Skip to content

Conversation

@ronlieb
Copy link
Collaborator

@ronlieb ronlieb commented Nov 11, 2025

No description provided.

naveen-seth and others added 30 commits November 10, 2025 21:24
…w clangOptions library" (llvm#167374)

This relands llvm#167348.

The original PR was reverted due to a reported build failure, which was
later diagnosed as a local issue in the developer’s checkout or build
state. See discussion here:
llvm#163659 (comment)

No additional changes have been made in this reland.
This isn't reachable today but will come into play once we reorder
passes for llvm#147352 and llvm#147351.

Note that the `CBufferRowIntrin` helper struct is copied from the
`DXILCBufferAccess` pass, but it will be removed from there when we
simplify that pass in llvm#147351
trans bf16 insts do not support clamp and omod
…m#167276)

Preparing to moving most of implementation out of the header file.

* llvm#167280

---------

Co-authored-by: Naveen Seth Hanig <naveen.hanig@outlook.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
…lvm#167383)

Re-land: Add test for Complex imag literal GNU extension after updating
the name
…#152830)

This adds the codegen support for the dyn_groupprivate clause.
It is undefined behavior to call a function with a mismatched calling
convention. Rather than crash on this behavior, it should compile.

This LLVM defect was identified via the AMD Fuzzing project.
…166658) (llvm#166846)

The previous implementation in llvm#166252 (rolled back in llvm#166658) caused
buildbot failures due to a bug in an added test. The modified
`UseAfterClose` did not pass a `struct flock` value to `fcntl`:


https://github.com/llvm/llvm-project/blob/2d5170594147b42a37698760d6e0194eec4f1390/libc/test/src/fcntl/fcntl_test.cpp#L175

Which ASAN caught and errored in the `fcntl` implementation when the
unspecified argument was accessed:


https://github.com/llvm/llvm-project/blob/c12cb2892c808af459eaa270b8738a2b375ecc9b/libc/src/__support/OSUtil/linux/fcntl.cpp#L59
Use getDefiningRecipe to future-proof the code. Split off from
llvm#156262 as suggested.
This commit introduces `SpecialCaseList::Match`, a small struct to hold
the matched rule and its line number. This simplifies the `match`
methods by allowing them to return a single value instead of using a
callback.

---------

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Make the systemSupportsMemoryTagging() function return even on system
that don't support memory tagging. This avoids the need to always check
if memory tagging is supported before calling th function.

Make systemSupportsMemoryTagging() cache the getauxval return value
instead of calling the function every time.

Updated the code that calls systemSupportsMemoryTagging().
This hardens the code to check based on WideMember0's operands. This
ensures each call will go through the same check. Should be NFC
currently but needed when generalizing in follow-up patches.
)

Refactor computeWidth from CUFOpConversion into a shared helper function
computeElementByteSize in CUFCommon.
…m#153608)

This should be the peephole's job. Because and sets V flag to 0, this is
why signed comparisons with 0 are okay to replace with tst. Note this is
only for AArch64, because ANDS on ARM leaves the V flag the same.

Fixes: llvm#154387
… of pattern matching (llvm#166806)

Flang alias analysis used to find allocation site by pattern matching
allocation ops in mainly FIR dialect. This MR extends the
characterization to instead characterize based on whether the result of
an op has MemAlloc effect.
After llvm#164165, we emit warnings
from non-system headers by default.
This change only preserves functionality of `clang-tidy` as it was
before the change.
…m#158224)

Both conceptually belong to the same subtarget, so it should not
be necessary to pass in the context TargetRegisterInfo to any
TargetInstrInfo member. Add this reference so those superfluous
arguments can be removed.

Most targets placed their TargetRegisterInfo as a member
in TargetInstrInfo. A few had this owned by the TargetSubtargetInfo,
so unify all targets to look the same.
The save stats test would fail downstream due to the tests being run
from a read only directory. The cwd flag would attempt to place the
statistics in that read only directory, causing an error. This patch
ensures that the tests are always run from a temp directory.
llvm#167350)

fwrite and friends don't modify errno if no error occurred. Therefore
frite_unlocked's return value shouldn't be constructed from errno
without checking if an error actually occurred.

This fixes an error introduced by
llvm@9e2f73f
@ronlieb ronlieb requested review from a team and dpalermo November 11, 2025 12:01
@z1-cciauto
Copy link
Collaborator

@z1-cciauto z1-cciauto merged commit 6f22abd into amd-staging Nov 11, 2025
10 checks passed
@z1-cciauto z1-cciauto deleted the amd/merge/upstream_merge_20251111045254 branch November 11, 2025 14:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.