[AVR] Fix for issue #163015, occasional corruption in stack passed params #164001
Conversation
|
Closes #163015 |
…register pressure. See this forum discussion for details (or the GitHub issue) https://discourse.llvm.org/t/avr-register-allocation-issue-help-advice-with-debugging/88498/11 @aykevl @benshi001
carlos4242
force-pushed
the
fix-163082
branch
from
64b7aad to
aa54b9a
Compare
There was a problem hiding this comment.
please use update_llc_test_checks.py to generate CHECK lines. Also, is there any reason you need -v -dump-input always?
There was a problem hiding this comment.
I think it is not needed to run update_llc_test_checks.py, since the key check point in this test are the CHECK-NOT lines.
There was a problem hiding this comment.
It is up to reviewer to mark resolved/unresolved.
@mshockwave Do you have any more concerns for that?
|
previous PR that I accidentally closed: #163082 |
|
Hi @mshockwave @benshi001 are we anywhere near getting this signed off and merged? It's not a very big PR and it would be a shame if it just ends up in limbo. I've got one or two more AVR PRs I wanted to try and get in after this one, before I step back from AVR. C |
Though I am the maintainer of the AVR part, I have to respect other reviewer's suggestion, if @mshockwave didn't response by Friday, I will merge it. @mshockwave what's your opinion? |
3 participants
Corruption can occur with passing parameters on the stack when under register pressure.
See this forum discussion for details (or the GitHub issue) https://discourse.llvm.org/t/avr-register-allocation-issue-help-advice-with-debugging/88498/11
@aykevl @benshi001