From 0b7c01533aa9f4a228d07d2768d084acb3a387bc Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:39 -0400 Subject: NFS: add a struct nfs_commit_data to replace nfs_write_data in commits Commits don't need the vectors of pages, etc. that writes do. Split out a separate structure for the commit operation. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 4 ++-- include/linux/nfs_xdr.h | 45 ++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 44 insertions(+), 5 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 52a1bdb4ee2b..d5d68f322bf0 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -552,8 +552,8 @@ extern int nfs_wb_page(struct inode *inode, struct page* page); extern int nfs_wb_page_cancel(struct inode *inode, struct page* page); #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4) extern int nfs_commit_inode(struct inode *, int); -extern struct nfs_write_data *nfs_commitdata_alloc(void); -extern void nfs_commit_free(struct nfs_write_data *wdata); +extern struct nfs_commit_data *nfs_commitdata_alloc(void); +extern void nfs_commit_free(struct nfs_commit_data *data); #else static inline int nfs_commit_inode(struct inode *inode, int how) diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 7ba3551a0414..8fb036a0d489 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -518,6 +518,24 @@ struct nfs_writeres { struct nfs4_sequence_res seq_res; }; +/* + * Arguments to the commit call. + */ +struct nfs_commitargs { + struct nfs_fh *fh; + __u64 offset; + __u32 count; + const u32 *bitmask; + struct nfs4_sequence_args seq_args; +}; + +struct nfs_commitres { + struct nfs_fattr *fattr; + struct nfs_writeverf *verf; + const struct nfs_server *server; + struct nfs4_sequence_res seq_res; +}; + /* * Common arguments to the unlink call */ @@ -1171,6 +1189,8 @@ struct nfs_read_data { struct page *page_array[NFS_PAGEVEC_SIZE]; }; +struct nfs_direct_req; + struct nfs_write_data { struct rpc_task task; struct inode *inode; @@ -1186,7 +1206,6 @@ struct nfs_write_data { struct nfs_writeres res; /* result struct */ struct pnfs_layout_segment *lseg; struct nfs_client *ds_clp; /* pNFS data server */ - int ds_commit_index; const struct rpc_call_ops *mds_ops; int (*write_done_cb) (struct rpc_task *task, struct nfs_write_data *data); #ifdef CONFIG_NFS_V4 @@ -1197,6 +1216,25 @@ struct nfs_write_data { struct page *page_array[NFS_PAGEVEC_SIZE]; }; +struct nfs_commit_data { + struct rpc_task task; + struct inode *inode; + struct rpc_cred *cred; + struct nfs_fattr fattr; + struct nfs_writeverf verf; + struct list_head pages; /* Coalesced requests we wish to flush */ + struct list_head list; /* lists of struct nfs_write_data */ + struct nfs_direct_req *dreq; /* O_DIRECT request */ + struct nfs_commitargs args; /* argument struct */ + struct nfs_commitres res; /* result struct */ + struct nfs_open_context *context; + struct pnfs_layout_segment *lseg; + struct nfs_client *ds_clp; /* pNFS data server */ + int ds_commit_index; + const struct rpc_call_ops *mds_ops; + int (*commit_done_cb) (struct rpc_task *task, struct nfs_commit_data *data); +}; + struct nfs_unlinkdata { struct hlist_node list; struct nfs_removeargs args; @@ -1277,8 +1315,9 @@ struct nfs_rpc_ops { void (*write_setup) (struct nfs_write_data *, struct rpc_message *); void (*write_rpc_prepare)(struct rpc_task *, struct nfs_write_data *); int (*write_done) (struct rpc_task *, struct nfs_write_data *); - void (*commit_setup) (struct nfs_write_data *, struct rpc_message *); - int (*commit_done) (struct rpc_task *, struct nfs_write_data *); + void (*commit_setup) (struct nfs_commit_data *, struct rpc_message *); + void (*commit_rpc_prepare)(struct rpc_task *, struct nfs_commit_data *); + int (*commit_done) (struct rpc_task *, struct nfs_commit_data *); int (*lock)(struct file *, int, struct file_lock *); int (*lock_check_bounds)(const struct file_lock *); void (*clear_acl_cache)(struct inode *); -- cgit v1.2.3 From cd841605f7a721878d8a2d1362484723d8abf569 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:44 -0400 Subject: NFS: create common nfs_pgio_header for both read and write In order to avoid duplicating all the data in nfs_read_data whenever we split it up into multiple RPC calls (either due to a short read result or due to rsize < PAGE_SIZE), we split out the bits that are the same per RPC call into a separate "header" structure. The goal this patch moves towards is to have a single header refcounted by several rpc_data structures. Thus, want to always refer from rpc_data to the header, and not the other way. This patch comes close to that ideal, but the directio code currently needs some special casing, isolated in the nfs_direct_[read_write]hdr_release() functions. This will be dealt with in a future patch. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 12 ------------ include/linux/nfs_xdr.h | 48 +++++++++++++++++++++++++++--------------------- 2 files changed, 27 insertions(+), 33 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index d5d68f322bf0..8d3a2b804201 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -568,12 +568,6 @@ nfs_have_writebacks(struct inode *inode) return NFS_I(inode)->npages != 0; } -/* - * Allocate nfs_write_data structures - */ -extern struct nfs_write_data *nfs_writedata_alloc(unsigned int npages); -extern void nfs_writedata_free(struct nfs_write_data *); - /* * linux/fs/nfs/read.c */ @@ -584,12 +578,6 @@ extern int nfs_readpage_result(struct rpc_task *, struct nfs_read_data *); extern int nfs_readpage_async(struct nfs_open_context *, struct inode *, struct page *); -/* - * Allocate nfs_read_data structures - */ -extern struct nfs_read_data *nfs_readdata_alloc(unsigned int npages); -extern void nfs_readdata_free(struct nfs_read_data *); - /* * linux/fs/nfs3proc.c */ diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 8fb036a0d489..fee324175391 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1168,52 +1168,58 @@ struct nfs_page; #define NFS_PAGEVEC_SIZE (8U) struct nfs_read_data { + struct nfs_pgio_header *header; + struct list_head list; struct rpc_task task; - struct inode *inode; - struct rpc_cred *cred; struct nfs_fattr fattr; /* fattr storage */ - struct list_head pages; /* Coalesced read requests */ - struct list_head list; /* lists of struct nfs_read_data */ - struct nfs_page *req; /* multi ops per nfs_page */ struct page **pagevec; unsigned int npages; /* Max length of pagevec */ struct nfs_readargs args; struct nfs_readres res; unsigned long timestamp; /* For lease renewal */ - struct pnfs_layout_segment *lseg; - struct