Fix continuation stacks when breakpointing #18413

gacholio · 2023-11-07T15:38:05Z

The breakpoint handling code was only fixing stacks actively running on
a thread. The stacks for continuations also need to be fixed.

Fixes: #18088

Signed-off-by: Graham Chapman graham_chapman@ca.ibm.com

gacholio · 2023-11-09T05:52:38Z

@babsingh I'm looking into the JVMTI test failures. Please do an initial review of my use of the continuations.

babsingh

Is exclusive VM access acquired for all j9mm_iterate_all_continuation_objects calls?
Should j9gc_flush_nonAllocationCaches_for_walk and j9mm_iterate_all_continuation_objects be invoked together?
Need to pass the correct threadObject to walkContinuationStackFrames.

runtime/jvmti/jvmtiHelpers.c

runtime/codert_vm/decomp.cpp

babsingh · 2023-11-09T15:16:38Z

++ @fengxue-IS for a second set of eyes.

fengxue-IS

code lgtm, few minor format picks

runtime/jvmti/jvmtiHelpers.c

runtime/codert_vm/decomp.cpp

gacholio · 2023-11-09T21:36:39Z

I found more places that need the flush, so I've helperized acquiring exclusive in JVMTI (some of the callers don't strictly need the flush, but it's a tiny amount of time compared to acquiring exclusive).

gacholio · 2023-11-10T06:56:51Z

On reflection, I don't like the exclusive/flush solution. I think @babsingh initial idea is best - do the flush before the iterate. I'll update tomorrow.

gacholio · 2023-11-10T12:36:16Z

@dmitripivkine Please confirm it's OK to call j9gc_flush_nonAllocationCaches_for_walk multiple times in the same exclusive VM access.

dmitripivkine · 2023-11-10T16:43:27Z

@dmitripivkine Please confirm it's OK to call j9gc_flush_nonAllocationCaches_for_walk multiple times in the same exclusive VM access.

Yes, I believe so. However it might be not optimum because once threads local buffers are flushed nothing should be added under the same exclusive.

babsingh

accidentally approved while juggling through multiple PRs.

Dismissing

gacholio · 2023-11-13T11:45:05Z

@babsingh Ready for final review.

babsingh

minor formatting nits; otherwise, lgtm.

runtime/codert_vm/decomp.cpp

The breakpoint handling code was only fixing stacks actively running on a thread. The stacks for continuations also need to be fixed. Fixes: eclipse-openj9#18088 Signed-off-by: Graham Chapman <graham_chapman@ca.ibm.com>

babsingh · 2023-11-13T20:30:46Z

jenkins test sanity zlinux jdk8,jdk21

babsingh · 2023-11-13T20:31:00Z

jenkins compile win jdk8,jdk21

babsingh

https://openj9-jenkins.osuosl.org/job/Test_openjdk21_j9_sanity.functional_s390x_linux_Personal_testList_1/57/consoleFull

Only a known failure is seen in the PR builds, which is unrelated to this PR: cmdLineTester_criu_nonPortableRestore_4_FAILED FAILED: testMXBeanUpTime() - uptimeAfterCheckpoint 9437 can't be greater than uptimeBeforeCheckpoint 7751 + 1500 #18384

gacholio changed the title ~~Fix continutation stacks when breakpointing~~ Fix continuation stacks when breakpointing Nov 7, 2023

gacholio force-pushed the break branch 5 times, most recently from c76af27 to e5c2ee9 Compare November 8, 2023 21:39

gacholio requested a review from babsingh November 9, 2023 05:51

babsingh reviewed Nov 9, 2023

View reviewed changes

runtime/jvmti/jvmtiHelpers.c Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Show resolved Hide resolved

babsingh requested a review from fengxue-IS November 9, 2023 15:17

gacholio force-pushed the break branch 2 times, most recently from aa66bbe to 88cbd28 Compare November 9, 2023 16:01

fengxue-IS reviewed Nov 9, 2023

View reviewed changes

runtime/jvmti/jvmtiHelpers.c Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

gacholio force-pushed the break branch 2 times, most recently from 7ac5535 to 2c3dddc Compare November 9, 2023 16:55

babsingh reviewed Nov 9, 2023

View reviewed changes

runtime/codert_vm/decomp.cpp Show resolved Hide resolved

babsingh approved these changes Nov 9, 2023

View reviewed changes

gacholio force-pushed the break branch 2 times, most recently from 371257d to c38e8e0 Compare November 9, 2023 21:08

gacholio force-pushed the break branch 2 times, most recently from 0b5d0b8 to 854e613 Compare November 9, 2023 21:54

gacholio force-pushed the break branch 4 times, most recently from 075368a to 2506ab6 Compare November 10, 2023 12:35

gacholio force-pushed the break branch 6 times, most recently from 6d84cca to 29a20a6 Compare November 11, 2023 13:19

babsingh previously requested changes Nov 11, 2023

View reviewed changes

gacholio requested a review from babsingh November 12, 2023 09:54

gacholio force-pushed the break branch from d1f073d to 7a45f71 Compare November 13, 2023 11:43

gacholio marked this pull request as ready for review November 13, 2023 11:44

babsingh reviewed Nov 13, 2023

View reviewed changes

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

runtime/codert_vm/decomp.cpp Outdated Show resolved Hide resolved

Fix continuation stacks when breakpointing

a4ac0f5

The breakpoint handling code was only fixing stacks actively running on a thread. The stacks for continuations also need to be fixed. Fixes: eclipse-openj9#18088 Signed-off-by: Graham Chapman <graham_chapman@ca.ibm.com>

gacholio force-pushed the break branch from a7c98c0 to a4ac0f5 Compare November 13, 2023 20:25

babsingh approved these changes Nov 14, 2023

View reviewed changes

babsingh merged commit 5fde7f7 into eclipse-openj9:master Nov 14, 2023
7 of 9 checks passed

babsingh mentioned this pull request Nov 14, 2023

Update the fully mounted check in JVMTI Suspend/Resume Thread #18448

Closed

gacholio deleted the break branch November 14, 2023 10:24

babsingh mentioned this pull request Nov 16, 2023

Convert jvmtiThread.c to jvmtiThread.cpp #18459

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix continuation stacks when breakpointing #18413

Fix continuation stacks when breakpointing #18413

gacholio commented Nov 7, 2023 •

edited

Loading

gacholio commented Nov 9, 2023

babsingh left a comment •

edited

Loading

babsingh commented Nov 9, 2023

fengxue-IS left a comment

gacholio commented Nov 9, 2023

gacholio commented Nov 10, 2023

gacholio commented Nov 10, 2023

dmitripivkine commented Nov 10, 2023

babsingh left a comment

gacholio commented Nov 13, 2023

babsingh left a comment

babsingh commented Nov 13, 2023

babsingh commented Nov 13, 2023

babsingh left a comment •

edited

Loading

Fix continuation stacks when breakpointing #18413

Fix continuation stacks when breakpointing #18413

Conversation

gacholio commented Nov 7, 2023 • edited Loading

gacholio commented Nov 9, 2023

babsingh left a comment • edited Loading

Choose a reason for hiding this comment

babsingh commented Nov 9, 2023

fengxue-IS left a comment

Choose a reason for hiding this comment

gacholio commented Nov 9, 2023

gacholio commented Nov 10, 2023

gacholio commented Nov 10, 2023

dmitripivkine commented Nov 10, 2023

babsingh left a comment

Choose a reason for hiding this comment

gacholio commented Nov 13, 2023

babsingh left a comment

Choose a reason for hiding this comment

babsingh commented Nov 13, 2023

babsingh commented Nov 13, 2023

babsingh left a comment • edited Loading

Choose a reason for hiding this comment

gacholio commented Nov 7, 2023 •

edited

Loading

babsingh left a comment •

edited

Loading

babsingh left a comment •

edited

Loading