Add OpenCL 2.0 atomic builtin functions as Clang builtin

[yaxunl]

Summary

OpenCL 2.0 atomic builtin functions have a scope argument which is ideally represented as synchronization scope argument in LLVM atomic instructions.

Clang supports translating Clang atomic builtin functions to LLVM atomic instructions. However it currently does not support synchronization scope of LLVM atomic instructions. Without this, users have to use LLVM assembly code to implement OpenCL atomic builtin functions.

This patch adds OpenCL 2.0 atomic builtin functions as Clang builtin functions, which supports generating LLVM atomic instructions with synchronization scope operand.

Currently only constant memory scope argument is supported. Support of non-constant memory scope argument will be added later.

[yaxunl]

Fix typos and formatting issues.

[rjmccall]

This patch changes the language design of the atomic builtins, which is outside the normal scope of patch review. You need to post an RFC to cfe-dev. I've gone ahead and made some material comments, but the concept itself needs debate.

Your proposed language design exposes LLVM internals (the specific values used by llvm::SynchronizationScope) directly to users; that is also unacceptable.

lib/AST/Expr.cpp

3917

Add an extra newline here, please, to be consistent with the other cases in the switch.

lib/CodeGen/CGAtomic.cpp

116

It is not acceptable to test this exclusively with an assertion; you need to be checking that the argument is an integer constant expression in Sema.

lib/Sema/SemaChecking.cpp

2922

1 is a magic number here.

[yaxunl]

In D28691#648267, @rjmccall wrote:

This patch changes the language design of the atomic builtins, which is outside the normal scope of patch review. You need to post an RFC to cfe-dev. I've gone ahead and made some material comments, but the concept itself needs debate.

Your proposed language design exposes LLVM internals (the specific values used by llvm::SynchronizationScope) directly to users; that is also unacceptable.

I sent an RFC to cfe-dev.

For the synchronization scope, I will add Clang's enums for synchronization scopes in a similar fashion as memory order. Different languages can have their own specific synchronization scopes. Then clang translates them to synchronization scopes used by LLVM atomic instructions.

lib/CodeGen/CGAtomic.cpp

116

Will add diagnostics to Sema.

lib/Sema/SemaChecking.cpp

2922

Will change that to enum.

[yaxunl]

The link for the RFC:
http://lists.llvm.org/pipermail/cfe-dev/2017-January/052304.html

[yaxunl]

Add __opencl_atomic_ builtins.

[rjmccall]

Looks like you haven't introduced proper constants in the header for scopes.

docs/LanguageExtensions.rst

1855

1. "an", not "in".
2. There's no need to put "explicit" in `` quotes.
3. The prefix you're using is "__opencl_", with only one trailing underscore.

lib/CodeGen/CGAtomic.cpp

114–1

This is equivalent to checking for the two AO kinds, right? Probably better to just do that.

lib/CodeGen/CGExpr.cpp

1

I think it would be better to keep this function simple and just add the cast outside in the two places you need it.

Also, please remember to use the target hook instead of using an addrspacecast directly.

lib/Sema/SemaChecking.cpp

3046

This clause now covers the scope argument as well.

3060

You need to check that is a constant expression in the correct range, and you should add a test for that.

3062

Please document the meaning of 1.

[yaxunl]

Revised by John's comments.

[Anastasia]

include/clang/Basic/Builtins.def

27

What about min/max? I believe they will need to have the scope too.

include/clang/Basic/DiagnosticSemaKinds.td

6956

Btw, is this disallowed by the spec? Can't find anything relevant.

lib/CodeGen/CGAtomic.cpp

114–1

The same here... not sure adding an extra variable is helping here. :)

115

Seems short enough to introduce an extra variable here. :)

117

Variable name doesn't follow the style.

117

could we avoid using C style cast?

124

Formatting seems to be a bit odd here...

lib/Sema/SemaChecking.cpp

3146

formatting seems odd.

test/CodeGenOpenCL/atomic-ops-libcall.cl

2

GEN4 -> SPIR

3

GEN0 -> AMDGPU

5

Could we use different scopes?

test/CodeGenOpenCL/atomic-ops.cl

8

why do we need this?

16

I think we could use different scope types all over...

33

do we have an addrspacecast before?

test/SemaOpenCL/atomic-ops.cl

23

could we have the full error for consistency?

31

could we also test with constant AS? Also I would generally improve testing wrt address spaces...

[yaxunl]

include/clang/Basic/Builtins.def

27

They are not 2.0 atomic builtin functions. They can be implemented as library functions through 2.0 atomic builtin functions.

include/clang/Basic/DiagnosticSemaKinds.td

6956

Just temporarily not supported by Clang. Will add support later.

lib/CodeGen/CGAtomic.cpp

114–1

removed the variable

115

removed the variable

117

will fix

117

will change to static_cast

124

this is done by clang-format

lib/Sema/SemaChecking.cpp

3146

this is done by clang-format

test/CodeGenOpenCL/atomic-ops-libcall.cl

2

will change

3

Actually this triple is armv5e. This test requires a target not supporting atomic instructions. Will change GEN0 -> ARM

5

Yes. will add them.

[yaxunl]

test/CodeGenOpenCL/atomic-ops.cl

8

This is to test the builtin works in pch. When generating pch, ALREADY_INCLUDED is undefined, therefore pch will include all functions. When including pch, since ALREADY_INCLUDED is defined through pch, the cl file is empty and function definitions from pch is used for codegen.

16

will add them.

33

No,

test/SemaOpenCL/atomic-ops.cl

23

will do

31

will add tests for different addr spaces.

[yaxunl]

Rebased to ToT and revised by Anastasia's comments.

[b-sumner]

include/clang/Basic/Builtins.def

27

Yes, they are. Please look again at 6.13.11.7.5 in the 2.0 C spec.

[yaxunl]

include/clang/Basic/Builtins.def

27

sorry I thought they are just some atomic extensions since C++11 atomic builtins do not have those. I will add them.

[kzhuravl]

Seems like SyncScope.h file is missing?

[b-sumner]

Can we drop the "opencl" part of the name and use something like __scoped_atomic_*? Also, it may not make sense to support non-constant scope here since we can't predict what other scopes may be added by other languages in the future.

[yaxunl]

In D28691#820375, @kzhuravl wrote:

Seems like SyncScope.h file is missing?

Right. I will add it.

[yaxunl]

Add min/max and missing file.

[yaxunl]

In D28691#820466, @b-sumner wrote:

Can we drop the "opencl" part of the name and use something like __scoped_atomic_*? Also, it may not make sense to support non-constant scope here since we can't predict what other scopes may be added by other languages in the future.

we could use the approach of LangAS, i.e. we allow targets to map all language specific scopes to target-specific scope names, since IR only cares about scope names, which are target specific. And this is what the current implementation does.

I have no objection to use the __scoped_atomic_ name. It is more general and extensible. John/Anastasia, any comments? Thanks.

[rjmccall]

In D28691#820489, @yaxunl wrote:

In D28691#820466, @b-sumner wrote:

Can we drop the "opencl" part of the name and use something like __scoped_atomic_*? Also, it may not make sense to support non-constant scope here since we can't predict what other scopes may be added by other languages in the future.

we could use the approach of LangAS, i.e. we allow targets to map all language specific scopes to target-specific scope names, since IR only cares about scope names, which are target specific. And this is what the current implementation does.

I have no objection to use the __scoped_atomic_ name. It is more general and extensible. John/Anastasia, any comments? Thanks.

I think I would prefer __opencl_atomic_* until we have some evidence that this concept is more general than just OpenCL.

[b-sumner]

In D28691#820526, @rjmccall wrote:

In D28691#820489, @yaxunl wrote:

In D28691#820466, @b-sumner wrote:

Can we drop the "opencl" part of the name and use something like __scoped_atomic_*? Also, it may not make sense to support non-constant scope here since we can't predict what other scopes may be added by other languages in the future.

we could use the approach of LangAS, i.e. we allow targets to map all language specific scopes to target-specific scope names, since IR only cares about scope names, which are target specific. And this is what the current implementation does.

I have no objection to use the __scoped_atomic_ name. It is more general and extensible. John/Anastasia, any comments? Thanks.

I think I would prefer __opencl_atomic_* until we have some evidence that this concept is more general than just OpenCL.

There are other languages for heterogeneous compute that have scopes, although not exposed quite as explicitly as OpenCL. For example AMD's "HC" language. And any language making use of clang and targeting SPIR-V would likely use these builtins. I think a more generic prefix is appropriate, and "scoped" tells us exactly when these are needed.

[rjmccall]

In D28691#820541, @b-sumner wrote:

There are other languages for heterogeneous compute that have scopes, although not exposed quite as explicitly as OpenCL. For example AMD's "HC" language. And any language making use of clang and targeting SPIR-V would likely use these builtins. I think a more generic prefix is appropriate, and "scoped" tells us exactly when these are needed.

But would those languages use the same language design for these scopes as OpenCL if they did expose them, as opposed to some more elaborate scoping specification? My objection is not that the concept is inherently OpenCL-specific, it's that the presentation in the language might be inherently OpenCL-specific, which makes staying in the opencl namespace is prudent.

[b-sumner]

In D28691#820595, @rjmccall wrote:

In D28691#820541, @b-sumner wrote:

There are other languages for heterogeneous compute that have scopes, although not exposed quite as explicitly as OpenCL. For example AMD's "HC" language. And any language making use of clang and targeting SPIR-V would likely use these builtins. I think a more generic prefix is appropriate, and "scoped" tells us exactly when these are needed.

But would those languages use the same language design for these scopes as OpenCL if they did expose them, as opposed to some more elaborate scoping specification? My objection is not that the concept is inherently OpenCL-specific, it's that the presentation in the language might be inherently OpenCL-specific, which makes staying in the opencl namespace is prudent.

Are you envisioning a language far enough from C/C++ that a standard library or header would not be able to map a scoped atomic operation into a call to one of these new builtins? Would we expect more of such languages than languages that would do such a mapping?

[rjmccall]

In D28691#820641, @b-sumner wrote:

In D28691#820595, @rjmccall wrote:

In D28691#820541, @b-sumner wrote:

There are other languages for heterogeneous compute that have scopes, although not exposed quite as explicitly as OpenCL. For example AMD's "HC" language. And any language making use of clang and targeting SPIR-V would likely use these builtins. I think a more generic prefix is appropriate, and "scoped" tells us exactly when these are needed.

But would those languages use the same language design for these scopes as OpenCL if they did expose them, as opposed to some more elaborate scoping specification? My objection is not that the concept is inherently OpenCL-specific, it's that the presentation in the language might be inherently OpenCL-specific, which makes staying in the opencl namespace is prudent.

Are you envisioning a language far enough from C/C++ that a standard library or header would not be able to map a scoped atomic operation into a call to one of these new builtins? Would we expect more of such languages than languages that would do such a mapping?

If you're using Clang as a frontend for your language, it must be similar enough to C to call a builtin function. That's not at issue. The central question here is whether these builtins are meaningfully general outside of OpenCL. The concept of heterogenous computation is certainly not specific to OpenCL; however, these builtins are defined in terms of scopes — "work item", "work group", "device", and "all svm devices" — which, it seems to me, are basically only defined by reference to the OpenCL architecture. A different heterogenous compute environment might reasonably formalize scopes in a completely different way; for example, it might wish to be more explicit about exactly which peers / devices to synchronize with.

SPIR is explicitly defined on top of the OpenCL model. Users should be able to use OpenCL builtins when targeting it. That does not make those builtins more general than OpenCL.

John.

[t-tye]

include/clang/Basic/Builtins.def

30

Will the OpenCL 2.0 memory fences also be supported which also have a memory order and memory scope?

include/clang/Basic/SyncScope.h

26–30

Since the builtins are being named as __opencl then should these also be named as opencl_ since they are the memory scopes for OpenCL using the OpenCL numeric values?

If another language wants to use memory scopes, would it then add its own langx_* names in a similar way that is done for address spaces where the LangAS enumeration type has values for each distinct language. Each target is then responsible for mapping each language scope to the appropriate target specific scope as is done for address spaces?

If so then the builtins are really supporting the concept of memory scopes and are not language specific as this enumeration can support multiple languages in the same way as the LangAS enumeration supports multiple languages. If so the builtins would be better named to reflect this as @b-sumner suggested.

[Anastasia]

In D28691#820684, @rjmccall wrote:

In D28691#820641, @b-sumner wrote:

In D28691#820595, @rjmccall wrote:

In D28691#820541, @b-sumner wrote:

There are other languages for heterogeneous compute that have scopes, although not exposed quite as explicitly as OpenCL. For example AMD's "HC" language. And any language making use of clang and targeting SPIR-V would likely use these builtins. I think a more generic prefix is appropriate, and "scoped" tells us exactly when these are needed.

But would those languages use the same language design for these scopes as OpenCL if they did expose them, as opposed to some more elaborate scoping specification? My objection is not that the concept is inherently OpenCL-specific, it's that the presentation in the language might be inherently OpenCL-specific, which makes staying in the opencl namespace is prudent.

Are you envisioning a language far enough from C/C++ that a standard library or header would not be able to map a scoped atomic operation into a call to one of these new builtins? Would we expect more of such languages than languages that would do such a mapping?

If you're using Clang as a frontend for your language, it must be similar enough to C to call a builtin function. That's not at issue. The central question here is whether these builtins are meaningfully general outside of OpenCL. The concept of heterogenous computation is certainly not specific to OpenCL; however, these builtins are defined in terms of scopes — "work item", "work group", "device", and "all svm devices" — which, it seems to me, are basically only defined by reference to the OpenCL architecture. A different heterogenous compute environment might reasonably formalize scopes in a completely different way; for example, it might wish to be more explicit about exactly which peers / devices to synchronize with.

SPIR is explicitly defined on top of the OpenCL model. Users should be able to use OpenCL builtins when targeting it. That does not make those builtins more general than OpenCL.

John.

The scope concept in OpenCL is fairly generic. And the builtins just take one extra argument on top of the existing C11 builtin style. The OpenCL scopes have of course specific meaning to the OpenCL model, but there is nothing preventing other uses of the scope feature. As far as I understand atomic scope implementation in LLVM is fairly generic wrt scope types and it is intended for broader functionality than just OpenCL. So I would vote for having this as generic as possible in Clang too even though I don't think name prefix __opencl is preventing from using this feature in other languages unless we would disallow the builtins in the other language dialects. Which is not the case with the current patch because the builtins are added as ATOMIC_BUILTIN and not LANGBUILTIN. We have for example used C11 builtin in OpenCL already. So other cases are possible too.

include/clang/Basic/SyncScope.h

26–30

We generally prefix the names of OpenCL specific implementations. So perhaps we should do some renaming here in case we don't intend this to be generic implementation.

lib/Sema/SemaChecking.cpp

3160

Could we merge this and the line after please.

[rjmccall]

In D28691#823810, @Anastasia wrote:

In D28691#820684, @rjmccall wrote:

In D28691#820641, @b-sumner wrote:

In D28691#820595, @rjmccall wrote:

In D28691#820541, @b-sumner wrote:

There are other languages for heterogeneous compute that have scopes, although not exposed quite as explicitly as OpenCL. For example AMD's "HC" language. And any language making use of clang and targeting SPIR-V would likely use these builtins. I think a more generic prefix is appropriate, and "scoped" tells us exactly when these are needed.

But would those languages use the same language design for these scopes as OpenCL if they did expose them, as opposed to some more elaborate scoping specification? My objection is not that the concept is inherently OpenCL-specific, it's that the presentation in the language might be inherently OpenCL-specific, which makes staying in the opencl namespace is prudent.

Are you envisioning a language far enough from C/C++ that a standard library or header would not be able to map a scoped atomic operation into a call to one of these new builtins? Would we expect more of such languages than languages that would do such a mapping?

If you're using Clang as a frontend for your language, it must be similar enough to C to call a builtin function. That's not at issue. The central question here is whether these builtins are meaningfully general outside of OpenCL. The concept of heterogenous computation is certainly not specific to OpenCL; however, these builtins are defined in terms of scopes — "work item", "work group", "device", and "all svm devices" — which, it seems to me, are basically only defined by reference to the OpenCL architecture. A different heterogenous compute environment might reasonably formalize scopes in a completely different way; for example, it might wish to be more explicit about exactly which peers / devices to synchronize with.

SPIR is explicitly defined on top of the OpenCL model. Users should be able to use OpenCL builtins when targeting it. That does not make those builtins more general than OpenCL.

John.

The scope concept in OpenCL is fairly generic. And the builtins just take one extra argument on top of the existing C11 builtin style. The OpenCL scopes have of course specific meaning to the OpenCL model, but there is nothing preventing other uses of the scope feature.

Yes, it is possible that some other language could introduce exactly the same scope-atomics concept only with a slightly different enumeration of scopes. But then we really shouldn't allow the OpenCL scopes to be used in that language, which means there would still not be a language-independent way of using these builtins.

As far as I understand atomic scope implementation in LLVM is fairly generic wrt scope types and it is intended for broader functionality than just OpenCL.

In some ways this is reasoning the wrong way around. I am not deeply informed about heterogenous computing, so I am happy to accept that llvm::SynchronizationScope is well-designed for our current needs. But LLVM's representation is, by design, ultimately just an implementation detail and can be easily updated — it's just a matter of changing a few calls and adding an upgrade path to the bitcode loader, exactly as we did when we introduced llvm::SynchronizationScope. That is not true of source language features, even builtins; their design is a contract with programmers, and the bar is substantially higher.

We do not have a compelling reason to claim that these are generally useful, so we should not. They should stay namespaced and be flagged as language-specific builtins.

John.

[rjmccall]

include/clang/Basic/SyncScope.h

2

Capitalization.

21

LLVM uses this namespace pattern in code that predates the addition of scoped enums to C++. Those days are behind us; we should just use a scoped enum.

23

It defines the synch scope values used by the atomic builtins and expressions. LLVM's headers define the values used by the instructions.

26–30

I agree that we should name the OpenCL-specific ones, like WorkGroup, with an OpenCL prefix.

33

This is C++; please just use () instead of (void).

lib/CodeGen/CGAtomic.cpp

126

None of the .getTypePtr() stuff here is necessary.

This function shouldn't really be necessary. I would encourage you to add a getValueType() accessor to AtomicExpr:

QualType AtomicExpr::getValueType() const {
  auto T = getPtr()->getType()->castTo<PointerType>()->getPointeeType();
  if (auto AT = T->getAs<AtomicType>()) {
    return AT->getValueType();
  } else {
    return T;
  }
}

You can then just use E->getValueType()->isSignedIntegerType() and eliminate this helper function.

128

Again you're using getTypePtr() unnecessarily.

The check should be whether the AST-level address spaces match, not whether the lowered address spaces match.

Please pass E->getType() instead of E here.

You should remove the DoIt parameter and just check E->isOpenCL() (where E is the AtomicExpr already in scope).

lib/CodeGen/CGCall.cpp

3911 ↗

(On Diff #108127)

No. Callers should ensure that they've added the right argument type, at least at the level of address spaces.

lib/CodeGen/TargetInfo.h

8

Why does this return a StringRef instead of an llvm::SynchronizationScope?

[yaxunl]

Revised by reviewers' comments.

[t-tye]

include/clang/Basic/SyncScope.h

24

The OpenCL workitem scope is only used for image fences and does not apply to atomic operations so not sure that it should be in this enumeration which is used only for memory atomics.

lib/CodeGen/TargetInfo.cpp

7554–7555

OpenCL only uses workitem for image fences which are not the same as atomic memory fences.

For image fences I don't think it would map to singlethread which is intended for signal handler support to ensure a signal handler has visibility of the updates done by a thread which is more of an optimization barrier. In contrast an image fence may need to flush caches to make the image and vector access paths coherent in a single thread.

Since this patch does not support fences probably want to leave workitem scope out. Current AMDGPU targets do not need to do anything for an OpenCL image fence, but other targets may need to generate an intrinsic since there is no LLVMIR instruction for this.

lib/Headers/opencl-c.h

13143–13151

Do these have to have the same values as the SycnScope enumeration? Should that be ensured in a similar way to the memory_order enumeration?

lib/Sema/SemaChecking.cpp

3103

IIRC OpenCL allows the scope to be a runtime value. So will doing this will likely cause failures in conformance?

[rjmccall]

The patch generally looks good, but if you need to handle non-constant scopes, you should submit a new patch to address that.

lib/CodeGen/CGAtomic.cpp

896

You can sink this line and the next.

lib/Sema/SemaChecking.cpp

3103

Ah, if that's true, you'll need to emit a switch in IRGen, the same way we handle non-constant memory orders.

[yaxunl]

include/clang/Basic/Builtins.def

30

I am considering supporting it with a separate patch since this patch is already quite large.

include/clang/Basic/SyncScope.h

24

You are right. I think we should drop it from this enum for now.

lib/CodeGen/CGAtomic.cpp

896

will do.

lib/CodeGen/CGCall.cpp

3911 ↗

(On Diff #108127)

Will remove.

lib/CodeGen/TargetInfo.cpp

7554–7555

Thanks. I will remove this for now.

lib/Headers/opencl-c.h

13143–13151

It is desirable to have the same value as SyncScope enumeration, otherwise the library has to do the translation when calling __opencl_atomic_* builtins.

Will do.

lib/Sema/SemaChecking.cpp

3103

Will support it in another patch since this one is already quite large.

3160

will do

[yaxunl]

Revised by Tony's and John's comments.

[yaxunl]

Add assert to make sure pre-defined macros __OPENCL_MEMORY_SCOP_* to be consistent with SyncScope enum.

[rjmccall]

LGTM. I'm fine with the plan to handle potentially non-constant scopes in a follow-up patch.

include/clang/Basic/SyncScope.h

22

If the numeric values here were chosen to align with the arguments to some runtime function, that's important to leave as a comment.

lib/Headers/opencl-c.h

13143–13151

Since we're defining these builtins ourselves de novo, it's fine to pick argument values that align with what the existing runtime functions expect. Once the builtins are defined and in-use, of course, we cannot subsequently change the builtin values, even if the runtime changes.

[yaxunl]

Add comments to SyncScope.h

[t-tye]

docs/LanguageExtensions.rst

1935

Should it also say:

The macros ``__OPENCL_MEMORY_SCOPE_WORK_ITEM``, ``__OPENCL_MEMORY_SCOPE_WORK_GROUP``, ``__OPENCL_MEMORY_SCOPE_DEVICE``, ``__OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES``, and ``__OPENCL_MEMORY_SCOPE_SUB_GROUP`` are provided, with values corresponding to the enumerators of OpenCL's ``memory_scope`` enumeration.

[yaxunl]

docs/LanguageExtensions.rst

1935

Thanks. Will do.

[yaxunl]

Added documentation about __OPENCL_MEMORY_SCOPE_* by Tony's comments.

[t-tye]

LGTM

[yaxunl]

include/clang/Basic/SyncScope.h

22

I have added the comments and updated the review.. On the Phabricator web page, it may take a bit effort to find it since it is interleaved with reviewers' comments.

[bader]

lib/AST/Expr.cpp

4000–4004

No need in else branch after return:

if (...) {
  return AT->getValueType();
}

return T;

http://llvm.org/docs/CodingStandards.html#don-t-use-else-after-a-return

test/SemaOpenCL/atomic-ops.cl

2

It's a pity, we have to parse the whole opencl-c.h file to get two enums and one typedef...

[yaxunl]

Ping. Any other comments? Thanks.

lib/AST/Expr.cpp

4000–4004

will fix when committing.

test/SemaOpenCL/atomic-ops.cl

2

Since there is change in opencl-c.h, using the header can test that.

[rjmccall]

Still LGTM.

docs/LanguageExtensions.rst

 .. code-block:: c
 
    double __builtin_canonicalize(double);
    float __builtin_canonicalizef(float);
    long double__builtin_canonicalizel(long double);
 
 Returns the platform specific canonical encoding of a floating point
 number. This canonicalization is useful for implementing certain
 numeric primitives such as frexp. See `LLVM canonicalize intrinsic
 <http://llvm.org/docs/LangRef.html#llvm-canonicalize-intrinsic>`_ for
 more information on the semantics.
 
 String builtins
 ---------------
 
 Clang provides constant expression evaluation support for builtins forms of
 the following functions from the C standard library ``<string.h>`` header:

 * ``__c11_atomic_fetch_xor``
 
 The macros ``__ATOMIC_RELAXED``, ``__ATOMIC_CONSUME``, ``__ATOMIC_ACQUIRE``,
 ``__ATOMIC_RELEASE``, ``__ATOMIC_ACQ_REL``, and ``__ATOMIC_SEQ_CST`` are
 provided, with values corresponding to the enumerators of C11's
 ``memory_order`` enumeration.
 
 (Note that Clang additionally provides GCC-compatible ``__atomic_*``
-builtins)
+builtins and OpenCL 2.0 ``__opencl_atomic_*`` builtins. The OpenCL 2.0
+atomic builtins are an explicit form of the corresponding OpenCL 2.0
+builtin function, and are named with a ``__opencl_`` prefix.)
 
 Low-level ARM exclusive memory builtins
 ---------------------------------------
 
 Clang provides overloaded builtins giving direct access to the three key ARM
 instructions for implementing atomic operations.
 
 .. code-block:: c
 

include/clang/AST/Expr.h

   }
 
   friend TrailingObjects;
   friend class ASTStmtReader;
 };
 
 /// AtomicExpr - Variadic atomic builtins: __atomic_exchange, __atomic_fetch_*,
 /// __atomic_load, __atomic_store, and __atomic_compare_exchange_*, for the
-/// similarly-named C++11 instructions, and __c11 variants for <stdatomic.h>.
-/// All of these instructions take one primary pointer and at least one memory
-/// order.
+/// similarly-named C++11 instructions, and __c11 variants for <stdatomic.h>,
+/// and corresponding __opencl_atomic_* for OpenCL 2.0.
+/// All of these instructions take one primary pointer, at least one memory
+/// order, and one synchronization scope. The C++11 and __c11 atomic AtomicExpr
+/// always take the default synchronization scope.
 class AtomicExpr : public Expr {
 public:
   enum AtomicOp {
 #define BUILTIN(ID, TYPE, ATTRS)
 #define ATOMIC_BUILTIN(ID, TYPE, ATTRS) AO ## ID,
 #include "clang/Basic/Builtins.def"
     // Avoid trailing comma
     BI_First = 0
   };
 
 private:
-  enum { PTR, ORDER, VAL1, ORDER_FAIL, VAL2, WEAK, END_EXPR };
+  enum { PTR, ORDER, SCOPE, VAL1, ORDER_FAIL, VAL2, WEAK, END_EXPR };
   Stmt* SubExprs[END_EXPR];
   unsigned NumSubExprs;
   SourceLocation BuiltinLoc, RParenLoc;
   AtomicOp Op;
 
   friend class ASTStmtReader;
 
 public:
   AtomicExpr(SourceLocation BLoc, ArrayRef<Expr*> args, QualType t,
              AtomicOp op, SourceLocation RP);
 
   /// \brief Determine the number of arguments the specified atomic builtin
   /// should have.
   static unsigned getNumSubExprs(AtomicOp Op);
 
   /// \brief Build an empty AtomicExpr.
   explicit AtomicExpr(EmptyShell Empty) : Expr(AtomicExprClass, Empty) { }
 
   Expr *getPtr() const {
     return cast<Expr>(SubExprs[PTR]);
   }
   Expr *getOrder() const {
     return cast<Expr>(SubExprs[ORDER]);
   }
+  Expr *getScope() const {
+    return cast<Expr>(SubExprs[SCOPE]);
+  }
   Expr *getVal1() const {
-    if (Op == AO__c11_atomic_init)
+    if (Op == AO__c11_atomic_init || Op == AO__opencl_atomic_init)
       return cast<Expr>(SubExprs[ORDER]);
     assert(NumSubExprs > VAL1);
     return cast<Expr>(SubExprs[VAL1]);
   }
   Expr *getOrderFail() const {
     assert(NumSubExprs > ORDER_FAIL);
     return cast<Expr>(SubExprs[ORDER_FAIL]);
   }
   Expr *getVal2() const {
     if (Op == AO__atomic_exchange)
       return cast<Expr>(SubExprs[ORDER_FAIL]);
     assert(NumSubExprs > VAL2);
     return cast<Expr>(SubExprs[VAL2]);
   }
   Expr *getWeak() const {
     assert(NumSubExprs > WEAK);
     return cast<Expr>(SubExprs[WEAK]);
   }
+  QualType getValueType() const;
 
   AtomicOp getOp() const { return Op; }
   unsigned getNumSubExprs() const { return NumSubExprs; }
 
   Expr **getSubExprs() { return reinterpret_cast<Expr **>(SubExprs); }
   const Expr * const *getSubExprs() const {
     return reinterpret_cast<Expr * const *>(SubExprs);
   }
 
   bool isVolatile() const {
     return getPtr()->getType()->getPointeeType().isVolatileQualified();
   }
 
   bool isCmpXChg() const {
     return getOp() == AO__c11_atomic_compare_exchange_strong ||
            getOp() == AO__c11_atomic_compare_exchange_weak ||
+           getOp() == AO__opencl_atomic_compare_exchange_strong ||
+           getOp() == AO__opencl_atomic_compare_exchange_weak ||
            getOp() == AO__atomic_compare_exchange ||
            getOp() == AO__atomic_compare_exchange_n;
   }
 
+  bool isOpenCL() const {
+    return getOp() >= AO__opencl_atomic_init &&
+           getOp() <= AO__opencl_atomic_fetch_max;
+  }
+
   SourceLocation getBuiltinLoc() const { return BuiltinLoc; }
   SourceLocation getRParenLoc() const { return RParenLoc; }
 
   SourceLocation getLocStart() const LLVM_READONLY { return BuiltinLoc; }
   SourceLocation getLocEnd() const LLVM_READONLY { return RParenLoc; }
 
   static bool classof(const Stmt *T) {
     return T->getStmtClass() == AtomicExprClass;

include/clang/Basic/Builtins.def

 // builtin, and results in a clang::builtin::BIXX enum value for XX.
 
 // The second value provided to the macro specifies the type of the function
 // (result value, then each argument) as follows:
 //  v -> void
 //  b -> boolean
 //  c -> char
 //  s -> short
 //  i -> int
 //  h -> half
 //  f -> float
 //  d -> double
 //  z -> size_t
 //  w -> wchar_t
 //  F -> constant CFString
 //  G -> id
 //  H -> SEL
 //  M -> struct objc_super
 //  a -> __builtin_va_list
 //  A -> "reference" to __builtin_va_list

 ATOMIC_BUILTIN(__atomic_nand_fetch, "v.", "t")
 BUILTIN(__atomic_test_and_set, "bvD*i", "n")
 BUILTIN(__atomic_clear, "vvD*i", "n")
 BUILTIN(__atomic_thread_fence, "vi", "n")
 BUILTIN(__atomic_signal_fence, "vi", "n")
 BUILTIN(__atomic_always_lock_free, "izvCD*", "n")
 BUILTIN(__atomic_is_lock_free, "izvCD*", "n")
 
+// OpenCL 2.0 atomic builtins.
+ATOMIC_BUILTIN(__opencl_atomic_init, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_load, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_store, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_exchange, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_compare_exchange_strong, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_compare_exchange_weak, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_add, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_sub, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_and, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_or, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_xor, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_min, "v.", "t")
+ATOMIC_BUILTIN(__opencl_atomic_fetch_max, "v.", "t")
+
 #undef ATOMIC_BUILTIN
 
 // Non-overloaded atomic builtins.
 BUILTIN(__sync_synchronize, "v", "n")
 // GCC does not support these, they are a Clang extension.
 BUILTIN(__sync_fetch_and_min, "iiD*i", "n")
 BUILTIN(__sync_fetch_and_max, "iiD*i", "n")
 BUILTIN(__sync_fetch_and_umin, "UiUiD*Ui", "n")

include/clang/Basic/DiagnosticSemaKinds.td

   "too many %select{|||execution configuration }0arguments to "
   "%select{function|block|method|kernel function}0 call, "
   "expected %1, have %2">;
 def err_typecheck_call_too_many_args_one : Error<
   "too many %select{|||execution configuration }0arguments to "
   "%select{function|block|method|kernel function}0 call, "
   "expected single argument %1, have %2 arguments">;
 def err_typecheck_call_too_many_args_at_most : Error<
   "too many %select{|||execution configuration }0arguments to "
   "%select{function|block|method|kernel function}0 call, "
   "expected at most %1, have %2">;
 def err_typecheck_call_too_many_args_at_most_one : Error<
   "too many %select{|||execution configuration }0arguments to "
   "%select{function|block|method|kernel function}0 call, "
   "expected at most single argument %1, have %2 arguments">;
 def err_typecheck_call_too_many_args_suggest : Error<
   "too many %select{|||execution configuration }0arguments to "

   "type (%0 invalid)">;
 def err_atomic_exclusive_builtin_pointer_size : Error<
   "address argument to load or store exclusive builtin must be a pointer to"
   " 1,2,4 or 8 byte type (%0 invalid)">;
 def err_atomic_op_needs_atomic : Error<
   "address argument to atomic operation must be a pointer to _Atomic "
   "type (%0 invalid)">;
 def err_atomic_op_needs_non_const_atomic : Error<
-  "address argument to atomic operation must be a pointer to non-const _Atomic "
-  "type (%0 invalid)">;
+  "address argument to atomic operation must be a pointer to non-%select{const|constant}0 _Atomic "
+  "type (%1 invalid)">;
 def err_atomic_op_needs_non_const_pointer : Error<
   "address argument to atomic operation must be a pointer to non-const "
   "type (%0 invalid)">;
 def err_atomic_op_needs_trivial_copy : Error<
   "address argument to atomic operation must be a pointer to a "
   "trivially-copyable type (%0 invalid)">;
 def err_atomic_op_needs_atomic_int_or_ptr : Error<
   "address argument to atomic operation must be a pointer to %select{|atomic }0"
   "integer or pointer (%1 invalid)">;
 def err_atomic_op_bitwise_needs_atomic_int : Error<
   "address argument to bitwise atomic operation must be a pointer to "
   "%select{|atomic }0integer (%1 invalid)">;
 def warn_atomic_op_has_invalid_memory_order : Warning<
   "memory order argument to atomic operation is invalid">,
   InGroup<DiagGroup<"atomic-memory-ordering">>;
+def err_atomic_op_has_invalid_synch_scope : Error<
+  "synchronization scope argument to atomic operation is invalid">;
+def err_atomic_op_has_non_constant_synch_scope : Error<
+  "non-constant synchronization scope argument to atomic operation is not supported">;
 
 def err_overflow_builtin_must_be_int : Error<
   "operand argument to overflow builtin must be an integer (%0 invalid)">;
 def err_overflow_builtin_must_be_ptr_int : Error<
   "result argument to overflow builtin must be a pointer "
   "to a non-const integer (%0 invalid)">;
 
 def err_atomic_load_store_uses_lib : Error<

include/clang/Basic/SyncScope.h

+//===--- SyncScope.h - Atomic synchronization scopes ------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// \brief Provides definitions for the atomic synchronization scopes.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_BASIC_SYNCSCOPE_H
+#define LLVM_CLANG_BASIC_SYNCSCOPE_H
+
+namespace clang {
+
+/// \brief Defines the synch scope values used by the atomic builtins and
+/// expressions
+enum class SyncScope {
+  OpenCLWorkGroup = 1,
+  OpenCLDevice = 2,
+  OpenCLAllSVMDevices = 3,
+  OpenCLSubGroup = 4,
+};
+
+inline bool isValidSyncScopeValue(unsigned Scope) {
+  return Scope >= static_cast<unsigned>(SyncScope::OpenCLWorkGroup) &&
+         Scope <= static_cast<unsigned>(SyncScope::OpenCLSubGroup);
+}
+}
+
+#endif

lib/AST/ASTContext.cpp

   ObjCBuiltinBoolTy = (Target.useSignedCharForObjCBool() ?
                        SignedCharTy : BoolTy);
   
