There was a long discussion on cfe-dev about this issue approximately January-February 2020 including @scanon @andrew.w.kaylor and many others. While there wasn't 100% consensus to move to ffp-contract=on there are many reasons to do it including better numerical results and debuggability. I'm hoping that @andrew.w.kaylor @kbsmith1 @rjmccall and others will join in here.

In D74436#2917357, @haowei wrote:

We are seeing float related unit tests failing in Fuchsia after this patch. What flags should we add to our build to match llvm's previous float point behavior?

@MaskRay Thanks a lot -- yes!

commit 79f2b03c5111392d647fa542e120f70c8f39a991
Author: Zahira Ammarguellat <zahira.ammarguellat@intel.com>
Date: Mon Jul 26 11:36:23 2021 -0700

Adding @zahiraam FYI

Thanks for fixing this Aaron, just curious was there a bug report?

This was a DUP to trigger pre-commit testing. It didn't work. Abandon

I tested this patch downstream on Windows with Microsoft (R) C/C++ Optimizing Compiler Version 19.28.29334 for x64

LGTM - please see changes requested inline. thanks a lot

Whoops, shouldn't mess with the mkdir line

@bnbarham
I think you mean a patch like this, is it right? I'd like to fix the test in a pre-commit and then re-push my patch

@MaskRay I experience that the failing test clang/test/Index/preamble-reparse-changed-module.m is non-deterministic and currently I cannot reproduce the fail. I added details in D95159. Are you able to reproduce if you clean your build area? Do you have a suggestion for how I can solve this problem? Thanks a lot

More info about the test failure I'm seeing for clang/test/Index/preamble-reparse-changed-module.m

I had to revert again, it's still failing on Intel buildbots. Cannot reproduce on Intel-internal Broadwell server.

@aaron.ballman suggested I may have made a mistake adding the new option, I'll look there.

The newly added test case clang/test/Index/preamble-reparse-changed-module.m is failing on the patch that I'm trying to commit, https://reviews.llvm.org/D93769

In D93769#2891552, @MaskRay wrote:

In D93769#2891549, @mibintc wrote:

In D93769#2891545, @MaskRay wrote:

This breaks clang/test/Index/preamble-reparse-changed-module.m

Seems like a libclang crash (an assert failure) if you enable LLVM_ENABLE_ASSERTIONS=on.

You can reproduce with check-clang-index

Thanks for this, I'm building with assertions on now. This patch doesn't expand FLT_EVAL_METHOD in -E mode, I'm guessing that's why it fails. It can't expand the macro during -E because the context showing the value of the macro setting is only available in Sema. I haven't yet studied the test but do you know have an idea how I might be able to solve the problem?

In D93769#2891545, @MaskRay wrote:

This breaks clang/test/Index/preamble-reparse-changed-module.m

http://45.33.8.238/linux/51638/step_7.txt

Removed use of auto, and used different capitalization for local var name

Question for Aaron

Partially respond to @aaron.ballman 's review by refactoring a change into a separate commit, but I'll push back on another request, I'll add that reply inline.

LGTM

I've rebased and applied clang-format. I'd like to push this, looking for your +1, thank you!

Respond to @aaron.ballman 's review, rebased & used clang-format

Intel compiles VS2019 #include files regularly with clang, and the file <filesystem> compiled with -std:c++latest encounters this error report, which @aaron.ballman suggests is related to this effort.

More buildbot failures

buildbot fails in lldb, will submit updates

buildbot fails on lldb need to update this patch

Rebase

Rebasing. Hope this clears lit fails.

some inline replies. I think this is all set now.

Respond to @aaron.ballman 's review

A couple inline replies to go along with the updated patch

The patch I uploaded for review yesterday wasn't correct, not sure what happened. This one looks better.

This patch addresses almost all the review comments, not yet sure about @aaron.ballman 's question about CoreOptions

I’ve been trying to commit this patch. However, when I commit, the tests fail some 21 floating point tests in LNT. Also there is both performance improvement as well as regression which shouldn’t happen because the sum total of my clang patch + my LNT patch should effectively be a no-op.

Reply to @joerg proposing new wording for the option description

In D102861#2819675, @rengolin wrote:

In D102861#2817795, @rengolin wrote:

I can resurrect some Armv7/v8 machines at home and try

Turns out my boards are too old and need some love that I'm unable to give them.

Thanks for looking

If you can't find anyone that can help you, an alternative is to merge and lookout for breakages, and if you only see one of two, (revert and) liaise with the bot owners to fix them.

I plan to try pushing again when I have a block of time to babysit the bots. Is the place I should watch on Green Dragon? e.g. https://green.lab.llvm.org/green/job/test-suite-verify-machineinstrs-x86_64h-O3/
I can see there's aarch64 and x86 but what about other arch?

Fixed an initialization bug in 2mm.c

The previous revision had a compilation error in 2mm.c I've corrected that.