On Sep 23, 2010, at 1:29 PM, Trond Myklebust wrote: > On Thu, 2010-09-23 at 19:20 +0200, Benny Halevy wrote: >> On 2010-09-21 23:58, Jeff Layton wrote: >>> On Tue, 21 Sep 2010 16:46:21 -0400 >>> Trond Myklebust <trond.myklebust@xxxxxxxxxx> wrote: >>> >>>> On Tue, 2010-09-21 at 16:24 -0400, J. Bruce Fields wrote: >>>>> On Tue, Sep 21, 2010 at 04:00:32PM -0400, Trond Myklebust wrote: >>>>>> On Tue, 2010-09-21 at 15:36 -0400, Trond Myklebust wrote: >>>>>>> On Tue, 2010-09-21 at 15:20 -0400, J. Bruce Fields wrote: >>>>>>>> All I need to do is mount nfs4.1 and run cthon -s. The last output I >>>>>>>> see from the test is: >>>>>>>> >>>>>>>> check for proper open/unlink operation >>>>>>>> nfsjunk files before unlink: >>>>>>>> >>>>>>> >>>>>>> Oh... I bet I see what it is. >>>>>>> >>>>>>> It's that damned res->sr_slotid = NFS4_MAX_SLOT_TABLE and related >>>>>>> initialisation junk that's biting us in the arse again... >>>>>>> >>>>>>> I'll fix it... >>>>>> >>>>>> Does this work? >>>>> >>>>> Yup. >>>> >>>> It seems to fix things for me too... Once you mentioned NFSv4.1 it was >>>> easy to reproduce the bug. >>>> >>>> OK. I'll merge these into nfs-for-2.6.37... >>> >>> Thanks for fixing it. I should have mentioned that the 4.1 parts were >>> written using the "cargo-cult" programming method of copying what >>> unlink does, and that they needed careful review. Mea culpa! >>> >> >> I still hate it that the sr_slotid requires explicit, non-trivial initialization. >> Once upon a time, the equivalent of sr_slotid used to be a pointer to >> struct slot. Using a pointer, implicitly initialized to NULL is significantly >> more straight forward as in the patch below. >> (passes cthon tests over your nfs-for-2.6.37 branch + the patch I >> sent earlier that fixes __put_nfs_open_context. w/o it I readily reproduce the >> crash reported on this thread) >> >> From 50221fb5bd8fbd46f6cabe1fc795db7efc71bff7 Mon Sep 17 00:00:00 2001 >> From: Benny Halevy <bhalevy@xxxxxxxxxxx> >> Date: Thu, 23 Sep 2010 19:08:21 +0200 >> Subject: [PATCH] NFSv4.1: keep seq_res.sr_slot as pointer rather than an index >> >> Having to explicitly initialize sr_slotid to NFS4_MAX_SLOT_TABLE >> resulted in numerous bugs. Keeping the current slot as a pointer >> to the slot table is more straight forward and robust as it's >> implicitly set up to NULL wherever the seq_res member is initialized >> to zeroes. >> >> Signed-off-by: Benny Halevy <bhalevy@xxxxxxxxxxx> > > Looks fine to me, and I prefer this approach to the one we have today. > > Note, however, that I've already applied commit > d688e11007419fd060ae74d8d952a5c4ece735aa (NFSv4.1: Fix the slotid > initialisation in nfs_async_rename()) to the nfs-for-2.6.37 branch. > > Are people happy with me rebasing nfs-for-2.6.37 to remove the above > commit, and apply this one instead? Yes, I like this approach as well and think it should be applied. -->Andy > > Cheers > Trond > -- > To unsubscribe from this list: send the line "unsubscribe linux-nfs" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
