-
Notifications
You must be signed in to change notification settings - Fork 728
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
HCRLateAttachWorkload_previewEnabled_1_FAILED - Segmentation error vmState=0x0002000f #13504
Comments
Bad O-slot is reported:
this O-slot points to stack allocated object
Unfortunately my |
@fengxue-IS Can you please take a look at this |
Tagging vm as well since vm is taking the first look. |
This failure is highly intermittent, reproduced with a recent nightly build in grinder (1/50).
Will look into the generated cores to see what caused the crash |
Just fyi 1/50 is not a blocker, and this isn't marked as a blocker, we could move it out. |
I haven't had any luck reproducing this failure with either the original build or a more recent build. Grinder runs on each showed 0/50 failures. |
An occurrence at JDK17
|
Moved to #13499 (comment) |
@babsingh Can you investigate this issue, Jack is busy with another one |
@JasonFengJ9 Is the This issue primarily covers the |
Yeah, #13504 (comment) (and an earlier #13504 (comment)) are similar w/ #13499 (comment) instead, in which Please focus on |
GrindersPersonal grinder 1: JIT enabled. 37 failures out of 1950 tests. Link to core files. Personal grinder 2: Related Test Code
Stackslots
ProblemThe two O-slots for Looking at
I am re-running the grinder with |
@llxia I ran a grinder (Link to Grinder #19186) for @fengxue-IS The stack corruption seen in this issue may be related to the stack corruption seen in #13162. |
As the failure rate seems to have increased, I've moved this out of the backlog and into the 0.41 milestone. Previously it was claimed to be highly intermittent. @hzongaro fyi |
I took a look at the core from build 141 (
The btyecode frame of [1]
[2]
[3]
[4]
|
The failure is highly intermittent. The test running in multiple threads adds random method calls to the public methods in
The invalid O-Slot value [1]
[2]
[3]
|
Yup, thanks. |
After adding JIT options to try to capture some compilation logs [1], I reproduced the issue once (FR: about 1/400) with all interpreted methods on the stack frames [2]. @babsingh So far I'm not sure how the object could be corrupted by JIT'd methods since there is no JIT'd method on the stack. Could this problem be investigated again from the VM's perspective? Thanks! [1]
[2]
|
@a7ehuo Can you share your core files (e.g. via Box)? |
The logs are updated to a Box folder and have been shared offline. Please check. Thank you! |
Although Source code
Details from DDR
|
@gacholio Is |
Yes, that will dump the stack at various points throughout the decompilation process. |
JDK11 s390x_linux(
50x grinder - 2/50 failed |
JDK17 s390x_linux(
50x grinder - 1/50 failed |
Thanks @babsingh for narrowing down the issue! This one is on my TODO list. I'll get to it as soon as possible.
Interesting! Thanks for pointing it out @hzongaro ! I'll take a look along with the information from @babsingh |
Moving this to 0.44 while @a7ehuo continues to investigate. |
After many runs, I reproduced the issue once with an old build from September with OSR and EA traces. With help from @hzongaro, we concluded this is the same issue that has been solved by PR #14074.
Below is the detailed analysis from the reproduced case: The VM trace shows the PC is The invalid object
Henry found [1]
[2]
[3]
[4]
[5]
[6]
[7]
|
Closing as duplicate. |
Failure link
From an internal build
job/Test_openjdk17_j9_extended.system_s390x_linux_Personal/12/tapResults/
(rhel7s390x-3-1
):Rerun in Grinder
Optional info
Failure output (captured from console output)
fyi @dmitripivkine
The text was updated successfully, but these errors were encountered: