Messages in this thread | | | Date | Wed, 6 Jun 2012 20:29:14 -0400 | From | Dave Jones <> | Subject | Re: processes hung after sys_renameat, and 'missing' processes |
| |
On Thu, Jun 07, 2012 at 12:54:04AM +0100, Al Viro wrote:
> What we need is ->i_mutex on parents. And I'm much more concerned about > this: 7732a557b1342c6e6966efb5f07effcf99f56167 and > 3f50fff4dace23d3cfeb195d5cd4ee813cee68b7. > > Dave, you seem to be able to reproduce it; could you try with those two > commits reverted? This stuff is *definitely* wrong with the way it > treats d_move(); there we might get it with parents not locked at all. > > FWIW, I'd suggest adding a check into d_move(); new parent must be > locked in all cases and old one whenever dentry has one (i.e. isn't > disconnected). If you can find a violation of that, you very likely > have found the cause of that bug.
Like this ?
void d_move(struct dentry *dentry, struct dentry *target) { write_seqlock(&rename_lock); + + BUG_ON(!spin_is_locked(&target->d_parent->d_lock)); + + if (dentry->d_parent != NULL) + BUG_ON(!spin_is_locked(&dentry->d_parent->d_lock)); + __d_move(dentry, target); write_sequnlock(&rename_lock); }
To be clear, do you want me to try that with or without the reverts ?
Dave
| |