesn't care about efficiency, but every project has a differing set of needs and makes different tradeoffs. This is an issue for LLVM, and impacting performance of the compiler to make a particular workflow possible is not valid justification. There are technical grounds for preferring the semi-monolithic library for the tooling and the compiler.

Preferring != requiring. Of course if there's sufficient technical justification for removing that option then fine, but that's for the community to decide, not you.

[lebedev.ri]

FWIW, i personally think that the current set of libs BUILD_SHARED_LIBS produces should not change, it's intentional that way, to both be fast and fine-grained.
And LLVM_BUILD_DYLIB is also intentionally producing a huge monolithic library, so that only a single library needs to be dlopened for speed reasons.
IIRC someone from MLIR also wanted to do something about library restructurization, i would suggest to talk to them.

All that being said, if either BUILD_SHARED_LIBS or LLVM_BUILD_DYLIB is being changed from what they are now,
i think this requires an RFC to llvm-dev with precise explanation of what why and how,
and why it's okay to regress existing granularity/speed/performance expectations.

[compnerd]

In D109192#2984318, @lebedev.ri wrote:

FWIW, i personally think that the current set of libs BUILD_SHARED_LIBS produces should not change, it's intentional that way, to both be fast and fine-grained.
And LLVM_BUILD_DYLIB is also intentionally producing a huge monolithic library, so that only a single library needs to be dlopened for speed reasons.
IIRC someone from MLIR also wanted to do something about library restructurization, i would suggest to talk to them.

All that being said, if either BUILD_SHARED_LIBS or LLVM_BUILD_DYLIB is being changed from what they are now,
i think this requires an RFC to llvm-dev with precise explanation of what why and how,
and why it's okay to regress existing granularity/speed/performance expectations.

This change has no impact on what BUILD_SHARED_LIBS does. It does however fix the current problem that LLVMSupport on WIndows when built as a shared library has no public functions and on ELF and MachO targets exports everything. Instead, this will explicitly control the public surface of the library's interface when built as a shared library.

There are similar changes that will be needed to support building LLVM as a shared library, but that is a separate larger change and I have no intention of merging that change into this change.

[MaskRay]

If I am to summarize the state. The currently ELF build models are:

• default: libLLVMSupport.a
• -DBUILD_SHARED_LIBS=on: libLLVMSupport.so
• -DLLVM_LINK_LLVM_DYLIB=on: libLLVM-14git.so

MSVC dynamic linking doesn't work.

From

• minimal set of interfaces required for llvm-ar (archiver/librarian)
• minimal set of interfaces required for lld (linker)
• minimal set of interfaces required for clang (compiler)

it seems that the intention is to have libLLVM{foo,bar,qux}.so as the end state? And that is a controversial point among at least @compnerd, @jrtc27, and me.

I think the main purpose of this patch is to introduce these source file annotations (LLVM_SUPPORT_ABI #include "llvm/Support/LLVMSupportExports.h") which will be the backbone of whatever dynamic linking mode we end up doing.
And we should ensure all parties are satisfied with the solution.
If you remove the behavior changing CMake changes, I will give a strong support.

Specifically, any risky CMake or visibility annotation changes should be dropped from this starting point patch. More concretely,

• #define LLVM_SUPPORT_ABI is acceptable since it is clearly a no-op.
• #define LLVM_SUPPORT_ABI __attribute__((visibility("protected"))) can be risky and is thus unacceptable
• Changing -DBUILD_SHARED_LIBS=on behavior is unacceptable (this presumably doesn't work on MSVC, so changing MSVC is fine). MinGW seems to support this config, so don't change its behavior (CC @mstorsjo)
• Changing -DLLVM_LINK_LLVM_DYLIB=on behavior is unacceptable (this doesn't work on MSVC, so changing MSVC is fine)

---

Currently there is a hacking change in lib/Support/CMakeLists.txt: add_llvm_component_library(LLVMSupport SHARED
This breaks the regular libLLVMSupport.a build, so it should be fixed.

The libLLVM{foo,bar,qux}.so design is related to the Windows DLL limitation about 64k exported symbols.
To reduce platform differences (ELF/Mach-O vs PE/COFF), @compnerd wants the libLLVM{foo,bar,qux}.so design to be the way forward and the monolithic libLLVM-14git.so scheme to be dropped.
I agree with @lebedev.ri that such change needs to discussed on llvm-dev.
The libLLVM-14git.so scheme is adopted by multiple major Linux distributions with many LLVM dependent packages.
Such a split should not be taken lightly.

FWIW For my https://github.com/MaskRay/ccls/blob/master/CMakeLists.txt#L95 , I'll need to adjust

if(LLVM_LINK_LLVM_DYLIB)
  target_link_libraries(ccls PRIVATE LLVM)

Other packages may need similar adaptation as well.
Personally I consider that ELF/Mach-O changing libLLVM-14git.so to cater to Windows DLL limitation unfortunate, but I can be convinced if the split argument (platform difference, maintenance cost consideration) is sufficiently strong.

[compnerd]

In D109192#2984321, @MaskRay wrote:

If I am to summarize the state. The currently ELF build models are:

• default: libLLVMSupport.a
• -DBUILD_SHARED_LIBS=on: libLLVMSupport.so
• -DLLVM_LINK_LLVM_DYLIB=on: libLLVM-14git.so

Windows dynamic linking doesn't work.

From

• minimal set of interfaces required for llvm-ar (archiver/librarian)
• minimal set of interfaces required for lld (linker)
• minimal set of interfaces required for clang (compiler)

it seems that the intention is to have libLLVM{foo,bar,qux}.so as the end state? And that is a controversial point among at least @compnerd, @jrtc27, and me.

I think the main purpose of this patch is to introduce these source file annotations (LLVM_SUPPORT_ABI #include "llvm/Support/LLVMSupportExports.h") which will be the backbone of whatever dynamic linking mode we end up doing.
And we should ensure all parties are satisfied with the solution.
If you remove the behavior changing CMake changes, I will give a strong support.

More to introduce the model for the annotations. The annotations are module dependent, that is each module must introduce its own decoration and the function must have a clear owner defined, which is why LLVMSupport is a good basis for the conceptual change - the intent is that the functions in LLVMSupport are owned by LLVMSupport and thus will be annotated with the LLVM_SUPPORT_ABI.

Specifically, any risky CMake or visibility annotation should be dropped from this starting point.

I think that this we disagree on. I think that the visibility handling needs to be part of this - otherwise the behaviour becomes too complicated to reason about. The default beahviour on Windows is that the symbol is dso_local and the default behaviour on ELF and MachO targets is *not* dso_local. This means that we have different defaults and any future change will need to contend with this. By changing the visibility for LLVMSupport when building LLVMSupport as a shared library to hidden visibility, we should get consistent behaviour across the board and make it easier to reason through for changes.

E.g.

• #define LLVM_SUPPORT_ABI is acceptable since it is clearly a no-op.
• #define LLVM_SUPPORT_ABI __attribute__((visibility("protected"))) can be risky and is thus unacceptable
• Changing -DBUILD_SHARED_LIBS=on behavior is unacceptable (this presumably doesn't work on Windows, so changing Windows is fine)
• Changing -DLLVM_LINK_LLVM_DYLIB=on behavior is unacceptable (this doesn't work on Windows, so changing Windows is fine)

---

Currently there is a hacking change in lib/Support/CMakeLists.txt: add_llvm_component_library(LLVMSupport SHARED
This breaks the regular libLLVMSupport.a build, so it should be fixed.

Yes, and I even commented as such that it is unacceptable, but is done to allow for quick turnaround on testing. Once the pre-commit checks have verified that it is happy with the build, this will be changed to restore the build of LLVMSupport to the way that it is handled today.

The libLLVM{foo,bar,qux}.so design is related to the Windows DLL limitation about 64k exported symbols.
To reduce platform differences (ELF/Mach-O vs PE/COFF), @compnerd wants the libLLVM{foo,bar,qux}.so design to be the way forward and the monolithic libLLVM-14git.so scheme to be dropped.
I agree with @lebedev.ri that such change needs to discussed on llvm-dev.
The libLLVM-14git.so scheme is adopted by multiple major Linux distributions with many LLVM dependent packages.
Such a split should not be taken lightly.

Yes I do expect that we are going to hit the limit of the exported symbols and thus its very likely to require that split. However, I cannot say with certainty that will happen until we get to that point. Once LLVMSupport changes are baked enough to demonstrate the pattern, the changes to allow building LLVM as a library can be tested to see what the current limitation is. The more fragmented the build, the worse it is for optimizations and size reduction possibilities. It also makes distribution more complicated.

FWIW For my https://github.com/MaskRay/ccls/blob/master/CMakeLists.txt#L95 , I'll need to adjust

if(LLVM_LINK_LLVM_DYLIB)
  target_link_libraries(ccls PRIVATE LLVM)

Other packages may need similar adaptation as well.
Personally I consider that ELF/Mach-O changing libLLVM-14git.so to cater to Windows DLL limitation unfortunate, but I can be convinced if the split argument (platform difference, maintenance cost consideration) is sufficiently strong.

I don't think that it is "catering" nor "unfortunate". There are tradeoffs in software engineering, and this is one of them. There are similar issues where LLVM is doing things that are non-optimal and other platforms suffer from them. An example is the use of the anchor key functions which are meaningless on some platforms.

The current state of the shared library build is another example of the "unfortunate" results of supporting ELF targets (I say this as a counterpoint to your argument that it is unfortunate that ELF must do something different, I tend to care about ELF targets just as much as PE/COFF targets).

Platforms are different, and software engineering requires that you evaluate different tradeoffs. There is a rich design space for structuring binaries, and choices aren't inherently unfortunate. They are evaluating with different weights being ascribed to different features, and that results in different choices.

Keeping LLVM friendly to a variety of platforms with contributors developing and testing on a different set of platforms means that it is best to have it behave as consistently as possible. This reduces the likelihood of someone unintentionally regressing a system other than the one that they are developing on.

[compnerd]

rebase

[compnerd]

The c-index-test failure is interesting. libclang will link against LLVMPasses and have the enable-ml-inliner option and thus register it. c-index-tool itself links against LLVMPasses and thus registers the option again, resulting in the duplicate registration.

[echristo]

Hi Saleem,

I think I understand where you're going, but I think this deserves a writeup and consideration on llvm-dev with your desired endpoint for the source base and what changes to both current and future development will need to happen with some benefits and costs outlined. Might want to include a "support story" for the project along these lines as well - or at least be clear that this doesn't change or affect what long term APIs we are willing to support.

Thoughts?

[compnerd]

In D109192#2990039, @echristo wrote:

Hi Saleem,

I think I understand where you're going, but I think this deserves a writeup and consideration on llvm-dev with your desired endpoint for the source base and what changes to both current and future development will need to happen with some benefits and costs outlined. Might want to include a "support story" for the project along these lines as well - or at least be clear that this doesn't change or affect what long term APIs we are willing to support.

Thoughts?

I think that is inline with what I had expected. I actually asked @rnk about his thoughts on this as well, and he was of the opinion that this wouldn't be a big surprise and to send out a proposed change first. I wanted to get through LLVMSupport to the point that it would be possible to show what an approach for enabling a shared library build that could support all the platforms would look like. I don't have any concerns with sending out a small write up of what the changes once at least the Windows pre-commit CI is green. I already had expected to have to do that subsequently as I don't expect to tackle LLVM in a single change. I had expected this to serve as a preparatory change to discuss the necessary changes for permitting LLVM_ENABLE_DYLIB builds on Windows.

[arichardson]

llvm/include/llvm/Support/LLVMSupportExports.h

12

If we need to add this for every library, it seems like the header should be generated?

Maybe we could even re-use https://cmake.org/cmake/help/latest/module/GenerateExportHeader.html?

[rnk]

llvm/include/llvm/Support/LLVMSupportExports.h

12

CMake is the official build system for LLVM, so maybe this should not weigh on the project priorities, but there are two unofficial build systems in-tree that would have to replicate this functionality (Bazel & gn).

I think we should strive to keep the build system as simple as possible, even if it has ways to help us reduce boilerplate code. Most LLVM developers are not CMake experts, but we can all read and reason about C++ and C pre-processor directives.

I believe there are ways to factor out the selection of the appropriate attribute using token pasting.

[compnerd]

llvm/include/llvm/Support/LLVMSupportExports.h

12

We would in order to support BUILD_SHARED_LIBS universally. However, my goal is not to enable BUILD_SHARED_LIBS but rather allow LLVM_BUILD_LLVM_DYLIB which builds a single shared library. LLVMSupport is a good target to demonstrate what is needed without having to implement the full coverage for all of LLVM without actually reaching some consensus. That said, I would certainly be interested in having LLVMSupport as a shared library as well.

[fortsnek9348]

This comment was removed by fortsnek9348.

[tstellar]

After talking with @compnerd on Discord, I'm planning on helping to push this
patch through. At this point my goal is just to annotate the ABI functions and
not make any changes to how -DBUILD_SHARED_LIBS=ON or -DLLVM_LINK_LLVM_DYLIB=ON
behave. We can make changes to those configurations in a future patch (if we
want to).

My plan is to:

1. Annotate all the ABI functions in the LLVM support library with LLVM_SUPPORT_ABI macro (this patch), and use this an example for how to do the rest of the annotations.
2. Annotate the rest of the LLVM ABI functions using a different macro (e.g. LLVM_ABI)
3. Annotate the C ABI with a third macro (e.g. LLVM_C_ABI).
4. Annotate the clang C++ ABI with another macro (e.g. CLANG_CPP_ABI).
5. Do an analysis of which ABI functions usage inside and outside of the tree, and remove unused functions from the public ABI (i.e. mark them as visibility hidden).

For this patch, I've rebased it on main and made a few changes:

• Add ABI annotations to classes and remove them from class member functions.

This will make the patch smaller and easier to maintain. We can always revisit
this in the future once we have the complete set of ABIs annotated.

• Use default visibility instead of protected visibility on Linux.

• Drop the add_llvm_component_library(LLVMSupport SHARED change.

[nikic]

In D109192#4407274, @tstellar wrote:

After talking with @compnerd on Discord, I'm planning on helping to push this
patch through. At this point my goal is just to annotate the ABI functions and
not make any changes to how -DBUILD_SHARED_LIBS=ON or -DLLVM_LINK_LLVM_DYLIB=ON
behave. We can make changes to those configurations in a future patch (if we
want to).

My plan is to:

1. Annotate all the ABI functions in the LLVM support library with LLVM_SUPPORT_ABI macro (this patch), and use this an example for how to do the rest of the annotations.
2. Annotate the rest of the LLVM ABI functions using a different macro (e.g. LLVM_ABI)

Why do we need/want to annotate the support library with a different macro from everything else? Isn't LLVMSupport part of libLLVM just like the rest?

We already have the LLVM_EXTERNAL_VISIBILITY and LLVM_LIBRARY_VISIBILITY macros. LLVM_EXTERNAL_VISIBILITY is used in the places that already use the -fvisibility=hidden default, namely the backend. LLVM_LIBRARY_VISIBILTY is used in a few places that want to explicitly exclude symbols in code that still uses default visibilty.

I would have expected that this would annotate symbols using the existing LLVM_EXTERNAL_VISIBILITY macro (and LLVM_LIBRARY_VISIBILITY could be retired once everything uses -fvisibility=hidden).

llvm/include/llvm/Support/LLVMSupportExports.h

14

What is the LLVM_SUPPORT_STATIC variable for? I don't see it defined anywhere.

[compnerd]

In D109192#4443741, @nikic wrote:

In D109192#4407274, @tstellar wrote:

After talking with @compnerd on Discord, I'm planning on helping to push this
patch through. At this point my goal is just to annotate the ABI functions and
not make any changes to how -DBUILD_SHARED_LIBS=ON or -DLLVM_LINK_LLVM_DYLIB=ON
behave. We can make changes to those configurations in a future patch (if we
want to).

My plan is to:

1. Annotate all the ABI functions in the LLVM support library with LLVM_SUPPORT_ABI macro (this patch), and use this an example for how to do the rest of the annotations.
2. Annotate the rest of the LLVM ABI functions using a different macro (e.g. LLVM_ABI)

Why do we need/want to annotate the support library with a different macro from everything else? Isn't LLVMSupport part of libLLVM just like the rest?

LLVMSupport is used very commonly as a support library outside of LLVM.

We already have the LLVM_EXTERNAL_VISIBILITY and LLVM_LIBRARY_VISIBILITY macros. LLVM_EXTERNAL_VISIBILITY is used in the places that already use the -fvisibility=hidden default, namely the backend. LLVM_LIBRARY_VISIBILTY is used in a few places that want to explicitly exclude symbols in code that still uses default visibilty.

I would have expected that this would annotate symbols using the existing LLVM_EXTERNAL_VISIBILITY macro (and LLVM_LIBRARY_VISIBILITY could be retired once everything uses -fvisibility=hidden).

Visibility and DLL Storage are technically different and provide subtly different things. LLVM_EXTERNAL_VISIBILITY does not take the library into account nor does it take into account the static vs dynamic linking nature and thus would not be sufficient for the purposes of DLL Storage.

[rnk]

In D109192#4443741, @nikic wrote:

Why do we need/want to annotate the support library with a different macro from everything else? Isn't LLVMSupport part of libLLVM just like the rest?

We already have the LLVM_EXTERNAL_VISIBILITY and LLVM_LIBRARY_VISIBILITY macros. LLVM_EXTERNAL_VISIBILITY is used in the places that already use the -fvisibility=hidden default, namely the backend. LLVM_LIBRARY_VISIBILTY is used in a few places that want to explicitly exclude symbols in code that still uses default visibilty.

I would have expected that this would annotate symbols using the existing LLVM_EXTERNAL_VISIBILITY macro (and LLVM_LIBRARY_VISIBILITY could be retired once everything uses -fvisibility=hidden).

For DLL import/export annotations, we need to have distinct macros for every DLL we plan to make. The question then becomes, do we need libLLVMSupport.dll?

Building Support as a DLL is by far the most conventional and straightforward approach. There are many tools (FileCheck, llvm-tblgen) that use Support but not all of LLVM, and they need to be built first.

The alternative is that we place dllexport annotations across the Support APIs, but then we come up with some creative build rules to produce a special static Support library for all of the tools that don't or can't link to the main LLVM DSO. This approach has the risk of leading users into accidentally linking both libraries, which then results in linker errors or silent ODR violations with duplicated global state. We can avoid all this if we are willing to tolerate the complexity of using different ABI annotations in Support headers.

Personally, I think in the long run, having a separate support DLL will be the least confusing approach and will have the fewest surprises. However, I'm speaking as someone who contributed to the implementation of dllexport semantics in clang, so I am probably much more familiar with them than the average LLVM developer.

[tstellar]

Given that my plan right now is just to annotate the ABI and not create any new shared objects, should we still use a separate annotation for Support or should we just use the same annotation for the whole library and then specialize it later?

[nikic]

In D109192#4444582, @rnk wrote:

In D109192#4443741, @nikic wrote:

Why do we need/want to annotate the support library with a different macro from everything else? Isn't LLVMSupport part of libLLVM just like the rest?

We already have the LLVM_EXTERNAL_VISIBILITY and LLVM_LIBRARY_VISIBILITY macros. LLVM_EXTERNAL_VISIBILITY is used in the places that already use the -fvisibility=hidden default, namely the backend. LLVM_LIBRARY_VISIBILTY is used in a few places that want to explicitly exclude symbols in code that still uses default visibilty.

I would have expected that this would annotate symbols using the existing LLVM_EXTERNAL_VISIBILITY macro (and LLVM_LIBRARY_VISIBILITY could be retired once everything uses -fvisibility=hidden).

For DLL import/export annotations, we need to have distinct macros for every DLL we plan to make. The question then becomes, do we need libLLVMSupport.dll?

Building Support as a DLL is by far the most conventional and straightforward approach. There are many tools (FileCheck, llvm-tblgen) that use Support but not all of LLVM, and they need to be built first.

The alternative is that we place dllexport annotations across the Support APIs, but then we come up with some creative build rules to produce a special static Support library for all of the tools that don't or can't link to the main LLVM DSO. This approach has the risk of leading users into accidentally linking both libraries, which then results in linker errors or silent ODR violations with duplicated global state. We can avoid all this if we are willing to tolerate the complexity of using different ABI annotations in Support headers.

Personally, I think in the long run, having a separate support DLL will be the least confusing approach and will have the fewest surprises. However, I'm speaking as someone who contributed to the implementation of dllexport semantics in clang, so I am probably much more familiar with them than the average LLVM developer.

So I think I understand why we need a separate ABI macro and can't reuse the existing VISIBILTY ones, but I'm still not clear on why Support is special. Currently (if we ignore BUILD_SHARED_LIBS, which is orthogonal to this) we only support a single libLLVM dylib. Cases that cannot link the dylib for some reason will link against static libraries instead. (Things like tablegen I assume, but also things like llvm-exegesis that use private symbols.)

Is this about some hypothetical future extension where we split up libLLVM.so into smaller libraries (something I am somewhat skeptical about) or is there some Windows idiosyncrasy that requires the special treatment of the Support library?

[tstellar]

I'm leaning towards not having a separate macro for LLVMSupport and just using the LLVM_ABI macro for LLVMSupport and everything else.

[nikic]

In D109192#4474895, @tstellar wrote:

I'm leaning towards not having a separate macro for LLVMSupport and just using the LLVM_ABI macro for LLVMSupport and everything else.

I agree. We should have one macro for each currently existing dylib -- which is just the one for LLVM. If we want to introduce additional dylibs in the future, we should split the macro at that point in time.

[tstellar]

• Use generic LLVM_ABI macro instead of LLVM_SUPPORT_ABI
• Rebase

[rnk]

If we're not making a separate DLL for LLVM support, that sounds fine to me.

I'm worried that by limiting scope here, we're just deferring the discovery and resolution of problems that @compnerd already grappled with in his attempt to make LLVM DLLs, but it's also good to be incremental and get things moving.

[tstellar]

In D109192#4486226, @rnk wrote:

If we're not making a separate DLL for LLVM support, that sounds fine to me.

I'm worried that by limiting scope here, we're just deferring the discovery and resolution of problems that @compnerd already grappled with in his attempt to make LLVM DLLs, but it's also good to be incremental and get things moving.

I've got a WIP patch to add annotations to the rest of the ABI, I can merge that into this one if you think that's better.

[rnk]

In D109192#4486728, @tstellar wrote:

In D109192#4486226, @rnk wrote:

If we're not making a separate DLL for LLVM support, that sounds fine to me.

I'm worried that by limiting scope here, we're just deferring the discovery and resolution of problems that @compnerd already grappled with in his attempt to make LLVM DLLs, but it's also good to be incremental and get things moving.

I've got a WIP patch to add annotations to the rest of the ABI, I can merge that into this one if you think that's better.

Either approach seems fine to me.

[mstorsjo]

Even if this is approved, make sure that it actually successfully builds in Windows configs (last time I tried an earlier version of this patch it broke the build very severely). The premerge CI doesn't seem to have succeeded in applying the patch to even try building it.

llvm/include/llvm-c/Error.h

 /*===------- llvm-c/Error.h - llvm::Error class C Interface -------*- C -*-===*\
 |*                                                                            *|
 |* Part of the LLVM Project, under the Apache License v2.0 with LLVM          *|
 |* Exceptions.                                                                *|
 |* See https://llvm.org/LICENSE.txt for license information.                  *|
 |* SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception                    *|
 |*                                                                            *|
 |*===----------------------------------------------------------------------===*|
 |*                                                                            *|
 |* This file defines the C interface to LLVM's Error class.                   *|
 |*                                                                            *|
 \*===----------------------------------------------------------------------===*/
 
 #ifndef LLVM_C_ERROR_H
 #define LLVM_C_ERROR_H
 
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm-c/ExternC.h"
 
 LLVM_C_EXTERN_C_BEGIN
 
 #define LLVMErrorSuccess 0
 
 /**
  * Opaque reference to an error instance. Null serves as the 'success' value.
  */
 typedef struct LLVMOpaqueError *LLVMErrorRef;
 
 /**
  * Error type identifier.
  */
 typedef const void *LLVMErrorTypeId;
 
 /**
  * Returns the type id for the given error instance, which must be a failure
  * value (i.e. non-null).
  */
-LLVMErrorTypeId LLVMGetErrorTypeId(LLVMErrorRef Err);
+LLVM_SUPPORT_ABI LLVMErrorTypeId LLVMGetErrorTypeId(LLVMErrorRef Err);
 
 /**
  * Dispose of the given error without handling it. This operation consumes the
  * error, and the given LLVMErrorRef value is not usable once this call returns.
  * Note: This method *only* needs to be called if the error is not being passed
  * to some other consuming operation, e.g. LLVMGetErrorMessage.
  */
-void LLVMConsumeError(LLVMErrorRef Err);
+LLVM_SUPPORT_ABI void LLVMConsumeError(LLVMErrorRef Err);
 
 /**
  * Returns the given string's error message. This operation consumes the error,
  * and the given LLVMErrorRef value is not usable once this call returns.
  * The caller is responsible for disposing of the string by calling
  * LLVMDisposeErrorMessage.
  */
-char *LLVMGetErrorMessage(LLVMErrorRef Err);
+LLVM_SUPPORT_ABI char *LLVMGetErrorMessage(LLVMErrorRef Err);
 
 /**
  * Dispose of the given error message.
  */
-void LLVMDisposeErrorMessage(char *ErrMsg);
+LLVM_SUPPORT_ABI void LLVMDisposeErrorMessage(char *ErrMsg);
 
 /**
  * Returns the type id for llvm StringError.
  */
-LLVMErrorTypeId LLVMGetStringErrorTypeId(void);
+LLVM_SUPPORT_ABI LLVMErrorTypeId LLVMGetStringErrorTypeId(void);
 
 /**
  * Create a StringError.
  */
 LLVMErrorRef LLVMCreateStringError(const char *ErrMsg);
 
 LLVM_C_EXTERN_C_END
 
 #endif

llvm/include/llvm-c/ErrorHandling.h

 |* This file defines the C interface to LLVM's error handling mechanism.      *|
 |*                                                                            *|
 \*===----------------------------------------------------------------------===*/
 
 #ifndef LLVM_C_ERRORHANDLING_H
 #define LLVM_C_ERRORHANDLING_H
 
 #include "llvm-c/ExternC.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 LLVM_C_EXTERN_C_BEGIN
 
 typedef void (*LLVMFatalErrorHandler)(const char *Reason);
 
 /**
  * Install a fatal error handler. By default, if LLVM detects a fatal error, it
  * will call exit(1). This may not be appropriate in many contexts. For example,

  */
 void LLVMResetFatalErrorHandler(void);
 
 /**
  * Enable LLVM's built-in stack trace code. This intercepts the OS's crash
  * signals and prints which component of LLVM you were in at the time if the
  * crash.
  */
-void LLVMEnablePrettyStackTrace(void);
+LLVM_SUPPORT_ABI void LLVMEnablePrettyStackTrace(void);
 
 LLVM_C_EXTERN_C_END
 
 #endif

llvm/include/llvm/ADT/APFixedPoint.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_APFIXEDPOINT_H
 #define LLVM_ADT_APFIXEDPOINT_H
 
 #include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/SmallString.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 class APFloat;
 struct fltSemantics;
 
 /// The fixed point semantics work similarly to fltSemantics. The width
 /// specifies the whole bit width of the underlying scaled integer (with padding
 /// if any). The scale represents the number of fractional bits in this type.
 /// When HasUnsignedPadding is true and this type is unsigned, the first bit
 /// in the value this represents is treated as padding.
-class FixedPointSemantics {
+class LLVM_SUPPORT_ABI FixedPointSemantics {
 public:
   FixedPointSemantics(unsigned Width, unsigned Scale, bool IsSigned,
                       bool IsSaturated, bool HasUnsignedPadding)
       : Width(Width), Scale(Scale), IsSigned(IsSigned),
         IsSaturated(IsSaturated), HasUnsignedPadding(HasUnsignedPadding) {
     assert(Width >= Scale && "Not enough room for the scale");
     assert(!(IsSigned && HasUnsignedPadding) &&
            "Cannot have unsigned padding on a signed type.");

 };
 
 /// The APFixedPoint class works similarly to APInt/APSInt in that it is a
 /// functional replacement for a scaled integer. It is meant to replicate the
 /// fixed point types proposed in ISO/IEC JTC1 SC22 WG14 N1169. The class carries
 /// info about the fixed point type's width, sign, scale, and saturation, and
 /// provides different operations that would normally be performed on fixed point
 /// types.
-class APFixedPoint {
+class LLVM_SUPPORT_ABI APFixedPoint {
 public:
   APFixedPoint(const APInt &Val, const FixedPointSemantics &Sema)
       : Val(Val, !Sema.isSigned()), Sema(Sema) {
     assert(Val.getBitWidth() == Sema.getWidth() &&
            "The value should have a bit width that matches the Sema width");
   }
 
   APFixedPoint(uint64_t Val, const FixedPointSemantics &Sema)

llvm/include/llvm/ADT/APFloat.h

 
 #ifndef LLVM_ADT_APFLOAT_H
 #define LLVM_ADT_APFLOAT_H
 
 #include "llvm/ADT/APInt.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/FloatingPointMode.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <memory>
 
 #define APFLOAT_DISPATCH_ON_SEMANTICS(METHOD_CALL)                             \
   do {                                                                         \
     if (usesLayout<IEEEFloat>(getSemantics()))                                 \
       return U.IEEE.METHOD_CALL;                                               \
     if (usesLayout<DoubleAPFloat>(getSemantics()))                             \
       return U.Double.METHOD_CALL;                                             \

 /// extended exponent range) (hard).
 ///
 /// New operations: sqrt, IEEE remainder, C90 fmod, nexttoward.
 ///
 
 // This is the common type definitions shared by APFloat and its internal
 // implementation classes. This struct should not define any non-static data
 // members.
-struct APFloatBase {
+struct LLVM_SUPPORT_ABI APFloatBase {
   typedef APInt::WordType integerPart;
   static constexpr unsigned integerPartWidth = APInt::APINT_BITS_PER_WORD;
 
   /// A signed type to represent a floating point numbers unbiased exponent.
   typedef int32_t ExponentType;
 
   /// \name Floating Point Semantics.
   /// @{
struct LLVM_SUPPORT_ABI APFloatBase {
 
   /// Returns the size of the floating point number (in bits) in the given
   /// semantics.
   static unsigned getSizeInBits(const fltSemantics &Sem);
 };
 
 namespace detail {
 
-class IEEEFloat final : public APFloatBase {
+class LLVM_SUPPORT_ABI IEEEFloat final : public APFloatBase {
 public:
   /// \name Constructors
   /// @{
 
   IEEEFloat(const fltSemantics &); // Default construct to +0.0
   IEEEFloat(const fltSemantics &, integerPart);
   IEEEFloat(const fltSemantics &, uninitializedTag);
   IEEEFloat(const fltSemantics &, const APInt &);
public:
   ///
   /// Note that the use of hash codes for floating point values is in general
   /// frought with peril. Equality is hard to define for these values. For
   /// example, should negative and positive zero hash to different codes? Are
   /// they equal or not? This hash value implementation specifically
   /// emphasizes producing different codes for different inputs in order to
   /// be used in canonicalization and memoization. As such, equality is
   /// bitwiseIsEqual, and 0 != -0.
-  friend hash_code hash_value(const IEEEFloat &Arg);
+  friend LLVM_SUPPORT_ABI hash_code hash_value(const IEEEFloat &Arg);
 
   /// Converts this value into a decimal string.
   ///
   /// \param FormatPrecision The maximum number of digits of
   ///   precision to output.  If there are fewer digits available,
   ///   zero padding will not be used unless the value is
   ///   integral and small enough to be expressed in
   ///   FormatPrecision digits.  0 means to use the natural
public:
   ///
   /// Because the radix of APFloat is 2, this is equivalent to floor(log2(x)).
   /// For special APFloat values, this returns special error codes:
   ///
   ///   NaN -> \c IEK_NaN
   ///   0   -> \c IEK_Zero
   ///   Inf -> \c IEK_Inf
   ///
-  friend int ilogb(const IEEEFloat &Arg);
+  friend LLVM_SUPPORT_ABI int ilogb(const IEEEFloat &Arg);
 
   /// Returns: X * 2^Exp for integral exponents.
-  friend IEEEFloat scalbn(IEEEFloat X, int Exp, roundingMode);
+  friend LLVM_SUPPORT_ABI IEEEFloat scalbn(IEEEFloat X, int Exp, roundingMode);
 
-  friend IEEEFloat frexp(const IEEEFloat &X, int &Exp, roundingMode);
+  friend LLVM_SUPPORT_ABI IEEEFloat frexp(const IEEEFloat &X, int &Exp, roundingMode);
 
   /// \name Special value setters.
   /// @{
 
   void makeLargest(bool Neg = false);
   void makeSmallest(bool Neg = false);
   void makeNaN(bool SNaN = false, bool Neg = false,
                const APInt *fill = nullptr);
private:
   /// Only 2 bits are required, but VisualStudio incorrectly sign extends it.
   /// Using the extra bit keeps it from failing under VisualStudio.
   fltCategory category : 3;
 
   /// Sign bit of the number.
   unsigned int sign : 1;
 };
 
-hash_code hash_value(const IEEEFloat &Arg);
-int ilogb(const IEEEFloat &Arg);
-IEEEFloat scalbn(IEEEFloat X, int Exp, IEEEFloat::roundingMode);
-IEEEFloat frexp(const IEEEFloat &Val, int &Exp, IEEEFloat::roundingMode RM);
+LLVM_SUPPORT_ABI hash_code hash_value(const IEEEFloat &Arg);
+LLVM_SUPPORT_ABI int ilogb(const IEEEFloat &Arg);
+LLVM_SUPPORT_ABI IEEEFloat scalbn(IEEEFloat X, int Exp, IEEEFloat::roundingMode);
+LLVM_SUPPORT_ABI IEEEFloat frexp(const IEEEFloat &Val, int &Exp, IEEEFloat::roundingMode RM);
 
 // This mode implements more precise float in terms of two APFloats.
 // The interface and layout is designed for arbitrary underlying semantics,
 // though currently only PPCDoubleDouble semantics are supported, whose
 // corresponding underlying semantics are IEEEdouble.
-class DoubleAPFloat final : public APFloatBase {
+class LLVM_SUPPORT_ABI DoubleAPFloat final : public APFloatBase {
   // Note: this must be the first data member.
   const fltSemantics *Semantics;
   std::unique_ptr<APFloat[]> Floats;
 
   opStatus addImpl(const APFloat &a, const APFloat &aa, const APFloat &c,
                    const APFloat &cc, roundingMode RM);
 
   opStatus addWithSpecial(const DoubleAPFloat &LHS, const DoubleAPFloat &RHS,
public:
   bool isLargest() const;
   bool isInteger() const;
 
   void toString(SmallVectorImpl<char> &Str, unsigned FormatPrecision,
                 unsigned FormatMaxPadding, bool TruncateZero = true) const;
 
   bool getExactInverse(APFloat *inv) const;
 
-  friend DoubleAPFloat scalbn(const DoubleAPFloat &X, int Exp, roundingMode);
-  friend DoubleAPFloat frexp(const DoubleAPFloat &X, int &Exp, roundingMode);
+  friend LLVM_SUPPORT_ABI DoubleAPFloat scalbn(const DoubleAPFloat &X, int Exp, roundingMode);
+  friend LLVM_SUPPORT_ABI DoubleAPFloat frexp(const DoubleAPFloat &X, int &Exp, roundingMode);
   friend hash_code hash_value(const DoubleAPFloat &Arg);
 };
 
 hash_code hash_value(const DoubleAPFloat &Arg);
 
 } // End detail namespace
 
 // This is a interface class that is currently forwarding functionalities from
 // detail::IEEEFloat.
-class APFloat : public APFloatBase {
+class LLVM_SUPPORT_ABI APFloat : public APFloatBase {
   typedef detail::IEEEFloat IEEEFloat;
   typedef detail::DoubleAPFloat DoubleAPFloat;
 
   static_assert(std::is_standard_layout<IEEEFloat>::value, "");
 
   union Storage {
     const fltSemantics *semantics;
     IEEEFloat IEEE;
public:
 
   void print(raw_ostream &) const;
   void dump() const;
 
   bool getExactInverse(APFloat *inv) const {
     APFLOAT_DISPATCH_ON_SEMANTICS(getExactInverse(inv));
   }
 
-  friend hash_code hash_value(const APFloat &Arg);
+  friend LLVM_SUPPORT_ABI hash_code hash_value(const APFloat &Arg);
   friend int ilogb(const APFloat &Arg) { return ilogb(Arg.getIEEE()); }
   friend APFloat scalbn(APFloat X, int Exp, roundingMode RM);
   friend APFloat frexp(const APFloat &X, int &Exp, roundingMode RM);
   friend IEEEFloat;
   friend DoubleAPFloat;
 };
 
 /// See friend declarations above.
 ///
 /// These additional declarations are required in order to compile LLVM with IBM
 /// xlC compiler.
-hash_code hash_value(const APFloat &Arg);
+LLVM_SUPPORT_ABI hash_code hash_value(const APFloat &Arg);
+
 inline APFloat scalbn(APFloat X, int Exp, APFloat::roundingMode RM) {
   if (APFloat::usesLayout<detail::IEEEFloat>(X.getSemantics()))
     return APFloat(scalbn(X.U.IEEE, Exp, RM), X.getSemantics());
   if (APFloat::usesLayout<detail::DoubleAPFloat>(X.getSemantics()))
     return APFloat(scalbn(X.U.Double, Exp, RM), X.getSemantics());
   llvm_unreachable("Unexpected semantics");
 }
 

llvm/include/llvm/ADT/APInt.h

 /// integral constant values and operations on them.
 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_APINT_H
 #define LLVM_ADT_APINT_H
 
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MathExtras.h"
+
 #include <cassert>
 #include <climits>
 #include <cstring>
 #include <utility>
 
 namespace llvm {
 class FoldingSetNodeID;
 class StringRef;

 ///     where it makes a difference, there are both signed and unsigned variants
 ///     of the operation. For example, sdiv and udiv. However, because the bit
 ///     widths must be the same, operations such as Mul and Add produce the same
 ///     results regardless of whether the values are interpreted as signed or
 ///     not.
 ///   * In general, the class tries to follow the style of computation that LLVM
 ///     uses in its IR. This simplifies its use for LLVM.
 ///
-class LLVM_NODISCARD APInt {
+class LLVM_NODISCARD LLVM_SUPPORT_ABI APInt {
 public:
   typedef uint64_t WordType;
 
   /// This enum is used to hold the constants we needed for APInt.
   enum : unsigned {
     /// Byte size of a word.
     APINT_WORD_SIZE = sizeof(WordType),
     /// Bits in a word.
static bool isSameValue(const APInt &I1, const APInt &I2) {
 
     if (I1.getBitWidth() > I2.getBitWidth())
       return I1 == I2.zext(I1.getBitWidth());
 
     return I1.zext(I2.getBitWidth()) == I2;
   }
 
   /// Overload to compute a hash_code for an APInt value.
-  friend hash_code hash_value(const APInt &Arg);
+  friend LLVM_SUPPORT_ABI hash_code hash_value(const APInt &Arg);
 
   /// This function returns a pointer to the internal storage of the APInt.
   /// This is useful for writing out the APInt in binary form without any
   /// conversions.
   const uint64_t *getRawData() const {
     if (isSingleWord())
       return &U.VAL;
     return &U.pVal[0];

 
   /// debug method
   void dump() const;
 
   /// @}
 };
 
 /// Magic data for optimising signed division by a constant.
 struct APInt::ms {
   APInt m;    ///< magic number
   unsigned s; ///< shift amount
 };
 
 /// Magic data for optimising unsigned division by a constant.
 struct APInt::mu {
   APInt m;    ///< magic number
   bool a;     ///< add indicator
   unsigned s; ///< shift amount
 };
 
 inline bool operator==(uint64_t V1, const APInt &V2) { return V2 == V1; }
 
 inline bool operator!=(uint64_t V1, const APInt &V2) { return V2 != V1; }
 
 /// Unary bitwise complement operator.
 ///
 /// \returns an APInt that is the bitwise complement of \p v.
 inline APInt operator~(APInt v) {
   v.flipAllBits();
   return v;
 }
 
 inline APInt operator&(APInt a, const APInt &b) {
   a &= b;
   return a;
 }
 
 inline APInt operator&(const APInt &a, APInt &&b) {
   b &= a;
   return std::move(b);
 }
 
 inline APInt operator&(APInt a, uint64_t RHS) {
   a &= RHS;
   return a;
 }
 
 inline APInt operator&(uint64_t LHS, APInt b) {
   b &= LHS;
   return b;
 }
 
 inline APInt operator|(APInt a, const APInt &b) {
   a |= b;
   return a;
 }
 
 inline APInt operator|(const APInt &a, APInt &&b) {
   b |= a;
   return std::move(b);
 }
 
 inline APInt operator|(APInt a, uint64_t RHS) {
   a |= RHS;
   return a;
 }

 }
 
 /// Compute GCD of two unsigned APInt values.
 ///
 /// This function returns the greatest common divisor of the two APInt values
 /// using Stein's algorithm.
 ///
 /// \returns the greatest common divisor of A and B.
-APInt GreatestCommonDivisor(APInt A, APInt B);
+LLVM_SUPPORT_ABI APInt GreatestCommonDivisor(APInt A, APInt B);
 
 /// Converts the given APInt to a double value.
 ///
 /// Treats the APInt as an unsigned value for conversion purposes.
 inline double RoundAPIntToDouble(const APInt &APIVal) {
   return APIVal.roundToDouble();
 }
 

 /// Treats the APInt as a signed value for conversion purposes.
 inline float RoundSignedAPIntToFloat(const APInt &APIVal) {
   return float(APIVal.signedRoundToDouble());
 }
 
 /// Converts the given double value into a APInt.
 ///
 /// This function convert a double value to an APInt value.
-APInt RoundDoubleToAPInt(double Double, unsigned width);
+LLVM_SUPPORT_ABI APInt RoundDoubleToAPInt(double Double, unsigned width);
 
 /// Converts a float value into a APInt.
 ///
 /// Converts a float value into an APInt value.
 inline APInt RoundFloatToAPInt(float Float, unsigned width) {
   return RoundDoubleToAPInt(double(Float), width);
 }
 
 /// Return A unsign-divided by B, rounded by the given rounding mode.
-APInt RoundingUDiv(const APInt &A, const APInt &B, APInt::Rounding RM);
+LLVM_SUPPORT_ABI APInt RoundingUDiv(const APInt &A, const APInt &B, APInt::Rounding RM);
 
 /// Return A sign-divided by B, rounded by the given rounding mode.
-APInt RoundingSDiv(const APInt &A, const APInt &B, APInt::Rounding RM);
+LLVM_SUPPORT_ABI APInt RoundingSDiv(const APInt &A, const APInt &B, APInt::Rounding RM);
 
 /// Let q(n) = An^2 + Bn + C, and BW = bit width of the value range
 /// (e.g. 32 for i32).
 /// This function finds the smallest number n, such that
 /// (a) n >= 0 and q(n) = 0, or
 /// (b) n >= 1 and q(n-1) and q(n), when evaluated in the set of all
 ///     integers, belong to two different intervals [Rk, Rk+R),
 ///     where R = 2^BW, and k is an integer.

 /// virtue of *signed* overflow. This function will *not* find such an n,
 /// however it may find a value of n satisfying the inequalities due to
 /// an *unsigned* overflow (if the values are treated as unsigned).
 /// To find a solution for a signed overflow, treat it as a problem of
 /// finding an unsigned overflow with a range with of BW-1.
 ///
 /// The returned value may have a different bit width from the input
 /// coefficients.
-Optional<APInt> SolveQuadraticEquationWrap(APInt A, APInt B, APInt C,
+LLVM_SUPPORT_ABI Optional<APInt> SolveQuadraticEquationWrap(APInt A, APInt B, APInt C,
                                            unsigned RangeWidth);
 
 /// Compare two values, and if they are different, return the position of the
 /// most significant bit that is different in the values.
-Optional<unsigned> GetMostSignificantDifferentBit(const APInt &A,
+LLVM_SUPPORT_ABI Optional<unsigned> GetMostSignificantDifferentBit(const APInt &A,
                                                   const APInt &B);
 
 } // End of APIntOps namespace
 
 // See friend declaration above. This additional declaration is required in
 // order to compile LLVM with IBM xlC compiler.
-hash_code hash_value(const APInt &Arg);
+LLVM_SUPPORT_ABI hash_code hash_value(const APInt &Arg);
 
 /// StoreIntToMemory - Fills the StoreBytes bytes of memory starting from Dst
 /// with the integer held in IntVal.
-void StoreIntToMemory(const APInt &IntVal, uint8_t *Dst, unsigned StoreBytes);
+LLVM_SUPPORT_ABI void StoreIntToMemory(const APInt &IntVal, uint8_t *Dst, unsigned StoreBytes);
 
 /// LoadIntFromMemory - Loads the integer stored in the LoadBytes bytes starting
 /// from Src into IntVal, which is assumed to be wide enough and to hold zero.
-void LoadIntFromMemory(APInt &IntVal, const uint8_t *Src, unsigned LoadBytes);
+LLVM_SUPPORT_ABI void LoadIntFromMemory(APInt &IntVal, const uint8_t *Src, unsigned LoadBytes);
 
 /// Provide DenseMapInfo for APInt.
-template <> struct DenseMapInfo<APInt> {
+template <> struct LLVM_SUPPORT_ABI DenseMapInfo<APInt> {
   static inline APInt getEmptyKey() {
     APInt V(nullptr, 0);
     V.U.VAL = 0;
     return V;
   }
 
   static inline APInt getTombstoneKey() {
     APInt V(nullptr, 0);

llvm/include/llvm/ADT/APSInt.h

 // represents an arbitrary sized integer that knows its signedness.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_APSINT_H
 #define LLVM_ADT_APSINT_H
 
 #include "llvm/ADT/APInt.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 /// An arbitrary precision integer that knows its signedness.
-class LLVM_NODISCARD APSInt : public APInt {
+class LLVM_NODISCARD LLVM_SUPPORT_ABI APSInt : public APInt {
   bool IsUnsigned;
 
 public:
   /// Default constructor that creates an uninitialized APInt.
   explicit APSInt() : IsUnsigned(false) {}
 
   /// Create an APSInt with the specified width, default to unsigned.
   explicit APSInt(uint32_t BitWidth, bool isUnsigned = true)

llvm/include/llvm/ADT/DAGDeltaAlgorithm.h

 //===- DAGDeltaAlgorithm.h - A DAG Minimization Algorithm ------*- C++ -*--===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_DAGDELTAALGORITHM_H
 #define LLVM_ADT_DAGDELTAALGORITHM_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <set>
 #include <utility>
 #include <vector>
 
 namespace llvm {
 
 /// DAGDeltaAlgorithm - Implements a "delta debugging" algorithm for minimizing
 /// directed acyclic graphs using a predicate function.

