DragonFly commits List (threaded) for 2011-03
Re: git: kernel -- vm_object locking: DEBUG_LOCKS check for hold_wait vs hold deadlock
On Sun, Mar 27, 2011 at 12:01:19PM -0700, Matthew Dillon wrote:
: If a thread has a hold on a vm_object and enters hold_wait (via either
: vm_object_terminate or vm_object_collapse), it will wait forever for the hold
: count to hit 0. Record the threads holding an object in a per-object array.
Hmm. That's a bit iffy, it is possible for there to be more than 32
holds on a VM object.
Agreed; this was something that was useful while debugging something locally,
but I kept it under DEBUG_LOCKS rather than INVARIANTS because it was not
generally safe. Would it be alright for me to drag bitstring.h into sys/
(where it is on FreeBSD at least) so that I can use that instead (and make
a growable bitvector)?
I thought this was okay, since DEBUG_LOCKS is not on by default; in the future
would you recommend not committing in stuff like this?