   ObjCConstantStringType = QualType();
   
   ObjCSuperType = QualType();
 
   // void * type
-  VoidPtrTy = getPointerType(VoidTy);
+  if (LangOpts.OpenCLVersion >= 200) {
+    auto Q = VoidTy.getQualifiers();
+    Q.setAddressSpace(LangAS::opencl_generic);
+    VoidPtrTy = getPointerType(getCanonicalType(
+        getQualifiedType(VoidTy.getUnqualifiedType(), Q)));
+  } else {
+    VoidPtrTy = getPointerType(VoidTy);
+  }
 
   // nullptr type (C++0x 2.14.7)
   InitBuiltinType(NullPtrTy,           BuiltinType::NullPtr);
 
   // half type (OpenCL 6.1.1.1) / ARM NEON __fp16
   InitBuiltinType(HalfTy, BuiltinType::Half);
 
   // Builtin type used to help define __builtin_va_list.

lib/AST/Expr.cpp

     if (const VariableArrayType *T =
             dyn_cast<VariableArrayType>(getArgumentType().getTypePtr()))
       return const_child_range(const_child_iterator(T), const_child_iterator());
     return const_child_range(const_child_iterator(), const_child_iterator());
   }
   return const_child_range(&Argument.Ex, &Argument.Ex + 1);
 }
 
 AtomicExpr::AtomicExpr(SourceLocation BLoc, ArrayRef<Expr*> args,
                        QualType t, AtomicOp op, SourceLocation RP)
   : Expr(AtomicExprClass, t, VK_RValue, OK_Ordinary,
          false, false, false, false),
     NumSubExprs(args.size()), BuiltinLoc(BLoc), RParenLoc(RP), Op(op)
 {
   assert(args.size() == getNumSubExprs(op) && "wrong number of subexpressions");
   for (unsigned i = 0; i != args.size(); i++) {
     if (args[i]->isTypeDependent())
       ExprBits.TypeDependent = true;
     if (args[i]->isValueDependent())
       ExprBits.ValueDependent = true;
     if (args[i]->isInstantiationDependent())
       ExprBits.InstantiationDependent = true;
     if (args[i]->containsUnexpandedParameterPack())
       ExprBits.ContainsUnexpandedParameterPack = true;
 
     SubExprs[i] = args[i];
   }
 }
 
 unsigned AtomicExpr::getNumSubExprs(AtomicOp Op) {
   switch (Op) {
   case AO__c11_atomic_init:
+  case AO__opencl_atomic_init:
+    return 2;
   case AO__c11_atomic_load:
+  case AO__opencl_atomic_load:
   case AO__atomic_load_n:
-    return 2;
+    return 3;
 
   case AO__c11_atomic_store:
   case AO__c11_atomic_exchange:
+  case AO__opencl_atomic_store:
+  case AO__opencl_atomic_exchange:
   case AO__atomic_load:
   case AO__atomic_store:
   case AO__atomic_store_n:
   case AO__atomic_exchange_n:
   case AO__c11_atomic_fetch_add:
   case AO__c11_atomic_fetch_sub:
   case AO__c11_atomic_fetch_and:
   case AO__c11_atomic_fetch_or:
   case AO__c11_atomic_fetch_xor:
+  case AO__opencl_atomic_fetch_add:
+  case AO__opencl_atomic_fetch_sub:
+  case AO__opencl_atomic_fetch_and:
+  case AO__opencl_atomic_fetch_or:
+  case AO__opencl_atomic_fetch_xor:
+  case AO__opencl_atomic_fetch_min:
+  case AO__opencl_atomic_fetch_max:
   case AO__atomic_fetch_add:
   case AO__atomic_fetch_sub:
   case AO__atomic_fetch_and:
   case AO__atomic_fetch_or:
   case AO__atomic_fetch_xor:
   case AO__atomic_fetch_nand:
   case AO__atomic_add_fetch:
   case AO__atomic_sub_fetch:
   case AO__atomic_and_fetch:
   case AO__atomic_or_fetch:
   case AO__atomic_xor_fetch:
   case AO__atomic_nand_fetch:
-    return 3;
+    return 4;
 
   case AO__atomic_exchange:
-    return 4;
+    return 5;
 
   case AO__c11_atomic_compare_exchange_strong:
   case AO__c11_atomic_compare_exchange_weak:
-    return 5;
+  case AO__opencl_atomic_compare_exchange_strong:
+  case AO__opencl_atomic_compare_exchange_weak:
+    return 6;
 
   case AO__atomic_compare_exchange:
   case AO__atomic_compare_exchange_n:
-    return 6;
+    return 7;
   }
   llvm_unreachable("unknown atomic op");
 }
 
+QualType AtomicExpr::getValueType() const {
+  auto T = getPtr()->getType()->castAs<PointerType>()->getPointeeType();
+  if (auto AT = T->getAs<AtomicType>()) {
+    return AT->getValueType();
+  } else {
+    return T;
+  }
+}
+
 QualType OMPArraySectionExpr::getBaseOriginalType(const Expr *Base) {
   unsigned ArraySectionCount = 0;
   while (auto *OASE = dyn_cast<OMPArraySectionExpr>(Base->IgnoreParens())) {
     Base = OASE->getBase();
     ++ArraySectionCount;
   }
   while (auto *ASE =
              dyn_cast<ArraySubscriptExpr>(Base->IgnoreParenImpCasts())) {

lib/AST/StmtPrinter.cpp

     break;
 #include "clang/Basic/Builtins.def"
   }
   OS << Name;
 
   // AtomicExpr stores its subexpressions in a permuted order.
   PrintExpr(Node->getPtr());
   if (Node->getOp() != AtomicExpr::AO__c11_atomic_load &&
-      Node->getOp() != AtomicExpr::AO__atomic_load_n) {
+      Node->getOp() != AtomicExpr::AO__atomic_load_n &&
+      Node->getOp() != AtomicExpr::AO__opencl_atomic_load) {
     OS << ", ";
     PrintExpr(Node->getVal1());
   }
   if (Node->getOp() == AtomicExpr::AO__atomic_exchange ||
       Node->isCmpXChg()) {
     OS << ", ";
     PrintExpr(Node->getVal2());
   }
   if (Node->getOp() == AtomicExpr::AO__atomic_compare_exchange ||
       Node->getOp() == AtomicExpr::AO__atomic_compare_exchange_n) {
     OS << ", ";
     PrintExpr(Node->getWeak());
   }
-  if (Node->getOp() != AtomicExpr::AO__c11_atomic_init) {
+  if (Node->getOp() != AtomicExpr::AO__c11_atomic_init &&
+      Node->getOp() != AtomicExpr::AO__opencl_atomic_init) {
     OS << ", ";
     PrintExpr(Node->getOrder());
   }
   if (Node->isCmpXChg()) {
     OS << ", ";
     PrintExpr(Node->getOrderFail());
   }
   OS << ")";

lib/Basic/Targets/AMDGPU.cpp

AMDGPUTargetInfo::AMDGPUTargetInfo(const llvm::Triple &Triple,
   // Set pointer width and alignment for target address space 0.
   PointerWidth = PointerAlign = DataLayout->getPointerSizeInBits();
   if (getMaxPointerWidth() == 64) {
     LongWidth = LongAlign = 64;
     SizeType = UnsignedLong;
     PtrDiffType = SignedLong;
     IntPtrType = SignedLong;
   }
+
+  MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
 }
 
 void AMDGPUTargetInfo::adjust(LangOptions &Opts) {
   TargetInfo::adjust(Opts);
   setAddressSpaceMap(Opts.OpenCL || !isAMDGCN(getTriple()));
 }
 
 ArrayRef<Builtin::Info> AMDGPUTargetInfo::getTargetBuiltins() const {

lib/CodeGen/CGAtomic.cpp

 // This file contains the code for emitting atomic operations.
 //
 //===----------------------------------------------------------------------===//
 
 #include "CGCall.h"
 #include "CGRecordLayout.h"
 #include "CodeGenFunction.h"
 #include "CodeGenModule.h"
+#include "TargetInfo.h"
 #include "clang/AST/ASTContext.h"
 #include "clang/CodeGen/CGFunctionInfo.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/Operator.h"
 
 using namespace clang;
 using namespace CodeGen;
AtomicInfo(CodeGenFunction &CGF, LValue &lvalue)
                                             /*IndexTypeQuals=*/0);
         }
         AtomicAlign = ValueAlign = lvalue.getAlignment();
       } else if (lvalue.isVectorElt()) {
         ValueTy = lvalue.getType()->getAs<VectorType>()->getElementType();
         ValueSizeInBits = C.getTypeSize(ValueTy);
         AtomicTy = lvalue.getType();
         AtomicSizeInBits = C.getTypeSize(AtomicTy);
         AtomicAlign = ValueAlign = lvalue.getAlignment();
         LVal = lvalue;
       } else {
         assert(lvalue.isExtVectorElt());
         ValueTy = lvalue.getType();
         ValueSizeInBits = C.getTypeSize(ValueTy);
         AtomicTy = ValueTy = CGF.getContext().getExtVectorType(
             lvalue.getType(), lvalue.getExtVectorAddress()
                                   .getElementType()->getVectorNumElements());
         AtomicSizeInBits = C.getTypeSize(AtomicTy);
         AtomicAlign = ValueAlign = lvalue.getAlignment();
         LVal = lvalue;
       }
       UseLibcall = !C.getTargetInfo().hasBuiltinAtomic(
           AtomicSizeInBits, C.toBits(lvalue.getAlignment()));
     }
 
     QualType getAtomicType() const { return AtomicTy; }
     QualType getValueType() const { return ValueTy; }
     CharUnits getAtomicAlignment() const { return AtomicAlign; }
     CharUnits getValueAlignment() const { return ValueAlign; }
     uint64_t getAtomicSizeInBits() const { return AtomicSizeInBits; }
bool AtomicInfo::emitMemSetZeroIfNecessary() const {
   return true;
 }
 
 static void emitAtomicCmpXchg(CodeGenFunction &CGF, AtomicExpr *E, bool IsWeak,
                               Address Dest, Address Ptr,
                               Address Val1, Address Val2,
                               uint64_t Size,
                               llvm::AtomicOrdering SuccessOrder,
-                              llvm::AtomicOrdering FailureOrder) {
+                              llvm::AtomicOrdering FailureOrder,
+                              llvm::SyncScope::ID Scope) {
   // Note that cmpxchg doesn't support weak cmpxchg, at least at the moment.
   llvm::Value *Expected = CGF.Builder.CreateLoad(Val1);
   llvm::Value *Desired = CGF.Builder.CreateLoad(Val2);
 
   llvm::AtomicCmpXchgInst *Pair = CGF.Builder.CreateAtomicCmpXchg(
-      Ptr.getPointer(), Expected, Desired, SuccessOrder, FailureOrder);
+      Ptr.getPointer(), Expected, Desired, SuccessOrder, FailureOrder,
+      Scope);
   Pair->setVolatile(E->isVolatile());
   Pair->setWeak(IsWeak);
 
   // Cmp holds the result of the compare-exchange operation: true on success,
   // false on failure.
   llvm::Value *Old = CGF.Builder.CreateExtractValue(Pair, 0);
   llvm::Value *Cmp = CGF.Builder.CreateExtractValue(Pair, 1);
 

 /// Given an ordering required on success, emit all possible cmpxchg
 /// instructions to cope with the provided (but possibly only dynamically known)
 /// FailureOrder.
 static void emitAtomicCmpXchgFailureSet(CodeGenFunction &CGF, AtomicExpr *E,
                                         bool IsWeak, Address Dest, Address Ptr,
                                         Address Val1, Address Val2,
                                         llvm::Value *FailureOrderVal,
                                         uint64_t Size,
-                                        llvm::AtomicOrdering SuccessOrder) {
+                                        llvm::AtomicOrdering SuccessOrder,
+                                        llvm::SyncScope::ID Scope) {
   llvm::AtomicOrdering FailureOrder;
   if (llvm::ConstantInt *FO = dyn_cast<llvm::ConstantInt>(FailureOrderVal)) {
     auto FOS = FO->getSExtValue();
     if (!llvm::isValidAtomicOrderingCABI(FOS))
       FailureOrder = llvm::AtomicOrdering::Monotonic;
     else
       switch ((llvm::AtomicOrderingCABI)FOS) {
       case llvm::AtomicOrderingCABI::relaxed:
else
       }
     if (isStrongerThan(FailureOrder, SuccessOrder)) {
       // Don't assert on undefined behavior "failure argument shall be no
       // stronger than the success argument".
       FailureOrder =
           llvm::AtomicCmpXchgInst::getStrongestFailureOrdering(SuccessOrder);
     }
     emitAtomicCmpXchg(CGF, E, IsWeak, Dest, Ptr, Val1, Val2, Size, SuccessOrder,
-                      FailureOrder);
+                      FailureOrder, Scope);
     return;
   }
 
   // Create all the relevant BB's
   llvm::BasicBlock *MonotonicBB = nullptr, *AcquireBB = nullptr,
                    *SeqCstBB = nullptr;
   MonotonicBB = CGF.createBasicBlock("monotonic_fail", CGF.CurFn);
   if (SuccessOrder != llvm::AtomicOrdering::Monotonic &&
       SuccessOrder != llvm::AtomicOrdering::Release)
     AcquireBB = CGF.createBasicBlock("acquire_fail", CGF.CurFn);
   if (SuccessOrder == llvm::AtomicOrdering::SequentiallyConsistent)
     SeqCstBB = CGF.createBasicBlock("seqcst_fail", CGF.CurFn);
 
   llvm::BasicBlock *ContBB = CGF.createBasicBlock("atomic.continue", CGF.CurFn);
 
   llvm::SwitchInst *SI = CGF.Builder.CreateSwitch(FailureOrderVal, MonotonicBB);
 
   // Emit all the different atomics
 
   // MonotonicBB is arbitrarily chosen as the default case; in practice, this
   // doesn't matter unless someone is crazy enough to use something that
   // doesn't fold to a constant for the ordering.
   CGF.Builder.SetInsertPoint(MonotonicBB);
   emitAtomicCmpXchg(CGF, E, IsWeak, Dest, Ptr, Val1, Val2,
-                    Size, SuccessOrder, llvm::AtomicOrdering::Monotonic);
+                    Size, SuccessOrder, llvm::AtomicOrdering::Monotonic, Scope);
   CGF.Builder.CreateBr(ContBB);
 
   if (AcquireBB) {
     CGF.Builder.SetInsertPoint(AcquireBB);
     emitAtomicCmpXchg(CGF, E, IsWeak, Dest, Ptr, Val1, Val2,
-                      Size, SuccessOrder, llvm::AtomicOrdering::Acquire);
+                      Size, SuccessOrder, llvm::AtomicOrdering::Acquire, Scope);
     CGF.Builder.CreateBr(ContBB);
     SI->addCase(CGF.Builder.getInt32((int)llvm::AtomicOrderingCABI::consume),
                 AcquireBB);
     SI->addCase(CGF.Builder.getInt32((int)llvm::AtomicOrderingCABI::acquire),
                 AcquireBB);
   }
   if (SeqCstBB) {
     CGF.Builder.SetInsertPoint(SeqCstBB);
     emitAtomicCmpXchg(CGF, E, IsWeak, Dest, Ptr, Val1, Val2, Size, SuccessOrder,
-                      llvm::AtomicOrdering::SequentiallyConsistent);
+                      llvm::AtomicOrdering::SequentiallyConsistent, Scope);
     CGF.Builder.CreateBr(ContBB);
     SI->addCase(CGF.Builder.getInt32((int)llvm::AtomicOrderingCABI::seq_cst),
                 SeqCstBB);
   }
 
   CGF.Builder.SetInsertPoint(ContBB);
 }
 
 static void EmitAtomicOp(CodeGenFunction &CGF, AtomicExpr *E, Address Dest,
                          Address Ptr, Address Val1, Address Val2,
                          llvm::Value *IsWeak, llvm::Value *FailureOrder,
-                         uint64_t Size, llvm::AtomicOrdering Order) {
+                         uint64_t Size, llvm::AtomicOrdering Order,
+                         llvm::SyncScope::ID Scope) {
   llvm::AtomicRMWInst::BinOp Op = llvm::AtomicRMWInst::Add;
   llvm::Instruction::BinaryOps PostOp = (llvm::Instruction::BinaryOps)0;
 
   switch (E->getOp()) {
   case AtomicExpr::AO__c11_atomic_init:
+  case AtomicExpr::AO__opencl_atomic_init:
     llvm_unreachable("Already handled!");
 
   case AtomicExpr::AO__c11_atomic_compare_exchange_strong:
+  case AtomicExpr::AO__opencl_atomic_compare_exchange_strong:
     emitAtomicCmpXchgFailureSet(CGF, E, false, Dest, Ptr, Val1, Val2,
-                                FailureOrder, Size, Order);
+                                FailureOrder, Size, Order, Scope);
     return;
   case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
+  case AtomicExpr::AO__opencl_atomic_compare_exchange_weak:
     emitAtomicCmpXchgFailureSet(CGF, E, true, Dest, Ptr, Val1, Val2,
-                                FailureOrder, Size, Order);
+                                FailureOrder, Size, Order, Scope);
     return;
   case AtomicExpr::AO__atomic_compare_exchange:
   case AtomicExpr::AO__atomic_compare_exchange_n: {
     if (llvm::ConstantInt *IsWeakC = dyn_cast<llvm::ConstantInt>(IsWeak)) {
       emitAtomicCmpXchgFailureSet(CGF, E, IsWeakC->getZExtValue(), Dest, Ptr,
-                                  Val1, Val2, FailureOrder, Size, Order);
+                                  Val1, Val2, FailureOrder, Size, Order, Scope);
     } else {
       // Create all the relevant BB's
       llvm::BasicBlock *StrongBB =
           CGF.createBasicBlock("cmpxchg.strong", CGF.CurFn);
       llvm::BasicBlock *WeakBB = CGF.createBasicBlock("cmxchg.weak", CGF.CurFn);
       llvm::BasicBlock *ContBB =
           CGF.createBasicBlock("cmpxchg.continue", CGF.CurFn);
 
       llvm::SwitchInst *SI = CGF.Builder.CreateSwitch(IsWeak, WeakBB);
       SI->addCase(CGF.Builder.getInt1(false), StrongBB);
 
