Page MenuHomePhabricator

jdoerfert (Johannes Doerfert)
Argonne National Laboratory

Projects

User does not belong to any projects.

User Details

User Since
Jan 30 2014, 6:27 AM (364 w, 6 d)

Recent Activity

Yesterday

jdoerfert accepted D95585: [OpenMP] Disabled profiling in `libomp` by default to unblock link errors.

wrong selection :(

Wed, Jan 27, 8:18 PM · Restricted Project
jdoerfert requested changes to D95585: [OpenMP] Disabled profiling in `libomp` by default to unblock link errors.

LG, thanks

Wed, Jan 27, 8:18 PM · Restricted Project
jdoerfert accepted D95585: [OpenMP] Disabled profiling in `libomp` by default to unblock link errors.

LG, we need to backport this too. And we need to document that profiling multi-threaded applications needs this flag in the build.

Wed, Jan 27, 8:17 PM · Restricted Project
jdoerfert accepted D95584: [OpenMP][NFC] Added release note for new `deviceRTLs` and hidden helper task.

LG, some suggestions for the final version.

Wed, Jan 27, 7:45 PM · Restricted Project
jdoerfert added inline comments to D95282: [OpenMP] Add source location information to the libomptarget profile.
Wed, Jan 27, 7:28 PM · Restricted Project
jdoerfert added a comment to D95572: [libomptarget][NFC] Link plugins with threads support library due to std::call_once usage..

I've tested the change and everything seems to work fine. So let's commit it. Thanks for the patch!

Do you mean I can ignore the pre-merge checks fail? I do not see how my changes could have affected libomp build.

Wed, Jan 27, 7:23 PM · Restricted Project
jdoerfert added a comment to D95572: [libomptarget][NFC] Link plugins with threads support library due to std::call_once usage..

I've tested the change and everything seems to work fine. So let's commit it. Thanks for the patch!

Wed, Jan 27, 6:57 PM · Restricted Project
jdoerfert added a comment to D95574: [OpenMP][Libomptarget] Fix conditional in CMake for remote plugin.

please add to the backport list

Wed, Jan 27, 5:13 PM · Restricted Project
jdoerfert accepted D95574: [OpenMP][Libomptarget] Fix conditional in CMake for remote plugin.

LG

Wed, Jan 27, 5:13 PM · Restricted Project
jdoerfert accepted D95572: [libomptarget][NFC] Link plugins with threads support library due to std::call_once usage..

This is not NFC but other than that reasonable. LG, @grokos ?

Wed, Jan 27, 5:11 PM · Restricted Project
jdoerfert added a comment to D95556: [OpenMP][NVPTX] Disable building NVPTX deviceRTL by default on a non-CUDA system.

also need to be backported to 12

Wed, Jan 27, 1:12 PM · Restricted Project
jdoerfert added a comment to D95466: [OpenMP][NVPTX] Drop dependence on CUDA to build NVPTX `deviceRTLs`.

turn this build off by default if cuda is missing (what we used to have)

Wed, Jan 27, 11:40 AM · Restricted Project
jdoerfert accepted D95550: [OpenMP][deviceRTLs] Added `[[clang::loader_uninitialized]]` explicitly.

LG

Wed, Jan 27, 11:19 AM · Restricted Project
jdoerfert added a comment to D95398: [OpenMP] Fix building using LLVM_ENABLE_RUNTIMES.

This should probably be applied to the 12.0 release branch as well

@JonChesterfield I landed, is there anything else I should do?

Wed, Jan 27, 11:17 AM · Restricted Project
jdoerfert added a comment to D95545: [OpenMP][NVPTX] Added the missing -O1 when building NVPTX bitcode libraries.

we need to put this in 12 as well, I think we need to open a bug that blocks the 12 meta-bug and start adding these patches (or git hashes) to be included in the release branch.

Wed, Jan 27, 10:39 AM · Restricted Project
jdoerfert added a comment to D95545: [OpenMP][NVPTX] Added the missing -O1 when building NVPTX bitcode libraries.

Does O1 create TBAA? we might need -O3 -Xclang -disable-llvm-optzns or -Xclang -disable-O0-optnone if we can also remove the noinline that way.

Wed, Jan 27, 10:30 AM · Restricted Project
jdoerfert added inline comments to D95539: [LV] Add analysis remark for mixed precision conversions.
Wed, Jan 27, 10:19 AM · Restricted Project
jdoerfert added a comment to D95539: [LV] Add analysis remark for mixed precision conversions.

