-
Notifications
You must be signed in to change notification settings - Fork 243
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
perfparser raises SIGABRT in PerfUnwind::unwindStack #431
Comments
this report is not actionable without a reliable way to reproduce this. Furthermore, the crash happens within elfutils. What assertion do you see? With that you might be able to report this upstream to the elfutils developers. |
Not sure if this is an elfutils problem or an issue with how elfutils is used ;-)
Can you please add a non-stripped elfutils to the AppImage and specify the version used? This would allow me to check what actually is happening. |
ping @milianw on the AppImage - could you create that with non-stripped elfutils? |
I don't have time for that in the next 1-2 weeks I'm afraid @GitMensch - but it should be pretty easy for you to hack the docker file. Ideally we would extract the debug symbols for all things we put into the appimage and then upload that separately as an artifact. That way we can then manually debug issues. Patches welcome! |
docker and GH actions are both not in my skill stack, will wait for 2-8 weeks ;-) |
I got some time for this after all today and set everything up: https://github.com/KDAB/hotspot#debugging-the-appimage Note that for elfutils debug symbols itself, you won't need most of this - just use the official centos debuginfod, that should be enough to get you debug symbols for that. |
Hm, shouldn't the docker package be available via https://github.com/orgs/KDAB/packages?tab=packages&q=kdesrc, too? I don't have ghcr.io setup... ... tested without that - just to find that the GDB instances are either too old or:
Installation of an updated GDB likely will take a while :-/ I guess I could download the necessary debuginfo manually, but no clue how to do that. Side note: You may want to add From inspecting the current source (which elfutils is contained in the appimage?) there are only two asserts... https://github.com/evverx/elfutils/blob/0bf61bc17ebf002dbf58fc5540bfb22343e7b354/libdwfl/frame_unwind.c#L499-L520 |
Are you sure? The appimage uses centos7 (I think only centos8+ has debuginfod) and also elfutils seems to be statically linked in the appimage, not sure if debuginfod can help here for self-built binaries. Also see #560. |
please create a new issue instead of commenting on old ones |
This isn't a new issue and the "old one" is still open. This issue is about a SIGABRT in PerfUnwind, as seen in the AppImage. Also you've told me to use the debuginfod option to get the missing symols - but I question if there is a debuginfod for the elfutils used for the appimage - as that is CentOS7 - and also that it would be used because elfutils are statically linked in (both could be me confusing things). Don't you agree that this issue is the right to comment about those to issues? What would a new issue about that look like ?!? |
Describe the bug
SIGABRT in https://github.com/KDAB/perfparser/blob/b455d4092aef0861b52a75bb818e34af6cb6ee32/app/perfunwind.cpp#L548-L574
To Reproduce
Steps to reproduce the behavior:
sudo perf record -o perf.data.6-50 --call-graph dwarf,50 -a --uid somuser -- sleep 30m
, which was aborted by CTRL+C because of size issues (no max size option on this older perf)Additional info
perf report -i
Expected behavior
perf.data which can be opened by
perf report
should be able to be opened byhotspot-perfparser
Check if it is possible to circumvent this specific error.
Improved error message in the GUI, when possible
-->The question: "What was invalid" would need more communication from perfparser to hotspot than a signal.
GDB info
The stacktrace is the same in all cases, so I did not check the objects on any abort but the first).
Version Info (please complete the following information):
3.10.0-1160.31.1.el7.x86_64
3.10.0-1160.el7.x86_64.debug
hotspot-v1.3.0-364-g322803e-x86_64.AppImage
The text was updated successfully, but these errors were encountered: