RFR: JDK-8297184: Test runtime/ErrorHandling/TestSigInfoInHsErrFile.java is failing

[Thomas Stuefe]

On Thu, 17 Nov 2022 11:13:23 GMT, David Holmes <dholmes at openjdk.org> wrote:

>> https://bugs.openjdk.org/browse/JDK-8296906 added a new test that artificially crashes the VM with a known signal/code/crashaddress and checks that we see those informations precisely in the hs-err file. It did this for SIGSEGV and SIGFPE.
>> 
>> SIGFPE, however, is problematic. We fall back to pthread_kill() if we cannot trigger a real FPE fault. In that case the code will be SI_KILL and the sending frame will be pthread_kill(). 
>> 
>> Since we cannot guarantee that a real signal is sent, I opted for just removing the part of the test that tests FPE. The other part, testing SIGSEGV, is enough to cover what the test should cover.
>> 
>> *update*
>> 
>> I also added a fix for the second error mode we see, where we don't crash with SIGSEGV SEGV_MAPERR but SIGSEGV ACCERR.
>
> @tstuefe  I am also seeing the SIGSEGV test failing on macos.
> 
> siginfo: si_signo: 11 (SIGSEGV), si_code: 2 (SEGV_ACCERR), si_addr: 0x0000000000000400
> 
> It isn't MAPERR.

@dholmes-ora @dcubed-ojdk @coleenp thanks for the reviews. Sorry for the inconvenience!

-------------

PR: https://git.openjdk.org/jdk/pull/11202