nfs_client *ds_clp; /* pNFS data server */ - 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]; + struct nfs_client *ds_clp; /* pNFS data server */ +}; + +struct nfs_pgio_header { + struct inode *inode; + struct rpc_cred *cred; + struct list_head pages; + struct nfs_page *req; + struct pnfs_layout_segment *lseg; + const struct rpc_call_ops *mds_ops; + int pnfs_error; +}; + +struct nfs_read_header { + struct nfs_pgio_header header; + struct nfs_read_data rpc_data; }; struct nfs_direct_req; struct nfs_write_data { + struct nfs_pgio_header *header; + struct list_head list; struct rpc_task task; - struct inode *inode; - struct rpc_cred *cred; struct nfs_fattr fattr; struct nfs_writeverf verf; - struct list_head pages; /* Coalesced requests we wish to flush */ - struct list_head list; /* lists of struct nfs_write_data */ - struct nfs_page *req; /* multi ops per nfs_page */ struct page **pagevec; unsigned int npages; /* Max length of pagevec */ struct nfs_writeargs args; /* argument struct */ struct nfs_writeres res; /* result struct */ - struct pnfs_layout_segment *lseg; - struct nfs_client *ds_clp; /* pNFS data server */ - const struct rpc_call_ops *mds_ops; - int (*write_done_cb) (struct rpc_task *task, struct nfs_write_data *data); -#ifdef CONFIG_NFS_V4 unsigned long timestamp; /* For lease renewal */ -#endif + int (*write_done_cb) (struct rpc_task *task, struct nfs_write_data *data); __u64 mds_offset; /* Filelayout dense stripe */ - int pnfs_error; struct page *page_array[NFS_PAGEVEC_SIZE]; + struct nfs_client *ds_clp; /* pNFS data server */ +}; + +struct nfs_write_header { + struct nfs_pgio_header header; + struct nfs_write_data rpc_data; }; struct nfs_commit_data { -- cgit v1.2.3 From 30dd374f6fc1b202db3a1b57b61afff1326bad92 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:45 -0400 Subject: NFS: create struct nfs_page_array Both nfs_read_data and nfs_write_data devote several fields which can be combined into a single shared struct. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index fee324175391..e34beaf86e9c 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1167,19 +1167,23 @@ struct nfs_page; #define NFS_PAGEVEC_SIZE (8U) +struct nfs_page_array { + struct page **pagevec; + unsigned int npages; /* Max length of pagevec */ + struct page *page_array[NFS_PAGEVEC_SIZE]; +}; + struct nfs_read_data { struct nfs_pgio_header *header; struct list_head list; struct rpc_task task; struct nfs_fattr fattr; /* fattr storage */ - struct page **pagevec; - unsigned int npages; /* Max length of pagevec */ struct nfs_readargs args; struct nfs_readres res; unsigned long timestamp; /* For lease renewal */ int (*read_done_cb) (struct rpc_task *task, struct nfs_read_data *data); __u64 mds_offset; - struct page *page_array[NFS_PAGEVEC_SIZE]; + struct nfs_page_array pages; struct nfs_client *ds_clp; /* pNFS data server */ }; @@ -1206,14 +1210,12 @@ struct nfs_write_data { struct rpc_task task; struct nfs_fattr fattr; struct nfs_writeverf verf; - struct page **pagevec; - unsigned int npages; /* Max length of pagevec */ struct nfs_writeargs args; /* argument struct */ struct nfs_writeres res; /* result struct */ unsigned long timestamp; /* For lease renewal */ int (*write_done_cb) (struct rpc_task *task, struct nfs_write_data *data); __u64 mds_offset; /* Filelayout dense stripe */ - struct page *page_array[NFS_PAGEVEC_SIZE]; + struct nfs_page_array pages; struct nfs_client *ds_clp; /* pNFS data server */ }; -- cgit v1.2.3 From 4db6e0b74c0f6dfc2f9c0690e8df512e3b635983 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:46 -0400 Subject: NFS: merge _full and _partial read rpc_ops Decouple nfs_pgio_header and nfs_read_data, and have (possibly multiple) nfs_read_datas each take a refcount on nfs_pgio_header. For the moment keeps nfs_read_header as a way to preallocate a single nfs_read_data with the nfs_pgio_header. The code doesn't need this, and would be prettier without, but given the amount of churn I am already introducing I didn't want to play with tuning new mempools. This also fixes bug in pnfs_ld_handle_read_error. In the case of desc->pg_bsize < PAGE_CACHE_SIZE, the pages list was empty, causing replay attempt to do nothing. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_page.h | 1 - include/linux/nfs_xdr.h | 16 ++++++++++++++++ 2 files changed, 16 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index eac30d6bec17..5c520344d8ad 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -27,7 +27,6 @@ enum { PG_CLEAN, PG_NEED_COMMIT, PG_NEED_RESCHED, - PG_PARTIAL_READ_FAILED, PG_COMMIT_TO_DS, }; diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index e34beaf86e9c..164862148ba0 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1187,14 +1187,30 @@ struct nfs_read_data { struct nfs_client *ds_clp; /* pNFS data server */ }; +/* used as flag bits in nfs_pgio_header */ +enum { + NFS_IOHDR_ERROR = 0, + NFS_IOHDR_EOF, + NFS_IOHDR_REDO, +}; + struct nfs_pgio_header { struct inode *inode; struct rpc_cred *cred; struct list_head pages; + struct list_head rpc_list; + atomic_t refcnt; struct nfs_page *req; struct pnfs_layout_segment *lseg; + loff_t io_start; const struct rpc_call_ops *mds_ops; + void (*release) (struct nfs_pgio_header *hdr); + spinlock_t lock; + /* fields protected by lock */ int pnfs_error; + int error; /* merge with pnfs_error */ + unsigned long good_bytes; /* boundary of good data */ + unsigned long flags; }; struct nfs_read_header { -- cgit v1.2.3 From 6c75dc0d498caa402fb17b1bf769835a9db875c8 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:47 -0400 Subject: NFS: merge _full and _partial write rpc_ops Decouple nfs_pgio_header and nfs_write_data, and have (possibly multiple) nfs_write_datas each take a refcount on nfs_pgio_header. For the moment keeps nfs_write_header as a way to preallocate a single nfs_write_data with the nfs_pgio_header. The code doesn't need this, and would be prettier without, but given the amount of churn I am already introducing I didn't want to play with tuning new mempools. This also fixes bug in pnfs_ld_handle_write_error. In the case of desc->pg_bsize < PAGE_CACHE_SIZE, the pages list was empty, causing replay attempt to do nothing. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 164862148ba0..0d17db7973de 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1192,6 +1192,8 @@ enum { NFS_IOHDR_ERROR = 0, NFS_IOHDR_EOF, NFS_IOHDR_REDO, + NFS_IOHDR_NEED_COMMIT, + NFS_IOHDR_NEED_RESCHED, }; struct nfs_pgio_header { -- cgit v1.2.3 From 061ae2edb7375ab6776468b075da71008a098b55 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:48 -0400 Subject: NFS: create completion structure to pass into page_init functions Factors out the code that will need to change when directio starts using these code paths. This will allow directio to use the generic pagein and flush routines Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_page.h | 2 ++ include/linux/nfs_xdr.h | 6 ++++++ 2 files changed, 8 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index 5c520344d8ad..bc5b7a5e787e 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -67,6 +67,7 @@ struct nfs_pageio_descriptor { int pg_ioflags; int pg_error; const struct rpc_call_ops *pg_rpc_callops; + const struct nfs_pgio_completion_ops *pg_completion_ops; struct pnfs_layout_segment *pg_lseg; }; @@ -83,6 +84,7 @@ extern void nfs_release_request(struct nfs_page *req); extern void nfs_pageio_init(struct nfs_pageio_descriptor *desc, struct inode *inode, const struct nfs_pageio_ops *pg_ops, + const struct nfs_pgio_completion_ops *compl_ops, size_t bsize, int how); extern int nfs_pageio_add_request(struct nfs_pageio_descriptor *, diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 0d17db7973de..6fa1d2278c9d 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1207,6 +1207,7 @@ struct nfs_pgio_header { loff_t io_start; const struct rpc_call_ops *mds_ops; void (*release) (struct nfs_pgio_header *hdr); + const struct nfs_pgio_completion_ops *completion_ops; spinlock_t lock; /* fields protected by lock */ int pnfs_error; @@ -1261,6 +1262,11 @@ struct nfs_commit_data { int (*commit_done_cb) (struct rpc_task *task, struct nfs_commit_data *data); }; +struct nfs_pgio_completion_ops { + void (*error_cleanup)(struct list_head *head); + void (*completion)(struct nfs_pgio_header *hdr); +}; + struct nfs_unlinkdata { struct hlist_node list; struct nfs_removeargs args; -- cgit v1.2.3 From 9533da2979757258d3fd5429d830a297013d69ed Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:49 -0400 Subject: NFS: remove unused wb_complete field from struct nfs_page Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_page.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index bc5b7a5e787e..0a5b63f16116 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -36,7 +36,6 @@ struct nfs_page { struct page *wb_page; /* page to read in/write out */ struct nfs_open_context *wb_context; /* File state context info */ struct nfs_lock_context *wb_lock_context; /* lock context info */ - atomic_t wb_complete; /* i/os we're waiting for */ pgoff_t wb_index; /* Offset >> PAGE_CACHE_SHIFT */ unsigned int wb_offset, /* Offset & ~PAGE_CACHE_MASK */ wb_pgbase, /* Start of page data */ -- cgit v1.2.3 From 584aa810b6240d88c28113a90c5029449814a3b5 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:51 -0400 Subject: NFS: rewrite directio read to use async coalesce code This also has the advantage that it allows directio to use pnfs. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_page.h | 1 + include/linux/nfs_xdr.h | 4 ++-- 2 files changed, 3 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index 0a5b63f16116..f9ee9eba7f88 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -68,6 +68,7 @@ struct nfs_pageio_descriptor { const struct rpc_call_ops *pg_rpc_callops; const struct nfs_pgio_completion_ops *pg_completion_ops; struct pnfs_layout_segment *pg_lseg; + struct nfs_direct_req *pg_dreq; }; #define NFS_WBACK_BUSY(req) (test_bit(PG_BUSY,&(req)->wb_flags)) diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 6fa1d2278c9d..38687b87ca9b 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1208,6 +1208,7 @@ struct nfs_pgio_header { const struct rpc_call_ops *mds_ops; void (*release) (struct nfs_pgio_header *hdr); const struct nfs_pgio_completion_ops *completion_ops; + struct nfs_direct_req *dreq; spinlock_t lock; /* fields protected by lock */ int pnfs_error; @@ -1221,8 +1222,6 @@ struct nfs_read_header { struct nfs_read_data rpc_data; }; -struct nfs_direct_req; - struct nfs_write_data { struct nfs_pgio_header *header; struct list_head list; @@ -1264,6 +1263,7 @@ struct nfs_commit_data { struct nfs_pgio_completion_ops { void (*error_cleanup)(struct list_head *head); + void (*init_hdr)(struct nfs_pgio_header *hdr); void (*completion)(struct nfs_pgio_header *hdr); }; -- cgit v1.2.3 From ea2cf2282b4278461266013e9c002ee1c66700ff Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:53 -0400 Subject: NFS: create struct nfs_commit_info It is COMMIT that is handled the most differently between the paged and direct paths. Create a structure that encapsulates everything either path needs to know about the commit state. We could use void to hide some of the layout driver stuff, but Trond suggests pulling it out to ensure type checking, given the huge changes being made, and the fact that it doesn't interfere with other drivers. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 5 +---- include/linux/nfs_xdr.h | 27 +++++++++++++++++++++++++++ 2 files changed, 28 insertions(+), 4 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 8d3a2b804201..8a88c16662c5 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -179,8 +179,7 @@ struct nfs_inode { __be32 cookieverf[2]; unsigned long npages; - unsigned long ncommit; - struct list_head commit_list; + struct nfs_mds_commit_info commit_info; /* Open contexts for shared mmap writes */ struct list_head open_files; @@ -201,7 +200,6 @@ struct nfs_inode { /* pNFS layout information */ struct pnfs_layout_hdr *layout; - atomic_t commits_outstanding; #endif /* CONFIG_NFS_V4*/ #ifdef CONFIG_NFS_FSCACHE struct fscache_cookie *fscache; @@ -230,7 +228,6 @@ struct nfs_inode { #define NFS_INO_FSCACHE (5) /* inode can be cached by FS-Cache */ #define NFS_INO_FSCACHE_LOCK (6) /* FS-Cache cookie management lock */ #define NFS_INO_COMMIT (7) /* inode is committing unstable writes */ -#define NFS_INO_PNFS_COMMIT (8) /* use pnfs code for commit */ #define NFS_INO_LAYOUTCOMMIT (9) /* layoutcommit required */ #define NFS_INO_LAYOUTCOMMITTING (10) /* layoutcommit inflight */ diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 38687b87ca9b..224e1e82670c 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1079,6 +1079,21 @@ struct nfstime4 { }; #ifdef CONFIG_NFS_V4_1 + +struct pnfs_commit_bucket { + struct list_head written; + struct list_head committing; + struct pnfs_layout_segment *wlseg; + struct pnfs_layout_segment *clseg; +}; + +struct pnfs_ds_commit_info { + int nwritten; + int ncommitting; + int nbuckets; + struct pnfs_commit_bucket *buckets; +}; + #define NFS4_EXCHANGE_ID_LEN (48) struct nfs41_exchange_id_args { struct nfs_client *client; @@ -1242,6 +1257,18 @@ struct nfs_write_header { struct nfs_write_data rpc_data; }; +struct nfs_mds_commit_info { + atomic_t rpcs_out; + unsigned long ncommit; + struct list_head list; +}; + +struct nfs_commit_info { + spinlock_t *lock; + struct nfs_mds_commit_info *mds; + struct pnfs_ds_commit_info *ds; +}; + struct nfs_commit_data { struct rpc_task task; struct inode *inode; -- cgit v1.2.3 From f453a54a01c7c0453ad9550906e3d2663dd486ac Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:54 -0400 Subject: NFS: create nfs_commit_completion_ops Factors out the code that needs to change when directio starts using these code paths. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 224e1e82670c..0e8b88ad9ae2 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1263,10 +1263,18 @@ struct nfs_mds_commit_info { struct list_head list; }; +struct nfs_commit_data; +struct nfs_inode; +struct nfs_commit_completion_ops { + void (*error_cleanup) (struct nfs_inode *nfsi); + void (*completion) (struct nfs_commit_data *data); +}; + struct nfs_commit_info { spinlock_t *lock; struct nfs_mds_commit_info *mds; struct pnfs_ds_commit_info *ds; + const struct nfs_commit_completion_ops *completion_ops; }; struct nfs_commit_data { @@ -1285,6 +1293,7 @@ struct nfs_commit_data { struct nfs_client *ds_clp; /* pNFS data server */ int ds_commit_index; const struct rpc_call_ops *mds_ops; + const struct nfs_commit_completion_ops *completion_ops; int (*commit_done_cb) (struct rpc_task *task, struct nfs_commit_data *data); }; -- cgit v1.2.3 From b359f9d09bcbaede09243cfe844172ba055d89fd Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Fri, 20 Apr 2012 14:47:55 -0400 Subject: NFS: add dreq to nfs_commit_info Need this to pass into nfs_commitdata_init, in order to keep data->dreq accurate. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 0e8b88ad9ae2..5f563bd113e8 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1274,6 +1274,7 @@ struct nfs_commit_info { spinlock_t *lock; struct nfs_mds_commit_info *mds; struct pnfs_ds_commit_info *ds; + struct nfs_direct_req *dreq; /* O_DIRECT request */ const struct nfs_commit_completion_ops *completion_ops; }; -- cgit v1.2.3 From 2671bfc3beb44e70636bd0208274426db57f73b5 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Fri, 27 Apr 2012 13:27:44 -0400 Subject: NFS: Remove secinfo knowledge out of the generic client And also remove the unneeded rpc_op. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 5f563bd113e8..eb1f143042f4 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1397,7 +1397,6 @@ struct nfs_rpc_ops { struct iattr *iattr); int (*init_client) (struct nfs_client *, const struct rpc_timeout *, const char *, rpc_authflavor_t, int); - int (*secinfo)(struct inode *, const struct qstr *, struct nfs4_secinfo_flavors *); }; /* -- cgit v1.2.3 From 281cad46b34db4dbb1d1e603f7b9cfe25d1ae7c9 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Fri, 27 Apr 2012 13:27:45 -0400 Subject: NFS: Create a submount rpc_op This simplifies the code for v2 and v3 and gives v4 a chance to decide on referrals without needing to modify the generic client. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index eb1f143042f4..4dada94eba7d 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1341,6 +1341,8 @@ struct nfs_rpc_ops { int (*getroot) (struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *); + struct vfsmount *(*submount) (struct nfs_server *, struct dentry *, + struct nfs_fh *, struct nfs_fattr *); int (*getattr) (struct nfs_server *, struct nfs_fh *, struct nfs_fattr *); int (*setattr) (struct dentry *, struct nfs_fattr *, -- cgit v1.2.3 From 80a16b21a81eb639f0b726549f4c46c0e9aff92e Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Fri, 27 Apr 2012 13:27:46 -0400 Subject: NFS: Remove extra rpc_clnt argument to proc_lookup Now that I'm doing secinfo automatically in the v4 code this extra argument isn't needed. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 4dada94eba7d..c940d46eb423 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1347,7 +1347,7 @@ struct nfs_rpc_ops { struct nfs_fattr *); int (*setattr) (struct dentry *, struct nfs_fattr *, struct iattr *); - int (*lookup) (struct rpc_clnt *clnt, struct inode *, struct qstr *, + int (*lookup) (struct inode *, struct qstr *, struct nfs_fh *, struct nfs_fattr *); int (*access) (struct inode *, struct nfs_access_entry *); int (*readlink)(struct inode *, struct page *, unsigned int, -- cgit v1.2.3 From 4f97615d19c370d1d907ef37f8bcd9c3672851ca Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 30 Apr 2012 18:39:20 -0400 Subject: NFS: Fix a compile issue when CONFIG_NFS_V4_1 is undefined struct nfs_direct_req can't compile when struct pnfs_ds_commit_info is undefined. Reported-by: Bryan Schumaker Signed-off-by: Trond Myklebust Cc: Fred Isaman --- include/linux/nfs_xdr.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index c940d46eb423..6deb8f097c42 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1176,6 +1176,11 @@ struct nfs41_free_stateid_res { struct nfs4_sequence_res seq_res; }; +#else + +struct pnfs_ds_commit_info { +}; + #endif /* CONFIG_NFS_V4_1 */ struct nfs_page; -- cgit v1.2.3 From 3a1556e8662cc425c433b463fcdae138908ca467 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 27 Apr 2012 13:48:18 -0400 Subject: NFSv2/v3: Simulate the change attribute Use the ctime to simulate a change attribute for NFSv2 and NFSv3. Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 6deb8f097c42..bc3680885428 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -106,14 +106,14 @@ struct nfs_fattr { | NFS_ATTR_FATTR_FILEID \ | NFS_ATTR_FATTR_ATIME \ | NFS_ATTR_FATTR_MTIME \ - | NFS_ATTR_FATTR_CTIME) + | NFS_ATTR_FATTR_CTIME \ + | NFS_ATTR_FATTR_CHANGE) #define NFS_ATTR_FATTR_V2 (NFS_ATTR_FATTR \ | NFS_ATTR_FATTR_BLOCKS_USED) #define NFS_ATTR_FATTR_V3 (NFS_ATTR_FATTR \ | NFS_ATTR_FATTR_SPACE_USED) #define NFS_ATTR_FATTR_V4 (NFS_ATTR_FATTR \ - | NFS_ATTR_FATTR_SPACE_USED \ - | NFS_ATTR_FATTR_CHANGE) + | NFS_ATTR_FATTR_SPACE_USED) /* * Info on the file system -- cgit v1.2.3 From 90ff0c548d1220d31f80e498b587393895705e6c Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 27 Apr 2012 13:48:18 -0400 Subject: NFSv4: Simplify the NFSv4 OPEN compound Get rid of the post-op GETATTR on the directory in order to reduce the amount of processing done on the server. The cost is that if we later need to stat() the directory, then we know that the ctime and mtime are likely to be invalid. Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 2 -- 1 file changed, 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index bc3680885428..92a929fc97c8 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -338,7 +338,6 @@ struct nfs_openargs { const struct qstr * name; const struct nfs_server *server; /* Needed for ID mapping */ const u32 * bitmask; - const u32 * dir_bitmask; __u32 claim; struct nfs4_sequence_args seq_args; }; @@ -349,7 +348,6 @@ struct nfs_openres { struct nfs4_change_info cinfo; __u32 rflags; struct nfs_fattr * f_attr; - struct nfs_fattr * dir_attr; struct nfs_seqid * seqid; const struct nfs_server *server; fmode_t delegation_type; -- cgit v1.2.3 From 7c317fcfbae773e493ecee1c53738db774b1d0ca Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 27 Apr 2012 13:48:18 -0400 Subject: NFSv4: Simplify the NFSv4 CREATE compound Get rid of the post-op GETATTR on the directory in order to reduce the amount of processing done on the server. Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 92a929fc97c8..696a17e047be 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -855,7 +855,6 @@ struct nfs4_create_res { struct nfs_fh * fh; struct nfs_fattr * fattr; struct nfs4_change_info dir_cinfo; - struct nfs_fattr * dir_fattr; struct nfs4_sequence_res seq_res; }; -- cgit v1.2.3 From 778d28172f710184855bcfeadcdd6b46997c4de2 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 27 Apr 2012 13:48:19 -0400 Subject: NFSv4: Simplify the NFSv4 REMOVE, LINK and RENAME compounds Get rid of the post-op GETATTR on the directory in order to reduce the amount of processing done on the server. Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 2 -- 1 file changed, 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 696a17e047be..2e53a3f1d2ff 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -540,7 +540,6 @@ struct nfs_commitres { struct nfs_removeargs { const struct nfs_fh *fh; struct qstr name; - const u32 * bitmask; struct nfs4_sequence_args seq_args; }; @@ -559,7 +558,6 @@ struct nfs_renameargs { const struct nfs_fh *new_dir; const struct qstr *old_name; const struct qstr *new_name; - const u32 *bitmask; struct nfs4_sequence_args seq_args; }; -- cgit v1.2.3 From d69ee9b85541a69a1092f5da675bd23256dc62af Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 1 May 2012 17:37:59 -0400 Subject: NFS: Adapt readdirplus to application usage patterns While the use of READDIRPLUS is significantly more efficient than READDIR followed by many LOOKUP calls, it is still less efficient than just READDIR if the attributes are not required. This patch tracks when lookups are attempted on the directory, and uses that information to selectively disable READDIRPLUS on that directory. The first 'readdir' call is always served using READDIRPLUS. Subsequent calls only use READDIRPLUS if there was a successful lookup or revalidation on a child in the mean time. Credit for the original idea should go to Neil Brown. See: http://www.spinics.net/lists/linux-nfs/msg19996.html However, the implementation in this patch differs from Neil's in that it focuses on tracking lookups rather than calls to stat(). Signed-off-by: Trond Myklebust Cc: Neil Brown --- include/linux/nfs_fs.h | 5 ----- 1 file changed, 5 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 8a88c16662c5..6cc7dbaf0695 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -314,11 +314,6 @@ static inline int nfs_server_capable(struct inode *inode, int cap) return NFS_SERVER(inode)->caps & cap; } -static inline int NFS_USE_READDIRPLUS(struct inode *inode) -{ - return test_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags); -} - static inline void nfs_set_verifier(struct dentry * dentry, unsigned long verf) { dentry->d_time = verf; -- cgit v1.2.3 From 3aff4ebb95b20ad8db2c1447e8c52097d89af5a7 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 9 May 2012 14:30:35 -0400 Subject: NFS: Prevent a deadlock in the new writeback code We have to unlock the nfs_page before we call nfs_end_page_writeback to avoid races with functions that expect the page to be unlocked when PG_locked and PG_writeback are not set. The problem is that nfs_unlock_request also releases the nfs_page, causing a deadlock if the release of the nfs_open_context triggers an iput() while the PG_writeback flag is still set... The solution is to separate the unlocking and release of the nfs_page, so that we can do the former before nfs_end_page_writeback and the latter after. Signed-off-by: Trond Myklebust Cc: Fred Isaman --- include/linux/nfs_page.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index f9ee9eba7f88..ef7504215446 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -96,6 +96,7 @@ extern bool nfs_generic_pg_test(struct nfs_pageio_descriptor *desc, struct nfs_page *req); extern int nfs_wait_on_request(struct nfs_page *); extern void nfs_unlock_request(struct nfs_page *req); +extern void nfs_unlock_request_dont_release(struct nfs_page *req); /* * Lock the page of an asynchronous request without getting a new reference -- cgit v1.2.3 From 7ad84aa9448571678c243f0c5ef383fbe5b50f4f Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 9 May 2012 13:19:15 -0400 Subject: NFS: Clean up - simplify nfs_lock_request() We only have two places where we need to grab a reference when trying to lock the nfs_page. We're better off making that explicit. Signed-off-by: Trond Myklebust Cc: Fred Isaman --- include/linux/nfs_page.h | 14 ++------------ 1 file changed, 2 insertions(+), 12 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index ef7504215446..263f30a5e10d 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -99,24 +99,14 @@ extern void nfs_unlock_request(struct nfs_page *req); extern void nfs_unlock_request_dont_release(struct nfs_page *req); /* - * Lock the page of an asynchronous request without getting a new reference + * Lock the page of an asynchronous request */ -static inline int -nfs_lock_request_dontget(struct nfs_page *req) -{ - return !test_and_set_bit(PG_BUSY, &req->wb_flags); -} - static inline int nfs_lock_request(struct nfs_page *req) { - if (test_and_set_bit(PG_BUSY, &req->wb_flags)) - return 0; - kref_get(&req->wb_kref); - return 1; + return !test_and_set_bit(PG_BUSY, &req->wb_flags); } - /** * nfs_list_add_request - Insert a request into a list * @req: request -- cgit v1.2.3 From 1d1afcbc294cc7c788eb5c7b6b98e8d63caf002c Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 9 May 2012 14:04:55 -0400 Subject: NFS: Clean up - Rename nfs_unlock_request and nfs_unlock_request_dont_release Function rename to ensure that the functionality of nfs_unlock_request() mirrors that of nfs_lock_request(). Then let nfs_unlock_and_release_request() do the work of what used to be called nfs_unlock_request()... Signed-off-by: Trond Myklebust Cc: Fred Isaman --- include/linux/nfs_page.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index 263f30a5e10d..88d166b555e8 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -96,7 +96,7 @@ extern bool nfs_generic_pg_test(struct nfs_pageio_descriptor *desc, struct nfs_page *req); extern int nfs_wait_on_request(struct nfs_page *); extern void nfs_unlock_request(struct nfs_page *req); -extern void nfs_unlock_request_dont_release(struct nfs_page *req); +extern void nfs_unlock_and_release_request(struct nfs_page *req); /* * Lock the page of an asynchronous request -- cgit v1.2.3 From e3c0fb7ef515852619932b0da993baa2d107684d Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:45:24 -0400 Subject: NFS: Add NFSDBG_STATE fs/nfs/nfs4state.c does not yet have any dprintk() call sites, and I'm about to introduce some. We will need a new flag for enabling them. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 6cc7dbaf0695..80a9385b88ab 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -634,6 +634,7 @@ nfs_fileid_to_ino_t(u64 fileid) #define NFSDBG_FSCACHE 0x0800 #define NFSDBG_PNFS 0x1000 #define NFSDBG_PNFS_LD 0x2000 +#define NFSDBG_STATE 0x4000 #define NFSDBG_ALL 0xFFFF #ifdef __KERNEL__ -- cgit v1.2.3 From 722baafc9e638714a69aa66e9ed24ef961ff350c Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:44:22 -0400 Subject: NFS: Fix comment misspelling in struct nfs_client definition Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 7073fc74481c..5498e9d9ba84 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -79,7 +79,7 @@ struct nfs_client { u32 cl_seqid; /* The flags used for obtaining the clientid during EXCHANGE_ID */ u32 cl_exchange_flags; - struct nfs4_session *cl_session; /* sharred session */ + struct nfs4_session *cl_session; /* shared session */ #endif /* CONFIG_NFS_V4 */ #ifdef CONFIG_NFS_FSCACHE -- cgit v1.2.3 From 79d4e1f0d8910f0214a57832ca6d589640d572c0 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:44:31 -0400 Subject: NFS: Use proper naming conventions for NFSv4.1 server scope fields Clean up: When naming fields and data types, follow established conventions to facilitate accurate grep/cscope searches. Additionally, for consistency, move the scope field into the NFSv4- specific part of the nfs_client, and free that memory in the logic that shuts down NFSv4 nfs_clients. Introduced by commit 99fe60d0 "nfs41: exchange_id operation", April 1 2009. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 4 ++-- include/linux/nfs_xdr.h | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 5498e9d9ba84..900d733668eb 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -17,7 +17,7 @@ struct nfs4_sequence_args; struct nfs4_sequence_res; struct nfs_server; struct nfs4_minor_version_ops; -struct server_scope; +struct nfs41_server_scope; struct nfs41_impl_id; /* @@ -80,13 +80,13 @@ struct nfs_client { /* The flags used for obtaining the clientid during EXCHANGE_ID */ u32 cl_exchange_flags; struct nfs4_session *cl_session; /* shared session */ + struct nfs41_server_scope *cl_serverscope; #endif /* CONFIG_NFS_V4 */ #ifdef CONFIG_NFS_FSCACHE struct fscache_cookie *fscache; /* client index cache cookie */ #endif - struct server_scope *server_scope; /* from exchange_id */ struct nfs41_impl_id *impl_id; /* from exchange_id */ struct net *net; }; diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 2e53a3f1d2ff..c420b8d60a55 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1104,7 +1104,7 @@ struct server_owner { char major_id[NFS4_OPAQUE_LIMIT]; }; -struct server_scope { +struct nfs41_server_scope { uint32_t server_scope_sz; char server_scope[NFS4_OPAQUE_LIMIT]; }; @@ -1118,7 +1118,7 @@ struct nfs41_impl_id { struct nfs41_exchange_id_res { struct nfs_client *client; u32 flags; - struct server_scope *server_scope; + struct nfs41_server_scope *server_scope; struct nfs41_impl_id *impl_id; }; -- cgit v1.2.3 From 591555465ec513c42416392d392fd56866cb220c Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:44:41 -0400 Subject: NFS: Use proper naming conventions for nfs_client.impl_id field Clean up: When naming fields and data types, follow established conventions to facilitate accurate grep/cscope searches. Additionally, for consistency, move the impl_id field into the NFSv4- specific part of the nfs_client, and free that memory in the logic that shuts down NFSv4 nfs_clients. Introduced by commit 7d2ed9ac "NFSv4: parse and display server implementation ids," Fri Feb 17, 2012. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 900d733668eb..773e02135903 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -81,13 +81,13 @@ struct nfs_client { u32 cl_exchange_flags; struct nfs4_session *cl_session; /* shared session */ struct nfs41_server_scope *cl_serverscope; + struct nfs41_impl_id *cl_implid; #endif /* CONFIG_NFS_V4 */ #ifdef CONFIG_NFS_FSCACHE struct fscache_cookie *fscache; /* client index cache cookie */ #endif - struct nfs41_impl_id *impl_id; /* from exchange_id */ struct net *net; }; -- cgit v1.2.3 From 73ea666c2bb536f2862cefdb3e014ed62b262ba5 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:44:50 -0400 Subject: NFS: Use proper naming conventions for the nfs_client.net field Clean up: When naming fields and data types, follow established conventions to facilitate accurate grep/cscope searches. Introduced by commit e50a7a1a "NFS: make NFS client allocated per network namespace context," Tue Jan 10, 2012. