[PATCH] D120305: [Driver] Default CLANG_DEFAULT_PIE_ON_LINUX to ON

Fangrui Song via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Sat Feb 26 01:37:36 PST 2022


MaskRay added a comment.

In D120305#3347193 <https://reviews.llvm.org/D120305#3347193>, @nikic wrote:

> In D120305#3347192 <https://reviews.llvm.org/D120305#3347192>, @tstellar wrote:
>
>> In D120305#3347177 <https://reviews.llvm.org/D120305#3347177>, @nikic wrote:
>>
>>> Yes, because you reverted the change for that one buildbot, of course it is green now. You could have also made the buildbot green by disabling tests on that bot. Or disabling sanitizers on it. Doesn't change the fact that the configuration it was originally testing is still broken, you just hid the failure.
>>
>> The build is green because of this commit: https://github.com/llvm/llvm-project/commit/274ec425dcc3e3f637dd006c5e9ae33bd0e2e917  The buildbot change you are referring to, which is https://github.com/llvm/llvm-zorg/commit/b6ddf02ce3a54da2df29e7e599b1838167e0e3ad, has not taken effect yet, because the buildbot server has not been restarted.
>
> Wow, that's even worse. So now it's not just a change to that one buildbot, but sanitizer tests were disabled for powerpc entirely?!

Only few sanitizer_common and lsan tests, not entirely. It should be re-enabled pretty soon once the llvm-zorg change is made live.
That was I mentioned that "we can enable the tests".

I think at this point, if you prefer, I am happy to revert this change the disabling (it needs quite a bit of tests), so that we can know whether -fno-pic and -fpie have a large difference on sqlite3 performance.

I lost my control when I saw https://reviews.llvm.org/D120305#3347058 and Tom's first reply to it.
For quite few hours yesterday, I did not know my fixed did not fix the problem or that nobody tried making llvm-zorg change work.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D120305/new/

https://reviews.llvm.org/D120305



More information about the cfe-commits mailing list