 ///   P(S) => P(S union pred(S))
 ///
 /// The minimization algorithm uses this dependency information to attempt to
 /// eagerly prune large subsets of changes. As with \see DeltaAlgorithm, the DAG
 /// is not required to satisfy this property, but the algorithm will run
 /// substantially fewer tests with appropriate dependencies. \see DeltaAlgorithm
 /// for more information on the properties which the predicate function itself
 /// should satisfy.
-class DAGDeltaAlgorithm {
+class LLVM_SUPPORT_ABI DAGDeltaAlgorithm {
   virtual void anchor();
 
 public:
   using change_ty = unsigned;
   using edge_ty = std::pair<change_ty, change_ty>;
 
   // FIXME: Use a decent data structure.
   using changeset_ty = std::set<change_ty>;

llvm/include/llvm/ADT/DeltaAlgorithm.h

 //===- DeltaAlgorithm.h - A Set Minimization Algorithm ---------*- C++ -*--===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_DELTAALGORITHM_H
 #define LLVM_ADT_DELTAALGORITHM_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <set>
 #include <vector>
 
 namespace llvm {
 
 /// DeltaAlgorithm - Implements the delta debugging algorithm (A. Zeller '99)
 /// for minimizing arbitrary sets using a predicate function.
 ///
 /// The result of the algorithm is a subset of the input change set which is
 /// guaranteed to satisfy the predicate, assuming that the input set did. For
 /// well formed predicates, the result set is guaranteed to be such that
 /// removing any single element would falsify the predicate.
 ///
 /// For best results the predicate function *should* (but need not) satisfy
 /// certain properties, in particular:
 ///  (1) The predicate should return false on an empty set and true on the full
 ///  set.
 ///  (2) If the predicate returns true for a set of changes, it should return
 ///  true for all supersets of that set.
 ///
 /// It is not an error to provide a predicate that does not satisfy these
 /// requirements, and the algorithm will generally produce reasonable
 /// results. However, it may run substantially more tests than with a good
 /// predicate.
-class DeltaAlgorithm {
+class LLVM_SUPPORT_ABI DeltaAlgorithm {
 public:
   using change_ty = unsigned;
   // FIXME: Use a decent data structure.
   using changeset_ty = std::set<change_ty>;
   using changesetlist_ty = std::vector<changeset_ty>;
 
 private:
   /// Cache of failed test results. Successful test results are never cached

llvm/include/llvm/ADT/EpochTracker.h

 // with asserts enabled.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_EPOCHTRACKER_H
 #define LLVM_ADT_EPOCHTRACKER_H
 
 #include "llvm/Config/abi-breaking.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 #include <cstdint>
 
 namespace llvm {
 
 #if LLVM_ENABLE_ABI_BREAKING_CHECKS
 
 /// A base class for data structure classes wishing to make iterators
 /// ("handles") pointing into themselves fail-fast.  When building without
 /// asserts, this class is empty and does nothing.
 ///
 /// DebugEpochBase does not by itself track handles pointing into itself.  The
 /// expectation is that routines touching the handles will poll on
 /// isHandleInSync at appropriate points to assert that the handle they're using
 /// is still valid.
 ///
-class DebugEpochBase {
+class LLVM_SUPPORT_ABI DebugEpochBase {
   uint64_t Epoch;
 
 public:
   DebugEpochBase() : Epoch(0) {}
 
   /// Calling incrementEpoch invalidates all handles pointing into the
   /// calling instance.
   void incrementEpoch() { ++Epoch; }
public:
     /// this handle points into.  Can be used to check if two iterators point
     /// into the same data structure.
     const void *getEpochAddress() const { return EpochAddress; }
   };
 };
 
 #else
 
-class DebugEpochBase {
+class LLVM_SUPPORT_ABI DebugEpochBase {
 public:
   void incrementEpoch() {}
 
   class HandleBase {
   public:
     HandleBase() = default;
     explicit HandleBase(const DebugEpochBase *) {}
     bool isHandleInSync() const { return true; }

llvm/include/llvm/ADT/FoldingSet.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_FOLDINGSET_H
 #define LLVM_ADT_FOLDINGSET_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/iterator.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <cstddef>
 #include <cstdint>
 #include <utility>
 
 namespace llvm {
 
 /// This folding set used for two purposes:

 
 //===----------------------------------------------------------------------===//
 /// FoldingSetBase - Implements the folding set functionality.  The main
 /// structure is an array of buckets.  Each bucket is indexed by the hash of
 /// the nodes it contains.  The bucket itself points to the nodes contained
 /// in the bucket via a singly linked list.  The last node in the list points
 /// back to the bucket to facilitate node removal.
 ///
-class FoldingSetBase {
+class LLVM_SUPPORT_ABI FoldingSetBase {
 protected:
   /// Buckets - Array of bucket chains.
   void **Buckets;
 
   /// NumBuckets - Length of the Buckets array.  Always a power of 2.
   unsigned NumBuckets;
 
   /// NumNodes - Number of nodes in the folding set. Growth occurs when NumNodes
template<typename T, typename Ctx> struct ContextualFoldingSetTrait
   : public DefaultContextualFoldingSetTrait<T, Ctx> {};
 
 //===--------------------------------------------------------------------===//
 /// FoldingSetNodeIDRef - This class describes a reference to an interned
 /// FoldingSetNodeID, which can be a useful to store node id data rather
 /// than using plain FoldingSetNodeIDs, since the 32-element SmallVector
 /// is often much larger than necessary, and the possibility of heap
 /// allocation means it requires a non-trivial destructor call.
-class FoldingSetNodeIDRef {
+class LLVM_SUPPORT_ABI FoldingSetNodeIDRef {
   const unsigned *Data = nullptr;
   size_t Size = 0;
 
 public:
   FoldingSetNodeIDRef() = default;
   FoldingSetNodeIDRef(const unsigned *D, size_t S) : Data(D), Size(S) {}
 
   /// ComputeHash - Compute a strong hash value for this FoldingSetNodeIDRef,
public:
   const unsigned *getData() const { return Data; }
   size_t getSize() const { return Size; }
 };
 
 //===--------------------------------------------------------------------===//
 /// FoldingSetNodeID - This class is used to gather all the unique data bits of
 /// a node.  When all the bits are gathered this class is used to produce a
 /// hash value for the node.
-class FoldingSetNodeID {
+class LLVM_SUPPORT_ABI FoldingSetNodeID {
   /// Bits - Vector of all the data bits that make the node unique.
   /// Use a SmallVector to avoid a heap allocation in the common case.
   SmallVector<unsigned, 32> Bits;
 
 public:
   FoldingSetNodeID() = default;
 
   FoldingSetNodeID(FoldingSetNodeIDRef Ref)
public:
 
   /// empty - Returns true if there are no nodes in the folding set.
   bool empty() const { return Set.empty(); }
 };
 
 //===----------------------------------------------------------------------===//
 /// FoldingSetIteratorImpl - This is the common iterator support shared by all
 /// folding sets, which knows how to walk the folding set hash table.
-class FoldingSetIteratorImpl {
+class LLVM_SUPPORT_ABI FoldingSetIteratorImpl {
 protected:
   FoldingSetNode *NodePtr;
 
   FoldingSetIteratorImpl(void **Bucket);
 
   void advance();
 
 public:

llvm/include/llvm/ADT/Hashing.h

 };
 
 
 /// A global, fixed seed-override variable.
 ///
 /// This variable can be set using the \see llvm::set_fixed_execution_seed
 /// function. See that function for details. Do not, under any circumstances,
 /// set or read this variable.
-extern uint64_t fixed_seed_override;
+extern LLVM_SUPPORT_ABI uint64_t fixed_seed_override;
 
 inline uint64_t get_execution_seed() {
   // FIXME: This needs to be a per-execution seed. This is just a placeholder
   // implementation. Switching to a per-execution seed is likely to flush out
   // instability bugs and so will happen as its own commit.
   //
   // However, if there is a fixed seed override set the first time this is
   // called, return that instead of the per-execution seed.

llvm/include/llvm/ADT/IntEqClasses.h

 // further changes.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_INTEQCLASSES_H
 #define LLVM_ADT_INTEQCLASSES_H
 
 #include "llvm/ADT/SmallVector.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
-class IntEqClasses {
+class LLVM_SUPPORT_ABI IntEqClasses {
   /// EC - When uncompressed, map each integer to a smaller member of its
   /// equivalence class. The class leader is the smallest member and maps to
   /// itself.
   ///
   /// When compressed, EC[i] is the equivalence class of i.
   SmallVector<unsigned, 8> EC;
 
   /// NumClasses - The number of equivalence classes when compressed, or 0 when

llvm/include/llvm/ADT/IntervalMap.h

 #ifndef LLVM_ADT_INTERVALMAP_H
 #define LLVM_ADT_INTERVALMAP_H
 
 #include "llvm/ADT/PointerIntPair.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/bit.h"
 #include "llvm/Support/AlignOf.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/RecyclingAllocator.h"
+
 #include <algorithm>
 #include <cassert>
 #include <cstdint>
 #include <iterator>
 #include <new>
 #include <utility>
 
 namespace llvm {

 /// @param Nodes    The number of nodes.
 /// @param Elements Total elements in all nodes.
 /// @param Capacity The capacity of each node.
 /// @param CurSize  Array[Nodes] of current node sizes, or NULL.
 /// @param NewSize  Array[Nodes] to receive the new node sizes.
 /// @param Position Insert position.
 /// @param Grow     Reserve space for a new element at Position.
 /// @return         (node, offset) for Position.
+LLVM_SUPPORT_ABI
 IdxPair distribute(unsigned Nodes, unsigned Elements, unsigned Capacity,
                    const unsigned *CurSize, unsigned NewSize[],
                    unsigned Position, bool Grow);
 
 //===----------------------------------------------------------------------===//
 //---                   IntervalMapImpl::NodeSizer                         ---//
 //===----------------------------------------------------------------------===//
 //

 // A NodeRef doesn't know whether it references a leaf node or a branch node.
 // It is the responsibility of the caller to use the correct types.
 //
 // Nodes are never supposed to be empty, and it is invalid to store a node size
 // of 0 in a NodeRef. The valid range of sizes is 1-64.
 //
 //===----------------------------------------------------------------------===//
 
-class NodeRef {
+class LLVM_SUPPORT_ABI NodeRef {
   struct CacheAlignedPointerTraits {
     static inline void *getAsVoidPointer(void *P) { return P; }
     static inline void *getFromVoidPointer(void *P) { return P; }
     static constexpr int NumLowBitsAvailable = Log2CacheLine;
   };
   PointerIntPair<void*, Log2CacheLine, unsigned, CacheAlignedPointerTraits> pip;
 
 public:

 // A Path is used by iterators to represent a position in a B+-tree, and the
 // path to get there from the root.
 //
 // The Path class also contains the tree navigation code that doesn't have to
 // be templatized.
 //
 //===----------------------------------------------------------------------===//
 
-class Path {
+class LLVM_SUPPORT_ABI Path {
   /// Entry - Each step in the path is a node pointer and an offset into that
   /// node.
   struct Entry {
     void *node;
     unsigned size;
     unsigned offset;
 
     Entry(void *Node, unsigned Size, unsigned Offset)

llvm/include/llvm/ADT/Optional.h

 
 #ifndef LLVM_ADT_OPTIONAL_H
 #define LLVM_ADT_OPTIONAL_H
 
 #include "llvm/ADT/Hashing.h"
 #include "llvm/ADT/None.h"
 #include "llvm/ADT/STLForwardCompat.h"
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/type_traits.h"
+
 #include <cassert>
 #include <memory>
 #include <new>
 #include <utility>
 
 namespace llvm {
 
 class raw_ostream;
constexpr bool operator>=(const Optional<T> &X, const T &Y) {
   return !(X < Y);
 }
 
 template <typename T>
 constexpr bool operator>=(const T &X, const Optional<T> &Y) {
   return !(X < Y);
 }
 
-raw_ostream &operator<<(raw_ostream &OS, NoneType);
+LLVM_SUPPORT_ABI raw_ostream &operator<<(raw_ostream &OS, NoneType);
 
 template <typename T, typename = decltype(std::declval<raw_ostream &>()
                                           << std::declval<const T &>())>
 raw_ostream &operator<<(raw_ostream &OS, const Optional<T> &O) {
   if (O)
     OS << *O;
   else
     OS << None;
   return OS;
 }
 
 } // end namespace llvm
 
 #endif // LLVM_ADT_OPTIONAL_H

llvm/include/llvm/ADT/SmallPtrSet.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_SMALLPTRSET_H
 #define LLVM_ADT_SMALLPTRSET_H
 
 #include "llvm/ADT/EpochTracker.h"
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/ReverseIteration.h"
 #include "llvm/Support/type_traits.h"
+
 #include <cassert>
 #include <cstddef>
 #include <cstdlib>
 #include <cstring>
 #include <initializer_list>
 #include <iterator>
 #include <utility>
 

 /// light-weight and cache-efficient scanning is used.
 ///
 /// Large sets use a classic exponentially-probed hash table.  Empty buckets are
 /// represented with an illegal pointer value (-1) to allow null pointers to be
 /// inserted.  Tombstones are represented with another illegal pointer value
 /// (-2), to allow deletion.  The hash table is resized when the table is 3/4 or
 /// more.  When this happens, the table is doubled in size.
 ///
-class SmallPtrSetImplBase : public DebugEpochBase {
+class LLVM_SUPPORT_ABI SmallPtrSetImplBase : public DebugEpochBase {
   friend class SmallPtrSetIteratorImpl;
 
 protected:
   /// SmallArray - Points to a fixed size set of buckets, used in 'small mode'.
   const void **SmallArray;
   /// CurArray - This is the current set of buckets.  If equal to SmallArray,
   /// then the set is in 'small mode'.
   const void **CurArray;

llvm/include/llvm/ADT/SmallVector.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_SMALLVECTOR_H
 #define LLVM_ADT_SMALLVECTOR_H
 
 #include "llvm/ADT/iterator_range.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MemAlloc.h"
 #include "llvm/Support/type_traits.h"
+
 #include <algorithm>
 #include <cassert>
 #include <cstddef>
 #include <cstdlib>
 #include <cstring>
 #include <functional>
 #include <initializer_list>
 #include <iterator>
protected:
 
   SmallVectorBase() = delete;
   SmallVectorBase(void *FirstEl, size_t TotalCapacity)
       : BeginX(FirstEl), Capacity(TotalCapacity) {}
 
   /// This is a helper for \a grow() that's out of line to reduce code
   /// duplication.  This function will report a fatal error if it can't grow at
   /// least to \p MinSize.
-  void *mallocForGrow(size_t MinSize, size_t TSize, size_t &NewCapacity);
+  LLVM_SUPPORT_ABI void *mallocForGrow(size_t MinSize, size_t TSize, size_t &NewCapacity);
 
   /// This is an implementation of the grow() method which only works
   /// on POD-like data types and is out of line to reduce code duplication.
   /// This function will report a fatal error if it cannot increase capacity.
-  void grow_pod(void *FirstEl, size_t MinSize, size_t TSize);
+  LLVM_SUPPORT_ABI void grow_pod(void *FirstEl, size_t MinSize, size_t TSize);
 
 public:
   size_t size() const { return Size; }
   size_t capacity() const { return Capacity; }
 
   LLVM_NODISCARD bool empty() const { return !Size; }
 
   /// Set the array size to \p N, which the current array must have enough
   /// capacity for.
   ///
   /// This does not construct or destroy any elements in the vector.
   ///
   /// Clients can use this in conjunction with capacity() to write past the end
   /// of the buffer when they know that more elements are available, and only
   /// update the size later. This avoids the cost of value initializing elements
   /// which will only be overwritten.
   void set_size(size_t N) {
     assert(N <= capacity());
     Size = N;
   }
 };
 
+extern template LLVM_SUPPORT_ABI SmallVectorBase<uint32_t>;
+#if SIZE_MAX > UINT32_MAX
+extern template LLVM_SUPPORT_ABI SmallVectorBase<uint64_t>;
+#endif
+
 template <class T>
 using SmallVectorSizeType =
     typename std::conditional<sizeof(T) < 4 && sizeof(void *) >= 8, uint64_t,
                               uint32_t>::type;
 
 /// Figure out the offset of the first element.
 template <class T, typename = void> struct SmallVectorAlignmentAndSize {
   alignas(SmallVectorBase<SmallVectorSizeType<T>>) char Base[sizeof(

llvm/include/llvm/ADT/Statistic.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_STATISTIC_H
 #define LLVM_ADT_STATISTIC_H
 
 #include "llvm/Config/llvm-config.h"
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <atomic>
 #include <memory>
 #include <vector>
 
 // Determine whether statistics should be enabled. We must do it here rather
 // than in CMake because multi-config generators cannot determine this at
 // configure time.
 #if !defined(NDEBUG) || LLVM_FORCE_ENABLE_STATS
 #define LLVM_ENABLE_STATS 1
 #else
 #define LLVM_ENABLE_STATS 0
 #endif
 
 namespace llvm {
 
 class raw_ostream;
 class raw_fd_ostream;
 class StringRef;
 
-class TrackingStatistic {
+class LLVM_SUPPORT_ABI TrackingStatistic {
 public:
   const char *const DebugType;
   const char *const Name;
   const char *const Desc;
 
   std::atomic<unsigned> Value;
   std::atomic<bool> Initialized;
 
#define STATISTIC(VARNAME, DESC)                                               \
   static llvm::Statistic VARNAME = {DEBUG_TYPE, #VARNAME, DESC}
 
 // ALWAYS_ENABLED_STATISTIC - A macro to define a statistic like STATISTIC but
 // it is enabled even if LLVM_ENABLE_STATS is off.
 #define ALWAYS_ENABLED_STATISTIC(VARNAME, DESC)                                \
   static llvm::TrackingStatistic VARNAME = {DEBUG_TYPE, #VARNAME, DESC}
 
 /// Enable the collection and printing of statistics.
-void EnableStatistics(bool DoPrintOnExit = true);
+LLVM_SUPPORT_ABI void EnableStatistics(bool DoPrintOnExit = true);
 
 /// Check if statistics are enabled.
-bool AreStatisticsEnabled();
+LLVM_SUPPORT_ABI bool AreStatisticsEnabled();
 
 /// Return a file stream to print our output on.
-std::unique_ptr<raw_fd_ostream> CreateInfoOutputFile();
+LLVM_SUPPORT_ABI std::unique_ptr<raw_fd_ostream> CreateInfoOutputFile();
 
 /// Print statistics to the file returned by CreateInfoOutputFile().
-void PrintStatistics();
+LLVM_SUPPORT_ABI void PrintStatistics();
 
 /// Print statistics to the given output stream.
-void PrintStatistics(raw_ostream &OS);
+LLVM_SUPPORT_ABI void PrintStatistics(raw_ostream &OS);
 
 /// Print statistics in JSON format. This does include all global timers (\see
 /// Timer, TimerGroup). Note that the timers are cleared after printing and will
 /// not be printed in human readable form or in a second call of
 /// PrintStatisticsJSON().
-void PrintStatisticsJSON(raw_ostream &OS);
+LLVM_SUPPORT_ABI void PrintStatisticsJSON(raw_ostream &OS);
 
 /// Get the statistics. This can be used to look up the value of
 /// statistics without needing to parse JSON.
 ///
 /// This function does not prevent statistics being updated by other threads
 /// during it's execution. It will return the value at the point that it is
 /// read. However, it will prevent new statistics from registering until it
 /// completes.
-const std::vector<std::pair<StringRef, unsigned>> GetStatistics();
+LLVM_SUPPORT_ABI const std::vector<std::pair<StringRef, unsigned>> GetStatistics();
 
 /// Reset the statistics. This can be used to zero and de-register the
 /// statistics in order to measure a compilation.
 ///
 /// When this function begins to call destructors prior to returning, all
 /// statistics will be zero and unregistered. However, that might not remain the
 /// case by the time this function finishes returning. Whether update from other
 /// threads are lost or merely deferred until during the function return is
 /// timing sensitive.
 ///
 /// Callers who intend to use this to measure statistics for a single
 /// compilation should ensure that no compilations are in progress at the point
 /// this function is called and that only one compilation executes until calling
 /// GetStatistics().
-void ResetStatistics();
+LLVM_SUPPORT_ABI void ResetStatistics();
 
 } // end namespace llvm
 
 #endif // LLVM_ADT_STATISTIC_H

llvm/include/llvm/ADT/StringExtras.h

 #ifndef LLVM_ADT_STRINGEXTRAS_H
 #define LLVM_ADT_STRINGEXTRAS_H
 
 #include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <cstddef>
 #include <cstdint>
 #include <cstdlib>
 #include <cstring>
 #include <iterator>
 #include <string>
 #include <utility>

 
 inline std::string toString(const APSInt &I, unsigned Radix) {
   return toString(I, Radix, I.isSigned());
 }
 
 /// StrInStrNoCase - Portable version of strcasestr.  Locates the first
 /// occurrence of string 's1' in string 's2', ignoring case.  Returns
 /// the offset of s2 in s1 or npos if s2 cannot be found.
-StringRef::size_type StrInStrNoCase(StringRef s1, StringRef s2);
+LLVM_SUPPORT_ABI StringRef::size_type StrInStrNoCase(StringRef s1, StringRef s2);
 
 /// getToken - This function extracts one token from source, ignoring any
 /// leading characters that appear in the Delimiters string, and ending the
 /// token at any of the characters that appear in the Delimiters string.  If
 /// there are no tokens in the source string, an empty string is returned.
 /// The function returns a pair containing the extracted token and the
 /// remaining tail string.
+LLVM_SUPPORT_ABI
 std::pair<StringRef, StringRef> getToken(StringRef Source,
                                          StringRef Delimiters = " \t\n\v\f\r");
 
 /// SplitString - Split up the specified string according to the specified
 /// delimiters, appending the result fragments to the output list.
+LLVM_SUPPORT_ABI
 void SplitString(StringRef Source,
                  SmallVectorImpl<StringRef> &OutFragments,
                  StringRef Delimiters = " \t\n\v\f\r");
 
 /// Returns the English suffix for an ordinal integer (-st, -nd, -rd, -th).
 inline StringRef getOrdinalSuffix(unsigned Val) {
   // It is critically important that we do this perfectly for
   // user-written sequences with over 100 elements.
switch (Val % 10) {
       case 3: return "rd";
       default: return "th";
     }
   }
 }
 
 /// Print each character of the specified string, escaping it if it is not
 /// printable or if it is an escape char.
-void printEscapedString(StringRef Name, raw_ostream &Out);
+LLVM_SUPPORT_ABI void printEscapedString(StringRef Name, raw_ostream &Out);
 
 /// Print each character of the specified string, escaping HTML special
 /// characters.
-void printHTMLEscaped(StringRef String, raw_ostream &Out);
+LLVM_SUPPORT_ABI void printHTMLEscaped(StringRef String, raw_ostream &Out);
 
 /// printLowerCase - Print each character as lowercase if it is uppercase.
-void printLowerCase(StringRef String, raw_ostream &Out);
+LLVM_SUPPORT_ABI void printLowerCase(StringRef String, raw_ostream &Out);
 
 /// Converts a string from camel-case to snake-case by replacing all uppercase
 /// letters with '_' followed by the letter in lowercase, except if the
 /// uppercase letter is the first character of the string.
-std::string convertToSnakeFromCamelCase(StringRef input);
+LLVM_SUPPORT_ABI std::string convertToSnakeFromCamelCase(StringRef input);
 
 /// Converts a string from snake-case to camel-case by replacing all occurrences
 /// of '_' followed by a lowercase letter with the letter in uppercase.
 /// Optionally allow capitalization of the first letter (if it is a lowercase
 /// letter)
-std::string convertToCamelFromSnakeCase(StringRef input,
+LLVM_SUPPORT_ABI std::string convertToCamelFromSnakeCase(StringRef input,
                                         bool capitalizeFirst = false);
 
 namespace detail {
 
 template <typename IteratorT>
 inline std::string join_impl(IteratorT Begin, IteratorT End,
                              StringRef Separator, std::input_iterator_tag) {
   std::string S;

llvm/include/llvm/ADT/StringMap.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_STRINGMAP_H
 #define LLVM_ADT_STRINGMAP_H
 
 #include "llvm/ADT/StringMapEntry.h"
 #include "llvm/Support/AllocatorBase.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/PointerLikeTypeTraits.h"
 #include <initializer_list>
 #include <iterator>
 
 namespace llvm {
 
 template <typename ValueTy> class StringMapConstIterator;
 template <typename ValueTy> class StringMapIterator;
 template <typename ValueTy> class StringMapKeyIterator;
 
 /// StringMapImpl - This is the base class of StringMap that is shared among
 /// all of its instantiations.
-class StringMapImpl {
+class LLVM_SUPPORT_ABI StringMapImpl {
 protected:
   // Array of NumBuckets pointers to entries, null pointers are holes.
   // TheTable[NumBuckets] contains a sentinel value for easy iteration. Followed
   // by an array of the actual hash values as unsigned integers.
   StringMapEntryBase **TheTable = nullptr;
   unsigned NumBuckets = 0;
   unsigned NumItems = 0;
   unsigned NumTombstones = 0;

llvm/include/llvm/ADT/StringRef.h

 //===- StringRef.h - Constant String Reference Wrapper ----------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_STRINGREF_H
 #define LLVM_ADT_STRINGREF_H
 
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/iterator_range.h"
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <algorithm>
 #include <cassert>
 #include <cstddef>
 #include <cstring>
 #include <limits>
 #include <string>
 #if __cplusplus > 201402L
 #include <string_view>
namespace llvm {
 
   class APInt;
   class hash_code;
   template <typename T> class SmallVectorImpl;
   template <typename T> struct DenseMapInfo;
   class StringRef;
 
   /// Helper functions for StringRef::getAsInteger.
+  LLVM_SUPPORT_ABI
   bool getAsUnsignedInteger(StringRef Str, unsigned Radix,
                             unsigned long long &Result);
 
+  LLVM_SUPPORT_ABI
   bool getAsSignedInteger(StringRef Str, unsigned Radix, long long &Result);
 
+  LLVM_SUPPORT_ABI
   bool consumeUnsignedInteger(StringRef &Str, unsigned Radix,
                               unsigned long long &Result);
+
+  LLVM_SUPPORT_ABI
   bool consumeSignedInteger(StringRef &Str, unsigned Radix, long long &Result);
 
   /// StringRef - Represent a constant reference to a string, i.e. a character
   /// array and a length, which need not be null terminated.
   ///
   /// This class does not own the string data, it is expected to be used in
   /// situations where the character data resides in some other buffer, whose
   /// lifetime extends past that of the StringRef. For this reason, it is not in
   /// general safe to store a StringRef.
-  class LLVM_GSL_POINTER StringRef {
+  class LLVM_SUPPORT_ABI LLVM_GSL_POINTER StringRef {
   public:
     static constexpr size_t npos = ~size_t(0);
 
     using iterator = const char *;
     using const_iterator = const char *;
     using size_type = size_t;
 
   private:
#endif
 
   inline std::string &operator+=(std::string &buffer, StringRef string) {
     return buffer.append(string.data(), string.size());
   }
 
   /// @}
 
   /// Compute a hash_code for a StringRef.
-  LLVM_NODISCARD
+  LLVM_NODISCARD LLVM_SUPPORT_ABI
   hash_code hash_value(StringRef S);
 
   // Provide DenseMapInfo for StringRefs.
   template <> struct DenseMapInfo<StringRef> {
     static inline StringRef getEmptyKey() {
       return StringRef(
           reinterpret_cast<const char *>(~static_cast<uintptr_t>(0)), 0);
     }

llvm/include/llvm/ADT/Triple.h

 //===-- llvm/ADT/Triple.h - Target triple helper class ----------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_TRIPLE_H
 #define LLVM_ADT_TRIPLE_H
 
 #include "llvm/ADT/Twine.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 // Some system headers or GCC predefined macros conflict with identifiers in
 // this file.  Undefine them here.
 #undef NetBSD
 #undef mips
 #undef sparc
 
 namespace llvm {

 ///
 /// At its core the Triple class is designed to be a wrapper for a triple
 /// string; the constructor does not change or normalize the triple string.
 /// Clients that need to handle the non-canonical triples that users often
 /// specify should use the normalize method.
 ///
 /// See autoconf/config.guess for a glimpse into what configuration names
 /// look like in practice.
-class Triple {
+class LLVM_SUPPORT_ABI Triple {
 public:
   enum ArchType {
     UnknownArch,
 
     arm,            // ARM (little endian): arm, armv.*, xscale
     armeb,          // ARM (big endian): armeb
     aarch64,        // AArch64 (little endian): aarch64
     aarch64_be,     // AArch64 (big endian): aarch64_be

llvm/include/llvm/ADT/Twine.h

 //===- Twine.h - Fast Temporary String Concatenation ------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_ADT_TWINE_H
 #define LLVM_ADT_TWINE_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <cstdint>
 #include <string>
 #if __cplusplus > 201402L
 #include <string_view>
 #endif
 
 namespace llvm {
namespace llvm {
   /// situations. For this reason, the concat() method is inlined so that
   /// concatenations of leaf nodes can be optimized into stores directly into a
   /// single stack allocated object.
   ///
   /// In practice, not all compilers can be trusted to optimize concat() fully,
   /// so we provide two additional methods (and accompanying operator+
   /// overloads) to guarantee that particularly important cases (cstring plus
   /// StringRef) codegen as desired.
-  class Twine {
+  class LLVM_SUPPORT_ABI Twine {
     /// NodeKind - Represent the type of an argument.
     enum NodeKind : unsigned char {
       /// An empty string; the result of concatenating anything with it is also
       /// empty.
       NullKind,
 
       /// The empty string.
       EmptyKind,

llvm/include/llvm/Support/AArch64TargetParser.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_AARCH64TARGETPARSER_H
 #define LLVM_SUPPORT_AARCH64TARGETPARSER_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/ARMTargetParser.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <vector>
 
 // FIXME:This should be made into class design,to avoid dupplication.
 namespace llvm {
 
 class Triple;
 
 namespace AArch64 {

 
 const ArchKind ArchKinds[] = {
 #define AARCH64_ARCH(NAME, ID, CPU_ATTR, SUB_ARCH, ARCH_ATTR, ARCH_FPU, ARCH_BASE_EXT) \
     ArchKind::ID,
 #include "AArch64TargetParser.def"
 };
 
 // FIXME: These should be moved to TargetTuple once it exists
-bool getExtensionFeatures(uint64_t Extensions,
+LLVM_SUPPORT_ABI bool getExtensionFeatures(uint64_t Extensions,
                           std::vector<StringRef> &Features);
-bool getArchFeatures(ArchKind AK, std::vector<StringRef> &Features);
+LLVM_SUPPORT_ABI bool getArchFeatures(ArchKind AK, std::vector<StringRef> &Features);
 
 StringRef getArchName(ArchKind AK);
 unsigned getArchAttr(ArchKind AK);
-StringRef getCPUAttr(ArchKind AK);
-StringRef getSubArch(ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getCPUAttr(ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getSubArch(ArchKind AK);
 StringRef getArchExtName(unsigned ArchExtKind);
-StringRef getArchExtFeature(StringRef ArchExt);
+LLVM_SUPPORT_ABI StringRef getArchExtFeature(StringRef ArchExt);
 
 // Information by Name
-unsigned getDefaultFPU(StringRef CPU, ArchKind AK);
-uint64_t getDefaultExtensions(StringRef CPU, ArchKind AK);
-StringRef getDefaultCPU(StringRef Arch);
-ArchKind getCPUArchKind(StringRef CPU);
+LLVM_SUPPORT_ABI unsigned getDefaultFPU(StringRef CPU, ArchKind AK);
+LLVM_SUPPORT_ABI uint64_t getDefaultExtensions(StringRef CPU, ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getDefaultCPU(StringRef Arch);
+LLVM_SUPPORT_ABI ArchKind getCPUArchKind(StringRef CPU);
 
 // Parser
-ArchKind parseArch(StringRef Arch);
-ArchExtKind parseArchExt(StringRef ArchExt);
-ArchKind parseCPUArch(StringRef CPU);
+LLVM_SUPPORT_ABI ArchKind parseArch(StringRef Arch);
+LLVM_SUPPORT_ABI ArchExtKind parseArchExt(StringRef ArchExt);
+LLVM_SUPPORT_ABI ArchKind parseCPUArch(StringRef CPU);
 // Used by target parser tests
-void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values);
+LLVM_SUPPORT_ABI void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values);
 
-bool isX18ReservedByDefault(const Triple &TT);
+LLVM_SUPPORT_ABI bool isX18ReservedByDefault(const Triple &TT);
 
 struct ParsedBranchProtection {
   StringRef Scope;
   StringRef Key;
   bool BranchTargetEnforcement;
 };
 
-bool parseBranchProtection(StringRef Spec, ParsedBranchProtection &PBP,
+LLVM_SUPPORT_ABI bool parseBranchProtection(StringRef Spec, ParsedBranchProtection &PBP,
                            StringRef &Err);
 
 } // namespace AArch64
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/AMDGPUMetadata.h

 ///
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_AMDGPUMETADATA_H
 #define LLVM_SUPPORT_AMDGPUMETADATA_H
 
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cstdint>
 #include <string>
 #include <system_error>
 #include <vector>
 
 namespace llvm {
 namespace AMDGPU {
 
struct Metadata final {
   /// Kernels metadata. Required.
   std::vector<Kernel::Metadata> mKernels = std::vector<Kernel::Metadata>();
 
   /// Default constructor.
   Metadata() = default;
 };
 
 /// Converts \p String to \p HSAMetadata.
-std::error_code fromString(StringRef String, Metadata &HSAMetadata);
+LLVM_SUPPORT_ABI std::error_code fromString(StringRef String, Metadata &HSAMetadata);
 
 /// Converts \p HSAMetadata to \p String.
-std::error_code toString(Metadata HSAMetadata, std::string &String);
+LLVM_SUPPORT_ABI std::error_code toString(Metadata HSAMetadata, std::string &String);
 
 //===----------------------------------------------------------------------===//
 // HSA metadata for v3 code object.
 //===----------------------------------------------------------------------===//
 namespace V3 {
 /// HSA metadata major version.
 constexpr uint32_t VersionMajor = 1;
 /// HSA metadata minor version.

llvm/include/llvm/Support/ARMAttributeParser.h

 
 #include "ARMBuildAttributes.h"
 #include "ELFAttributeParser.h"
 #include "ScopedPrinter.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/Support/DataExtractor.h"
 #include "llvm/Support/Endian.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 class StringRef;
 
-class ARMAttributeParser : public ELFAttributeParser {
+class LLVM_SUPPORT_ABI ARMAttributeParser : public ELFAttributeParser {
   struct DisplayHandler {
     ARMBuildAttrs::AttrType attribute;
     Error (ARMAttributeParser::*routine)(ARMBuildAttrs::AttrType);
   };
   static const DisplayHandler displayRoutines[];
 
   Error handler(uint64_t tag, bool &handled) override;
 

llvm/include/llvm/Support/ARMBuildAttributes.h

 // http://infocenter.arm.com/help/topic/com.arm.doc.ihi0044e/IHI0044E_aaelf.pdf
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ARMBUILDATTRIBUTES_H
 #define LLVM_SUPPORT_ARMBUILDATTRIBUTES_H
 
 #include "llvm/Support/ELFAttributes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 namespace ARMBuildAttrs {
 
-const TagNameMap &getARMAttributeTags();
+LLVM_SUPPORT_ABI const TagNameMap &getARMAttributeTags();
 
 enum SpecialAttr {
   // This is for the .cpu asm attr. It translates into one or more
   // AttrType (below) entries in the .ARM.attributes section in the ELF.
   SEL_CPU
 };
 
 enum AttrType : unsigned {

llvm/include/llvm/Support/ARMTargetParser.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ARMTARGETPARSER_H
 #define LLVM_SUPPORT_ARMTARGETPARSER_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/ARMBuildAttributes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <vector>
 
 namespace llvm {
 
 class Triple;
 
 namespace ARM {
 
#define ARM_ARCH(NAME, ID, CPU_ATTR, SUB_ARCH, ARCH_ATTR, ARCH_FPU,            \
    CPU_ATTR,     sizeof(CPU_ATTR) - 1,                                         \
    SUB_ARCH,     sizeof(SUB_ARCH) - 1,                                         \
    ARCH_FPU,     ARCH_BASE_EXT,                                                \
    ArchKind::ID, ARCH_ATTR},
 #include "llvm/Support/ARMTargetParser.def"
 };
 
 // Information by ID
-StringRef getFPUName(unsigned FPUKind);
-FPUVersion getFPUVersion(unsigned FPUKind);
-NeonSupportLevel getFPUNeonSupportLevel(unsigned FPUKind);
-FPURestriction getFPURestriction(unsigned FPUKind);
+LLVM_SUPPORT_ABI StringRef getFPUName(unsigned FPUKind);
+LLVM_SUPPORT_ABI FPUVersion getFPUVersion(unsigned FPUKind);
+LLVM_SUPPORT_ABI NeonSupportLevel getFPUNeonSupportLevel(unsigned FPUKind);
+LLVM_SUPPORT_ABI FPURestriction getFPURestriction(unsigned FPUKind);
 
 // FIXME: These should be moved to TargetTuple once it exists
-bool getFPUFeatures(unsigned FPUKind, std::vector<StringRef> &Features);
-bool getHWDivFeatures(uint64_t HWDivKind, std::vector<StringRef> &Features);
-bool getExtensionFeatures(uint64_t Extensions,
+LLVM_SUPPORT_ABI bool getFPUFeatures(unsigned FPUKind, std::vector<StringRef> &Features);
+LLVM_SUPPORT_ABI bool getHWDivFeatures(uint64_t HWDivKind, std::vector<StringRef> &Features);
+LLVM_SUPPORT_ABI bool getExtensionFeatures(uint64_t Extensions,
                           std::vector<StringRef> &Features);
 
-StringRef getArchName(ArchKind AK);
-unsigned getArchAttr(ArchKind AK);
-StringRef getCPUAttr(ArchKind AK);
-StringRef getSubArch(ArchKind AK);
-StringRef getArchExtName(uint64_t ArchExtKind);
-StringRef getArchExtFeature(StringRef ArchExt);
-bool appendArchExtFeatures(StringRef CPU, ARM::ArchKind AK, StringRef ArchExt,
+LLVM_SUPPORT_ABI StringRef getArchName(ArchKind AK);
+LLVM_SUPPORT_ABI unsigned getArchAttr(ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getCPUAttr(ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getSubArch(ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getArchExtName(uint64_t ArchExtKind);
+LLVM_SUPPORT_ABI StringRef getArchExtFeature(StringRef ArchExt);
+LLVM_SUPPORT_ABI bool appendArchExtFeatures(StringRef CPU, ARM::ArchKind AK, StringRef ArchExt,
                            std::vector<StringRef> &Features,
                            unsigned &ArgFPUKind);
 
 // Information by Name
-unsigned getDefaultFPU(StringRef CPU, ArchKind AK);
-uint64_t getDefaultExtensions(StringRef CPU, ArchKind AK);
-StringRef getDefaultCPU(StringRef Arch);
+LLVM_SUPPORT_ABI unsigned getDefaultFPU(StringRef CPU, ArchKind AK);
+LLVM_SUPPORT_ABI uint64_t getDefaultExtensions(StringRef CPU, ArchKind AK);
+LLVM_SUPPORT_ABI StringRef getDefaultCPU(StringRef Arch);
 StringRef getCanonicalArchName(StringRef Arch);
 StringRef getFPUSynonym(StringRef FPU);
 StringRef getArchSynonym(StringRef Arch);
 
 // Parser
-uint64_t parseHWDiv(StringRef HWDiv);
-unsigned parseFPU(StringRef FPU);
-ArchKind parseArch(StringRef Arch);
-uint64_t parseArchExt(StringRef ArchExt);
-ArchKind parseCPUArch(StringRef CPU);
-ISAKind parseArchISA(StringRef Arch);
-EndianKind parseArchEndian(StringRef Arch);
-ProfileKind parseArchProfile(StringRef Arch);
-unsigned parseArchVersion(StringRef Arch);
+LLVM_SUPPORT_ABI uint64_t parseHWDiv(StringRef HWDiv);
+LLVM_SUPPORT_ABI unsigned parseFPU(StringRef FPU);
+LLVM_SUPPORT_ABI ArchKind parseArch(StringRef Arch);
+LLVM_SUPPORT_ABI uint64_t parseArchExt(StringRef ArchExt);
+LLVM_SUPPORT_ABI ArchKind parseCPUArch(StringRef CPU);
+LLVM_SUPPORT_ABI ISAKind parseArchISA(StringRef Arch);
+LLVM_SUPPORT_ABI EndianKind parseArchEndian(StringRef Arch);
+LLVM_SUPPORT_ABI ProfileKind parseArchProfile(StringRef Arch);
+LLVM_SUPPORT_ABI unsigned parseArchVersion(StringRef Arch);
 
-void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values);
-StringRef computeDefaultTargetABI(const Triple &TT, StringRef CPU);
+LLVM_SUPPORT_ABI void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values);
+LLVM_SUPPORT_ABI StringRef computeDefaultTargetABI(const Triple &TT, StringRef CPU);
 
 } // namespace ARM
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/ARMWinEH.h

 //===-- llvm/Support/ARMWinEH.h - Windows on ARM EH Constants ---*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ARMWINEH_H
 #define LLVM_SUPPORT_ARMWINEH_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/Support/Endian.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 namespace ARM {
 namespace WinEH {
 enum class RuntimeFunctionFlag {
   RFF_Unpacked,       /// unpacked entry
   RFF_Packed,         /// packed entry
   RFF_PackedFragment, /// packed entry representing a fragment
inline uint16_t StackAdjustment(const RuntimeFunction &RF) {
   uint16_t Adjustment = RF.StackAdjust();
   if (Adjustment >= 0x3f4)
     return (Adjustment & 0x3) ? ((Adjustment & 0x3) << 2) - 1 : 0;
   return Adjustment;
 }
 
 /// SavedRegisterMask - Utility function to calculate the set of saved general
 /// purpose (r0-r15) and VFP (d0-d31) registers.
-std::pair<uint16_t, uint32_t> SavedRegisterMask(const RuntimeFunction &RF);
+LLVM_SUPPORT_ABI std::pair<uint16_t, uint32_t>
+SavedRegisterMask(const RuntimeFunction &RF);
 
 /// RuntimeFunctionARM64 - An entry in the table of procedure data (.pdata)
 ///
 ///  3 3 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0
 ///  1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0
 /// +---------------------------------------------------------------+
 /// |                     Function Start RVA                        |
 /// +-----------------+---+-+-------+-----+---------------------+---+

llvm/include/llvm/Support/Allocator.h

 #define LLVM_SUPPORT_ALLOCATOR_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Support/Alignment.h"
 #include "llvm/Support/AllocatorBase.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/MemAlloc.h"
+
 #include <algorithm>
 #include <cassert>
 #include <cstddef>
 #include <cstdint>
 #include <cstdlib>
 #include <iterator>
 #include <type_traits>
 #include <utility>
 
 namespace llvm {
 
 namespace detail {
 
 // We call out to an external function to actually print the message as the
 // printing code uses Allocator.h in its implementation.
-void printBumpPtrAllocatorStats(unsigned NumSlabs, size_t BytesAllocated,
+LLVM_SUPPORT_ABI void printBumpPtrAllocatorStats(unsigned NumSlabs, size_t BytesAllocated,
                                 size_t TotalMemory);
 
 } // end namespace detail
 
 /// Allocate memory in an ever growing pool, as if by bump-pointer.
 ///
 /// This isn't strictly a bump-pointer allocator as it uses backing slabs of
 /// memory rather than relying on a boundless contiguous heap. However, it has

llvm/include/llvm/Support/BinaryStreamError.h

 //===- BinaryStreamError.h - Error extensions for Binary Streams *- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_BINARYSTREAMERROR_H
 #define LLVM_SUPPORT_BINARYSTREAMERROR_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 #include <string>
 
 namespace llvm {
 enum class stream_error_code {
   unspecified,
   stream_too_short,
   invalid_array_size,
   invalid_offset,
   filesystem_error
 };
 
 /// Base class for errors originating when parsing raw PDB files
-class BinaryStreamError : public ErrorInfo<BinaryStreamError> {
+class LLVM_SUPPORT_ABI BinaryStreamError : public ErrorInfo<BinaryStreamError> {
 public:
   static char ID;
   explicit BinaryStreamError(stream_error_code C);
   explicit BinaryStreamError(StringRef Context);
   BinaryStreamError(stream_error_code C, StringRef Context);
 
   void log(raw_ostream &OS) const override;
   std::error_code convertToErrorCode() const override;

llvm/include/llvm/Support/BinaryStreamReader.h

 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Alignment.h"
 #include "llvm/Support/BinaryStreamArray.h"
 #include "llvm/Support/BinaryStreamRef.h"
 #include "llvm/Support/ConvertUTF.h"
 #include "llvm/Support/Endian.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/type_traits.h"
+
 #include <type_traits>
 
 namespace llvm {
 
 /// Provides read only access to a subclass of `BinaryStream`.  Provides
 /// bounds checking and helpers for writing certain common data types such as
 /// null-terminated strings, integers in various flavors of endianness, etc.
 /// Can be subclassed to provide reading of custom datatypes, although no
 /// are overridable.
-class BinaryStreamReader {
+class LLVM_SUPPORT_ABI BinaryStreamReader {
 public:
   BinaryStreamReader() = default;
   explicit BinaryStreamReader(BinaryStreamRef Ref);
   explicit BinaryStreamReader(BinaryStream &Stream);
   explicit BinaryStreamReader(ArrayRef<uint8_t> Data,
                               llvm::support::endianness Endian);
   explicit BinaryStreamReader(StringRef Data, llvm::support::endianness Endian);
 

llvm/include/llvm/Support/BinaryStreamRef.h

 //===- BinaryStreamRef.h - A copyable reference to a stream -----*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_BINARYSTREAMREF_H
 #define LLVM_SUPPORT_BINARYSTREAMREF_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/Support/BinaryStream.h"
 #include "llvm/Support/BinaryStreamError.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <algorithm>
 #include <cstdint>
 #include <memory>
 
 namespace llvm {
 
 /// Common stuff for mutable and immutable StreamRefs.
 template <class RefType, class StreamType> class BinaryStreamRefBase {

 
 /// BinaryStreamRef is to BinaryStream what ArrayRef is to an Array.  It
 /// provides copy-semantics and read only access to a "window" of the underlying
 /// BinaryStream. Note that BinaryStreamRef is *not* a BinaryStream.  That is to
 /// say, it does not inherit and override the methods of BinaryStream.  In
 /// general, you should not pass around pointers or references to BinaryStreams
 /// and use inheritance to achieve polymorphism.  Instead, you should pass
 /// around BinaryStreamRefs by value and achieve polymorphism that way.
-class BinaryStreamRef
+class LLVM_SUPPORT_ABI BinaryStreamRef
     : public BinaryStreamRefBase<BinaryStreamRef, BinaryStream> {
   friend BinaryStreamRefBase<BinaryStreamRef, BinaryStream>;
   friend class WritableBinaryStreamRef;
   BinaryStreamRef(std::shared_ptr<BinaryStream> Impl, uint32_t ViewOffset,
                   Optional<uint32_t> Length)
       : BinaryStreamRefBase(Impl, ViewOffset, Length) {}
 
 public:
struct BinarySubstreamRef {
   split(uint32_t Off) const {
     return std::make_pair(keep_front(Off), drop_front(Off));
   }
 
   uint32_t size() const { return StreamData.getLength(); }
   bool empty() const { return size() == 0; }
 };
 
-class WritableBinaryStreamRef
+class LLVM_SUPPORT_ABI WritableBinaryStreamRef
     : public BinaryStreamRefBase<WritableBinaryStreamRef,
                                  WritableBinaryStream> {
   friend BinaryStreamRefBase<WritableBinaryStreamRef, WritableBinaryStream>;
   WritableBinaryStreamRef(std::shared_ptr<WritableBinaryStream> Impl,
                           uint32_t ViewOffset, Optional<uint32_t> Length)
       : BinaryStreamRefBase(Impl, ViewOffset, Length) {}
 