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 773e02135903..59410b365ba4 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -88,7 +88,7 @@ struct nfs_client { struct fscache_cookie *fscache; /* client index cache cookie */ #endif - struct net *net; + struct net *cl_net; }; /* -- cgit v1.2.3 From f092075dd33ea04000590e8ffea65c2e7d03d764 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:45:41 -0400 Subject: NFS: Always use the same SETCLIENTID boot verifier Currently our NFS client assigns a unique SETCLIENTID boot verifier for each server IP address it knows about. It's set to CURRENT_TIME when the struct nfs_client for that server IP is created. During the SETCLIENTID operation, our client also presents an nfs_client_id4 string to servers, as an identifier on which the server can hang all of this client's NFSv4 state. Our client's nfs_client_id4 string is unique for each server IP address. An NFSv4 server is obligated to wipe all NFSv4 state associated with an nfs_client_id4 string when the client presents the same nfs_client_id4 string along with a changed SETCLIENTID boot verifier. When our client unmounts the last of a server's shares, it destroys that server's struct nfs_client. The next time the client mounts that NFS server, it creates a fresh struct nfs_client with a fresh boot verifier. On seeing the fresh verifer, the server wipes any previous NFSv4 state associated with that nfs_client_id4. However, NFSv4.1 clients are supposed to present the same nfs_client_id4 string to all servers. And, to support Transparent State Migration, the same nfs_client_id4 string should be presented to all NFSv4.0 servers so they recognize that migrated state for this client belongs with state a server may already have for this client. (This is known as the Uniform Client String model). If the nfs_client_id4 string is the same but the boot verifier changes for each server IP address, SETCLIENTID and EXCHANGE_ID operations from such a client could unintentionally result in a server wiping a client's previously obtained lease. Thus, if our NFS client is going to use a fixed nfs_client_id4 string, either for NFSv4.0 or NFSv4.1 mounts, our NFS client should use a boot verifier that does not change depending on server IP address. Replace our current per-nfs_client boot verifier with a per-nfs_net boot verifier. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 3 --- 1 file changed, 3 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 59410b365ba4..fbec57d6dc0a 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -61,9 +61,6 @@ struct nfs_client { struct rpc_wait_queue cl_rpcwaitq; - /* used for the setclientid verifier */ - struct timespec cl_boot_time; - /* idmapper */ struct idmap * cl_idmap; -- cgit v1.2.3 From 8cab4c390b43fe34c07bd33799c1bc24be648122 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:45:59 -0400 Subject: NFS: Refactor nfs_get_client(): initialize nfs_client Clean up: Continue to rationalize the locking in nfs_get_client() by moving the logic that handles the case where a matching server IP address is not found. When we support server trunking detection, client initialization may return a different nfs_client struct than was passed to it. Change the synopsis of the init_client methods to return an nfs_client. The client initialization logic in nfs_get_client() is not much more than a wrapper around ->init_client. It's simpler to keep the little bits of error handling in the version-specific init_client methods. No behavior change is expected. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index c420b8d60a55..0c521cd496a7 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1397,7 +1397,8 @@ struct nfs_rpc_ops { struct nfs_open_context *ctx, int open_flags, struct iattr *iattr); - int (*init_client) (struct nfs_client *, const struct rpc_timeout *, + struct nfs_client * + (*init_client) (struct nfs_client *, const struct rpc_timeout *, const char *, rpc_authflavor_t, int); }; -- cgit v1.2.3 From 4bf590e08f6db3395c181618a4c14f1c39b7c4af Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:46:07 -0400 Subject: NFS: Add nfs_client behavior flags "noresvport" and "discrtry" can be passed to nfs_create_rpc_client() by setting flags in the passed-in nfs_client. This change makes it easy to add new flags. Note that these settings are now "sticky" over the lifetime of a struct nfs_client, and may even be copied when an nfs_client is cloned. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 3 +++ include/linux/nfs_xdr.h | 2 +- 2 files changed, 4 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index fbec57d6dc0a..3a99f5252340 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -35,6 +35,9 @@ struct nfs_client { #define NFS_CS_RENEWD 3 /* - renewd started */ #define NFS_CS_STOP_RENEW 4 /* no more state to renew */ #define NFS_CS_CHECK_LEASE_TIME 5 /* need to check lease time */ + unsigned long cl_flags; /* behavior switches */ +#define NFS_CS_NORESVPORT 0 /* - use ephemeral src port */ +#define NFS_CS_DISCRTRY 1 /* - disconnect on RPC retry */ struct sockaddr_storage cl_addr; /* server identifier */ size_t cl_addrlen; char * cl_hostname; /* hostname of server */ diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 0c521cd496a7..07048c012dec 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1399,7 +1399,7 @@ struct nfs_rpc_ops { struct iattr *iattr); struct nfs_client * (*init_client) (struct nfs_client *, const struct rpc_timeout *, - const char *, rpc_authflavor_t, int); + const char *, rpc_authflavor_t); }; /* -- cgit v1.2.3 From acdeb69d9c5934a678a732b4e24770326bf9471e Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 21 May 2012 22:46:16 -0400 Subject: NFS: EXCHANGE_ID should save the server major and minor ID Save the server major and minor ID results from EXCHANGE_ID, as they are needed for detecting server trunking. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs_sb.h | 1 + include/linux/nfs_xdr.h | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 3a99f5252340..fbb78fb09bd2 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -80,6 +80,7 @@ struct nfs_client { /* The flags used for obtaining the clientid during EXCHANGE_ID */ u32 cl_exchange_flags; struct nfs4_session *cl_session; /* shared session */ + struct nfs41_server_owner *cl_serverowner; struct nfs41_server_scope *cl_serverscope; struct nfs41_impl_id *cl_implid; #endif /* CONFIG_NFS_V4 */ diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 07048c012dec..0872f32c8eef 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1098,7 +1098,7 @@ struct nfs41_exchange_id_args { u32 flags; }; -struct server_owner { +struct nfs41_server_owner { uint64_t minor_id; uint32_t major_id_sz; char major_id[NFS4_OPAQUE_LIMIT]; @@ -1118,6 +1118,7 @@ struct nfs41_impl_id { struct nfs41_exchange_id_res { struct nfs_client *client; u32 flags; + struct nfs41_server_owner *server_owner; struct nfs41_server_scope *server_scope; struct nfs41_impl_id *impl_id; }; -- cgit v1.2.3 From 88034c3d88c2c48b215f2cc5eb22e564aa817f9c Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Wed, 23 May 2012 05:02:34 -0400 Subject: NFSv4.1 mdsthreshold attribute xdr We only support one layout type per file system, so one threshold_item4 per mdsthreshold4. Signed-off-by: Andy Adamson Signed-off-by: Trond Myklebust --- include/linux/nfs4.h | 7 +++++++ include/linux/nfs_xdr.h | 10 ++++++++++ 2 files changed, 17 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index 