[issue1975] signals not always delivered to main thread, since other threads have the signal unmasked

[Duncan Findlay]

Duncan Findlay <duncan at duncf.ca> added the comment:

This is definitely still an issue.

With the "pthread_sig" patch attached to this bug, both on FreeBSD and on linux, any processes spawned from a thread seem to have their signals blocked, so they can not be killed.

Without it, on FreeBSD, the behavior described by bamby is still a problem.

I've attached a test case that adapts bamby's example code into a test, and shows the "unkillable subprocess" problem I described above.

On FreeBSD without the patch, test_signal fails, and with the patch test_thr fails.

On Linux and OS X, without the patch, all tests pass. With the patch, test_thr fails.

I hope somebody can come up with a better fix.

----------
Added file: http://bugs.python.org/file20046/thread_test.py

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue1975>
_______________________________________