This is already caught by -Wimplicit-float-conversion/-Wdouble-promotion though
https://godbolt.org/z/o5Pzh4

Wed, Jan 27, 10:17 AM · Restricted Project

Tue, Jan 26

jdoerfert added a comment to D93727: [OpenMP] Add using bit flags to select Libomptarget Information.

offloading/dynamic_module_load.c fails due to this commit in my environment with x86_64 offload. It fails because std::call_once cannot be used without libpthread, and libomptarget.rtl.x86_64.so is not linked with libpthread. Basically, gdb shows that include/c++/7.3.0/x86_64-linux-gnu/bits/gthr-default.h::__gthread_active_p() returns false, when std::call_once is called from libomptarget.rtl.x86_64.so. It does return true, though, when std::call_once is called from libomptarget.so, because libomptarget.so is linked with libpthread.

It looks like GCC11 will have a fix: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55394
For the time being, all plugins have to be linked with libpthread, since Debug.h is considered to be common code for all plugins.

Any disagreement?

Tue, Jan 26, 9:48 PM · Restricted Project
jdoerfert retitled D95483: [OpenMP][WIP] Pass auxiliary YAML along with device image for OpenMP offload. from [WIP] Pass auxiliary YAML along with device image for OpenMP offload. to [OpenMP][WIP] Pass auxiliary YAML along with device image for OpenMP offload..
Tue, Jan 26, 6:46 PM
jdoerfert requested changes to D95483: [OpenMP][WIP] Pass auxiliary YAML along with device image for OpenMP offload..

vzakhari removed a reviewer: jdoerfert.

Tue, Jan 26, 6:46 PM
jdoerfert added a comment to D93734: [LoopDeletion] Insert an early exit from dead path in loop.

I may be doing something wrong, but D95468 did not help very much looking at these numbers it seems...

Tue, Jan 26, 6:18 PM · Restricted Project
jdoerfert accepted D95485: [OpenMP] libomp: fix clang-cl build for vs2019.

LG

Tue, Jan 26, 3:05 PM · Restricted Project
jdoerfert accepted D95467: [libomptarget][cuda] Only run tests when sure there is cuda available.

LG

Tue, Jan 26, 12:30 PM · Restricted Project
jdoerfert added inline comments to D95468: [LoopUnswitch] Add shortcut if unswitched path is a no-op..
Tue, Jan 26, 12:26 PM · Restricted Project
jdoerfert accepted D95314: [OpenMP][Libomptarget] Introduce Remote Offloading Plugin.

LG, with the nits I mentioned before.

Tue, Jan 26, 8:54 AM · Restricted Project
jdoerfert added a comment to D95314: [OpenMP][Libomptarget] Introduce Remote Offloading Plugin.

A few nits but given the opt-in semantics now I don't see a reason not to include this plugin (as we included other plugins before).

Tue, Jan 26, 8:40 AM · Restricted Project

Mon, Jan 25

jdoerfert committed rG8c7fdc4c61bf: [OpenMP] Add source location information to the libomptarget profile (authored by jdoerfert).
[OpenMP] Add source location information to the libomptarget profile
Mon, Jan 25, 8:44 PM
jdoerfert closed D95282: [OpenMP] Add source location information to the libomptarget profile.
Mon, Jan 25, 8:44 PM · Restricted Project
jdoerfert committed rGbd756286d2e7: [OpenMP][FIX] Enforce a function boundary for a new data environment (authored by jdoerfert).
[OpenMP][FIX] Enforce a function boundary for a new data environment
Mon, Jan 25, 8:43 PM
jdoerfert closed D94315: [OpenMP][FIX] Enforce a function boundary for a new data environment.
Mon, Jan 25, 8:43 PM · Restricted Project
jdoerfert requested review of D95416: [Docs] Remove `no-aa` from the alias analysis documentation.
Mon, Jan 25, 8:38 PM · Restricted Project
jdoerfert accepted D95415: [libomptarget][cuda] Handle missing _v2 symbols gracefully.

LG, with the nit @grokos pointed out.

Mon, Jan 25, 8:10 PM · Restricted Project
jdoerfert accepted D77609: [OpenMP] Added the support for hidden helper task in RTL.

Known issues resolved, AMDGPU is not yet a supported target and hard to test right now. LG

Mon, Jan 25, 7:12 PM · Restricted Project
jdoerfert accepted D95412: [libomptarget][cuda] Gracefully handle missing cuda library.

LG

Mon, Jan 25, 6:32 PM · Restricted Project
jdoerfert added inline comments to D95314: [OpenMP][Libomptarget] Introduce Remote Offloading Plugin.
Mon, Jan 25, 4:38 PM · Restricted Project
jdoerfert added a comment to D95371: [OpenMP] Use C++ to link libomp.so when LLVM libraries are included.

That's right. Regardless that the code is written in C++ the intention is to not have any dependency on C++ runtime. The libomp.so should work fine with C applications as well as with Fortran codes those often cannot link C++ runtime.

Right. With profile support we got C++ dependences. So if you disable the profile support you can remove the dependence. What if we set the linker language to CXX if the profile support is enabled? Is that a problem?

But if profile support is enabled by default and will got into LLVM release, then people with pure C codes of Fortran codes will be surprised to see their OpenMP application depends on C++ runtime.
E.g. if they provide Fortran redistributables with Fortran application, it won't work because of C++ runtime dependency.
Or somebody may setup flang without clang, or gfortran without gcc, and link with libomp.
Not sure it is immediate problem, but in length of time such things will happen.

Mon, Jan 25, 1:14 PM · Restricted Project
jdoerfert added a comment to D95371: [OpenMP] Use C++ to link libomp.so when LLVM libraries are included.

And all exported from libomp.so symbols supposed to be C symbols.

Mon, Jan 25, 11:15 AM · Restricted Project
jdoerfert added a comment to D95274: [libomptarget][cuda] Call v2 functions explicitly.

What version of cuda are you using? It might be useful to know when things changed.

I think we can dlopen cuda, and try to load cuMemcpyDtoHAsync_v2, and then try to load cuMemcpyDtoHAsync on failure. That would get us some degree of robustness to different symbol versions.

$ cat /usr/local/cuda/version.txt 
CUDA Version 10.2.89
Mon, Jan 25, 8:18 AM · Restricted Project
jdoerfert accepted D95367: [libomptarget] Compile with older cuda, revert D95274.

LGTM, thanks

Mon, Jan 25, 8:12 AM · Restricted Project
jdoerfert added a comment to D95288: [ValueTracking] Don't assume readonly function will return.

Yeah, I agree that mustprogress, argmemonly and nosync should imply willreturn. However, FuncAttrs currently infers neither nosync nor argmemonly, so it would need those first. I think that the argmemonly check here was there to cover some common intrinsics like llvm.memcpy, but those are explicitly willreturn now.

Mon, Jan 25, 8:09 AM · Restricted Project
jdoerfert added inline comments to D95288: [ValueTracking] Don't assume readonly function will return.
Mon, Jan 25, 7:55 AM · Restricted Project
jdoerfert resigned from D95306: openmp: stop checking /proc/cpuinfo on arm mac.
Mon, Jan 25, 7:52 AM · Restricted Project
jdoerfert added a comment to D95274: [libomptarget][cuda] Call v2 functions explicitly.

We can unblock you by going back to the non-suffixed versions in rtl.cpp and adding the macros back to dynamic_cuda/cuda.h. That will mean the following combinations work:

Mon, Jan 25, 7:51 AM · Restricted Project

Sun, Jan 24

jdoerfert added a comment to D95282: [OpenMP] Add source location information to the libomptarget profile.

Seems like this patch introduces many unrelated changes.

Sun, Jan 24, 5:23 PM · Restricted Project
jdoerfert added a comment to D95318: [OpenMP][NFC] Format all files in OpenMP.

I am afraid such action may be needed regularly, unless we freeze the clang-format version and write it in the documentation.
Because each version of clang-format may produce different result of formatting, unfortunately.
At least that was the case couple of years ago.

Sun, Jan 24, 4:20 PM · Restricted Project

Sat, Jan 23

jdoerfert updated the diff for D95282: [OpenMP] Add source location information to the libomptarget profile.

Remove leftover

Sat, Jan 23, 10:29 PM · Restricted Project
jdoerfert updated the diff for D95282: [OpenMP] Add source location information to the libomptarget profile.

More profiling and selective at to avoid too coarse grained scopes

Sat, Jan 23, 10:26 PM · Restricted Project
jdoerfert added a comment to D93040: [InlineFunction] Use llvm.experimental.noalias.scope.decl for noalias arguments..
Sat, Jan 23, 1:11 PM · Restricted Project, Restricted Project
jdoerfert added a comment to D95141: [InstCombine] Remove unused llvm.experimental.noalias.scope.decl.

Logic seems sounds to me. Code as well, one minor nit below. I have no objections to this.

Sat, Jan 23, 1:03 PM · Restricted Project
jdoerfert accepted D95293: [OpenMP] Fixed test environment of `check-libomptarget-nvptx`.

LG

Sat, Jan 23, 12:47 PM · Restricted Project
jdoerfert accepted D95295: [libomptarget][amdgpu][nfc] Update comments.

LG

Sat, Jan 23, 12:46 PM · Restricted Project

Fri, Jan 22

jdoerfert added a comment to D95282: [OpenMP] Add source location information to the libomptarget profile.

Still not perfect but getting there:

Fri, Jan 22, 5:45 PM · Restricted Project
jdoerfert requested review of D95282: [OpenMP] Add source location information to the libomptarget profile.
Fri, Jan 22, 5:44 PM · Restricted Project
jdoerfert accepted D95278: [InstCombine] remove incompatible attribute when simplifying some lib calls.

LGTM. Thanks for following up on this!

Fri, Jan 22, 5:16 PM · Restricted Project
jdoerfert accepted D95277: [OpenMP] Return NULL from unsuccessful OMPT function lookup.

Switch the commit message with the subject though. LGTM

Fri, Jan 22, 5:11 PM · Restricted Project
jdoerfert accepted D95274: [libomptarget][cuda] Call v2 functions explicitly.

LGTM

Fri, Jan 22, 4:34 PM · Restricted Project
jdoerfert accepted D94106: [Local] Treat calls that may not return as being alive (WIP)..
Fri, Jan 22, 3:11 PM · Restricted Project
jdoerfert added a comment to D94106: [Local] Treat calls that may not return as being alive (WIP)..

Can we add something along the lines of the commit message for D95260

Fri, Jan 22, 3:11 PM · Restricted Project
jdoerfert added a comment to D95260: [Local] Don't DCE calls that may not return.

Calls that aren't willreturn should not be eliminated, as this may remove an infinite loop.

Fri, Jan 22, 1:30 PM · Restricted Project
jdoerfert accepted D95155: [libomptarget] Build cuda plugin without cuda installed locally.

One nit wrt. _v2 not _v2, otherwise LGTM.

Fri, Jan 22, 12:13 PM · Restricted Project
jdoerfert added inline comments to D94106: [Local] Treat calls that may not return as being alive (WIP)..
Fri, Jan 22, 7:54 AM · Restricted Project
jdoerfert added a comment to D95155: [libomptarget] Build cuda plugin without cuda installed locally.

Pretty much good to go. CMAKE variable, one potential change in the comment below, maybe name the declarations _v2.

Fri, Jan 22, 7:47 AM · Restricted Project
jdoerfert accepted D95126: [llvm-link] Fix for an assertion when linking global with appending linkage.

Looks reasonable to me. Test coverage is nice now, thanks.

Fri, Jan 22, 7:43 AM · Restricted Project
jdoerfert accepted D95137: [update_(llc_)test_checks.py] Support pre-processing commands.

LGTM. Maybe wait a bit if someone else wants to take a look.

Fri, Jan 22, 7:40 AM · Restricted Project
jdoerfert added a comment to D95238: [LangRef] Update memory access ops to raise UB if ptrs are not well defined.

+1 on the motivation. When we landed the other patch we didn't consider all the implications and, so far, they clearly outweigh the benefits of having partial undef offsets "work".

Fri, Jan 22, 7:37 AM · Restricted Project

Thu, Jan 21

jdoerfert added inline comments to D95155: [libomptarget] Build cuda plugin without cuda installed locally.
Thu, Jan 21, 9:36 PM · Restricted Project
jdoerfert added inline comments to D94745: [OpenMP][deviceRTLs] Build the deviceRTLs with OpenMP instead of target dependent language.
Thu, Jan 21, 3:47 PM · Restricted Project, Restricted Project
jdoerfert added a comment to D95155: [libomptarget] Build cuda plugin without cuda installed locally.

Interesting. Should we do this instead of the pure macro stuff? Is this going to break if the argument type doesn't match the declaration parameter type, e.g.,

void f(int);
Thu, Jan 21, 3:03 PM · Restricted Project
jdoerfert added a comment to D95122: [Libcalls, Attrs] Annotate libcalls with noundef.

