From: Trond Myklebust <trond.myklebust@primarydata.com>
To: linux-nfs@vger.kernel.org
Subject: [PATCH 3/4] NFSv4.1/flexfile: Ensure uniqueness of mirrors across layout segments
Date: Mon, 24 Aug 2015 21:43:05 -0400 [thread overview]
Message-ID: <1440466986-36826-3-git-send-email-trond.myklebust@primarydata.com> (raw)
In-Reply-To: <1440466986-36826-2-git-send-email-trond.myklebust@primarydata.com>
Keep the full list of mirrors in the struct nfs4_ff_layout_mirror so that
they can be shared among the layout segments that use them.
Also ensure that we send out only one copy of the layoutstats per mirror.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
---
fs/nfs/flexfilelayout/flexfilelayout.c | 116 ++++++++++++++++++++++++++-------
fs/nfs/flexfilelayout/flexfilelayout.h | 3 +
2 files changed, 95 insertions(+), 24 deletions(-)
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index f3efff640989..210352741177 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -34,6 +34,7 @@ ff_layout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags)
ffl = kzalloc(sizeof(*ffl), gfp_flags);
if (ffl) {
INIT_LIST_HEAD(&ffl->error_list);
+ INIT_LIST_HEAD(&ffl->mirrors);
return &ffl->generic_hdr;
} else
return NULL;
@@ -135,6 +136,66 @@ decode_name(struct xdr_stream *xdr, u32 *id)
return 0;
}
+static bool ff_mirror_match_fh(const struct nfs4_ff_layout_mirror *m1,
+ const struct nfs4_ff_layout_mirror *m2)
+{
+ int i, j;
+
+ if (m1->fh_versions_cnt != m2->fh_versions_cnt)
+ return false;
+ for (i = 0; i < m1->fh_versions_cnt; i++) {
+ bool found_fh = false;
+ for (j = 0; j < m2->fh_versions_cnt; i++) {
+ if (nfs_compare_fh(&m1->fh_versions[i],
+ &m2->fh_versions[j])) {
+ found_fh = true;
+ break;
+ }
+ }
+ if (!found_fh)
+ return false;
+ }
+ return true;
+}
+
+static struct nfs4_ff_layout_mirror *
+ff_layout_add_mirror(struct pnfs_layout_hdr *lo,
+ struct nfs4_ff_layout_mirror *mirror)
+{
+ struct nfs4_flexfile_layout *ff_layout = FF_LAYOUT_FROM_HDR(lo);
+ struct nfs4_ff_layout_mirror *pos;
+ struct inode *inode = lo->plh_inode;
+
+ spin_lock(&inode->i_lock);
+ list_for_each_entry(pos, &ff_layout->mirrors, mirrors) {
+ if (mirror->mirror_ds != pos->mirror_ds)
+ continue;
+ if (!ff_mirror_match_fh(mirror, pos))
+ continue;
+ if (atomic_inc_not_zero(&pos->ref)) {
+ spin_unlock(&inode->i_lock);
+ return pos;
+ }
+ }
+ list_add(&mirror->mirrors, &ff_layout->mirrors);
+ mirror->layout = lo;
+ spin_unlock(&inode->i_lock);
+ return mirror;
+}
+
+void
+ff_layout_remove_mirror(struct nfs4_ff_layout_mirror *mirror)
+{
+ struct inode *inode;
+ if (mirror->layout == NULL)
+ return;
+ inode = mirror->layout->plh_inode;
+ spin_lock(&inode->i_lock);
+ list_del(&mirror->mirrors);
+ spin_unlock(&inode->i_lock);
+ mirror->layout = NULL;
+}
+
static struct nfs4_ff_layout_mirror *ff_layout_alloc_mirror(gfp_t gfp_flags)
{
struct nfs4_ff_layout_mirror *mirror;
@@ -143,12 +204,14 @@ static struct nfs4_ff_layout_mirror *ff_layout_alloc_mirror(gfp_t gfp_flags)
if (mirror != NULL) {
spin_lock_init(&mirror->lock);
atomic_set(&mirror->ref, 1);
+ INIT_LIST_HEAD(&mirror->mirrors);
}
return mirror;
}
static void ff_layout_free_mirror(struct nfs4_ff_layout_mirror *mirror)
{
+ ff_layout_remove_mirror(mirror);
kfree(mirror->fh_versions);
nfs4_ff_layout_put_deviceid(mirror->mirror_ds);
kfree(mirror);
@@ -267,6 +330,7 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
goto out_err_free;
for (i = 0; i < fls->mirror_array_cnt; i++) {
+ struct nfs4_ff_layout_mirror *mirror;
struct nfs4_deviceid devid;
struct nfs4_deviceid_node *idnode;
u32 ds_count;
@@ -355,6 +419,12 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
if (rc)
goto out_err_free;
+ mirror = ff_layout_add_mirror(lh, fls->mirror_array[i]);
+ if (mirror != fls->mirror_array[i]) {
+ ff_layout_free_mirror(fls->mirror_array[i]);
+ fls->mirror_array[i] = mirror;
+ }
+
dprintk("%s: uid %d gid %d\n", __func__,
fls->mirror_array[i]->uid,
fls->mirror_array[i]->gid);
@@ -1883,24 +1953,27 @@ ff_layout_encode_layoutstats(struct xdr_stream *xdr,
*start = cpu_to_be32((xdr->p - start - 1) * 4);
}
-static bool
+static int
ff_layout_mirror_prepare_stats(struct nfs42_layoutstat_args *args,
struct pnfs_layout_segment *pls,
- int *dev_count, int dev_limit)
+ int dev_limit)
{
+ struct nfs4_flexfile_layout *ff_layout = FF_LAYOUT_FROM_HDR(pls->pls_layout);
struct nfs4_ff_layout_mirror *mirror;
struct nfs4_deviceid_node *dev;
struct nfs42_layoutstat_devinfo *devinfo;
- int i;
+ int i = 0;
- for (i = 0; i < FF_LAYOUT_MIRROR_COUNT(pls); i++) {
- if (*dev_count >= dev_limit)
+ list_for_each_entry(mirror, &ff_layout->mirrors, mirrors) {
+ if (i >= dev_limit)
break;
- mirror = FF_LAYOUT_COMP(pls, i);
- if (!mirror || !mirror->mirror_ds)
+ if (!mirror->mirror_ds)
continue;
- dev = FF_LAYOUT_DEVID_NODE(pls, i);
- devinfo = &args->devinfo[*dev_count];
+ /* mirror refcount put in cleanup_layoutstats */
+ if (!atomic_inc_not_zero(&mirror->ref))
+ continue;
+ dev = &mirror->mirror_ds->id_node;
+ devinfo = &args->devinfo[i];
memcpy(&devinfo->dev_id, &dev->deviceid, NFS4_DEVICEID4_SIZE);
devinfo->offset = pls->pls_range.offset;
devinfo->length = pls->pls_range.length;
@@ -1911,24 +1984,25 @@ ff_layout_mirror_prepare_stats(struct nfs42_layoutstat_args *args,
devinfo->layout_type = LAYOUT_FLEX_FILES;
devinfo->layoutstats_encode = ff_layout_encode_layoutstats;
devinfo->layout_private = mirror;
- /* mirror refcount put in cleanup_layoutstats */
- atomic_inc(&mirror->ref);
- ++(*dev_count);
+ i++;
}
-
- return *dev_count < dev_limit;
+ return i;
}
static int
ff_layout_prepare_layoutstats(struct nfs42_layoutstat_args *args)
{
+ struct nfs4_flexfile_layout *ff_layout;
+ struct nfs4_ff_layout_mirror *mirror;
struct pnfs_layout_segment *pls;
int dev_count = 0;
spin_lock(&args->inode->i_lock);
- list_for_each_entry(pls, &NFS_I(args->inode)->layout->plh_segs, pls_list) {
- dev_count += FF_LAYOUT_MIRROR_COUNT(pls);
+ ff_layout = FF_LAYOUT_FROM_HDR(NFS_I(args->inode)->layout);
+ list_for_each_entry(mirror, &ff_layout->mirrors, mirrors) {
+ if (atomic_read(&mirror->ref) != 0)
+ dev_count ++;
}
spin_unlock(&args->inode->i_lock);
/* For now, send at most PNFS_LAYOUTSTATS_MAXDEV statistics */
@@ -1937,20 +2011,14 @@ ff_layout_prepare_layoutstats(struct nfs42_layoutstat_args *args)
__func__, dev_count, PNFS_LAYOUTSTATS_MAXDEV);
dev_count = PNFS_LAYOUTSTATS_MAXDEV;
}
- args->devinfo = kmalloc(dev_count * sizeof(*args->devinfo), GFP_KERNEL);
+ args->devinfo = kmalloc_array(dev_count, sizeof(*args->devinfo), GFP_NOIO);
if (!args->devinfo)
return -ENOMEM;
dev_count = 0;
spin_lock(&args->inode->i_lock);
- list_for_each_entry(pls, &NFS_I(args->inode)->layout->plh_segs, pls_list) {
- if (!ff_layout_mirror_prepare_stats(args, pls, &dev_count,
- PNFS_LAYOUTSTATS_MAXDEV)) {
- break;
- }
- }
+ args->num_dev = ff_layout_mirror_prepare_stats(args, pls, dev_count);
spin_unlock(&args->inode->i_lock);
- args->num_dev = dev_count;
return 0;
}
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.h b/fs/nfs/flexfilelayout/flexfilelayout.h
index fe9d3ff7cf85..68cc0d9828f9 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.h
+++ b/fs/nfs/flexfilelayout/flexfilelayout.h
@@ -67,6 +67,8 @@ struct nfs4_ff_layoutstat {
};
struct nfs4_ff_layout_mirror {
+ struct pnfs_layout_hdr *layout;
+ struct list_head mirrors;
u32 ds_count;
u32 efficiency;
struct nfs4_ff_layout_ds *mirror_ds;
@@ -95,6 +97,7 @@ struct nfs4_ff_layout_segment {
struct nfs4_flexfile_layout {
struct pnfs_layout_hdr generic_hdr;
struct pnfs_ds_commit_info commit_info;
+ struct list_head mirrors;
struct list_head error_list; /* nfs4_ff_layout_ds_err */
};
--
2.4.3
next prev parent reply other threads:[~2015-08-25 1:43 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-08-25 1:43 [PATCH 1/4] NFSv4.1/flexfiles: Add refcounting to struct nfs4_ff_layout_mirror Trond Myklebust
2015-08-25 1:43 ` [PATCH 2/4] NFSv4.1/flexfiles: Remove mirror backpointer to lseg Trond Myklebust
2015-08-25 1:43 ` Trond Myklebust [this message]
2015-08-25 1:43 ` [PATCH 4/4] NFSv4.2/pnfs: Make the layoutstats timer configurable Trond Myklebust
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=1440466986-36826-3-git-send-email-trond.myklebust@primarydata.com \
--to=trond.myklebust@primarydata.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