From: Brian F. <bf...@re...> - 2012-09-26 13:06:01
|
On 09/26/2012 06:14 AM, Maxim V. Patlasov wrote: > Brian, > >> I had a couple nit-level comments/questions inline with the associated >> patches and am hesitantly confident I understand the >> fuse_get_user_pages() bits, but all in all this looks pretty good to me: >> >> Reviewed-by: Brian Foster <bf...@re...> > > Thanks a lot for review. Highly appreciated. > Thanks for the patches! >> >> As Avati mentioned, we're very interested in incorporating this >> functionality. We've seen very good results in preliminary performance >> testing. If this can get upstream relatively soon, we can probably >> contribute some functional testing against the core bits soonish as >> well. Thanks. > > Thanks for support. Glad to hear that the patches helps in your > use-cases too. Btw, what do you, guys, think about making fuse_release() > synchronous? This issue seemed to bite pretty many people and it was > partially fixed for fuseblk use-case. Are you interested in improving > fuse in this area? > I'm not totally sure this is an issue for us (Avati can jump in if he's familiar with it), so I'd have to reserve judgement on that until I understand the issue better. Brian > Thanks, > Maxim |