- User Since
- Jun 17 2015, 7:07 AM (174 w, 4 d)
Thu, Oct 4
- Address @pekka.jaaskelainen's review.
Wed, Oct 3
I actually already created a review for the Clang part which is D52117, but uploaded the wrong diff. Corrected. I'll wait for both being accepted before committing.
- Upload diff for clang portion (instead of D52116)
Thank you for your feedback. If you don'r mind, before I commit this, I will prepare a patch for clang generating this new kind of metadata (since atm nothing is generating it) and give others some time for feedback.
Mon, Oct 1
Fri, Sep 28
- Address @hfinkel's remarks
Thu, Sep 27
Altough I fixed the related bug in r343212, this patch might still be useful in case we ever would want to allow nested parallelism.
Wed, Sep 26
- Use call access group if instruction's access group is not set
Mon, Sep 24
Commited as rL342861.
Sep 19 2018
Sep 17 2018
Currently, when I apply this, I get the following error (Windows and Linux):
CMake Error at cmake/modules/LLVM-Config.cmake:105 (target_link_libraries): Target "LLVMPolly" of type MODULE_LIBRARY may not be linked into another target. One may link only to STATIC or SHARED libraries, or to executables with the ENABLE_EXPORTS property set. Call Stack (most recent call first): cmake/modules/LLVM-Config.cmake:93 (explicit_llvm_config) cmake/modules/AddLLVM.cmake:774 (llvm_config) cmake/modules/AddLLVM.cmake:1108 (add_llvm_executable) tools/polly/unittests/CMakeLists.txt:9 (add_unittest) tools/polly/unittests/Isl/CMakeLists.txt:1 (add_polly_unittest)
Sep 14 2018
I talked with @hfinkel about what to do in such cases. He's ok with just reverting. So go ahead and commit the revert.
Sep 13 2018
Can you add a test case? Since I never contributed any code to bugpoint, I am not comfortable with just accepting it (usually one asks the original author, who, as you mentioned, is not active anymore). But it would obvious if there was a test case that illustrates the problem/ensure the bug does not reappear.
Does this fix the Windows or "non-MSVC" builds (as D51904)? For which configurations?
Sep 10 2018
Sep 8 2018
Sorry, this does not fix bug http://llvm.org/PR38390 for me. I still get:
$ bugpoint polly-timeout-grtestutils.ll -verify -opt-command=C:\Users\meinersbur\build\llvm\release\bin\opt.exe Read input file : 'polly-timeout-grtestutils.ll' *** All input ok Running selected passes on program to test for crash: C:\Users\meinersbur\build\llvm\release\bin\opt.exe: bugpoint-input-18caf6b.bc: error: Could not open input file: permission denied Exited with error code '1'
Sep 4 2018
Aug 30 2018
Thanks for investing time into the documentation
Aug 29 2018
Aug 22 2018
Can you outline the intended workflow? My guess:
Aug 20 2018
I am ok with committing this, but maybe we should have someone else's opinion as well?
Aug 17 2018
- Report unroll-and-jam as not applied even if unroll is present as well.
- rename followup_cyclic/followup_noncyclic to followup_sequential/followup_coincident
- Move hasUnrollAndJamTransformation in LoopUnrollAndJamPass to different place
- Remove some unrelated whitespace changes made by clang-format
- Extract followup attribute names into constant
Aug 16 2018
Aug 14 2018
I am thinking about adding a LoopMetadataTacker (sort of a combination of LoopVectorizeHints and AssumptionTracker) analysis pass which would centralize the interpretation of that metadata and avoid the linear search through the metadata list when looking up a specific attribute.
Aug 13 2018
To add some details: The original failure was external/skia/src/gpu/GrTestUtils.cpp of the aosp buildbot. It expects a library function __muloti4 which was lowered from the intrinsic llvm.smul.with.overflow.i128. The next smaller llvm.smul.with.overflow.i64 should be able to handle 64 bit values (not just 63 bits).
Aug 10 2018
Aug 9 2018
- Explicitly document followup of not applied transformations to be ignored
- Unroll/UnrollAndJam: Interpret enable/count/full as forced
- Unroll/UnrollAndJam: Add tests for disable_nonforced combined with enable/count/full
- Reduce size of unroll-pragmas_transform.ll
Should the total execution time of benchmarks be around second or less?
Maybe reduce the number of tested parameters to reduce the total runtime?
Could you close the review if it's not needed anymore?
Aug 7 2018
Looks like an obvious change, but r338787 is by @MatzeB I'd wait for his ok.
@santosh Do you need someone to commit?
Thank you. LGTM.
Aug 6 2018
Aug 4 2018
Aug 3 2018
For this patch the goal is to have the attributes in the AST in an order that is less surprising to consumers (including out-of-tree). If we change it now, new/revised code/diagnostics will be written to match this order.