On 19 Mar 2008, at 07:13, Jeff Roberson wrote:
jeff 2008-03-19 07:13:24 UTC
FreeBSD src repository
Modified files:
sys/kern kern_lockf.c
Log:
- Fix the last of the threading bugs that were introduced as far
back as
1.38 in 2001. Break out of the FOREACH_THREAD_IN_PROC loop when
we've
discovered a new proc in the chain.
- Increment i and check for maxlockdepth once per matching process
not
once per thread. This didn't properly terminate the loop before.
- Fix a bug which has existed potentially since rev 1.1.
waitblock->lf_next
can be NULL when a thread has been woken-up but not yet
scheduled. Check
for this condition rather than blindly dereferencing.
Found by: libMicro
Revision Changes Path
1.59 +12 -6 src/sys/kern/kern_lockf.c
I have completely removed all this proc tomfoolery in my perforce
branch and have replaced it with a real deadlock detection system
instead of this horrible code. Apart from anything else, the existing
code has serious locking problems - if the deadlock loop spans two or
more vnodes, it doesn't lock any but the first one.
I should have a patch ready for review (along with the new kernel NFS
lock manager) towards the end of this week and I hope to get the thing
into -current some time next week.