Since this issue only happens when the undef value is involved, another solution is to revert D87994 as suggested & make load/store's pointer noundef & rely on the future that the undef value will be finally removed.
I don't think this will cause a problem in practice as well even if undef persists; memory sanitizer is already catching the case.

Thu, Jan 21, 1:09 PM · Restricted Project
jdoerfert accepted D95133: [OpenMP] Fix failing test due to change in offloading flags.

LGTM one nit

Thu, Jan 21, 8:32 AM · Restricted Project
jdoerfert added a comment to D95122: [Libcalls, Attrs] Annotate libcalls with noundef.

LGTM. FWIW, nonnull should imply noundef because if it was undef we could pick null.

I did not look into the changes in this patch yet, but does it mean that from

f(i8* nonnull %p)

it is allowed to attach noundef?

f(i8* nonnull noundef %p)

I think this shouldn't be allowed, because the former one isn't UB when %p was null but the latter one raises UB.

Thu, Jan 21, 8:28 AM · Restricted Project
jdoerfert added a comment to D93734: [LoopDeletion] Insert an early exit from dead path in loop.

What is left after we merged the loop unswitch solution?

Thu, Jan 21, 7:13 AM · Restricted Project
jdoerfert accepted D95122: [Libcalls, Attrs] Annotate libcalls with noundef.
Thu, Jan 21, 7:07 AM · Restricted Project
jdoerfert added a comment to D95122: [Libcalls, Attrs] Annotate libcalls with noundef.

LGTM. FWIW, nonnull should imply noundef because if it was undef we could pick null.

Thu, Jan 21, 7:07 AM · Restricted Project
jdoerfert added a comment to D95126: [llvm-link] Fix for an assertion when linking global with appending linkage.

Given the langref text I'm unsure what it means to link an append linkage variable to a non-append linkage one. I can see how extern is different but what would happen before & now if you remove the extern. And what happens before/now if you have append on both @var? We should add those tests if possible.

Thu, Jan 21, 7:05 AM · Restricted Project
jdoerfert added inline comments to D95088: [InstCombine] avoid crashing on attribute propagation.
Thu, Jan 21, 7:00 AM · Restricted Project

Wed, Jan 20

jdoerfert added a comment to D94745: [OpenMP][deviceRTLs] Build the deviceRTLs with OpenMP instead of target dependent language.

Yeah... That's not gone well. Unless @jdoerfert can point us at a better way to spell these in openmp, let's add 'fix target atomics' to the todo pile and use clang intrinsics for the time being.

Wed, Jan 20, 9:12 PM · Restricted Project, Restricted Project
jdoerfert accepted D94633: [FunctionAttrs] Infer willreturn for functions without loops.

LGTM

Wed, Jan 20, 9:09 PM · Restricted Project, Restricted Project
jdoerfert updated the diff for D95104: [OpenMP] Replace `cuda.h` with (forward) declarations in the offload plugin.

Look for CUDA functions on demand and once only, inform libomptarget if CUDA was not found.

Wed, Jan 20, 5:09 PM · Restricted Project
jdoerfert updated the summary of D95104: [OpenMP] Replace `cuda.h` with (forward) declarations in the offload plugin.
Wed, Jan 20, 4:49 PM · Restricted Project
jdoerfert updated the diff for D95104: [OpenMP] Replace `cuda.h` with (forward) declarations in the offload plugin.

Load the cuda functions at runtime

Wed, Jan 20, 4:47 PM · Restricted Project
jdoerfert requested review of D95104: [OpenMP] Replace `cuda.h` with (forward) declarations in the offload plugin.
Wed, Jan 20, 4:06 PM · Restricted Project
jdoerfert added a comment to D95023: [test-suite] SPEC2017 CPU Exchange2 integer tests..

Looks OK to me, @Meinersbur can accept at some point.

Wed, Jan 20, 3:32 PM
jdoerfert accepted D95095: [BuildLibcalls, Attrs] Support more variants of C++'s new, add attributes for C++'s delete.

LGTM

Wed, Jan 20, 2:38 PM · Restricted Project
jdoerfert accepted D95088: [InstCombine] avoid crashing on attribute propagation.

LGTM, one comment below though.

Wed, Jan 20, 2:00 PM · Restricted Project
jdoerfert added a comment to D95062: [libomptarget][devicertl][nfc] Simplify target_atomic abstraction.

None of the three atomic operations provided by OpenMP, read, write, and update, sound like similar semantics as the four operations here.

Wed, Jan 20, 1:31 PM · Restricted Project
jdoerfert accepted D95085: [OpenMP][NVPTX] Added forward declaration for atomic operations.

