llvm-project.git, branch users/zhaoqi5/test-vec-addressmode Unnamed repository; edit this file 'description' to name the repository. pass format 2025-08-11T02:28:07+00:00 Qi Zhao zhaoqi01@loongson.cn 2025-08-04T08:50:54+00:00 e0ba234672430cff502506210a7b2db6ef704fc7 






[LoongArch] Pre-commit tests for vector type isLegalAddressingMode implementation
2025-08-11T02:28:07+00:00

Qi Zhao
zhaoqi01@loongson.cn

2025-08-04T07:51:57+00:00

1358e2fec8e20378157da6fe5a6a1ef913bace2a












[PatternMatch] Add `m_[Shift]OrSelf` matchers. (#152924)
2025-08-11T01:58:16+00:00

Yingwei Zheng
dtcxzyw2333@gmail.com

2025-08-11T01:58:16+00:00

1c499351d682aa46c2f087a6f757d22b01d18aa7

Address the comment
https://github.com/llvm/llvm-project/pull/147414/files#r2228612726.
As they are usually used to match integer packing patterns, it is enough
to handle constant shamts.




Address the comment
https://github.com/llvm/llvm-project/pull/147414/files#r2228612726.
As they are usually used to match integer packing patterns, it is enough
to handle constant shamts.






[libc][math][c++23] Add bf16mul{,f,l,f128} math functions (#152847)
2025-08-11T01:38:27+00:00

Krishna Pandey
kpandey81930@gmail.com

2025-08-11T01:38:27+00:00

628c0e33e4f1ca74787dede621677d8eaca4f726

This PR adds the following basic math functions for BFloat16 type along
with the tests:
- bf16mul
- bf16mulf
- bf16mull
- bf16mulf128

---------

Signed-off-by: Krishna Pandey <kpandey81930@gmail.com>




This PR adds the following basic math functions for BFloat16 type along
with the tests:
- bf16mul
- bf16mulf
- bf16mull
- bf16mulf128

---------

Signed-off-by: Krishna Pandey <kpandey81930@gmail.com>






[libclc] Implement clc_log/sinpi/sqrt with __nv_* functions (#150174)
2025-08-11T00:53:49+00:00

Wenju He
wenju.he@intel.com

2025-08-11T00:53:49+00:00

12cec437c664e1b671d85f38418c97f222340ce4

This is to upstream implementations in
https://github.com/intel/llvm/tree/sycl/libclc/clc/lib/ptx-nvidiacl/math




This is to upstream implementations in
https://github.com/intel/llvm/tree/sycl/libclc/clc/lib/ptx-nvidiacl/math






AMDGPU: Add debug printing for early exit if there are no AGPRs allocated
2025-08-11T00:27:05+00:00

Matt Arsenault
Matthew.Arsenault@amd.com

2025-08-11T00:27:05+00:00

1f86deb5a42517444378442e7c38580622069081












[Docs] Fix the description about %clang_cpp (#152842)
2025-08-10T23:41:36+00:00

parabola94
heavybaby5000@toki.waseda.jp

2025-08-10T23:41:36+00:00

d218789d3efee246c9627256c8a44e1579af6b21












[Offload][NFC] Re-enable clang-format for omptarget.h (#152937)
2025-08-10T22:46:37+00:00

Kevin Sala Penades
salapenades1@llnl.gov

2025-08-10T22:46:37+00:00

5751e96f9a749c5ddb1fc8d9831a835f7f4c6189












[VPlan] Add dedicated user for resume phi with epilogue vectorization.
2025-08-10T20:21:16+00:00

Florian Hahn
flo@fhahn.com

2025-08-10T20:18:46+00:00

86813aa786190005f51008f6419901e738fd18f3

Epilogue vectorization currently relies on the resume phi for the
canonical induction being always available, which is why VPPhi are
considered to have side-effects, to prevent their removal.

This patch adds a new ResumeForEpilogue opcode to mark the resume phi as
used for epilogue vectorization. This allows treating VPPhis in general
as not having side-effects, enabling removal of unused VPPhis.





Epilogue vectorization currently relies on the resume phi for the
canonical induction being always available, which is why VPPhi are
considered to have side-effects, to prevent their removal.

This patch adds a new ResumeForEpilogue opcode to mark the resume phi as
used for epilogue vectorization. This allows treating VPPhis in general
as not having side-effects, enabling removal of unused VPPhis.







[Support] Fix CRTP for GraphWriter (NFC) (#152811)
2025-08-10T19:41:34+00:00

Naveen Seth Hanig
naveen.hanig@outlook.com

2025-08-10T19:41:34+00:00

a2d353ea17e14a1fc193ec1cf4a6cd39cd73ee44

Commit 474bbc1 forgot to change the members of GraphWriterBase from private 
to protected. 
Because of this, GraphWriter can't be properly partially specialized using CRTP.
This commit corrects that, allowing GraphWriterBase to be partially specialized
using CRTP as intended.




Commit 474bbc1 forgot to change the members of GraphWriterBase from private 
to protected. 
Because of this, GraphWriter can't be properly partially specialized using CRTP.
This commit corrects that, allowing GraphWriterBase to be partially specialized
using CRTP as intended.