   Error checkOffsetForWrite(uint32_t Offset, uint32_t DataSize) const {

llvm/include/llvm/Support/BinaryStreamWriter.h

 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/BinaryStreamArray.h"
 #include "llvm/Support/BinaryStreamError.h"
 #include "llvm/Support/BinaryStreamRef.h"
 #include "llvm/Support/Endian.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cstdint>
 #include <type_traits>
 #include <utility>
 
 namespace llvm {
 
 /// Provides write only access to a subclass of `WritableBinaryStream`.
 /// Provides bounds checking and helpers for writing certain common data types
 /// such as null-terminated strings, integers in various flavors of endianness,
 /// etc.  Can be subclassed to provide reading and writing of custom datatypes,
 /// although no methods are overridable.
-class BinaryStreamWriter {
+class LLVM_SUPPORT_ABI BinaryStreamWriter {
 public:
   BinaryStreamWriter() = default;
   explicit BinaryStreamWriter(WritableBinaryStreamRef Ref);
   explicit BinaryStreamWriter(WritableBinaryStream &Stream);
   explicit BinaryStreamWriter(MutableArrayRef<uint8_t> Data,
                               llvm::support::endianness Endian);
 
   BinaryStreamWriter(const BinaryStreamWriter &Other)

llvm/include/llvm/Support/BlockFrequency.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_BLOCKFREQUENCY_H
 #define LLVM_SUPPORT_BLOCKFREQUENCY_H
 
 #include "llvm/Support/BranchProbability.h"
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 class raw_ostream;
 
 // This class represents Block Frequency as a 64-bit value.
-class BlockFrequency {
+class LLVM_SUPPORT_ABI BlockFrequency {
   uint64_t Frequency;
 
 public:
   BlockFrequency(uint64_t Freq = 0) : Frequency(Freq) { }
 
   /// Returns the maximum possible frequency, the saturation value.
   static uint64_t getMaxFrequency() { return -1ULL; }
 
public:
   bool operator>=(BlockFrequency RHS) const {
     return Frequency >= RHS.Frequency;
   }
 
   bool operator==(BlockFrequency RHS) const {
     return Frequency == RHS.Frequency;
   }
 };
 
 }
 
 #endif

llvm/include/llvm/Support/BranchProbability.h

 //===- BranchProbability.h - Branch Probability Wrapper ---------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // Definition of BranchProbability shared by IR and Machine Instructions.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_BRANCHPROBABILITY_H
 #define LLVM_SUPPORT_BRANCHPROBABILITY_H
 
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <algorithm>
 #include <cassert>
 #include <climits>
 #include <numeric>
 
 namespace llvm {
 
 class raw_ostream;
 
 // This class represents Branch Probability as a non-negative fraction that is
 // no greater than 1. It uses a fixed-point-like implementation, in which the
 // denominator is always a constant value (here we use 1<<31 for maximum
 // precision).
-class BranchProbability {
+class LLVM_SUPPORT_ABI BranchProbability {
   // Numerator
   uint32_t N;
 
   // Denominator, which is a constant value.
   static constexpr uint32_t D = 1u << 31;
   static constexpr uint32_t UnknownN = UINT32_MAX;
 
   // Construct a BranchProbability with only numerator assuming the denominator

llvm/include/llvm/Support/BuryPointer.h

 //===- llvm/Support/BuryPointer.h - Memory Manipulation/Leak ----*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_BURYPOINTER_H
 #define LLVM_SUPPORT_BURYPOINTER_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <memory>
 
 namespace llvm {
 
 // In tools that will exit soon anyway, going through the process of explicitly
 // deallocating resources can be unnecessary - better to leak the resources and
 // let the OS clean them up when the process ends. Use this function to ensure
 // the memory is not misdiagnosed as an unintentional leak by leak detection
 // tools (this is achieved by preserving pointers to the object in a globally
 // visible array).
-void BuryPointer(const void *Ptr);
+LLVM_SUPPORT_ABI void BuryPointer(const void *Ptr);
+
 template <typename T> void BuryPointer(std::unique_ptr<T> Ptr) {
   BuryPointer(Ptr.release());
 }
 
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/COM.h

 //===- llvm/Support/COM.h ---------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 /// \file
 ///
 /// Provides a library for accessing COM functionality of the Host OS.
 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_COM_H
 #define LLVM_SUPPORT_COM_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 namespace llvm {
 namespace sys {
 
 enum class COMThreadingMode { SingleThreaded, MultiThreaded };
 
-class InitializeCOMRAII {
+class LLVM_SUPPORT_ABI InitializeCOMRAII {
 public:
   explicit InitializeCOMRAII(COMThreadingMode Threading,
                              bool SpeedOverMemory = false);
   ~InitializeCOMRAII();
 
 private:
   InitializeCOMRAII(const InitializeCOMRAII &) = delete;
   void operator=(const InitializeCOMRAII &) = delete;
 };
 }
 }
 
 #endif

llvm/include/llvm/Support/CRC.h

 //===-- llvm/Support/CRC.h - Cyclic Redundancy Check-------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains implementations of CRC functions.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_CRC_H
 #define LLVM_SUPPORT_CRC_H
 
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 template <typename T> class ArrayRef;
 
 // Compute the CRC-32 of Data.
-uint32_t crc32(ArrayRef<uint8_t> Data);
+LLVM_SUPPORT_ABI uint32_t crc32(ArrayRef<uint8_t> Data);
 
 // Compute the running CRC-32 of Data, with CRC being the previous value of the
 // checksum.
-uint32_t crc32(uint32_t CRC, ArrayRef<uint8_t> Data);
+LLVM_SUPPORT_ABI uint32_t crc32(uint32_t CRC, ArrayRef<uint8_t> Data);
 
 // Class for computing the JamCRC.
 //
 // We will use the "Rocksoft^tm Model CRC Algorithm" to describe the properties
 // of this CRC:
 //   Width  : 32
 //   Poly   : 04C11DB7
 //   Init   : FFFFFFFF
 //   RefIn  : True
 //   RefOut : True
 //   XorOut : 00000000
 //   Check  : 340BC6D9 (result of CRC for "123456789")
 //
 // In other words, this is the same as CRC-32, except that XorOut is 0 instead
 // of FFFFFFFF.
 //
 // N.B.  We permit flexibility of the "Init" value.  Some consumers of this need
 //       it to be zero.
-class JamCRC {
+class LLVM_SUPPORT_ABI JamCRC {
 public:
   JamCRC(uint32_t Init = 0xFFFFFFFFU) : CRC(Init) {}
 
   // Update the CRC calculation with Data.
   void update(ArrayRef<uint8_t> Data);
 
   uint32_t getCRC() const { return CRC; }
 
 private:
   uint32_t CRC;
 };
 
 } // end namespace llvm
 
 #endif

llvm/include/llvm/Support/CachePruning.h

 // various policies.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_CACHEPRUNING_H
 #define LLVM_SUPPORT_CACHEPRUNING_H
 
 #include "llvm/ADT/Optional.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <chrono>
 
 namespace llvm {
 
 template <typename T> class Expected;
 class StringRef;
 
 /// Policy for the pruneCache() function. A default constructed
struct CachePruningPolicy {
   uint64_t MaxSizeFiles = 1000000;
 };
 
 /// Parse the given string as a cache pruning policy. Defaults are taken from a
 /// default constructed CachePruningPolicy object.
 /// For example: "prune_interval=30s:prune_after=24h:cache_size=50%"
 /// which means a pruning interval of 30 seconds, expiration time of 24 hours
 /// and maximum cache size of 50% of available disk space.
-Expected<CachePruningPolicy> parseCachePruningPolicy(StringRef PolicyStr);
+LLVM_SUPPORT_ABI Expected<CachePruningPolicy> parseCachePruningPolicy(StringRef PolicyStr);
 
 /// Peform pruning using the supplied policy, returns true if pruning
 /// occurred, i.e. if Policy.Interval was expired.
 ///
 /// As a safeguard against data loss if the user specifies the wrong directory
 /// as their cache directory, this function will ignore files not matching the
 /// pattern "llvmcache-*".
-bool pruneCache(StringRef Path, CachePruningPolicy Policy);
+LLVM_SUPPORT_ABI bool pruneCache(StringRef Path, CachePruningPolicy Policy);
 
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/Chrono.h

 //===- llvm/Support/Chrono.h - Utilities for Timing Manipulation-*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_CHRONO_H
 #define LLVM_SUPPORT_CHRONO_H
 
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/FormatProviders.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 #include <chrono>
 #include <ctime>
 
 namespace llvm {
 
 class raw_ostream;
 

 toTimePoint(std::time_t T, uint32_t nsec) {
   using namespace std::chrono;
   return time_point_cast<nanoseconds>(system_clock::from_time_t(T))
     + nanoseconds(nsec);
 }
 
 } // namespace sys
 
+LLVM_SUPPORT_ABI
 raw_ostream &operator<<(raw_ostream &OS, sys::TimePoint<> TP);
 
 /// Format provider for TimePoint<>
 ///
 /// The options string is a strftime format string, with extensions:
 ///   - %L is millis: 000-999
 ///   - %f is micros: 000000-999999
 ///   - %N is nanos: 000000000 - 999999999
 ///
 /// If no options are given, the default format is "%Y-%m-%d %H:%M:%S.%N".
 template <>
-struct format_provider<sys::TimePoint<>> {
+struct LLVM_SUPPORT_ABI format_provider<sys::TimePoint<>> {
   static void format(const sys::TimePoint<> &TP, llvm::raw_ostream &OS,
                      StringRef Style);
 };
 
 namespace detail {
 template <typename Period> struct unit { static const char value[]; };
 template <typename Period> const char unit<Period>::value[] = "";
 
-template <> struct unit<std::ratio<3600>> { static const char value[]; };
-template <> struct unit<std::ratio<60>> { static const char value[]; };
-template <> struct unit<std::ratio<1>> { static const char value[]; };
-template <> struct unit<std::milli> { static const char value[]; };
-template <> struct unit<std::micro> { static const char value[]; };
-template <> struct unit<std::nano> { static const char value[]; };
+template <> struct LLVM_SUPPORT_ABI unit<std::ratio<3600>> { static const char value[]; };
+template <> struct LLVM_SUPPORT_ABI unit<std::ratio<60>> { static const char value[]; };
+template <> struct LLVM_SUPPORT_ABI unit<std::ratio<1>> { static const char value[]; };
+template <> struct LLVM_SUPPORT_ABI unit<std::milli> { static const char value[]; };
+template <> struct LLVM_SUPPORT_ABI unit<std::micro> { static const char value[]; };
+template <> struct LLVM_SUPPORT_ABI unit<std::nano> { static const char value[]; };
 } // namespace detail
 
 /// Implementation of format_provider<T> for duration types.
 ///
 /// The options string of a duration type has the grammar:
 ///
 ///   duration_options  ::= [unit][show_unit [number_options]]
 ///   unit              ::= `h`|`m`|`s`|`ms|`us`|`ns`

llvm/include/llvm/Support/CodeGenCoverage.h

 //== llvm/Support/CodeGenCoverage.h ------------------------------*- C++ -*-==//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 /// \file This file provides rule coverage tracking for tablegen-erated CodeGen.
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_CODEGENCOVERAGE_H
 #define LLVM_SUPPORT_CODEGENCOVERAGE_H
 
 #include "llvm/ADT/BitVector.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 class LLVMContext;
 class MemoryBuffer;
 
-class CodeGenCoverage {
+class LLVM_SUPPORT_ABI CodeGenCoverage {
 protected:
   BitVector RuleCoverage;
 
 public:
   using const_covered_iterator = BitVector::const_set_bits_iterator;
 
   CodeGenCoverage();
 

llvm/include/llvm/Support/CommandLine.h

 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/ADT/iterator_range.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/ManagedStatic.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <cassert>
 #include <climits>
 #include <cstddef>
 #include <functional>
 #include <initializer_list>
 #include <string>
 #include <type_traits>
 #include <vector>

 //
 // If EnvVar is not nullptr, command-line options are also parsed from the
 // environment variable named by EnvVar.  Precedence is given to occurrences
 // from argv.  This precedence is currently implemented by parsing argv after
 // the environment variable, so it is only implemented correctly for options
 // that give precedence to later occurrences.  If your program supports options
 // that give precedence to earlier occurrences, you will need to extend this
 // function to support it correctly.
+LLVM_SUPPORT_ABI
 bool ParseCommandLineOptions(int argc, const char *const *argv,
                              StringRef Overview = "",
                              raw_ostream *Errs = nullptr,
                              const char *EnvVar = nullptr,
                              bool LongOptionsUseDoubleDash = false);
 
 // Function pointer type for printing version information.
 using VersionPrinterTy = std::function<void(raw_ostream &)>;
 
 ///===---------------------------------------------------------------------===//
 /// SetVersionPrinter - Override the default (LLVM specific) version printer
 ///                     used to print out the version when --version is given
 ///                     on the command line. This allows other systems using the
 ///                     CommandLine utilities to print their own version string.
-void SetVersionPrinter(VersionPrinterTy func);
+LLVM_SUPPORT_ABI void SetVersionPrinter(VersionPrinterTy func);
 
 ///===---------------------------------------------------------------------===//
 /// AddExtraVersionPrinter - Add an extra printer to use in addition to the
 ///                          default one. This can be called multiple times,
 ///                          and each time it adds a new function to the list
 ///                          which will be called after the basic LLVM version
 ///                          printing is complete. Each can then add additional
 ///                          information specific to the tool.
-void AddExtraVersionPrinter(VersionPrinterTy func);
+LLVM_SUPPORT_ABI void AddExtraVersionPrinter(VersionPrinterTy func);
 
 // PrintOptionValues - Print option values.
 // With -print-options print the difference between option values and defaults.
 // With -print-all-options print all option values.
 // (Currently not perfect, but best-effort.)
-void PrintOptionValues();
+LLVM_SUPPORT_ABI void PrintOptionValues();
 
 // Forward declaration - AddLiteralOption needs to be up here to make gcc happy.
 class Option;
 
 /// Adds a new option for parsing and provides the option it refers to.
 ///
 /// \param O pointer to the option
 /// \param Name the string name for the option to handle during parsing
 ///
 /// Literal options are used by some parsers to register special option values.
 /// This is how the PassNameParser registers pass names for opt.
+LLVM_SUPPORT_ABI
 void AddLiteralOption(Option &O, StringRef Name);
 
 //===----------------------------------------------------------------------===//
 // Flags permitted to be passed to command line arguments
 //
 
 enum NumOccurrencesFlag { // Flags for the number of occurrences allowed
   Optional = 0x00,        // Zero or One occurrence
enum MiscFlags {             // Miscellaneous flags to adjust argument
 
   // Default option
   DefaultOption = 0x10
 };
 
 //===----------------------------------------------------------------------===//
 // Option Category class
 //
-class OptionCategory {
+class LLVM_SUPPORT_ABI OptionCategory {
 private:
   StringRef const Name;
   StringRef const Description;
 
   void registerCategory();
 
 public:
   OptionCategory(StringRef const Name,
                  StringRef const Description = "")
       : Name(Name), Description(Description) {
     registerCategory();
   }
 
   StringRef getName() const { return Name; }
   StringRef getDescription() const { return Description; }
 };
 
 // The general Option Category (used as default category).
-OptionCategory &getGeneralCategory();
+LLVM_SUPPORT_ABI OptionCategory &getGeneralCategory();
 
 //===----------------------------------------------------------------------===//
 // SubCommand class
 //
-class SubCommand {
+class LLVM_SUPPORT_ABI SubCommand {
 private:
   StringRef Name;
   StringRef Description;
 
 protected:
   void registerSubCommand();
   void unregisterSubCommand();
 
public:
   SmallVector<Option *, 4> PositionalOpts;
   SmallVector<Option *, 4> SinkOpts;
   StringMap<Option *> OptionsMap;
 
   Option *ConsumeAfterOpt = nullptr; // The ConsumeAfter option if it exists.
 };
 
 // A special subcommand representing no subcommand
-extern ManagedStatic<SubCommand> TopLevelSubCommand;
+extern LLVM_SUPPORT_ABI ManagedStatic<SubCommand> TopLevelSubCommand;
 
 // A special subcommand that can be used to put an option into all subcommands.
-extern ManagedStatic<SubCommand> AllSubCommands;
+extern LLVM_SUPPORT_ABI ManagedStatic<SubCommand> AllSubCommands;
 
 //===----------------------------------------------------------------------===//
 // Option Base class
 //
-class Option {
+class LLVM_SUPPORT_ABI Option {
   friend class alias;
 
   // handleOccurrences - Overriden by subclasses to handle the value passed into
   // an argument.  Should return true if there was an error processing the
   // argument and the program should exit.
   //
   virtual bool handleOccurrence(unsigned pos, StringRef ArgName,
                                 StringRef Arg) = 0;
callback(F CB) {
   using arg_type = typename detail::callback_traits<F>::arg_type;
   return cb<result_type, arg_type>(CB);
 }
 
 //===----------------------------------------------------------------------===//
 // OptionValue class
 
 // Support value comparison outside the template.
-struct GenericOptionValue {
+struct LLVM_SUPPORT_ABI GenericOptionValue {
   virtual bool compare(const GenericOptionValue &V) const = 0;
 
 protected:
   GenericOptionValue() = default;
   GenericOptionValue(const GenericOptionValue&) = default;
   GenericOptionValue &operator=(const GenericOptionValue &) = default;
   ~GenericOptionValue() = default;
 
struct OptionValue<cl::boolOrDefault> final
   OptionValue(const cl::boolOrDefault &V) { this->setValue(V); }
 
   OptionValue<cl::boolOrDefault> &operator=(const cl::boolOrDefault &V) {
     setValue(V);
     return *this;
   }
 
 private:
-  void anchor() override;
+  LLVM_SUPPORT_ABI void anchor() override;
 };
 
 template <>
 struct OptionValue<std::string> final : OptionValueCopy<std::string> {
   using WrapperType = StringRef;
 
   OptionValue() = default;
 
   OptionValue(const std::string &V) { this->setValue(V); }
 
   OptionValue<std::string> &operator=(const std::string &V) {
     setValue(V);
     return *this;
   }
 
 private:
-  void anchor() override;
+  LLVM_SUPPORT_ABI void anchor() override;
 };
 
 //===----------------------------------------------------------------------===//
 // Enum valued command line option
 //
 
 // This represents a single enum value, using "int" as the underlying type.
 struct OptionEnumValue {

 // built-in, uses a mapping table to map specific options to values, which is
 // used, among other things, to handle enum types.
 
 //--------------------------------------------------
 // generic_parser_base - This class holds all the non-generic code that we do
 // not need replicated for every instance of the generic parser.  This also
 // allows us to put stuff into CommandLine.cpp
 //
-class generic_parser_base {
+class LLVM_SUPPORT_ABI generic_parser_base {
 protected:
   class GenericOptionInfo {
   public:
     GenericOptionInfo(StringRef name, StringRef helpStr)
         : Name(name), HelpStr(helpStr) {}
     StringRef Name;
     StringRef HelpStr;
   };
void removeLiteralOption(StringRef Name) {
     assert(N != Values.size() && "Option not found!");
     Values.erase(Values.begin() + N);
   }
 };
 
 //--------------------------------------------------
 // basic_parser - Super class of parsers to provide boilerplate code
 //
-class basic_parser_impl { // non-template implementation of basic_parser<t>
+class LLVM_SUPPORT_ABI basic_parser_impl { // non-template implementation of basic_parser<t>
 public:
   basic_parser_impl(Option &) {}
 
   virtual ~basic_parser_impl() {}
 
   enum ValueExpected getValueExpectedFlagDefault() const {
     return ValueRequired;
   }
public:
   using OptVal = OptionValue<DataType>;
 
   basic_parser(Option &O) : basic_parser_impl(O) {}
 };
 
 //--------------------------------------------------
 // parser<bool>
 //
-template <> class parser<bool> : public basic_parser<bool> {
+template <> class LLVM_SUPPORT_ABI parser<bool> : public basic_parser<bool> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, bool &Val);
 
   void initialize() {}
 
public:
   // An out-of-line virtual method to provide a 'home' for this class.
   void anchor() override;
 };
 
 extern template class basic_parser<bool>;
 
 //--------------------------------------------------
 // parser<boolOrDefault>
-template <> class parser<boolOrDefault> : public basic_parser<boolOrDefault> {
+template <> class LLVM_SUPPORT_ABI parser<boolOrDefault> : public basic_parser<boolOrDefault> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, boolOrDefault &Val);
 
   enum ValueExpected getValueExpectedFlagDefault() const {
     return ValueOptional;
public:
   void anchor() override;
 };
 
 extern template class basic_parser<boolOrDefault>;
 
 //--------------------------------------------------
 // parser<int>
 //
-template <> class parser<int> : public basic_parser<int> {
+template <> class LLVM_SUPPORT_ABI parser<int> : public basic_parser<int> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, int &Val);
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "int"; }
public:
   void anchor() override;
 };
 
 extern template class basic_parser<long>;
 
 //--------------------------------------------------
 // parser<long long>
 //
-template <> class parser<long long> : public basic_parser<long long> {
+template <> class LLVM_SUPPORT_ABI parser<long long> : public basic_parser<long long> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, long long &Val);
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "long"; }
 
   void printOptionDiff(const Option &O, long long V, OptVal Default,
                        size_t GlobalWidth) const;
 
   // An out-of-line virtual method to provide a 'home' for this class.
   void anchor() override;
 };
 
 extern template class basic_parser<long long>;
 
 //--------------------------------------------------
 // parser<unsigned>
 //
-template <> class parser<unsigned> : public basic_parser<unsigned> {
+template <> class LLVM_SUPPORT_ABI parser<unsigned> : public basic_parser<unsigned> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, unsigned &Val);
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "uint"; }

 };
 
 extern template class basic_parser<unsigned long>;
 
 //--------------------------------------------------
 // parser<unsigned long long>
 //
 template <>
-class parser<unsigned long long> : public basic_parser<unsigned long long> {
+class LLVM_SUPPORT_ABI parser<unsigned long long> : public basic_parser<unsigned long long> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg,
              unsigned long long &Val);
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "ulong"; }
 
   void printOptionDiff(const Option &O, unsigned long long V, OptVal Default,
                        size_t GlobalWidth) const;
 
   // An out-of-line virtual method to provide a 'home' for this class.
   void anchor() override;
 };
 
 extern template class basic_parser<unsigned long long>;
 
 //--------------------------------------------------
 // parser<double>
 //
-template <> class parser<double> : public basic_parser<double> {
+template <> class LLVM_SUPPORT_ABI parser<double> : public basic_parser<double> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, double &Val);
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "number"; }
 
   void printOptionDiff(const Option &O, double V, OptVal Default,
                        size_t GlobalWidth) const;
 
   // An out-of-line virtual method to provide a 'home' for this class.
   void anchor() override;
 };
 
 extern template class basic_parser<double>;
 
 //--------------------------------------------------
 // parser<float>
 //
-template <> class parser<float> : public basic_parser<float> {
+template <> class LLVM_SUPPORT_ABI parser<float> : public basic_parser<float> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &O, StringRef ArgName, StringRef Arg, float &Val);
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "number"; }
 
   void printOptionDiff(const Option &O, float V, OptVal Default,
                        size_t GlobalWidth) const;
 
   // An out-of-line virtual method to provide a 'home' for this class.
   void anchor() override;
 };
 
 extern template class basic_parser<float>;
 
 //--------------------------------------------------
 // parser<std::string>
 //
-template <> class parser<std::string> : public basic_parser<std::string> {
+template <> class LLVM_SUPPORT_ABI parser<std::string> : public basic_parser<std::string> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &, StringRef, StringRef Arg, std::string &Value) {
     Value = Arg.str();
     return false;
   }
 
   // getValueName - Overload in subclass to provide a better default value.
   StringRef getValueName() const override { return "string"; }
 
   void printOptionDiff(const Option &O, StringRef V, const OptVal &Default,
                        size_t GlobalWidth) const;
 
   // An out-of-line virtual method to provide a 'home' for this class.
   void anchor() override;
 };
 
 extern template class basic_parser<std::string>;
 
 //--------------------------------------------------
 // parser<char>
 //
-template <> class parser<char> : public basic_parser<char> {
+template <> class LLVM_SUPPORT_ABI parser<char> : public basic_parser<char> {
 public:
   parser(Option &O) : basic_parser(O) {}
 
   // parse - Return true on error.
   bool parse(Option &, StringRef, StringRef Arg, char &Value) {
     Value = Arg[0];
     return false;
   }
public:
   std::function<void(const typename ParserClass::parser_data_type &)> Callback =
       [](const typename ParserClass::parser_data_type &) {};
 };
 
 //===----------------------------------------------------------------------===//
 // Aliased command line option (alias this name to a preexisting name)
 //
 
-class alias : public Option {
+class LLVM_SUPPORT_ABI alias : public Option {
   Option *AliasFor;
 
   bool handleOccurrence(unsigned pos, StringRef /*ArgName*/,
                         StringRef Arg) override {
     return AliasFor->handleOccurrence(pos, AliasFor->ArgStr, Arg);
   }
 
   bool addOccurrence(unsigned pos, StringRef /*ArgName*/, StringRef Value,
struct aliasopt {
 
   void apply(alias &A) const { A.setAliasFor(Opt); }
 };
 
 // extrahelp - provide additional help at the end of the normal help
 // output. All occurrences of cl::extrahelp will be accumulated and
 // printed to stderr at the end of the regular help, just before
 // exit is called.
-struct extrahelp {
+struct LLVM_SUPPORT_ABI extrahelp {
   StringRef morehelp;
 
   explicit extrahelp(StringRef help);
 };
 
-void PrintVersionMessage();
+LLVM_SUPPORT_ABI void PrintVersionMessage();
 
 /// This function just prints the help message, exactly the same way as if the
 /// -help or -help-hidden option had been given on the command line.
 ///
 /// \param Hidden if true will print hidden options
 /// \param Categorized if true print options in categories
-void PrintHelpMessage(bool Hidden = false, bool Categorized = false);
+LLVM_SUPPORT_ABI void PrintHelpMessage(bool Hidden = false, bool Categorized = false);
 
 //===----------------------------------------------------------------------===//
 // Public interface for accessing registered options.
 //
 
 /// Use this to get a StringMap to all registered named options
 /// (e.g. -help).
 ///

 ///
 /// This interface is useful for modifying options in libraries that are out of
 /// the control of the client. The options should be modified before calling
 /// llvm::cl::ParseCommandLineOptions().
 ///
 /// Hopefully this API can be deprecated soon. Any situation where options need
 /// to be modified by tools or libraries should be handled by sane APIs rather
 /// than just handing around a global list.
-StringMap<Option *> &getRegisteredOptions(SubCommand &Sub = *TopLevelSubCommand);
+LLVM_SUPPORT_ABI StringMap<Option *> &getRegisteredOptions(SubCommand &Sub = *TopLevelSubCommand);
 
 /// Use this to get all registered SubCommands from the provided parser.
 ///
 /// \return A range of all SubCommand pointers registered with the parser.
 ///
 /// Typical usage:
 /// \code
 /// main(int argc, char* argv[]) {
 ///   llvm::cl::ParseCommandLineOptions(argc, argv);
 ///   for (auto* S : llvm::cl::getRegisteredSubcommands()) {
 ///     if (*S) {
 ///       std::cout << "Executing subcommand: " << S->getName() << std::endl;
 ///       // Execute some function based on the name...
 ///     }
 ///   }
 /// }
 /// \endcode
 ///
 /// This interface is useful for defining subcommands in libraries and
 /// the dispatch from a single point (like in the main function).
-iterator_range<typename SmallPtrSet<SubCommand *, 4>::iterator>
+LLVM_SUPPORT_ABI iterator_range<typename SmallPtrSet<SubCommand *, 4>::iterator>
 getRegisteredSubcommands();
 
 //===----------------------------------------------------------------------===//
 // Standalone command line processing utilities.
 //
 
 /// Tokenizes a command line that can contain escapes and quotes.
 //
 /// The quoting rules match those used by GCC and other tools that use
 /// libiberty's buildargv() or expandargv() utilities, and do not match bash.
 /// They differ from buildargv() on treatment of backslashes that do not escape
 /// a special character to make it possible to accept most Windows file paths.
 ///
 /// \param [in] Source The string to be split on whitespace with quotes.
 /// \param [in] Saver Delegates back to the caller for saving parsed strings.
 /// \param [in] MarkEOLs true if tokenizing a response file and you want end of
 /// lines and end of the response file to be marked with a nullptr string.
 /// \param [out] NewArgv All parsed strings are appended to NewArgv.
-void TokenizeGNUCommandLine(StringRef Source, StringSaver &Saver,
+LLVM_SUPPORT_ABI void TokenizeGNUCommandLine(StringRef Source, StringSaver &Saver,
                             SmallVectorImpl<const char *> &NewArgv,
                             bool MarkEOLs = false);
 
 /// Tokenizes a Windows command line which may contain quotes and escaped
 /// quotes.
 ///
 /// See MSDN docs for CommandLineToArgvW for information on the quoting rules.
 /// http://msdn.microsoft.com/en-us/library/windows/desktop/17w5ykft(v=vs.85).aspx
 ///
 /// \param [in] Source The string to be split on whitespace with quotes.
 /// \param [in] Saver Delegates back to the caller for saving parsed strings.
 /// \param [in] MarkEOLs true if tokenizing a response file and you want end of
 /// lines and end of the response file to be marked with a nullptr string.
 /// \param [out] NewArgv All parsed strings are appended to NewArgv.
-void TokenizeWindowsCommandLine(StringRef Source, StringSaver &Saver,
+LLVM_SUPPORT_ABI void TokenizeWindowsCommandLine(StringRef Source, StringSaver &Saver,
                                 SmallVectorImpl<const char *> &NewArgv,
                                 bool MarkEOLs = false);
 
 /// Tokenizes a Windows command line while attempting to avoid copies. If no
 /// quoting or escaping was used, this produces substrings of the original
 /// string. If a token requires unquoting, it will be allocated with the
 /// StringSaver.
 void TokenizeWindowsCommandLineNoCopy(StringRef Source, StringSaver &Saver,

 ///
 /// \param [in] Source The string representing content of config file.
 /// \param [in] Saver Delegates back to the caller for saving parsed strings.
 /// \param [out] NewArgv All parsed strings are appended to NewArgv.
 /// \param [in] MarkEOLs Added for compatibility with TokenizerCallback.
 ///
 /// It works like TokenizeGNUCommandLine with ability to skip comment lines.
 ///
-void tokenizeConfigFile(StringRef Source, StringSaver &Saver,
+LLVM_SUPPORT_ABI void tokenizeConfigFile(StringRef Source, StringSaver &Saver,
                         SmallVectorImpl<const char *> &NewArgv,
                         bool MarkEOLs = false);
 
 /// Reads command line options from the given configuration file.
 ///
 /// \param [in] CfgFileName Path to configuration file.
 /// \param [in] Saver  Objects that saves allocated strings.
 /// \param [out] Argv Array to which the read options are added.
 /// \return true if the file was successfully read.
 ///
 /// It reads content of the specified file, tokenizes it and expands "@file"
 /// commands resolving file names in them relative to the directory where
 /// CfgFilename resides.
 ///
-bool readConfigFile(StringRef CfgFileName, StringSaver &Saver,
+LLVM_SUPPORT_ABI bool readConfigFile(StringRef CfgFileName, StringSaver &Saver,
                     SmallVectorImpl<const char *> &Argv);
 
 /// Expand response files on a command line recursively using the given
 /// StringSaver and tokenization strategy.  Argv should contain the command line
 /// before expansion and will be modified in place. If requested, Argv will
 /// also be populated with nullptrs indicating where each response file line
 /// ends, which is useful for the "/link" argument that needs to consume all
 /// remaining arguments only until the next end of line, when in a response
 /// file.
 ///
 /// \param [in] Saver Delegates back to the caller for saving parsed strings.
 /// \param [in] Tokenizer Tokenization strategy. Typically Unix or Windows.
 /// \param [in,out] Argv Command line into which to expand response files.
 /// \param [in] MarkEOLs Mark end of lines and the end of the response file
 /// with nullptrs in the Argv vector.
 /// \param [in] RelativeNames true if names of nested response files must be
 /// resolved relative to including file.
 /// \param [in] FS File system used for all file access when running the tool.
 /// \param [in] CurrentDir Path used to resolve relative rsp files. If set to
 /// None, process' cwd is used instead.
 /// \return true if all @files were expanded successfully or there were none.
-bool ExpandResponseFiles(StringSaver &Saver, TokenizerCallback Tokenizer,
+LLVM_SUPPORT_ABI bool ExpandResponseFiles(StringSaver &Saver, TokenizerCallback Tokenizer,
                          SmallVectorImpl<const char *> &Argv, bool MarkEOLs,
                          bool RelativeNames,
                          llvm::Optional<llvm::StringRef> CurrentDir,
                          llvm::vfs::FileSystem &FS);
 
 /// An overload of ExpandResponseFiles() that uses
 /// llvm::vfs::getRealFileSystem().
-bool ExpandResponseFiles(
+LLVM_SUPPORT_ABI bool ExpandResponseFiles(
     StringSaver &Saver, TokenizerCallback Tokenizer,
     SmallVectorImpl<const char *> &Argv, bool MarkEOLs = false,
     bool RelativeNames = false,
     llvm::Optional<llvm::StringRef> CurrentDir = llvm::None);
 
 /// A convenience helper which concatenates the options specified by the
 /// environment variable EnvVar and command line options, then expands response
 /// files recursively. The tokenizer is a predefined GNU or Windows one.
 /// \return true if all @files were expanded successfully or there were none.
-bool expandResponseFiles(int Argc, const char *const *Argv, const char *EnvVar,
+LLVM_SUPPORT_ABI bool expandResponseFiles(int Argc, const char *const *Argv, const char *EnvVar,
                          StringSaver &Saver,
                          SmallVectorImpl<const char *> &NewArgv);
 
 /// Mark all options not part of this category as cl::ReallyHidden.
 ///
 /// \param Category the category of options to keep displaying
 ///
 /// Some tools (like clang-format) like to be able to hide all options that are
 /// not specific to the tool. This function allows a tool to specify a single
 /// option category to display in the -help output.
+LLVM_SUPPORT_ABI
 void HideUnrelatedOptions(cl::OptionCategory &Category,
                           SubCommand &Sub = *TopLevelSubCommand);
 
 /// Mark all options not part of the categories as cl::ReallyHidden.
 ///
 /// \param Categories the categories of options to keep displaying.
 ///
 /// Some tools (like clang-format) like to be able to hide all options that are
 /// not specific to the tool. This function allows a tool to specify a single
 /// option category to display in the -help output.
+LLVM_SUPPORT_ABI
 void HideUnrelatedOptions(ArrayRef<const cl::OptionCategory *> Categories,
                           SubCommand &Sub = *TopLevelSubCommand);
 
 /// Reset all command line options to a state that looks as if they have
 /// never appeared on the command line.  This is useful for being able to parse
 /// a command line multiple times (especially useful for writing tests).
-void ResetAllOptionOccurrences();
+LLVM_SUPPORT_ABI void ResetAllOptionOccurrences();
 
 /// Reset the command line parser back to its initial state.  This
 /// removes
 /// all options, categories, and subcommands and returns the parser to a state
 /// where no options are supported.
-void ResetCommandLineParser();
+LLVM_SUPPORT_ABI void ResetCommandLineParser();
 
 /// Parses `Arg` into the option handler `Handler`.
 bool ProvidePositionalOption(Option *Handler, StringRef Arg, int i);
 
 } // end namespace cl
 
 } // end namespace llvm
 
 #endif // LLVM_SUPPORT_COMMANDLINE_H

llvm/include/llvm/Support/Compression.h

 //===-- llvm/Support/Compression.h ---Compression----------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains basic functions for compression/uncompression.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_COMPRESSION_H
 #define LLVM_SUPPORT_COMPRESSION_H
 
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 template <typename T> class SmallVectorImpl;
 class Error;
 class StringRef;
 
 namespace zlib {
 
 static constexpr int NoCompression = 0;
 static constexpr int BestSpeedCompression = 1;
 static constexpr int DefaultCompression = 6;
 static constexpr int BestSizeCompression = 9;
 
-bool isAvailable();
+LLVM_SUPPORT_ABI bool isAvailable();
 
-Error compress(StringRef InputBuffer, SmallVectorImpl<char> &CompressedBuffer,
+LLVM_SUPPORT_ABI Error compress(StringRef InputBuffer, SmallVectorImpl<char> &CompressedBuffer,
                int Level = DefaultCompression);
 
-Error uncompress(StringRef InputBuffer, char *UncompressedBuffer,
+LLVM_SUPPORT_ABI Error uncompress(StringRef InputBuffer, char *UncompressedBuffer,
                  size_t &UncompressedSize);
 
-Error uncompress(StringRef InputBuffer,
+LLVM_SUPPORT_ABI Error uncompress(StringRef InputBuffer,
                  SmallVectorImpl<char> &UncompressedBuffer,
                  size_t UncompressedSize);
 
 uint32_t crc32(StringRef Buffer);
 
 }  // End of namespace zlib
 
 } // End of namespace llvm
 
 #endif

llvm/include/llvm/Support/ConvertUTF.h

/* ---------------------------------------------------------------------
     Rev History: Rick McGowan, fixes & updates May 2001.
          Fixes & updates, Sept 2001.
 
 ------------------------------------------------------------------------ */
 
 #ifndef LLVM_SUPPORT_CONVERTUTF_H
 #define LLVM_SUPPORT_CONVERTUTF_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cstddef>
 #include <string>
 #include <system_error>
 
 // Wrap everything in namespace llvm so that programs can link with llvm and
 // their own version of the unicode libraries.
 
 namespace llvm {
typedef enum {
   sourceIllegal           /* source sequence is illegal/malformed */
 } ConversionResult;
 
 typedef enum {
   strictConversion = 0,
   lenientConversion
 } ConversionFlags;
 
-ConversionResult ConvertUTF8toUTF16 (
+LLVM_SUPPORT_ABI ConversionResult ConvertUTF8toUTF16 (
   const UTF8** sourceStart, const UTF8* sourceEnd,
   UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags);
 
 /**
  * Convert a partial UTF8 sequence to UTF32.  If the sequence ends in an
  * incomplete code unit sequence, returns \c sourceExhausted.
  */
-ConversionResult ConvertUTF8toUTF32Partial(
+LLVM_SUPPORT_ABI ConversionResult ConvertUTF8toUTF32Partial(
   const UTF8** sourceStart, const UTF8* sourceEnd,
   UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags);
 
 /**
  * Convert a partial UTF8 sequence to UTF32.  If the sequence ends in an
  * incomplete code unit sequence, returns \c sourceIllegal.
  */
-ConversionResult ConvertUTF8toUTF32(
+LLVM_SUPPORT_ABI ConversionResult ConvertUTF8toUTF32(
   const UTF8** sourceStart, const UTF8* sourceEnd,
   UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags);
 
-ConversionResult ConvertUTF16toUTF8 (
+LLVM_SUPPORT_ABI ConversionResult ConvertUTF16toUTF8 (
   const UTF16** sourceStart, const UTF16* sourceEnd,
   UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags);
 
 ConversionResult ConvertUTF32toUTF8 (
   const UTF32** sourceStart, const UTF32* sourceEnd,
   UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags);
 
 ConversionResult ConvertUTF16toUTF32 (
   const UTF16** sourceStart, const UTF16* sourceEnd,
   UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags);
 
 ConversionResult ConvertUTF32toUTF16 (
   const UTF32** sourceStart, const UTF32* sourceEnd,
   UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags);
 
-Boolean isLegalUTF8Sequence(const UTF8 *source, const UTF8 *sourceEnd);
+LLVM_SUPPORT_ABI Boolean isLegalUTF8Sequence(const UTF8 *source, const UTF8 *sourceEnd);
 
-Boolean isLegalUTF8String(const UTF8 **source, const UTF8 *sourceEnd);
+LLVM_SUPPORT_ABI Boolean isLegalUTF8String(const UTF8 **source, const UTF8 *sourceEnd);
 
-unsigned getNumBytesForUTF8(UTF8 firstByte);
+LLVM_SUPPORT_ABI unsigned getNumBytesForUTF8(UTF8 firstByte);
 
 /*************************************************************************/
 /* Below are LLVM-specific wrappers of the functions above. */
 
 template <typename T> class ArrayRef;
 template <typename T> class SmallVectorImpl;
 class StringRef;
 
 /**
  * Convert an UTF8 StringRef to UTF8, UTF16, or UTF32 depending on
  * WideCharWidth. The converted data is written to ResultPtr, which needs to
  * point to at least WideCharWidth * (Source.Size() + 1) bytes. On success,
  * ResultPtr will point one after the end of the copied string. On failure,
  * ResultPtr will not be changed, and ErrorPtr will be set to the location of
  * the first character which could not be converted.
  * \return true on success.
  */
+LLVM_SUPPORT_ABI
 bool ConvertUTF8toWide(unsigned WideCharWidth, llvm::StringRef Source,
                        char *&ResultPtr, const UTF8 *&ErrorPtr);
 
 /**
 * Converts a UTF-8 StringRef to a std::wstring.
 * \return true on success.
 */
+LLVM_SUPPORT_ABI
 bool ConvertUTF8toWide(llvm::StringRef Source, std::wstring &Result);
 
 /**
 * Converts a UTF-8 C-string to a std::wstring.
 * \return true on success.
 */
+LLVM_SUPPORT_ABI
 bool ConvertUTF8toWide(const char *Source, std::wstring &Result);
 
 /**
 * Converts a std::wstring to a UTF-8 encoded std::string.
 * \return true on success.
 */
+LLVM_SUPPORT_ABI
 bool convertWideToUTF8(const std::wstring &Source, std::string &Result);
 
 
 /**
  * Convert an Unicode code point to UTF8 sequence.
  *
  * \param Source a Unicode code point.
  * \param [in,out] ResultPtr pointer to the output buffer, needs to be at least
  * \c UNI_MAX_UTF8_BYTES_PER_CODE_POINT bytes.  On success \c ResultPtr is
  * updated one past end of the converted sequence.
  *
  * \returns true on success.
  */
-bool ConvertCodePointToUTF8(unsigned Source, char *&ResultPtr);
+LLVM_SUPPORT_ABI bool ConvertCodePointToUTF8(unsigned Source, char *&ResultPtr);
 
 /**
  * Convert the first UTF8 sequence in the given source buffer to a UTF32
  * code point.
  *
  * \param [in,out] source A pointer to the source buffer. If the conversion
  * succeeds, this pointer will be updated to point to the byte just past the
  * end of the converted sequence.
if ((ptrdiff_t)size > sourceEnd - *source)
     return sourceExhausted;
   return ConvertUTF8toUTF32(source, *source + size, &target, target + 1, flags);
 }
 
 /**
  * Returns true if a blob of text starts with a UTF-16 big or little endian byte
  * order mark.
  */
-bool hasUTF16ByteOrderMark(ArrayRef<char> SrcBytes);
+LLVM_SUPPORT_ABI bool hasUTF16ByteOrderMark(ArrayRef<char> SrcBytes);
 
 /**
  * Converts a stream of raw bytes assumed to be UTF16 into a UTF8 std::string.
  *
  * \param [in] SrcBytes A buffer of what is assumed to be UTF-16 encoded text.
  * \param [out] Out Converted UTF-8 is stored here on success.
  * \returns true on success
  */
-bool convertUTF16ToUTF8String(ArrayRef<char> SrcBytes, std::string &Out);
+LLVM_SUPPORT_ABI bool convertUTF16ToUTF8String(ArrayRef<char> SrcBytes, std::string &Out);
 
 /**
 * Converts a UTF16 string into a UTF8 std::string.
 *
 * \param [in] Src A buffer of UTF-16 encoded text.
 * \param [out] Out Converted UTF-8 is stored here on success.
 * \returns true on success
 */
-bool convertUTF16ToUTF8String(ArrayRef<UTF16> Src, std::string &Out);
+LLVM_SUPPORT_ABI bool convertUTF16ToUTF8String(ArrayRef<UTF16> Src, std::string &Out);
 
 /**
  * Converts a UTF-8 string into a UTF-16 string with native endianness.
  *
  * \returns true on success
  */
-bool convertUTF8ToUTF16String(StringRef SrcUTF8,
+LLVM_SUPPORT_ABI bool convertUTF8ToUTF16String(StringRef SrcUTF8,
                               SmallVectorImpl<UTF16> &DstUTF16);
 
 #if defined(_WIN32)
 namespace sys {
 namespace windows {
-std::error_code UTF8ToUTF16(StringRef utf8, SmallVectorImpl<wchar_t> &utf16);
+LLVM_SUPPORT_ABI std::error_code UTF8ToUTF16(StringRef utf8, SmallVectorImpl<wchar_t> &utf16);
 /// Convert to UTF16 from the current code page used in the system
 std::error_code CurCPToUTF16(StringRef utf8, SmallVectorImpl<wchar_t> &utf16);
 std::error_code UTF16ToUTF8(const wchar_t *utf16, size_t utf16_len,
                             SmallVectorImpl<char> &utf8);
 /// Convert from UTF16 to the current code page used in the system
 std::error_code UTF16ToCurCP(const wchar_t *utf16, size_t utf16_len,
                              SmallVectorImpl<char> &utf8);
 } // namespace windows
 } // namespace sys
 #endif
 
 } /* end namespace llvm */
 
 #endif

llvm/include/llvm/Support/CrashRecoveryContext.h

 //===--- CrashRecoveryContext.h - Crash Recovery ----------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_CRASHRECOVERYCONTEXT_H
 #define LLVM_SUPPORT_CRASHRECOVERYCONTEXT_H
 
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 class CrashRecoveryContextCleanup;
 
 /// Crash recovery helper object.
 ///
 /// This class implements support for running operations in a safe context so
 /// that crashes (memory errors, stack overflow, assertion violations) can be

 ///
 ///      ... no crash was detected ...
 ///    }
 /// \endcode
 ///
 /// To assist recovery the class allows specifying set of actions that will be
 /// executed in any case, whether crash occurs or not. These actions may be used
 /// to reclaim resources in the case of crash.
-class CrashRecoveryContext {
+class LLVM_SUPPORT_ABI CrashRecoveryContext {
   void *Impl = nullptr;
   CrashRecoveryContextCleanup *head = nullptr;
 
 public:
   CrashRecoveryContext();
   ~CrashRecoveryContext();
 
   /// Register cleanup handler, which is used when the recovery context is

 
 /// Abstract base class of cleanup handlers.
 ///
 /// Derived classes override method recoverResources, which makes actual work on
 /// resource recovery.
 ///
 /// Cleanup handlers are stored in a double list, which is owned and managed by
 /// a crash recovery context.
-class CrashRecoveryContextCleanup {
+class LLVM_SUPPORT_ABI CrashRecoveryContextCleanup {
 protected:
   CrashRecoveryContext *context = nullptr;
   CrashRecoveryContextCleanup(CrashRecoveryContext *context)
       : context(context) {}
 
 public:
   bool cleanupFired = false;
 

llvm/include/llvm/Support/DJB.h

 //===-- llvm/Support/DJB.h ---DJB Hash --------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains support for the DJ Bernstein hash function.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_DJB_H
 #define LLVM_SUPPORT_DJB_H
 
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 /// The Bernstein hash function used by the DWARF accelerator tables.
 inline uint32_t djbHash(StringRef Buffer, uint32_t H = 5381) {
   for (unsigned char C : Buffer.bytes())
     H = (H << 5) + H + C;
   return H;
 }
 
 /// Computes the Bernstein hash after folding the input according to the Dwarf 5
 /// standard case folding rules.
-uint32_t caseFoldingDjbHash(StringRef Buffer, uint32_t H = 5381);
+LLVM_SUPPORT_ABI uint32_t caseFoldingDjbHash(StringRef Buffer, uint32_t H = 5381);
 } // namespace llvm
 
