| commit | 81b18e513c41e0e025fc785a16df2c1a63e9551c | |
| author | Matthew Dillon <dillon@apollo.backplane.com> | |
| Sun, 25 Jan 2009 07:13:46 +0000 (24 23:13 -0800) | ||
| committer | Matthew Dillon <dillon@apollo.backplane.com> | |
| Sun, 25 Jan 2009 07:13:46 +0000 (24 23:13 -0800) | ||
| tree | 55611cb1180ab3a2445a774cc93b45a9b333bf5f | treesnapshot (tar.gz zip) |
| parent | 3f9b4cfa354fa847568bf60bb411929ef858e6a5 | commitdiff |
Fix another process stuck-during-exit bug - stuck due to signal.
Fix a bug where an exiting process becomes a zombie, entering the SZOMB
state, but then receives a stop signal before it can be reaped. If this
occurs the process was incorrectly placed in SSTOP and then later made
SACTIVE again, preventing it from being reaped.
Symptoms were: process shows up in ps in a DEL state but does not show
up in kgdb's 'info thread' list. Process is the zombie list and is still
a child of init, but it's state is SACTIVE instead of SZOMB and p_xstat
(last signal taken) is SIGSTOP or SIGTSTP.
Fix a bug where an exiting process becomes a zombie, entering the SZOMB
state, but then receives a stop signal before it can be reaped. If this
occurs the process was incorrectly placed in SSTOP and then later made
SACTIVE again, preventing it from being reaped.
Symptoms were: process shows up in ps in a DEL state but does not show
up in kgdb's 'info thread' list. Process is the zombie list and is still
a child of init, but it's state is SACTIVE instead of SZOMB and p_xstat
(last signal taken) is SIGSTOP or SIGTSTP.
| sys/kern/kern_sig.c | diffblobblamehistory |