Always leverage fast unwinding to obtain the native stack #141
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
pablogsal
force-pushed
the
better_inlining
branch
from
8066922
to
98ad428
Compare
pablogsal
force-pushed
the
better_inlining
branch
from
98ad428
to
c550f53
Compare
godlygeek
previously approved these changes
Jun 13, 2022
|
Also - does this deserve a changelog entry? This fixes a bug where aggregating call stacks could give misleading results on aarch64 under some circumstances, right? |
pablogsal
force-pushed
the
better_inlining
branch
from
c550f53
to
208bb91
Compare
Yes and not, is not mostly user-visible because the actual stack traces were the same (they will produce weird aggregations but not incorrect ones), is just that the test suite fails because the native_frame_ids are different although they produce the same results. I will add a news entry nevertheless |
pablogsal
force-pushed
the
better_inlining
branch
2 times, most recently
from
55d7d84
to
db90c91
Compare
|
@godlygeek Can you please review again? |
Seems that libunwind just chokes if the function that calls unw_get_reg is not inlined in aarch64. Additionally, there are some problems with inlined functions that call libunwind's APIs in powerpc and other architectures. To avoid all this trouble, leverage just the fast unwinding APIs, which will make the program counters more consistent and will eliminate the inline shenanigans. Additionally increase the default size for the native stack vector to reduce the need to call multiple times into the unwinding API. Signed-off-by: Pablo Galindo <pablogsal@gmail.com>
pablogsal
force-pushed
the
better_inlining
branch
from
db90c91
to
19e55f0
Compare
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.
Seems that libunwind just chokes if the function that calls unw_get_reg
is not inlined in aarch64. Additionally, there are some problems with
inlined functions that call libunwind's APIs in powerpc and other
architectures.
To avoid all this trouble, leverage just the fast unwinding APIs, which
will make the program counters more consistent and will eliminate the
inline shenanigans.
Closes: #126