KONNO Kazuhiro

Issue #14621 looks the same. - Segmentation error in old_slow_jitHandleInternalErrorTrap - The InaccessibleAddress is near the end of a memory page

> It doesn't make sense to me that a dedicated JIT compilation thread would be executing such Java code. I agree it is confusing. What I know is the javacore...

It is likely these macOS issues share the same problem: #14621, #15352, #15518 (this issue)

I ran Grinder jobs for SCM23.MultiCL_0, but the crash was not reproduced in 80 runs in total. internal job/Grinder/27331, 27351

I tried two more Ginder jobs (40x each) for SharedClasses.SCM23.MultiCL_0. No crashes. internal job/Grinder/27363, 27454

Yet another Grinder job (40x) with no crash: job/Grinder/27459

My local testing with debug prints (2 samples) shows the following: - `jitARM64Handler()` is called with `sigType` = 40 (`J9PORT_SIG_FLAG_SIGBUS`) repeatedly. - `pcPtr` points to the entry address of jitted...

I reproduced the crash with repeated calls to `jitARM64Handler()` in an internal Grinder job 27566. The following is from the debug print added in `jitARM64Handler()`. ``` [2022-09-15T00:49:38.395Z] MTM4 stderr @@1...

I found a "Unable to locate JIT stack map" case in my local testing, as shown below. ``` MTM4 15:41:20 >> Loaded 20000 classes... MTM4 stderr @@1 sigType = 40,...

I opened PR #15907 as a fix for this issue. I think the PR also fixes some other intermittent issues on macOS.