Ping.

Gentle ping.

LG.

Already covered by https://reviews.llvm.org/D127609.

In D143561#4117072, @qiucf wrote:

Is this an AIX only restriction or for all DWARF <= 4?

Check dwarf version number per @benmxwl-arm 's suggestion.

Is there any progress with this patch? This patch can fix our downstream test failure which is caused by insistent alignment between IR and backend, thus affects alias analysis.

Complete error messages.

Changes are included in https://reviews.llvm.org/D138986.

Add more checks on buffer boundary.

You can reproduce using the upstream buildbots (https://github.com/google/sanitizers/wiki/SanitizerBotReproduceBuild), but a quick-asan-build that's hacked together might be easier for iterative testing:

Will add more check on buffer boundary when reading archive content.

@arsenm Much appreciated for the suggestion.

Will seek another method that's not hacky.

Ping.

In D122118#4019373, @nikic wrote:

Compile-time: http://llvm-compile-time-tracker.com/compare.php?from=781eabeb40b8e47e3a46b0b927784e63f0aad9ab&to=0af2744a89bf0ed05e83ac1ed9d21d6d74cdfeca&stat=instructions%3Au

Use range loop.

Do you know on average how many chains we see per function?

Add target hook and command line option.

Check if the offset of the first member is still in the buffer. Fixed asan errors.

Thanks for reverting it in time. I'll have a look at the failure and fix it.

Address comments.

LGTM. Though there might be some pseudo instruction expansion changed the CFG, they are not breaking assumptions PPCCTRLoop made.

I would expect not only memset, some consecutive stores could also reuse the result of vector split, see https://godbolt.org/z/77aMvncb4.
For

void foo(long a[3]) {
    a[0] = 12;
    a[1] = 12;
    a[2] = 12;
}

LGTM.