From: "J. Bruce Fields" Subject: Re: [PATCH] knfsd: nfsd: Handle ERESTARTSYS from syscalls. Date: Sat, 14 Jun 2008 13:11:09 -0400 Message-ID: <20080614171109.GA27041@fieldses.org> References: <20080613213759.26929.patches@notabene> <1080613114215.27095@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-nfs@vger.kernel.org, linux-kernel@vger.kernel.org To: NeilBrown Return-path: Received: from mail.fieldses.org ([66.93.2.214]:35746 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753062AbYFNRLL (ORCPT ); Sat, 14 Jun 2008 13:11:11 -0400 In-Reply-To: <1080613114215.27095@suse.de> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Fri, Jun 13, 2008 at 09:42:15PM +1000, NeilBrown wrote: > > OCFS2 can return -ERESTARTSYS from write requests (and possibly > elsewhere) if there is a signal pending. > > If nfsd is shutdown (by sending a signal to each thread) while there > is still an IO load from the client, each thread could handle one last > request with a signal pending. This can result in -ERESTARTSYS > which is not understood by nfserrno() and so is reflected back to > the client as nfserr_io aka -EIO. This is wrong. > > Instead, interpret ERESTARTSYS to mean "don't send a reply". > The client will resend and - if the server is restarted - the write will > (hopefully) be successful and everyone will be happy. Thanks, applied--with a trivial change: > > Signed-off-by: Neil Brown > > ### Diffstat output > ./fs/nfsd/nfsproc.c | 1 + > 1 file changed, 1 insertion(+) > > ---- > Funny how the shortest patches sometimes have the longest > descriptions. I added this to the body of the patch description: > > The symptom that I narrowed down to this was: > copy a large file via NFS to an OCFS2 filesystem, and restart > the nfs server during the copy. > The 'cp' might get an -EIO, and the file will be corrupted - > presumably holes in the middle were writes appeared to fail. On the grounds that someone might run across the same problem or need to reproduce it for some reason, so a good, simple description of the case that found it might help. --b. > > diff .prev/fs/nfsd/nfsproc.c ./fs/nfsd/nfsproc.c > --- .prev/fs/nfsd/nfsproc.c 2008-06-13 21:31:53.000000000 +1000 > +++ ./fs/nfsd/nfsproc.c 2008-06-13 21:31:57.000000000 +1000 > @@ -614,6 +614,7 @@ nfserrno (int errno) > #endif > { nfserr_stale, -ESTALE }, > { nfserr_jukebox, -ETIMEDOUT }, > + { nfserr_dropit, -ERESTARTSYS }, > { nfserr_dropit, -EAGAIN }, > { nfserr_dropit, -ENOMEM }, > { nfserr_badname, -ESRCH }, > > ### Diffstat output > ./fs/nfsd/nfsproc.c | 1 + > 1 file changed, 1 insertion(+) > > diff .prev/fs/nfsd/nfsproc.c ./fs/nfsd/nfsproc.c > --- .prev/fs/nfsd/nfsproc.c 2008-06-13 21:31:53.000000000 +1000 > +++ ./fs/nfsd/nfsproc.c 2008-06-13 21:31:57.000000000 +1000 > @@ -614,6 +614,7 @@ nfserrno (int errno) > #endif > { nfserr_stale, -ESTALE }, > { nfserr_jukebox, -ETIMEDOUT }, > + { nfserr_dropit, -ERESTARTSYS }, > { nfserr_dropit, -EAGAIN }, > { nfserr_dropit, -ENOMEM }, > { nfserr_badname, -ESRCH },