       CGF.Builder.SetInsertPoint(StrongBB);
       emitAtomicCmpXchgFailureSet(CGF, E, false, Dest, Ptr, Val1, Val2,
-                                  FailureOrder, Size, Order);
+                                  FailureOrder, Size, Order, Scope);
       CGF.Builder.CreateBr(ContBB);
 
       CGF.Builder.SetInsertPoint(WeakBB);
       emitAtomicCmpXchgFailureSet(CGF, E, true, Dest, Ptr, Val1, Val2,
-                                  FailureOrder, Size, Order);
+                                  FailureOrder, Size, Order, Scope);
       CGF.Builder.CreateBr(ContBB);
 
       CGF.Builder.SetInsertPoint(ContBB);
     }
     return;
   }
   case AtomicExpr::AO__c11_atomic_load:
+  case AtomicExpr::AO__opencl_atomic_load:
   case AtomicExpr::AO__atomic_load_n:
   case AtomicExpr::AO__atomic_load: {
     llvm::LoadInst *Load = CGF.Builder.CreateLoad(Ptr);
-    Load->setAtomic(Order);
+    Load->setAtomic(Order, Scope);
     Load->setVolatile(E->isVolatile());
     CGF.Builder.CreateStore(Load, Dest);
     return;
   }
 
   case AtomicExpr::AO__c11_atomic_store:
+  case AtomicExpr::AO__opencl_atomic_store:
   case AtomicExpr::AO__atomic_store:
   case AtomicExpr::AO__atomic_store_n: {
     llvm::Value *LoadVal1 = CGF.Builder.CreateLoad(Val1);
     llvm::StoreInst *Store = CGF.Builder.CreateStore(LoadVal1, Ptr);
-    Store->setAtomic(Order);
+    Store->setAtomic(Order, Scope);
     Store->setVolatile(E->isVolatile());
     return;
   }
 
   case AtomicExpr::AO__c11_atomic_exchange:
+  case AtomicExpr::AO__opencl_atomic_exchange:
   case AtomicExpr::AO__atomic_exchange_n:
   case AtomicExpr::AO__atomic_exchange:
     Op = llvm::AtomicRMWInst::Xchg;
     break;
 
   case AtomicExpr::AO__atomic_add_fetch:
     PostOp = llvm::Instruction::Add;
     // Fall through.
   case AtomicExpr::AO__c11_atomic_fetch_add:
+  case AtomicExpr::AO__opencl_atomic_fetch_add:
   case AtomicExpr::AO__atomic_fetch_add:
     Op = llvm::AtomicRMWInst::Add;
     break;
 
   case AtomicExpr::AO__atomic_sub_fetch:
     PostOp = llvm::Instruction::Sub;
     // Fall through.
   case AtomicExpr::AO__c11_atomic_fetch_sub:
+  case AtomicExpr::AO__opencl_atomic_fetch_sub:
   case AtomicExpr::AO__atomic_fetch_sub:
     Op = llvm::AtomicRMWInst::Sub;
     break;
 
+  case AtomicExpr::AO__opencl_atomic_fetch_min:
+    Op = E->getValueType()->isSignedIntegerType() ? llvm::AtomicRMWInst::Min
+                                                  : llvm::AtomicRMWInst::UMin;
+    break;
+
+  case AtomicExpr::AO__opencl_atomic_fetch_max:
+    Op = E->getValueType()->isSignedIntegerType() ? llvm::AtomicRMWInst::Max
+                                                  : llvm::AtomicRMWInst::UMax;
+    break;
+
   case AtomicExpr::AO__atomic_and_fetch:
     PostOp = llvm::Instruction::And;
     // Fall through.
   case AtomicExpr::AO__c11_atomic_fetch_and:
+  case AtomicExpr::AO__opencl_atomic_fetch_and:
   case AtomicExpr::AO__atomic_fetch_and:
     Op = llvm::AtomicRMWInst::And;
     break;
 
   case AtomicExpr::AO__atomic_or_fetch:
     PostOp = llvm::Instruction::Or;
     // Fall through.
   case AtomicExpr::AO__c11_atomic_fetch_or:
+  case AtomicExpr::AO__opencl_atomic_fetch_or:
   case AtomicExpr::AO__atomic_fetch_or:
     Op = llvm::AtomicRMWInst::Or;
     break;
 
   case AtomicExpr::AO__atomic_xor_fetch:
     PostOp = llvm::Instruction::Xor;
     // Fall through.
   case AtomicExpr::AO__c11_atomic_fetch_xor:
+  case AtomicExpr::AO__opencl_atomic_fetch_xor:
   case AtomicExpr::AO__atomic_fetch_xor:
     Op = llvm::AtomicRMWInst::Xor;
     break;
 
   case AtomicExpr::AO__atomic_nand_fetch:
     PostOp = llvm::Instruction::And; // the NOT is special cased below
   // Fall through.
   case AtomicExpr::AO__atomic_fetch_nand:
     Op = llvm::AtomicRMWInst::Nand;
     break;
   }
 
   llvm::Value *LoadVal1 = CGF.Builder.CreateLoad(Val1);
   llvm::AtomicRMWInst *RMWI =
-      CGF.Builder.CreateAtomicRMW(Op, Ptr.getPointer(), LoadVal1, Order);
+      CGF.Builder.CreateAtomicRMW(Op, Ptr.getPointer(), LoadVal1, Order, Scope);
   RMWI->setVolatile(E->isVolatile());
 
   // For __atomic_*_fetch operations, perform the operation again to
   // determine the value which was written.
   llvm::Value *Result = RMWI;
   if (PostOp)
     Result = CGF.Builder.CreateBinOp(PostOp, RMWI, LoadVal1);
   if (E->getOp() == AtomicExpr::AO__atomic_nand_fetch)
RValue CodeGenFunction::EmitAtomicExpr(AtomicExpr *E) {
 
   llvm::Value *IsWeak = nullptr, *OrderFail = nullptr;
 
   Address Val1 = Address::invalid();
   Address Val2 = Address::invalid();
   Address Dest = Address::invalid();
   Address Ptr(EmitScalarExpr(E->getPtr()), alignChars);
 
-  if (E->getOp() == AtomicExpr::AO__c11_atomic_init) {
+  if (E->getOp() == AtomicExpr::AO__c11_atomic_init ||
+      E->getOp() == AtomicExpr::AO__opencl_atomic_init) {
     LValue lvalue = MakeAddrLValue(Ptr, AtomicTy);
     EmitAtomicInit(E->getVal1(), lvalue);
     return RValue::get(nullptr);
   }
 
   llvm::Value *Order = EmitScalarExpr(E->getOrder());
+  llvm::Value *Scope = EmitScalarExpr(E->getScope());
 
   switch (E->getOp()) {
   case AtomicExpr::AO__c11_atomic_init:
+  case AtomicExpr::AO__opencl_atomic_init:
     llvm_unreachable("Already handled above with EmitAtomicInit!");
 
   case AtomicExpr::AO__c11_atomic_load:
+  case AtomicExpr::AO__opencl_atomic_load:
   case AtomicExpr::AO__atomic_load_n:
     break;
 
   case AtomicExpr::AO__atomic_load:
     Dest = EmitPointerWithAlignment(E->getVal1());
     break;
 
   case AtomicExpr::AO__atomic_store:
     Val1 = EmitPointerWithAlignment(E->getVal1());
     break;
 
   case AtomicExpr::AO__atomic_exchange:
     Val1 = EmitPointerWithAlignment(E->getVal1());
     Dest = EmitPointerWithAlignment(E->getVal2());
     break;
 
   case AtomicExpr::AO__c11_atomic_compare_exchange_strong:
   case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
+  case AtomicExpr::AO__opencl_atomic_compare_exchange_strong:
+  case AtomicExpr::AO__opencl_atomic_compare_exchange_weak:
   case AtomicExpr::AO__atomic_compare_exchange_n:
   case AtomicExpr::AO__atomic_compare_exchange:
     Val1 = EmitPointerWithAlignment(E->getVal1());
     if (E->getOp() == AtomicExpr::AO__atomic_compare_exchange)
       Val2 = EmitPointerWithAlignment(E->getVal2());
     else
       Val2 = EmitValToTemp(*this, E->getVal2());
     OrderFail = EmitScalarExpr(E->getOrderFail());
-    if (E->getNumSubExprs() == 6)
+    if (E->getOp() == AtomicExpr::AO__atomic_compare_exchange_n ||
+        E->getOp() == AtomicExpr::AO__atomic_compare_exchange)
       IsWeak = EmitScalarExpr(E->getWeak());
     break;
 
   case AtomicExpr::AO__c11_atomic_fetch_add:
   case AtomicExpr::AO__c11_atomic_fetch_sub:
+  case AtomicExpr::AO__opencl_atomic_fetch_add:
+  case AtomicExpr::AO__opencl_atomic_fetch_sub:
     if (MemTy->isPointerType()) {
       // For pointer arithmetic, we're required to do a bit of math:
       // adding 1 to an int* is not the same as adding 1 to a uintptr_t.
       // ... but only for the C11 builtins. The GNU builtins expect the
       // user to multiply by sizeof(T).
       QualType Val1Ty = E->getVal1()->getType();
       llvm::Value *Val1Scalar = EmitScalarExpr(E->getVal1());
       CharUnits PointeeIncAmt =
           getContext().getTypeSizeInChars(MemTy->getPointeeType());
       Val1Scalar = Builder.CreateMul(Val1Scalar, CGM.getSize(PointeeIncAmt));
       auto Temp = CreateMemTemp(Val1Ty, ".atomictmp");
       Val1 = Temp;
       EmitStoreOfScalar(Val1Scalar, MakeAddrLValue(Temp, Val1Ty));
       break;
     }
     // Fall through.
   case AtomicExpr::AO__atomic_fetch_add:
   case AtomicExpr::AO__atomic_fetch_sub:
   case AtomicExpr::AO__atomic_add_fetch:
   case AtomicExpr::AO__atomic_sub_fetch:
   case AtomicExpr::AO__c11_atomic_store:
   case AtomicExpr::AO__c11_atomic_exchange:
+  case AtomicExpr::AO__opencl_atomic_store:
+  case AtomicExpr::AO__opencl_atomic_exchange:
   case AtomicExpr::AO__atomic_store_n:
   case AtomicExpr::AO__atomic_exchange_n:
   case AtomicExpr::AO__c11_atomic_fetch_and:
   case AtomicExpr::AO__c11_atomic_fetch_or:
   case AtomicExpr::AO__c11_atomic_fetch_xor:
+  case AtomicExpr::AO__opencl_atomic_fetch_and:
+  case AtomicExpr::AO__opencl_atomic_fetch_or:
+  case AtomicExpr::AO__opencl_atomic_fetch_xor:
+  case AtomicExpr::AO__opencl_atomic_fetch_min:
+  case AtomicExpr::AO__opencl_atomic_fetch_max:
   case AtomicExpr::AO__atomic_fetch_and:
   case AtomicExpr::AO__atomic_fetch_or:
   case AtomicExpr::AO__atomic_fetch_xor:
   case AtomicExpr::AO__atomic_fetch_nand:
   case AtomicExpr::AO__atomic_and_fetch:
   case AtomicExpr::AO__atomic_or_fetch:
   case AtomicExpr::AO__atomic_xor_fetch:
   case AtomicExpr::AO__atomic_nand_fetch:
RValue CodeGenFunction::EmitAtomicExpr(AtomicExpr *E) {
   else if (!RValTy->isVoidType())
     Dest = Atomics.emitCastToAtomicIntPointer(Atomics.CreateTempAlloca());
 
   // Use a library call.  See: http://gcc.gnu.org/wiki/Atomic/GCCMM/LIbrary .
   if (UseLibcall) {
     bool UseOptimizedLibcall = false;
     switch (E->getOp()) {
     case AtomicExpr::AO__c11_atomic_init:
+    case AtomicExpr::AO__opencl_atomic_init:
       llvm_unreachable("Already handled above with EmitAtomicInit!");
 
     case AtomicExpr::AO__c11_atomic_fetch_add:
+    case AtomicExpr::AO__opencl_atomic_fetch_add:
     case AtomicExpr::AO__atomic_fetch_add:
     case AtomicExpr::AO__c11_atomic_fetch_and:
+    case AtomicExpr::AO__opencl_atomic_fetch_and:
     case AtomicExpr::AO__atomic_fetch_and:
     case AtomicExpr::AO__c11_atomic_fetch_or:
+    case AtomicExpr::AO__opencl_atomic_fetch_or:
     case AtomicExpr::AO__atomic_fetch_or:
     case AtomicExpr::AO__atomic_fetch_nand:
     case AtomicExpr::AO__c11_atomic_fetch_sub:
+    case AtomicExpr::AO__opencl_atomic_fetch_sub:
     case AtomicExpr::AO__atomic_fetch_sub:
     case AtomicExpr::AO__c11_atomic_fetch_xor:
+    case AtomicExpr::AO__opencl_atomic_fetch_xor:
+    case AtomicExpr::AO__opencl_atomic_fetch_min:
+    case AtomicExpr::AO__opencl_atomic_fetch_max:
     case AtomicExpr::AO__atomic_fetch_xor:
     case AtomicExpr::AO__atomic_add_fetch:
     case AtomicExpr::AO__atomic_and_fetch:
     case AtomicExpr::AO__atomic_nand_fetch:
     case AtomicExpr::AO__atomic_or_fetch:
     case AtomicExpr::AO__atomic_sub_fetch:
     case AtomicExpr::AO__atomic_xor_fetch:
       // For these, only library calls for certain sizes exist.
       UseOptimizedLibcall = true;
       break;
 
     case AtomicExpr::AO__c11_atomic_load:
     case AtomicExpr::AO__c11_atomic_store:
     case AtomicExpr::AO__c11_atomic_exchange:
     case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
     case AtomicExpr::AO__c11_atomic_compare_exchange_strong:
+    case AtomicExpr::AO__opencl_atomic_load:
+    case AtomicExpr::AO__opencl_atomic_store:
+    case AtomicExpr::AO__opencl_atomic_exchange:
+    case AtomicExpr::AO__opencl_atomic_compare_exchange_weak:
+    case AtomicExpr::AO__opencl_atomic_compare_exchange_strong:
     case AtomicExpr::AO__atomic_load_n:
     case AtomicExpr::AO__atomic_load:
     case AtomicExpr::AO__atomic_store_n:
     case AtomicExpr::AO__atomic_store:
     case AtomicExpr::AO__atomic_exchange_n:
     case AtomicExpr::AO__atomic_exchange:
     case AtomicExpr::AO__atomic_compare_exchange_n:
     case AtomicExpr::AO__atomic_compare_exchange:
       // Only use optimized library calls for sizes for which they exist.
       if (Size == 1 || Size == 2 || Size == 4 || Size == 8)
         UseOptimizedLibcall = true;
       break;
     }
 
     CallArgList Args;
     if (!UseOptimizedLibcall) {
       // For non-optimized library calls, the size is the first parameter
       Args.add(RValue::get(llvm::ConstantInt::get(SizeTy, Size)),
                getContext().getSizeType());
     }
     // Atomic address is the first or second parameter
-    Args.add(RValue::get(EmitCastToVoidPtr(Ptr.getPointer())),
+    // The OpenCL atomic library functions only accept pointer arguments to
+    // generic address space.
+    auto CastToGenericAddrSpace = [&](llvm::Value *V, QualType PT) {
+      if (!E->isOpenCL())
+        return V;
+      auto AS = PT->getAs<PointerType>()->getPointeeType().getAddressSpace();
+      if (AS == LangAS::opencl_generic)
+        return V;
+      auto DestAS = getContext().getTargetAddressSpace(LangAS::opencl_generic);
+      auto T = V->getType();
+      auto *DestType = T->getPointerElementType()->getPointerTo(DestAS);
+
+      return getTargetHooks().performAddrSpaceCast(
+          *this, V, AS, LangAS::opencl_generic, DestType, false);
+    };
+
+    Args.add(RValue::get(CastToGenericAddrSpace(
+                 EmitCastToVoidPtr(Ptr.getPointer()), E->getPtr()->getType())),
              getContext().VoidPtrTy);
 
     std::string LibCallName;
     QualType LoweredMemTy =
       MemTy->isPointerType() ? getContext().getIntPtrType() : MemTy;
     QualType RetTy;
     bool HaveRetTy = false;
     llvm::Instruction::BinaryOps PostOp = (llvm::Instruction::BinaryOps)0;
     switch (E->getOp()) {
     case AtomicExpr::AO__c11_atomic_init:
+    case AtomicExpr::AO__opencl_atomic_init:
       llvm_unreachable("Already handled!");
 
