Skip to content

Add missing Q type checks - #16411

Merged
tajila merged 1 commit into
eclipse-openj9:masterfrom
ehrenjulzert:missing_q_type_all_files_final
Feb 8, 2023
Merged

tajila merged 1 commit into
eclipse-openj9:masterfrom
ehrenjulzert:missing_q_type_all_files_final

Conversation

@ehrenjulzert

Copy link
Copy Markdown

Add a Q type check in all locations in which a
Q type should be handled the same way as an L type

For issue #15861

Signed-off-by: Ehren Julien-Neitzert ehren.julien-neitzert@ibm.com

@ehrenjulzert

Copy link
Copy Markdown
Author

@hangshao0

Comment thread runtime/rastrace/method_trace.c Resolved
@ehrenjulzert
ehrenjulzert force-pushed the missing_q_type_all_files_final branch from 95d591c to f180a75 Compare December 19, 2022 15:24
@hangshao0
hangshao0 requested a review from tajila December 19, 2022 18:12
@hangshao0

Copy link
Copy Markdown
Contributor

There are DDR and minor GC changes, adding @keithc-ca @dmitripivkine as reviewers.

@dmitripivkine dmitripivkine left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

GC related code (couple of files) looks ok to me. There is one remark left for commented out lines - I would prefer to not have them in final commit but not strong on this (what purpose these lines suppose to serve?)

Comment thread runtime/bcutil/ClassFileWriter.cpp Resolved
Comment thread runtime/jnichk/jnicheck.c Resolved
@ehrenjulzert
ehrenjulzert force-pushed the missing_q_type_all_files_final branch 2 times, most recently from 3478e09 to 9987875 Compare January 9, 2023 16:47
Comment thread runtime/bcverify/bcverify.c Resolved
Comment thread runtime/jnichk/jnicheck.c Resolved
Comment thread runtime/verbose/errormessagehelper.c Outdated

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

While you're here, please fix the preprocessor lines 654 & 658: # should be at the beginning of the line and #if should be omitted from the #endif comment.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I've noticed there are a lot more places in the codebase that have these inconsistencies in both C and Java preprocessor comments (most of which are my fault). I could fix all of them in a separate PR if you think it's worthwhile? Alternatively I could fix them in this PR (would only be about 40 more lines changed but it would touch files that haven't been changed yet in this PR).

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it's reasonable to fix those other files in this pull request.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I fixed all the existing inconsistent C preprocessor comments in 6d7788e, but not all the Java ones because there turned out to be way more than I thought (around 150 instances). Is that still reasonable for this PR do you think?

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thoughts on this @keithc-ca?

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm happy to defer a cleanup of those Java source files.

Comment thread runtime/vm/bindnatv.cpp Resolved
Comment thread runtime/vm/extendedMessageNPE.cpp Resolved
@keithc-ca

Copy link
Copy Markdown
Contributor

Please also correct the typo in the commit message: "location" should be "locations" and sentences should end with a period.

@ehrenjulzert
ehrenjulzert force-pushed the missing_q_type_all_files_final branch 3 times, most recently from 7678ed1 to 6d7788e Compare January 10, 2023 20:37
@ehrenjulzert
ehrenjulzert force-pushed the missing_q_type_all_files_final branch 2 times, most recently from c1370a2 to 89dde4a Compare January 11, 2023 19:29
@tajila

tajila commented Jan 12, 2023

Copy link
Copy Markdown
Contributor

Jenkins test sanity,extended xlinuxval jdknext

@tajila

tajila commented Jan 12, 2023

Copy link
Copy Markdown
Contributor

Jenkins test sanity plinuxvalst jdknext

@tajila

tajila commented Jan 12, 2023

Copy link
Copy Markdown
Contributor

Jenkins test sanity win jdk8

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I found the source of the build failure (I used sigOffset here by accident instead of index), and when I tested this code earlier I didn't test it on JDK8. But I'm not sure why this was only an error on JDK8 builds?

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The same error appears on line 465 as well.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

fixed in e51ed56

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The same error appears on line 465 as well.

- Add a Q type check in all locations in which a
Q type should be handled the same way as an L type

- Change comments after #else and #endif directives
to be consistent

For issue eclipse-openj9#15861

Signed-off-by: Ehren Julien-Neitzert <ehren.julien-neitzert@ibm.com>
@ehrenjulzert
ehrenjulzert force-pushed the missing_q_type_all_files_final branch from 89dde4a to e51ed56 Compare January 17, 2023 21:54
@tajila

tajila commented Jan 19, 2023

Copy link
Copy Markdown
Contributor

@hzongaro Do you know why this PR would affect the inlineMap?

@tajila

tajila commented Jan 19, 2023

Copy link
Copy Markdown
Contributor

@ehrenjulzert can you see how reproduceable that test failure is?

@JasonFengJ9

Copy link
Copy Markdown
Member

In case the error in question is not related to this PR, @ehrenjulzert please create an issue so we can track it separately.

@ehrenjulzert

Copy link
Copy Markdown
Author

Ok, I'm rerunning it in test grinder now.

@ehrenjulzert

ehrenjulzert commented Jan 19, 2023 •

Copy link
Copy Markdown
Author

I re-ran the test on grinder but it failed again (https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30501/). I also tried to find an available nightly s390x_linux_vt_standard build of the master branch to test but couldn't find any. Am I just supposed to use the regular s390x_linux platform (eg this nightly build: https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Build_JDKnext_s390x_linux_Nightly/)?

