From: The pNFS Team <linux-nfs@xxxxxxxxxxxxxxx>
Signed-off-by: Andy Adamson <andros@xxxxxxxxxx>
---
fs/nfs/internal.h | 8 ++++
fs/nfs/nfs4proc.c | 14 ++++++-
fs/nfs/pnfs.c | 37 ++++++++++++++++
fs/nfs/pnfs.h | 3 +
fs/nfs/write.c | 99 ++++++++++++++++++++++++++++++++------------
include/linux/nfs4_pnfs.h | 7 +++
include/linux/nfs_iostat.h | 1 +
7 files changed, 141 insertions(+), 28 deletions(-)
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index 02f0da8..ae8b895 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -276,6 +276,14 @@ extern int pnfs_initiate_write(struct nfs_write_data *data,
struct rpc_clnt *clnt,
const struct rpc_call_ops *call_ops,
int how);
+extern int nfs_initiate_commit(struct nfs_write_data *data,
+ struct rpc_clnt *clnt,
+ const struct rpc_call_ops *call_ops,
+ int how);
+extern int pnfs_initiate_commit(struct nfs_write_data *data,
+ struct rpc_clnt *clnt,
+ const struct rpc_call_ops *call_ops,
+ int how, int pnfs);
extern void nfs_write_prepare(struct rpc_task *task, void *calldata);
extern void nfs_mark_list_commit(struct list_head *head);
#ifdef CONFIG_MIGRATION
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 44ffa33..55aba4c 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -3233,6 +3233,17 @@ static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_messag
static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
{
struct inode *inode = data->inode;
+ struct nfs_server *server = NFS_SERVER(data->inode);
+ struct nfs_client *client = server->nfs_client;
+
+#ifdef CONFIG_NFS_V4_1
+ /* Is this a DS session */
+ if (data->fldata.ds_nfs_client) {
+ dprintk("%s DS commit\n", __func__);
+ client = data->fldata.ds_nfs_client;
+ }
+#endif /* CONFIG_NFS_V4_1 */
+
if (!nfs4_sequence_done(task, &data->res.seq_res))
return -EAGAIN;
@@ -3240,7 +3251,8 @@ static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
nfs_restart_rpc(task, NFS_SERVER(inode)->nfs_client);
return -EAGAIN;
}
- nfs_refresh_inode(inode, data->res.fattr);
+ if (client == server->nfs_client)
+ nfs_refresh_inode(inode, data->res.fattr);
return 0;
}
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 424efce..393855e 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -260,6 +260,14 @@ pnfs_register_layoutdriver(struct pnfs_layoutdriver_type *ld_type)
return NULL;
}
+ if (!io_ops->read_pagelist || !io_ops->write_pagelist ||
+ !io_ops->commit) {
+ printk(KERN_ERR "%s Layout driver must provide "
+ "read_pagelist, write_pagelist, and commit.\n",
+ __func__);
+ return NULL;
+ }
+
pnfs_mod = kmalloc(sizeof(struct pnfs_module), GFP_KERNEL);
if (pnfs_mod != NULL) {
dprintk("%s Registering id:%u name:%s\n",
@@ -1323,6 +1331,35 @@ pnfs_try_to_read_data(struct nfs_read_data *rdata,
return trypnfs;
}
+enum pnfs_try_status
+pnfs_try_to_commit(struct nfs_write_data *data,
+ const struct rpc_call_ops *call_ops, int sync)
+{
+ struct inode *inode = data->inode;
+ struct nfs_server *nfss = NFS_SERVER(data->inode);
+ enum pnfs_try_status trypnfs;
+
+ dprintk("%s: Begin\n", __func__);
+
+ /* We need to account for possibility that
+ * each nfs_page can point to a different lseg (or be NULL).
+ * For the immediate case of whole-file-only layouts, we at
+ * least know there can be only a single lseg.
+ * We still have to account for the possibility of some being NULL.
+ * This will be done by passing the buck to the layout driver.
+ */
+ data->pdata.call_ops = call_ops;
+ data->pdata.how = sync;
+ data->pdata.lseg = NULL;
+ trypnfs = nfss->pnfs_curr_ld->ld_io_ops->commit(data, sync);
+ if (trypnfs == PNFS_NOT_ATTEMPTED)
+ _pnfs_clear_lseg_from_pages(&data->pages);
+ else
+ nfs_inc_stats(inode, NFSIOS_PNFS_COMMIT);
+ dprintk("%s End (trypnfs:%d)\n", __func__, trypnfs);
+ return trypnfs;
+}
+
/*
* Set up the argument/result storage required for the RPC call.
*/
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index b110f4e..80f67c7 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -50,6 +50,9 @@ void pnfs_layoutcommit_free(struct pnfs_layoutcommit_data *data);
int pnfs_layoutcommit_inode(struct inode *inode, int sync);
void pnfs_update_last_write(struct nfs_inode *nfsi, loff_t offset, size_t extent);
void pnfs_need_layoutcommit(struct nfs_inode *nfsi, struct nfs_open_context *ctx);
+unsigned int pnfs_getiosize(struct nfs_server *server);
+enum pnfs_try_status pnfs_try_to_commit(struct nfs_write_data *,
+ const struct rpc_call_ops *, int);
void pnfs_pageio_init_read(struct nfs_pageio_descriptor *, struct inode *,
struct nfs_open_context *, struct list_head *);
void pnfs_pageio_init_write(struct nfs_pageio_descriptor *, struct inode *);
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 18926d3..668c4c1 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -1311,40 +1311,72 @@ static void nfs_commitdata_release(void *data)
nfs_commit_free(wdata);
}
-/*
- * Set up the argument/result storage required for the RPC call.
- */
-static int nfs_commit_rpcsetup(struct list_head *head,
- struct nfs_write_data *data,
- int how)
+int nfs_initiate_commit(struct nfs_write_data *data,
+ struct rpc_clnt *clnt,
+ const struct rpc_call_ops *call_ops,
+ int how)
{
- struct nfs_page *first = nfs_list_entry(head->next);
- struct inode *inode = first->wb_context->path.dentry->d_inode;
+ struct inode *inode = data->inode;
int priority = flush_task_priority(how);
struct rpc_task *task;
struct rpc_message msg = {
.rpc_argp = &data->args,
.rpc_resp = &data->res,
- .rpc_cred = first->wb_context->cred,
+ .rpc_cred = data->cred,
};
struct rpc_task_setup task_setup_data = {
.task = &data->task,
- .rpc_client = NFS_CLIENT(inode),
+ .rpc_client = clnt,
.rpc_message = &msg,
- .callback_ops = &nfs_commit_ops,
+ .callback_ops = call_ops,
.callback_data = data,
.workqueue = nfsiod_workqueue,
.flags = RPC_TASK_ASYNC,
.priority = priority,
};
+ /* Set up the initial task struct. */
+ NFS_PROTO(inode)->commit_setup(data, &msg);
+
+ dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
+
+ task = rpc_run_task(&task_setup_data);
+ if (IS_ERR(task))
+ return PTR_ERR(task);
+ rpc_put_task(task);
+ return 0;
+}
+EXPORT_SYMBOL(nfs_initiate_commit);
+
+
+int pnfs_initiate_commit(struct nfs_write_data *data,
+ struct rpc_clnt *clnt,
+ const struct rpc_call_ops *call_ops,
+ int how, int pnfs)
+{
+ if (pnfs &&
+ (pnfs_try_to_commit(data, &nfs_commit_ops, how) == PNFS_ATTEMPTED))
+ return 0;
+ return nfs_initiate_commit(data, clnt, &nfs_commit_ops, how);
+}
+
+/*
+ * Set up the argument/result storage required for the RPC call.