     // There is only one libcall for compare an exchange, because there is no
     // optimisation benefit possible from a libcall version of a weak compare
     // and exchange.
     // bool __atomic_compare_exchange(size_t size, void *mem, void *expected,
     //                                void *desired, int success, int failure)
     // bool __atomic_compare_exchange_N(T *mem, T *expected, T desired,
     //                                  int success, int failure)
     case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
     case AtomicExpr::AO__c11_atomic_compare_exchange_strong:
+    case AtomicExpr::AO__opencl_atomic_compare_exchange_weak:
+    case AtomicExpr::AO__opencl_atomic_compare_exchange_strong:
     case AtomicExpr::AO__atomic_compare_exchange:
     case AtomicExpr::AO__atomic_compare_exchange_n:
       LibCallName = "__atomic_compare_exchange";
       RetTy = getContext().BoolTy;
       HaveRetTy = true;
-      Args.add(RValue::get(EmitCastToVoidPtr(Val1.getPointer())),
-               getContext().VoidPtrTy);
+      Args.add(
+          RValue::get(CastToGenericAddrSpace(
+              EmitCastToVoidPtr(Val1.getPointer()), E->getVal1()->getType())),
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val2.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       Args.add(RValue::get(Order), getContext().IntTy);
       Order = OrderFail;
       break;
     // void __atomic_exchange(size_t size, void *mem, void *val, void *return,
     //                        int order)
     // T __atomic_exchange_N(T *mem, T val, int order)
     case AtomicExpr::AO__c11_atomic_exchange:
+    case AtomicExpr::AO__opencl_atomic_exchange:
     case AtomicExpr::AO__atomic_exchange_n:
     case AtomicExpr::AO__atomic_exchange:
       LibCallName = "__atomic_exchange";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       break;
     // void __atomic_store(size_t size, void *mem, void *val, int order)
     // void __atomic_store_N(T *mem, T val, int order)
     case AtomicExpr::AO__c11_atomic_store:
+    case AtomicExpr::AO__opencl_atomic_store:
     case AtomicExpr::AO__atomic_store:
     case AtomicExpr::AO__atomic_store_n:
       LibCallName = "__atomic_store";
       RetTy = getContext().VoidTy;
       HaveRetTy = true;
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       break;
     // void __atomic_load(size_t size, void *mem, void *return, int order)
     // T __atomic_load_N(T *mem, int order)
     case AtomicExpr::AO__c11_atomic_load:
+    case AtomicExpr::AO__opencl_atomic_load:
     case AtomicExpr::AO__atomic_load:
     case AtomicExpr::AO__atomic_load_n:
       LibCallName = "__atomic_load";
       break;
     // T __atomic_add_fetch_N(T *mem, T val, int order)
     // T __atomic_fetch_add_N(T *mem, T val, int order)
     case AtomicExpr::AO__atomic_add_fetch:
       PostOp = llvm::Instruction::Add;
     // Fall through.
     case AtomicExpr::AO__c11_atomic_fetch_add:
+    case AtomicExpr::AO__opencl_atomic_fetch_add:
     case AtomicExpr::AO__atomic_fetch_add:
       LibCallName = "__atomic_fetch_add";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         LoweredMemTy, E->getExprLoc(), sizeChars);
       break;
     // T __atomic_and_fetch_N(T *mem, T val, int order)
     // T __atomic_fetch_and_N(T *mem, T val, int order)
     case AtomicExpr::AO__atomic_and_fetch:
       PostOp = llvm::Instruction::And;
     // Fall through.
     case AtomicExpr::AO__c11_atomic_fetch_and:
+    case AtomicExpr::AO__opencl_atomic_fetch_and:
     case AtomicExpr::AO__atomic_fetch_and:
       LibCallName = "__atomic_fetch_and";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       break;
     // T __atomic_or_fetch_N(T *mem, T val, int order)
     // T __atomic_fetch_or_N(T *mem, T val, int order)
     case AtomicExpr::AO__atomic_or_fetch:
       PostOp = llvm::Instruction::Or;
     // Fall through.
     case AtomicExpr::AO__c11_atomic_fetch_or:
+    case AtomicExpr::AO__opencl_atomic_fetch_or:
     case AtomicExpr::AO__atomic_fetch_or:
       LibCallName = "__atomic_fetch_or";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       break;
     // T __atomic_sub_fetch_N(T *mem, T val, int order)
     // T __atomic_fetch_sub_N(T *mem, T val, int order)
     case AtomicExpr::AO__atomic_sub_fetch:
       PostOp = llvm::Instruction::Sub;
     // Fall through.
     case AtomicExpr::AO__c11_atomic_fetch_sub:
+    case AtomicExpr::AO__opencl_atomic_fetch_sub:
     case AtomicExpr::AO__atomic_fetch_sub:
       LibCallName = "__atomic_fetch_sub";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         LoweredMemTy, E->getExprLoc(), sizeChars);
       break;
     // T __atomic_xor_fetch_N(T *mem, T val, int order)
     // T __atomic_fetch_xor_N(T *mem, T val, int order)
     case AtomicExpr::AO__atomic_xor_fetch:
       PostOp = llvm::Instruction::Xor;
     // Fall through.
     case AtomicExpr::AO__c11_atomic_fetch_xor:
+    case AtomicExpr::AO__opencl_atomic_fetch_xor:
     case AtomicExpr::AO__atomic_fetch_xor:
       LibCallName = "__atomic_fetch_xor";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       break;
+    case AtomicExpr::AO__opencl_atomic_fetch_min:
+      LibCallName = E->getValueType()->isSignedIntegerType()
+                        ? "__atomic_fetch_min"
+                        : "__atomic_fetch_umin";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
+                        LoweredMemTy, E->getExprLoc(), sizeChars);
+      break;
+    case AtomicExpr::AO__opencl_atomic_fetch_max:
+      LibCallName = E->getValueType()->isSignedIntegerType()
+                        ? "__atomic_fetch_max"
+                        : "__atomic_fetch_umax";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
+                        LoweredMemTy, E->getExprLoc(), sizeChars);
+      break;
     // T __atomic_nand_fetch_N(T *mem, T val, int order)
     // T __atomic_fetch_nand_N(T *mem, T val, int order)
     case AtomicExpr::AO__atomic_nand_fetch:
       PostOp = llvm::Instruction::And; // the NOT is special cased below
     // Fall through.
     case AtomicExpr::AO__atomic_fetch_nand:
       LibCallName = "__atomic_fetch_nand";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1.getPointer(),
                         MemTy, E->getExprLoc(), sizeChars);
       break;
     }
 
+    if (E->isOpenCL()) {
+      LibCallName = std::string("__opencl") +
+          StringRef(LibCallName).drop_front(1).str();
+
+    }
     // Optimized functions have the size in their name.
     if (UseOptimizedLibcall)
       LibCallName += "_" + llvm::utostr(Size);
     // By default, assume we return a value of the atomic type.
     if (!HaveRetTy) {
       if (UseOptimizedLibcall) {
         // Value is returned directly.
         // The function returns an appropriately sized integer type.
         RetTy = getContext().getIntTypeForBitwidth(
             getContext().toBits(sizeChars), /*Signed=*/false);
       } else {
         // Value is returned through parameter before the order.
         RetTy = getContext().VoidTy;
         Args.add(RValue::get(EmitCastToVoidPtr(Dest.getPointer())),
                  getContext().VoidPtrTy);
       }
     }
     // order is always the last parameter
     Args.add(RValue::get(Order),
              getContext().IntTy);
+    if (E->isOpenCL())
+      Args.add(RValue::get(Scope), getContext().IntTy);
 
     // PostOp is only needed for the atomic_*_fetch operations, and
     // thus is only needed for and implemented in the
     // UseOptimizedLibcall codepath.
     assert(UseOptimizedLibcall || !PostOp);
 
     RValue Res = emitAtomicLibcall(*this, LibCallName, RetTy, Args);
     // The value is returned directly from the libcall.
if (RValTy->isVoidType())
       return RValue::get(nullptr);
 
     return convertTempToRValue(
         Builder.CreateBitCast(Dest, ConvertTypeForMem(RValTy)->getPointerTo()),
         RValTy, E->getExprLoc());
   }
 
   bool IsStore = E->getOp() == AtomicExpr::AO__c11_atomic_store ||
+                 E->getOp() == AtomicExpr::AO__opencl_atomic_store ||
                  E->getOp() == AtomicExpr::AO__atomic_store ||
                  E->getOp() == AtomicExpr::AO__atomic_store_n;
   bool IsLoad = E->getOp() == AtomicExpr::AO__c11_atomic_load ||
+                E->getOp() == AtomicExpr::AO__opencl_atomic_load ||
                 E->getOp() == AtomicExpr::AO__atomic_load ||
                 E->getOp() == AtomicExpr::AO__atomic_load_n;
 
+  assert(isa<llvm::ConstantInt>(Scope) &&
+      "Non-constant synchronization scope not supported");
+  auto SCID = getTargetHooks().getLLVMSyncScopeID(
+      static_cast<SyncScope>(cast<llvm::ConstantInt>(Scope)->getZExtValue()),
+      getLLVMContext());
+
   if (isa<llvm::ConstantInt>(Order)) {
     auto ord = cast<llvm::ConstantInt>(Order)->getZExtValue();
     // We should not ever get to a case where the ordering isn't a valid C ABI
     // value, but it's hard to enforce that in general.
     if (llvm::isValidAtomicOrderingCABI(ord))
       switch ((llvm::AtomicOrderingCABI)ord) {
       case llvm::AtomicOrderingCABI::relaxed:
         EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
-                     llvm::AtomicOrdering::Monotonic);
+                     llvm::AtomicOrdering::Monotonic, SCID);
         break;
       case llvm::AtomicOrderingCABI::consume:
       case llvm::AtomicOrderingCABI::acquire:
         if (IsStore)
           break; // Avoid crashing on code with undefined behavior
         EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
-                     llvm::AtomicOrdering::Acquire);
+                     llvm::AtomicOrdering::Acquire, SCID);
         break;
       case llvm::AtomicOrderingCABI::release:
         if (IsLoad)
           break; // Avoid crashing on code with undefined behavior
         EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
-                     llvm::AtomicOrdering::Release);
+                     llvm::AtomicOrdering::Release, SCID);
         break;
       case llvm::AtomicOrderingCABI::acq_rel:
         if (IsLoad || IsStore)
           break; // Avoid crashing on code with undefined behavior
         EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
-                     llvm::AtomicOrdering::AcquireRelease);
+                     llvm::AtomicOrdering::AcquireRelease, SCID);
         break;
       case llvm::AtomicOrderingCABI::seq_cst:
         EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
-                     llvm::AtomicOrdering::SequentiallyConsistent);
+                     llvm::AtomicOrdering::SequentiallyConsistent, SCID);
         break;
       }
     if (RValTy->isVoidType())
       return RValue::get(nullptr);
 
     return convertTempToRValue(
         Builder.CreateBitCast(Dest, ConvertTypeForMem(RValTy)->getPointerTo()),
         RValTy, E->getExprLoc());
RValue CodeGenFunction::EmitAtomicExpr(AtomicExpr *E) {
   // MonotonicBB is arbitrarily chosen as the default case; in practice, this
   // doesn't matter unless someone is crazy enough to use something that
   // doesn't fold to a constant for the ordering.
   Order = Builder.CreateIntCast(Order, Builder.getInt32Ty(), false);
   llvm::SwitchInst *SI = Builder.CreateSwitch(Order, MonotonicBB);
 
   // Emit all the different atomics
   Builder.SetInsertPoint(MonotonicBB);
-  EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail,
-               Size, llvm::AtomicOrdering::Monotonic);
+  EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
+               llvm::AtomicOrdering::Monotonic, SCID);
   Builder.CreateBr(ContBB);
   if (!IsStore) {
     Builder.SetInsertPoint(AcquireBB);
-    EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail,
-                 Size, llvm::AtomicOrdering::Acquire);
+    EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
+                 llvm::AtomicOrdering::Acquire, SCID);
     Builder.CreateBr(ContBB);
     SI->addCase(Builder.getInt32((int)llvm::AtomicOrderingCABI::consume),
                 AcquireBB);
     SI->addCase(Builder.getInt32((int)llvm::AtomicOrderingCABI::acquire),
                 AcquireBB);
   }
   if (!IsLoad) {
     Builder.SetInsertPoint(ReleaseBB);
-    EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail,
-                 Size, llvm::AtomicOrdering::Release);
+    EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
+                 llvm::AtomicOrdering::Release, SCID);
     Builder.CreateBr(ContBB);
     SI->addCase(Builder.getInt32((int)llvm::AtomicOrderingCABI::release),
                 ReleaseBB);
   }
   if (!IsLoad && !IsStore) {
     Builder.SetInsertPoint(AcqRelBB);
-    EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail,
-                 Size, llvm::AtomicOrdering::AcquireRelease);
+    EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
+                 llvm::AtomicOrdering::AcquireRelease, SCID);
     Builder.CreateBr(ContBB);
     SI->addCase(Builder.getInt32((int)llvm::AtomicOrderingCABI::acq_rel),
                 AcqRelBB);
   }
   Builder.SetInsertPoint(SeqCstBB);
-  EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail,
-               Size, llvm::AtomicOrdering::SequentiallyConsistent);
+  EmitAtomicOp(*this, E, Dest, Ptr, Val1, Val2, IsWeak, OrderFail, Size,
+               llvm::AtomicOrdering::SequentiallyConsistent, SCID);
   Builder.CreateBr(ContBB);
   SI->addCase(Builder.getInt32((int)llvm::AtomicOrderingCABI::seq_cst),
               SeqCstBB);
 
   // Cleanup and return
   Builder.SetInsertPoint(ContBB);
   if (RValTy->isVoidType())
     return RValue::get(nullptr);

lib/CodeGen/CGExpr.cpp

 //===--- CGExpr.cpp - Emit LLVM Code from Expressions ---------------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //

 using namespace CodeGen;
 
 //===--------------------------------------------------------------------===//
 //                        Miscellaneous Helper Methods
 //===--------------------------------------------------------------------===//
 
 llvm::Value *CodeGenFunction::EmitCastToVoidPtr(llvm::Value *value) {
   unsigned addressSpace =
       cast<llvm::PointerType>(value->getType())->getAddressSpace();
 
   llvm::PointerType *destType = Int8PtrTy;
   if (addressSpace)
     destType = llvm::Type::getInt8PtrTy(getLLVMContext(), addressSpace);
 
   if (value->getType() == destType) return value;
   return Builder.CreateBitCast(value, destType);
 }

lib/CodeGen/TargetInfo.h

 //===---- TargetInfo.h - Encapsulate target details -------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 // These classes wrap the information about a call or function
 // definition used to handle ABI compliancy.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_CLANG_LIB_CODEGEN_TARGETINFO_H
 #define LLVM_CLANG_LIB_CODEGEN_TARGETINFO_H
 
 #include "CodeGenModule.h"
 #include "CGValue.h"
 #include "clang/AST/Type.h"
 #include "clang/Basic/LLVM.h"
+#include "clang/Basic/SyncScope.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringRef.h"
 
 namespace llvm {
 class Constant;
 class GlobalValue;
 class Type;
 class Value;
public:
   /// Perform address space cast of a constant expression of pointer type.
   /// \param V is the LLVM constant to be casted to another address space.
   /// \param SrcAddr is the language address space of \p V.
   /// \param DestAddr is the targeted language address space.
   /// \param DestTy is the destination LLVM pointer type.
   virtual llvm::Constant *
   performAddrSpaceCast(CodeGenModule &CGM, llvm::Constant *V, unsigned SrcAddr,
                        unsigned DestAddr, llvm::Type *DestTy) const;
+
+  /// Get the syncscope used in LLVM IR.
+  virtual llvm::SyncScope::ID getLLVMSyncScopeID(SyncScope S,
+                                                 llvm::LLVMContext &C) const;
 };
 
 } // namespace CodeGen
 } // namespace clang
 
 #endif // LLVM_CLANG_LIB_CODEGEN_TARGETINFO_H

lib/CodeGen/TargetInfo.cpp

 TargetCodeGenInfo::performAddrSpaceCast(CodeGenModule &CGM, llvm::Constant *Src,
                                         unsigned SrcAddr, unsigned DestAddr,
                                         llvm::Type *DestTy) const {
   // Since target may map different address spaces in AST to the same address
   // space, an address space conversion may end up as a bitcast.
   return llvm::ConstantExpr::getPointerCast(Src, DestTy);
 }
 
+llvm::SyncScope::ID
+TargetCodeGenInfo::getLLVMSyncScopeID(SyncScope S, llvm::LLVMContext &C) const {
+  return C.getOrInsertSyncScopeID(""); /* default sync scope */
+}
+
 static bool isEmptyRecord(ASTContext &Context, QualType T, bool AllowArrays);
 
 /// isEmptyField - Return true iff a the field is "empty", that is it
 /// is an unnamed bit-field or an (array of) empty record(s).
 static bool isEmptyField(ASTContext &Context, const FieldDecl *FD,
                          bool AllowArrays) {
   if (FD->isUnnamedBitfield())
     return true;

       llvm::PointerType *T, QualType QT) const override;
 
   unsigned getASTAllocaAddressSpace() const override {
     return LangAS::FirstTargetAddressSpace +
            getABIInfo().getDataLayout().getAllocaAddrSpace();
   }
   unsigned getGlobalVarAddressSpace(CodeGenModule &CGM,
                                     const VarDecl *D) const override;
+  llvm::SyncScope::ID getLLVMSyncScopeID(SyncScope S,
+                                         llvm::LLVMContext &C) const override;
 };
 }
 
 void AMDGPUTargetCodeGenInfo::setTargetAttributes(
     const Decl *D, llvm::GlobalValue *GV, CodeGen::CodeGenModule &M,
     ForDefinition_t IsForDefinition) const {
   if (!IsForDefinition)
     return;

 
   if (CGM.isTypeConstant(D->getType(), false)) {
     if (auto ConstAS = CGM.getTarget().getConstantAddressSpace())
       return ConstAS.getValue();
   }
   return DefaultGlobalAS;
 }
 
+llvm::SyncScope::ID
+AMDGPUTargetCodeGenInfo::getLLVMSyncScopeID(SyncScope S,
+                                            llvm::LLVMContext &C) const {
+  StringRef Name;
+  switch (S) {
+  case SyncScope::OpenCLWorkGroup:
+    Name = "workgroup";
+    break;
+  case SyncScope::OpenCLDevice:
+    Name = "agent";
+    break;
+  case SyncScope::OpenCLAllSVMDevices:
+    Name = "";
+    break;
+  case SyncScope::OpenCLSubGroup:
+    Name = "subgroup";
+  }
+  return C.getOrInsertSyncScopeID(Name);
+}
+
 //===----------------------------------------------------------------------===//
 // SPARC v8 ABI Implementation.
 // Based on the SPARC Compliance Definition version 2.4.1.
 //
 // Ensures that complex values are passed in registers.
 //
 namespace {
 class SparcV8ABIInfo : public DefaultABIInfo {

lib/Frontend/InitPreprocessor.cpp

#undef TOSTR2
   // Define macros for the C11 / C++11 memory orderings
   Builder.defineMacro("__ATOMIC_RELAXED", "0");
   Builder.defineMacro("__ATOMIC_CONSUME", "1");
   Builder.defineMacro("__ATOMIC_ACQUIRE", "2");
   Builder.defineMacro("__ATOMIC_RELEASE", "3");
   Builder.defineMacro("__ATOMIC_ACQ_REL", "4");
   Builder.defineMacro("__ATOMIC_SEQ_CST", "5");
 
