@@ -310,6 +310,7 @@ extern int nfs_migrate_page(struct address_space *,
#endif
/* nfs4proc.c */
+extern void __nfs4_read_done_cb(struct nfs_read_data *);
extern void nfs4_reset_read(struct rpc_task *task, struct nfs_read_data *data);
extern int nfs4_init_client(struct nfs_client *clp,
const struct rpc_timeout *timeparms,
@@ -3174,6 +3174,11 @@ static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
return err;
}
+void __nfs4_read_done_cb(struct nfs_read_data *data)
+{
+ nfs_invalidate_atime(data->inode);
+}
+
static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_read_data *data)
{
struct nfs_server *server = NFS_SERVER(data->inode);
@@ -3183,7 +3188,7 @@ static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_read_data *data)
return -EAGAIN;
}
- nfs_invalidate_atime(data->inode);
+ __nfs4_read_done_cb(data);
if (task->tk_status > 0)
renew_lease(server, data->timestamp);
return 0;
@@ -1059,6 +1059,30 @@ pnfs_pageio_init_write(struct nfs_pageio_descriptor *pgio, struct inode *inode)
pgio->pg_test = (ld && ld->pg_test) ? pnfs_write_pg_test : NULL;
}
+/*
+ * Called by non rpc-based layout drivers
+ */
+int
+pnfs_ld_write_done(struct nfs_write_data *data)
+{
+ int status;
+
+ put_lseg(data->lseg);
+ data->lseg = NULL;
+ if (!data->pnfs_error) {
+ pnfs_set_layoutcommit(data);
+ data->mds_ops->rpc_call_done(NULL, data);
+ data->mds_ops->rpc_release(data);
+ return 0;
+ }
+
+ dprintk("%s: pnfs_error=%d, retry via MDS\n", __func__,
+ data->pnfs_error);
+ status = nfs_initiate_write(data, NFS_CLIENT(data->inode), data->mds_ops, NFS_FILE_SYNC);
+ return status ? : -EAGAIN;
+}
+EXPORT_SYMBOL_GPL(pnfs_ld_write_done);
+
enum pnfs_try_status
pnfs_try_to_write_data(struct nfs_write_data *wdata,
const struct rpc_call_ops *call_ops, int how)
@@ -1084,6 +1108,30 @@ pnfs_try_to_write_data(struct nfs_write_data *wdata,
}
/*
+ * Called by non rpc-based layout drivers
+ */
+int
+pnfs_ld_read_done(struct nfs_read_data *data)
+{
+ int status;
+
+ put_lseg(data->lseg);
+ data->lseg = NULL;
+ if (!data->pnfs_error) {
+ __nfs4_read_done_cb(data);
+ data->mds_ops->rpc_call_done(NULL, data);
+ data->mds_ops->rpc_release(data);
+ return 0;
+ }
+
+ dprintk("%s: pnfs_error=%d, retry via MDS\n", __func__,
+ data->pnfs_error);
+ status = nfs_initiate_read(data, NFS_CLIENT(data->inode), data->mds_ops);
+ return status ? : -EAGAIN;
+}
+EXPORT_SYMBOL_GPL(pnfs_ld_read_done);
+
+/*
* Call the appropriate parallel I/O subsystem read function.
*/
enum pnfs_try_status
@@ -166,6 +166,8 @@ void pnfs_roc_set_barrier(struct inode *ino, u32 barrier);
bool pnfs_roc_drain(struct inode *ino, u32 *barrier);
void pnfs_set_layoutcommit(struct nfs_write_data *wdata);
int pnfs_layoutcommit_inode(struct inode *inode, bool sync);
+int pnfs_ld_write_done(struct nfs_write_data *);
+int pnfs_ld_read_done(struct nfs_read_data *);
static inline int lo_fail_bit(u32 iomode)
{
@@ -1086,6 +1086,7 @@ struct nfs_read_data {
const struct rpc_call_ops *mds_ops;
int (*read_done_cb) (struct rpc_task *task, struct nfs_read_data *data);
__u64 mds_offset;
+ int pnfs_error;
struct page *page_array[NFS_PAGEVEC_SIZE];
};
@@ -1111,6 +1112,7 @@ struct nfs_write_data {
unsigned long timestamp; /* For lease renewal */
#endif
__u64 mds_offset; /* Filelayout dense stripe */
+ int pnfs_error;
struct page *page_array[NFS_PAGEVEC_SIZE];
};
Non-rpc layout driver such as for objects and blocks implement their own I/O path and error handling logic. Therefore bypass NFS-based error handling for these layout drivers. [get rid of PNFS_USE_RPC_CODE] [get rid of __nfs4_write_done_cb] [revert useless change in nfs4_write_done_cb] Signed-off-by: Benny Halevy <bhalevy@panasas.com> --- fs/nfs/internal.h | 1 + fs/nfs/nfs4proc.c | 7 +++++- fs/nfs/pnfs.c | 48 +++++++++++++++++++++++++++++++++++++++++++++++ fs/nfs/pnfs.h | 2 + include/linux/nfs_xdr.h | 2 + 5 files changed, 59 insertions(+), 1 deletions(-)