[PATCH][LoopStrengthReduce] Don't trim a formula that uses a subset of required registers

[Adam Nemet]

On Apr 28, 2014, at 5:05 PM, Chandler Carruth <chandlerc at google.com> wrote:

> I don't object to having the fix checked in but:
> 
> On Mon, Apr 28, 2014 at 4:51 PM, Andrew Trick <atrick at apple.com> wrote:
> one of the major difficulties in getting it fixed was we just have absolutely no way to force it in llc
> 
> This is *exactly* why we should absolutely spend time isolating what it is about this test case that doesn't reproduce with llc. This is *critical* to the LLVM test infrastructure and project. It's not something we can or should paper over, for exactly the reason you described: it makes finding these bugs nearly impossible.

OK, so as I understand this is the plan:

* file a PR on the original LSR problem with the C test case
* check in the fix again (without the test case), referencing the PR
* keep the PR open at this point until we figure out a way to have an LLVM IR test case for this
* remove clang from the substitutions in lit.cfg so that the next person won’t get confused like I got

Sounds good?

Adam

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140428/b5853cc0/attachment.html>