 #endif // LLVM_SUPPORT_DJB_H

llvm/include/llvm/Support/DataExtractor.h

 //===-- DataExtractor.h -----------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_DATAEXTRACTOR_H
 #define LLVM_SUPPORT_DATAEXTRACTOR_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 /// An auxiliary type to facilitate extraction of 3-byte entities.
 struct Uint24 {
   uint8_t Bytes[3];
   Uint24(uint8_t U) {
     Bytes[0] = Bytes[1] = Bytes[2] = U;

 using uint24_t = Uint24;
 static_assert(sizeof(uint24_t) == 3, "sizeof(uint24_t) != 3");
 
 /// Needed by swapByteOrder().
 inline uint24_t getSwappedBytes(uint24_t C) {
   return uint24_t(C.Bytes[2], C.Bytes[1], C.Bytes[0]);
 }
 
-class DataExtractor {
+class LLVM_SUPPORT_ABI DataExtractor {
   StringRef Data;
   uint8_t IsLittleEndian;
   uint8_t AddressSize;
 public:
   /// A class representing a position in a DataExtractor, as well as any error
   /// encountered during extraction. It enables one to extract a sequence of
   /// values without error-checking and then checking for errors in bulk at the
   /// end. The class holds an Error object, so failing to check the result of

llvm/include/llvm/Support/Debug.h

 // LLVM_DEBUG() statements disappears, so it does not affect the runtime of the
 // code.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_DEBUG_H
 #define LLVM_SUPPORT_DEBUG_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 namespace llvm {
 
 class raw_ostream;
 
 #ifndef NDEBUG
 
 /// isCurrentDebugType - Return true if the specified string is the debug type
 /// specified on the command line, or if none was specified on the command line

 ///\}
 
 /// EnableDebugBuffering - This defaults to false.  If true, the debug
 /// stream will install signal handlers to dump any buffered debug
 /// output.  It allows clients to selectively allow the debug stream
 /// to install signal handlers if they are certain there will be no
 /// conflict.
 ///
-extern bool EnableDebugBuffering;
+extern LLVM_SUPPORT_ABI bool EnableDebugBuffering;
 
 /// dbgs() - This returns a reference to a raw_ostream for debugging
 /// messages.  If debugging is disabled it returns errs().  Use it
 /// like: dbgs() << "foo" << "bar";
-raw_ostream &dbgs();
+LLVM_SUPPORT_ABI raw_ostream &dbgs();
 
 // DEBUG macro - This macro should be used by passes to emit debug information.
 // In the '-debug' option is specified on the commandline, and if this is a
 // debug build, then the code specified as the option to the macro will be
 // executed.  Otherwise it will not be.  Example:
 //
 // LLVM_DEBUG(dbgs() << "Bitset contains: " << Bitset << "\n");
 //
 #define LLVM_DEBUG(X) DEBUG_WITH_TYPE(DEBUG_TYPE, X)
 
 } // end namespace llvm
 
 #endif // LLVM_SUPPORT_DEBUG_H

llvm/include/llvm/Support/DebugCounter.h

 
 #ifndef LLVM_SUPPORT_DEBUGCOUNTER_H
 #define LLVM_SUPPORT_DEBUGCOUNTER_H
 
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/UniqueVector.h"
 #include "llvm/Support/Debug.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 
 namespace llvm {
 
 class raw_ostream;
 
-class DebugCounter {
+class LLVM_SUPPORT_ABI DebugCounter {
 public:
   ~DebugCounter();
 
   /// Returns a reference to the singleton instance.
   static DebugCounter &instance();
 
   // Used by the command line option parser to push a new value it parsed.
   void push_back(const std::string &);

llvm/include/llvm/Support/DynamicLibrary.h

 //===-- llvm/Support/DynamicLibrary.h - Portable Dynamic Library -*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file declares the sys::DynamicLibrary class.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_DYNAMICLIBRARY_H
 #define LLVM_SUPPORT_DYNAMICLIBRARY_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 
 namespace llvm {
 
 class StringRef;
 
 namespace sys {
 
   /// This class provides a portable interface to dynamic libraries which also
   /// might be known as shared libraries, shared objects, dynamic shared
   /// objects, or dynamic link libraries. Regardless of the terminology or the
   /// operating system interface, this class provides a portable interface that
   /// allows dynamic libraries to be loaded and searched for externally
   /// defined symbols. This is typically used to provide "plug-in" support.
   /// It also allows for symbols to be defined which don't live in any library,
   /// but rather the main program itself, useful on Windows where the main
   /// executable cannot be searched.
   ///
   /// Note: there is currently no interface for temporarily loading a library,
   /// or for unloading libraries when the LLVM library is unloaded.
-  class DynamicLibrary {
+  class LLVM_SUPPORT_ABI DynamicLibrary {
     // Placeholder whose address represents an invalid library.
     // We use this instead of NULL or a pointer-int pair because the OS library
     // might define 0 or 1 to be "special" handles, such as "search all".
     static char Invalid;
 
     // Opaque data used to interface with OS-specific dynamic library handling.
     void *Data;
 
public:
 
     /// This functions permanently adds the symbol \p symbolName with the
     /// value \p symbolValue.  These symbols are searched before any
     /// libraries.
     /// Add searchable symbol/value pair.
     static void AddSymbol(StringRef symbolName, void *symbolValue);
 
     class HandleSet;
   };
 
 } // End sys namespace
 } // End llvm namespace
 
 #endif

llvm/include/llvm/Support/ELFAttributeParser.h

 //===- ELF AttributeParser.h - ELF Attribute Parser -------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ELFATTRIBUTEPARSER_H
 #define LLVM_SUPPORT_ELFATTRIBUTEPARSER_H
 
 #include "ELFAttributes.h"
 #include "ScopedPrinter.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/Support/DataExtractor.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 #include <unordered_map>
 
 namespace llvm {
 class StringRef;
 
-class ELFAttributeParser {
+class LLVM_SUPPORT_ABI ELFAttributeParser {
   StringRef vendor;
   std::unordered_map<unsigned, unsigned> attributes;
   std::unordered_map<unsigned, StringRef> attributesStr;
 
   virtual Error handler(uint64_t tag, bool &handled) = 0;
 
 protected:
   ScopedPrinter *sw;

llvm/include/llvm/Support/ELFAttributes.h

 //===-- ELFAttributes.h - ELF Attributes ------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ELFATTRIBUTES_H
 #define LLVM_SUPPORT_ELFATTRIBUTES_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 struct TagNameItem {
   unsigned attr;
   StringRef tagName;
 };
 
 using TagNameMap = ArrayRef<TagNameItem>;
 
 namespace ELFAttrs {
 
 enum AttrType : unsigned { File = 1, Section = 2, Symbol = 3 };
 
-StringRef attrTypeAsString(unsigned attr, TagNameMap tagNameMap,
+LLVM_SUPPORT_ABI StringRef attrTypeAsString(unsigned attr, TagNameMap tagNameMap,
                            bool hasTagPrefix = true);
-Optional<unsigned> attrTypeFromString(StringRef tag, TagNameMap tagNameMap);
+LLVM_SUPPORT_ABI Optional<unsigned> attrTypeFromString(StringRef tag, TagNameMap tagNameMap);
 
 // Magic numbers for ELF attributes.
 enum AttrMagic { Format_Version = 0x41 };
 
 } // namespace ELFAttrs
 } // namespace llvm
 #endif

llvm/include/llvm/Support/Error.h

 #include "llvm/ADT/Twine.h"
 #include "llvm/Config/abi-breaking.h"
 #include "llvm/Support/AlignOf.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/ErrorOr.h"
 #include "llvm/Support/Format.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <algorithm>
 #include <cassert>
 #include <cstdint>
 #include <cstdlib>
 #include <functional>
 #include <memory>
 #include <new>
 #include <string>
 #include <system_error>
 #include <type_traits>
 #include <utility>
 #include <vector>
 
 namespace llvm {
 
 class ErrorSuccess;
 
 /// Base class for error info classes. Do not extend this directly: Extend
 /// the ErrorInfo template subclass instead.
-class ErrorInfoBase {
+class LLVM_SUPPORT_ABI ErrorInfoBase {
 public:
   virtual ~ErrorInfoBase() = default;
 
   /// Print an error message to an output stream.
   virtual void log(raw_ostream &OS) const = 0;
 
   /// Return the error message as a string.
   virtual std::string message() const {
public:
   void log(raw_ostream &OS) const override {
     OS << "Multiple errors:\n";
     for (auto &ErrPayload : Payloads) {
       ErrPayload->log(OS);
       OS << "\n";
     }
   }
 
+  // FIXME(compnerd) is this portable?
+  LLVM_SUPPORT_ABI
   std::error_code convertToErrorCode() const override;
 
   // Used by ErrorInfo::classID.
-  static char ID;
+  // FIXME(compnerd) is this portable?
+  LLVM_SUPPORT_ABI static char ID;
 
 private:
   ErrorList(std::unique_ptr<ErrorInfoBase> Payload1,
             std::unique_ptr<ErrorInfoBase> Payload2) {
     assert(!Payload1->isA<ErrorList>() && !Payload2->isA<ErrorList>() &&
            "ErrorList constructor payloads should be singleton errors");
     Payloads.push_back(std::move(Payload1));
     Payloads.push_back(std::move(Payload2));
#endif
   bool HasError : 1;
 #if LLVM_ENABLE_ABI_BREAKING_CHECKS
   bool Unchecked : 1;
 #endif
 };
 
 /// Report a serious error, calling any installed error handler. See
 /// ErrorHandling.h.
-[[noreturn]] void report_fatal_error(Error Err, bool gen_crash_diag = true);
+[[noreturn]] LLVM_SUPPORT_ABI void report_fatal_error(Error Err, bool gen_crash_diag = true);
 
 /// Report a fatal error if Err is a failure value.
 ///
 /// This function can be used to wrap calls to fallible functions ONLY when it
 /// is known that the Error will always be a success value. E.g.
 ///
 ///   @code{.cpp}
 ///   // foo only attempts the fallible operation if DoFallibleOperation is

 ///
 /// This function is compatible with the helpers from Support/WithColor.h. You
 /// can pass any of them as the OS. Please consider using them instead of
 /// including 'error: ' in the ErrorBanner.
 ///
 /// This is useful in the base level of your program to allow clean termination
 /// (allowing clean deallocation of resources, etc.), while reporting error
 /// information to the user.
-void logAllUnhandledErrors(Error E, raw_ostream &OS, Twine ErrorBanner = {});
+LLVM_SUPPORT_ABI void logAllUnhandledErrors(Error E, raw_ostream &OS, Twine ErrorBanner = {});
 
 /// Write all error messages (if any) in E to a string. The newline character
 /// is used to separate error messages.
 inline std::string toString(Error E) {
   SmallVector<std::string, 2> Errors;
   handleAllErrors(std::move(E), [&Errors](const ErrorInfoBase &EI) {
     Errors.push_back(EI.message());
   });
private:
   Expected<T> *ValOrErr;
 };
 
 /// This class wraps a std::error_code in a Error.
 ///
 /// This is useful if you're writing an interface that returns a Error
 /// (or Expected) and you want to call code that still returns
 /// std::error_codes.
-class ECError : public ErrorInfo<ECError> {
-  friend Error errorCodeToError(std::error_code);
+class LLVM_SUPPORT_ABI ECError : public ErrorInfo<ECError> {
+  friend LLVM_SUPPORT_ABI Error errorCodeToError(std::error_code);
 
   virtual void anchor() override;
 
 public:
   void setErrorCode(std::error_code EC) { this->EC = EC; }
   std::error_code convertToErrorCode() const override { return EC; }
   void log(raw_ostream &OS) const override { OS << EC.message(); }
 
   // Used by ErrorInfo::classID.
   static char ID;
 
 protected:
   ECError() = default;
   ECError(std::error_code EC) : EC(EC) {}
 
   std::error_code EC;
 };
 
 /// The value returned by this function can be returned from convertToErrorCode
 /// for Error values where no sensible translation to std::error_code exists.
 /// It should only be used in this situation, and should never be used where a
 /// sensible conversion to std::error_code is available, as attempts to convert
 /// to/from this error will result in a fatal error. (i.e. it is a programmatic
 ///error to try to convert such a value).
-std::error_code inconvertibleErrorCode();
+LLVM_SUPPORT_ABI std::error_code inconvertibleErrorCode();
 
 /// Helper for converting an std::error_code to a Error.
-Error errorCodeToError(std::error_code EC);
+LLVM_SUPPORT_ABI Error errorCodeToError(std::error_code EC);
 
 /// Helper for converting an ECError to a std::error_code.
 ///
 /// This method requires that Err be Error() or an ECError, otherwise it
 /// will trigger a call to abort().
-std::error_code errorToErrorCode(Error Err);
+LLVM_SUPPORT_ABI std::error_code errorToErrorCode(Error Err);
 
 /// Convert an ErrorOr<T> to an Expected<T>.
 template <typename T> Expected<T> errorOrToExpected(ErrorOr<T> &&EO) {
   if (auto EC = EO.getError())
     return errorCodeToError(EC);
   return std::move(*EO);
 }
 

 ///
 ///   @code{.cpp}
 ///   Expected<int> foo() {
 ///      return llvm::make_error<PDBError>(pdb_error_code::dia_failed_loading,
 ///                                        "Additional information");
 ///   }
 ///   @endcode
 ///
-class StringError : public ErrorInfo<StringError> {
+class LLVM_SUPPORT_ABI StringError : public ErrorInfo<StringError> {
 public:
   static char ID;
 
   // Prints EC + S and converts to EC
   StringError(std::error_code EC, const Twine &S = Twine());
 
   // Prints S and converts to EC
   StringError(const Twine &S, std::error_code EC);

 inline Error createStringError(std::error_code EC, char const *Fmt,
                                const Ts &... Vals) {
   std::string Buffer;
   raw_string_ostream Stream(Buffer);
   Stream << format(Fmt, Vals...);
   return make_error<StringError>(Stream.str(), EC);
 }
 
-Error createStringError(std::error_code EC, char const *Msg);
+LLVM_SUPPORT_ABI Error createStringError(std::error_code EC, char const *Msg);
 
 inline Error createStringError(std::error_code EC, const Twine &S) {
   return createStringError(EC, S.str().c_str());
 }
 
 template <typename... Ts>
 inline Error createStringError(std::errc EC, char const *Fmt,
                                const Ts &... Vals) {
   return createStringError(std::make_error_code(EC), Fmt, Vals...);
 }
 
 /// This class wraps a filename and another Error.
 ///
 /// In some cases, an error needs to live along a 'source' name, in order to
 /// show more detailed information to the user.
-class FileError final : public ErrorInfo<FileError> {
+class LLVM_SUPPORT_ABI FileError final : public ErrorInfo<FileError> {
 
   friend Error createFileError(const Twine &, Error);
   friend Error createFileError(const Twine &, size_t, Error);
 
 public:
   void log(raw_ostream &OS) const override {
     assert(Err && !FileName.empty() && "Trying to log after takeError().");
     OS << "'" << FileName << "': ";

llvm/include/llvm/Support/ErrorHandling.h

 // errors (most of them) should be handled through LLVMContext.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ERRORHANDLING_H
 #define LLVM_SUPPORT_ERRORHANDLING_H
 
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 
 namespace llvm {
 class StringRef;
   class Twine;
 
   /// An error handler callback.
   typedef void (*fatal_error_handler_t)(void *user_data,
class StringRef;
   ///
   /// It is dangerous to naively use an error handler which throws an exception.
   /// Even though some applications desire to gracefully recover from arbitrary
   /// faults, blindly throwing exceptions through unfamiliar code isn't a way to
   /// achieve this.
   ///
   /// \param user_data - An argument which will be passed to the install error
   /// handler.
-  void install_fatal_error_handler(fatal_error_handler_t handler,
+  LLVM_SUPPORT_ABI void install_fatal_error_handler(fatal_error_handler_t handler,
                                    void *user_data = nullptr);
 
   /// Restores default error handling behaviour.
-  void remove_fatal_error_handler();
+  LLVM_SUPPORT_ABI void remove_fatal_error_handler();
 
   /// ScopedFatalErrorHandler - This is a simple helper class which just
   /// calls install_fatal_error_handler in its constructor and
   /// remove_fatal_error_handler in its destructor.
   struct ScopedFatalErrorHandler {
     explicit ScopedFatalErrorHandler(fatal_error_handler_t handler,
                                      void *user_data = nullptr) {
       install_fatal_error_handler(handler, user_data);
     }
 
     ~ScopedFatalErrorHandler() { remove_fatal_error_handler(); }
   };
 
 /// Reports a serious error, calling any installed error handler. These
 /// functions are intended to be used for error conditions which are outside
 /// the control of the compiler (I/O errors, invalid user input, etc.)
 ///
 /// If no error handler is installed the default is to print the message to
 /// standard error, followed by a newline.
 /// After the error handler is called this function will call abort(), it
 /// does not return.
-[[noreturn]] void report_fatal_error(const char *reason,
+[[noreturn]] LLVM_SUPPORT_ABI void report_fatal_error(const char *reason,
                                      bool gen_crash_diag = true);
-[[noreturn]] void report_fatal_error(const std::string &reason,
+[[noreturn]] LLVM_SUPPORT_ABI void report_fatal_error(const std::string &reason,
                                      bool gen_crash_diag = true);
-[[noreturn]] void report_fatal_error(StringRef reason,
+[[noreturn]] LLVM_SUPPORT_ABI void report_fatal_error(StringRef reason,
                                      bool gen_crash_diag = true);
-[[noreturn]] void report_fatal_error(const Twine &reason,
+[[noreturn]] LLVM_SUPPORT_ABI void report_fatal_error(const Twine &reason,
                                      bool gen_crash_diag = true);
 
 /// Installs a new bad alloc error handler that should be used whenever a
 /// bad alloc error, e.g. failing malloc/calloc, is encountered by LLVM.
 ///
 /// The user can install a bad alloc handler, in order to define the behavior
 /// in case of failing allocations, e.g. throwing an exception. Note that this
 /// handler must not trigger any additional allocations itself.
 ///
 /// If no error handler is installed the default is to print the error message
 /// to stderr, and call exit(1).  If an error handler is installed then it is
 /// the handler's responsibility to log the message, it will no longer be
 /// printed to stderr.  If the error handler returns, then exit(1) will be
 /// called.
 ///
 ///
 /// \param user_data - An argument which will be passed to the installed error
 /// handler.
 void install_bad_alloc_error_handler(fatal_error_handler_t handler,
                                      void *user_data = nullptr);
 
 /// Restores default bad alloc error handling behavior.
 void remove_bad_alloc_error_handler();
 
 void install_out_of_memory_new_handler();
 
 /// Reports a bad alloc error, calling any user defined bad alloc
 /// error handler. In contrast to the generic 'report_fatal_error'
 /// functions, this function might not terminate, e.g. the user
 /// defined error handler throws an exception, but it won't return.
 ///
 /// Note: When throwing an exception in the bad alloc handler, make sure that
 /// the following unwind succeeds, e.g. do not trigger additional allocations
 /// in the unwind chain.
 ///
 /// If no error handler is installed (default), throws a bad_alloc exception
 /// if LLVM is compiled with exception support. Otherwise prints the error
 /// to standard error and calls abort().
-[[noreturn]] void report_bad_alloc_error(const char *Reason,
+[[noreturn]] LLVM_SUPPORT_ABI void report_bad_alloc_error(const char *Reason,
                                          bool GenCrashDiag = true);
 
 /// This function calls abort(), and prints the optional message to stderr.
 /// Use the llvm_unreachable macro (that adds location info), instead of
 /// calling this function directly.
-[[noreturn]] void
+[[noreturn]] LLVM_SUPPORT_ABI void
 llvm_unreachable_internal(const char *msg = nullptr, const char *file = nullptr,
                           unsigned line = 0);
 }
 
 /// Marks that the current location is not supposed to be reachable.
 /// In !NDEBUG builds, prints the message and location info to stderr.
 /// In NDEBUG builds, becomes an optimizer hint that the current location
 /// is not supposed to be reachable.  On compilers that don't support

llvm/include/llvm/Support/ExtensibleRTTI.h

 //
 //   @endcode
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_EXTENSIBLERTTI_H
 #define LLVM_SUPPORT_EXTENSIBLERTTI_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 namespace llvm {
 
 template <typename ThisT, typename ParentT> class RTTIExtends;
 
 /// Base class for the extensible RTTI hierarchy.
 ///
 /// This class defines virtual methods, dynamicClassID and isA, that enable
 /// type comparisons.
-class RTTIRoot {
+class LLVM_SUPPORT_ABI RTTIRoot {
 public:
   virtual ~RTTIRoot() = default;
 
   /// Returns the class ID for this type.
   static const void *classID() { return &ID; }
 
   /// Returns the class ID for the dynamic type of this RTTIRoot instance.
   virtual const void *dynamicClassID() const = 0;

llvm/include/llvm/Support/FileCollector.h

 //===-- FileCollector.h -----------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_FILECOLLECTOR_H
 #define LLVM_SUPPORT_FILECOLLECTOR_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringSet.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/VirtualFileSystem.h"
+
 #include <mutex>
 #include <string>
 
 namespace llvm {
 class FileCollectorFileSystem;
 class Twine;
 
-class FileCollectorBase {
+class LLVM_SUPPORT_ABI FileCollectorBase {
 public:
   FileCollectorBase();
   virtual ~FileCollectorBase();
 
   void addFile(const Twine &file);
   void addDirectory(const Twine &Dir);
 
 protected:

 ///
 /// We generate file copies and mapping lazily - see writeMapping and copyFiles.
 /// We don't try to capture the state of the file at the exact time when it's
 /// accessed. Files might get changed, deleted ... we record only the "final"
 /// state.
 ///
 /// In order to preserve the relative topology of files we use their real paths
 /// as relative paths inside of the Root.
-class FileCollector : public FileCollectorBase {
+class LLVM_SUPPORT_ABI FileCollector : public FileCollectorBase {
 public:
   /// Helper utility that encapsulates the logic for canonicalizing a virtual
   /// path and a path to copy from.
-  class PathCanonicalizer {
+  class LLVM_SUPPORT_ABI PathCanonicalizer {
   public:
-    struct PathStorage {
+    struct LLVM_SUPPORT_ABI PathStorage {
       SmallString<256> CopyFrom;
       SmallString<256> VirtualPath;
     };
 
     /// Canonicalize a pair of virtual and real paths.
     PathStorage canonicalize(StringRef SrcPath);
 
   private:

llvm/include/llvm/Support/FileOutputBuffer.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_FILEOUTPUTBUFFER_H
 #define LLVM_SUPPORT_FILEOUTPUTBUFFER_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 /// FileOutputBuffer - This interface provides simple way to create an in-memory
 /// buffer which will be written to a file. During the lifetime of these
 /// objects, the content or existence of the specified file is undefined. That
 /// is, creating an OutputBuffer for a file may immediately remove the file.
 /// If the FileOutputBuffer is committed, the target file's content will become
 /// the buffer content at the time of the commit.  If the FileOutputBuffer is
 /// not committed, the file will be deleted in the FileOutputBuffer destructor.
-class FileOutputBuffer {
+class LLVM_SUPPORT_ABI FileOutputBuffer {
 public:
   enum {
     /// Set the 'x' bit on the resulting file.
     F_executable = 1,
 
     /// Don't use mmap and instead write an in-memory buffer to a file when this
     /// buffer is closed.
     F_no_mmap = 2,

llvm/include/llvm/Support/FileSystem.h

 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Config/llvm-config.h"
 #include "llvm/Support/Chrono.h"
 #include "llvm/Support/Error.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/ErrorOr.h"
 #include "llvm/Support/FileSystem/UniqueID.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MD5.h"
+
 #include <cassert>
 #include <cstdint>
 #include <ctime>
 #include <memory>
 #include <stack>
 #include <string>
 #include <system_error>
 #include <vector>
inline perms operator~(perms x) {
   // Avoid UB by explicitly truncating the (unsigned) ~ result.
   return static_cast<perms>(
       static_cast<unsigned short>(~static_cast<unsigned short>(x)));
 }
 
 /// Represents the result of a call to directory_iterator::status(). This is a
 /// subset of the information returned by a regular sys::fs::status() call, and
 /// represents the information provided by Windows FileFirstFile/FindNextFile.
-class basic_file_status {
+class LLVM_SUPPORT_ABI basic_file_status {
 protected:
   #if defined(LLVM_ON_UNIX)
   time_t fs_st_atime = 0;
   time_t fs_st_mtime = 0;
   uint32_t fs_st_atime_nsec = 0;
   uint32_t fs_st_mtime_nsec = 0;
   uid_t fs_st_uid = 0;
   gid_t fs_st_gid = 0;
#elif defined(_WIN32)
 
   // setters
   void type(file_type v) { Type = v; }
   void permissions(perms p) { Perms = p; }
 };
 
 /// Represents the result of a call to sys::fs::status().
 class file_status : public basic_file_status {
-  friend bool equivalent(file_status A, file_status B);
+  friend LLVM_SUPPORT_ABI bool equivalent(file_status A, file_status B);
 
   #if defined(LLVM_ON_UNIX)
   dev_t fs_st_dev = 0;
   nlink_t fs_st_nlinks = 0;
   ino_t fs_st_ino = 0;
   #elif defined (_WIN32)
   uint32_t NumLinks = 0;
   uint32_t VolumeSerialNumber = 0;
   uint32_t FileIndexHigh = 0;
   uint32_t FileIndexLow = 0;

 ///
 /// Makes \a path absolute using the \a current_directory if it is not already.
 /// An empty \a path will result in the \a current_directory.
 ///
 /// /absolute/path   => /absolute/path
 /// relative/../path => <current-directory>/relative/../path
 ///
 /// @param path A path that is modified to be an absolute path.
-void make_absolute(const Twine &current_directory, SmallVectorImpl<char> &path);
+LLVM_SUPPORT_ABI void make_absolute(const Twine &current_directory, SmallVectorImpl<char> &path);
 
 /// Make \a path an absolute path.
 ///
 /// Makes \a path absolute using the current directory if it is not already. An
 /// empty \a path will result in the current directory.
 ///
 /// /absolute/path   => /absolute/path
 /// relative/../path => <current-directory>/relative/../path
 ///
 /// @param path A path that is modified to be an absolute path.
 /// @returns errc::success if \a path has been made absolute, otherwise a
 ///          platform-specific error_code.
-std::error_code make_absolute(SmallVectorImpl<char> &path);
+LLVM_SUPPORT_ABI std::error_code make_absolute(SmallVectorImpl<char> &path);
 
 /// Create all the non-existent directories in path.
 ///
 /// @param path Directories to create.
 /// @returns errc::success if is_directory(path), otherwise a platform
 ///          specific error_code. If IgnoreExisting is false, also returns
 ///          error if the directory already existed.
-std::error_code create_directories(const Twine &path,
+LLVM_SUPPORT_ABI std::error_code create_directories(const Twine &path,
                                    bool IgnoreExisting = true,
                                    perms Perms = owner_all | group_all);
 
 /// Create the directory in path.
 ///
 /// @param path Directory to create.
 /// @returns errc::success if is_directory(path), otherwise a platform
 ///          specific error_code. If IgnoreExisting is false, also returns
 ///          error if the directory already existed.
-std::error_code create_directory(const Twine &path, bool IgnoreExisting = true,
+LLVM_SUPPORT_ABI std::error_code create_directory(const Twine &path, bool IgnoreExisting = true,
                                  perms Perms = owner_all | group_all);
 
 /// Create a link from \a from to \a to.
 ///
 /// The link may be a soft or a hard link, depending on the platform. The caller
 /// may not assume which one. Currently on windows it creates a hard link since
 /// soft links require extra privileges. On unix, it creates a soft link since
 /// hard links don't work on SMB file systems.
 ///
 /// @param to The path to hard link to.
 /// @param from The path to hard link from. This is created.
 /// @returns errc::success if the link was created, otherwise a platform
 /// specific error_code.
-std::error_code create_link(const Twine &to, const Twine &from);
+LLVM_SUPPORT_ABI std::error_code create_link(const Twine &to, const Twine &from);
 
 /// Create a hard link from \a from to \a to, or return an error.
 ///
 /// @param to The path to hard link to.
 /// @param from The path to hard link from. This is created.
 /// @returns errc::success if the link was created, otherwise a platform
 /// specific error_code.
-std::error_code create_hard_link(const Twine &to, const Twine &from);
+LLVM_SUPPORT_ABI std::error_code create_hard_link(const Twine &to, const Twine &from);
 
 /// Collapse all . and .. patterns, resolve all symlinks, and optionally
 ///        expand ~ expressions to the user's home directory.
 ///
 /// @param path The path to resolve.
 /// @param output The location to store the resolved path.
 /// @param expand_tilde If true, resolves ~ expressions to the user's home
 ///                     directory.
-std::error_code real_path(const Twine &path, SmallVectorImpl<char> &output,
+LLVM_SUPPORT_ABI std::error_code real_path(const Twine &path, SmallVectorImpl<char> &output,
                           bool expand_tilde = false);
 
 /// Expands ~ expressions to the user's home directory. On Unix ~user
 /// directories are resolved as well.
 ///
 /// @param path The path to resolve.
-void expand_tilde(const Twine &path, SmallVectorImpl<char> &output);
+LLVM_SUPPORT_ABI void expand_tilde(const Twine &path, SmallVectorImpl<char> &output);
 
 /// Get the current path.
 ///
 /// @param result Holds the current path on return.
 /// @returns errc::success if the current path has been stored in result,
 ///          otherwise a platform-specific error_code.
-std::error_code current_path(SmallVectorImpl<char> &result);
+LLVM_SUPPORT_ABI std::error_code current_path(SmallVectorImpl<char> &result);
 
 /// Set the current path.
 ///
 /// @param path The path to set.
 /// @returns errc::success if the current path was successfully set,
 ///          otherwise a platform-specific error_code.
-std::error_code set_current_path(const Twine &path);
+LLVM_SUPPORT_ABI std::error_code set_current_path(const Twine &path);
 
 /// Remove path. Equivalent to POSIX remove().
 ///
 /// @param path Input path.
 /// @returns errc::success if path has been removed or didn't exist, otherwise a
 ///          platform-specific error code. If IgnoreNonExisting is false, also
 ///          returns error if the file didn't exist.
-std::error_code remove(const Twine &path, bool IgnoreNonExisting = true);
+LLVM_SUPPORT_ABI std::error_code remove(const Twine &path, bool IgnoreNonExisting = true);
 
 /// Recursively delete a directory.
 ///
 /// @param path Input path.
 /// @returns errc::success if path has been removed or didn't exist, otherwise a
 ///          platform-specific error code.
-std::error_code remove_directories(const Twine &path, bool IgnoreErrors = true);
+LLVM_SUPPORT_ABI std::error_code remove_directories(const Twine &path, bool IgnoreErrors = true);
 
 /// Rename \a from to \a to.
 ///
 /// Files are renamed as if by POSIX rename(), except that on Windows there may
 /// be a short interval of time during which the destination file does not
 /// exist.
 ///
 /// @param from The path to rename from.
 /// @param to The path to rename to. This is created.
-std::error_code rename(const Twine &from, const Twine &to);
+LLVM_SUPPORT_ABI std::error_code rename(const Twine &from, const Twine &to);
 
 /// Copy the contents of \a From to \a To.
 ///
 /// @param From The path to copy from.
 /// @param To The path to copy to. This is created.
-std::error_code copy_file(const Twine &From, const Twine &To);
+LLVM_SUPPORT_ABI std::error_code copy_file(const Twine &From, const Twine &To);
 
 /// Copy the contents of \a From to \a To.
 ///
 /// @param From The path to copy from.
 /// @param ToFD The open file descriptor of the destination file.
-std::error_code copy_file(const Twine &From, int ToFD);
+LLVM_SUPPORT_ABI std::error_code copy_file(const Twine &From, int ToFD);
 
 /// Resize path to size. File is resized as if by POSIX truncate().
 ///
 /// @param FD Input file descriptor.
 /// @param Size Size to resize to.
 /// @returns errc::success if \a path has been resized to \a size, otherwise a
 ///          platform-specific error_code.
-std::error_code resize_file(int FD, uint64_t Size);
+LLVM_SUPPORT_ABI std::error_code resize_file(int FD, uint64_t Size);
 
 /// Resize \p FD to \p Size before mapping \a mapped_file_region::readwrite. On
 /// non-Windows, this calls \a resize_file(). On Windows, this is a no-op,
 /// since the subsequent mapping (via \c CreateFileMapping) automatically
 /// extends the file.
 inline std::error_code resize_file_before_mapping_readwrite(int FD,
                                                             uint64_t Size) {
 #ifdef _WIN32
   (void)FD;
   (void)Size;
   return std::error_code();
 #else
   return resize_file(FD, Size);
 #endif
 }
 
 /// Compute an MD5 hash of a file's contents.
 ///
 /// @param FD Input file descriptor.
 /// @returns An MD5Result with the hash computed, if successful, otherwise a
 ///          std::error_code.
-ErrorOr<MD5::MD5Result> md5_contents(int FD);
+LLVM_SUPPORT_ABI ErrorOr<MD5::MD5Result> md5_contents(int FD);
 
 /// Version of compute_md5 that doesn't require an open file descriptor.
-ErrorOr<MD5::MD5Result> md5_contents(const Twine &Path);
+LLVM_SUPPORT_ABI ErrorOr<MD5::MD5Result> md5_contents(const Twine &Path);
 
 /// @}
 /// @name Physical Observers
 /// @{
 
 /// Does file exist?
 ///
 /// @param status A basic_file_status previously returned from stat.
 /// @returns True if the file represented by status exists, false if it does
 ///          not.
-bool exists(const basic_file_status &status);
+LLVM_SUPPORT_ABI bool exists(const basic_file_status &status);
 
 enum class AccessMode { Exist, Write, Execute };
 
 /// Can the file be accessed?
 ///
 /// @param Path Input path.
 /// @returns errc::success if the path can be accessed, otherwise a
 ///          platform-specific error_code.
-std::error_code access(const Twine &Path, AccessMode Mode);
+LLVM_SUPPORT_ABI std::error_code access(const Twine &Path, AccessMode Mode);
 
 /// Does file exist?
 ///
 /// @param Path Input path.
 /// @returns True if it exists, false otherwise.
 inline bool exists(const Twine &Path) {
   return !access(Path, AccessMode::Exist);
 }
 
 /// Can we execute this file?
 ///
 /// @param Path Input path.
 /// @returns True if we can execute it, false otherwise.
-bool can_execute(const Twine &Path);
+LLVM_SUPPORT_ABI bool can_execute(const Twine &Path);
 
 /// Can we write this file?
 ///
 /// @param Path Input path.
 /// @returns True if we can write to it, false otherwise.
 inline bool can_write(const Twine &Path) {
   return !access(Path, AccessMode::Write);
 }
 
 /// Do file_status's represent the same thing?
 ///
 /// @param A Input file_status.
 /// @param B Input file_status.
 ///
 /// assert(status_known(A) || status_known(B));
 ///
 /// @returns True if A and B both represent the same file system entity, false
 ///          otherwise.
-bool equivalent(file_status A, file_status B);
+LLVM_SUPPORT_ABI bool equivalent(file_status A, file_status B);
 
 /// Do paths represent the same thing?
 ///
 /// assert(status_known(A) || status_known(B));
 ///
 /// @param A Input path A.
 /// @param B Input path B.
 /// @param result Set to true if stat(A) and stat(B) have the same device and
 ///               inode (or equivalent).
 /// @returns errc::success if result has been successfully set, otherwise a
 ///          platform-specific error_code.
-std::error_code equivalent(const Twine &A, const Twine &B, bool &result);
+LLVM_SUPPORT_ABI std::error_code equivalent(const Twine &A, const Twine &B, bool &result);
 
 /// Simpler version of equivalent for clients that don't need to
 ///        differentiate between an error and false.
 inline bool equivalent(const Twine &A, const Twine &B) {
   bool result;
   return !equivalent(A, B, result) && result;
 }
 
 /// Is the file mounted on a local filesystem?
 ///
 /// @param path Input path.
 /// @param result Set to true if \a path is on fixed media such as a hard disk,
 ///               false if it is not.
 /// @returns errc::success if result has been successfully set, otherwise a
 ///          platform specific error_code.
-std::error_code is_local(const Twine &path, bool &result);
+LLVM_SUPPORT_ABI std::error_code is_local(const Twine &path, bool &result);
 
 /// Version of is_local accepting an open file descriptor.
-std::error_code is_local(int FD, bool &result);
+LLVM_SUPPORT_ABI std::error_code is_local(int FD, bool &result);
 
 /// Simpler version of is_local for clients that don't need to
 ///        differentiate between an error and false.
 inline bool is_local(const Twine &Path) {
   bool Result;
   return !is_local(Path, Result) && Result;
 }
 

 ///               of the link itself, or of the target.
 /// @returns A value from the file_type enumeration indicating the type of file.
 file_type get_file_type(const Twine &Path, bool Follow = true);
 
 /// Does status represent a directory?
 ///
 /// @param status A basic_file_status previously returned from status.
 /// @returns status.type() == file_type::directory_file.
-bool is_directory(const basic_file_status &status);
+LLVM_SUPPORT_ABI bool is_directory(const basic_file_status &status);
 
 /// Is path a directory?
 ///
 /// @param path Input path.
 /// @param result Set to true if \a path is a directory (after following
 ///               symlinks, false if it is not. Undefined otherwise.
 /// @returns errc::success if result has been successfully set, otherwise a
 ///          platform-specific error_code.
-std::error_code is_directory(const Twine &path, bool &result);
+LLVM_SUPPORT_ABI std::error_code is_directory(const Twine &path, bool &result);
 
 /// Simpler version of is_directory for clients that don't need to
 ///        differentiate between an error and false.
 inline bool is_directory(const Twine &Path) {
   bool Result;
   return !is_directory(Path, Result) && Result;
 }
 
 /// Does status represent a regular file?
 ///
 /// @param status A basic_file_status previously returned from status.
 /// @returns status_known(status) && status.type() == file_type::regular_file.
-bool is_regular_file(const basic_file_status &status);
+LLVM_SUPPORT_ABI bool is_regular_file(const basic_file_status &status);
 
 /// Is path a regular file?
 ///
 /// @param path Input path.
 /// @param result Set to true if \a path is a regular file (after following
 ///               symlinks), false if it is not. Undefined otherwise.
 /// @returns errc::success if result has been successfully set, otherwise a
 ///          platform-specific error_code.
-std::error_code is_regular_file(const Twine &path, bool &result);
+LLVM_SUPPORT_ABI std::error_code is_regular_file(const Twine &path, bool &result);
 
 /// Simpler version of is_regular_file for clients that don't need to
 ///        differentiate between an error and false.
 inline bool is_regular_file(const Twine &Path) {
   bool Result;
   if (is_regular_file(Path, Result))
     return false;
   return Result;

 /// Get file status as if by POSIX stat().
 ///
 /// @param path Input path.
 /// @param result Set to the file status.
 /// @param follow When true, follows symlinks.  Otherwise, the symlink itself is
 ///               statted.
 /// @returns errc::success if result has been successfully set, otherwise a
 ///          platform-specific error_code.
-std::error_code status(const Twine &path, file_status &result,
+LLVM_SUPPORT_ABI std::error_code status(const Twine &path, file_status &result,
                        bool follow = true);
 
 /// A version for when a file descriptor is already available.
-std::error_code status(int FD, file_status &Result);
+LLVM_SUPPORT_ABI std::error_code status(int FD, file_status &Result);
 
 #ifdef _WIN32
 /// A version for when a file descriptor is already available.
-std::error_code status(file_t FD, file_status &Result);
+LLVM_SUPPORT_ABI std::error_code status(file_t FD, file_status &Result);
 #endif
 
 /// Get file creation mode mask of the process.
 ///
 /// @returns Mask reported by umask(2)
 /// @note There is no umask on Windows. This function returns 0 always
 ///       on Windows. This function does not return an error_code because
 ///       umask(2) never fails. It is not thread safe.
-unsigned getUmask();
+LLVM_SUPPORT_ABI unsigned getUmask();
 
 /// Set file permissions.
 ///
 /// @param Path File to set permissions on.
 /// @param Permissions New file permissions.
 /// @returns errc::success if the permissions were successfully set, otherwise
 ///          a platform-specific error_code.
 /// @note On Windows, all permissions except *_write are ignored. Using any of
 ///       owner_write, group_write, or all_write will make the file writable.
 ///       Otherwise, the file will be marked as read-only.
-std::error_code setPermissions(const Twine &Path, perms Permissions);
+LLVM_SUPPORT_ABI std::error_code setPermissions(const Twine &Path, perms Permissions);
 
 /// Vesion of setPermissions accepting a file descriptor.
 /// TODO Delete the path based overload once we implement the FD based overload
 /// on Windows.
-std::error_code setPermissions(int FD, perms Permissions);
+LLVM_SUPPORT_ABI std::error_code setPermissions(int FD, perms Permissions);
 
 /// Get file permissions.
 ///
 /// @param Path File to get permissions from.
 /// @returns the permissions if they were successfully retrieved, otherwise a
 ///          platform-specific error_code.
 /// @note On Windows, if the file does not have the FILE_ATTRIBUTE_READONLY
 ///       attribute, all_all will be returned. Otherwise, all_read | all_exe
 ///       will be returned.
-ErrorOr<perms> getPermissions(const Twine &Path);
+LLVM_SUPPORT_ABI ErrorOr<perms> getPermissions(const Twine &Path);
 
 /// Get file size.
 ///
 /// @param Path Input path.
 /// @param Result Set to the size of the file in \a Path.
 /// @returns errc::success if result has been successfully set, otherwise a
 ///          platform-specific error_code.
 inline std::error_code file_size(const Twine &Path, uint64_t &Result) {
   file_status Status;
   std::error_code EC = status(Path, Status);
   if (EC)
     return EC;
   Result = Status.getSize();
   return std::error_code();
 }
 
 /// Set the file modification and access time.
 ///
 /// @returns errc::success if the file times were successfully set, otherwise a
 ///          platform-specific error_code or errc::function_not_supported on
 ///          platforms where the functionality isn't available.
-std::error_code setLastAccessAndModificationTime(int FD, TimePoint<> AccessTime,
+LLVM_SUPPORT_ABI std::error_code setLastAccessAndModificationTime(int FD, TimePoint<> AccessTime,
                                                  TimePoint<> ModificationTime);
 
 /// Simpler version that sets both file modification and access time to the same
 /// time.
 inline std::error_code setLastAccessAndModificationTime(int FD,
                                                         TimePoint<> Time) {
   return setLastAccessAndModificationTime(FD, Time, Time);
 }

 ///
 /// @param Model Name to base unique path off of.
 /// @param ResultFD Set to the opened file's file descriptor.
 /// @param ResultPath Set to the opened file's absolute path.
 /// @param Flags Set to the opened file's flags.
 /// @param Mode Set to the opened file's permissions.
 /// @returns errc::success if Result{FD,Path} have been successfully set,
 ///          otherwise a platform-specific error_code.
-std::error_code createUniqueFile(const Twine &Model, int &ResultFD,
+LLVM_SUPPORT_ABI std::error_code createUniqueFile(const Twine &Model, int &ResultFD,
                                  SmallVectorImpl<char> &ResultPath,
                                  OpenFlags Flags = OF_None,
                                  unsigned Mode = all_read | all_write);
 
 /// Simpler version for clients that don't want an open file. An empty
 /// file will still be created.
-std::error_code createUniqueFile(const Twine &Model,
+LLVM_SUPPORT_ABI std::error_code createUniqueFile(const Twine &Model,
                                  SmallVectorImpl<char> &ResultPath,
                                  unsigned Mode = all_read | all_write);
 
 /// Represents a temporary file.
 ///
 /// The temporary file must be eventually discarded or given a final name and
 /// kept.
 ///
 /// The destructor doesn't implicitly discard because there is no way to
 /// properly handle errors in a destructor.
-class TempFile {
+class LLVM_SUPPORT_ABI TempFile {
   bool Done = false;
   TempFile(StringRef Name, int FD);
 
 public:
   /// This creates a temporary file with createUniqueFile and schedules it for
   /// deletion with sys::RemoveFileOnSignal.
   static Expected<TempFile> create(const Twine &Model,
                                    unsigned Mode = all_read | all_write,

 /// Create a file in the system temporary directory.
 ///
 /// The filename is of the form prefix-random_chars.suffix. Since the directory
 /// is not know to the caller, Prefix and Suffix cannot have path separators.
 /// The files are created with mode 0600.
 ///
 /// This should be used for things like a temporary .s that is removed after
 /// running the assembler.
-std::error_code createTemporaryFile(const Twine &Prefix, StringRef Suffix,
+LLVM_SUPPORT_ABI std::error_code createTemporaryFile(const Twine &Prefix, StringRef Suffix,
                                     int &ResultFD,
                                     SmallVectorImpl<char> &ResultPath,
                                     OpenFlags Flags = OF_None);
 
 /// Simpler version for clients that don't want an open file. An empty
 /// file will still be created.
-std::error_code createTemporaryFile(const Twine &Prefix, StringRef Suffix,
+LLVM_SUPPORT_ABI std::error_code createTemporaryFile(const Twine &Prefix, StringRef Suffix,
                                     SmallVectorImpl<char> &ResultPath,
                                     OpenFlags Flags = OF_None);
 
-std::error_code createUniqueDirectory(const Twine &Prefix,
+LLVM_SUPPORT_ABI std::error_code createUniqueDirectory(const Twine &Prefix,
                                       SmallVectorImpl<char> &ResultPath);
 
 /// Get a unique name, not currently exisiting in the filesystem. Subject
 /// to race conditions, prefer to use createUniqueFile instead.
 ///
 /// Similar to createUniqueFile, but instead of creating a file only
 /// checks if it exists. This function is subject to race conditions, if you
 /// want to use the returned name to actually create a file, use

 ///                 is stored in this location. Otherwise, this is set to -1.
 /// @param Disp Value specifying the existing-file behavior.
 /// @param Access Value specifying whether to open the file in read, write, or
 ///               read-write mode.
 /// @param Flags Additional flags.
 /// @param Mode The access permissions of the file, represented in octal.
 /// @returns errc::success if \a Name has been opened, otherwise a
 ///          platform-specific error_code.
-std::error_code openFile(const Twine &Name, int &ResultFD,
+LLVM_SUPPORT_ABI std::error_code openFile(const Twine &Name, int &ResultFD,
                          CreationDisposition Disp, FileAccess Access,
                          OpenFlags Flags, unsigned Mode = 0666);
 
 /// @brief Opens a file with the specified creation disposition, access mode,
 /// and flags and returns a platform-specific file object.
 ///
 /// The caller is responsible for closing the file object once they are
 /// finished with it.
 ///
 /// @param Name The path of the file to open, relative or absolute.
 /// @param Disp Value specifying the existing-file behavior.
 /// @param Access Value specifying whether to open the file in read, write, or
 ///               read-write mode.
 /// @param Flags Additional flags.
 /// @param Mode The access permissions of the file, represented in octal.
 /// @returns errc::success if \a Name has been opened, otherwise a
 ///          platform-specific error_code.
-Expected<file_t> openNativeFile(const Twine &Name, CreationDisposition Disp,
+LLVM_SUPPORT_ABI Expected<file_t> openNativeFile(const Twine &Name, CreationDisposition Disp,
                                 FileAccess Access, OpenFlags Flags,
                                 unsigned Mode = 0666);
 
 /// Converts from a Posix file descriptor number to a native file handle.
 /// On Windows, this retreives the underlying handle. On non-Windows, this is a
 /// no-op.
-file_t convertFDToNativeFile(int FD);
-
-#ifndef _WIN32
+#if defined(_WIN32)
+LLVM_SUPPORT_ABI file_t convertFDToNativeFile(int FD);
+#else
 inline file_t convertFDToNativeFile(int FD) { return FD; }
 #endif
 
 /// Return an open handle to standard in. On Unix, this is typically FD 0.
 /// Returns kInvalidFile when the stream is closed.
 file_t getStdinHandle();
 
 /// Return an open handle to standard out. On Unix, this is typically FD 1.
 /// Returns kInvalidFile when the stream is closed.
 file_t getStdoutHandle();
 
 /// Return an open handle to standard error. On Unix, this is typically FD 2.
 /// Returns kInvalidFile when the stream is closed.
 file_t getStderrHandle();
 
 /// Reads \p Buf.size() bytes from \p FileHandle into \p Buf. Returns the number
 /// of bytes actually read. On Unix, this is equivalent to `return ::read(FD,
 /// Buf.data(), Buf.size())`, with error reporting. Returns 0 when reaching EOF.
 ///
 /// @param FileHandle File to read from.
 /// @param Buf Buffer to read into.
 /// @returns The number of bytes read, or error.
-Expected<size_t> readNativeFile(file_t FileHandle, MutableArrayRef<char> Buf);
+LLVM_SUPPORT_ABI Expected<size_t> readNativeFile(file_t FileHandle, MutableArrayRef<char> Buf);
 
 /// Reads \p Buf.size() bytes from \p FileHandle at offset \p Offset into \p
 /// Buf. If 'pread' is available, this will use that, otherwise it will use
 /// 'lseek'. Returns the number of bytes actually read. Returns 0 when reaching
 /// EOF.
 ///
 /// @param FileHandle File to read from.
 /// @param Buf Buffer to read into.
 /// @param Offset Offset into the file at which the read should occur.
 /// @returns The number of bytes read, or error.
-Expected<size_t> readNativeFileSlice(file_t FileHandle,
+LLVM_SUPPORT_ABI Expected<size_t> readNativeFileSlice(file_t FileHandle,
                                      MutableArrayRef<char> Buf,
                                      uint64_t Offset);
 
 /// @brief Opens the file with the given name in a write-only or read-write
 /// mode, returning its open file descriptor. If the file does not exist, it
 /// is created.
 ///
 /// The caller is responsible for closing the file descriptor once they are
 /// finished with it.

