Messages in this thread | | | From | Godmar Back <> | Subject | Re: oops in nfs_wback_result 2.2.3 | Date | Thu, 18 Mar 1999 12:06:35 -0700 (MST) |
| |
FWIW, nfs/write.c contains this code:
/* * Ok, there's another user of this page with the new request.. * The IO completion will then free the page and the dentry. */ atomic_inc(&page->count); file->f_count++;
Now if the problem is with f_count and it's on an SMP only, could it be that that's the problem up there?
If page->count requires an atomic update in this particular situation, wouldn't f_count also require an atomic update? Or is f_count guaranteed to be accessed only be one processor?
I'm just guessing here, of course.
- Godmar
> > > Hi, > > Thanks for your report. > > Can you repeat this Oops? I've seen reports of it before, and it > seems to indicate that the 'f_count' from the file pointer is out of > sync (the Oops is due to the f_dentry being NULL, so that the attempt > to reference f_dentry->inode fails). > > If you have found a way of reproducing it, do you think you could > 'echo 72 >/proc/sys/sunrpc/nfs_debug' and then provoke the Oops? I > need to find out what is going wrong with the file count... > > BTW: is anybody seeing this problem on non-SMP machines? > > Cheers, > Trond >
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |