-
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
Do not create class unload PIC site assumption if not required #18063
Do not create class unload PIC site assumption if not required #18063
Conversation
The issue is `TR_RelocationRecordSymbolFromManager::activatePointer` creates `TR_UnloadedClassPicSite` for a J9Method in the pre-prologue of the current method body. The class of the J9Method is the same as the class of the current method body. In this case, the class unload assumption is not needed because once the class is unloaded, no code will ever reach the J9Method in the pre-prologue. Fixes: eclipse-openj9#17734 Signed-off-by: Annabelle Huo <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGMT
Are there other parts of the code that may need a similar change? |
#17734 is only reproduced so far with AOT loaded methods. Other places in openj9/runtime/compiler/runtime/RelocationRecord.cpp Lines 2696 to 2698 in 2cbea41
openj9/runtime/compiler/runtime/RelocationRecord.cpp Lines 5677 to 5679 in 2cbea41
openj9/runtime/compiler/runtime/RelocationRecord.cpp Lines 5883 to 5884 in 2cbea41
No class pointer openj9/runtime/compiler/runtime/RelocationRecord.cpp Lines 6138 to 6140 in 2cbea41
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
I started Jenkins test with this change and having the assert as TR_ASSERT_FATAL to double check if there are other cases that hit this assert. |
jenkins test sanity all jdk17 |
@a7ehuo Unless the result from the grinder mentioned in comment #18063 (comment) is not good, I am ready to perform the merge. |
Jenkins result looks good. I ran |
The issue is
TR_RelocationRecordSymbolFromManager::activatePointer
createsTR_UnloadedClassPicSite
for a J9Method in the pre-prologue of the current method body. The class of the J9Method is the same as the class of the current method body. In this case, the class unload assumption is not needed because once the class is unloaded, no code will ever reach the J9Method in the pre-prologue.Fixes: #17734