From: Boaz Harrosh <bharrosh@panasas.com>
To: "William A. (Andy) Adamson" <androsadamson@gmail.com>,
Benny Halevy <bhalevy@panasas.com>,
NFS list <linux-nfs@vger.kernel.org>
Subject: ONE More BAD bug with 2.6.34-pnfs branch found - watch out.
Date: Thu, 13 May 2010 18:58:59 +0300 [thread overview]
Message-ID: <4BEC21C3.50007@panasas.com> (raw)
The NFS_FILE_SYNC return from write_done is one big bug. Do not use it!
Both Pan_shim and osd_objio where returning NFS_FILE_SYN. For pan_shim it
is true and for osd_objio it was out of laziness and was actually a lie.
The patch below will work around the situation and will finally after
many month return my 2.6.34 client to it's previous glory. Fffffff
If any one is returning NFS_FILE_SYNC from write then they have a problem
Boaz
---
git diff --stat -p -M fs/nfs/objlayout
fs/nfs/objlayout/objio_osd.c | 2 +-
fs/nfs/objlayout/objlayout.c | 17 ++++++++++++++++-
2 files changed, 17 insertions(+), 2 deletions(-)
diff --git a/fs/nfs/objlayout/objio_osd.c b/fs/nfs/objlayout/objio_osd.c
index 642d6fa..73e8b85 100644
--- a/fs/nfs/objlayout/objio_osd.c
+++ b/fs/nfs/objlayout/objio_osd.c
@@ -852,7 +852,7 @@ static ssize_t _write_done(struct objio_state *ios)
if (likely(!ret)) {
/* FIXME: should be based on the OSD's persistence model
* See OSD2r05 Section 4.13 Data persistence model */
- ios->ol_state.committed = NFS_FILE_SYNC;
+ ios->ol_state.committed = NFS_UNSTABLE; //NFS_FILE_SYNC;
status = ios->length;
} else {
status = ret;
diff --git a/fs/nfs/objlayout/objlayout.c b/fs/nfs/objlayout/objlayout.c
index 880d987..60f64b7 100644
--- a/fs/nfs/objlayout/objlayout.c
+++ b/fs/nfs/objlayout/objlayout.c
@@ -287,15 +287,30 @@ objlayout_io_set_result(struct objlayout_io_state *state, unsigned index,
}
}
+static void _rpc_commit_complete(struct work_struct *work)
+{
+ struct rpc_task *task;
+ struct nfs_write_data *wdata;
+
+ dprintk("%s enter\n", __func__);
+ task = container_of(work, struct rpc_task, u.tk_work);
+ wdata = container_of(task, struct nfs_write_data, task);
+
+ pnfs_client_ops->nfs_commit_complete(wdata);
+}
+
/*
* Commit data remotely on OSDs
*/
enum pnfs_try_status
objlayout_commit(struct pnfs_layout_type *pnfslay,
int sync,
- struct nfs_write_data *data)
+ struct nfs_write_data *wdata)
{
int status = PNFS_ATTEMPTED;
+
+ INIT_WORK(&wdata->task.u.tk_work, _rpc_commit_complete);
+ schedule_work(&wdata->task.u.tk_work);
dprintk("%s: Return %d\n", __func__, status);
return status;
}
reply other threads:[~2010-05-13 15:59 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4BEC21C3.50007@panasas.com \
--to=bharrosh@panasas.com \
--cc=androsadamson@gmail.com \
--cc=bhalevy@panasas.com \
--cc=linux-nfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).