Correctly initialize walkState values when calling walkStackFrames #18967
Conversation
After a thread's stackWalkState is used, we don't reset the fields to 0 but expects future code uses to set the parameters to the expected value before calling walkStackFrames. This change fixes walkWrapperImpl where userData1 is not properly set. Signed-off-by: Jack Lu <Jack.S.Lu@ibm.com>
|
Personal build, build failures due to network issue, sanity.functional test failure is unrelated. @keithc-ca can you please take a look at this. |
| @@ -146,6 +147,8 @@ Java_java_lang_StackWalker_walkWrapperImpl(JNIEnv *env, jclass clazz, jint flags | |||
| if (NULL == walkerMethodChars) { /* native out of memory exception pending */ | |||
| return NULL; | |||
| } | |||
| /* Ensure userData1/2 used by stackFrameFilter function is set properly. */ | |||
| walkState->userData1 = NULL; | |||
There was a problem hiding this comment.
Please explain why this is appropriate. Should line 156 not be here instead?
walkState->userData1 = (void *)(UDATA)flags;Same question for continuations (lines 216-220).
There was a problem hiding this comment.
You are right, I placed the NULL to avoid conflict with getCallerClass call, but given #18868 is already merged, it should be fine to move the assignment earlier.
I will update the PR and run local testing before asking for another round of review.
There was a problem hiding this comment.
local testing passed, @keithc-ca can you please take another look
fengxue-IS
changed the title
Signed-off-by: Jack Lu <Jack.S.Lu@ibm.com>
fengxue-IS
changed the title
|
Jenkins test sanity alinux jdk17,jdk22 |
|
Exclude test ( |
|
The most recent personal build is here. |
|
There was one CRIU test failure which seems unrelated to this change. |
After a thread's stackWalkState is used, we don't reset the fields to 0 but expects future code uses to set the parameters to the expected value before calling walkStackFrames.
This change fixes walkWrapperImpl where userData1 is not properly set.
Fixes: #18696
Fixes: #18961