 /// @param Name The path of the file to open, relative or absolute.
 /// @param ResultFD If the file could be opened successfully, its descriptor
 ///                 is stored in this location. Otherwise, this is set to -1.
 /// @param RealPath If nonnull, extra work is done to determine the real path
 ///                 of the opened file, and that path is stored in this
 ///                 location.
 /// @returns errc::success if \a Name has been opened, otherwise a
 ///          platform-specific error_code.
-std::error_code openFileForRead(const Twine &Name, int &ResultFD,
+LLVM_SUPPORT_ABI std::error_code openFileForRead(const Twine &Name, int &ResultFD,
                                 OpenFlags Flags = OF_None,
                                 SmallVectorImpl<char> *RealPath = nullptr);
 
 /// @brief Opens the file with the given name in a read-only mode, returning
 /// its open file descriptor.
 ///
 /// The caller is responsible for closing the freeing the file once they are
 /// finished with it.
 ///
 /// @param Name The path of the file to open, relative or absolute.
 /// @param RealPath If nonnull, extra work is done to determine the real path
 ///                 of the opened file, and that path is stored in this
 ///                 location.
 /// @returns a platform-specific file descriptor if \a Name has been opened,
 ///          otherwise an error object.
-Expected<file_t>
+LLVM_SUPPORT_ABI Expected<file_t>
 openNativeFileForRead(const Twine &Name, OpenFlags Flags = OF_None,
                       SmallVectorImpl<char> *RealPath = nullptr);
 
 /// Try to locks the file during the specified time.
 ///
 /// This function implements advisory locking on entire file. If it returns
 /// <em>errc::success</em>, the file is locked by the calling process. Until the
 /// process unlocks the file by calling \a unlockFile, all attempts to lock the
 /// same file will fail/block. The process that locked the file may assume that
 /// none of other processes read or write this file, provided that all processes
 /// lock the file prior to accessing its content.
 ///
 /// @param FD      The descriptor representing the file to lock.
 /// @param Timeout Time in milliseconds that the process should wait before
 ///                reporting lock failure. Zero value means try to get lock only
 ///                once.
 /// @returns errc::success if lock is successfully obtained,
 /// errc::no_lock_available if the file cannot be locked, or platform-specific
 /// error_code otherwise.
 ///
 /// @note Care should be taken when using this function in a multithreaded
 /// context, as it may not prevent other threads in the same process from
 /// obtaining a lock on the same file, even if they are using a different file
 /// descriptor.
-std::error_code
+LLVM_SUPPORT_ABI std::error_code
 tryLockFile(int FD,
             std::chrono::milliseconds Timeout = std::chrono::milliseconds(0));
 
 /// Lock the file.
 ///
 /// This function acts as @ref tryLockFile but it waits infinitely.
 std::error_code lockFile(int FD);
 
 /// Unlock the file.
 ///
 /// @param FD The descriptor representing the file to unlock.
 /// @returns errc::success if lock is successfully released or platform-specific
 /// error_code otherwise.
-std::error_code unlockFile(int FD);
+LLVM_SUPPORT_ABI std::error_code unlockFile(int FD);
 
 /// @brief Close the file object.  This should be used instead of ::close for
 /// portability. On error, the caller should assume the file is closed, as is
 /// the case for Process::SafelyCloseFileDescriptor
 ///
 /// @param F On input, this is the file to close.  On output, the file is
 /// set to kInvalidFile.
 ///
 /// @returns An error code if closing the file failed. Typically, an error here
 /// means that the filesystem may have failed to perform some buffered writes.
-std::error_code closeFile(file_t &F);
+LLVM_SUPPORT_ABI std::error_code closeFile(file_t &F);
 
 #ifdef LLVM_ON_UNIX
 /// @brief Change ownership of a file.
 ///
 /// @param Owner The owner of the file to change to.
 /// @param Group The group of the file to change to.
 /// @returns errc::success if successfully updated file ownership, otherwise an
 ///          error code is returned.
if (FD != -1) {
       std::error_code Result = unlockFile(FD);
       FD = -1;
       return Result;
     }
     return std::error_code();
   }
 };
 
-std::error_code getUniqueID(const Twine Path, UniqueID &Result);
+LLVM_SUPPORT_ABI std::error_code getUniqueID(const Twine Path, UniqueID &Result);
 
 /// Get disk space usage information.
 ///
 /// Note: Users must be careful about "Time Of Check, Time Of Use" kind of bug.
 /// Note: Windows reports results according to the quota allocated to the user.
 ///
 /// @param Path Input path.
 /// @returns a space_info structure filled with the capacity, free, and
 /// available space on the device \a Path is on. A platform specific error_code
 /// is returned on error.
 ErrorOr<space_info> disk_space(const Twine &Path);
 
 /// This class represents a memory mapped file. It is based on
 /// boost::iostreams::mapped_file.
-class mapped_file_region {
+class LLVM_SUPPORT_ABI mapped_file_region {
 public:
   enum mapmode {
     readonly, ///< May only access map via const_data as read only.
     readwrite, ///< May access map via data and modify it. Written to path.
     priv ///< May modify via data, but changes are lost on destruction.
   };
 
 private:
public:
 
   /// \returns The minimum alignment offset must be.
   static int alignment();
 };
 
 /// Return the path to the main executable, given the value of argv[0] from
 /// program startup and the address of main itself. In extremis, this function
 /// may fail and return an empty path.
-std::string getMainExecutable(const char *argv0, void *MainExecAddr);
+LLVM_SUPPORT_ABI std::string getMainExecutable(const char *argv0, void *MainExecAddr);
 
 /// @}
 /// @name Iterators
 /// @{
 
 /// directory_entry - A single entry in a directory.
-class directory_entry {
+class LLVM_SUPPORT_ABI directory_entry {
   // FIXME: different platforms make different information available "for free"
   // when traversing a directory. The design of this class wraps most of the
   // information in basic_file_status, so on platforms where we can't populate
   // that whole structure, callers end up paying for a stat().
   // std::filesystem::directory_entry may be a better model.
   std::string Path;
   file_type Type = file_type::type_unknown; // Most platforms can provide this.
   bool FollowSymlinks = true;               // Affects the behavior of status().
public:
   bool operator> (const directory_entry& RHS) const;
   bool operator>=(const directory_entry& RHS) const;
 };
 
 namespace detail {
 
   struct DirIterState;
 
-  std::error_code directory_iterator_construct(DirIterState &, StringRef, bool);
-  std::error_code directory_iterator_increment(DirIterState &);
-  std::error_code directory_iterator_destruct(DirIterState &);
+  LLVM_SUPPORT_ABI std::error_code directory_iterator_construct(DirIterState &, StringRef, bool);
+  LLVM_SUPPORT_ABI std::error_code directory_iterator_increment(DirIterState &);
+  LLVM_SUPPORT_ABI std::error_code directory_iterator_destruct(DirIterState &);
 
   /// Keeps state for the directory_iterator.
   struct DirIterState {
     ~DirIterState() {
       directory_iterator_destruct(*this);
     }
 
     intptr_t IterationHandle = 0;

llvm/include/llvm/Support/FileUtilities.h

 
 #ifndef LLVM_SUPPORT_FILEUTILITIES_H
 #define LLVM_SUPPORT_FILEUTILITIES_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Errc.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/FileSystem.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/Path.h"
 
 namespace llvm {
 
   /// DiffFilesWithTolerance - Compare the two files specified, returning 0 if
   /// the files match, 1 if they are different, and 2 if there is a file error.
   /// This function allows you to specify an absolute and relative FP error that
   /// is allowed to exist.  If you specify a string to fill in for the error
   /// option, it will set the string to an error message if an error occurs, or
   /// if the files are different.
   ///
-  int DiffFilesWithTolerance(StringRef FileA,
+  LLVM_SUPPORT_ABI int DiffFilesWithTolerance(StringRef FileA,
                              StringRef FileB,
                              double AbsTol, double RelTol,
                              std::string *Error = nullptr);
 
 
   /// FileRemover - This class is a simple object meant to be stack allocated.
   /// If an exception is thrown from a region, the object removes the filename
   /// specified (if deleteIt is true).
   ///
   class FileRemover {
     SmallString<128> Filename;
     bool DeleteIt;
   public:
     FileRemover() : DeleteIt(false) {}
 
     explicit FileRemover(const Twine& filename, bool deleteIt = true)
       : DeleteIt(deleteIt) {
       filename.toVector(Filename);
     }
 
     ~FileRemover() {
       if (DeleteIt) {
         // Ignore problems deleting the file.
         sys::fs::remove(Filename);
       }
     }
 
     /// setFile - Give ownership of the file to the FileRemover so it will
     /// be removed when the object is destroyed.  If the FileRemover already
     /// had ownership of a file, remove it first.
     void setFile(const Twine& filename, bool deleteIt = true) {
       if (DeleteIt) {
         // Ignore problems deleting the file.
         sys::fs::remove(Filename);
       }
 
       Filename.clear();
       filename.toVector(Filename);
       DeleteIt = deleteIt;
     }
 
     /// releaseFile - Take ownership of the file away from the FileRemover so it
     /// will not be removed when the object is destroyed.
     void releaseFile() { DeleteIt = false; }
   };
 
   enum class atomic_write_error {
     failed_to_create_uniq_file = 0,
     output_stream_error,
     failed_to_rename_temp_file
   };
 
-  class AtomicFileWriteError : public llvm::ErrorInfo<AtomicFileWriteError> {
+  class LLVM_SUPPORT_ABI AtomicFileWriteError : public llvm::ErrorInfo<AtomicFileWriteError> {
   public:
     AtomicFileWriteError(atomic_write_error Error) : Error(Error) {}
 
     void log(raw_ostream &OS) const override;
 
     const atomic_write_error Error;
     static char ID;
 
   private:
     // Users are not expected to use error_code.
     std::error_code convertToErrorCode() const override {
       return llvm::inconvertibleErrorCode();
     }
   };
 
   // atomic_write_error + whatever the Writer can return
 
   /// Creates a unique file with name according to the given \p TempPathModel,
   /// writes content of \p Buffer to the file and renames it to \p FinalPath.
   ///
   /// \returns \c AtomicFileWriteError in case of error.
-  llvm::Error writeFileAtomically(StringRef TempPathModel, StringRef FinalPath,
+  LLVM_SUPPORT_ABI llvm::Error writeFileAtomically(StringRef TempPathModel, StringRef FinalPath,
                                   StringRef Buffer);
 
-  llvm::Error
+  LLVM_SUPPORT_ABI llvm::Error
   writeFileAtomically(StringRef TempPathModel, StringRef FinalPath,
                       std::function<llvm::Error(llvm::raw_ostream &)> Writer);
 } // End llvm namespace
 
 #endif

llvm/include/llvm/Support/Format.h

 
 #ifndef LLVM_SUPPORT_FORMAT_H
 #define LLVM_SUPPORT_FORMAT_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <cstdio>
 #include <tuple>
 #include <utility>
 
 namespace llvm {
 
 /// This is a helper class used for handling formatted output.  It is the
 /// abstract base class of a templated derived class.
-class format_object_base {
+class LLVM_SUPPORT_ABI format_object_base {
 protected:
   const char *Fmt;
   ~format_object_base() = default; // Disallow polymorphic deletion.
   format_object_base(const format_object_base &) = default;
   virtual void home(); // Out of line virtual method.
 
   /// Call snprintf() for this object, on the given buffer and size.
   virtual int snprint(char *Buffer, unsigned BufferSize) const = 0;

llvm/include/llvm/Support/FormatProviders.h

 #ifndef LLVM_SUPPORT_FORMATPROVIDERS_H
 #define LLVM_SUPPORT_FORMATPROVIDERS_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/FormatVariadicDetails.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/NativeFormatting.h"
 
 #include <type_traits>
 #include <vector>
 
 namespace llvm {
 namespace detail {
 template <typename T>

llvm/include/llvm/Support/FormatVariadic.h

 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/FormatCommon.h"
 #include "llvm/Support/FormatProviders.h"
 #include "llvm/Support/FormatVariadicDetails.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <cstddef>
 #include <string>
 #include <tuple>
 #include <utility>
 #include <vector>
 
 namespace llvm {
 
struct ReplacementItem {
   StringRef Spec;
   size_t Index = 0;
   size_t Align = 0;
   AlignStyle Where = AlignStyle::Right;
   char Pad = 0;
   StringRef Options;
 };
 
-class formatv_object_base {
+class LLVM_SUPPORT_ABI formatv_object_base {
 protected:
   StringRef Fmt;
   ArrayRef<detail::format_adapter *> Adapters;
 
   static bool consumeFieldLayout(StringRef &Spec, AlignStyle &Where,
                                  size_t &Align, char &Pad);
 
   static std::pair<ReplacementItem, StringRef>

llvm/include/llvm/Support/FormatVariadicDetails.h

 //===- FormatVariadicDetails.h - Helpers for FormatVariadic.h ----*- C++-*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_FORMATVARIADICDETAILS_H
 #define LLVM_SUPPORT_FORMATVARIADICDETAILS_H
 
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 #include <type_traits>
 
 namespace llvm {
 template <typename T, typename Enable = void> struct format_provider {};
 class Error;
 
 namespace detail {
-class format_adapter {
+class LLVM_SUPPORT_ABI format_adapter {
   virtual void anchor();
 
 protected:
   virtual ~format_adapter() {}
 
 public:
   virtual void format(raw_ostream &S, StringRef Options) = 0;
 };

llvm/include/llvm/Support/FormattedStream.h

 // things like pretty-print comments.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_FORMATTEDSTREAM_H
 #define LLVM_SUPPORT_FORMATTEDSTREAM_H
 
 #include "llvm/ADT/SmallString.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <utility>
 
 namespace llvm {
 
 /// formatted_raw_ostream - A raw_ostream that wraps another one and keeps track
 /// of line and column position, allowing padding out to specific column
 /// boundaries and querying the number of lines written to the stream. This
 /// assumes that the contents of the stream is valid UTF-8 encoded text. This
 /// doesn't attempt to handle everything Unicode can do (combining characters,
 /// right-to-left markers, etc), but should cover the cases likely to appear in
 /// source code or diagnostic messages.
-class formatted_raw_ostream : public raw_ostream {
+class LLVM_SUPPORT_ABI formatted_raw_ostream : public raw_ostream {
   /// TheStream - The real stream we output to. We set it to be
   /// unbuffered, since we're already doing our own buffering.
   ///
   raw_ostream *TheStream;
 
   /// Position - The current output column and line of the data that's
   /// been flushed and the portion of the buffer that's been
   /// scanned.  The line and column scheme is zero-based.
if (size_t BufferSize = GetBufferSize())
       TheStream->SetBufferSize(BufferSize);
     else
       TheStream->SetUnbuffered();
   }
 };
 
 /// fouts() - This returns a reference to a formatted_raw_ostream for
 /// standard output.  Use it like: fouts() << "foo" << "bar";
-formatted_raw_ostream &fouts();
+LLVM_SUPPORT_ABI formatted_raw_ostream &fouts();
 
 /// ferrs() - This returns a reference to a formatted_raw_ostream for
 /// standard error.  Use it like: ferrs() << "foo" << "bar";
 formatted_raw_ostream &ferrs();
 
 /// fdbgs() - This returns a reference to a formatted_raw_ostream for
 /// debug output.  Use it like: fdbgs() << "foo" << "bar";
 formatted_raw_ostream &fdbgs();
 
 } // end llvm namespace
 
 
 #endif

llvm/include/llvm/Support/GlobPattern.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_GLOBPATTERN_H
 #define LLVM_SUPPORT_GLOBPATTERN_H
 
 #include "llvm/ADT/BitVector.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <vector>
 
 // This class represents a glob pattern. Supported metacharacters
 // are "*", "?", "\", "[<chars>]", "[^<chars>]", and "[!<chars>]".
 namespace llvm {
 
 template <typename T> class ArrayRef;
 class StringRef;
 
-class GlobPattern {
+class LLVM_SUPPORT_ABI GlobPattern {
 public:
   static Expected<GlobPattern> create(StringRef Pat);
   bool match(StringRef S) const;
 
   // Returns true for glob pattern "*". Can be used to avoid expensive
   // preparation/acquisition of the input for match().
   bool isTrivialMatchAll() const {
     if (Prefix && Prefix->empty()) {

llvm/include/llvm/Support/GraphWriter.h

 #ifndef LLVM_SUPPORT_GRAPHWRITER_H
 #define LLVM_SUPPORT_GRAPHWRITER_H
 
 #include "llvm/ADT/GraphTraits.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/DOTGraphTraits.h"
 #include "llvm/Support/FileSystem.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <algorithm>
 #include <cstddef>
 #include <iterator>
 #include <string>
 #include <type_traits>
 #include <vector>
 
 namespace llvm {
 
 namespace DOT {  // Private functions...
 
-std::string EscapeString(const std::string &Label);
+LLVM_SUPPORT_ABI std::string EscapeString(const std::string &Label);
 
 /// Get a color string for this node number. Simply round-robin selects
 /// from a reasonable number of colors.
 StringRef getColorString(unsigned NodeNumber);
 
 } // end namespace DOT
 
 namespace GraphProgram {
 
 enum Name {
   DOT,
   FDP,
   NEATO,
   TWOPI,
   CIRCO
 };
 
 } // end namespace GraphProgram
 
-bool DisplayGraph(StringRef Filename, bool wait = true,
+LLVM_SUPPORT_ABI bool DisplayGraph(StringRef Filename, bool wait = true,
                   GraphProgram::Name program = GraphProgram::DOT);
 
 template<typename GraphType>
 class GraphWriter {
   raw_ostream &O;
   const GraphType &G;
 
   using DOTTraits = DOTGraphTraits<GraphType>;
raw_ostream &WriteGraph(raw_ostream &O, const GraphType &G,
   GraphWriter<GraphType> W(O, G, ShortNames);
 
   // Emit the graph.
   W.writeGraph(Title.str());
 
   return O;
 }
 
-std::string createGraphFilename(const Twine &Name, int &FD);
+LLVM_SUPPORT_ABI std::string createGraphFilename(const Twine &Name, int &FD);
 
 /// Writes graph into a provided @c Filename.
 /// If @c Filename is empty, generates a random one.
 /// \return The resulting filename, or an empty string if writing
 /// failed.
 template <typename GraphType>
 std::string WriteGraph(const GraphType &G, const Twine &Name,
                        bool ShortNames = false,

llvm/include/llvm/Support/Host.h

 //===- llvm/Support/Host.h - Host machine characteristics --------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // Methods for querying the nature of the host machine.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_HOST_H
 #define LLVM_SUPPORT_HOST_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 
 namespace llvm {
 class MallocAllocator;
 class StringRef;
 template <typename ValueTy, typename AllocatorTy> class StringMap;
 
 namespace sys {
 
   /// getDefaultTargetTriple() - Return the default target triple the compiler
   /// has been configured to produce code for.
   ///
   /// The target triple is a string in the format of:
   ///   CPU_TYPE-VENDOR-OPERATING_SYSTEM
   /// or
   ///   CPU_TYPE-VENDOR-KERNEL-OPERATING_SYSTEM
-  std::string getDefaultTargetTriple();
+  LLVM_SUPPORT_ABI std::string getDefaultTargetTriple();
 
   /// getProcessTriple() - Return an appropriate target triple for generating
   /// code to be loaded into the current process, e.g. when using the JIT.
-  std::string getProcessTriple();
+  LLVM_SUPPORT_ABI std::string getProcessTriple();
 
   /// getHostCPUName - Get the LLVM name for the host CPU. The particular format
   /// of the name is target dependent, and suitable for passing as -mcpu to the
   /// target which matches the host.
   ///
   /// \return - The host CPU name, or empty if the CPU could not be determined.
-  StringRef getHostCPUName();
+  LLVM_SUPPORT_ABI StringRef getHostCPUName();
 
   /// getHostCPUFeatures - Get the LLVM names for the host CPU features.
   /// The particular format of the names are target dependent, and suitable for
   /// passing as -mattr to the target which matches the host.
   ///
   /// \param Features - A string mapping feature names to either
   /// true (if enabled) or false (if disabled). This routine makes no guarantees
   /// about exactly which features may appear in this map, except that they are
   /// all valid LLVM feature names.
   ///
   /// \return - True on success.
-  bool getHostCPUFeatures(StringMap<bool, MallocAllocator> &Features);
+  LLVM_SUPPORT_ABI bool getHostCPUFeatures(StringMap<bool, MallocAllocator> &Features);
 
   /// Get the number of physical cores (as opposed to logical cores returned
   /// from thread::hardware_concurrency(), which includes hyperthreads).
   /// Returns -1 if unknown for the current host system.
-  int getHostNumPhysicalCores();
+  LLVM_SUPPORT_ABI int getHostNumPhysicalCores();
 
   namespace detail {
   /// Helper functions to extract HostCPUName from /proc/cpuinfo on linux.
-  StringRef getHostCPUNameForPowerPC(StringRef ProcCpuinfoContent);
-  StringRef getHostCPUNameForARM(StringRef ProcCpuinfoContent);
-  StringRef getHostCPUNameForS390x(StringRef ProcCpuinfoContent);
-  StringRef getHostCPUNameForBPF();
+  LLVM_SUPPORT_ABI StringRef getHostCPUNameForPowerPC(StringRef ProcCpuinfoContent);
+  LLVM_SUPPORT_ABI StringRef getHostCPUNameForARM(StringRef ProcCpuinfoContent);
+  LLVM_SUPPORT_ABI StringRef getHostCPUNameForS390x(StringRef ProcCpuinfoContent);
+  LLVM_SUPPORT_ABI StringRef getHostCPUNameForBPF();
 
   /// Helper functions to extract CPU details from CPUID on x86.
   namespace x86 {
   enum class VendorSignatures {
     UNKNOWN,
     GENUINE_INTEL,
     AUTHENTIC_AMD,
   };
 
   /// Returns the host CPU's vendor.
   /// MaxLeaf: if a non-nullptr pointer is specified, the EAX value will be
   /// assigned to its pointee.
   VendorSignatures getVendorSignature(unsigned *MaxLeaf = nullptr);
   } // namespace x86
   }
 }
 }
 
 #endif

llvm/include/llvm/Support/InitLLVM.h

 //===- InitLLVM.h -----------------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_INITLLVM_H
 #define LLVM_SUPPORT_INITLLVM_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/PrettyStackTrace.h"
 
 // The main() functions in typical LLVM tools start with InitLLVM which does
 // the following one-time initializations:
 //
 //  1. Setting up a signal handler so that pretty stack trace is printed out
 //     if a process crashes. A signal handler that exits when a failed write to
 //     a pipe occurs may optionally be installed: this is on-by-default.
 //
 //  2. Set up the global new-handler which is called when a memory allocation
 //     attempt fails.
 //
 //  3. If running on Windows, obtain command line arguments using a
 //     multibyte character-aware API and convert arguments into UTF-8
 //     encoding, so that you can assume that command line arguments are
 //     always encoded in UTF-8 on any platform.
 //
 // InitLLVM calls llvm_shutdown() on destruction, which cleans up
 // ManagedStatic objects.
 namespace llvm {
-class InitLLVM {
+class LLVM_SUPPORT_ABI InitLLVM {
 public:
   InitLLVM(int &Argc, const char **&Argv,
            bool InstallPipeSignalExitHandler = true);
   InitLLVM(int &Argc, char **&Argv, bool InstallPipeSignalExitHandler = true)
       : InitLLVM(Argc, const_cast<const char **&>(Argv),
                  InstallPipeSignalExitHandler) {}
 
   ~InitLLVM();

llvm/include/llvm/Support/InstructionCost.h

 /// accumulating costs on large cost-values don't overflow.
 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_INSTRUCTIONCOST_H
 #define LLVM_SUPPORT_INSTRUCTIONCOST_H
 
 #include "llvm/ADT/Optional.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MathExtras.h"
+
 #include <limits>
 
 namespace llvm {
 
 class raw_ostream;
 
-class InstructionCost {
+class LLVM_SUPPORT_ABI InstructionCost {
 public:
   using CostType = int64_t;
 
   /// CostState describes the state of a cost.
   enum CostState {
     Valid,  /// < The cost value represents a valid cost, even when the
             /// cost-value is large.
     Invalid /// < Invalid indicates there is no way to represent the cost as a

llvm/include/llvm/Support/ItaniumManglingCanonicalizer.h

 //===--- ItaniumManglingCanonicalizer.h -------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines a class for computing equivalence classes of mangled names
 // given a set of equivalences between name fragments.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_ITANIUMMANGLINGCANONICALIZER_H
 #define LLVM_SUPPORT_ITANIUMMANGLINGCANONICALIZER_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cstddef>
 #include <cstdint>
 
 namespace llvm {
 
 class StringRef;
 
 /// Canonicalizer for mangled names.
 ///
 /// This class allows specifying a list of "equivalent" manglings. For example,
 /// you can specify that Ss is equivalent to
 ///   NSt3__112basic_stringIcNS_11char_traitsIcEENS_9allocatorIcEEEE
 /// and then manglings that refer to libstdc++'s 'std::string' will be
 /// considered equivalent to manglings that are the same except that they refer
 /// to libc++'s 'std::string'.
 ///
 /// This can be used when data (eg, profiling data) is available for a version
 /// of a program built in a different configuration, with correspondingly
 /// different manglings.
-class ItaniumManglingCanonicalizer {
+class LLVM_SUPPORT_ABI ItaniumManglingCanonicalizer {
 public:
   ItaniumManglingCanonicalizer();
   ItaniumManglingCanonicalizer(const ItaniumManglingCanonicalizer &) = delete;
   void operator=(const ItaniumManglingCanonicalizer &) = delete;
   ~ItaniumManglingCanonicalizer();
 
   enum class EquivalenceError {
     Success,

llvm/include/llvm/Support/JSON.h

 #ifndef LLVM_SUPPORT_JSON_H
 #define LLVM_SUPPORT_JSON_H
 
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Error.h"
 #include "llvm/Support/FormatVariadic.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <map>
 
 namespace llvm {
 namespace json {
 
 // === String encodings ===
 //
 // JSON strings are character sequences (not byte sequences like std::string).
 // We need to know the encoding, and for simplicity only support UTF-8.
 //
 //   - When parsing, invalid UTF-8 is a syntax error like any other
 //
 //   - When creating Values from strings, callers must ensure they are UTF-8.
 //        with asserts on, invalid UTF-8 will crash the program
 //        with asserts off, we'll substitute the replacement character (U+FFFD)
 //     Callers can use json::isUTF8() and json::fixUTF8() for validation.
 //
 //   - When retrieving strings from Values (e.g. asString()), the result will
 //     always be valid UTF-8.
 
 /// Returns true if \p S is valid UTF-8, which is required for use as JSON.
 /// If it returns false, \p Offset is set to a byte offset near the first error.
-bool isUTF8(llvm::StringRef S, size_t *ErrOffset = nullptr);
+LLVM_SUPPORT_ABI bool isUTF8(llvm::StringRef S, size_t *ErrOffset = nullptr);
 /// Replaces invalid UTF-8 sequences in \p S with the replacement character
 /// (U+FFFD). The returned string is valid UTF-8.
 /// This is much slower than isUTF8, so test that first.
-std::string fixUTF8(llvm::StringRef S);
+LLVM_SUPPORT_ABI std::string fixUTF8(llvm::StringRef S);
 
 class Array;
 class ObjectKey;
 class Value;
 template <typename T> Value toJSON(const llvm::Optional<T> &Opt);
 
 /// An Object is a JSON object, which maps strings to heterogenous JSON values.
 /// It simulates DenseMap<ObjectKey, Value>. ObjectKey is a maybe-owned string.
-class Object {
+class LLVM_SUPPORT_ABI Object {
   using Storage = DenseMap<ObjectKey, Value, llvm::DenseMapInfo<StringRef>>;
   Storage M;
 
 public:
   using key_type = ObjectKey;
   using mapped_type = Value;
   using value_type = Storage::value_type;
   using iterator = Storage::iterator;
public:
   llvm::Optional<double> getNumber(StringRef K) const;
   llvm::Optional<int64_t> getInteger(StringRef K) const;
   llvm::Optional<llvm::StringRef> getString(StringRef K) const;
   const json::Object *getObject(StringRef K) const;
   json::Object *getObject(StringRef K);
   const json::Array *getArray(StringRef K) const;
   json::Array *getArray(StringRef K);
 };
-bool operator==(const Object &LHS, const Object &RHS);
+
+LLVM_SUPPORT_ABI bool operator==(const Object &LHS, const Object &RHS);
+
 inline bool operator!=(const Object &LHS, const Object &RHS) {
   return !(LHS == RHS);
 }
 
 /// An Array is a JSON array, which contains heterogeneous JSON values.
 /// It simulates std::vector<Value>.
-class Array {
+class LLVM_SUPPORT_ABI Array {
   std::vector<Value> V;
 
 public:
   using value_type = Value;
   using iterator = std::vector<Value>::iterator;
   using const_iterator = std::vector<Value>::const_iterator;
 
   Array() = default;

 /// Values can be serialized to JSON:
 ///   1) raw_ostream << Value                    // Basic formatting.
 ///   2) raw_ostream << formatv("{0}", Value)    // Basic formatting.
 ///   3) raw_ostream << formatv("{0:2}", Value)  // Pretty-print with indent 2.
 ///
 /// And parsed:
 ///   Expected<Value> E = json::parse("[1, 2, null]");
 ///   assert(E && E->kind() == Value::Array);
-class Value {
+class LLVM_SUPPORT_ABI Value {
 public:
   enum Kind {
     Null,
     Boolean,
     /// Number values can store both int64s and doubles at full precision,
     /// depending on what they were constructed/parsed from.
     Number,
     String,
enum ValueType : char {
     T_Object,
     T_Array,
   };
   // All members mutable, see moveFrom().
   mutable ValueType Type;
   mutable llvm::AlignedCharArrayUnion<bool, double, int64_t, llvm::StringRef,
                                       std::string, json::Array, json::Object>
       Union;
-  friend bool operator==(const Value &, const Value &);
+
+  friend LLVM_SUPPORT_ABI bool operator==(const Value &, const Value &);
 };
 
-bool operator==(const Value &, const Value &);
+LLVM_SUPPORT_ABI bool operator==(const Value &, const Value &);
+
 inline bool operator!=(const Value &L, const Value &R) { return !(L == R); }
 
 /// ObjectKey is a used to capture keys in Object. Like Value but:
 ///   - only strings are allowed
 ///   - it's optimized for the string literal case (Owned == nullptr)
 /// Like Value, strings must be UTF-8. See isUTF8 documentation for details.
 class ObjectKey {
 public:

 }
 inline bool Object::erase(StringRef K) {
   return M.erase(ObjectKey(K));
 }
 
 /// A "cursor" marking a position within a Value.
 /// The Value is a tree, and this is the path from the root to the current node.
 /// This is used to associate errors with particular subobjects.
-class Path {
+class LLVM_SUPPORT_ABI Path {
 public:
   class Root;
 
   /// Records that the value at the current path is invalid.
   /// Message is e.g. "expected number" and becomes part of the final error.
   /// This overwrites any previously written error message in the root.
   void report(llvm::StringLiteral Message);
 
private:
   const Path *Parent;
   Segment Seg;
 
   Path(const Path *Parent, Segment S) : Parent(Parent), Seg(S) {}
 };
 
 /// The root is the trivial Path to the root value.
 /// It also stores the latest reported error and the path where it occurred.
-class Path::Root {
+class LLVM_SUPPORT_ABI Path::Root {
   llvm::StringRef Name;
   llvm::StringLiteral ErrorMessage;
   std::vector<Path::Segment> ErrorPath; // Only valid in error state. Reversed.
 
   friend void Path::report(llvm::StringLiteral Message);
 
 public:
   Root(llvm::StringRef Name = "") : Name(Name), ErrorMessage("") {}

 private:
   const Object *O;
   Path P;
 };
 
 /// Parses the provided JSON source, or returns a ParseError.
 /// The returned Value is self-contained and owns its strings (they do not refer
 /// to the original source).
-llvm::Expected<Value> parse(llvm::StringRef JSON);
+LLVM_SUPPORT_ABI llvm::Expected<Value> parse(llvm::StringRef JSON);
 
 class ParseError : public llvm::ErrorInfo<ParseError> {
   const char *Msg;
   unsigned Line, Column, Offset;
 
 public:
   static char ID;
   ParseError(const char *Msg, unsigned Line, unsigned Column, unsigned Offset)

 ///     J.objectEnd();
 ///   }
 ///   J.arrayEnd();
 ///
 /// If the call sequence isn't valid JSON, asserts will fire in debug mode.
 /// This can be mismatched begin()/end() pairs, trying to emit attributes inside
 /// an array, and so on.
 /// With asserts disabled, this is undefined behavior.
-class OStream {
+class LLVM_SUPPORT_ABI OStream {
  public:
   using Block = llvm::function_ref<void()>;
   // If IndentSize is nonzero, output is pretty-printed.
   explicit OStream(llvm::raw_ostream &OS, unsigned IndentSize = 0)
       : OS(OS), IndentSize(IndentSize) {
     Stack.emplace_back();
   }
   ~OStream() {
     assert(Stack.size() == 1 && "Unmatched begin()/end()");
inline llvm::raw_ostream &operator<<(llvm::raw_ostream &OS, const Value &V) {
   OStream(OS).value(V);
   return OS;
 }
 } // namespace json
 
 /// Allow printing json::Value with formatv().
 /// The default style is basic/compact formatting, like operator<<.
 /// A format string like formatv("{0:2}", Value) pretty-prints with indent 2.
-template <> struct format_provider<llvm::json::Value> {
+template <> struct LLVM_SUPPORT_ABI format_provider<llvm::json::Value> {
   static void format(const llvm::json::Value &, raw_ostream &, StringRef);
 };
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/KnownBits.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_KNOWNBITS_H
 #define LLVM_SUPPORT_KNOWNBITS_H
 
 #include "llvm/ADT/APInt.h"
 #include "llvm/ADT/Optional.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 // Struct for tracking the known zeros and ones of a value.
-struct KnownBits {
+struct LLVM_SUPPORT_ABI KnownBits {
   APInt Zero;
   APInt One;
 
 private:
   // Internal constructor for creating a KnownBits from two APInts.
   KnownBits(APInt Zero, APInt One)
       : Zero(std::move(Zero)), One(std::move(One)) {}
 

llvm/include/llvm/Support/LEB128.h

 //===- llvm/Support/LEB128.h - [SU]LEB128 utility functions -----*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file declares some utility functions for encoding SLEB128 and
 // ULEB128 values.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_LEB128_H
 #define LLVM_SUPPORT_LEB128_H
 
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 /// Utility function to encode a SLEB128 value to an output stream. Returns
 /// the length in bytes of the encoded value.
 inline unsigned encodeSLEB128(int64_t Value, raw_ostream &OS,
                               unsigned PadTo = 0) {
inline int64_t decodeSLEB128(const uint8_t *p, unsigned *n = nullptr,
   if (Shift < 64 && (Byte & 0x40))
     Value |= (-1ULL) << Shift;
   if (n)
     *n = (unsigned)(p - orig_p);
   return Value;
 }
 
 /// Utility function to get the size of the ULEB128-encoded value.
-extern unsigned getULEB128Size(uint64_t Value);
+LLVM_SUPPORT_ABI unsigned getULEB128Size(uint64_t Value);
 
 /// Utility function to get the size of the SLEB128-encoded value.
-extern unsigned getSLEB128Size(int64_t Value);
+LLVM_SUPPORT_ABI unsigned getSLEB128Size(int64_t Value);
 
 } // namespace llvm
 
 #endif // LLVM_SUPPORT_LEB128_H

llvm/include/llvm/Support/LLVMSupportExports.h

+//===--- SupportMacros.h - ABI Macro Definitions ----------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_SUPPORT_LLVMSUPPORT_EXPORTS_H
+#define LLVM_SUPPORT_LLVMSUPPORT_EXPORTS_H
+
+#if defined(__ELF__)
+# if defined(LLVM_SUPPORT_STATIC)
+#   define LLVM_SUPPORT_ABI
+# else
+#   if defined(LLVMSupport_EXPORTS)
+#     define LLVM_SUPPORT_ABI __attribute__((__visibility__("protected")))
+#   else
+#     define LLVM_SUPPORT_ABI __attribute__((__visibility__("default")))
+#   endif
+# endif
+#elif defined(__MACH__) || defined(__WASM__)
+# if defined(LLVM_SUPPORT_STATIC)
+#   define LLVM_SUPPORT_ABI
+# else
+#   define LLVM_SUPPORT_ABI __attribute__((__visibility__("default")))
+# endif
+#else
+# if defined(LLVM_SUPPORT_STATIC)
+#   define LLVM_SUPPORT_abi
+# else
+#   if defined(LLVMSupport_EXPORTS)
+#     define LLVM_SUPPORT_ABI __declspec(dllexport)
+#   else
+#     define LLVM_SUPPORT_ABI __declspec(dllimport)
+#   endif
+# endif
+#endif
+
+#endif

llvm/include/llvm/Support/LineIterator.h

 //===- LineIterator.h - Iterator to read a text buffer's lines --*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_LINEITERATOR_H
 #define LLVM_SUPPORT_LINEITERATOR_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MemoryBufferRef.h"
+
 #include <iterator>
 
 namespace llvm {
 
 class MemoryBuffer;
 
 /// A forward iterator which reads text lines from a buffer.
 ///
 /// This class provides a forward iterator interface for reading one line at
 /// a time from a buffer. When default constructed the iterator will be the
 /// "end" iterator.
 ///
 /// The iterator is aware of what line number it is currently processing. It
 /// strips blank lines by default, and comment lines given a comment-starting
 /// character.
 ///
 /// Note that this iterator requires the buffer to be nul terminated.
-class line_iterator {
+class LLVM_SUPPORT_ABI line_iterator {
   Optional<MemoryBufferRef> Buffer;
   char CommentMarker = '\0';
   bool SkipBlanks = true;
 
   unsigned LineNumber = 1;
   StringRef CurrentLine;
 
 public:

llvm/include/llvm/Support/Locale.h

+//===--- Locale.h - Locale helpers -----------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===---------------------------------------------------------------------===//
+
 #ifndef LLVM_SUPPORT_LOCALE_H
 #define LLVM_SUPPORT_LOCALE_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 namespace llvm {
 class StringRef;
 
 namespace sys {
 namespace locale {
 
-int columnWidth(StringRef s);
-bool isPrint(int c);
+LLVM_SUPPORT_ABI int columnWidth(StringRef s);
+LLVM_SUPPORT_ABI bool isPrint(int c);
 
 }
 }
 }
 
 #endif // LLVM_SUPPORT_LOCALE_H

llvm/include/llvm/Support/LockFileManager.h

 //===--- LockFileManager.h - File-level locking utility ---------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 #ifndef LLVM_SUPPORT_LOCKFILEMANAGER_H
 #define LLVM_SUPPORT_LOCKFILEMANAGER_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/SmallString.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <system_error>
 #include <utility> // for std::pair
 
 namespace llvm {
 class StringRef;
 
 /// Class that manages the creation of a lock file to aid
 /// implicit coordination between different processes.
 ///
 /// The implicit coordination works by creating a ".lock" file alongside
 /// the file that we're coordinating for, using the atomicity of the file
 /// system to ensure that only a single process can create that ".lock" file.
 /// When the lock file is removed, the owning process has finished the
 /// operation.
-class LockFileManager {
+class LLVM_SUPPORT_ABI LockFileManager {
 public:
   /// Describes the state of a lock file.
   enum LockFileState {
     /// The lock file has been created and is owned by this instance
     /// of the object.
     LFS_Owned,
     /// The lock file already exists and is owned by some other
     /// instance.

llvm/include/llvm/Support/LowLevelTypeImpl.h

 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_LOWLEVELTYPEIMPL_H
 #define LLVM_SUPPORT_LOWLEVELTYPEIMPL_H
 
 #include "llvm/ADT/DenseMapInfo.h"
 #include "llvm/Support/Debug.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MachineValueType.h"
+
 #include <cassert>
 
 namespace llvm {
 
 class DataLayout;
 class Type;
 class raw_ostream;
 
-class LLT {
+class LLVM_SUPPORT_ABI LLT {
 public:
   /// Get a low-level scalar or aggregate "bag of bits".
   static LLT scalar(unsigned SizeInBits) {
     return LLT{/*isPointer=*/false, /*isVector=*/false, /*isScalar=*/true,
                ElementCount::getFixed(0), SizeInBits,
                /*AddressSpace=*/0};
   }
 

llvm/include/llvm/Support/MD5.h

  *
  * See md5.c for more information.
  */
 
 #ifndef LLVM_SUPPORT_MD5_H
 #define LLVM_SUPPORT_MD5_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Endian.h"
+#include "llvm/SUpport/LLVMSupportExports.h"
+
 #include <array>
 #include <cstdint>
 
 namespace llvm {
 
 template <unsigned N> class SmallString;
 template <typename T> class ArrayRef;
 
-class MD5 {
+class LLVM_SUPPORT_ABI MD5 {
 public:
-  struct MD5Result {
+  struct LLVM_SUPPORT_ABI MD5Result {
     std::array<uint8_t, 16> Bytes;
 
     operator std::array<uint8_t, 16>() const { return Bytes; }
 
     const uint8_t &operator[](size_t I) const { return Bytes[I]; }
     uint8_t &operator[](size_t I) { return Bytes[I]; }
 
     SmallString<32> digest() const;

llvm/include/llvm/Support/ManagedStatic.h

 //===-- llvm/Support/ManagedStatic.h - Static Global wrapper ----*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines the ManagedStatic class and the llvm_shutdown() function.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_MANAGEDSTATIC_H
 #define LLVM_SUPPORT_MANAGEDSTATIC_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <atomic>
 #include <cstddef>
 
 namespace llvm {
 
 /// object_creator - Helper method for ManagedStatic.
 template <class C> struct object_creator {
   static void *call() { return new C(); }

 // settings. For the affected versions of MSVC, use the old linker
 // initialization pattern of not providing a constructor and leaving the fields
 // uninitialized. See http://llvm.org/PR41367 for details.
 #if !defined(_MSC_VER) || (_MSC_VER >= 1925) || defined(__clang__)
 #define LLVM_USE_CONSTEXPR_CTOR
 #endif
 
 /// ManagedStaticBase - Common base class for ManagedStatic instances.
-class ManagedStaticBase {
+class LLVM_SUPPORT_ABI ManagedStaticBase {
 protected:
 #ifdef LLVM_USE_CONSTEXPR_CTOR
   mutable std::atomic<void *> Ptr{};
   mutable void (*DeleterFn)(void *) = nullptr;
   mutable const ManagedStaticBase *Next = nullptr;
 #else
   // This should only be used as a static variable, which guarantees that this
   // will be zero initialized.
public:
   // Extract the instance, leaving the ManagedStatic uninitialized. The
   // user is then responsible for the lifetime of the returned instance.
   C *claim() {
     return static_cast<C *>(Ptr.exchange(nullptr));
   }
 };
 
 /// llvm_shutdown - Deallocate and destroy all ManagedStatic variables.
-void llvm_shutdown();
+LLVM_SUPPORT_ABI void llvm_shutdown();
 
 /// llvm_shutdown_obj - This is a simple helper class that calls
 /// llvm_shutdown() when it is destroyed.
 struct llvm_shutdown_obj {
   llvm_shutdown_obj() = default;
   ~llvm_shutdown_obj() { llvm_shutdown(); }
 };
 
 } // end namespace llvm
 
 #endif // LLVM_SUPPORT_MANAGEDSTATIC_H

llvm/include/llvm/Support/MathExtras.h

 //===-- llvm/Support/MathExtras.h - Useful math functions -------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains some functions that are useful for math stuff.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_MATHEXTRAS_H
 #define LLVM_SUPPORT_MATHEXTRAS_H
 
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <climits>
 #include <cmath>
 #include <cstdint>
 #include <cstring>
 #include <limits>
 #include <type_traits>
 
SaturatingMultiplyAdd(T X, T Y, T A, bool *ResultOverflowed = nullptr) {
   T Product = SaturatingMultiply(X, Y, &Overflowed);
   if (Overflowed)
     return Product;
 
   return SaturatingAdd(A, Product, &Overflowed);
 }
 
 /// Use this rather than HUGE_VALF; the latter causes warnings on MSVC.
-extern const float huge_valf;
+extern LLVM_SUPPORT_ABI const float huge_valf;
 
 
 /// Add two signed integers, computing the two's complement truncated result,
 /// returning true if overflow occured.
 template <typename T>
 std::enable_if_t<std::is_signed<T>::value, T> AddOverflow(T X, T Y, T &Result) {
 #if __has_builtin(__builtin_add_overflow)
   return __builtin_add_overflow(X, Y, &Result);
 #else
   // Perform the unsigned addition.

llvm/include/llvm/Support/MemAlloc.h

 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_MEMALLOC_H
 #define LLVM_SUPPORT_MEMALLOC_H
 
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include <cstdlib>
 
 namespace llvm {
 
 LLVM_ATTRIBUTE_RETURNS_NONNULL inline void *safe_malloc(size_t Sz) {
   void *Result = std::malloc(Sz);
   if (Result == nullptr) {
     // It is implementation-defined whether allocation occurs if the space

 ///
 /// When the compiler supports aligned operator new, this will use it to to
 /// handle even over-aligned allocations.
 ///
 /// However, this doesn't make any attempt to leverage the fancier techniques
 /// like posix_memalign due to portability. It is mostly intended to allow
 /// compatibility with platforms that, after aligned allocation was added, use
 /// reduced default alignment.
-LLVM_ATTRIBUTE_RETURNS_NONNULL LLVM_ATTRIBUTE_RETURNS_NOALIAS void *
+LLVM_SUPPORT_ABI LLVM_ATTRIBUTE_RETURNS_NONNULL LLVM_ATTRIBUTE_RETURNS_NOALIAS void *
 allocate_buffer(size_t Size, size_t Alignment);
 
 /// Deallocate a buffer of memory with the given size and alignment.
 ///
 /// If supported, this will used the sized delete operator. Also if supported,
 /// this will pass the alignment to the delete operator.
 ///
 /// The pointer must have been allocated with the corresponding new operator,
 /// most likely using the above helper.
-void deallocate_buffer(void *Ptr, size_t Size, size_t Alignment);
+LLVM_SUPPORT_ABI void deallocate_buffer(void *Ptr, size_t Size, size_t Alignment);
 
 } // namespace llvm
 #endif

llvm/include/llvm/Support/Memory.h

 //===- llvm/Support/Memory.h - Memory Support -------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file declares the llvm::sys::Memory class.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_MEMORY_H
 #define LLVM_SUPPORT_MEMORY_H
 
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <system_error>
 
 namespace llvm {
 
 // Forward declare raw_ostream: it is used for debug dumping below.
 class raw_ostream;
 
 namespace sys {
private:
     unsigned Flags = 0;
     friend class Memory;
   };
 
   /// This class provides various memory handling functions that manipulate
   /// MemoryBlock instances.
   /// @since 1.4
   /// An abstraction for memory operations.
-  class Memory {
+  class LLVM_SUPPORT_ABI Memory {
   public:
     enum ProtectionFlags {
       MF_READ = 0x1000000,
       MF_WRITE = 0x2000000,
       MF_EXEC = 0x4000000,
       MF_RWE_MASK = 0x7000000,
 
       /// The \p MF_HUGE_HINT flag is used to indicate that the request for

llvm/include/llvm/Support/MemoryBuffer.h

 #define LLVM_SUPPORT_MEMORYBUFFER_H
 
 #include "llvm-c/Types.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/CBindingWrapping.h"
 #include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MemoryBufferRef.h"
+
 #include <cstddef>
 #include <cstdint>
 #include <memory>
 
 namespace llvm {
 namespace sys {
 namespace fs {
 // Duplicated from FileSystem.h to avoid a dependency.

