Fix misunwind during Profile test under rr #39553
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Actually, I was able to reproduce the original problem in a docker container, and this does seem to fix it, so I'll go ahead and go work on the Yggdrasil bump. |
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 6, 2021
Keno
force-pushed
the
kf/unwindcfarsp
branch
2 times, most recently
from
44aa9ef to
0fef25f
Compare
|
Updated the patch. Hopefully that should fix the backtrace quality regression of the previous patch while keeping the Profile fix. |
Keno
force-pushed
the
kf/unwindcfarsp
branch
2 times, most recently
from
a7ce2de to
b2ad240
Compare
|
I'm gonna go with this method of CI'ing doesn't work. Works fine locally with the patch applied as well as with a BB-built version of the binary, but not with the one from the buildbot - I assume the corruption has something to do with it. I'll just bump BB and let's try that. |
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 8, 2021
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 8, 2021
Keno
changed the title
|
Freebsd issue turned out to be the fallback unwinder (I guess some library on freebsd doesn't have DWARF info), which needs to be updated to this change. Aarch64 issue, I didn't debug, but I'll just drop the non-x86_64 changes here since we don't (currently) run under rr on aarch64. |
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 8, 2021
Also drop changes for other architectures (c.f. JuliaLang/julia#39553).
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 8, 2021
Also drop changes for other architectures (c.f. JuliaLang/julia#39553).
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 8, 2021
Also drop changes for other architectures (c.f. JuliaLang/julia#39553).
Keno
added a commit
to JuliaPackaging/Yggdrasil
that referenced
this pull request
Feb 8, 2021
Also drop changes for other architectures (c.f. JuliaLang/julia#39553).
Keno
added a commit
that referenced
this pull request
Feb 9, 2021
I noticed in #39553 that JIT unwind info didn't actually work on FreeBSD. As a general policy, we should register our frame info to make sure that our unwind is accurate. Libunwind has some heuristics to fall back to, but they can break e.g. for large stack frames.
Libunwind improperly aliases RSP and CFA, which are separate concepts. Fix that.
vtjnash
pushed a commit
that referenced
this pull request
Feb 9, 2021
I noticed in #39553 that JIT unwind info didn't actually work on FreeBSD. As a general policy, we should register our frame info to make sure that our unwind is accurate. Libunwind has some heuristics to fall back to, but they can break e.g. for large stack frames.
|
This is broken on i686. Looks like the failures appeared on earlier runs of the PR, but, awkwardly, tester_linux32 never ran on the final update, so it has 17 green marks without showing the 1 red. |
KristofferC
pushed a commit
that referenced
this pull request
Feb 17, 2021
Libunwind improperly aliases RSP and CFA, which are separate concepts. Fix that. (cherry picked from commit bbf7f97)
ElOceanografo
pushed a commit
to ElOceanografo/julia
that referenced
this pull request
May 4, 2021
Libunwind improperly aliases RSP and CFA, which are separate concepts. Fix that.
ElOceanografo
pushed a commit
to ElOceanografo/julia
that referenced
this pull request
May 4, 2021
I noticed in JuliaLang#39553 that JIT unwind info didn't actually work on FreeBSD. As a general policy, we should register our frame info to make sure that our unwind is accurate. Libunwind has some heuristics to fall back to, but they can break e.g. for large stack frames.
antoine-levitt
pushed a commit
to antoine-levitt/julia
that referenced
this pull request
May 9, 2021
Libunwind improperly aliases RSP and CFA, which are separate concepts. Fix that.
antoine-levitt
pushed a commit
to antoine-levitt/julia
that referenced
this pull request
May 9, 2021
I noticed in JuliaLang#39553 that JIT unwind info didn't actually work on FreeBSD. As a general policy, we should register our frame info to make sure that our unwind is accurate. Libunwind has some heuristics to fall back to, but they can break e.g. for large stack frames.
staticfloat
pushed a commit
that referenced
this pull request
Dec 23, 2022
Libunwind improperly aliases RSP and CFA, which are separate concepts. Fix that. (cherry picked from commit bbf7f97)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
See the included patch for details. In short, libunwind has a bug that causes Profile test failures. Unfortunately, since I'm not seeing the issue locally, I cannot say whether this actually fixes it, so I added a second commit that just builds libunwind locally on the buildbot and hopefully if we run this a couple of times, we'll find out if it works or not. If it does, we can submit this upstream and spin a Yggdrasil release.