+ */
+static int nfs_commit_rpcsetup(struct list_head *head,
+ struct nfs_write_data *data,
+ int how, int pnfs)
+{
+ struct nfs_page *first = nfs_list_entry(head->next);
+ struct inode *inode = first->wb_context->path.dentry->d_inode;
+
/* Set up the RPC argument and reply structs
* NB: take care not to mess about with data->commit et al. */
list_splice_init(head, &data->pages);
data->inode = inode;
- data->cred = msg.rpc_cred;
+ data->cred = first->wb_context->cred;
data->args.fh = NFS_FH(data->inode);
/* Note: we always request a commit of the entire inode */
@@ -1355,24 +1387,19 @@ static int nfs_commit_rpcsetup(struct list_head *head,
data->res.fattr = &data->fattr;
data->res.verf = &data->verf;
nfs_fattr_init(&data->fattr);
+ kref_init(&data->refcount);
+ data->parent = NULL;
+ data->args.context = first->wb_context; /* used by commit done */
- /* Set up the initial task struct. */
- NFS_PROTO(inode)->commit_setup(data, &msg);
-
- dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
-
- task = rpc_run_task(&task_setup_data);
- if (IS_ERR(task))
- return PTR_ERR(task);
- rpc_put_task(task);
- return 0;
+ return pnfs_initiate_commit(data, NFS_CLIENT(inode), &nfs_commit_ops,
+ how, pnfs);
}
/*
* Commit dirty pages
*/
static int
-nfs_commit_list(struct inode *inode, struct list_head *head, int how)
+nfs_commit_list(struct inode *inode, struct list_head *head, int how, int pnfs)
{
struct nfs_write_data *data;
struct nfs_page *req;
@@ -1383,7 +1410,7 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how)
goto out_bad;
/* Set up the argument struct */
- return nfs_commit_rpcsetup(head, data, how);
+ return nfs_commit_rpcsetup(head, data, how, pnfs);
out_bad:
while (!list_empty(head)) {
req = nfs_list_entry(head->next);
@@ -1413,6 +1440,19 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
return;
}
+static inline void nfs_commit_cleanup(struct kref *kref)
+{
+ struct nfs_write_data *data;
+
+ data = container_of(kref, struct nfs_write_data, refcount);
+ /* Clear lock only when all cloned commits are finished */
+ if (data->parent)
+ kref_put(&data->parent->refcount, nfs_commit_cleanup);
+ else
+ nfs_commit_clear_lock(NFS_I(data->inode));
+ nfs_commitdata_release(data);
+}
+
static void nfs_commit_release(void *calldata)
{
struct nfs_write_data *data = calldata;
@@ -1430,6 +1470,11 @@ static void nfs_commit_release(void *calldata)
req->wb_bytes,
(long long)req_offset(req));
if (status < 0) {
+ if (req->wb_lseg) {
+ nfs_mark_request_nopnfs(req);
+ nfs_mark_request_dirty(req);
+ goto next;
+ }
nfs_context_set_write_error(req->wb_context, status);
nfs_inode_remove_request(req);
dprintk(", error = %d\n", status);
@@ -1446,12 +1491,12 @@ static void nfs_commit_release(void *calldata)
}
/* We have a mismatch. Write the page again */
dprintk(" mismatch\n");
+ nfs_mark_request_nopnfs(req);
nfs_mark_request_dirty(req);
next:
nfs_clear_page_tag_locked(req);
}
- nfs_commit_clear_lock(NFS_I(data->inode));
- nfs_commitdata_release(calldata);
+ kref_put(&data->refcount, nfs_commit_cleanup);
}
static const struct rpc_call_ops nfs_commit_ops = {
@@ -1475,7 +1520,7 @@ int nfs_commit_inode(struct inode *inode, int how)
res = nfs_scan_commit(inode, &head, 0, 0, &use_pnfs);
spin_unlock(&inode->i_lock);
if (res) {
- int error = nfs_commit_list(inode, &head, how);
+ int error = nfs_commit_list(inode, &head, how, use_pnfs);
if (error < 0)
return error;
if (may_wait) {
diff --git a/include/linux/nfs4_pnfs.h b/include/linux/nfs4_pnfs.h
index b010ff1..ef160e6 100644
--- a/include/linux/nfs4_pnfs.h
+++ b/include/linux/nfs4_pnfs.h
@@ -124,6 +124,13 @@ struct layoutdriver_io_operations {
enum pnfs_try_status
(*write_pagelist) (struct nfs_write_data *nfs_data, unsigned nr_pages, int how);
+ /* Consistency ops */
+ /* 2 problems:
+ * 1) the page list contains nfs_pages, NOT pages
+ * 2) currently the NFS code doesn't create a page array (as it does with read/write)
+ */
+ enum pnfs_try_status
+ (*commit) (struct nfs_write_data *nfs_data, int how);
/* Layout information. For each inode, alloc_layout is executed once to retrieve an
* inode specific layout structure. Each subsequent layoutget operation results in
diff --git a/include/linux/nfs_iostat.h b/include/linux/nfs_iostat.h
index 8866bb3..f9b5f44 100644
--- a/include/linux/nfs_iostat.h
+++ b/include/linux/nfs_iostat.h
@@ -115,6 +115,7 @@ enum nfs_stat_eventcounters {
NFSIOS_DELAY,
NFSIOS_PNFS_READ,
NFSIOS_PNFS_WRITE,
+ NFSIOS_PNFS_COMMIT,
__NFSIOS_COUNTSMAX,
};
--
1.6.2.5
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
