PHP 8.2.5 - 8.2.7 crash #11357
Labels
Comments
|
It looks like a heap corruption, so the origin of the actual problem could be anywhere I'm afraid. Some questions to begin with:
|
|
1 - yes 30-50% requests will crash this is prod server and i can't run this version again without fix. You can see screenshot and log from dnf
2023-06-02T00:07:01+0300 DEBUG ---> Package php82-fpm.x86_64 8.2.1-1 will be upgraded |
|
If you can give a way so we can reproduce the issue on our own systems, that would be great. Otherwise it's going to be difficult to debug this. Alternatively if you are able to trigger the issue with a custom build in a test environment, this would allow you to get more info. If that's possible, then you can configure your PHP build with the Alternatively, performing a bisect between the commits of 8.2.1 and 8.2.5 would also work I guess. Of course I understand if this is not really possible in your situation. The problem right now is that there's too little information to figure out what goes wrong I'm afraid... |
|
No feedback was provided. The issue is being suspended because we assume that you are no longer experiencing the problem. If this is not the case and you are able to provide the information that was requested earlier, please do so. Thank you. |
|
Crash still exist, please don't close ticket |
|
the crash was observed on the prod, I'm looking for options to reproduce it in the dev environment |
github-actions
bot
added
Status: Needs Triage
and removed
Status: Needs Feedback
labels
|
I made debug build 8.2.7 for dev environment without debug - 502 / core / sigsegv phperror_log - empty php-fpm.log has only this: |
github-actions
bot
added
Status: Needs Triage
and removed
Status: Needs Feedback
labels
|
non-debug crash bt |
|
bt from gdb & debug build |
|
@sitnikov Thank you! Can you try compiling with |
|
address-sanitizer enabled. 
(gdb) attach 1670293 http - 502 |
|
What else can I do to help? |
|
I'm surprised there's not output, afaict there should've been ASAN (Address SANitizer) output. @iluuu1994 any idea why there's no output? |
|
Unfortunately not no. I yet have to encounter a false negative in ASAN. Maybe the ASAN output is somewhere else? You could try setting the env variable |
|
|
I've seen crashes like that hitting that assertion, it's usually incorrect refcounting somewhere such that the refcount drops to 0 too early. This is going to be difficult to find the root cause and fix it without having a reproducer at hand. |
|
The codebase is very large and the business logic is complex. I have no idea how to extract a sufficient piece of code to reproduce the problem. |
|
As we know, version 8.2.1 works, 8.2.5 does not. If I check all the versions in between and we find the one where it broke, will that help in any way? |
|
Yes that would help a lot. |
|
If the first breaking version is 8.2.4 then I think it could be that one ext/date issue that I fixed recently (fix should be in 8.2.8). The behaviour is the same and the stacktrace looks similar with the same assert being hit. It's the only refcounting issue I'm aware of between 8.2.1 and 8.2.5. |
|
8.2.3 - OK |
|
Thanks for your efforts btw! |
|
Thanks! So this issue is actually a duplicate of #11455 . The fix for this will be included in the 8.2.8 release. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Description
I only have a core from version 8.2.5, but version 8.2.6 also crashed. I can't replicate the issue for 8.2.6 because this is a production system. Version 8.2.1 is working properly
(gdb) bt
(gdb) zbacktrace
DiscountCalculator.php:155
The text was updated successfully, but these errors were encountered: