Jlalond wrote:
Hey @DavidSpickett thanks for fixing the test.
I think you fixed it the right way, but I wanted to share my understanding of
how this probably happened
In ProcessElfCore, we check the siginfo status of all the threads
```
if (!siginfo_signal_found) {
// If we don't have signal from SIGINFO use the signal from each threads
// PRSTATUS note.
if (prstatus_signal_found) {
for (auto &thread_data : m_thread_data)
thread_data.signo = thread_data.prstatus_sig;
} else if (m_thread_data.size() > 0) {
// If all else fails force the first thread to be SIGSTOP
m_thread_data.begin()->signo =
GetUnixSignals()->GetSignalNumberFromName("SIGSTOP");
}
}
```
So in our case, I've changed this behavior, where `siginfo_signal_found` will
always be set to true if we've extracted any bytes from the PT_NOTE even if we
can't use them later due to a lack of a platform class. Because we passed the
siginfo bytes check we never populate the `signo` from the PRSTATUS
Then in ThreadElfCore I think we'd fail to calculate stop info because we don't
have a valid SIGNO.
I think a few fixes here:
Currently, we set the stop reason even for `signo = 0` so you get a confusing
stopped with signal 0 in LLDB. We should always stop for ThreadElfCore, but
only make stop info dependent on the actual siginfo.
This same infinite loop also can happen for Minidump, and I haven't had time to
fix it. I think in general we can wrap all of this into a nice postmortem
thread and have the postmortem thread know it should always stop, but
optionally get the signal description.
CC: @labath
https://github.com/llvm/llvm-project/pull/142143
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