+  // Define macros for the OpenCL memory scope
+  Builder.defineMacro("__OPENCL_MEMORY_SCOPE_WORK_ITEM", "0");
+  Builder.defineMacro("__OPENCL_MEMORY_SCOPE_WORK_GROUP", "1");
+  Builder.defineMacro("__OPENCL_MEMORY_SCOPE_DEVICE", "2");
+  Builder.defineMacro("__OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES", "3");
+  Builder.defineMacro("__OPENCL_MEMORY_SCOPE_SUB_GROUP", "4");
+
   // Support for #pragma redefine_extname (Sun compatibility)
   Builder.defineMacro("__PRAGMA_REDEFINE_EXTNAME", "1");
 
   // As sad as it is, enough software depends on the __VERSION__ for version
   // checks that it is necessary to report 4.2.1 (the base GCC version we claim
   // compatibility with) first.
   Builder.defineMacro("__VERSION__", "\"4.2.1 Compatible " + 
                       Twine(getClangFullCPPVersion()) + "\"");

lib/Headers/opencl-c.h

  * of memory operations to global memory. This can be
  * useful when work-items, for example, write to buffer or
  * image objects and then want to read the updated data.
  */
 
 void __ovld __conv barrier(cl_mem_fence_flags flags);
 
 #if __OPENCL_C_VERSION__ >= CL_VERSION_2_0
 
-typedef enum memory_scope
-{
-  memory_scope_work_item,
-  memory_scope_work_group,
-  memory_scope_device,
-  memory_scope_all_svm_devices,
-  memory_scope_sub_group
+typedef enum memory_scope {
+  memory_scope_work_item = __OPENCL_MEMORY_SCOPE_WORK_ITEM,
+  memory_scope_work_group = __OPENCL_MEMORY_SCOPE_WORK_GROUP,
+  memory_scope_device = __OPENCL_MEMORY_SCOPE_DEVICE,
+  memory_scope_all_svm_devices = __OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES,
+#if defined(cl_intel_subgroups) || defined(cl_khr_subgroups)
+  memory_scope_sub_group = __OPENCL_MEMORY_SCOPE_SUB_GROUP
+#endif
 } memory_scope;
 
 void __ovld __conv work_group_barrier(cl_mem_fence_flags flags, memory_scope scope);
 void __ovld __conv work_group_barrier(cl_mem_fence_flags flags);
 #endif //__OPENCL_C_VERSION__ >= CL_VERSION_2_0
 
 // OpenCL v1.1 s6.11.9, v1.2 s6.12.9 - Explicit Memory Fence Functions
 

 #ifndef ATOMIC_VAR_INIT
 #define ATOMIC_VAR_INIT(x) (x)
 #endif //ATOMIC_VAR_INIT
 #define ATOMIC_FLAG_INIT 0
 
 // enum values aligned with what clang uses in EmitAtomicExpr()
 typedef enum memory_order
 {
-  memory_order_relaxed,
-  memory_order_acquire,
-  memory_order_release,
-  memory_order_acq_rel,
-  memory_order_seq_cst
+  memory_order_relaxed = __ATOMIC_RELAXED,
+  memory_order_acquire = __ATOMIC_ACQUIRE,
+  memory_order_release = __ATOMIC_RELEASE,
+  memory_order_acq_rel = __ATOMIC_ACQ_REL,
+  memory_order_seq_cst = __ATOMIC_SEQ_CST
 } memory_order;
 
 // double atomics support requires extensions cl_khr_int64_base_atomics and cl_khr_int64_extended_atomics
 #if defined(cl_khr_int64_base_atomics) && defined(cl_khr_int64_extended_atomics)
 #pragma OPENCL EXTENSION cl_khr_int64_base_atomics : enable
 #pragma OPENCL EXTENSION cl_khr_int64_extended_atomics : enable
 #endif
 

lib/Sema/SemaChecking.cpp

 #include "clang/AST/Expr.h"
 #include "clang/AST/ExprCXX.h"
 #include "clang/AST/ExprObjC.h"
 #include "clang/AST/ExprOpenMP.h"
 #include "clang/AST/StmtCXX.h"
 #include "clang/AST/StmtObjC.h"
 #include "clang/Analysis/Analyses/FormatString.h"
 #include "clang/Basic/CharInfo.h"
+#include "clang/Basic/SyncScope.h"
 #include "clang/Basic/TargetBuiltins.h"
 #include "clang/Basic/TargetInfo.h"
 #include "clang/Lex/Lexer.h" // TODO: Extract static functions to fix layering.
 #include "clang/Sema/Initialization.h"
 #include "clang/Sema/Lookup.h"
 #include "clang/Sema/ScopeInfo.h"
 #include "clang/Sema/Sema.h"
 #include "clang/Sema/SemaInternal.h"

 
 static bool isValidOrderingForOp(int64_t Ordering, AtomicExpr::AtomicOp Op) {
   if (!llvm::isValidAtomicOrderingCABI(Ordering))
     return false;
 
   auto OrderingCABI = (llvm::AtomicOrderingCABI)Ordering;
   switch (Op) {
   case AtomicExpr::AO__c11_atomic_init:
+  case AtomicExpr::AO__opencl_atomic_init:
     llvm_unreachable("There is no ordering argument for an init");
 
   case AtomicExpr::AO__c11_atomic_load:
+  case AtomicExpr::AO__opencl_atomic_load:
   case AtomicExpr::AO__atomic_load_n:
   case AtomicExpr::AO__atomic_load:
     return OrderingCABI != llvm::AtomicOrderingCABI::release &&
            OrderingCABI != llvm::AtomicOrderingCABI::acq_rel;
 
   case AtomicExpr::AO__c11_atomic_store:
+  case AtomicExpr::AO__opencl_atomic_store:
   case AtomicExpr::AO__atomic_store:
   case AtomicExpr::AO__atomic_store_n:
     return OrderingCABI != llvm::AtomicOrderingCABI::consume &&
            OrderingCABI != llvm::AtomicOrderingCABI::acquire &&
            OrderingCABI != llvm::AtomicOrderingCABI::acq_rel;
 
   default:
     return true;
   }
 }
 
 ExprResult Sema::SemaAtomicOpsOverloaded(ExprResult TheCallResult,
                                          AtomicExpr::AtomicOp Op) {
   CallExpr *TheCall = cast<CallExpr>(TheCallResult.get());
   DeclRefExpr *DRE =cast<DeclRefExpr>(TheCall->getCallee()->IgnoreParenCasts());
 
-  // All these operations take one of the following forms:
+  // All the non-OpenCL operations take one of the following forms.
+  // The OpenCL operations take the __c11 forms with one extra argument for
+  // synchronization scope.
   enum {
     // C    __c11_atomic_init(A *, C)
     Init,
     // C    __c11_atomic_load(A *, int)
     Load,
     // void __atomic_load(A *, CP, int)
     LoadCopy,
     // void __atomic_store(A *, CP, int)
     Copy,
     // C    __c11_atomic_add(A *, M, int)
     Arithmetic,
     // C    __atomic_exchange_n(A *, CP, int)
     Xchg,
     // void __atomic_exchange(A *, C *, CP, int)
     GNUXchg,
     // bool __c11_atomic_compare_exchange_strong(A *, C *, CP, int, int)
     C11CmpXchg,
     // bool __atomic_compare_exchange(A *, C *, CP, bool, int, int)
     GNUCmpXchg
   } Form = Init;
+  const unsigned NumForm = GNUCmpXchg + 1;
   const unsigned NumArgs[] = { 2, 2, 3, 3, 3, 3, 4, 5, 6 };
   const unsigned NumVals[] = { 1, 0, 1, 1, 1, 1, 2, 2, 3 };
   // where:
   //   C is an appropriate type,
   //   A is volatile _Atomic(C) for __c11 builtins and is C for GNU builtins,
   //   CP is C for __c11 builtins and GNU _n builtins and is C * otherwise,
   //   M is C if C is an integer, and ptrdiff_t if C is a pointer, and
   //   the int parameters are for orderings.
 
+  static_assert(sizeof(NumArgs)/sizeof(NumArgs[0]) == NumForm
+      && sizeof(NumVals)/sizeof(NumVals[0]) == NumForm,
+      "need to update code for modified forms");
   static_assert(AtomicExpr::AO__c11_atomic_init == 0 &&
                     AtomicExpr::AO__c11_atomic_fetch_xor + 1 ==
                         AtomicExpr::AO__atomic_load,
                 "need to update code for modified C11 atomics");
-  bool IsC11 = Op >= AtomicExpr::AO__c11_atomic_init &&
-               Op <= AtomicExpr::AO__c11_atomic_fetch_xor;
+  bool IsOpenCL = Op >= AtomicExpr::AO__opencl_atomic_init &&
+                  Op <= AtomicExpr::AO__opencl_atomic_fetch_max;
+  bool IsC11 = (Op >= AtomicExpr::AO__c11_atomic_init &&
+               Op <= AtomicExpr::AO__c11_atomic_fetch_xor) ||
+               IsOpenCL;
   bool IsN = Op == AtomicExpr::AO__atomic_load_n ||
              Op == AtomicExpr::AO__atomic_store_n ||
              Op == AtomicExpr::AO__atomic_exchange_n ||
              Op == AtomicExpr::AO__atomic_compare_exchange_n;
   bool IsAddSub = false;
 
   switch (Op) {
   case AtomicExpr::AO__c11_atomic_init:
+  case AtomicExpr::AO__opencl_atomic_init:
     Form = Init;
     break;
 
   case AtomicExpr::AO__c11_atomic_load:
+  case AtomicExpr::AO__opencl_atomic_load:
   case AtomicExpr::AO__atomic_load_n:
     Form = Load;
     break;
 
   case AtomicExpr::AO__atomic_load:
     Form = LoadCopy;
     break;
 
   case AtomicExpr::AO__c11_atomic_store:
+  case AtomicExpr::AO__opencl_atomic_store:
   case AtomicExpr::AO__atomic_store:
   case AtomicExpr::AO__atomic_store_n:
     Form = Copy;
     break;
 
   case AtomicExpr::AO__c11_atomic_fetch_add:
   case AtomicExpr::AO__c11_atomic_fetch_sub:
+  case AtomicExpr::AO__opencl_atomic_fetch_add:
+  case AtomicExpr::AO__opencl_atomic_fetch_sub:
+  case AtomicExpr::AO__opencl_atomic_fetch_min:
+  case AtomicExpr::AO__opencl_atomic_fetch_max:
   case AtomicExpr::AO__atomic_fetch_add:
   case AtomicExpr::AO__atomic_fetch_sub:
   case AtomicExpr::AO__atomic_add_fetch:
   case AtomicExpr::AO__atomic_sub_fetch:
     IsAddSub = true;
     // Fall through.
   case AtomicExpr::AO__c11_atomic_fetch_and:
   case AtomicExpr::AO__c11_atomic_fetch_or:
   case AtomicExpr::AO__c11_atomic_fetch_xor:
+  case AtomicExpr::AO__opencl_atomic_fetch_and:
+  case AtomicExpr::AO__opencl_atomic_fetch_or:
+  case AtomicExpr::AO__opencl_atomic_fetch_xor:
   case AtomicExpr::AO__atomic_fetch_and:
   case AtomicExpr::AO__atomic_fetch_or:
   case AtomicExpr::AO__atomic_fetch_xor:
   case AtomicExpr::AO__atomic_fetch_nand:
   case AtomicExpr::AO__atomic_and_fetch:
   case AtomicExpr::AO__atomic_or_fetch:
   case AtomicExpr::AO__atomic_xor_fetch:
   case AtomicExpr::AO__atomic_nand_fetch:
     Form = Arithmetic;
     break;
 
   case AtomicExpr::AO__c11_atomic_exchange:
+  case AtomicExpr::AO__opencl_atomic_exchange:
   case AtomicExpr::AO__atomic_exchange_n:
     Form = Xchg;
     break;
 
   case AtomicExpr::AO__atomic_exchange:
     Form = GNUXchg;
     break;
 
   case AtomicExpr::AO__c11_atomic_compare_exchange_strong:
   case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
+  case AtomicExpr::AO__opencl_atomic_compare_exchange_strong:
+  case AtomicExpr::AO__opencl_atomic_compare_exchange_weak:
     Form = C11CmpXchg;
     break;
 
   case AtomicExpr::AO__atomic_compare_exchange:
   case AtomicExpr::AO__atomic_compare_exchange_n:
     Form = GNUCmpXchg;
     break;
   }
 
+  unsigned AdjustedNumArgs = NumArgs[Form];
+  if (IsOpenCL && Op != AtomicExpr::AO__opencl_atomic_init)
+    ++AdjustedNumArgs;
   // Check we have the right number of arguments.
-  if (TheCall->getNumArgs() < NumArgs[Form]) {
+  if (TheCall->getNumArgs() < AdjustedNumArgs) {
     Diag(TheCall->getLocEnd(), diag::err_typecheck_call_too_few_args)
-      << 0 << NumArgs[Form] << TheCall->getNumArgs()
+      << 0 << AdjustedNumArgs << TheCall->getNumArgs()
       << TheCall->getCallee()->getSourceRange();
     return ExprError();
-  } else if (TheCall->getNumArgs() > NumArgs[Form]) {
-    Diag(TheCall->getArg(NumArgs[Form])->getLocStart(),
+  } else if (TheCall->getNumArgs() > AdjustedNumArgs) {
+    Diag(TheCall->getArg(AdjustedNumArgs)->getLocStart(),
          diag::err_typecheck_call_too_many_args)
-      << 0 << NumArgs[Form] << TheCall->getNumArgs()
+      << 0 << AdjustedNumArgs << TheCall->getNumArgs()
       << TheCall->getCallee()->getSourceRange();
     return ExprError();
   }
 
   // Inspect the first argument of the atomic operation.
   Expr *Ptr = TheCall->getArg(0);
   ExprResult ConvertedPtr = DefaultFunctionArrayLvalueConversion(Ptr);
   if (ConvertedPtr.isInvalid())

   QualType AtomTy = pointerType->getPointeeType(); // 'A'
   QualType ValType = AtomTy; // 'C'
   if (IsC11) {
     if (!AtomTy->isAtomicType()) {
       Diag(DRE->getLocStart(), diag::err_atomic_op_needs_atomic)
         << Ptr->getType() << Ptr->getSourceRange();
       return ExprError();
     }
-    if (AtomTy.isConstQualified()) {
+    if (AtomTy.isConstQualified() ||
+        AtomTy.getAddressSpace() == LangAS::opencl_constant) {
       Diag(DRE->getLocStart(), diag::err_atomic_op_needs_non_const_atomic)
-        << Ptr->getType() << Ptr->getSourceRange();
+          << (AtomTy.isConstQualified() ? 0 : 1) << Ptr->getType()
+          << Ptr->getSourceRange();
       return ExprError();
     }
     ValType = AtomTy->getAs<AtomicType>()->getValueType();
   } else if (Form != Load && Form != LoadCopy) {
     if (ValType.isConstQualified()) {
       Diag(DRE->getLocStart(), diag::err_atomic_op_needs_non_const_pointer)
         << Ptr->getType() << Ptr->getSourceRange();
       return ExprError();

   }
 
   // atomic_fetch_or takes a pointer to a volatile 'A'.  We shouldn't let the
   // volatile-ness of the pointee-type inject itself into the result or the
   // other operands. Similarly atomic_load can take a pointer to a const 'A'.
   ValType.removeLocalVolatile();
   ValType.removeLocalConst();
   QualType ResultType = ValType;
-  if (Form == Copy || Form == LoadCopy || Form == GNUXchg || Form == Init)
+  if (Form == Copy || Form == LoadCopy || Form == GNUXchg ||
+      Form == Init)
     ResultType = Context.VoidTy;
   else if (Form == C11CmpXchg || Form == GNUCmpXchg)
     ResultType = Context.BoolTy;
 
   // The type of a parameter passed 'by value'. In the GNU atomics, such
   // arguments are actually passed as pointers.
   QualType ByValType = ValType; // 'CP'
   if (!IsC11 && !IsN)
     ByValType = Ptr->getType();
 
   // The first argument --- the pointer --- has a fixed type; we
   // deduce the types of the rest of the arguments accordingly.  Walk
   // the remaining arguments, converting them to the deduced value type.
-  for (unsigned i = 1; i != NumArgs[Form]; ++i) {
+  for (unsigned i = 1; i != TheCall->getNumArgs(); ++i) {
     QualType Ty;
     if (i < NumVals[Form] + 1) {
       switch (i) {
       case 1:
         // The second argument is the non-atomic operand. For arithmetic, this
         // is always passed by value, and for a compare_exchange it is always
         // passed by address. For the rest, GNU uses by-address and C11 uses
         // by-value.
         assert(Form != Load);
         if (Form == Init || (Form == Arithmetic && ValType->isIntegerType()))
           Ty = ValType;
         else if (Form == Copy || Form == Xchg)
           Ty = ByValType;
         else if (Form == Arithmetic)
           Ty = Context.getPointerDiffType();
         else {
           Expr *ValArg = TheCall->getArg(i);
           // Treat this argument as _Nonnull as we want to show a warning if
           // NULL is passed into it.
           CheckNonNullArgument(*this, ValArg, DRE->getLocStart());
           unsigned AS = 0;
           // Keep address space of non-atomic pointer type.
           if (const PointerType *PtrTy =
                   ValArg->getType()->getAs<PointerType>()) {
             AS = PtrTy->getPointeeType().getAddressSpace();
           }
           Ty = Context.getPointerType(
               Context.getAddrSpaceQualType(ValType.getUnqualifiedType(), AS));
         }
         break;
       case 2:
         // The third argument to compare_exchange / GNU exchange is a
         // (pointer to a) desired value.
         Ty = ByValType;
         break;
       case 3:
         // The fourth argument to GNU compare_exchange is a 'weak' flag.
         Ty = Context.BoolTy;
         break;
       }
     } else {
-      // The order(s) are always converted to int.
+      // The order(s) and scope are always converted to int.
       Ty = Context.IntTy;
     }
 