@JasonFengJ9

Copy link
Copy Markdown
Member

There are no regular s390x_linux_vt_standard nightly builds.
Can the failure be reproduced with last PR build?

@tajila

tajila commented Jan 20, 2023

Copy link
Copy Markdown
Contributor

Am I just supposed to use the regular s390x_linux platform (eg this nightly build: https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Build_JDKnext_s390x_linux_Nightly/)?

No youll need a VT build

@tajila

tajila commented Jan 20, 2023

Copy link
Copy Markdown
Contributor

@ehrenjulzert you can make a VT build without your changes with the s390x_linux_valhalla SPEC and see if that fails

@ehrenjulzert

Copy link
Copy Markdown
Author

Can the failure be reproduced with last PR build?

It looks like the last PR build doesn't exist anymore, I assume because we don't run s390x_linux_vt_standard very often (unless I'm misunderstanding how to use artifactory)

@ehrenjulzert

Copy link
Copy Markdown
Author

@ehrenjulzert you can make a VT build without your changes with the s390x_linux_valhalla SPEC and see if that fails

Sure I can run that, although to be clear: is s390x_linux_valhalla what we want? Don't we want s390x_linux_vt_standard?

@JasonFengJ9

Copy link
Copy Markdown
Member

It looks like the last PR build doesn't exist anymore, I assume because we don't run s390x_linux_vt_standard very often (unless I'm misunderstanding how to use artifactory)

Yeah, last PR build has been purged, just launched one w/ default branches - https://openj9-jenkins.osuosl.org/job/Build_JDKnext_s390x_linux_vt_standard_Personal/11

@tajila

tajila commented Jan 20, 2023 •

Copy link
Copy Markdown
Contributor

Sure I can run that, although to be clear: is s390x_linux_valhalla what we want? Don't we want s390x_linux_vt_standard?

s390x_linux_vt_standard is the test spec, s390x_linux_valhalla is the JDK build spec. So you need to make a build without your changes, then run the failing test against it.

I guess you could also rerun the s390x_linux_vt_standard test spec with an openj9 SHA without your changes.

EDIT: nvm I see Jason has launched it

@ehrenjulzert

Copy link
Copy Markdown
Author

Yeah, last PR build has been purged, just launched one w/ default branches - https://openj9-jenkins.osuosl.org/job/Build_JDKnext_s390x_linux_vt_standard_Personal/11

Thanks Jason! Sorry about all the confusion.

@ehrenjulzert

ehrenjulzert commented Jan 20, 2023 •

Copy link
Copy Markdown
Author

I ran the tests and they passed this time (https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30541/), which surprises me since my decomp.cpp changes look equivalent to what was there previously. Looking more into it now

@tajila

tajila commented Jan 24, 2023

Copy link
Copy Markdown
Contributor

Can you get a core file when the error occurs? You can look at https://www.ibm.com/docs/en/sdk-java-technology/8?topic=options-xtrace#trigger to see how you can trigger a core on that tracepoint (tracepoint IDs can be found in jdk/lib/J9TraceFormat.dat

@hzongaro

Copy link
Copy Markdown
Member

I ran the tests and they passed this time. . . which surprises me

Ehren @ehrenjulzert, FYI - I was able to reproduce the failure in a Grinder run of a build without this change, https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30530/ - assuming I didn't mess up my own build somehow.

@ehrenjulzert

Copy link
Copy Markdown
Author

I was able to reproduce the failure in a Grinder run of a build without this change, https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30530/

This seems correct to me, looks like the test only sometimes fails. That would be a bit of a coincidence since I ran the test twice with my changes and the error was reproduced each time, and I also ran the test twice without my changes and it passed both times.

I'm running multiple iterations of the same test on my branch here to see if it sometimes passes: https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30642/

@tajila

tajila commented Jan 31, 2023

Copy link
Copy Markdown
Contributor

@ehrenjulzert Have you concluded that the failure predates your changes?

@ehrenjulzert

Copy link
Copy Markdown
Author

Not yet, I'm unsure how to interpret the different results that Henry and I got. Assuming we both ran our Jenkins tests correctly the results indicate that the test sometimes fails on master but always fails on my branch. I was thinking that it may have to do with the machine that my tests were running on, but didn’t get a chance to look into it last week. I was planning on running more tests some time today.

@tajila

tajila commented Jan 31, 2023

Copy link
Copy Markdown
Contributor

Jenkins test sanity zlinuxvalst jdknext

@tajila

tajila commented Jan 31, 2023

Copy link
Copy Markdown
Contributor

@ehrenjulzert can you see if that test passes without the changes in decomp.c

@ehrenjulzert

ehrenjulzert commented Feb 1, 2023 •

Copy link
Copy Markdown
Author

@ehrenjulzert can you see if that test passes without the changes in decomp.c

Sure, I'll try that now

@ehrenjulzert

Copy link
Copy Markdown
Author

Yes, there's still intermittent failures without the decomp.cpp changes (https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30845/)

@ehrenjulzert

Copy link
Copy Markdown
Author

Oops I didn't notice until now that cmdLineTester_decompilationTests_0 is also failing in the same way as cmdLineTester_decompilationTests_1 (same assertion error on the same line). It is also failing on master though (see https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/30893/), so not a problem with my changes.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

8 participants