On Mon, 2009-01-05 at 19:34 +0300, malc wrote:

Before this change (atleast) mono_handle_native_sigsegv was executed
(before machine locks-up hard) after the change this code path is
never touched.

The fact that machine locks up hard and not even magic sysrq works
is what deserves a better look.

Indeed. It's possible that the normally not hit error case is triggering
a different bug. Definitely worth looking at.

I didn't have a chance so far, I'm a bit swamped at the moment, but keep
pining me please :-)

Cheers,
Ben.