Messages in this thread | | | Date | Tue, 10 Jan 2012 17:50:28 +0100 | Subject | Re: [RFC PATCH] shrink_dcache_parent() deadlock | From | Miklos Szeredi <> |
| |
On Tue, Jan 10, 2012 at 5:33 PM, Linus Torvalds <torvalds@linux-foundation.org> wrote: > Hmm. Even if they are never run in parallel, I think it would be much > nicer to do it in both, just so that there would be a conceptual > consistency of "when we remove the dentry from the LRU list and put it > on our pruning list, we set the bit". That cacheline will be dirty > anyway (due to the list move and getting the dentry lock), so setting > a bit is not expensive - but having odd inconsistent ad-hoc rules is > nasty.
Makes sense.
I'm testing the modified patch right now and will post shortly.
Thanks, Miklos
| |