All of lore.kernel.org
 help / color / mirror / Atom feed
From: Boaz Harrosh <bharrosh@panasas.com>
To: Trond Myklebust <Trond.Myklebust@netapp.com>,
	Benny Halevy <bhalevy@panasas.com>,
	NFS list <linux-nfs@vger.kernel.org>,
	open-osd <osd-dev@open-osd.org>
Cc: Benny Halevy <bhalevy@panasas.com>
Subject: [PATCH 02/32] NFSv4.1: use struct nfs_client to qualify deviceid
Date: Sun, 29 May 2011 21:17:28 +0300	[thread overview]
Message-ID: <1306693048-7759-1-git-send-email-bharrosh@panasas.com> (raw)
In-Reply-To: <4DE28CFB.4060608@panasas.com>

From: Benny Halevy <bhalevy@panasas.com>

deviceids are unique per server, per layout type.
Therefore, in the global cache in the files layout driver
deviceids from different servers may clash so we need
to qualify them with a struct nfs_client that represents
the nfs server that returned the deviceid.

Introduced in 2.6.39 commit ea8eecdd
"NFSv4.1 move deviceid cache to filelayout driver"

Signed-off-by: Benny Halevy <bhalevy@panasas.com>
---
 fs/nfs/nfs4filelayout.c    |    2 +-
 fs/nfs/nfs4filelayout.h    |    3 ++-
 fs/nfs/nfs4filelayoutdev.c |    9 ++++-----
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c
index dd6ccf0..571c1b0 100644
--- a/fs/nfs/nfs4filelayout.c
+++ b/fs/nfs/nfs4filelayout.c
@@ -440,7 +440,7 @@ filelayout_check_layout(struct pnfs_layout_hdr *lo,
 	}
 
 	/* find and reference the deviceid */
-	dsaddr = nfs4_fl_find_get_deviceid(id);
+	dsaddr = nfs4_fl_find_get_deviceid(NFS_SERVER(lo->plh_inode)->nfs_client, id);
 	if (dsaddr == NULL) {
 		dsaddr = get_device_info(lo->plh_inode, id, gfp_flags);
 		if (dsaddr == NULL)
diff --git a/fs/nfs/nfs4filelayout.h b/fs/nfs/nfs4filelayout.h
index 2b461d7..301b955 100644
--- a/fs/nfs/nfs4filelayout.h
+++ b/fs/nfs/nfs4filelayout.h
@@ -60,6 +60,7 @@ struct nfs4_pnfs_ds {
 
 struct nfs4_file_layout_dsaddr {
 	struct hlist_node		node;
+	struct nfs_client		*nfs_client;
 	struct nfs4_deviceid		deviceid;
 	atomic_t			ref;
 	unsigned long			flags;
@@ -101,7 +102,7 @@ u32 nfs4_fl_calc_ds_index(struct pnfs_layout_segment *lseg, u32 j);
 struct nfs4_pnfs_ds *nfs4_fl_prepare_ds(struct pnfs_layout_segment *lseg,
 					u32 ds_idx);
 extern struct nfs4_file_layout_dsaddr *
-nfs4_fl_find_get_deviceid(struct nfs4_deviceid *dev_id);
+nfs4_fl_find_get_deviceid(struct nfs_client *, struct nfs4_deviceid *dev_id);
 extern void nfs4_fl_put_deviceid(struct nfs4_file_layout_dsaddr *dsaddr);
 struct nfs4_file_layout_dsaddr *
 get_device_info(struct inode *inode, struct nfs4_deviceid *dev_id, gfp_t gfp_flags);
diff --git a/fs/nfs/nfs4filelayoutdev.c b/fs/nfs/nfs4filelayoutdev.c
index db07c7a..42e3266 100644
--- a/fs/nfs/nfs4filelayoutdev.c
+++ b/fs/nfs/nfs4filelayoutdev.c
@@ -431,7 +431,7 @@ decode_device(struct inode *ino, struct pnfs_device *pdev, gfp_t gfp_flags)
 	dsaddr->stripe_indices = stripe_indices;
 	stripe_indices = NULL;
 	dsaddr->ds_num = num;
-
+	dsaddr->nfs_client = NFS_SERVER(ino)->nfs_client;
 	memcpy(&dsaddr->deviceid, &pdev->dev_id, sizeof(pdev->dev_id));
 
 	for (i = 0; i < dsaddr->ds_num; i++) {
@@ -516,7 +516,7 @@ decode_and_add_device(struct inode *inode, struct pnfs_device *dev, gfp_t gfp_fl
 	}
 
 	spin_lock(&filelayout_deviceid_lock);
-	d = nfs4_fl_find_get_deviceid(&new->deviceid);
+	d = nfs4_fl_find_get_deviceid(new->nfs_client, &new->deviceid);
 	if (d) {
 		spin_unlock(&filelayout_deviceid_lock);
 		nfs4_fl_free_deviceid(new);
@@ -610,16 +610,15 @@ nfs4_fl_put_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
 }
 
 struct nfs4_file_layout_dsaddr *
-nfs4_fl_find_get_deviceid(struct nfs4_deviceid *id)
+nfs4_fl_find_get_deviceid(struct nfs_client *clp, struct nfs4_deviceid *id)
 {
 	struct nfs4_file_layout_dsaddr *d;
 	struct hlist_node *n;
 	long hash = nfs4_fl_deviceid_hash(id);
 
-
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(d, n, &filelayout_deviceid_cache[hash], node) {
-		if (!memcmp(&d->deviceid, id, sizeof(*id))) {
+		if (d->nfs_client == clp && !memcmp(&d->deviceid, id, sizeof(*id))) {
 			if (!atomic_inc_not_zero(&d->ref))
 				goto fail;
 			rcu_read_unlock();
-- 
1.7.2.3


  parent reply	other threads:[~2011-05-29 18:17 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-05-29 18:14 [PATCHSET final 00/32] pnfs-obj: pnfs-objects layout driver submission Boaz Harrosh
2011-05-29 18:16 ` [PATCH 01/32] NFSv4.1: fix typo in filelayout_check_layout Boaz Harrosh
2011-05-29 18:17 ` Boaz Harrosh [this message]
2011-05-29 18:17 ` [PATCH 03/32] pnfs: resolve header dependency in pnfs.h Boaz Harrosh
2011-05-29 18:18 ` [PATCH 04/32] NFSv4.1: make deviceid cache global Boaz Harrosh
2011-05-29 18:18 ` [PATCH 05/32] NFSv4.1: purge deviceid cache on nfs_free_client Boaz Harrosh
2011-05-29 18:19 ` [PATCH 06/32] pnfs: CB_NOTIFY_DEVICEID Boaz Harrosh
2011-05-29 18:19 ` [PATCH 07/32] NFSv4.1: use layout driver in global device cache Boaz Harrosh
2011-05-29 18:19 ` [PATCH 08/32] SUNRPC: introduce xdr_init_decode_pages Boaz Harrosh
2011-05-29 18:19 ` [PATCH 09/32] pnfs: Use byte-range for layoutget Boaz Harrosh
2011-05-29 18:20 ` [PATCH 10/32] pnfs: align layoutget requests on page boundaries Boaz Harrosh
2011-05-29 18:20 ` [PATCH 11/32] pnfs: Use byte-range for cb_layoutrecall Boaz Harrosh
2011-05-29 18:21 ` [PATCH 12/32] pnfs: client stats Boaz Harrosh
2011-05-29 18:21 ` [PATCH 13/32] pnfs-obj: objlayoutdriver module skeleton Boaz Harrosh
2011-05-29 18:21 ` [PATCH 14/32] pnfs-obj: pnfs_osd XDR definitions Boaz Harrosh
2011-05-29 18:22 ` [PATCH 15/32] pnfs-obj: pnfs_osd XDR client implementation Boaz Harrosh
2011-05-29 18:22 ` [PATCH 16/32] pnfs-obj: decode layout, alloc/free lseg Boaz Harrosh
2011-05-29 18:22 ` [PATCH 17/32] pnfs-obj: objio_osd device information retrieval and caching Boaz Harrosh
2011-05-29 18:23 ` [PATCH 18/32] pnfs: alloc and free layout_hdr layoutdriver methods Boaz Harrosh
2011-05-29 18:23 ` [PATCH 19/32] pnfs-obj: define per-inode private structure Boaz Harrosh
2011-05-29 18:24 ` [PATCH 20/32] pnfs: support for non-rpc layout drivers Boaz Harrosh
2011-05-29 18:25 ` [PATCH 21/32] pnfs-obj: osd raid engine read/write implementation Boaz Harrosh
2011-05-29 18:25 ` [PATCH 22/32] pnfs: layoutreturn Boaz Harrosh
2011-05-29 18:26 ` [PATCH 23/32] pnfs: layoutret_on_setattr Boaz Harrosh
2011-05-29 18:26 ` [PATCH 24/32] pnfs: encode_layoutreturn Boaz Harrosh
2011-05-29 18:26 ` [PATCH 25/32] pnfs-obj: report errors and .encode_layoutreturn Implementation Boaz Harrosh
2011-05-29 18:27 ` [PATCH 26/32] pnfs: encode_layoutcommit Boaz Harrosh
2011-05-29 18:27 ` [PATCH 27/32] pnfs-obj: objlayout_encode_layoutcommit implementation Boaz Harrosh
2011-05-29 18:27 ` [PATCH 28/32] NFSv4.1: unify pnfs_pageio_init functions Boaz Harrosh
2011-05-29 18:28 ` [PATCH 29/32] NFSv4.1: change pg_test return type to bool Boaz Harrosh
2011-05-29 18:28 ` [PATCH 30/32] NFSv4.1: use pnfs_generic_pg_test directly by layout driver Boaz Harrosh
2011-05-29 18:29 ` [PATCH 31/32] NFSv4.1: define nfs_generic_pg_test Boaz Harrosh
2011-05-29 18:30 ` [PATCH 32/32] pnfs-obj: pg_test check for max_io_size Boaz Harrosh
2011-05-29 18:36 ` [osd-dev] [PATCHSET final 00/32] pnfs-obj: pnfs-objects layout driver submission Boaz Harrosh
2011-05-31  5:58 ` Benny Halevy
2011-05-31 23:32   ` Jim Rees
2011-06-01 11:55     ` Benny Halevy

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=1306693048-7759-1-git-send-email-bharrosh@panasas.com \
    --to=bharrosh@panasas.com \
    --cc=Trond.Myklebust@netapp.com \
    --cc=bhalevy@panasas.com \
    --cc=linux-nfs@vger.kernel.org \
    --cc=osd-dev@open-osd.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.