0987146b0637..72b6bada0d79 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -526,6 +526,13 @@ enum lock_type4 { #define FATTR4_WORD1_MOUNTED_ON_FILEID (1UL << 23) #define FATTR4_WORD1_FS_LAYOUT_TYPES (1UL << 30) #define FATTR4_WORD2_LAYOUT_BLKSIZE (1UL << 1) +#define FATTR4_WORD2_MDSTHRESHOLD (1UL << 4) + +/* MDS threshold bitmap bits */ +#define THRESHOLD_RD (1UL << 0) +#define THRESHOLD_WR (1UL << 1) +#define THRESHOLD_RD_IO (1UL << 2) +#define THRESHOLD_WR_IO (1UL << 3) #define NFSPROC4_NULL 0 #define NFSPROC4_COMPOUND 1 diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 0872f32c8eef..201c312152fb 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -35,6 +35,15 @@ static inline int nfs_fsid_equal(const struct nfs_fsid *a, const struct nfs_fsid return a->major == b->major && a->minor == b->minor; } +struct nfs4_threshold { + __u32 bm; + __u32 l_type; + __u64 rd_sz; + __u64 wr_sz; + __u64 rd_io_sz; + __u64 wr_io_sz; +}; + struct nfs_fattr { unsigned int valid; /* which fields are valid */ umode_t mode; @@ -67,6 +76,7 @@ struct nfs_fattr { unsigned long gencount; struct nfs4_string *owner_name; struct nfs4_string *group_name; + struct nfs4_threshold *mdsthreshold; /* pNFS threshold hints */ }; #define NFS_ATTR_FATTR_TYPE (1U << 0) -- cgit v1.2.3 From 82be417aa37c05116e310b0f2171187ea389f89b Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Wed, 23 May 2012 05:02:35 -0400 Subject: NFSv4.1 cache mdsthreshold values on OPEN Signed-off-by: Andy Adamson Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 80a9385b88ab..ce910cb7d761 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -102,6 +102,7 @@ struct nfs_open_context { int error; struct list_head list; + struct nfs4_threshold *mdsthreshold; }; struct nfs_open_dir_context { -- cgit v1.2.3 From 2701d086dbfca03b2d28b25c6dc11dd78d0e26ad Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 24 May 2012 13:13:24 -0400 Subject: NFSv4.1 add nfs_inode book keeping for mdsthreshold Keep track of the number of bytes read or written via buffered, direct, and mem-mapped i/o for use by mdsthreshold size_io hints. Signed-off-by: Andy Adamson Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index ce910cb7d761..b23cfc120edb 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -202,6 +202,9 @@ struct nfs_inode { /* pNFS layout information */ struct pnfs_layout_hdr *layout; #endif /* CONFIG_NFS_V4*/ + /* how many bytes have been written/read and how many bytes queued up */ + __u64 write_io; + __u64 read_io; #ifdef CONFIG_NFS_FSCACHE struct fscache_cookie *fscache; #endif -- cgit v1.2.3 From 7c44f1ae4a21458a1ea3d6482ffb3136f1df6d2b Mon Sep 17 00:00:00 2001 From: Weston Andros Adamson Date: Thu, 24 May 2012 13:22:50 -0400 Subject: nfs4.1: add BIND_CONN_TO_SESSION operation This patch adds the BIND_CONN_TO_SESSION operation which is needed for upcoming SP4_MACH_CRED work and useful for recovering from broken connections without destroying the session. Signed-off-by: Weston Andros Adamson Signed-off-by: Trond Myklebust --- include/linux/nfs4.h | 5 +++++ include/linux/nfs_xdr.h | 6 ++++++ 2 files changed, 11 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index 72b6bada0d79..a2b71cbfc44a 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -69,6 +69,10 @@ #define NFS4_CDFC4_FORE_OR_BOTH 0x3 #define NFS4_CDFC4_BACK_OR_BOTH 0x7 +#define NFS4_CDFS4_FORE 0x1 +#define NFS4_CDFS4_BACK 0x2 +#define NFS4_CDFS4_BOTH 0x3 + #define NFS4_SET_TO_SERVER_TIME 0 #define NFS4_SET_TO_CLIENT_TIME 1 @@ -589,6 +593,7 @@ enum { NFSPROC4_CLNT_SECINFO, /* nfs41 */ + NFSPROC4_CLNT_BIND_CONN_TO_SESSION, NFSPROC4_CLNT_EXCHANGE_ID, NFSPROC4_CLNT_CREATE_SESSION, NFSPROC4_CLNT_DESTROY_SESSION, diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 201c312152fb..6387fc0097fe 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1125,6 +1125,12 @@ struct nfs41_impl_id { struct nfstime4 date; }; +struct nfs41_bind_conn_to_session_res { + struct nfs4_session *session; + u32 dir; + bool use_conn_in_rdma_mode; +}; + struct nfs41_exchange_id_res { struct nfs_client *client; u32 flags; -- cgit v1.2.3 From ad24ecfbcddfa88541bccc980e753aeda8bf4031 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 25 May 2012 17:11:42 -0400 Subject: NFSv4.1: Move NFSPROC4_CLNT_BIND_CONN_TO_SESSION to the end of the operations For backward compatibility with nfs-utils. Signed-off-by: Trond Myklebust Cc: Weston Andros Adamson --- include/linux/nfs4.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index a2b71cbfc44a..54006a997dd0 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -593,7 +593,6 @@ enum { NFSPROC4_CLNT_SECINFO, /* nfs41 */ - NFSPROC4_CLNT_BIND_CONN_TO_SESSION, NFSPROC4_CLNT_EXCHANGE_ID, NFSPROC4_CLNT_CREATE_SESSION, NFSPROC4_CLNT_DESTROY_SESSION, @@ -608,6 +607,7 @@ enum { NFSPROC4_CLNT_TEST_STATEID, NFSPROC4_CLNT_FREE_STATEID, NFSPROC4_CLNT_GETDEVICELIST, + NFSPROC4_CLNT_BIND_CONN_TO_SESSION, }; /* nfs41 types */ -- cgit v1.2.3 From 662455391040a783b89d0232e743c27c23617dbd Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 25 May 2012 17:18:09 -0400 Subject: NFSv4.1: Add DESTROY_CLIENTID Ensure that we destroy our lease on last unmount Signed-off-by: Trond Myklebust --- include/linux/nfs4.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index 54006a997dd0..af2d2fa30eee 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -608,6 +608,7 @@ enum { NFSPROC4_CLNT_FREE_STATEID, NFSPROC4_CLNT_GETDEVICELIST, NFSPROC4_CLNT_BIND_CONN_TO_SESSION, + NFSPROC4_CLNT_DESTROY_CLIENTID, }; /* nfs41 types */ -- cgit v1.2.3 From 32b0131069c5bebf52368a9fe170f8d58b78fa8d Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 26 May 2012 13:41:04 -0400 Subject: NFSv4.1: Don't clobber the seqid if exchange_id returns a confirmed clientid If the EXCHGID4_FLAG_CONFIRMED_R flag is set, the client is in theory supposed to already know the correct value of the seqid, in which case RFC5661 states that it should ignore the value returned. Also ensure that if the sanity check in nfs4_check_cl_exchange_flags fails, then we must not change the nfs_client fields. Finally, clean up the code: we don't need to retest the value of 'status' unless it can change. Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 6387fc0097fe..d1a7bf51c326 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1132,7 +1132,8 @@ struct nfs41_bind_conn_to_session_res { }; struct nfs41_exchange_id_res { - struct nfs_client *client; + u64 clientid; + u32 seqid; u32 flags; struct nfs41_server_owner *server_owner; struct nfs41_server_scope *server_scope; -- cgit v1.2.3