 /// buffer.  In addition to basic access to the characters in the file, this
 /// interface guarantees you can read one character past the end of the file,
 /// and that this character will read as '\0'.
 ///
 /// The '\0' guarantee is needed to support an optimization -- it's intended to
 /// be more efficient for clients which are reading all the data to stop
 /// reading when they encounter a '\0' than to continually check the file
 /// position to see if it has reached the end of the file.
-class MemoryBuffer {
+class LLVM_SUPPORT_ABI MemoryBuffer {
   const char *BufferStart; // Start of the buffer.
   const char *BufferEnd;   // End of the buffer.
 
 protected:
   MemoryBuffer() = default;
 
   void init(const char *BufStart, const char *BufEnd,
             bool RequiresNullTerminator);
public:
 
   MemoryBufferRef getMemBufferRef() const;
 };
 
 /// This class is an extension of MemoryBuffer, which allows copy-on-write
 /// access to the underlying contents.  It only supports creation methods that
 /// are guaranteed to produce a writable buffer.  For example, mapping a file
 /// read-only is not supported.
-class WritableMemoryBuffer : public MemoryBuffer {
+class LLVM_SUPPORT_ABI WritableMemoryBuffer : public MemoryBuffer {
 protected:
   WritableMemoryBuffer() = default;
 
 public:
   using MemoryBuffer::getBuffer;
   using MemoryBuffer::getBufferEnd;
   using MemoryBuffer::getBufferStart;
 
private:
   using MemoryBuffer::getOpenFileSlice;
   using MemoryBuffer::getSTDIN;
 };
 
 /// This class is an extension of MemoryBuffer, which allows write access to
 /// the underlying contents and committing those changes to the original source.
 /// It only supports creation methods that are guaranteed to produce a writable
 /// buffer.  For example, mapping a file read-only is not supported.
-class WriteThroughMemoryBuffer : public MemoryBuffer {
+class LLVM_SUPPORT_ABI WriteThroughMemoryBuffer : public MemoryBuffer {
 protected:
   WriteThroughMemoryBuffer() = default;
 
 public:
   using MemoryBuffer::getBuffer;
   using MemoryBuffer::getBufferEnd;
   using MemoryBuffer::getBufferStart;
 

llvm/include/llvm/Support/MemoryBufferRef.h

 //===- MemoryBufferRef.h - Memory Buffer Reference --------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 //  This file defines the MemoryBuffer interface.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_MEMORYBUFFERREF_H
 #define LLVM_SUPPORT_MEMORYBUFFERREF_H
 
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 class MemoryBuffer;
 
-class MemoryBufferRef {
+class LLVM_SUPPORT_ABI MemoryBufferRef {
   StringRef Buffer;
   StringRef Identifier;
 
 public:
   MemoryBufferRef() = default;
   MemoryBufferRef(const MemoryBuffer &Buffer);
   MemoryBufferRef(StringRef Buffer, StringRef Identifier)
       : Buffer(Buffer), Identifier(Identifier) {}

llvm/include/llvm/Support/NativeFormatting.h

 //===- NativeFormatting.h - Low level formatting helpers ---------*- C++-*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_NATIVEFORMATTING_H
 #define LLVM_SUPPORT_NATIVEFORMATTING_H
 
 #include "llvm/ADT/Optional.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cstdint>
 
 namespace llvm {
 class raw_ostream;
 enum class FloatStyle { Exponent, ExponentUpper, Fixed, Percent };
 enum class IntegerStyle {
   Integer,
   Number,
 };
 enum class HexPrintStyle { Upper, Lower, PrefixUpper, PrefixLower };
 
-size_t getDefaultPrecision(FloatStyle Style);
+LLVM_SUPPORT_ABI size_t getDefaultPrecision(FloatStyle Style);
 
-bool isPrefixedHexStyle(HexPrintStyle S);
+LLVM_SUPPORT_ABI bool isPrefixedHexStyle(HexPrintStyle S);
 
+LLVM_SUPPORT_ABI
 void write_integer(raw_ostream &S, unsigned int N, size_t MinDigits,
                    IntegerStyle Style);
+
+LLVM_SUPPORT_ABI
 void write_integer(raw_ostream &S, int N, size_t MinDigits, IntegerStyle Style);
+
+LLVM_SUPPORT_ABI
 void write_integer(raw_ostream &S, unsigned long N, size_t MinDigits,
                    IntegerStyle Style);
+
+LLVM_SUPPORT_ABI
 void write_integer(raw_ostream &S, long N, size_t MinDigits,
                    IntegerStyle Style);
+
+LLVM_SUPPORT_ABI
 void write_integer(raw_ostream &S, unsigned long long N, size_t MinDigits,
                    IntegerStyle Style);
+
+LLVM_SUPPORT_ABI
 void write_integer(raw_ostream &S, long long N, size_t MinDigits,
                    IntegerStyle Style);
 
+LLVM_SUPPORT_ABI
 void write_hex(raw_ostream &S, uint64_t N, HexPrintStyle Style,
                Optional<size_t> Width = None);
+
+LLVM_SUPPORT_ABI
 void write_double(raw_ostream &S, double D, FloatStyle Style,
                   Optional<size_t> Precision = None);
 }
 
 #endif
 

llvm/include/llvm/Support/OptimizedStructLayout.h

 ///   impact on locality: less overall memory may be used, but adjacent
 ///   fields in the original array may be moved further from one another.
 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_OPTIMIZEDSTRUCTLAYOUT_H
 #define LLVM_SUPPORT_OPTIMIZEDSTRUCTLAYOUT_H
 
-#include "llvm/Support/Alignment.h"
 #include "llvm/ADT/ArrayRef.h"
+#include "llvm/Support/Alignment.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <utility>
 
 namespace llvm {
 
 /// A field in a structure.
 struct OptimizedStructLayoutField {
   /// A special value for Offset indicating that the field can be moved
   /// anywhere.

 /// On return, all fields will have been assigned a fixed offset, and the
 /// array will be sorted in order of ascending offsets.  Note that this
 /// means that the fixed-offset fields may no longer form a strict prefix
 /// if there's any padding before they end.
 ///
 /// The return value is the total size of the struct and its required
 /// alignment.  Note that the total size is not rounded up to a multiple
 /// of the required alignment; clients which require this can do so easily.
-std::pair<uint64_t, Align> performOptimizedStructLayout(
+LLVM_SUPPORT_ABI std::pair<uint64_t, Align> performOptimizedStructLayout(
                         MutableArrayRef<OptimizedStructLayoutField> Fields);
 
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/Parallel.h

 //===- llvm/Support/Parallel.h - Parallel algorithms ----------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_PARALLEL_H
 #define LLVM_SUPPORT_PARALLEL_H
 
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/Config/llvm-config.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/Threading.h"
 
 #include <algorithm>
 #include <condition_variable>
 #include <functional>
 #include <mutex>
 
 namespace llvm {
 
 namespace parallel {
 
 // Strategy for the default executor used by the parallel routines provided by
 // this file. It defaults to using all hardware threads and should be
 // initialized before the first use of parallel routines.
-extern ThreadPoolStrategy strategy;
+extern LLVM_SUPPORT_ABI ThreadPoolStrategy strategy;
 
 namespace detail {
 
 #if LLVM_ENABLE_THREADS
 
 class Latch {
   uint32_t Count;
   mutable std::mutex Mutex;
public:
   }
 
   void sync() const {
     std::unique_lock<std::mutex> lock(Mutex);
     Cond.wait(lock, [&] { return Count == 0; });
   }
 };
 
-class TaskGroup {
+class LLVM_SUPPORT_ABI TaskGroup {
   Latch L;
   bool Parallel;
 
 public:
   TaskGroup();
   ~TaskGroup();
 
   void spawn(std::function<void()> f);

llvm/include/llvm/Support/Path.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_PATH_H
 #define LLVM_SUPPORT_PATH_H
 
 #include "llvm/ADT/Twine.h"
 #include "llvm/ADT/iterator.h"
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <iterator>
 #include <system_error>
 
 namespace llvm {
 namespace sys {
 namespace path {
 
 enum class Style { windows, posix, native };
class const_iterator
     : public iterator_facade_base<const_iterator, std::input_iterator_tag,
                                   const StringRef> {
   StringRef Path;          ///< The entire path.
   StringRef Component;     ///< The current component. Not necessarily in Path.
   size_t    Position = 0;  ///< The iterators current position within Path.
   Style S = Style::native; ///< The path style to use.
 
   // An end iterator has Position = Path.size() + 1.
-  friend const_iterator begin(StringRef path, Style style);
-  friend const_iterator end(StringRef path);
+  friend LLVM_SUPPORT_ABI const_iterator begin(StringRef path, Style style);
+  friend LLVM_SUPPORT_ABI const_iterator end(StringRef path);
 
 public:
   reference operator*() const { return Component; }
-  const_iterator &operator++();    // preincrement
-  bool operator==(const const_iterator &RHS) const;
+  LLVM_SUPPORT_ABI const_iterator &operator++();    // preincrement
+  LLVM_SUPPORT_ABI bool operator==(const const_iterator &RHS) const;
 
   /// Difference in bytes between this and RHS.
-  ptrdiff_t operator-(const const_iterator &RHS) const;
+  LLVM_SUPPORT_ABI ptrdiff_t operator-(const const_iterator &RHS) const;
 };
 
 /// Reverse path iterator.
 ///
 /// This is an input iterator that iterates over the individual components in
 /// \a path in reverse order. The traversal order is exactly reversed from that
 /// of \a const_iterator
 class reverse_iterator
     : public iterator_facade_base<reverse_iterator, std::input_iterator_tag,
                                   const StringRef> {
   StringRef Path;          ///< The entire path.
   StringRef Component;     ///< The current component. Not necessarily in Path.
   size_t    Position = 0;  ///< The iterators current position within Path.
   Style S = Style::native; ///< The path style to use.
 
-  friend reverse_iterator rbegin(StringRef path, Style style);
-  friend reverse_iterator rend(StringRef path);
+  friend LLVM_SUPPORT_ABI reverse_iterator rbegin(StringRef path, Style style);
+  friend LLVM_SUPPORT_ABI reverse_iterator rend(StringRef path);
 
 public:
   reference operator*() const { return Component; }
-  reverse_iterator &operator++();    // preincrement
-  bool operator==(const reverse_iterator &RHS) const;
+  LLVM_SUPPORT_ABI reverse_iterator &operator++();    // preincrement
+  LLVM_SUPPORT_ABI bool operator==(const reverse_iterator &RHS) const;
 
   /// Difference in bytes between this and RHS.
-  ptrdiff_t operator-(const reverse_iterator &RHS) const;
+  LLVM_SUPPORT_ABI ptrdiff_t operator-(const reverse_iterator &RHS) const;
 };
 
 /// Get begin iterator over \a path.
 /// @param path Input path.
 /// @returns Iterator initialized with the first component of \a path.
-const_iterator begin(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI const_iterator begin(StringRef path, Style style = Style::native);
 
 /// Get end iterator over \a path.
 /// @param path Input path.
 /// @returns Iterator initialized to the end of \a path.
-const_iterator end(StringRef path);
+LLVM_SUPPORT_ABI const_iterator end(StringRef path);
 
 /// Get reverse begin iterator over \a path.
 /// @param path Input path.
 /// @returns Iterator initialized with the first reverse component of \a path.
-reverse_iterator rbegin(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI reverse_iterator rbegin(StringRef path, Style style = Style::native);
 
 /// Get reverse end iterator over \a path.
 /// @param path Input path.
 /// @returns Iterator initialized to the reverse end of \a path.
-reverse_iterator rend(StringRef path);
+LLVM_SUPPORT_ABI reverse_iterator rend(StringRef path);
 
 /// @}
 /// @name Lexical Modifiers
 /// @{
 
 /// Remove the last component from \a path unless it is the root dir.
 ///
 /// Similar to the POSIX "dirname" utility.
 ///
 /// @code
 ///   directory/filename.cpp => directory/
 ///   directory/             => directory
 ///   filename.cpp           => <empty>
 ///   /                      => /
 /// @endcode
 ///
 /// @param path A path that is modified to not have a file component.
-void remove_filename(SmallVectorImpl<char> &path, Style style = Style::native);
+LLVM_SUPPORT_ABI void remove_filename(SmallVectorImpl<char> &path, Style style = Style::native);
 
 /// Replace the file extension of \a path with \a extension.
 ///
 /// @code
 ///   ./filename.cpp => ./filename.extension
 ///   ./filename     => ./filename.extension
 ///   ./             => ./.extension
 /// @endcode
 ///
 /// @param path A path that has its extension replaced with \a extension.
 /// @param extension The extension to be added. It may be empty. It may also
 ///                  optionally start with a '.', if it does not, one will be
 ///                  prepended.
-void replace_extension(SmallVectorImpl<char> &path, const Twine &extension,
+LLVM_SUPPORT_ABI void replace_extension(SmallVectorImpl<char> &path, const Twine &extension,
                        Style style = Style::native);
 
 /// Replace matching path prefix with another path.
 ///
 /// @code
 ///   /foo, /old, /new => /foo
 ///   /old, /old, /new => /new
 ///   /old, /old/, /new => /old
 ///   /old/foo, /old, /new => /new/foo
 ///   /old/foo, /old/, /new => /new/foo
 ///   /old/foo, /old/, /new/ => /new/foo
 ///   /oldfoo, /old, /new => /oldfoo
 ///   /foo, <empty>, /new => /new/foo
 ///   /foo, <empty>, new => new/foo
 ///   /old/foo, /old, <empty> => /foo
 /// @endcode
 ///
 /// @param Path If \a Path starts with \a OldPrefix modify to instead
 ///        start with \a NewPrefix.
 /// @param OldPrefix The path prefix to strip from \a Path.
 /// @param NewPrefix The path prefix to replace \a NewPrefix with.
 /// @param style The style used to match the prefix. Exact match using
 /// Posix style, case/separator insensitive match for Windows style.
 /// @result true if \a Path begins with OldPrefix
-bool replace_path_prefix(SmallVectorImpl<char> &Path, StringRef OldPrefix,
+LLVM_SUPPORT_ABI bool replace_path_prefix(SmallVectorImpl<char> &Path, StringRef OldPrefix,
                          StringRef NewPrefix,
                          Style style = Style::native);
 
 /// Append to path.
 ///
 /// @code
 ///   /foo  + bar/f => /foo/bar/f
 ///   /foo/ + bar/f => /foo/bar/f
 ///   foo   + bar/f => foo/bar/f
 /// @endcode
 ///
 /// @param path Set to \a path + \a component.
 /// @param a The component to be appended to \a path.
-void append(SmallVectorImpl<char> &path, const Twine &a,
+LLVM_SUPPORT_ABI void append(SmallVectorImpl<char> &path, const Twine &a,
                                          const Twine &b = "",
                                          const Twine &c = "",
                                          const Twine &d = "");
 
-void append(SmallVectorImpl<char> &path, Style style, const Twine &a,
+LLVM_SUPPORT_ABI void append(SmallVectorImpl<char> &path, Style style, const Twine &a,
             const Twine &b = "", const Twine &c = "", const Twine &d = "");
 
 /// Append to path.
 ///
 /// @code
 ///   /foo  + [bar,f] => /foo/bar/f
 ///   /foo/ + [bar,f] => /foo/bar/f
 ///   foo   + [bar,f] => foo/bar/f
 /// @endcode
 ///
 /// @param path Set to \a path + [\a begin, \a end).
 /// @param begin Start of components to append.
 /// @param end One past the end of components to append.
-void append(SmallVectorImpl<char> &path, const_iterator begin,
+LLVM_SUPPORT_ABI void append(SmallVectorImpl<char> &path, const_iterator begin,
             const_iterator end, Style style = Style::native);
 
 /// @}
 /// @name Transforms (or some other better name)
 /// @{
 
 /// Convert path to the native form. This is used to give paths to users and
 /// operating system calls in the platform's normal way. For example, on Windows
 /// all '/' are converted to '\'.
 ///
 /// @param path A path that is transformed to native format.
 /// @param result Holds the result of the transformation.
-void native(const Twine &path, SmallVectorImpl<char> &result,
+LLVM_SUPPORT_ABI void native(const Twine &path, SmallVectorImpl<char> &result,
             Style style = Style::native);
 
 /// Convert path to the native form in place. This is used to give paths to
 /// users and operating system calls in the platform's normal way. For example,
 /// on Windows all '/' are converted to '\'.
 ///
 /// @param path A path that is transformed to native format.
-void native(SmallVectorImpl<char> &path, Style style = Style::native);
+LLVM_SUPPORT_ABI void native(SmallVectorImpl<char> &path, Style style = Style::native);
 
 /// Replaces backslashes with slashes if Windows.
 ///
 /// @param path processed path
 /// @result The result of replacing backslashes with forward slashes if Windows.
 /// On Unix, this function is a no-op because backslashes are valid path
 /// chracters.
-std::string convert_to_slash(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI std::string convert_to_slash(StringRef path, Style style = Style::native);
 
 /// @}
 /// @name Lexical Observers
 /// @{
 
 /// Get root name.
 ///
 /// @code
 ///   //net/hello => //net
 ///   c:/hello    => c: (on Windows, on other platforms nothing)
 ///   /hello      => <empty>
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The root name of \a path if it has one, otherwise "".
-StringRef root_name(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef root_name(StringRef path, Style style = Style::native);
 
 /// Get root directory.
 ///
 /// @code
 ///   /goo/hello => /
 ///   c:/hello   => /
 ///   d/file.txt => <empty>
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The root directory of \a path if it has one, otherwise
 ///               "".
-StringRef root_directory(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef root_directory(StringRef path, Style style = Style::native);
 
 /// Get root path.
 ///
 /// Equivalent to root_name + root_directory.
 ///
 /// @param path Input path.
 /// @result The root path of \a path if it has one, otherwise "".
-StringRef root_path(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef root_path(StringRef path, Style style = Style::native);
 
 /// Get relative path.
 ///
 /// @code
 ///   C:\hello\world => hello\world
 ///   foo/bar        => foo/bar
 ///   /foo/bar       => foo/bar
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The path starting after root_path if one exists, otherwise "".
-StringRef relative_path(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef relative_path(StringRef path, Style style = Style::native);
 
 /// Get parent path.
 ///
 /// @code
 ///   /          => <empty>
 ///   /foo       => /
 ///   foo/../bar => foo/..
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The parent path of \a path if one exists, otherwise "".
-StringRef parent_path(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef parent_path(StringRef path, Style style = Style::native);
 
 /// Get filename.
 ///
 /// @code
 ///   /foo.txt    => foo.txt
 ///   .          => .
 ///   ..         => ..
 ///   /          => /
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The filename part of \a path. This is defined as the last component
 ///         of \a path. Similar to the POSIX "basename" utility.
-StringRef filename(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef filename(StringRef path, Style style = Style::native);
 
 /// Get stem.
 ///
 /// If filename contains a dot but not solely one or two dots, result is the
 /// substring of filename ending at (but not including) the last dot. Otherwise
 /// it is filename.
 ///
 /// @code
 ///   /foo/bar.txt => bar
 ///   /foo/bar     => bar
 ///   /foo/.txt    => <empty>
 ///   /foo/.       => .
 ///   /foo/..      => ..
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The stem of \a path.
-StringRef stem(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef stem(StringRef path, Style style = Style::native);
 
 /// Get extension.
 ///
 /// If filename contains a dot but not solely one or two dots, result is the
 /// substring of filename starting at (and including) the last dot, and ending
 /// at the end of \a path. Otherwise "".
 ///
 /// @code
 ///   /foo/bar.txt => .txt
 ///   /foo/bar     => <empty>
 ///   /foo/.txt    => .txt
 /// @endcode
 ///
 /// @param path Input path.
 /// @result The extension of \a path.
-StringRef extension(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef extension(StringRef path, Style style = Style::native);
 
 /// Check whether the given char is a path separator on the host OS.
 ///
 /// @param value a character
 /// @result true if \a value is a path separator character on the host OS
-bool is_separator(char value, Style style = Style::native);
+LLVM_SUPPORT_ABI bool is_separator(char value, Style style = Style::native);
 
 /// Return the preferred separator for this platform.
 ///
 /// @result StringRef of the preferred separator, null-terminated.
-StringRef get_separator(Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef get_separator(Style style = Style::native);
 
 /// Get the typical temporary directory for the system, e.g.,
 /// "/var/tmp" or "C:/TEMP"
 ///
 /// @param erasedOnReboot Whether to favor a path that is erased on reboot
 /// rather than one that potentially persists longer. This parameter will be
 /// ignored if the user or system has set the typical environment variable
 /// (e.g., TEMP on Windows, TMPDIR on *nix) to specify a temporary directory.
 ///
 /// @param result Holds the resulting path name.
-void system_temp_directory(bool erasedOnReboot, SmallVectorImpl<char> &result);
+LLVM_SUPPORT_ABI void system_temp_directory(bool erasedOnReboot, SmallVectorImpl<char> &result);
 
 /// Get the user's home directory.
 ///
 /// @param result Holds the resulting path name.
 /// @result True if a home directory is set, false otherwise.
-bool home_directory(SmallVectorImpl<char> &result);
+LLVM_SUPPORT_ABI bool home_directory(SmallVectorImpl<char> &result);
 
 /// Get the directory where packages should read user-specific configurations.
 /// e.g. $XDG_CONFIG_HOME.
 ///
 /// @param result Holds the resulting path name.
 /// @result True if the appropriate path was determined, it need not exist.
-bool user_config_directory(SmallVectorImpl<char> &result);
+LLVM_SUPPORT_ABI bool user_config_directory(SmallVectorImpl<char> &result);
 
 /// Get the directory where installed packages should put their
 /// machine-local cache, e.g. $XDG_CACHE_HOME.
 ///
 /// @param result Holds the resulting path name.
 /// @result True if the appropriate path was determined, it need not exist.
-bool cache_directory(SmallVectorImpl<char> &result);
+LLVM_SUPPORT_ABI bool cache_directory(SmallVectorImpl<char> &result);
 
 /// Has root name?
 ///
 /// root_name != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a root name, false otherwise.
-bool has_root_name(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_root_name(const Twine &path, Style style = Style::native);
 
 /// Has root directory?
 ///
 /// root_directory != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a root directory, false otherwise.
-bool has_root_directory(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_root_directory(const Twine &path, Style style = Style::native);
 
 /// Has root path?
 ///
 /// root_path != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a root path, false otherwise.
-bool has_root_path(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_root_path(const Twine &path, Style style = Style::native);
 
 /// Has relative path?
 ///
 /// relative_path != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a relative path, false otherwise.
 bool has_relative_path(const Twine &path, Style style = Style::native);
 
 /// Has parent path?
 ///
 /// parent_path != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a parent path, false otherwise.
-bool has_parent_path(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_parent_path(const Twine &path, Style style = Style::native);
 
 /// Has filename?
 ///
 /// filename != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a filename, false otherwise.
-bool has_filename(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_filename(const Twine &path, Style style = Style::native);
 
 /// Has stem?
 ///
 /// stem != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a stem, false otherwise.
-bool has_stem(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_stem(const Twine &path, Style style = Style::native);
 
 /// Has extension?
 ///
 /// extension != ""
 ///
 /// @param path Input path.
 /// @result True if the path has a extension, false otherwise.
-bool has_extension(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool has_extension(const Twine &path, Style style = Style::native);
 
 /// Is path absolute?
 ///
 /// According to cppreference.com, C++17 states: "An absolute path is a path
 /// that unambiguously identifies the location of a file without reference to
 /// an additional starting location."
 ///
 /// In other words, the rules are:
 /// 1) POSIX style paths with nonempty root directory are absolute.
 /// 2) Windows style paths with nonempty root name and root directory are
 ///    absolute.
 /// 3) No other paths are absolute.
 ///
 /// \see has_root_name
 /// \see has_root_directory
 ///
 /// @param path Input path.
 /// @result True if the path is absolute, false if it is not.
-bool is_absolute(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool is_absolute(const Twine &path, Style style = Style::native);
 
 /// Is path absolute using GNU rules?
 ///
 /// GNU rules are:
 /// 1) Paths starting with a path separator are absolute.
 /// 2) Windows style paths are also absolute if they start with a character
 ///    followed by ':'.
 /// 3) No other paths are absolute.
 ///
 /// On Windows style the path "C:\Users\Default" has "C:" as root name and "\"
 /// as root directory.
 ///
 /// Hence "C:" on Windows is absolute under GNU rules and not absolute under
 /// C++17 because it has no root directory. Likewise "/" and "\" on Windows are
 /// absolute under GNU and are not absolute under C++17 due to empty root name.
 ///
 /// \see has_root_name
 /// \see has_root_directory
 ///
 /// @param path Input path.
 /// @param style The style of \p path (e.g. Windows or POSIX). "native" style
 /// means to derive the style from the host.
 /// @result True if the path is absolute following GNU rules, false if it is
 /// not.
-bool is_absolute_gnu(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool is_absolute_gnu(const Twine &path, Style style = Style::native);
 
 /// Is path relative?
 ///
 /// @param path Input path.
 /// @result True if the path is relative, false if it is not.
-bool is_relative(const Twine &path, Style style = Style::native);
+LLVM_SUPPORT_ABI bool is_relative(const Twine &path, Style style = Style::native);
 
 /// Remove redundant leading "./" pieces and consecutive separators.
 ///
 /// @param path Input path.
 /// @result The cleaned-up \a path.
-StringRef remove_leading_dotslash(StringRef path, Style style = Style::native);
+LLVM_SUPPORT_ABI StringRef remove_leading_dotslash(StringRef path, Style style = Style::native);
 
 /// In-place remove any './' and optionally '../' components from a path.
 ///
 /// @param path processed path
 /// @param remove_dot_dot specify if '../' (except for leading "../") should be
 /// removed
 /// @result True if path was changed
-bool remove_dots(SmallVectorImpl<char> &path, bool remove_dot_dot = false,
+LLVM_SUPPORT_ABI bool remove_dots(SmallVectorImpl<char> &path, bool remove_dot_dot = false,
                  Style style = Style::native);
 
 } // end namespace path
 } // end namespace sys
 } // end namespace llvm
 
 #endif

llvm/include/llvm/Support/PluginLoader.h

 // header can only be included by a program ONCE, so it should never to used by
 // library authors.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_PLUGINLOADER_H
 #define LLVM_SUPPORT_PLUGINLOADER_H
 
+#include "llvm/Support/LLVMSupportExports.h"
 #ifndef DONT_GET_PLUGIN_LOADER_OPTION
 #include "llvm/Support/CommandLine.h"
 #endif
 
 #include <string>
 
 namespace llvm {
-  struct PluginLoader {
+  struct LLVM_SUPPORT_ABI PluginLoader {
     void operator=(const std::string &Filename);
     static unsigned getNumPlugins();
     static std::string& getPlugin(unsigned num);
   };
 
 #ifndef DONT_GET_PLUGIN_LOADER_OPTION
   // This causes operator= above to be invoked for every -load option.
   static cl::opt<PluginLoader, false, cl::parser<std::string> >
     LoadOpt("load", cl::ZeroOrMore, cl::value_desc("pluginfilename"),
             cl::desc("Load the specified plugin"));
 #endif
 }
 
 #endif

llvm/include/llvm/Support/PrettyStackTrace.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_PRETTYSTACKTRACE_H
 #define LLVM_SUPPORT_PRETTYSTACKTRACE_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
   class raw_ostream;
 
   /// Enables dumping a "pretty" stack trace when the program crashes.
   ///
   /// \see PrettyStackTraceEntry
-  void EnablePrettyStackTrace();
+  LLVM_SUPPORT_ABI void EnablePrettyStackTrace();
 
   /// Enables (or disables) dumping a "pretty" stack trace when the user sends
   /// SIGINFO or SIGUSR1 to the current process.
   ///
   /// This is a per-thread decision so that a program can choose to print stack
   /// traces only on a primary thread, or on all threads that use
   /// PrettyStackTraceEntry.
   ///
   /// \see EnablePrettyStackTrace
   /// \see PrettyStackTraceEntry
-  void EnablePrettyStackTraceOnSigInfoForThisThread(bool ShouldEnable = true);
+  LLVM_SUPPORT_ABI void EnablePrettyStackTraceOnSigInfoForThisThread(bool ShouldEnable = true);
 
   /// Replaces the generic bug report message that is output upon
   /// a crash.
-  void setBugReportMsg(const char *Msg);
+  LLVM_SUPPORT_ABI void setBugReportMsg(const char *Msg);
 
   /// Get the bug report message that will be output upon a crash.
-  const char *getBugReportMsg();
+  LLVM_SUPPORT_ABI const char *getBugReportMsg();
 
   /// PrettyStackTraceEntry - This class is used to represent a frame of the
   /// "pretty" stack trace that is dumped when a program crashes. You can define
   /// subclasses of this and declare them on the program stack: when they are
   /// constructed and destructed, they will add their symbolic frames to a
   /// virtual stack trace.  This gets dumped out if the program crashes.
-  class PrettyStackTraceEntry {
+  class LLVM_SUPPORT_ABI PrettyStackTraceEntry {
     friend PrettyStackTraceEntry *ReverseStackTrace(PrettyStackTraceEntry *);
 
     PrettyStackTraceEntry *NextEntry;
     PrettyStackTraceEntry(const PrettyStackTraceEntry &) = delete;
     void operator=(const PrettyStackTraceEntry &) = delete;
   public:
     PrettyStackTraceEntry();
     virtual ~PrettyStackTraceEntry();
 
     /// print - Emit information about this stack frame to OS.
     virtual void print(raw_ostream &OS) const = 0;
 
     /// getNextEntry - Return the next entry in the list of frames.
     const PrettyStackTraceEntry *getNextEntry() const { return NextEntry; }
   };
 
   /// PrettyStackTraceString - This object prints a specified string (which
   /// should not contain newlines) to the stream as the stack trace when a crash
   /// occurs.
-  class PrettyStackTraceString : public PrettyStackTraceEntry {
+  class LLVM_SUPPORT_ABI PrettyStackTraceString : public PrettyStackTraceEntry {
     const char *Str;
   public:
     PrettyStackTraceString(const char *str) : Str(str) {}
     void print(raw_ostream &OS) const override;
   };
 
   /// PrettyStackTraceFormat - This object prints a string (which may use
   /// printf-style formatting but should not contain newlines) to the stream
   /// as the stack trace when a crash occurs.
   class PrettyStackTraceFormat : public PrettyStackTraceEntry {
     llvm::SmallVector<char, 32> Str;
   public:
     PrettyStackTraceFormat(const char *Format, ...);
     void print(raw_ostream &OS) const override;
   };
 
   /// PrettyStackTraceProgram - This object prints a specified program arguments
   /// to the stream as the stack trace when a crash occurs.
-  class PrettyStackTraceProgram : public PrettyStackTraceEntry {
+  class LLVM_SUPPORT_ABI PrettyStackTraceProgram : public PrettyStackTraceEntry {
     int ArgC;
     const char *const *ArgV;
   public:
     PrettyStackTraceProgram(int argc, const char * const*argv)
       : ArgC(argc), ArgV(argv) {
       EnablePrettyStackTrace();
     }
     void print(raw_ostream &OS) const override;
   };
 
   /// Returns the topmost element of the "pretty" stack state.
-  const void *SavePrettyStackState();
+  LLVM_SUPPORT_ABI const void *SavePrettyStackState();
 
   /// Restores the topmost element of the "pretty" stack state to State, which
   /// should come from a previous call to SavePrettyStackState().  This is
   /// useful when using a CrashRecoveryContext in code that also uses
   /// PrettyStackTraceEntries, to make sure the stack that's printed if a crash
   /// happens after a crash that's been recovered by CrashRecoveryContext
   /// doesn't have frames on it that were added in code unwound by the
   /// CrashRecoveryContext.
-  void RestorePrettyStackState(const void *State);
+  LLVM_SUPPORT_ABI void RestorePrettyStackState(const void *State);
 
 } // end namespace llvm
 
 #endif

llvm/include/llvm/Support/Process.h

 #ifndef LLVM_SUPPORT_PROCESS_H
 #define LLVM_SUPPORT_PROCESS_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/Support/AllocatorBase.h"
 #include "llvm/Support/Chrono.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/Program.h"
+
 #include <system_error>
 
 namespace llvm {
 template <typename T> class ArrayRef;
 class StringRef;
 
 namespace sys {
 
 
 /// A collection of legacy interfaces for querying information about the
 /// current executing process.
-class Process {
+class LLVM_SUPPORT_ABI Process {
 public:
   using Pid = int32_t;
 
   /// Get the process's identifier.
   static Pid getProcessId();
 
   /// Get the process's page size.
   /// This may fail if the underlying syscall returns an error. In most cases,
public:
   /// In that case, the control flow will resume after RunSafely(), like for a
   /// crash, rather than exiting the current process.
   /// Use \arg NoCleanup for calling _exit() instead of exit().
   [[noreturn]] static void Exit(int RetCode, bool NoCleanup = false);
 
 private:
   [[noreturn]] static void ExitNoCleanup(int RetCode);
 };
-
 }
 }
 
 #endif

llvm/include/llvm/Support/Program.h

 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/BitVector.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Config/llvm-config.h"
 #include "llvm/Support/ErrorOr.h"
 #include "llvm/Support/FileSystem.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <chrono>
 #include <system_error>
 
 namespace llvm {
 namespace sys {
 
   /// This is the OS-specific separator for PATH like environment variables:
   // a colon on Unix or a semicolon on Windows.
#endif
   ///
   /// \param Name name of the executable to find. If it contains any system
   ///   slashes, it will be returned as is.
   /// \param Paths optional list of paths to search for \p Name. If empty it
   ///   will use the system PATH environment instead.
   ///
   /// \returns The fully qualified path to the first \p Name in \p Paths if it
   ///   exists. \p Name if \p Name has slashes in it. Otherwise an error.
-  ErrorOr<std::string>
+  LLVM_SUPPORT_ABI ErrorOr<std::string>
   findProgramByName(StringRef Name, ArrayRef<StringRef> Paths = {});
 
   // These functions change the specified standard stream (stdin or stdout) mode
   // based on the Flags. They return errc::success if the specified stream was
   // changed. Otherwise, a platform dependent error is returned.
   std::error_code ChangeStdinMode(fs::OpenFlags Flags);
   std::error_code ChangeStdoutMode(fs::OpenFlags Flags);
 
   // These functions change the specified standard stream (stdin or stdout) to
   // binary mode. They return errc::success if the specified stream
   // was changed. Otherwise a platform dependent error is returned.
-  std::error_code ChangeStdinToBinary();
+  LLVM_SUPPORT_ABI std::error_code ChangeStdinToBinary();
   std::error_code ChangeStdoutToBinary();
 
   /// This function executes the program using the arguments provided.  The
   /// invoked program will inherit the stdin, stdout, and stderr file
   /// descriptors, the environment and other configuration settings of the
   /// invoking program.
   /// This function waits for the program to finish, so should be avoided in
   /// library functions that aren't expected to block. Consider using
   /// ExecuteNoWait() instead.
   /// \returns an integer result code indicating the status of the program.
   /// A zero or positive value indicates the result code of the program.
   /// -1 indicates failure to execute
   /// -2 indicates a crash during execution or timeout
+  LLVM_SUPPORT_ABI
   int ExecuteAndWait(
       StringRef Program, ///< Path of the program to be executed. It is
       ///< presumed this is the result of the findProgramByName method.
       ArrayRef<StringRef> Args, ///< An array of strings that are passed to the
       ///< program.  The first element should be the name of the program.
       ///< The array should **not** be terminated by an empty StringRef.
       Optional<ArrayRef<StringRef>> Env = None, ///< An optional vector of
       ///< strings to use for the program's environment. If not provided, the
int ExecuteAndWait(
                                         /// program shall run on.
   );
 
   /// Similar to ExecuteAndWait, but returns immediately.
   /// @returns The \see ProcessInfo of the newly launched process.
   /// \note On Microsoft Windows systems, users will need to either call
   /// \see Wait until the process finished execution or win32 CloseHandle() API
   /// on ProcessInfo.ProcessHandle to avoid memory leaks.
-  ProcessInfo ExecuteNoWait(StringRef Program, ArrayRef<StringRef> Args,
+  LLVM_SUPPORT_ABI ProcessInfo ExecuteNoWait(StringRef Program, ArrayRef<StringRef> Args,
                             Optional<ArrayRef<StringRef>> Env,
                             ArrayRef<Optional<StringRef>> Redirects = {},
                             unsigned MemoryLimit = 0,
                             std::string *ErrMsg = nullptr,
                             bool *ExecutionFailed = nullptr,
                             BitVector *AffinityMask = nullptr);
 
   /// Return true if the given arguments fit within system-specific
   /// argument length limits.
-  bool commandLineFitsWithinSystemLimits(StringRef Program,
+  LLVM_SUPPORT_ABI bool commandLineFitsWithinSystemLimits(StringRef Program,
                                          ArrayRef<StringRef> Args);
 
   /// Return true if the given arguments fit within system-specific
   /// argument length limits.
-  bool commandLineFitsWithinSystemLimits(StringRef Program,
+  LLVM_SUPPORT_ABI bool commandLineFitsWithinSystemLimits(StringRef Program,
                                          ArrayRef<const char *> Args);
 
   /// File encoding options when writing contents that a non-UTF8 tool will
   /// read (on Windows systems). For UNIX, we always use UTF-8.
   enum WindowsEncodingMethod {
     /// UTF-8 is the LLVM native encoding, being the same as "do not perform
     /// encoding conversion".
     WEM_UTF8,
#endif
   /// it to your desired encoding before writing to the file.
   ///
   /// FIXME: We use EM_CurrentCodePage to write response files for GNU tools in
   /// a MinGW/MinGW-w64 environment, which has serious flaws but currently is
   /// our best shot to make gcc/ld understand international characters. This
   /// should be changed as soon as binutils fix this to support UTF16 on mingw.
   ///
   /// \returns non-zero error_code if failed
-  std::error_code
+  LLVM_SUPPORT_ABI std::error_code
   writeFileWithEncoding(StringRef FileName, StringRef Contents,
                         WindowsEncodingMethod Encoding = WEM_UTF8);
 
   /// This function waits for the process specified by \p PI to finish.
   /// \returns A \see ProcessInfo struct with Pid set to:
   /// \li The process id of the child process if the child process has changed
   /// state.
   /// \li 0 if the child process has not changed state.
   /// \note Users of this function should always check the ReturnCode member of
   /// the \see ProcessInfo returned from this function.
-  ProcessInfo Wait(
+  LLVM_SUPPORT_ABI ProcessInfo Wait(
       const ProcessInfo &PI,  ///< The child process that should be waited on.
       unsigned SecondsToWait, ///< If non-zero, this specifies the amount of
       ///< time to wait for the child process to exit. If the time expires, the
       ///< child is killed and this function returns. If zero, this function
       ///< will perform a non-blocking wait on the child process.
       bool WaitUntilTerminates, ///< If true, ignores \p SecondsToWait and waits
       ///< until child has terminated.
       std::string *ErrMsg = nullptr, ///< If non-zero, provides a pointer to a
       ///< string instance in which error messages will be returned. If the
       ///< string is non-empty upon return an error occurred while invoking the
       ///< program.
       Optional<ProcessStatistics> *ProcStat = nullptr ///< If non-zero, provides
       /// a pointer to a structure in which process execution statistics will be
       /// stored.
   );
 
   /// Print a command argument, and optionally quote it.
-  void printArg(llvm::raw_ostream &OS, StringRef Arg, bool Quote);
+  LLVM_SUPPORT_ABI void printArg(llvm::raw_ostream &OS, StringRef Arg, bool Quote);
 
 #if defined(_WIN32)
   /// Given a list of command line arguments, quote and escape them as necessary
   /// to build a single flat command line appropriate for calling CreateProcess
   /// on
   /// Windows.
   ErrorOr<std::wstring> flattenWindowsCommandLine(ArrayRef<StringRef> Args);
 #endif
   }
 }
 
 #endif

llvm/include/llvm/Support/RISCVAttributeParser.h

 //===-- RISCVAttributeParser.h - RISCV Attribute Parser ---------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_RISCVATTRIBUTEPARSER_H
 #define LLVM_SUPPORT_RISCVATTRIBUTEPARSER_H
 
 #include "llvm/Support/ELFAttributeParser.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/RISCVAttributes.h"
 
 namespace llvm {
-class RISCVAttributeParser : public ELFAttributeParser {
+class LLVM_SUPPORT_ABI RISCVAttributeParser : public ELFAttributeParser {
   struct DisplayHandler {
     RISCVAttrs::AttrType attribute;
     Error (RISCVAttributeParser::*routine)(unsigned);
   };
   static const DisplayHandler displayRoutines[];
 
   Error handler(uint64_t tag, bool &handled) override;
 

llvm/include/llvm/Support/RISCVAttributes.h

 //
 // https://github.com/riscv/riscv-elf-psabi-doc/blob/master/riscv-elf.md
 //
 //===----------------------------------------------------------------------===//
 #ifndef LLVM_SUPPORT_RISCVATTRIBUTES_H
 #define LLVM_SUPPORT_RISCVATTRIBUTES_H
 
 #include "llvm/Support/ELFAttributes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 namespace RISCVAttrs {
 
-const TagNameMap &getRISCVAttributeTags();
+LLVM_SUPPORT_ABI const TagNameMap &getRISCVAttributeTags();
 
 enum AttrType : unsigned {
   // Attribute types in ELF/.riscv.attributes.
   STACK_ALIGN = 4,
   ARCH = 5,
   UNALIGNED_ACCESS = 6,
   PRIV_SPEC = 8,
   PRIV_SPEC_MINOR = 10,

llvm/include/llvm/Support/RandomNumberGenerator.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_RANDOMNUMBERGENERATOR_H_
 #define LLVM_SUPPORT_RANDOMNUMBERGENERATOR_H_
 
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/DataTypes.h" // Needed for uint64_t on Windows.
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <random>
 #include <system_error>
 
 namespace llvm {
 class StringRef;
 
 /// A random number generator.
 ///
 /// Instances of this class should not be shared across threads. The
 /// seed should be set by passing the -rng-seed=<uint64> option. Use
 /// Module::createRNG to create a new RNG instance for use with that
 /// module.
-class RandomNumberGenerator {
+class LLVM_SUPPORT_ABI RandomNumberGenerator {
 
   // 64-bit Mersenne Twister by Matsumoto and Nishimura, 2000
   // http://en.cppreference.com/w/cpp/numeric/random/mersenne_twister_engine
   // This RNG is deterministically portable across C++11
   // implementations.
   using generator_type = std::mt19937_64;
 
 public:

llvm/include/llvm/Support/Regex.h

 // This implementation also supports matching strings with embedded NUL chars.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_REGEX_H
 #define LLVM_SUPPORT_REGEX_H
 
 #include "llvm/ADT/BitmaskEnum.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 
 struct llvm_regex;
 
 namespace llvm {
   class StringRef;
   template<typename T> class SmallVectorImpl;
 
-  class Regex {
+  class LLVM_SUPPORT_ABI Regex {
   public:
     enum RegexFlags : unsigned {
       NoFlags = 0,
       /// Compile for matching that ignores upper/lower case distinctions.
       IgnoreCase = 1,
       /// Compile for newline-sensitive matching. With this flag '[^' bracket
       /// expressions and '.' never match newline. A ^ anchor matches the
       /// null string after any newline in the string in addition to its normal

llvm/include/llvm/Support/SHA1.h

 // and modified by wrapping it in a C++ interface for LLVM,
 // and removing unnecessary code.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SHA1_H
 #define LLVM_SUPPORT_SHA1_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <array>
 #include <cstdint>
 
 namespace llvm {
 template <typename T> class ArrayRef;
 class StringRef;
 
 /// A class that wrap the SHA1 algorithm.
-class SHA1 {
+class LLVM_SUPPORT_ABI SHA1 {
 public:
   SHA1() { init(); }
 
   /// Reinitialize the internal state
   void init();
 
   /// Digest more data.
   void update(ArrayRef<uint8_t> Data);

llvm/include/llvm/Support/SHA256.h

  *       code)
  *   [1] llvm/lib/Support/SHA1.{h,cpp}
  */
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SHA256_H
 #define LLVM_SUPPORT_SHA256_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <array>
 #include <cstdint>
 
 namespace llvm {
 
 template <typename T> class ArrayRef;
 class StringRef;
 
-class SHA256 {
+class LLVM_SUPPORT_ABI SHA256 {
 public:
   explicit SHA256() { init(); }
 
   /// Reinitialize the internal state
   void init();
 
   /// Digest more data.
   void update(ArrayRef<uint8_t> Data);

llvm/include/llvm/Support/SMTAPI.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SMTAPI_H
 #define LLVM_SUPPORT_SMTAPI_H
 
 #include "llvm/ADT/APFloat.h"
 #include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/FoldingSet.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <memory>
 
 namespace llvm {
 
 /// Generic base class for SMT sorts
 class SMTSort {
 public:
   SMTSort() = default;
public:
 
   virtual void print(raw_ostream &OS) const = 0;
 };
 
 /// Shared pointer for SMTSolvers.
 using SMTSolverRef = std::shared_ptr<SMTSolver>;
 
