thx for the added coverage.
LGTM

LGTM
Thx

Since @phosek requested changes, this now require his review to see if his concerns are addressed.
@phosek can you please take a look at this patch to see if it addresses your concerns? PowerPC require this patch for our libcxx builds.

In D126122#4100211, @arcivanov wrote:

@lei long time ago. Let me rebase it on main.
I assume after that is committed you'll "cherry-pick" into 16.x via GH issue?

Since Eric mentioned above that he didn't mean to block this patch and he hasn't come back to it, I am going to assume it is okay to land this patch now. I can help to commit this tomorrow if we don't get any blocking comments by then.

@arcivanov are you still waiting for Eric to respond or are you looking for help to commit this patch?

@tstellar PowerPC would need this patch enable to publish a build with libcxx. When does this patch need to land enable to get this into LLVM 16.0?

It would be great if we can get this in 16.

This patch fixes the exact problem I am seeing when building on PowerPC for -DLLVM_ENABLE_RUNTIMES=compiler-rt;libcxx;libcxxabi -DLLVM_ENABLE_LIBCXX:BOOL=ON.
I have applied this patch and tested it on the ppcle rhel/aix/flang/mlir bots and it does not cause any regressions so this LGTM.

Thanks! This LGTM

fix test doc wording

Please note that this patch broke any bots that build this project with -Werror (example: https://lab.llvm.org/buildbot/#/builders/19/builds/14037). As a result, I have reverted it since there has been no fix within a reasonable time.
May I advise ensuring that a project is clean wrt. to the additional warnings when turning on extra warnings as there are bound to be some builds that use -Werror. Also, new warnings are almost always disruptive changes as there are many different build configurations that may be affected in unforeseen ways. If you have such a disruptive change ready to commit, it is advisable to wait until Monday to commit it rather than committing it on a Friday evening as most people are disconnecting.

LGTM
Thx!

LGTM
Thx

LGTM
Thx

1 missed bitcast, otherwise LGTM

@phosek It seems the runtime tests are no longer running for us after this patch in PPC bot https://lab.llvm.org/buildbot/#/builders/57. I can see they are still ran when it's part of LLVM_ENABLE_PROJECTS, but this bot specifies compiler-rt as LLVM_ENABLE_RT=compiler-rt.

LGTM
Thx!

Address review comments from Nemanja

Simplify vec_replace_[elt|unaligned] code gen.

Move error checking for vec_replace_unaligned to during code gen.

rebase to ToT

LGTM, just a few nit that can be addressed on commit.

Add fix for vec_replace_elt and vec_replace_unaligned.

Please document all new functions added.

LGTM
Thx

LGTM

LGTM
Thx for fixing this.

LGTM with minor updates before commit.

This also failing on all our bots:
https://lab.llvm.org/buildbot/#/builders/105/builds/22623
https://lab.llvm.org/buildbot/#/builders/100/builds/13530

Commited to release/14.x via ce1e90fc8dbacdaff0310e50090ec3cfdfce4def

In D120907#3357433, @ldionne wrote:

LGTM for libc++. I assume this is targeting release/14.x only.

remove unintented change

LGTM

LGTM
Thx

LGTM
Thx for cleaning up.

Not a silly question at all!
Initially I thought that I didn't need to update the tests because I wan't changing code and the CHECK/CHECK-NEXT lines don't really care that there are comments that come after the instructions. This is true in most cases but there are tests that already have a comment after the instruction line (for example # 8-byte Folded Spill) and in those cases by adding a second comment I'm adding a new line and the CHECK-NEXT lines don't work anymore. Anyway, I'm going to update those tests now.

LGTM
Thx for the cleanup!

LGTM, just a minor comment.

fix indentation.

LGTM other then the comment about the unecessary code block.

Switch to using new cmake variable to control comments.

Instead of adding:

if (isMMAType(Src))
    return InstructionCost::getMax();

to all the different cost calculating functions, is it possible to just add them the underlying functions?

address review comments

LGTM
I think you went a bit overkill with the tests for this patch 🙂. Please cut down the number of run lines before committing.