     InitializedEntity Entity =
         InitializedEntity::InitializeParameter(Context, Ty, false);
     ExprResult Arg = TheCall->getArg(i);
     Arg = PerformCopyInitialization(Entity, SourceLocation(), Arg);
     if (Arg.isInvalid())
       return true;
     TheCall->setArg(i, Arg.get());
   }
 
+  Expr *Scope;
+  if (Form != Init) {
+    if (IsOpenCL) {
+      Scope = TheCall->getArg(TheCall->getNumArgs() - 1);
+      llvm::APSInt Result(32);
+      if (!Scope->isIntegerConstantExpr(Result, Context))
+        Diag(Scope->getLocStart(),
+             diag::err_atomic_op_has_non_constant_synch_scope)
+            << Scope->getSourceRange();
+      else if (!isValidSyncScopeValue(Result.getZExtValue()))
+        Diag(Scope->getLocStart(), diag::err_atomic_op_has_invalid_synch_scope)
+            << Scope->getSourceRange();
+    } else {
+      Scope = IntegerLiteral::Create(
+          Context,
+          llvm::APInt(Context.getTypeSize(Context.IntTy),
+                      static_cast<unsigned>(SyncScope::OpenCLAllSVMDevices)),
+          Context.IntTy, SourceLocation());
+    }
+  }
+
   // Permute the arguments into a 'consistent' order.
   SmallVector<Expr*, 5> SubExprs;
   SubExprs.push_back(Ptr);
   switch (Form) {
   case Init:
     // Note, AtomicExpr::getVal1() has a special case for this atomic.
     SubExprs.push_back(TheCall->getArg(1)); // Val1
     break;
   case Load:
     SubExprs.push_back(TheCall->getArg(1)); // Order
+    SubExprs.push_back(Scope);              // Scope
     break;
   case LoadCopy:
   case Copy:
   case Arithmetic:
   case Xchg:
     SubExprs.push_back(TheCall->getArg(2)); // Order
+    SubExprs.push_back(Scope);              // Scope
     SubExprs.push_back(TheCall->getArg(1)); // Val1
     break;
   case GNUXchg:
     // Note, AtomicExpr::getVal2() has a special case for this atomic.
     SubExprs.push_back(TheCall->getArg(3)); // Order
+    SubExprs.push_back(Scope);              // Scope
     SubExprs.push_back(TheCall->getArg(1)); // Val1
     SubExprs.push_back(TheCall->getArg(2)); // Val2
     break;
   case C11CmpXchg:
     SubExprs.push_back(TheCall->getArg(3)); // Order
+    SubExprs.push_back(Scope);              // Scope
     SubExprs.push_back(TheCall->getArg(1)); // Val1
     SubExprs.push_back(TheCall->getArg(4)); // OrderFail
     SubExprs.push_back(TheCall->getArg(2)); // Val2
     break;
   case GNUCmpXchg:
     SubExprs.push_back(TheCall->getArg(4)); // Order
+    SubExprs.push_back(Scope);              // Scope
     SubExprs.push_back(TheCall->getArg(1)); // Val1
     SubExprs.push_back(TheCall->getArg(5)); // OrderFail
     SubExprs.push_back(TheCall->getArg(2)); // Val2
     SubExprs.push_back(TheCall->getArg(3)); // Weak
     break;
   }
 
   if (SubExprs.size() >= 2 && Form != Init) {
     llvm::APSInt Result(32);
     if (SubExprs[1]->isIntegerConstantExpr(Result, Context) &&
         !isValidOrderingForOp(Result.getSExtValue(), Op))
       Diag(SubExprs[1]->getLocStart(),
            diag::warn_atomic_op_has_invalid_memory_order)
           << SubExprs[1]->getSourceRange();
   }
 
   AtomicExpr *AE = new (Context) AtomicExpr(TheCall->getCallee()->getLocStart(),
                                             SubExprs, ResultType, Op,
                                             TheCall->getRParenLoc());
   
   if ((Op == AtomicExpr::AO__c11_atomic_load ||
-       (Op == AtomicExpr::AO__c11_atomic_store)) &&
+       Op == AtomicExpr::AO__c11_atomic_store ||
+       Op == AtomicExpr::AO__opencl_atomic_load ||
+       Op == AtomicExpr::AO__opencl_atomic_store ) &&
       Context.AtomicUsesUnsupportedLibcall(AE))
-    Diag(AE->getLocStart(), diag::err_atomic_load_store_uses_lib) <<
-    ((Op == AtomicExpr::AO__c11_atomic_load) ? 0 : 1);
+    Diag(AE->getLocStart(), diag::err_atomic_load_store_uses_lib)
+        << ((Op == AtomicExpr::AO__c11_atomic_load ||
+            Op == AtomicExpr::AO__opencl_atomic_load)
+                ? 0 : 1);
 
   return AE;
 }
 
 /// checkBuiltinArgument - Given a call to a builtin function, perform
 /// normal type-checking on the given argument, updating the call in
 /// place.  This is useful when a builtin function requires custom
 /// type-checking for some of its arguments but not necessarily all of

test/CodeGenOpenCL/atomic-ops-libcall.cl

+// RUN: %clang_cc1 < %s -cl-std=CL2.0 -finclude-default-header -triple spir64 -emit-llvm | FileCheck -check-prefix=SPIR %s
+// RUN: %clang_cc1 < %s -cl-std=CL2.0 -finclude-default-header -triple armv5e-none-linux-gnueabi -emit-llvm | FileCheck -check-prefix=ARM %s
+
+void f(atomic_int *i, atomic_uint *ui, int cmp) {
+  int x;
+  // SPIR: {{%[^ ]*}} = call i32 @__opencl_atomic_load_4(i8 addrspace(4)* {{%[0-9]+}}, i32 5, i32 1)
+  // ARM: {{%[^ ]*}} = call i32 @__opencl_atomic_load_4(i8* {{%[0-9]+}}, i32 5, i32 1)
+  x = __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: call void @__opencl_atomic_store_4(i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  // ARM: call void @__opencl_atomic_store_4(i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  __opencl_atomic_store(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: {{%[^ ]*}} = call i32 @__opencl_atomic_fetch_add_4(i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  // ARM: {{%[^ ]*}} = call i32 @__opencl_atomic_fetch_add_4(i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  x = __opencl_atomic_fetch_add(i, 3, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: {{%[^ ]*}} = call i32 @__opencl_atomic_fetch_min_4(i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  // ARM: {{%[^ ]*}} = call i32 @__opencl_atomic_fetch_min_4(i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  x = __opencl_atomic_fetch_min(i, 3, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: {{%[^ ]*}} = call i32 @__opencl_atomic_fetch_umin_4(i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  // ARM: {{%[^ ]*}} = call i32 @__opencl_atomic_fetch_umin_4(i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 1)
+  x = __opencl_atomic_fetch_min(ui, 3, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8 addrspace(4)* {{%[0-9]+}}, i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 1)
+  // ARM: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8* {{%[0-9]+}}, i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 1)
+  x = __opencl_atomic_compare_exchange_strong(i, &cmp, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8 addrspace(4)* {{%[0-9]+}}, i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 1)
+  // ARM: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8* {{%[0-9]+}}, i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 1)
+  x = __opencl_atomic_compare_exchange_weak(i, &cmp, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+  // SPIR: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8 addrspace(4)* {{%[0-9]+}}, i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 2)
+  // ARM: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8* {{%[0-9]+}}, i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 2)
+  x = __opencl_atomic_compare_exchange_weak(i, &cmp, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_device);
+  // SPIR: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8 addrspace(4)* {{%[0-9]+}}, i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 3)
+  // ARM: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8* {{%[0-9]+}}, i8* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 3)
+  x = __opencl_atomic_compare_exchange_weak(i, &cmp, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_all_svm_devices);
+#ifdef cl_khr_subgroups
+  // SPIR: {{%[^ ]*}} = call zeroext i1 @__opencl_atomic_compare_exchange_4(i8 addrspace(4)* {{%[0-9]+}}, i8 addrspace(4)* {{%[0-9]+}}, i32 {{%[0-9]+}}, i32 5, i32 5, i32 4)
+  x = __opencl_atomic_compare_exchange_weak(i, &cmp, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_sub_group);
+#endif
+}

test/CodeGenOpenCL/atomic-ops.cl

+// RUN: %clang_cc1 %s -cl-std=CL2.0 -emit-llvm -finclude-default-header -O0 -o - -triple=amdgcn-amd-amdhsa-opencl | FileCheck %s
+
+// Also test serialization of atomic operations here, to avoid duplicating the test.
+// RUN: %clang_cc1 %s -cl-std=CL2.0 -finclude-default-header -emit-pch -O0 -o %t -triple=amdgcn-amd-amdhsa-opencl
+// RUN: %clang_cc1 %s -cl-std=CL2.0 -finclude-default-header -include-pch %t -O0 -triple=amdgcn-amd-amdhsa-opencl -emit-llvm -o - | FileCheck %s
+
+#ifndef ALREADY_INCLUDED
+#define ALREADY_INCLUDED
+
+atomic_int j;
+
+void fi1(atomic_int *i) {
+  // CHECK-LABEL: @fi1
+  // CHECK: load atomic i32, i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  int x = __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_work_group);
+  // CHECK: load atomic i32, i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}} syncscope("agent") seq_cst
+  x = __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_device);
+  // CHECK: load atomic i32, i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}} seq_cst
+  x = __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_all_svm_devices);
+  // CHECK: load atomic i32, i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}} syncscope("subgroup") seq_cst
+  x = __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_sub_group);
+}
+
+void fi2(atomic_int *i) {
+  // CHECK-LABEL: @fi2
+  // CHECK: store atomic i32 %{{[.0-9A-Z_a-z]+}}, i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  __opencl_atomic_store(i, 1, memory_order_seq_cst, memory_scope_work_group);
+}
+
+void fi3(atomic_int *i, atomic_uint *ui) {
+  // CHECK-LABEL: @fi3
+  // CHECK: atomicrmw and i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}}, i32 %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  int x = __opencl_atomic_fetch_and(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  // CHECK: atomicrmw min i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}}, i32 %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  x = __opencl_atomic_fetch_min(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  // CHECK: atomicrmw max i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}}, i32 %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  x = __opencl_atomic_fetch_max(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  // CHECK: atomicrmw umin i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}}, i32 %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  x = __opencl_atomic_fetch_min(ui, 1, memory_order_seq_cst, memory_scope_work_group);
+  // CHECK: atomicrmw umax i32 addrspace(4)* %{{[.0-9A-Z_a-z]+}}, i32 %{{[.0-9A-Z_a-z]+}} syncscope("workgroup") seq_cst
+  x = __opencl_atomic_fetch_max(ui, 1, memory_order_seq_cst, memory_scope_work_group);
+}
+
+bool fi4(atomic_int *i) {
+  // CHECK-LABEL: @fi4(
+  // CHECK: [[PAIR:%[.0-9A-Z_a-z]+]] = cmpxchg i32 addrspace(4)* [[PTR:%[.0-9A-Z_a-z]+]], i32 [[EXPECTED:%[.0-9A-Z_a-z]+]], i32 [[DESIRED:%[.0-9A-Z_a-z]+]] syncscope("workgroup") acquire acquire
+  // CHECK: [[OLD:%[.0-9A-Z_a-z]+]] = extractvalue { i32, i1 } [[PAIR]], 0
+  // CHECK: [[CMP:%[.0-9A-Z_a-z]+]] = extractvalue { i32, i1 } [[PAIR]], 1
+  // CHECK: br i1 [[CMP]], label %[[STORE_EXPECTED:[.0-9A-Z_a-z]+]], label %[[CONTINUE:[.0-9A-Z_a-z]+]]
+  // CHECK: store i32 [[OLD]]
+  int cmp = 0;
+  return __opencl_atomic_compare_exchange_strong(i, &cmp, 1, memory_order_acquire, memory_order_acquire, memory_scope_work_group);
+}
+
+float ff1(global atomic_float *d) {
+  // CHECK-LABEL: @ff1
+  // CHECK: load atomic i32, i32 addrspace(1)* {{.*}} syncscope("workgroup") monotonic
+  return __opencl_atomic_load(d, memory_order_relaxed, memory_scope_work_group);
+}
+
+void ff2(atomic_float *d) {
+  // CHECK-LABEL: @ff2
+  // CHECK: store atomic i32 {{.*}} syncscope("workgroup") release
+  __opencl_atomic_store(d, 1, memory_order_release, memory_scope_work_group);
+}
+
+float ff3(atomic_float *d) {
+  // CHECK-LABEL: @ff3
+  // CHECK: atomicrmw xchg i32 addrspace(4)* {{.*}} syncscope("workgroup") seq_cst
+  return __opencl_atomic_exchange(d, 2, memory_order_seq_cst, memory_scope_work_group);
+}
+
+// CHECK-LABEL: @atomic_init_foo
+void atomic_init_foo()
+{
+  // CHECK-NOT: atomic
+  // CHECK: store
+  __opencl_atomic_init(&j, 42);
+
+  // CHECK-NOT: atomic
+  // CHECK: }
+}
+
+// CHECK-LABEL: @failureOrder
+void failureOrder(atomic_int *ptr, int *ptr2) {
+  // CHECK: cmpxchg i32 addrspace(4)* {{%[0-9A-Za-z._]+}}, i32 {{%[0-9A-Za-z._]+}}, i32 {{%[0-9A-Za-z_.]+}} syncscope("workgroup") acquire monotonic
+  __opencl_atomic_compare_exchange_strong(ptr, ptr2, 43, memory_order_acquire, memory_order_relaxed, memory_scope_work_group);
+
+  // CHECK: cmpxchg weak i32 addrspace(4)* {{%[0-9A-Za-z._]+}}, i32 {{%[0-9A-Za-z._]+}}, i32 {{%[0-9A-Za-z_.]+}} syncscope("workgroup") seq_cst acquire
+  __opencl_atomic_compare_exchange_weak(ptr, ptr2, 43, memory_order_seq_cst, memory_order_acquire, memory_scope_work_group);
+}
+
+// CHECK-LABEL: @generalFailureOrder
+void generalFailureOrder(atomic_int *ptr, int *ptr2, int success, int fail) {
+  __opencl_atomic_compare_exchange_strong(ptr, ptr2, 42, success, fail, memory_scope_work_group);
+  // CHECK: switch i32 {{.*}}, label %[[MONOTONIC:[0-9a-zA-Z._]+]] [
+  // CHECK-NEXT: i32 1, label %[[ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 2, label %[[ACQUIRE]]
+  // CHECK-NEXT: i32 3, label %[[RELEASE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 4, label %[[ACQREL:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 5, label %[[SEQCST:[0-9a-zA-Z._]+]]
+
+  // CHECK: [[MONOTONIC]]
+  // CHECK: switch {{.*}}, label %[[MONOTONIC_MONOTONIC:[0-9a-zA-Z._]+]] [
+  // CHECK-NEXT: ]
+
+  // CHECK: [[ACQUIRE]]
+  // CHECK: switch {{.*}}, label %[[ACQUIRE_MONOTONIC:[0-9a-zA-Z._]+]] [
+  // CHECK-NEXT: i32 1, label %[[ACQUIRE_ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 2, label %[[ACQUIRE_ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: ]
+
+  // CHECK: [[RELEASE]]
+  // CHECK: switch {{.*}}, label %[[RELEASE_MONOTONIC:[0-9a-zA-Z._]+]] [
+  // CHECK-NEXT: ]
+
+  // CHECK: [[ACQREL]]
+  // CHECK: switch {{.*}}, label %[[ACQREL_MONOTONIC:[0-9a-zA-Z._]+]] [
+  // CHECK-NEXT: i32 1, label %[[ACQREL_ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 2, label %[[ACQREL_ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: ]
+
+  // CHECK: [[SEQCST]]
+  // CHECK: switch {{.*}}, label %[[SEQCST_MONOTONIC:[0-9a-zA-Z._]+]] [
+  // CHECK-NEXT: i32 1, label %[[SEQCST_ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 2, label %[[SEQCST_ACQUIRE:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: i32 5, label %[[SEQCST_SEQCST:[0-9a-zA-Z._]+]]
+  // CHECK-NEXT: ]
+
+  // CHECK: [[MONOTONIC_MONOTONIC]]
+  // CHECK: cmpxchg {{.*}} monotonic monotonic
+  // CHECK: br
+
+  // CHECK: [[ACQUIRE_MONOTONIC]]
+  // CHECK: cmpxchg {{.*}} acquire monotonic
+  // CHECK: br
+
+  // CHECK: [[ACQUIRE_ACQUIRE]]
+  // CHECK: cmpxchg {{.*}} acquire acquire
+  // CHECK: br
+
+  // CHECK: [[ACQREL_MONOTONIC]]
+  // CHECK: cmpxchg {{.*}} acq_rel monotonic
+  // CHECK: br
+
+  // CHECK: [[ACQREL_ACQUIRE]]
+  // CHECK: cmpxchg {{.*}} acq_rel acquire
+  // CHECK: br
+
+  // CHECK: [[SEQCST_MONOTONIC]]
+  // CHECK: cmpxchg {{.*}} seq_cst monotonic
+  // CHECK: br
+
+  // CHECK: [[SEQCST_ACQUIRE]]
+  // CHECK: cmpxchg {{.*}} seq_cst acquire
+  // CHECK: br
+
+  // CHECK: [[SEQCST_SEQCST]]
+  // CHECK: cmpxchg {{.*}} seq_cst seq_cst
+  // CHECK: br
+}
+
+int test_volatile(volatile atomic_int *i) {
+  // CHECK-LABEL: @test_volatile
+  // CHECK:      %[[i_addr:.*]] = alloca i32
+  // CHECK-NEXT: %[[atomicdst:.*]] = alloca i32
+  // CHECK-NEXT: store i32 addrspace(4)* %i, i32 addrspace(4)** %[[i_addr]]
+  // CHECK-NEXT: %[[addr:.*]] = load i32 addrspace(4)*, i32 addrspace(4)** %[[i_addr]]
+  // CHECK-NEXT: %[[res:.*]] = load atomic volatile i32, i32 addrspace(4)* %[[addr]] syncscope("workgroup") seq_cst
+  // CHECK-NEXT: store i32 %[[res]], i32* %[[atomicdst]]
+  // CHECK-NEXT: %[[retval:.*]] = load i32, i32* %[[atomicdst]]
+  // CHECK-NEXT: ret i32 %[[retval]]
+  return __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_work_group);
+}
+
+#endif

test/Preprocessor/init.c

 // WEBASSEMBLY32-NEXT:#define __LDBL_MIN_EXP__ (-16381)
 // WEBASSEMBLY32-NEXT:#define __LDBL_MIN__ 3.36210314311209350626267781732175260e-4932L
 // WEBASSEMBLY32-NEXT:#define __LITTLE_ENDIAN__ 1
 // WEBASSEMBLY32-NEXT:#define __LONG_LONG_MAX__ 9223372036854775807LL
 // WEBASSEMBLY32-NEXT:#define __LONG_MAX__ 2147483647L
 // WEBASSEMBLY32-NOT:#define __LP64__
 // WEBASSEMBLY32-NEXT:#define __NO_INLINE__ 1
 // WEBASSEMBLY32-NEXT:#define __OBJC_BOOL_IS_BOOL 0
+// WEBASSEMBLY32-NEXT:#define __OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES 3
+// WEBASSEMBLY32-NEXT:#define __OPENCL_MEMORY_SCOPE_DEVICE 2
+// WEBASSEMBLY32-NEXT:#define __OPENCL_MEMORY_SCOPE_SUB_GROUP 4
+// WEBASSEMBLY32-NEXT:#define __OPENCL_MEMORY_SCOPE_WORK_GROUP 1
+// WEBASSEMBLY32-NEXT:#define __OPENCL_MEMORY_SCOPE_WORK_ITEM 0
 // WEBASSEMBLY32-NEXT:#define __ORDER_BIG_ENDIAN__ 4321
 // WEBASSEMBLY32-NEXT:#define __ORDER_LITTLE_ENDIAN__ 1234
 // WEBASSEMBLY32-NEXT:#define __ORDER_PDP_ENDIAN__ 3412
 // WEBASSEMBLY32-NEXT:#define __POINTER_WIDTH__ 32
 // WEBASSEMBLY32-NEXT:#define __PRAGMA_REDEFINE_EXTNAME 1
 // WEBASSEMBLY32-NEXT:#define __PTRDIFF_FMTd__ "d"
 // WEBASSEMBLY32-NEXT:#define __PTRDIFF_FMTi__ "i"
 // WEBASSEMBLY32-NEXT:#define __PTRDIFF_MAX__ 2147483647

 // WEBASSEMBLY64-NEXT:#define __LDBL_MIN_EXP__ (-16381)
 // WEBASSEMBLY64-NEXT:#define __LDBL_MIN__ 3.36210314311209350626267781732175260e-4932L
 // WEBASSEMBLY64-NEXT:#define __LITTLE_ENDIAN__ 1
 // WEBASSEMBLY64-NEXT:#define __LONG_LONG_MAX__ 9223372036854775807LL
 // WEBASSEMBLY64-NEXT:#define __LONG_MAX__ 9223372036854775807L
 // WEBASSEMBLY64-NEXT:#define __LP64__ 1
 // WEBASSEMBLY64-NEXT:#define __NO_INLINE__ 1
 // WEBASSEMBLY64-NEXT:#define __OBJC_BOOL_IS_BOOL 0
+// WEBASSEMBLY64-NEXT:#define __OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES 3
+// WEBASSEMBLY64-NEXT:#define __OPENCL_MEMORY_SCOPE_DEVICE 2
+// WEBASSEMBLY64-NEXT:#define __OPENCL_MEMORY_SCOPE_SUB_GROUP 4
+// WEBASSEMBLY64-NEXT:#define __OPENCL_MEMORY_SCOPE_WORK_GROUP 1
+// WEBASSEMBLY64-NEXT:#define __OPENCL_MEMORY_SCOPE_WORK_ITEM 0
 // WEBASSEMBLY64-NEXT:#define __ORDER_BIG_ENDIAN__ 4321
 // WEBASSEMBLY64-NEXT:#define __ORDER_LITTLE_ENDIAN__ 1234
 // WEBASSEMBLY64-NEXT:#define __ORDER_PDP_ENDIAN__ 3412
 // WEBASSEMBLY64-NEXT:#define __POINTER_WIDTH__ 64
 // WEBASSEMBLY64-NEXT:#define __PRAGMA_REDEFINE_EXTNAME 1
 // WEBASSEMBLY64-NEXT:#define __PTRDIFF_FMTd__ "ld"
 // WEBASSEMBLY64-NEXT:#define __PTRDIFF_FMTi__ "li"
 // WEBASSEMBLY64-NEXT:#define __PTRDIFF_MAX__ 9223372036854775807L

test/Preprocessor/predefined-macros.c

 // CHECK-CL20: #define CL_VERSION_1_0 100
 // CHECK-CL20: #define CL_VERSION_1_1 110
 // CHECK-CL20: #define CL_VERSION_1_2 120
 // CHECK-CL20: #define CL_VERSION_2_0 200
 // CHECK-CL20: #define __OPENCL_C_VERSION__ 200
 // CHECK-CL20-NOT: #define __FAST_RELAXED_MATH__ 1
 // CHECK-FRM: #define __FAST_RELAXED_MATH__ 1
 
+// RUN: %clang_cc1 %s -E -dM -o - -x cl \
+// RUN:   | FileCheck %s --check-prefix=MSCOPE
+// MSCOPE:#define __OPENCL_MEMORY_SCOPE_ALL_SVM_DEVICES 3
+// MSCOPE:#define __OPENCL_MEMORY_SCOPE_DEVICE 2
+// MSCOPE:#define __OPENCL_MEMORY_SCOPE_SUB_GROUP 4
+// MSCOPE:#define __OPENCL_MEMORY_SCOPE_WORK_GROUP 1
+// MSCOPE:#define __OPENCL_MEMORY_SCOPE_WORK_ITEM 0
+
 // RUN: %clang_cc1 -triple aarch64-windows %s -E -dM -o - -x cl \
 // RUN:   | FileCheck -match-full-lines %s --check-prefix=CHECK-ARM64-WIN
 
 // CHECK-ARM64-WIN: #define _M_ARM64 1
 // CHECK-ARM64-WIN: #define _WIN32 1
 // CHECK-ARM64-WIN: #define _WIN64 1

test/SemaOpenCL/atomic-ops.cl

+// RUN: %clang_cc1 %s -cl-std=CL2.0 -finclude-default-header -verify -fsyntax-only -triple=spir64
+// RUN: %clang_cc1 %s -cl-std=CL2.0 -finclude-default-header -verify -fsyntax-only -triple=amdgcn-amdhsa-amd-opencl
+
+// Basic parsing/Sema tests for __opencl_atomic_*
+
+#pragma OPENCL EXTENSION cl_khr_int64_base_atomics : enable
+#pragma OPENCL EXTENSION cl_khr_int64_extended_atomics : enable
+
+struct S { char c[3]; };
+
+char i8;
+short i16;
+int i32;
+int8 i64;
+
+atomic_int gn;
+
+void f(atomic_int *i, const atomic_int *ci,
+       atomic_intptr_t *p, atomic_float *d,
+       int *I, const int *CI,
+       intptr_t *P, float *D, struct S *s1, struct S *s2,
+       global atomic_int *i_g, local atomic_int *i_l, private atomic_int *i_p,
+       constant atomic_int *i_c) {
+  __opencl_atomic_init(I, 5); // expected-error {{address argument to atomic operation must be a pointer to _Atomic type ('__generic int *' invalid)}}
+  __opencl_atomic_init(ci, 5); // expected-error {{address argument to atomic operation must be a pointer to non-const _Atomic type ('const __generic atomic_int *' (aka 'const __generic _Atomic(int) *') invalid)}}
+
+  __opencl_atomic_load(0); // expected-error {{too few arguments to function call, expected 3, have 1}}
+  __opencl_atomic_load(0, 0, 0, 0); // expected-error {{too many arguments to function call, expected 3, have 4}}
+  __opencl_atomic_store(0,0,0,0); // expected-error {{address argument to atomic builtin must be a pointer}}
+  __opencl_atomic_store((int *)0, 0, 0, 0); // expected-error {{address argument to atomic operation must be a pointer to _Atomic type ('__generic int *' invalid)}}
+  __opencl_atomic_store(i, 0, memory_order_relaxed, memory_scope_work_group);
+  __opencl_atomic_store(ci, 0, memory_order_relaxed, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to non-const _Atomic type ('const __generic atomic_int *' (aka 'const __generic _Atomic(int) *') invalid)}}
+  __opencl_atomic_store(i_g, 0, memory_order_relaxed, memory_scope_work_group);
+  __opencl_atomic_store(i_l, 0, memory_order_relaxed, memory_scope_work_group);
+  __opencl_atomic_store(i_p, 0, memory_order_relaxed, memory_scope_work_group);
+  __opencl_atomic_store(i_c, 0, memory_order_relaxed, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to non-constant _Atomic type ('__constant atomic_int *' (aka '__constant _Atomic(int) *') invalid)}}
+
+  __opencl_atomic_load(i, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_load(p, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_load(d, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_load(ci, memory_order_seq_cst, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to non-const _Atomic type ('const __generic atomic_int *' (aka 'const __generic _Atomic(int) *') invalid)}}
+
+  __opencl_atomic_store(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_store(p, 1, memory_order_seq_cst, memory_scope_work_group);
+  (int)__opencl_atomic_store(d, 1, memory_order_seq_cst, memory_scope_work_group); // expected-error {{operand of type 'void' where arithmetic or pointer type is required}}
+
+  int exchange_1 = __opencl_atomic_exchange(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  int exchange_2 = __opencl_atomic_exchange(I, 1, memory_order_seq_cst, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to _Atomic}}
+
+  __opencl_atomic_fetch_add(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_fetch_add(p, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_fetch_add(d, 1, memory_order_seq_cst, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to atomic integer or pointer ('__generic atomic_float *' (aka '__generic _Atomic(float) *') invalid)}}
+  __opencl_atomic_fetch_and(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_fetch_and(p, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_fetch_and(d, 1, memory_order_seq_cst, memory_scope_work_group); // expected-error {{address argument to bitwise atomic operation must be a pointer to atomic integer ('__generic atomic_float *' (aka '__generic _Atomic(float) *') invalid)}}
+
+  __opencl_atomic_fetch_min(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_fetch_max(i, 1, memory_order_seq_cst, memory_scope_work_group);
+  __opencl_atomic_fetch_min(d, 1, memory_order_seq_cst, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to atomic integer or pointer ('__generic atomic_float *' (aka '__generic _Atomic(float) *') invalid)}}
+  __opencl_atomic_fetch_max(d, 1, memory_order_seq_cst, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to atomic integer or pointer ('__generic atomic_float *' (aka '__generic _Atomic(float) *') invalid)}}
+
+  bool cmpexch_1 = __opencl_atomic_compare_exchange_strong(i, I, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+  bool cmpexch_2 = __opencl_atomic_compare_exchange_strong(p, P, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+  bool cmpexch_3 = __opencl_atomic_compare_exchange_strong(d, I, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group); // expected-warning {{incompatible pointer types passing '__generic int *' to parameter of type '__generic float *'}}
+  (void)__opencl_atomic_compare_exchange_strong(i, CI, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group); // expected-warning {{passing 'const __generic int *' to parameter of type '__generic int *' discards qualifiers}}
+
+  bool cmpexchw_1 = __opencl_atomic_compare_exchange_weak(i, I, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+  bool cmpexchw_2 = __opencl_atomic_compare_exchange_weak(p, P, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+  bool cmpexchw_3 = __opencl_atomic_compare_exchange_weak(d, I, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group); // expected-warning {{incompatible pointer types passing '__generic int *' to parameter of type '__generic float *'}}
+  (void)__opencl_atomic_compare_exchange_weak(i, CI, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group); // expected-warning {{passing 'const __generic int *' to parameter of type '__generic int *' discards qualifiers}}
+
+  // Pointers to different address spaces are allowed.
+  bool cmpexch_10 = __opencl_atomic_compare_exchange_strong((global atomic_int *)0x308, (constant int *)0x309, 1, memory_order_seq_cst, memory_order_seq_cst, memory_scope_work_group);
+
+  __opencl_atomic_init(ci, 0); // expected-error {{address argument to atomic operation must be a pointer to non-const _Atomic type ('const __generic atomic_int *' (aka 'const __generic _Atomic(int) *') invalid)}}
+  __opencl_atomic_store(ci, 0, memory_order_release, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to non-const _Atomic type ('const __generic atomic_int *' (aka 'const __generic _Atomic(int) *') invalid)}}
+  __opencl_atomic_load(ci, memory_order_acquire, memory_scope_work_group); // expected-error {{address argument to atomic operation must be a pointer to non-const _Atomic type ('const __generic atomic_int *' (aka 'const __generic _Atomic(int) *') invalid)}}
+
+  __opencl_atomic_init(&gn, 456);
+  __opencl_atomic_init(&gn, (void*)0); // expected-warning{{incompatible pointer to integer conversion passing '__generic void *' to parameter of type 'int'}}
+}
+
+void memory_checks(atomic_int *Ap, int *p, int val) {
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_load(Ap, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_load(Ap, memory_order_consume, memory_scope_work_group); // expected-error {{use of undeclared identifier 'memory_order_consume'}}
+  (void)__opencl_atomic_load(Ap, memory_order_release, memory_scope_work_group); // expected-warning {{memory order argument to atomic operation is invalid}}
+  (void)__opencl_atomic_load(Ap, memory_order_acq_rel, memory_scope_work_group); // expected-warning {{memory order argument to atomic operation is invalid}}
+  (void)__opencl_atomic_load(Ap, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_store(Ap, val, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_store(Ap, val, memory_order_acquire, memory_scope_work_group); // expected-warning {{memory order argument to atomic operation is invalid}}
+  (void)__opencl_atomic_store(Ap, val, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_store(Ap, val, memory_order_acq_rel, memory_scope_work_group); // expected-warning {{memory order argument to atomic operation is invalid}}
+  (void)__opencl_atomic_store(Ap, val, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_fetch_add(Ap, 1, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_add(Ap, 1, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_add(Ap, 1, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_add(Ap, 1, memory_order_acq_rel, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_add(Ap, 1, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_init(Ap, val);
+
+  (void)__opencl_atomic_fetch_sub(Ap, val, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_sub(Ap, val, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_sub(Ap, val, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_sub(Ap, val, memory_order_acq_rel, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_sub(Ap, val, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_fetch_and(Ap, val, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_and(Ap, val, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_and(Ap, val, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_and(Ap, val, memory_order_acq_rel, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_and(Ap, val, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_fetch_or(Ap, val, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_or(Ap, val, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_or(Ap, val, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_or(Ap, val, memory_order_acq_rel, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_or(Ap, val, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_fetch_xor(Ap, val, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_xor(Ap, val, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_xor(Ap, val, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_xor(Ap, val, memory_order_acq_rel, memory_scope_work_group);
+  (void)__opencl_atomic_fetch_xor(Ap, val, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_exchange(Ap, val, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_exchange(Ap, val, memory_order_acquire, memory_scope_work_group);
+  (void)__opencl_atomic_exchange(Ap, val, memory_order_release, memory_scope_work_group);
+  (void)__opencl_atomic_exchange(Ap, val, memory_order_acq_rel, memory_scope_work_group);
+  (void)__opencl_atomic_exchange(Ap, val, memory_order_seq_cst, memory_scope_work_group);
+
+  (void)__opencl_atomic_compare_exchange_strong(Ap, p, val, memory_order_relaxed, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_strong(Ap, p, val, memory_order_acquire, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_strong(Ap, p, val, memory_order_release, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_strong(Ap, p, val, memory_order_acq_rel, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_strong(Ap, p, val, memory_order_seq_cst, memory_order_relaxed, memory_scope_work_group);
+
+  (void)__opencl_atomic_compare_exchange_weak(Ap, p, val, memory_order_relaxed, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_weak(Ap, p, val, memory_order_acquire, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_weak(Ap, p, val, memory_order_release, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_weak(Ap, p, val, memory_order_acq_rel, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_compare_exchange_weak(Ap, p, val, memory_order_seq_cst, memory_order_relaxed, memory_scope_work_group);
+}
+
+void synchscope_checks(atomic_int *Ap, int scope) {
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, memory_scope_work_item); // expected-error{{synchronization scope argument to atomic operation is invalid}}
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, memory_scope_work_group);
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, memory_scope_device);
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, memory_scope_all_svm_devices);
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, memory_scope_sub_group);
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, scope); // expected-error{{non-constant synchronization scope argument to atomic operation is not supported}}
+  (void)__opencl_atomic_load(Ap, memory_order_relaxed, 10);    //expected-error{{synchronization scope argument to atomic operation is invalid}}
+}
+
+void nullPointerWarning(atomic_int *Ap, int *p, int val) {
+  // The 'expected' pointer shouldn't be NULL.
+  (void)__opencl_atomic_compare_exchange_strong(Ap, NULL, val, memory_order_relaxed, memory_order_relaxed, memory_scope_work_group); // expected-warning {{null passed to a callee that requires a non-null argument}}
+}