 /// Convenience method to create and Z3Solver object
-SMTSolverRef CreateZ3Solver();
+LLVM_SUPPORT_ABI SMTSolverRef CreateZ3Solver();
 
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/ScaledNumber.h

 // These might remind you of soft-floats.  If you want one of those, you're in
 // the wrong place.  Look at include/llvm/ADT/APFloat.h instead.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SCALEDNUMBER_H
 #define LLVM_SUPPORT_SCALEDNUMBER_H
 
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MathExtras.h"
+
 #include <algorithm>
 #include <cstdint>
 #include <limits>
 #include <string>
 #include <tuple>
 #include <utility>
 
 namespace llvm {

 inline std::pair<uint64_t, int16_t> getAdjusted64(uint64_t Digits,
                                                   int16_t Scale = 0) {
   return getAdjusted<uint64_t>(Digits, Scale);
 }
 
 /// Multiply two 64-bit integers to create a 64-bit scaled number.
 ///
 /// Implemented with four 64-bit integer multiplies.
-std::pair<uint64_t, int16_t> multiply64(uint64_t LHS, uint64_t RHS);
+LLVM_SUPPORT_ABI std::pair<uint64_t, int16_t> multiply64(uint64_t LHS, uint64_t RHS);
 
 /// Multiply two 32-bit integers to create a 32-bit scaled number.
 ///
 /// Implemented with one 64-bit integer multiply.
 template <class DigitsT>
 inline std::pair<DigitsT, int16_t> getProduct(DigitsT LHS, DigitsT RHS) {
   static_assert(!std::numeric_limits<DigitsT>::is_signed, "expected unsigned");
 
inline std::pair<uint64_t, int16_t> getProduct64(uint64_t LHS, uint64_t RHS) {
   return getProduct(LHS, RHS);
 }
 
 /// Divide two 64-bit integers to create a 64-bit scaled number.
 ///
 /// Implemented with long division.
 ///
 /// \pre \c Dividend and \c Divisor are non-zero.
-std::pair<uint64_t, int16_t> divide64(uint64_t Dividend, uint64_t Divisor);
+LLVM_SUPPORT_ABI std::pair<uint64_t, int16_t> divide64(uint64_t Dividend, uint64_t Divisor);
 
 /// Divide two 32-bit integers to create a 32-bit scaled number.
 ///
 /// Implemented with one 64-bit integer divide/remainder pair.
 ///
 /// \pre \c Dividend and \c Divisor are non-zero.
-std::pair<uint32_t, int16_t> divide32(uint32_t Dividend, uint32_t Divisor);
+LLVM_SUPPORT_ABI std::pair<uint32_t, int16_t> divide32(uint32_t Dividend, uint32_t Divisor);
 
 /// Divide two 32-bit numbers to create a 32-bit scaled number.
 ///
 /// Implemented with one 64-bit integer divide/remainder pair.
 ///
 /// Returns \c (DigitsT_MAX, MaxScale) for divide-by-zero (0 for 0/0).
 template <class DigitsT>
 std::pair<DigitsT, int16_t> getQuotient(DigitsT Dividend, DigitsT Divisor) {

 
 /// Implementation for comparing scaled numbers.
 ///
 /// Compare two 64-bit numbers with different scales.  Given that the scale of
 /// \c L is higher than that of \c R by \c ScaleDiff, compare them.  Return -1,
 /// 1, and 0 for less than, greater than, and equal, respectively.
 ///
 /// \pre 0 <= ScaleDiff < 64.
-int compareImpl(uint64_t L, uint64_t R, int ScaleDiff);
+LLVM_SUPPORT_ABI int compareImpl(uint64_t L, uint64_t R, int ScaleDiff);
 
 /// Compare two scaled numbers.
 ///
 /// Compare two scaled numbers.  Returns 0 for equal, -1 for less than, and 1
 /// for greater than.
 template <class DigitsT>
 int compare(DigitsT LDigits, int16_t LScale, DigitsT RDigits, int16_t RScale) {
   static_assert(!std::numeric_limits<DigitsT>::is_signed, "expected unsigned");

 }
 
 } // end namespace ScaledNumbers
 } // end namespace llvm
 
 namespace llvm {
 
 class raw_ostream;
-class ScaledNumberBase {
+class LLVM_SUPPORT_ABI ScaledNumberBase {
 public:
   static constexpr int DefaultPrecision = 10;
 
   static void dump(uint64_t D, int16_t E, int Width);
   static raw_ostream &print(raw_ostream &OS, uint64_t D, int16_t E, int Width,
                             unsigned Precision);
   static std::string toString(uint64_t D, int16_t E, int Width,
                               unsigned Precision);

llvm/include/llvm/Support/ScopedPrinter.h

 
 #include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Support/Endian.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <algorithm>
 
 namespace llvm {
 
 template <typename T> struct EnumEntry {
   StringRef Name;
   // While Name suffices in most of the cases, in certain cases
   // GNU style and LLVM style of ELFDumper do not
struct HexNumber {
   HexNumber(unsigned char Value) : Value(Value) {}
   HexNumber(unsigned short Value) : Value(Value) {}
   HexNumber(unsigned int Value) : Value(Value) {}
   HexNumber(unsigned long Value) : Value(Value) {}
   HexNumber(unsigned long long Value) : Value(Value) {}
   uint64_t Value;
 };
 
-raw_ostream &operator<<(raw_ostream &OS, const HexNumber &Value);
-std::string to_hexString(uint64_t Value, bool UpperCase = true);
+LLVM_SUPPORT_ABI raw_ostream &operator<<(raw_ostream &OS, const HexNumber &Value);
+
+LLVM_SUPPORT_ABI std::string to_hexString(uint64_t Value, bool UpperCase = true);
 
 template <class T> std::string to_string(const T &Value) {
   std::string number;
   llvm::raw_string_ostream stream(number);
   stream << Value;
   return stream.str();
 }
 
-class ScopedPrinter {
+class LLVM_SUPPORT_ABI ScopedPrinter {
 public:
   ScopedPrinter(raw_ostream &OS) : OS(OS), IndentLevel(0) {}
 
   void flush() { OS.flush(); }
 
   void indent(int Levels = 1) { IndentLevel += Levels; }
 
   void unindent(int Levels = 1) {

llvm/include/llvm/Support/Signals.h

 //===- llvm/Support/Signals.h - Signal Handling support ----------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines some helpful functions for dealing with the possibility of
 // unix signals occurring while your program is running.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SIGNALS_H
 #define LLVM_SUPPORT_SIGNALS_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 
 namespace llvm {
 class StringRef;
 class raw_ostream;
 
 namespace sys {
 
   /// This function runs all the registered interrupt handlers, including the
   /// removal of files registered by RemoveFileOnSignal.
-  void RunInterruptHandlers();
+  LLVM_SUPPORT_ABI void RunInterruptHandlers();
 
   /// This function registers signal handlers to ensure that if a signal gets
   /// delivered that the named file is removed.
   /// Remove a file if a fatal signal occurs.
-  bool RemoveFileOnSignal(StringRef Filename, std::string* ErrMsg = nullptr);
+  LLVM_SUPPORT_ABI bool RemoveFileOnSignal(StringRef Filename, std::string* ErrMsg = nullptr);
 
   /// This function removes a file from the list of files to be removed on
   /// signal delivery.
   void DontRemoveFileOnSignal(StringRef Filename);
 
   /// When an error signal (such as SIGABRT or SIGSEGV) is delivered to the
   /// process, print a stack trace and then exit.
   /// Print a stack trace if a fatal signal occurs.
   /// \param Argv0 the current binary name, used to find the symbolizer
   ///        relative to the current binary before searching $PATH; can be
   ///        StringRef(), in which case we will only search $PATH.
   /// \param DisableCrashReporting if \c true, disable the normal crash
   ///        reporting mechanisms on the underlying operating system.
-  void PrintStackTraceOnErrorSignal(StringRef Argv0,
+  LLVM_SUPPORT_ABI void PrintStackTraceOnErrorSignal(StringRef Argv0,
                                     bool DisableCrashReporting = false);
 
   /// Disable all system dialog boxes that appear when the process crashes.
-  void DisableSystemDialogsOnCrash();
+  LLVM_SUPPORT_ABI void DisableSystemDialogsOnCrash();
 
   /// Print the stack trace using the given \c raw_ostream object.
   /// \param Depth refers to the number of stackframes to print. If not
   ///        specified, the entire frame is printed.
-  void PrintStackTrace(raw_ostream &OS, int Depth = 0);
+  LLVM_SUPPORT_ABI void PrintStackTrace(raw_ostream &OS, int Depth = 0);
 
   // Run all registered signal handlers.
   void RunSignalHandlers();
 
   using SignalHandlerCallback = void (*)(void *);
 
   /// Add a function to be called when an abort/kill signal is delivered to the
   /// process. The handler can have a cookie passed to it to identify what
   /// instance of the handler it is.
-  void AddSignalHandler(SignalHandlerCallback FnPtr, void *Cookie);
+  LLVM_SUPPORT_ABI void AddSignalHandler(SignalHandlerCallback FnPtr, void *Cookie);
 
   /// This function registers a function to be called when the user "interrupts"
   /// the program (typically by pressing ctrl-c).  When the user interrupts the
   /// program, the specified interrupt function is called instead of the program
   /// being killed, and the interrupt function automatically disabled.
   ///
   /// Note that interrupt functions are not allowed to call any non-reentrant
   /// functions.  An null interrupt function pointer disables the current
   /// installed function.  Note also that the handler may be executed on a
   /// different thread on some platforms.
-  void SetInterruptFunction(void (*IF)());
+  LLVM_SUPPORT_ABI void SetInterruptFunction(void (*IF)());
 
   /// Registers a function to be called when an "info" signal is delivered to
   /// the process.
   ///
   /// On POSIX systems, this will be SIGUSR1; on systems that have it, SIGINFO
   /// will also be used (typically ctrl-t).
   ///
   /// Note that signal handlers are not allowed to call any non-reentrant

llvm/include/llvm/Support/SmallVectorMemoryBuffer.h

 // object will be generated.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SMALLVECTORMEMORYBUFFER_H
 #define LLVM_SUPPORT_SMALLVECTORMEMORYBUFFER_H
 
 #include "llvm/ADT/SmallVector.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MemoryBuffer.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 /// SmallVector-backed MemoryBuffer instance.
 ///
 /// This class enables efficient construction of MemoryBuffers from SmallVector
 /// instances. This is useful for MCJIT and Orc, where object files are streamed
 /// into SmallVectors, then inspected using ObjectFile (which takes a
 /// MemoryBuffer).
-class SmallVectorMemoryBuffer : public MemoryBuffer {
+class LLVM_SUPPORT_ABI SmallVectorMemoryBuffer : public MemoryBuffer {
 public:
   /// Construct an SmallVectorMemoryBuffer from the given SmallVector
   /// r-value.
   ///
   /// FIXME: It'd be nice for this to be a non-templated constructor taking a
   /// SmallVectorImpl here instead of a templated one taking a SmallVector<N>,
   /// but SmallVector's move-construction/assignment currently only take
   /// SmallVectors. If/when that is fixed we can simplify this constructor and

llvm/include/llvm/Support/SourceMgr.h

 // level things for simple parsers.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SOURCEMGR_H
 #define LLVM_SUPPORT_SOURCEMGR_H
 
 #include "llvm/ADT/SmallVector.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MemoryBuffer.h"
 #include "llvm/Support/SMLoc.h"
+
 #include <vector>
 
 namespace llvm {
 
 class raw_ostream;
 class SMDiagnostic;
 class SMFixIt;
 
 /// This owns the files read by a parser, handles include stacks,
 /// and handles diagnostic wrangling.
-class SourceMgr {
+class LLVM_SUPPORT_ABI SourceMgr {
 public:
   enum DiagKind {
     DK_Error,
     DK_Warning,
     DK_Remark,
     DK_Note,
   };
 
public:
   /// custom formatted message.
   ///
   /// \param IncludeLoc The location of the include.
   /// \param OS the raw_ostream to print on.
   void PrintIncludeStack(SMLoc IncludeLoc, raw_ostream &OS) const;
 };
 
 /// Represents a single fixit, a replacement of one range of text with another.
-class SMFixIt {
+class LLVM_SUPPORT_ABI SMFixIt {
   SMRange Range;
 
   std::string Text;
 
 public:
   SMFixIt(SMRange R, const Twine &Replacement);
 
   SMFixIt(SMLoc Loc, const Twine &Replacement)
       : SMFixIt(SMRange(Loc, Loc), Replacement) {}
 
   StringRef getText() const { return Text; }
   SMRange getRange() const { return Range; }
 
   bool operator<(const SMFixIt &Other) const {
     if (Range.Start.getPointer() != Other.Range.Start.getPointer())
       return Range.Start.getPointer() < Other.Range.Start.getPointer();
     if (Range.End.getPointer() != Other.Range.End.getPointer())
       return Range.End.getPointer() < Other.Range.End.getPointer();
     return Text < Other.Text;
   }
 };
 
 /// Instances of this class encapsulate one diagnostic report, allowing
 /// printing to a raw_ostream as a caret diagnostic.
-class SMDiagnostic {
+class LLVM_SUPPORT_ABI SMDiagnostic {
   const SourceMgr *SM = nullptr;
   SMLoc Loc;
   std::string Filename;
   int LineNo = 0;
   int ColumnNo = 0;
   SourceMgr::DiagKind Kind = SourceMgr::DK_Error;
   std::string Message, LineContents;
   std::vector<std::pair<unsigned, unsigned>> Ranges;

llvm/include/llvm/Support/SpecialCaseList.h

 // replaced with .*
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SPECIALCASELIST_H
 #define LLVM_SUPPORT_SPECIALCASELIST_H
 
 #include "llvm/ADT/StringMap.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/Regex.h"
 #include "llvm/Support/TrigramIndex.h"
+
 #include <memory>
 #include <string>
 #include <vector>
 
 namespace llvm {
 class MemoryBuffer;
 class StringRef;
 
 namespace vfs {
 class FileSystem;
 }
 
-class SpecialCaseList {
+class LLVM_SUPPORT_ABI SpecialCaseList {
 public:
   /// Parses the special case list entries from files. On failure, returns
   /// 0 and writes an error message to string.
   static std::unique_ptr<SpecialCaseList>
   create(const std::vector<std::string> &Paths, llvm::vfs::FileSystem &FS,
          std::string &Error);
   /// Parses the special case list from a memory buffer. On failure, returns
   /// 0 and writes an error message to string.
protected:
   SpecialCaseList &operator=(SpecialCaseList const &) = delete;
 
   /// Represents a set of regular expressions.  Regular expressions which are
   /// "literal" (i.e. no regex metacharacters) are stored in Strings.  The
   /// reason for doing so is efficiency; StringMap is much faster at matching
   /// literal strings than Regex.
   class Matcher {
   public:
-    bool insert(std::string Regexp, unsigned LineNumber, std::string &REError);
+    LLVM_SUPPORT_ABI bool insert(std::string Regexp, unsigned LineNumber, std::string &REError);
     // Returns the line number in the source file that this query matches to.
     // Returns zero if no match is found.
-    unsigned match(StringRef Query) const;
+    LLVM_SUPPORT_ABI unsigned match(StringRef Query) const;
 
   private:
     StringMap<unsigned> Strings;
     TrigramIndex Trigrams;
     std::vector<std::pair<std::unique_ptr<Regex>, unsigned>> RegExes;
   };
 
   using SectionEntries = StringMap<StringMap<Matcher>>;

llvm/include/llvm/Support/StringSaver.h

 //===- llvm/Support/StringSaver.h -------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_STRINGSAVER_H
 #define LLVM_SUPPORT_STRINGSAVER_H
 
 #include "llvm/ADT/DenseSet.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 
 /// Saves strings in the provided stable storage and returns a
 /// StringRef with a stable character pointer.
-class StringSaver final {
+class LLVM_SUPPORT_ABI StringSaver final {
   BumpPtrAllocator &Alloc;
 
 public:
   StringSaver(BumpPtrAllocator &Alloc) : Alloc(Alloc) {}
 
   // All returned strings are null-terminated: *save(S).end() == 0.
   StringRef save(const char *S) { return save(StringRef(S)); }
   StringRef save(StringRef S);
   StringRef save(const Twine &S) { return save(StringRef(S.str())); }
   StringRef save(const std::string &S) { return save(StringRef(S)); }
 };
 
 /// Saves strings in the provided stable storage and returns a StringRef with a
 /// stable character pointer. Saving the same string yields the same StringRef.
 ///
 /// Compared to StringSaver, it does more work but avoids saving the same string
 /// multiple times.
 ///
 /// Compared to StringPool, it performs fewer allocations but doesn't support
 /// refcounting/deletion.
-class UniqueStringSaver final {
+class LLVM_SUPPORT_ABI UniqueStringSaver final {
   StringSaver Strings;
   llvm::DenseSet<llvm::StringRef> Unique;
 
 public:
   UniqueStringSaver(BumpPtrAllocator &Alloc) : Strings(Alloc) {}
 
   // All returned strings are null-terminated: *save(S).end() == 0.
   StringRef save(const char *S) { return save(StringRef(S)); }
   StringRef save(StringRef S);
   StringRef save(const Twine &S) { return save(StringRef(S.str())); }
   StringRef save(const std::string &S) { return save(StringRef(S)); }
 };
 
 }
 #endif

llvm/include/llvm/Support/SuffixTree.h

 //
 //===----------------------------------------------------------------------===//
 #ifndef LLVM_SUPPORT_SUFFIXTREE_H
 #define LLVM_SUPPORT_SUFFIXTREE_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <vector>
 
 namespace llvm {
 
 /// Represents an undefined index in the suffix tree.
 const unsigned EmptyIdx = -1;
 
 /// A node in a suffix tree which represents a substring or suffix.

 /// contain 1 or many strings, which can then be queried as one large string.
 ///
 /// The suffix tree is implemented using Ukkonen's algorithm for linear-time
 /// suffix tree construction. Ukkonen's algorithm is explained in more detail
 /// in the paper by Esko Ukkonen "On-line construction of suffix trees. The
 /// paper is available at
 ///
 /// https://www.cs.helsinki.fi/u/ukkonen/SuffixT1withFigs.pdf
-class SuffixTree {
+class LLVM_SUPPORT_ABI SuffixTree {
 public:
   /// Each element is an integer representing an instruction in the module.
   llvm::ArrayRef<unsigned> Str;
 
   /// A repeated substring in the tree.
   struct RepeatedSubstring {
     /// The length of the string.
     unsigned Length;

llvm/include/llvm/Support/SymbolRemappingReader.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SYMBOLREMAPPINGREADER_H
 #define LLVM_SUPPORT_SYMBOLREMAPPINGREADER_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Error.h"
 #include "llvm/Support/ItaniumManglingCanonicalizer.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MemoryBuffer.h"
 
 namespace llvm {
 
-class SymbolRemappingParseError : public ErrorInfo<SymbolRemappingParseError> {
+class LLVM_SUPPORT_ABI SymbolRemappingParseError : public ErrorInfo<SymbolRemappingParseError> {
 public:
   SymbolRemappingParseError(StringRef File, int64_t Line, const Twine &Message)
       : File(File), Line(Line), Message(Message.str()) {}
 
   void log(llvm::raw_ostream &OS) const override {
     OS << File << ':' << Line << ": " << Message;
   }
   std::error_code convertToErrorCode() const override {
private:
   int64_t Line;
   std::string Message;
 };
 
 /// Reader for symbol remapping files.
 ///
 /// Remaps the symbol names in profile data to match those in the program
 /// according to a set of rules specified in a given file.
-class SymbolRemappingReader {
+class LLVM_SUPPORT_ABI SymbolRemappingReader {
 public:
   /// Read remappings from the given buffer, which must live as long as
   /// the remapper.
   Error read(MemoryBuffer &B);
 
   /// A Key represents an equivalence class of symbol names.
   using Key = uintptr_t;
 

llvm/include/llvm/Support/SystemUtils.h

 //===- SystemUtils.h - Utilities to do low-level system stuff ---*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains functions used to do a variety of low-level, often
 // system-specific, tasks.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_SYSTEMUTILS_H
 #define LLVM_SUPPORT_SYSTEMUTILS_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 namespace llvm {
 class raw_ostream;
 
 /// Determine if the raw_ostream provided is connected to a terminal. If so,
 /// generate a warning message to errs() advising against display of bitcode
 /// and return true. Otherwise just return false.
 /// Check for output written to a console
+LLVM_SUPPORT_ABI
 bool CheckBitcodeOutputToConsole(
     raw_ostream &stream_to_check ///< The stream to be checked
 );
 
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/TarWriter.h

 //===-- llvm/Support/TarWriter.h - Tar archive file creator -----*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TARWRITER_H
 #define LLVM_SUPPORT_TARWRITER_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/StringSet.h"
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
-class TarWriter {
+class LLVM_SUPPORT_ABI TarWriter {
 public:
   static Expected<std::unique_ptr<TarWriter>> create(StringRef OutputPath,
                                                      StringRef BaseDir);
 
   void append(StringRef Path, StringRef Data);
 
 private:
   TarWriter(int FD, StringRef BaseDir);
   raw_fd_ostream OS;
   std::string BaseDir;
   StringSet<> Files;
 };
 }
 
 #endif

llvm/include/llvm/Support/TargetParser.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TARGETPARSER_H
 #define LLVM_SUPPORT_TARGETPARSER_H
 
 // FIXME: vector is used because that's what clang uses for subtarget feature
 // lists, but SmallVector would probably be better
 #include "llvm/ADT/Triple.h"
 #include "llvm/Support/ARMTargetParser.h"
 #include "llvm/Support/AArch64TargetParser.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <vector>
 
 namespace llvm {
 class StringRef;
 
 // Target specific information in their own namespaces.
 // (ARM/AArch64/X86 are declared in ARM/AArch64/X86TargetParser.h)
 // These should be generated from TableGen because the information is already
enum ArchFeatureKind : uint32_t {
 
   // Xnack is available.
   FEATURE_XNACK = 1 << 7,
 
   // Sram-ecc is available.
   FEATURE_SRAMECC = 1 << 8,
 };
 
-StringRef getArchNameAMDGCN(GPUKind AK);
-StringRef getArchNameR600(GPUKind AK);
-StringRef getCanonicalArchName(const Triple &T, StringRef Arch);
-GPUKind parseArchAMDGCN(StringRef CPU);
-GPUKind parseArchR600(StringRef CPU);
-unsigned getArchAttrAMDGCN(GPUKind AK);
-unsigned getArchAttrR600(GPUKind AK);
+LLVM_SUPPORT_ABI StringRef getArchNameAMDGCN(GPUKind AK);
+LLVM_SUPPORT_ABI StringRef getArchNameR600(GPUKind AK);
+LLVM_SUPPORT_ABI StringRef getCanonicalArchName(const Triple &T, StringRef Arch);
+LLVM_SUPPORT_ABI GPUKind parseArchAMDGCN(StringRef CPU);
+LLVM_SUPPORT_ABI GPUKind parseArchR600(StringRef CPU);
+LLVM_SUPPORT_ABI unsigned getArchAttrAMDGCN(GPUKind AK);
+LLVM_SUPPORT_ABI unsigned getArchAttrR600(GPUKind AK);
 
-void fillValidArchListAMDGCN(SmallVectorImpl<StringRef> &Values);
-void fillValidArchListR600(SmallVectorImpl<StringRef> &Values);
+LLVM_SUPPORT_ABI void fillValidArchListAMDGCN(SmallVectorImpl<StringRef> &Values);
+LLVM_SUPPORT_ABI void fillValidArchListR600(SmallVectorImpl<StringRef> &Values);
 
-IsaVersion getIsaVersion(StringRef GPU);
+LLVM_SUPPORT_ABI IsaVersion getIsaVersion(StringRef GPU);
 
 } // namespace AMDGPU
 
 namespace RISCV {
 
 enum CPUKind : unsigned {
 #define PROC(ENUM, NAME, FEATURES, DEFAULT_MARCH) CK_##ENUM,
 #include "RISCVTargetParser.def"
 };
 
 enum FeatureKind : unsigned {
   FK_INVALID = 0,
   FK_NONE = 1,
   FK_STDEXTM = 1 << 2,
   FK_STDEXTA = 1 << 3,
   FK_STDEXTF = 1 << 4,
   FK_STDEXTD = 1 << 5,
   FK_STDEXTC = 1 << 6,
   FK_64BIT = 1 << 7,
 };
 
-bool checkCPUKind(CPUKind Kind, bool IsRV64);
-bool checkTuneCPUKind(CPUKind Kind, bool IsRV64);
-CPUKind parseCPUKind(StringRef CPU);
-CPUKind parseTuneCPUKind(StringRef CPU, bool IsRV64);
-StringRef getMArchFromMcpu(StringRef CPU);
-void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values, bool IsRV64);
-void fillValidTuneCPUArchList(SmallVectorImpl<StringRef> &Values, bool IsRV64);
-bool getCPUFeaturesExceptStdExt(CPUKind Kind, std::vector<StringRef> &Features);
-StringRef resolveTuneCPUAlias(StringRef TuneCPU, bool IsRV64);
+LLVM_SUPPORT_ABI bool checkCPUKind(CPUKind Kind, bool IsRV64);
+LLVM_SUPPORT_ABI bool checkTuneCPUKind(CPUKind Kind, bool IsRV64);
+LLVM_SUPPORT_ABI CPUKind parseCPUKind(StringRef CPU);
+LLVM_SUPPORT_ABI CPUKind parseTuneCPUKind(StringRef CPU, bool IsRV64);
+LLVM_SUPPORT_ABI StringRef getMArchFromMcpu(StringRef CPU);
+LLVM_SUPPORT_ABI void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values, bool IsRV64);
+LLVM_SUPPORT_ABI void fillValidTuneCPUArchList(SmallVectorImpl<StringRef> &Values, bool IsRV64);
+LLVM_SUPPORT_ABI bool getCPUFeaturesExceptStdExt(CPUKind Kind, std::vector<StringRef> &Features);
+LLVM_SUPPORT_ABI StringRef resolveTuneCPUAlias(StringRef TuneCPU, bool IsRV64);
 
 } // namespace RISCV
 
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/TargetRegistry.h

 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Triple.h"
 #include "llvm/ADT/iterator_range.h"
 #include "llvm/MC/MCObjectFileInfo.h"
 #include "llvm/Support/CodeGen.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/FormattedStream.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <algorithm>
 #include <cassert>
 #include <cstddef>
 #include <iterator>
 #include <memory>
 #include <string>
 
 namespace llvm {
if (InstrPostProcessCtorFn)
       return InstrPostProcessCtorFn(STI, MCII);
     return nullptr;
   }
 
   /// @}
 };
 
 /// TargetRegistry - Generic interface to target specific features.
-struct TargetRegistry {
+struct LLVM_SUPPORT_ABI TargetRegistry {
   // FIXME: Make this a namespace, probably just move all the Register*
   // functions into Target (currently they all just set members on the Target
   // anyway, and Target friends this class so those functions can...
   // function).
   TargetRegistry() = delete;
 
   class iterator {
     friend struct TargetRegistry;

llvm/include/llvm/Support/ThreadLocal.h

 //===- llvm/Support/ThreadLocal.h - Thread Local Data ------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file declares the llvm::sys::ThreadLocal class.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_THREADLOCAL_H
 #define LLVM_SUPPORT_THREADLOCAL_H
 
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/Threading.h"
+
 #include <cassert>
 
 namespace llvm {
   namespace sys {
     // ThreadLocalImpl - Common base class of all ThreadLocal instantiations.
     // YOU SHOULD NEVER USE THIS DIRECTLY.
-    class ThreadLocalImpl {
+    class LLVM_SUPPORT_ABI ThreadLocalImpl {
       typedef uint64_t ThreadLocalDataTy;
       /// Platform-specific thread local data.
       ///
       /// This is embedded in the class and we avoid malloc'ing/free'ing it,
       /// to make this class more safe for use along with CrashRecoveryContext.
       union {
         char data[sizeof(ThreadLocalDataTy)];
         ThreadLocalDataTy align_data;
       };
     public:
       ThreadLocalImpl();
       virtual ~ThreadLocalImpl();
       void setInstance(const void* d);
       void *getInstance();
       void removeInstance();
     };
 
     /// ThreadLocal - A class used to abstract thread-local storage.  It holds,
     /// for each thread, a pointer a single object of type T.
     template<class T>
     class ThreadLocal : public ThreadLocalImpl {
     public:
       ThreadLocal() : ThreadLocalImpl() { }
 
       /// get - Fetches a pointer to the object associated with the current

llvm/include/llvm/Support/ThreadPool.h

 //===-- llvm/Support/ThreadPool.h - A ThreadPool implementation -*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines a crude C++11 based thread pool.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_THREADPOOL_H
 #define LLVM_SUPPORT_THREADPOOL_H
 
 #include "llvm/Config/llvm-config.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/Threading.h"
 #include "llvm/Support/thread.h"
 
 #include <future>
 
 #include <atomic>
 #include <condition_variable>
 #include <functional>
 #include <memory>
 #include <mutex>
 #include <queue>
 #include <utility>
 
 namespace llvm {
 
 /// A ThreadPool for asynchronous parallel execution on a defined number of
 /// threads.
 ///
 /// The pool keeps a vector of threads alive, waiting on a condition variable
 /// for some work to become available.
-class ThreadPool {
+class LLVM_SUPPORT_ABI ThreadPool {
 public:
   using TaskTy = std::function<void()>;
   using PackagedTaskTy = std::packaged_task<void()>;
 
   /// Construct a pool using the hardware strategy \p S for mapping hardware
   /// execution resources (threads, cores, CPUs)
   /// Defaults to using the maximum execution resources in the system, but
   /// accounting for the affinity mask.

llvm/include/llvm/Support/Threading.h

 #define LLVM_SUPPORT_THREADING_H
 
 #include "llvm/ADT/BitVector.h"
 #include "llvm/ADT/FunctionExtras.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Config/llvm-config.h" // for LLVM_ON_UNIX
 #include "llvm/Support/Compiler.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <ciso646> // So we can check the C++ standard lib macros.
 #include <functional>
 
 #if defined(_MSC_VER)
 // MSVC's call_once implementation worked since VS 2015, which is the minimum
 // supported version as of this writing.
 #define LLVM_THREADING_USE_STD_CALL_ONCE 1
 #elif defined(LLVM_ON_UNIX) &&                                                 \

 #include "llvm/Support/Atomic.h"
 #endif
 
 namespace llvm {
 class Twine;
 
 /// Returns true if LLVM is compiled with support for multi-threading, and
 /// false otherwise.
-bool llvm_is_multithreaded();
+LLVM_SUPPORT_ABI bool llvm_is_multithreaded();
 
 #if LLVM_THREADING_USE_STD_CALL_ONCE
 
   typedef std::once_flag once_flag;
 
 #else
 
   enum InitStatus { Uninitialized = 0, Wait = 1, Done = 2 };
if (old_val == Uninitialized) {
         sys::MemoryFence();
       }
     }
     TsanHappensAfter(&flag.status);
 #endif
   }
 
   /// This tells how a thread pool will be used
-  class ThreadPoolStrategy {
+  class LLVM_SUPPORT_ABI ThreadPoolStrategy {
   public:
     // The default value (0) means all available threads should be used,
     // taking the affinity mask into account. If set, this value only represents
     // a suggested high bound, the runtime might choose a lower value (not
     // higher).
     unsigned ThreadsRequested = 0;
 
     // If SMT is active, use hyper threads. If false, there will be only one
public:
     Optional<unsigned> compute_cpu_socket(unsigned ThreadPoolNum) const;
   };
 
   /// Build a strategy from a number of threads as a string provided in \p Num.
   /// When Num is above the max number of threads specified by the \p Default
   /// strategy, we attempt to equally allocate the threads on all CPU sockets.
   /// "0" or an empty string will return the \p Default strategy.
   /// "all" for using all hardware threads.
-  Optional<ThreadPoolStrategy>
+  LLVM_SUPPORT_ABI Optional<ThreadPoolStrategy>
   get_threadpool_strategy(StringRef Num, ThreadPoolStrategy Default = {});
 
   /// Returns a thread strategy for tasks requiring significant memory or other
   /// resources. To be used for workloads where hardware_concurrency() proves to
   /// be less efficient. Avoid this strategy if doing lots of I/O. Currently
   /// based on physical cores, if available for the host system, otherwise falls
   /// back to hardware_concurrency(). Returns 1 when LLVM is configured with
   /// LLVM_ENABLE_THREADS = OFF.
inline ThreadPoolStrategy optimal_concurrency(unsigned TaskCount = 0) {
     S.ThreadsRequested = TaskCount;
     return S;
   }
 
   /// Return the current thread id, as used in various OS system calls.
   /// Note that not all platforms guarantee that the value returned will be
   /// unique across the entire system, so portable code should not assume
   /// this.
-  uint64_t get_threadid();
+  LLVM_SUPPORT_ABI uint64_t get_threadid();
 
   /// Get the maximum length of a thread name on this platform.
   /// A value of 0 means there is no limit.
   uint32_t get_max_thread_name_length();
 
   /// Set the name of the current thread.  Setting a thread's name can
   /// be helpful for enabling useful diagnostics under a debugger or when
   /// logging.  The level of support for setting a thread's name varies
   /// wildly across operating systems, and we only make a best effort to
   /// perform the operation on supported platforms.  No indication of success
   /// or failure is returned.
-  void set_thread_name(const Twine &Name);
+  LLVM_SUPPORT_ABI void set_thread_name(const Twine &Name);
 
   /// Get the name of the current thread.  The level of support for
   /// getting a thread's name varies wildly across operating systems, and it
   /// is not even guaranteed that if you can successfully set a thread's name
   /// that you can later get it back.  This function is intended for diagnostic
   /// purposes, and as with setting a thread's name no indication of whether
   /// the operation succeeded or failed is returned.
-  void get_thread_name(SmallVectorImpl<char> &Name);
+  LLVM_SUPPORT_ABI void get_thread_name(SmallVectorImpl<char> &Name);
 
   /// Returns a mask that represents on which hardware thread, core, CPU, NUMA
   /// group, the calling thread can be executed. On Windows, threads cannot
   /// cross CPU sockets boundaries.
-  llvm::BitVector get_thread_affinity_mask();
+  LLVM_SUPPORT_ABI llvm::BitVector get_thread_affinity_mask();
 
   /// Returns how many physical CPUs or NUMA groups the system has.
-  unsigned get_cpus();
+  LLVM_SUPPORT_ABI unsigned get_cpus();
 
   enum class ThreadPriority {
     Background = 0,
     Default = 1,
   };
   /// If priority is Background tries to lower current threads priority such
   /// that it does not affect foreground tasks significantly. Can be used for
   /// long-running, latency-insensitive tasks to make sure cpu is not hogged by
   /// this task.
   /// If the priority is default tries to restore current threads priority to
   /// default scheduling priority.
   enum class SetThreadPriorityResult { FAILURE, SUCCESS };
-  SetThreadPriorityResult set_thread_priority(ThreadPriority Priority);
+
+  LLVM_SUPPORT_ABI SetThreadPriorityResult set_thread_priority(ThreadPriority Priority);
 }
 
 #endif

llvm/include/llvm/Support/TimeProfiler.h

 //===- llvm/Support/TimeProfiler.h - Hierarchical Time Profiler -*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TIMEPROFILER_H
 #define LLVM_SUPPORT_TIMEPROFILER_H
 
 #include "llvm/Support/Error.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 struct TimeTraceProfiler;
-TimeTraceProfiler *getTimeTraceProfilerInstance();
+LLVM_SUPPORT_ABI TimeTraceProfiler *getTimeTraceProfilerInstance();
 
 /// Initialize the time trace profiler.
 /// This sets up the global \p TimeTraceProfilerInstance
 /// variable to be the profiler instance.
-void timeTraceProfilerInitialize(unsigned TimeTraceGranularity,
+LLVM_SUPPORT_ABI void timeTraceProfilerInitialize(unsigned TimeTraceGranularity,
                                  StringRef ProcName);
 
 /// Cleanup the time trace profiler, if it was initialized.
-void timeTraceProfilerCleanup();
+LLVM_SUPPORT_ABI void timeTraceProfilerCleanup();
 
 /// Finish a time trace profiler running on a worker thread.
-void timeTraceProfilerFinishThread();
+LLVM_SUPPORT_ABI void timeTraceProfilerFinishThread();
 
 /// Is the time trace profiler enabled, i.e. initialized?
 inline bool timeTraceProfilerEnabled() {
   return getTimeTraceProfilerInstance() != nullptr;
 }
 
 /// Write profiling data to output stream.
 /// Data produced is JSON, in Chrome "Trace Event" format, see
 /// https://docs.google.com/document/d/1CvAClvFfyA5R-PhYUmn5OOQtYMH4h6I0nSsKchNAySU/preview
-void timeTraceProfilerWrite(raw_pwrite_stream &OS);
+LLVM_SUPPORT_ABI void timeTraceProfilerWrite(raw_pwrite_stream &OS);
 
 /// Write profiling data to a file.
 /// The function will write to \p PreferredFileName if provided, if not
 /// then will write to \p FallbackFileName appending .time-trace.
 /// Returns a StringError indicating a failure if the function is
 /// unable to open the file for writing.
-Error timeTraceProfilerWrite(StringRef PreferredFileName,
+LLVM_SUPPORT_ABI Error timeTraceProfilerWrite(StringRef PreferredFileName,
                              StringRef FallbackFileName);
 
 /// Manually begin a time section, with the given \p Name and \p Detail.
 /// Profiler copies the string data, so the pointers can be given into
 /// temporaries. Time sections can be hierarchical; every Begin must have a
 /// matching End pair but they can nest.
-void timeTraceProfilerBegin(StringRef Name, StringRef Detail);
-void timeTraceProfilerBegin(StringRef Name,
+LLVM_SUPPORT_ABI void timeTraceProfilerBegin(StringRef Name, StringRef Detail);
+LLVM_SUPPORT_ABI void timeTraceProfilerBegin(StringRef Name,
                             llvm::function_ref<std::string()> Detail);
 
 /// Manually end the last time section.
-void timeTraceProfilerEnd();
+LLVM_SUPPORT_ABI void timeTraceProfilerEnd();
 
 /// The TimeTraceScope is a helper class to call the begin and end functions
 /// of the time trace profiler.  When the object is constructed, it begins
 /// the section; and when it is destroyed, it stops it. If the time profiler
 /// is not initialized, the overhead is a single branch.
 struct TimeTraceScope {
 
   TimeTraceScope() = delete;

llvm/include/llvm/Support/Timer.h

 //===-- llvm/Support/Timer.h - Interval Timing Support ----------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TIMER_H
 #define LLVM_SUPPORT_TIMER_H
 
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <string>
 #include <utility>
 #include <vector>
 
 namespace llvm {
 
 class Timer;
 class TimerGroup;
 class raw_ostream;
 
-class TimeRecord {
+class LLVM_SUPPORT_ABI TimeRecord {
   double WallTime;               ///< Wall clock time elapsed in seconds.
   double UserTime;               ///< User time elapsed.
   double SystemTime;             ///< System time elapsed.
   ssize_t MemUsed;               ///< Memory allocated (in bytes).
   uint64_t InstructionsExecuted; ///< Number of instructions executed
 public:
   TimeRecord()
       : WallTime(0), UserTime(0), SystemTime(0), MemUsed(0),

 
 /// This class is used to track the amount of time spent between invocations of
 /// its startTimer()/stopTimer() methods.  Given appropriate OS support it can
 /// also keep track of the RSS of the program at various points.  By default,
 /// the Timer will print the amount of time it has captured to standard error
 /// when the last timer is destroyed, otherwise it is printed when its
 /// TimerGroup is destroyed.  Timers do not print their information if they are
 /// never started.
-class Timer {
+class LLVM_SUPPORT_ABI Timer {
   TimeRecord Time;          ///< The total time captured.
   TimeRecord StartTime;     ///< The time startTimer() was last called.
   std::string Name;         ///< The name of this time variable.
   std::string Description;  ///< Description of this time variable.
   bool Running = false;     ///< Is the timer currently running?
   bool Triggered = false;   ///< Has the timer ever been triggered?
   TimerGroup *TG = nullptr; ///< The TimerGroup this Timer is in.
 

 private:
   friend class TimerGroup;
 };
 
 /// The TimeRegion class is used as a helper class to call the startTimer() and
 /// stopTimer() methods of the Timer class.  When the object is constructed, it
 /// starts the timer specified as its argument.  When it is destroyed, it stops
 /// the relevant timer.  This makes it easy to time a region of code.
-class TimeRegion {
+class LLVM_SUPPORT_ABI TimeRegion {
   Timer *T;
   TimeRegion(const TimeRegion &) = delete;
 
 public:
   explicit TimeRegion(Timer &t) : T(&t) {
     T->startTimer();
   }
   explicit TimeRegion(Timer *t) : T(t) {
     if (T) T->startTimer();
   }
   ~TimeRegion() {
     if (T) T->stopTimer();
   }
 };
 
 /// This class is basically a combination of TimeRegion and Timer.  It allows
 /// you to declare a new timer, AND specify the region to time, all in one
 /// statement.  All timers with the same name are merged.  This is primarily
 /// used for debugging and for hunting performance problems.
-struct NamedRegionTimer : public TimeRegion {
+struct LLVM_SUPPORT_ABI NamedRegionTimer : public TimeRegion {
   explicit NamedRegionTimer(StringRef Name, StringRef Description,
                             StringRef GroupName,
                             StringRef GroupDescription, bool Enabled = true);
 };
 
 /// The TimerGroup class is used to group together related timers into a single
 /// report that is printed when the TimerGroup is destroyed.  It is illegal to
 /// destroy a TimerGroup object before all of the Timers in it are gone.  A
 /// TimerGroup can be specified for a newly created timer in its constructor.
-class TimerGroup {
+class LLVM_SUPPORT_ABI TimerGroup {
   struct PrintRecord {
     TimeRecord Time;
     std::string Name;
     std::string Description;
 
     PrintRecord(const PrintRecord &Other) = default;
     PrintRecord &operator=(const PrintRecord &Other) = default;
     PrintRecord(const TimeRecord &Time, const std::string &Name,
public:
   static void ConstructTimerLists();
 
   /// This makes the default group unmanaged, and lets the user manage the
   /// group's lifetime.
   static std::unique_ptr<TimerGroup> aquireDefaultGroup();
 
 private:
   friend class Timer;
-  friend void PrintStatisticsJSON(raw_ostream &OS);
+  friend LLVM_SUPPORT_ABI void PrintStatisticsJSON(raw_ostream &OS);
   void addTimer(Timer &T);
   void removeTimer(Timer &T);
   void prepareToPrintList(bool reset_time = false);
   void PrintQueuedTimers(raw_ostream &OS);
   void printJSONValue(raw_ostream &OS, const PrintRecord &R,
                       const char *suffix, double Value);
 };
 
 } // end namespace llvm
 
 #endif

llvm/include/llvm/Support/ToolOutputFile.h

 //===- ToolOutputFile.h - Output files for compiler-like tools -----------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 //  This file defines the ToolOutputFile class.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TOOLOUTPUTFILE_H
 #define LLVM_SUPPORT_TOOLOUTPUTFILE_H
 
 #include "llvm/ADT/Optional.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 /// This class contains a raw_fd_ostream and adds a few extra features commonly
 /// needed for compiler-like tool output files:
 ///   - The file is automatically deleted if the process is killed.
 ///   - The file is automatically deleted when the ToolOutputFile
 ///     object is destroyed unless the client calls keep().
 class ToolOutputFile {
   /// This class is declared before the raw_fd_ostream so that it is constructed
   /// before the raw_fd_ostream is constructed and destructed after the
   /// raw_fd_ostream is destructed. It installs cleanups in its constructor and
   /// uninstalls them in its destructor.
-  class CleanupInstaller {
+  class LLVM_SUPPORT_ABI CleanupInstaller {
     /// The name of the file.
     std::string Filename;
   public:
     /// The flag which indicates whether we should not delete the file.
     bool Keep;
 
     StringRef getFilename() { return Filename; }
     explicit CleanupInstaller(StringRef Filename);
     ~CleanupInstaller();
   } Installer;
 
   /// Storage for the stream, if we're owning our own stream. This is
   /// intentionally declared after Installer.
   Optional<raw_fd_ostream> OSHolder;
 
   /// The actual stream to use.
   raw_fd_ostream *OS;
 
 public:
   /// This constructor's arguments are passed to raw_fd_ostream's
   /// constructor.
+  LLVM_SUPPORT_ABI
   ToolOutputFile(StringRef Filename, std::error_code &EC,
                  sys::fs::OpenFlags Flags);
 
+  LLVM_SUPPORT_ABI
   ToolOutputFile(StringRef Filename, int FD);
 
   /// Return the contained raw_fd_ostream.
   raw_fd_ostream &os() { return *OS; }
 
   /// Return the filename initialized with.
   StringRef getFilename() { return Installer.getFilename(); }
 
   /// Indicate that the tool's job wrt this output file has been successful and
   /// the file should not be deleted.
   void keep() { Installer.Keep = true; }
 };
 
 } // end llvm namespace
 
 #endif

llvm/include/llvm/Support/TrigramIndex.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TRIGRAMINDEX_H
 #define LLVM_SUPPORT_TRIGRAMINDEX_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 #include <unordered_map>
 #include <vector>
 
 namespace llvm {
 class StringRef;
 
-class TrigramIndex {
+class LLVM_SUPPORT_ABI TrigramIndex {
  public:
   /// Inserts a new Regex into the index.
   void insert(const std::string &Regex);
 
   /// Returns true, if special case list definitely does not have a line
   /// that matches the query. Returns false, if it's not sure.
   bool isDefinitelyOut(StringRef Query) const;
 
   /// Returned true, iff the heuristic is defeated and not useful.

llvm/include/llvm/Support/TypeSize.h

 // it can be used in much the same way as a single scalar value.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_TYPESIZE_H
 #define LLVM_SUPPORT_TYPESIZE_H
 
 #include "llvm/ADT/ArrayRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/WithColor.h"
 
 #include <algorithm>
 #include <array>
 #include <cassert>
 #include <cstdint>
 #include <type_traits>
 
 namespace llvm {
 
 /// Reports a diagnostic message to indicate an invalid size request has been
 /// done on a scalable vector. This function may not return.
-void reportInvalidSizeRequest(const char *Msg);
+LLVM_SUPPORT_ABI void reportInvalidSizeRequest(const char *Msg);
 
 template <typename LeafTy> struct LinearPolyBaseTypeTraits {};
 
 //===----------------------------------------------------------------------===//
 // LinearPolyBase - a base class for linear polynomials with multiple
 // dimensions. This can e.g. be used to describe offsets that are have both a
 // fixed and scalable component.
 //===----------------------------------------------------------------------===//

 };
 