LGTM

Wed, Jan 20, 1:14 PM · Restricted Project
jdoerfert added a comment to D76127: [TableGen] Do not set ReadOnly attribute on intrinsics with side effects.

TBH, I feel "X is readonly and has side effects" sends the wrong message to begin with. It is a contradiction (in the IR world) as basically shown by the need for this patch. Given that there are no examples in-tree I don't understand why one would mark a side-effect intrinsic as readonly (or similar). Long story short, I would argue this should be a loud error, not silently ignored.

Why shouldn't the intrinsics file be able to express the semantics we want even if we can't represent it in IR today?

Wed, Jan 20, 1:09 PM · Restricted Project
jdoerfert added a comment to D95062: [libomptarget][devicertl][nfc] Simplify target_atomic abstraction.

LG, can you replace it with clang or OpenMP atomics next?

Yep, expect so. OpenMP probably expects the variables to have atomic types, C++-style, and they're presently a mix of unadorned and volatile.

Wed, Jan 20, 12:11 PM · Restricted Project
jdoerfert accepted D95062: [libomptarget][devicertl][nfc] Simplify target_atomic abstraction.

LG, can you replace it with clang or OpenMP atomics next?

Wed, Jan 20, 11:48 AM · Restricted Project
jdoerfert accepted D95060: [libomptarget][devicertl][nfc] Remove some cuda intrinsics, simplify.

LGTM

Wed, Jan 20, 11:44 AM · Restricted Project
jdoerfert added a comment to D76127: [TableGen] Do not set ReadOnly attribute on intrinsics with side effects.

TBH, I feel "X is readonly and has side effects" sends the wrong message to begin with. It is a contradiction (in the IR world) as basically shown by the need for this patch. Given that there are no examples in-tree I don't understand why one would mark a side-effect intrinsic as readonly (or similar). Long story short, I would argue this should be a load error, not silently ignored.

Wed, Jan 20, 11:36 AM · Restricted Project
jdoerfert added a comment to D93525: [OpenMP] Add unbundling of archives containing bundled object files into device specific archives.

At the moment this patch defines compatibility as exact string match of bundler entry ID.
[...]
Supporting target ID requires little more work and discussion.

Let's get this in first, then revisit target ID support as we need it.

I do not think this patch should ignore target ID as that is now upstreamed and documented. What is involved in correcting the compatibility test to be correct by the target ID rules? There are examples of doing this in all the runtimes and I can help if that is useful.

Wed, Jan 20, 8:49 AM · Restricted Project
jdoerfert added a comment to D93525: [OpenMP] Add unbundling of archives containing bundled object files into device specific archives.

At the moment this patch defines compatibility as exact string match of bundler entry ID.
[...]
Supporting target ID requires little more work and discussion.

Wed, Jan 20, 8:05 AM · Restricted Project

Tue, Jan 19

jdoerfert added a comment to D94745: [OpenMP][deviceRTLs] Build the deviceRTLs with OpenMP instead of target dependent language.

Let's split this. Declare target in all the source files can go first, doesn't hurt anyone.
Forward declarations should also work fine with existing compilation, so that would be two.
Conditional defines and a SHARED(...) macro would be three.
Four is the cmake magic. Potentially a different folder to build it as C++ + OpenMP alternatively.

Tue, Jan 19, 9:21 PM · Restricted Project, Restricted Project
jdoerfert added inline comments to D94745: [OpenMP][deviceRTLs] Build the deviceRTLs with OpenMP instead of target dependent language.
Tue, Jan 19, 4:20 PM · Restricted Project, Restricted Project
jdoerfert accepted D94850: [BuildLibcalls] Mark some libcalls with inaccessiblememonly and inaccessiblemem_or_argmemonly.

LGTM, I think this is correct. Thanks a lot!

Tue, Jan 19, 4:13 PM · Restricted Project
jdoerfert added a comment to D95003: [OpenMP] libomp: implement OpenMP 5.1 nteams-var and teams-thread-limit-var ICVs.

We should replace the OMP_NUM_TEAMS handling in the plugins with this, right? Separate patch or can we do it right away?

Tue, Jan 19, 4:00 PM · Restricted Project
jdoerfert added inline comments to D94871: [Clang][OpenMP] Fixed an issue that clang crashed when compiling OpenMP program in device only mode without host IR.
Tue, Jan 19, 8:17 AM · Restricted Project, Restricted Project