 // TODO: Most functionality in this class will gradually be phased out
 // so it will resemble LinearPolySize as much as possible.
 //
 // TypeSize is used to represent the size of types. If the type is of fixed
 // size, it will represent the exact size. If the type is a scalable vector,
 // it will represent the known minimum size.
-class TypeSize : public LinearPolySize<TypeSize> {
+class LLVM_SUPPORT_ABI TypeSize : public LinearPolySize<TypeSize> {
 public:
   TypeSize(const LinearPolySize<TypeSize> &V) : LinearPolySize(V) {}
   TypeSize(ScalarTy MinVal, bool IsScalable)
       : LinearPolySize(LinearPolySize::get(MinVal, IsScalable)) {}
 
   static TypeSize Fixed(ScalarTy MinVal) { return TypeSize(MinVal, false); }
   static TypeSize Scalable(ScalarTy MinVal) { return TypeSize(MinVal, true); }
 

llvm/include/llvm/Support/Unicode.h

 //===- llvm/Support/Unicode.h - Unicode character properties  -*- C++ -*-=====//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines functions that allow querying certain properties of Unicode
 // characters.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_UNICODE_H
 #define LLVM_SUPPORT_UNICODE_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 namespace llvm {
 class StringRef;
 
 namespace sys {
 namespace unicode {
 
 enum ColumnWidthErrors {
   ErrorInvalidUTF8 = -2,

 ///   * default ignorable code points as per 5.21 of
 ///     http://www.unicode.org/versions/Unicode6.2.0/UnicodeStandard-6.2.pdf
 ///     except for U+00AD SOFT HYPHEN, as it's actually displayed on most
 ///     terminals;
 ///   * format characters (category = Cf);
 ///   * surrogates (category = Cs);
 ///   * unassigned characters (category = Cn).
 /// \return true if the character is considered printable.
-bool isPrintable(int UCS);
+LLVM_SUPPORT_ABI bool isPrintable(int UCS);
 
 /// Gets the number of positions the UTF8-encoded \p Text is likely to occupy
 /// when output on a terminal ("character width"). This depends on the
 /// implementation of the terminal, and there's no standard definition of
 /// character width.
 ///
 /// The implementation defines it in a way that is expected to be compatible
 /// with a generic Unicode-capable terminal.
 ///
 /// \return Character width:
 ///   * ErrorNonPrintableCharacter (-1) if \p Text contains non-printable
 ///     characters (as identified by isPrintable);
 ///   * 0 for each non-spacing and enclosing combining mark;
 ///   * 2 for each CJK character excluding halfwidth forms;
 ///   * 1 for each of the remaining characters.
-int columnWidthUTF8(StringRef Text);
+LLVM_SUPPORT_ABI int columnWidthUTF8(StringRef Text);
 
 /// Fold input unicode character according the Simple unicode case folding
 /// rules.
 int foldCharSimple(int C);
 
 } // namespace unicode
 } // namespace sys
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/Valgrind.h

 // under.  These are all no-ops unless LLVM was configured on a system with the
 // valgrind headers installed and valgrind is controlling this process.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_VALGRIND_H
 #define LLVM_SUPPORT_VALGRIND_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cstddef>
 
 namespace llvm {
 namespace sys {
   // True if Valgrind is controlling this process.
-  bool RunningOnValgrind();
+  LLVM_SUPPORT_ABI bool RunningOnValgrind();
 
   // Discard valgrind's translation of code in the range [Addr .. Addr + Len).
   // Otherwise valgrind may continue to execute the old version of the code.
   void ValgrindDiscardTranslations(const void *Addr, size_t Len);
 } // namespace sys
 } // end namespace llvm
 
 #endif // LLVM_SUPPORT_VALGRIND_H

llvm/include/llvm/Support/VersionTuple.h

 //===----------------------------------------------------------------------===//
 #ifndef LLVM_SUPPORT_VERSIONTUPLE_H
 #define LLVM_SUPPORT_VERSIONTUPLE_H
 
 #include "llvm/ADT/DenseMapInfo.h"
 #include "llvm/ADT/Hashing.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/Support/HashBuilder.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <string>
 #include <tuple>
 
 namespace llvm {
 class raw_ostream;
 class StringRef;
 
 /// Represents a version number in the form major[.minor[.subminor[.build]]].
-class VersionTuple {
+class LLVM_SUPPORT_ABI VersionTuple {
   unsigned Major : 32;
 
   unsigned Minor : 31;
   unsigned HasMinor : 1;
 
   unsigned Subminor : 31;
   unsigned HasSubminor : 1;
 
public:
 
   /// Try to parse the given string as a version number.
   /// \returns \c true if the string does not match the regular expression
   ///   [0-9]+(\.[0-9]+){0,3}
   bool tryParse(StringRef string);
 };
 
 /// Print a version number.
+LLVM_SUPPORT_ABI
 raw_ostream &operator<<(raw_ostream &Out, const VersionTuple &V);
 
 // Provide DenseMapInfo for version tuples.
 template <> struct DenseMapInfo<VersionTuple> {
   static inline VersionTuple getEmptyKey() { return VersionTuple(0x7FFFFFFF); }
   static inline VersionTuple getTombstoneKey() {
     return VersionTuple(0x7FFFFFFE);
   }

llvm/include/llvm/Support/VirtualFileSystem.h

 #include "llvm/ADT/IntrusiveRefCntPtr.h"
 #include "llvm/ADT/None.h"
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Chrono.h"
 #include "llvm/Support/ErrorOr.h"
 #include "llvm/Support/FileSystem.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/Path.h"
 #include "llvm/Support/SourceMgr.h"
+
 #include <cassert>
 #include <cstdint>
 #include <ctime>
 #include <memory>
 #include <stack>
 #include <string>
 #include <system_error>
 #include <utility>
 #include <vector>
 
 namespace llvm {
 
 class MemoryBuffer;
 class MemoryBufferRef;
 class Twine;
 
 namespace vfs {
 
 /// The result of a \p status operation.
-class Status {
+class LLVM_SUPPORT_ABI Status {
   std::string Name;
   llvm::sys::fs::UniqueID UID;
   llvm::sys::TimePoint<> MTime;
   uint32_t User;
   uint32_t Group;
   uint64_t Size;
   llvm::sys::fs::file_type Type = llvm::sys::fs::file_type::status_error;
   llvm::sys::fs::perms Perms;
public:
   bool isOther() const;
   bool isSymlink() const;
   bool isStatusKnown() const;
   bool exists() const;
   /// @}
 };
 
 /// Represents an open file.
-class File {
+class LLVM_SUPPORT_ABI File {
 public:
   /// Destroy the file after closing it (if open).
   /// Sub-classes should generally call close() inside their destructors.  We
   /// cannot do that from the base class, since close is virtual.
   virtual ~File();
 
   /// Get the status of the file.
   virtual llvm::ErrorOr<Status> status() = 0;
public:
   llvm::StringRef path() const { return Path; }
   llvm::sys::fs::file_type type() const { return Type; }
 };
 
 namespace detail {
 
 /// An interface for virtual file systems to provide an iterator over the
 /// (non-recursive) contents of a directory.
-struct DirIterImpl {
+struct LLVM_SUPPORT_ABI DirIterImpl {
   virtual ~DirIterImpl();
 
   /// Sets \c CurrentEntry to the next entry in the directory on success,
   /// to directory_entry() at end,  or returns a system-defined \c error_code.
   virtual std::error_code increment() = 0;
 
   directory_entry CurrentEntry;
 };
struct RecDirIterState {
   std::stack<directory_iterator, std::vector<directory_iterator>> Stack;
   bool HasNoPushRequest = false;
 };
 
 } // end namespace detail
 
 /// An input iterator over the recursive contents of a virtual path,
 /// similar to llvm::sys::fs::recursive_directory_iterator.
-class recursive_directory_iterator {
+class LLVM_SUPPORT_ABI recursive_directory_iterator {
   FileSystem *FS;
   std::shared_ptr<detail::RecDirIterState>
       State; // Input iterator semantics on copy.
 
 public:
   recursive_directory_iterator(FileSystem &FS, const Twine &Path,
                                std::error_code &EC);
 
assert(!State->Stack.empty() &&
            "Cannot get level without any iteration state");
     return State->Stack.size() - 1;
   }
 
   void no_push() { State->HasNoPushRequest = true; }
 };
 
 /// The virtual file system interface.
-class FileSystem : public llvm::ThreadSafeRefCountedBase<FileSystem> {
+class LLVM_SUPPORT_ABI FileSystem : public llvm::ThreadSafeRefCountedBase<FileSystem> {
 public:
   virtual ~FileSystem();
 
   /// Get the status of the entry at \p Path, if one exists.
   virtual llvm::ErrorOr<Status> status(const Twine &Path) = 0;
 
   /// Get a \p File object for the file at \p Path, if one exists.
   virtual llvm::ErrorOr<std::unique_ptr<File>>
public:
   ///          platform-specific error_code.
   virtual std::error_code makeAbsolute(SmallVectorImpl<char> &Path) const;
 };
 
 /// Gets an \p vfs::FileSystem for the 'real' file system, as seen by
 /// the operating system.
 /// The working directory is linked to the process's working directory.
 /// (This is usually thread-hostile).
-IntrusiveRefCntPtr<FileSystem> getRealFileSystem();
+LLVM_SUPPORT_ABI IntrusiveRefCntPtr<FileSystem> getRealFileSystem();
 
 /// Create an \p vfs::FileSystem for the 'real' file system, as seen by
 /// the operating system.
 /// It has its own working directory, independent of (but initially equal to)
 /// that of the process.
-std::unique_ptr<FileSystem> createPhysicalFileSystem();
+LLVM_SUPPORT_ABI std::unique_ptr<FileSystem> createPhysicalFileSystem();
 
 /// A file system that allows overlaying one \p AbstractFileSystem on top
 /// of another.
 ///
 /// Consists of a stack of >=1 \p FileSystem objects, which are treated as being
 /// one merged file system. When there is a directory that exists in more than
 /// one file system, the \p OverlayFileSystem contains a directory containing
 /// the union of their contents.  The attributes (permissions, etc.) of the
 /// top-most (most recently added) directory are used.  When there is a file
 /// that exists in more than one file system, the file in the top-most file
 /// system overrides the other(s).
-class OverlayFileSystem : public FileSystem {
+class LLVM_SUPPORT_ABI OverlayFileSystem : public FileSystem {
   using FileSystemList = SmallVector<IntrusiveRefCntPtr<FileSystem>, 1>;
 
   /// The stack of file systems, implemented as a list in order of
   /// their addition.
   FileSystemList FSList;
 
 public:
   OverlayFileSystem(IntrusiveRefCntPtr<FileSystem> Base);
public:
   /// Get an iterator pointing one-past the most recently added file system.
   reverse_iterator overlays_rend() { return FSList.end(); }
   const_reverse_iterator overlays_rend() const { return FSList.end(); }
 };
 
 /// By default, this delegates all calls to the underlying file system. This
 /// is useful when derived file systems want to override some calls and still
 /// proxy other calls.
-class ProxyFileSystem : public FileSystem {
+class LLVM_SUPPORT_ABI ProxyFileSystem : public FileSystem {
 public:
   explicit ProxyFileSystem(IntrusiveRefCntPtr<FileSystem> FS)
       : FS(std::move(FS)) {}
 
   llvm::ErrorOr<Status> status(const Twine &Path) override {
     return FS->status(Path);
   }
   llvm::ErrorOr<std::unique_ptr<File>>

 namespace detail {
 
 class InMemoryDirectory;
 class InMemoryFile;
 
 } // namespace detail
 
 /// An in-memory file system.
-class InMemoryFileSystem : public FileSystem {
+class LLVM_SUPPORT_ABI InMemoryFileSystem : public FileSystem {
   std::unique_ptr<detail::InMemoryDirectory> Root;
   std::string WorkingDirectory;
   bool UseNormalizedPaths = true;
 
   /// If HardLinkTarget is non-null, a hardlink is created to the To path which
   /// must be a file. If it is null then it adds the file as the public addFile.
   bool addFile(const Twine &Path, time_t ModificationTime,
                std::unique_ptr<llvm::MemoryBuffer> Buffer,
public:
   std::error_code setCurrentWorkingDirectory(const Twine &Path) override;
 };
 
 /// Get a globally unique ID for a virtual file or directory.
 llvm::sys::fs::UniqueID getNextVirtualUniqueID();
 
 /// Gets a \p FileSystem for a virtual file system described in YAML
 /// format.
-std::unique_ptr<FileSystem>
+LLVM_SUPPORT_ABI std::unique_ptr<FileSystem>
 getVFSFromYAML(std::unique_ptr<llvm::MemoryBuffer> Buffer,
                llvm::SourceMgr::DiagHandlerTy DiagHandler,
                StringRef YAMLFilePath, void *DiagContext = nullptr,
                IntrusiveRefCntPtr<FileSystem> ExternalFS = getRealFileSystem());
 
 struct YAMLVFSEntry {
   template <typename T1, typename T2>
   YAMLVFSEntry(T1 &&VPath, T2 &&RPath, bool IsDirectory = false)

 public:
   /// Looks up \p Path in \c Roots and returns a LookupResult giving the
   /// matched entry and, if the entry was a FileEntry or DirectoryRemapEntry,
   /// the path it redirects to in the external file system.
   ErrorOr<LookupResult> lookupPath(StringRef Path) const;
 
   /// Parses \p Buffer, which is expected to be in YAML format and
   /// returns a virtual file system representing its contents.
+  LLVM_SUPPORT_ABI
   static std::unique_ptr<RedirectingFileSystem>
   create(std::unique_ptr<MemoryBuffer> Buffer,
          SourceMgr::DiagHandlerTy DiagHandler, StringRef YAMLFilePath,
          void *DiagContext, IntrusiveRefCntPtr<FileSystem> ExternalFS);
 
   /// Redirect each of the remapped files from first to second.
   static std::unique_ptr<RedirectingFileSystem>
   create(ArrayRef<std::pair<std::string, std::string>> RemappedFiles,

 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
   LLVM_DUMP_METHOD void dump() const;
 #endif
 };
 
 /// Collect all pairs of <virtual path, real path> entries from the
 /// \p YAMLFilePath. This is used by the module dependency collector to forward
 /// the entries into the reproducer output VFS YAML file.
-void collectVFSFromYAML(
+LLVM_SUPPORT_ABI void collectVFSFromYAML(
     std::unique_ptr<llvm::MemoryBuffer> Buffer,
     llvm::SourceMgr::DiagHandlerTy DiagHandler, StringRef YAMLFilePath,
     SmallVectorImpl<YAMLVFSEntry> &CollectedEntries,
     void *DiagContext = nullptr,
     IntrusiveRefCntPtr<FileSystem> ExternalFS = getRealFileSystem());
 
-class YAMLVFSWriter {
+class LLVM_SUPPORT_ABI YAMLVFSWriter {
   std::vector<YAMLVFSEntry> Mappings;
   Optional<bool> IsCaseSensitive;
   Optional<bool> IsOverlayRelative;
   Optional<bool> UseExternalNames;
   std::string OverlayDir;
 
   void addEntry(StringRef VirtualPath, StringRef RealPath, bool IsDirectory);
 

llvm/include/llvm/Support/Windows/WindowsSupport.h

 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Config/llvm-config.h" // Get build system configuration settings
 #include "llvm/Support/Allocator.h"
 #include "llvm/Support/Chrono.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/VersionTuple.h"
+
 #include <cassert>
 #include <string>
 #include <system_error>
 #include <windows.h>
 
 // Must be included after windows.h
 #include <wincrypt.h>
 

 // Returns command line arguments. Unlike arguments given to main(),
 // this function guarantees that the returned arguments are encoded in
 // UTF-8 regardless of the current code page setting.
 std::error_code GetCommandLineArguments(SmallVectorImpl<const char *> &Args,
                                         BumpPtrAllocator &Alloc);
 
 /// Convert UTF-8 path to a suitable UTF-16 path for use with the Win32 Unicode
 /// File API.
-std::error_code widenPath(const Twine &Path8, SmallVectorImpl<wchar_t> &Path16,
+LLVM_SUPPORT_ABI std::error_code widenPath(const Twine &Path8, SmallVectorImpl<wchar_t> &Path16,
                           size_t MaxPathLen = MAX_PATH);
 
 } // end namespace windows
 } // end namespace sys
 } // end namespace llvm.
 
 #endif

llvm/include/llvm/Support/WindowsError.h

 //===-- WindowsError.h - Support for mapping windows errors to posix-------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_WINDOWSERROR_H
 #define LLVM_SUPPORT_WINDOWSERROR_H
 
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <system_error>
 
 namespace llvm {
-std::error_code mapWindowsError(unsigned EV);
+LLVM_SUPPORT_ABI std::error_code mapWindowsError(unsigned EV);
 }
 
 #endif

llvm/include/llvm/Support/WithColor.h

 //===- WithColor.h ----------------------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_WITHCOLOR_H
 #define LLVM_SUPPORT_WITHCOLOR_H
 
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
 
 namespace llvm {
 
 class Error;
 class StringRef;
 
 namespace cl {
 class OptionCategory;
 }
 
-extern cl::OptionCategory &getColorCategory();
+LLVM_SUPPORT_ABI cl::OptionCategory &getColorCategory();
 
 // Symbolic names for various syntax elements.
 enum class HighlightColor {
   Address,
   String,
   Tag,
   Attribute,
   Enumerator,
enum class ColorMode {
   /// disabled.
   Enable,
   /// Disable colors.
   Disable,
 };
 
 /// An RAII object that temporarily switches an output stream to a specific
 /// color.
-class WithColor {
+class LLVM_SUPPORT_ABI WithColor {
   raw_ostream &OS;
   ColorMode Mode;
 
 public:
   /// To be used like this: WithColor(OS, HighlightColor::String) << "text";
   /// @param OS The output stream
   /// @param S Symbolic name for syntax element to color
   /// @param Mode Enable, disable or compute whether to use colors.

llvm/include/llvm/Support/X86TargetParser.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_X86TARGETPARSER_H
 #define LLVM_SUPPORT_X86TARGETPARSER_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
 class StringRef;
 
 namespace X86 {
 
 // This should be kept in sync with libcc/compiler-rt as its included by clang
 // as a proxy for what's in libgcc/compiler-rt.
enum CPUKind {
   CK_x86_64_v2,
   CK_x86_64_v3,
   CK_x86_64_v4,
   CK_Geode,
 };
 
 /// Parse \p CPU string into a CPUKind. Will only accept 64-bit capable CPUs if
 /// \p Only64Bit is true.
-CPUKind parseArchX86(StringRef CPU, bool Only64Bit = false);
-CPUKind parseTuneCPU(StringRef CPU, bool Only64Bit = false);
+LLVM_SUPPORT_ABI CPUKind parseArchX86(StringRef CPU, bool Only64Bit = false);
+LLVM_SUPPORT_ABI CPUKind parseTuneCPU(StringRef CPU, bool Only64Bit = false);
 
 /// Provide a list of valid CPU names. If \p Only64Bit is true, the list will
 /// only contain 64-bit capable CPUs.
-void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values,
+LLVM_SUPPORT_ABI void fillValidCPUArchList(SmallVectorImpl<StringRef> &Values,
                           bool Only64Bit = false);
 /// Provide a list of valid -mtune names.
-void fillValidTuneCPUList(SmallVectorImpl<StringRef> &Values,
+LLVM_SUPPORT_ABI void fillValidTuneCPUList(SmallVectorImpl<StringRef> &Values,
                           bool Only64Bit = false);
 
 /// Get the key feature prioritizing target multiversioning.
-ProcessorFeatures getKeyFeature(CPUKind Kind);
+LLVM_SUPPORT_ABI ProcessorFeatures getKeyFeature(CPUKind Kind);
 
 /// Fill in the features that \p CPU supports into \p Features.
-void getFeaturesForCPU(StringRef CPU, SmallVectorImpl<StringRef> &Features);
+LLVM_SUPPORT_ABI void getFeaturesForCPU(StringRef CPU, SmallVectorImpl<StringRef> &Features);
 
 /// Set or clear entries in \p Features that are implied to be enabled/disabled
 /// by the provided \p Feature.
-void updateImpliedFeatures(StringRef Feature, bool Enabled,
+LLVM_SUPPORT_ABI void updateImpliedFeatures(StringRef Feature, bool Enabled,
                            StringMap<bool> &Features);
 
-uint64_t getCpuSupportsMask(ArrayRef<StringRef> FeatureStrs);
-unsigned getFeaturePriority(ProcessorFeatures Feat);
+LLVM_SUPPORT_ABI uint64_t getCpuSupportsMask(ArrayRef<StringRef> FeatureStrs);
+LLVM_SUPPORT_ABI unsigned getFeaturePriority(ProcessorFeatures Feat);
 
 } // namespace X86
 } // namespace llvm
 
 #endif

llvm/include/llvm/Support/YAMLParser.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_YAMLPARSER_H
 #define LLVM_SUPPORT_YAMLPARSER_H
 
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Allocator.h"
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/SMLoc.h"
 #include "llvm/Support/SourceMgr.h"
+
 #include <cassert>
 #include <cstddef>
 #include <iterator>
 #include <map>
 #include <memory>
 #include <string>
 #include <system_error>
 
 namespace llvm {
 
 class MemoryBufferRef;
 class raw_ostream;
 class Twine;
 
 namespace yaml {
 
 class Document;
 class document_iterator;
 class Node;
 class Scanner;
 struct Token;
 
 /// Dump all the tokens in this stream to OS.
 /// \returns true if there was an error, false otherwise.
-bool dumpTokens(StringRef Input, raw_ostream &);
+LLVM_SUPPORT_ABI bool dumpTokens(StringRef Input, raw_ostream &);
 
 /// Scans all tokens in input without outputting anything. This is used
 ///        for benchmarking the tokenizer.
 /// \returns true if there was an error, false otherwise.
-bool scanTokens(StringRef Input);
+LLVM_SUPPORT_ABI bool scanTokens(StringRef Input);
 
 /// Escape \a Input for a double quoted scalar; if \p EscapePrintable
 /// is true, all UTF8 sequences will be escaped, if \p EscapePrintable is
 /// false, those UTF8 sequences encoding printable unicode scalars will not be
 /// escaped, but emitted verbatim.
-std::string escape(StringRef Input, bool EscapePrintable = true);
+LLVM_SUPPORT_ABI std::string escape(StringRef Input, bool EscapePrintable = true);
 
 /// Parse \p S as a bool according to https://yaml.org/type/bool.html.
-llvm::Optional<bool> parseBool(StringRef S);
+LLVM_SUPPORT_ABI llvm::Optional<bool> parseBool(StringRef S);
 
 /// This class represents a YAML stream potentially containing multiple
 ///        documents.
-class Stream {
+class LLVM_SUPPORT_ABI Stream {
 public:
   /// This keeps a reference to the string referenced by \p Input.
   Stream(StringRef Input, SourceMgr &, bool ShowColors = true,
          std::error_code *EC = nullptr);
 
   Stream(MemoryBufferRef InputBuffer, SourceMgr &, bool ShowColors = true,
          std::error_code *EC = nullptr);
   ~Stream();

 private:
   friend class Document;
 
   std::unique_ptr<Scanner> scanner;
   std::unique_ptr<Document> CurrentDoc;
 };
 
 /// Abstract base class for all Nodes.
-class Node {
+class LLVM_SUPPORT_ABI Node {
   virtual void anchor();
 
 public:
   enum NodeKind {
     NK_Null,
     NK_Scalar,
     NK_BlockScalar,
     NK_KeyValue,
public:
   static bool classof(const Node *N) { return N->getType() == NK_Null; }
 };
 
 /// A scalar node is an opaque datum that can be presented as a
 ///        series of zero or more Unicode scalar values.
 ///
 /// Example:
 ///   Adena
-class ScalarNode final : public Node {
+class LLVM_SUPPORT_ABI ScalarNode final : public Node {
   void anchor() override;
 
 public:
   ScalarNode(std::unique_ptr<Document> &D, StringRef Anchor, StringRef Tag,
              StringRef Val)
       : Node(NK_Scalar, D, Anchor, Tag), Value(Val) {
     SMLoc Start = SMLoc::getFromPointer(Val.begin());
     SMLoc End = SMLoc::getFromPointer(Val.end());

 
 /// A key and value pair. While not technically a Node under the YAML
 ///        representation graph, it is easier to treat them this way.
 ///
 /// TODO: Consider making this not a child of Node.
 ///
 /// Example:
 ///   Section: .text
-class KeyValueNode final : public Node {
+class LLVM_SUPPORT_ABI KeyValueNode final : public Node {
   void anchor() override;
 
 public:
   KeyValueNode(std::unique_ptr<Document> &D)
       : Node(NK_KeyValue, D, StringRef(), StringRef()) {}
 
   /// Parse and return the key.
   ///

 
 /// Represents a YAML map created from either a block map for a flow map.
 ///
 /// This parses the YAML stream as increment() is called.
 ///
 /// Example:
 ///   Name: _main
 ///   Scope: Global
-class MappingNode final : public Node {
+class LLVM_SUPPORT_ABI MappingNode final : public Node {
   void anchor() override;
 
 public:
   enum MappingType {
     MT_Block,
     MT_Flow,
     MT_Inline ///< An inline mapping node is used for "[key: value]".
   };

 /// Represents a YAML sequence created from either a block sequence for a
 ///        flow sequence.
 ///
 /// This parses the YAML stream as increment() is called.
 ///
 /// Example:
 ///   - Hello
 ///   - World
-class SequenceNode final : public Node {
+class LLVM_SUPPORT_ABI SequenceNode final : public Node {
   void anchor() override;
 
 public:
   enum SequenceType {
     ST_Block,
     ST_Flow,
     // Use for:
     //
public:
   static bool classof(const Node *N) { return N->getType() == NK_Alias; }
 
 private:
   StringRef Name;
 };
 
 /// A YAML Stream is a sequence of Documents. A document contains a root
 ///        node.
-class Document {
+class LLVM_SUPPORT_ABI Document {
 public:
   Document(Stream &ParentStream);
 
   /// Root for parsing a node. Returns a single node.
   Node *parseBlockNode();
 
   /// Finish parsing the current document and return true if there are
   ///        more. Return false otherwise.

llvm/include/llvm/Support/YAMLTraits.h

 //===- llvm/Support/YAMLTraits.h --------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_YAMLTRAITS_H
 #define LLVM_SUPPORT_YAMLTRAITS_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/AlignOf.h"
 #include "llvm/Support/Allocator.h"
 #include "llvm/Support/Endian.h"
+#include "llvm/SUpport/LLVMSupportExports.h"
 #include "llvm/Support/Regex.h"
 #include "llvm/Support/SMLoc.h"
 #include "llvm/Support/SourceMgr.h"
 #include "llvm/Support/VersionTuple.h"
 #include "llvm/Support/YAMLParser.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <cassert>
 #include <cctype>
 #include <cstddef>
 #include <cstdint>
 #include <iterator>
 #include <map>
 #include <memory>
 #include <new>

 
 template <typename T, typename Context>
 struct unvalidatedMappingTraits
     : public std::integral_constant<
           bool, has_MappingTraits<T, Context>::value &&
                     !has_MappingValidateTraits<T, Context>::value> {};
 
 // Base class for Input and Output.
-class IO {
+class LLVM_SUPPORT_ABI IO {
 public:
   IO(void *Ctxt = nullptr);
   virtual ~IO();
 
   virtual bool outputting() const = 0;
 
   virtual unsigned beginSequence() = 0;
   virtual bool preflightElement(unsigned, void *&) = 0;
for(unsigned i=0; i < count; ++i) {
         yamlize(io, SequenceTraits<T>::element(io, Seq, i), true, Ctx);
         io.postflightElement(SaveInfo);
       }
     }
     io.endSequence();
   }
 }
 
 template<>
-struct ScalarTraits<bool> {
+struct LLVM_SUPPORT_ABI ScalarTraits<bool> {
   static void output(const bool &, void* , raw_ostream &);
   static StringRef input(StringRef, void *, bool &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<StringRef> {
+struct LLVM_SUPPORT_ABI ScalarTraits<StringRef> {
   static void output(const StringRef &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, StringRef &);
   static QuotingType mustQuote(StringRef S) { return needsQuotes(S); }
 };
 
 template<>
-struct ScalarTraits<std::string> {
+struct LLVM_SUPPORT_ABI ScalarTraits<std::string> {
   static void output(const std::string &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, std::string &);
   static QuotingType mustQuote(StringRef S) { return needsQuotes(S); }
 };
 
 template<>
-struct ScalarTraits<uint8_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<uint8_t> {
   static void output(const uint8_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, uint8_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<uint16_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<uint16_t> {
   static void output(const uint16_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, uint16_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<uint32_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<uint32_t> {
   static void output(const uint32_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, uint32_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<uint64_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<uint64_t> {
   static void output(const uint64_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, uint64_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<int8_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<int8_t> {
   static void output(const int8_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, int8_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<int16_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<int16_t> {
   static void output(const int16_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, int16_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<int32_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<int32_t> {
   static void output(const int32_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, int32_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<int64_t> {
+struct LLVM_SUPPORT_ABI ScalarTraits<int64_t> {
   static void output(const int64_t &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, int64_t &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<float> {
+struct LLVM_SUPPORT_ABI ScalarTraits<float> {
   static void output(const float &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, float &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<double> {
+struct LLVM_SUPPORT_ABI ScalarTraits<double> {
   static void output(const double &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, double &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 // For endian types, we use existing scalar Traits class for the underlying
 // type.  This way endian aware types are supported whenever the traits are
 // defined for the underlying type.

 /// It works by using YAMLParser to do a syntax parse of the entire yaml
 /// document, then the Input class builds a graph of HNodes which wraps
 /// each yaml Node.  The extra layer is buffering.  The low level yaml
 /// parser only lets you look at each node once.  The buffering layer lets
 /// you search and interate multiple times.  This is necessary because
 /// the mapRequired() method calls may not be in the same order
 /// as the keys in the document.
 ///
-class Input : public IO {
+class LLVM_SUPPORT_ABI Input : public IO {
 public:
   // Construct a yaml Input object from a StringRef and optional
   // user-data. The DiagHandler can be specified to provide
   // alternative error reporting.
   Input(StringRef InputContent,
         void *Ctxt = nullptr,
         SourceMgr::DiagHandlerTy DiagHandler = nullptr,
         void *DiagHandlerCtxt = nullptr);
private:
   bool                                ScalarMatchFound = false;
   bool AllowUnknownKeys = false;
 };
 
 ///
 /// The Output class is used to generate a yaml document from in-memory structs
 /// and vectors.
 ///
-class Output : public IO {
+class LLVM_SUPPORT_ABI Output : public IO {
 public:
   Output(raw_ostream &, void *Ctxt = nullptr, int WrapColumn = 70);
   ~Output() override;
 
   /// Set whether or not to output optional values which are equal
   /// to the default value.  By default, when outputting if you attempt
   /// to write a value that is equal to the default, the value gets ignored.
   /// Sometimes, it is useful to be able to see these in the resulting YAML

 /// Use these types instead of uintXX_t in any mapping to have
 /// its yaml output formatted as hexadecimal.
 ///
 LLVM_YAML_STRONG_TYPEDEF(uint8_t, Hex8)
 LLVM_YAML_STRONG_TYPEDEF(uint16_t, Hex16)
 LLVM_YAML_STRONG_TYPEDEF(uint32_t, Hex32)
 LLVM_YAML_STRONG_TYPEDEF(uint64_t, Hex64)
 
 template<>
-struct ScalarTraits<Hex8> {
+struct LLVM_SUPPORT_ABI ScalarTraits<Hex8> {
   static void output(const Hex8 &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, Hex8 &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<Hex16> {
+struct LLVM_SUPPORT_ABI ScalarTraits<Hex16> {
   static void output(const Hex16 &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, Hex16 &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<Hex32> {
+struct LLVM_SUPPORT_ABI ScalarTraits<Hex32> {
   static void output(const Hex32 &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, Hex32 &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 template<>
-struct ScalarTraits<Hex64> {
+struct LLVM_SUPPORT_ABI ScalarTraits<Hex64> {
   static void output(const Hex64 &, void *, raw_ostream &);
   static StringRef input(StringRef, void *, Hex64 &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
-template <> struct ScalarTraits<VersionTuple> {
+template <> struct LLVM_SUPPORT_ABI ScalarTraits<VersionTuple> {
   static void output(const VersionTuple &Value, void *, llvm::raw_ostream &Out);
   static StringRef input(StringRef, void *, VersionTuple &);
   static QuotingType mustQuote(StringRef) { return QuotingType::None; }
 };
 
 // Define non-member operator>> so that Input can stream in a document list.
 template <typename T>
 inline std::enable_if_t<has_DocumentListTraits<T>::value, Input &>

llvm/include/llvm/Support/raw_os_ostream.h

 //===- raw_os_ostream.h - std::ostream adaptor for raw_ostream --*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 //  This file defines the raw_os_ostream class.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_RAW_OS_OSTREAM_H
 #define LLVM_SUPPORT_RAW_OS_OSTREAM_H
 
+#include "llvm/Support/LLVMSupportExports.h"
 #include "llvm/Support/raw_ostream.h"
+
 #include <iosfwd>
 
 namespace llvm {
 
 /// raw_os_ostream - A raw_ostream that writes to an std::ostream.  This is a
 /// simple adaptor class.  It does not check for output errors; clients should
 /// use the underlying stream to detect errors.
-class raw_os_ostream : public raw_ostream {
+class LLVM_SUPPORT_ABI raw_os_ostream : public raw_ostream {
   std::ostream &OS;
 
   /// write_impl - See raw_ostream::write_impl.
   void write_impl(const char *Ptr, size_t Size) override;
 
   /// current_pos - Return the current position within the stream, not
   /// counting the bytes currently in the buffer.
   uint64_t current_pos() const override;

llvm/include/llvm/Support/raw_ostream.h

 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_RAW_OSTREAM_H
 #define LLVM_SUPPORT_RAW_OSTREAM_H
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/DataTypes.h"
+#include "llvm/Support/LLVMSupportExports.h"
+
 #include <cassert>
 #include <chrono>
 #include <cstddef>
 #include <cstdint>
 #include <cstring>
 #include <string>
 #if __cplusplus > 201402L
 #include <string_view>

 class FileLocker;
 } // end namespace fs
 } // end namespace sys
 
 /// This class implements an extremely fast bulk output stream that can *only*
 /// output to a stream.  It does not support seeking, reopening, rewinding, line
 /// buffered disciplines etc. It is a simple buffer that outputs
 /// a chunk at a time.
-class raw_ostream {
+class LLVM_SUPPORT_ABI raw_ostream {
 public:
   // Class kinds to support LLVM-style RTTI.
   enum class OStreamKind {
     OK_OStream,
     OK_FDStream,
   };
 
 private:

 operator<<(OStream &&OS, const T &Value) {
   OS << Value;
   return std::move(OS);
 }
 
 /// An abstract base class for streams implementations that also support a
 /// pwrite operation. This is useful for code that can mostly stream out data,
 /// but needs to patch in a header that needs to know the output size.
-class raw_pwrite_stream : public raw_ostream {
+class LLVM_SUPPORT_ABI raw_pwrite_stream : public raw_ostream {
   virtual void pwrite_impl(const char *Ptr, size_t Size, uint64_t Offset) = 0;
   void anchor() override;
 
 public:
   explicit raw_pwrite_stream(bool Unbuffered = false,
                              OStreamKind K = OStreamKind::OK_OStream)
       : raw_ostream(Unbuffered, K) {}
   void pwrite(const char *Ptr, size_t Size, uint64_t Offset) {

 };
 
 //===----------------------------------------------------------------------===//
 // File Output Streams
 //===----------------------------------------------------------------------===//
 
 /// A raw_ostream that writes to a file descriptor.
 ///
-class raw_fd_ostream : public raw_pwrite_stream {
+class LLVM_SUPPORT_ABI raw_fd_ostream : public raw_pwrite_stream {
   int FD;
   bool ShouldClose;
   bool SupportsSeeking = false;
   mutable Optional<bool> HasColors;
 
 #ifdef _WIN32
   /// True if this fd refers to a Windows console device. Mintty and other
   /// terminal emulators are TTYs, but they are not consoles.
public:
   ///
   /// It is used as @ref lock.
   LLVM_NODISCARD
   Expected<sys::fs::FileLocker> tryLockFor(std::chrono::milliseconds Timeout);
 };
 
 /// This returns a reference to a raw_fd_ostream for standard output. Use it
 /// like: outs() << "foo" << "bar";
-raw_fd_ostream &outs();
+LLVM_SUPPORT_ABI raw_fd_ostream &outs();
 
 /// This returns a reference to a raw_ostream for standard error.
 /// Use it like: errs() << "foo" << "bar";
 /// By default, the stream is tied to stdout to ensure stdout is flushed before
 /// stderr is written, to ensure the error messages are written in their
 /// expected place.
-raw_fd_ostream &errs();
+LLVM_SUPPORT_ABI raw_fd_ostream &errs();
 
 /// This returns a reference to a raw_ostream which simply discards output.
-raw_ostream &nulls();
+LLVM_SUPPORT_ABI raw_ostream &nulls();
 
 //===----------------------------------------------------------------------===//
 // File Streams
 //===----------------------------------------------------------------------===//
 
 /// A raw_ostream of a file for reading/writing/seeking.
 ///
-class raw_fd_stream : public raw_fd_ostream {
+class LLVM_SUPPORT_ABI raw_fd_stream : public raw_fd_ostream {
 public:
   /// Open the specified file for reading/writing/seeking. If an error occurs,
   /// information about the error is put into EC, and the stream should be
   /// immediately destroyed.
   raw_fd_stream(StringRef Filename, std::error_code &EC);
 
   /// This reads the \p Size bytes into a buffer pointed by \p Ptr.
   ///

 };
 
 //===----------------------------------------------------------------------===//
 // Output Stream Adaptors
 //===----------------------------------------------------------------------===//
 
 /// A raw_ostream that writes to an std::string.  This is a simple adaptor
 /// class. This class does not encounter output errors.
-class raw_string_ostream : public raw_ostream {
+class LLVM_SUPPORT_ABI raw_string_ostream : public raw_ostream {
   std::string &OS;
 
   /// See raw_ostream::write_impl.
   void write_impl(const char *Ptr, size_t Size) override;
 
   /// Return the current position within the stream, not counting the bytes
   /// currently in the buffer.
   uint64_t current_pos() const override { return OS.size(); }
public:
   }
 };
 
 /// A raw_ostream that writes to an SmallVector or SmallString.  This is a
 /// simple adaptor class. This class does not encounter output errors.
 /// raw_svector_ostream operates without a buffer, delegating all memory
 /// management to the SmallString. Thus the SmallString is always up-to-date,
 /// may be used directly and there is no need to call flush().
-class raw_svector_ostream : public raw_pwrite_stream {
+class LLVM_SUPPORT_ABI raw_svector_ostream : public raw_pwrite_stream {
   SmallVectorImpl<char> &OS;
 
   /// See raw_ostream::write_impl.
   void write_impl(const char *Ptr, size_t Size) override;
 
   void pwrite_impl(const char *Ptr, size_t Size, uint64_t Offset) override;
 
   /// Return the current position within the stream.
public:
   StringRef str() const { return StringRef(OS.data(), OS.size()); }
 
   void reserveExtraSpace(uint64_t ExtraSize) override {
     OS.reserve(tell() + ExtraSize);
   }
 };
 
 /// A raw_ostream that discards all output.
-class raw_null_ostream : public raw_pwrite_stream {
+class LLVM_SUPPORT_ABI raw_null_ostream : public raw_pwrite_stream {
   /// See raw_ostream::write_impl.
   void write_impl(const char *Ptr, size_t size) override;
   void pwrite_impl(const char *Ptr, size_t Size, uint64_t Offset) override;
 
   /// Return the current position within the stream, not counting the bytes
   /// currently in the buffer.
   uint64_t current_pos() const override;
 
 public:
   explicit raw_null_ostream() = default;
   ~raw_null_ostream() override;
 };
 
-class buffer_ostream : public raw_svector_ostream {
+class LLVM_SUPPORT_ABI buffer_ostream : public raw_svector_ostream {
   raw_ostream &OS;
   SmallVector<char, 0> Buffer;
 
   virtual void anchor() override;
 
 public:
   buffer_ostream(raw_ostream &OS) : raw_svector_ostream(Buffer), OS(OS) {}
   ~buffer_ostream() override { OS << str(); }
 };
 
-class buffer_unique_ostream : public raw_svector_ostream {
+class LLVM_SUPPORT_ABI buffer_unique_ostream : public raw_svector_ostream {
   std::unique_ptr<raw_ostream> OS;
   SmallVector<char, 0> Buffer;
 
   virtual void anchor() override;
 
 public:
   buffer_unique_ostream(std::unique_ptr<raw_ostream> OS)
       : raw_svector_ostream(Buffer), OS(std::move(OS)) {}
   ~buffer_unique_ostream() override { *OS << str(); }
 };
 
 class Error;
 
 /// This helper creates an output stream and then passes it to \p Write.
 /// The stream created is based on the specified \p OutputFileName:
 /// llvm::outs for "-", raw_null_ostream for "/dev/null", and raw_fd_ostream
 /// for other names. For raw_fd_ostream instances, the stream writes to
 /// a temporary file. The final output file is atomically replaced with the
 /// temporary file after the \p Write function is finished.
-Error writeToOutput(StringRef OutputFileName,
+LLVM_SUPPORT_ABI Error writeToOutput(StringRef OutputFileName,
                     std::function<Error(raw_ostream &)> Write);
 
 } // end namespace llvm
 
 #endif // LLVM_SUPPORT_RAW_OSTREAM_H

llvm/include/llvm/Support/thread.h

 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_SUPPORT_THREAD_H
 #define LLVM_SUPPORT_THREAD_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/Config/llvm-config.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 #ifdef _WIN32
 typedef unsigned long DWORD;
 typedef void *PVOID;
 typedef PVOID HANDLE;
 #endif
 
 #if LLVM_ENABLE_THREADS
 
 #include <thread>
 
 namespace llvm {
 
 #if LLVM_ON_UNIX || _WIN32
 
 /// LLVM thread following std::thread interface with added constructor to
 /// specify stack size.
-class thread {
+class LLVM_SUPPORT_ABI thread {
   template <typename FPtr, typename... Args, size_t... Indices>
   static void Apply(std::tuple<FPtr, Args...> &Callee,
                     std::index_sequence<Indices...>) {
     std::move(std::get<0>(Callee))(std::move(std::get<Indices + 1>(Callee))...);
   }
 
   template <typename CalleeTuple> static void GenericThreadProxy(void *Ptr) {
     std::unique_ptr<CalleeTuple> Callee(static_cast<CalleeTuple *>(Ptr));
#endif
   inline void detach();
 
   void swap(llvm::thread &Other) noexcept { std::swap(Thread, Other.Thread); }
 
 private:
   native_handle_type Thread;
 };
 
-thread::native_handle_type
+LLVM_SUPPORT_ABI thread::native_handle_type
 llvm_execute_on_thread_impl(thread::start_routine_type ThreadFunc, void *Arg,
                             llvm::Optional<unsigned> StackSizeInBytes);
-void llvm_thread_join_impl(thread::native_handle_type Thread);
-void llvm_thread_detach_impl(thread::native_handle_type Thread);
+
+LLVM_SUPPORT_ABI void llvm_thread_join_impl(thread::native_handle_type Thread);
+
+LLVM_SUPPORT_ABI void llvm_thread_detach_impl(thread::native_handle_type Thread);
+
 thread::id llvm_thread_get_id_impl(thread::native_handle_type Thread);
 thread::id llvm_thread_get_current_id_impl();
 
 template <class Function, class... Args>
 thread::thread(llvm::Optional<unsigned> StackSizeInBytes, Function &&f,
                Args &&...args) {
   typedef std::tuple<typename std::decay<Function>::type,
                      typename std::decay<Args>::type...>

llvm/include/llvm/Support/xxhash.h

 /* based on revision d2df04efcbef7d7f6886d345861e5dfda4edacc1 Removed
  * everything but a simple interface for computing XXh64. */
 
 #ifndef LLVM_SUPPORT_XXHASH_H
 #define LLVM_SUPPORT_XXHASH_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Support/LLVMSupportExports.h"
 
 namespace llvm {
-uint64_t xxHash64(llvm::StringRef Data);
-uint64_t xxHash64(llvm::ArrayRef<uint8_t> Data);
+LLVM_SUPPORT_ABI uint64_t xxHash64(llvm::StringRef Data);
+LLVM_SUPPORT_ABI uint64_t xxHash64(llvm::ArrayRef<uint8_t> Data);
 }
 
 #endif

llvm/lib/Support/CMakeLists.txt

elseif(LLVM_INTEGRATED_CRT_ALLOC MATCHES "mimalloc$")
     set(MIMALLOC_LIB "${LLVM_INTEGRATED_CRT_ALLOC}/out/msvc-x64/Release/mimalloc-static.lib")
     if(NOT EXISTS "${MIMALLOC_LIB}")
     message(FATAL_ERROR "Cannot find the mimalloc static library. To build it, first apply the patch from https://github.com/microsoft/mimalloc/issues/268 then build the Release x64 target through ${LLVM_INTEGRATED_CRT_ALLOC}\\ide\\vs2019\\mimalloc.sln")
     endif()
     set(system_libs ${system_libs} "${MIMALLOC_LIB}" "-INCLUDE:malloc")
   endif()
 endif()
 
-add_llvm_component_library(LLVMSupport
+add_llvm_component_library(LLVMSupport SHARED
   AArch64TargetParser.cpp
   ABIBreak.cpp
   ARMTargetParser.cpp
   AMDGPUMetadata.cpp
   APFixedPoint.cpp
   APFloat.cpp
   APInt.cpp
   APSInt.cpp
if(LLVM_ENABLE_ZLIB)
   set(llvm_system_libs ${llvm_system_libs} "${zlib_library}")
 endif()
 
 if(LLVM_ENABLE_TERMINFO)
   get_library_name(${TERMINFO_LIB} terminfo_library)
   set(llvm_system_libs ${llvm_system_libs} "${terminfo_library}")
 endif()
 
-set_property(TARGET LLVMSupport PROPERTY LLVM_SYSTEM_LIBS "${llvm_system_libs}")
+set_target_properties(LLVMSupport PROPERTIES
+  LLVM_SYSTEM_LIBS "${llvm_system_libs}"
+  DEFINE_SYMBOL "LLVMSupport_EXPORTS"
+  C_VISIBILITY_PRESET hidden
+  CXX_VISIBILITY_PRESET hidden
+  VISIBILITY_INLINES_HIDDEN YES)
 
 
 if(LLVM_INTEGRATED_CRT_ALLOC)
   if(LLVM_INTEGRATED_CRT_ALLOC MATCHES "snmalloc$")
     set_property(TARGET LLVMSupport PROPERTY CXX_STANDARD 17)
     add_definitions(-D_SILENCE_CXX17_ITERATOR_BASE_CLASS_DEPRECATION_WARNING)
     if ("${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang" AND
         "${CMAKE_SYSTEM_PROCESSOR}" MATCHES "x86_64")

llvm/lib/Support/Windows/Threading.inc

 #include <bitset>
 
 // Windows will at times define MemoryFence.
 #ifdef MemoryFence
 #undef MemoryFence
 #endif
 
 namespace llvm {
-HANDLE
+LLVM_SUPPORT_ABI HANDLE
 llvm_execute_on_thread_impl(unsigned(__stdcall *ThreadFunc)(void *), void *Arg,
                             llvm::Optional<unsigned> StackSizeInBytes) {
   HANDLE hThread = (HANDLE)::_beginthreadex(
       NULL, StackSizeInBytes.getValueOr(0), ThreadFunc, Arg, 0, NULL);
 
   if (!hThread) {
     ReportLastErrorFatal("_beginthreadex failed");
   }
 
   return hThread;
 }
 
+LLVM_SUPPORT_ABI
 void llvm_thread_join_impl(HANDLE hThread) {
   if (::WaitForSingleObject(hThread, INFINITE) == WAIT_FAILED) {
     ReportLastErrorFatal("WaitForSingleObject failed");
   }
 }
 
+LLVM_SUPPORT_ABI
 void llvm_thread_detach_impl(HANDLE hThread) {
   if (::CloseHandle(hThread) == FALSE) {
     ReportLastErrorFatal("CloseHandle failed");
   }
 }
 
 DWORD llvm_thread_get_id_impl(HANDLE hThread) {
   return ::GetThreadId(hThread);