From b31268ac793fd300da66b9c28bbf0a200339ab96 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 21 Mar 2011 17:02:00 -0400 Subject: FS: Use stable writes when not doing a bulk flush If we're only doing a single write, and there are no other unstable writes being queued up, we might want to just flip to using a stable write RPC call. Reviewed-by: NeilBrown Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 2 ++ include/linux/nfs_page.h | 1 + 2 files changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index f88522b10a38..cb2add401f25 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -33,6 +33,8 @@ #define FLUSH_STABLE 4 /* commit to stable storage */ #define FLUSH_LOWPRI 8 /* low priority background flush */ #define FLUSH_HIGHPRI 16 /* high priority memory reclaim flush */ +#define FLUSH_COND_STABLE 32 /* conditional stable write - only stable + * if everything fits in one RPC */ #ifdef __KERNEL__ diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index 90907ada6d52..92d54c81f51e 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -57,6 +57,7 @@ struct nfs_pageio_descriptor { size_t pg_count; size_t pg_bsize; unsigned int pg_base; + char pg_moreio; struct inode *pg_inode; int (*pg_doio)(struct nfs_pageio_descriptor *); -- cgit v1.2.3 From 480c2006ebb44ae03165695db7b3e38c04e0d102 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Wed, 23 Mar 2011 14:48:29 -0400 Subject: NFS: Create nfs_open_dir_context nfs_opendir() created a context that held much more information than we need for a readdir. This patch introduces a slimmed-down nfs_open_dir_context that contains only the cookie and the cred used for RPC operations. The new context will eventually be used to help detect readdir loops. Signed-off-by: Bryan Schumaker 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 cb2add401f25..4b851a02f7f3 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -95,7 +95,10 @@ struct nfs_open_context { int error; struct list_head list; +}; +struct nfs_open_dir_context { + struct rpc_cred *cred; __u64 dir_cookie; }; -- cgit v1.2.3 From 8ef2ce3e16d9bec6cf015207c1c82a5b864046ac Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Wed, 23 Mar 2011 15:04:31 -0400 Subject: NFS: Detect loops in a readdir due to bad cookies Some filesystems (such as ext4) can return the same cookie value for multiple files. If we try to start a readdir with one of these cookies, the server will return the first file found with a cookie of the same value. This can cause the client to enter an infinite loop. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 4b851a02f7f3..4179c368844b 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -100,6 +100,8 @@ struct nfs_open_context { struct nfs_open_dir_context { struct rpc_cred *cred; __u64 dir_cookie; + __u64 dup_cookie; + int duped; }; /* -- cgit v1.2.3 From a861a1e1c398fe34701569fd8ac9225dfe0a9a7e Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Wed, 23 Mar 2011 13:27:51 +0000 Subject: NFSv4.1: add generic layer hooks for pnfs COMMIT We create three major hooks for the pnfs code. pnfs_mark_request_commit() is called during writeback_done from nfs_mark_request_commit, which gives the driver an opportunity to claim it wants control over commiting a particular req. pnfs_choose_commit_list() is called from nfs_scan_list to choose which list a given req should be added to, based on where we intend to send it for COMMIT. It is up to the driver to have preallocated list headers for each destination it may need. pnfs_commit_list() is how the driver actually takes control, it is used instead of nfs_commit_list(). In order to pass information between the above functions, we create a union in nfs_page to hold a lseg (which is possible because the req is not on any list while in transition), and add some flags to indicate if we need to use the pnfs code. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 1 + include/linux/nfs_page.h | 6 +++++- 2 files changed, 6 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 4179c368844b..eddda6ce7c42 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -226,6 +226,7 @@ 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 */ static inline struct nfs_inode *NFS_I(const struct inode *inode) { diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index 92d54c81f51e..8023e4e25133 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -33,11 +33,15 @@ enum { PG_CLEAN, PG_NEED_COMMIT, PG_NEED_RESCHED, + PG_PNFS_COMMIT, }; struct nfs_inode; struct nfs_page { - struct list_head wb_list; /* Defines state of page: */ + union { + struct list_head wb_list; /* Defines state of page: */ + struct pnfs_layout_segment *wb_commit_lseg; /* Used when PG_PNFS_COMMIT set */ + }; 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 */ -- cgit v1.2.3 From e0c2b3801828aadb65dec9f67f7c6b7a675ad007 Mon Sep 17 00:00:00 2001 From: Fred Isaman Date: Wed, 23 Mar 2011 13:27:53 +0000 Subject: NFSv4.1: filelayout driver specific code for COMMIT Implement all the hooks created in the previous patches. This requires exporting quite a few functions and adding a few structure fields. Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 1 + include/linux/nfs_xdr.h | 1 + 2 files changed, 2 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index eddda6ce7c42..807e07c86b26 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -198,6 +198,7 @@ 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; diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 2c2c67d2eb42..ac0c0e51786e 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1040,6 +1040,7 @@ 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 -- cgit v1.2.3 From 863a3c6c686d5773f7192a4818769e15db12ce08 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Wed, 23 Mar 2011 13:27:54 +0000 Subject: NFSv4.1: layoutcommit The filelayout driver sends LAYOUTCOMMIT only when COMMIT goes to the data server (as opposed to the MDS) and the data server WRITE is not NFS_FILE_SYNC. Only whole file layout support means that there is only one IOMODE_RW layout segment. Signed-off-by: Andy Adamson Signed-off-by: Alexandros Batsakis Signed-off-by: Boaz Harrosh Signed-off-by: Dean Hildebrand Signed-off-by: Fred Isaman Signed-off-by: Mingyang Guo Signed-off-by: Tao Guo Signed-off-by: Zhang Jingwang Tested-by: Boaz Harrosh Signed-off-by: Benny Halevy Signed-off-by: Fred Isaman Signed-off-by: Trond Myklebust --- include/linux/nfs4.h | 1 + include/linux/nfs_fs.h | 1 + include/linux/nfs_xdr.h | 23 +++++++++++++++++++++++ 3 files changed, 25 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index 134716e5e350..bf22cfaaeee5 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -560,6 +560,7 @@ enum { NFSPROC4_CLNT_RECLAIM_COMPLETE, NFSPROC4_CLNT_LAYOUTGET, NFSPROC4_CLNT_GETDEVICEINFO, + NFSPROC4_CLNT_LAYOUTCOMMIT, }; /* nfs41 types */ diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 807e07c86b26..1b93b9c60e55 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -228,6 +228,7 @@ struct nfs_inode { #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 */ static inline struct nfs_inode *NFS_I(const struct inode *inode) { diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index ac0c0e51786e..84f3585c5728 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -236,6 +236,29 @@ struct nfs4_getdeviceinfo_res { struct nfs4_sequence_res seq_res; }; +struct nfs4_layoutcommit_args { + nfs4_stateid stateid; + __u64 lastbytewritten; + struct inode *inode; + const u32 *bitmask; + struct nfs4_sequence_args seq_args; +}; + +struct nfs4_layoutcommit_res { + struct nfs_fattr *fattr; + const struct nfs_server *server; + struct nfs4_sequence_res seq_res; +}; + +struct nfs4_layoutcommit_data { + struct rpc_task task; + struct nfs_fattr fattr; + struct pnfs_layout_segment *lseg; + struct rpc_cred *cred; + struct nfs4_layoutcommit_args args; + struct nfs4_layoutcommit_res res; +}; + /* * Arguments to the open call. */ -- cgit v1.2.3 From 7c5130588d691a3b34d02312f1bd1b6d56fe0100 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Thu, 24 Mar 2011 17:12:24 +0000 Subject: NFS: lookup supports alternate client A later patch will need to perform a lookup using an alternate client with a different security flavor. This patch adds support for doing that on NFS v4. 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 2c2c67d2eb42..71ee6799db9b 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1071,7 +1071,7 @@ struct nfs_rpc_ops { struct nfs_fattr *); int (*setattr) (struct dentry *, struct nfs_fattr *, struct iattr *); - int (*lookup) (struct inode *, struct qstr *, + int (*lookup) (struct rpc_clnt *clnt, 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 5a5ea0d485c9715c86bf858bbdc5f6d373b3db88 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Thu, 24 Mar 2011 17:12:29 +0000 Subject: NFS: Add secinfo procedure This patch adds the nfs4 operation secinfo as a valid nfs rpc operation. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- include/linux/nfs4.h | 1 + include/linux/nfs_xdr.h | 37 +++++++++++++++++++++++++++++++++++++ 2 files changed, 38 insertions(+) (limited to 'include/linux') diff --git a/include/linux/nfs4.h b/include/linux/nfs4.h index 134716e5e350..7e7f6b7e46b1 100644 --- a/include/linux/nfs4.h +++ b/include/linux/nfs4.h @@ -550,6 +550,7 @@ enum { NFSPROC4_CLNT_SETACL, NFSPROC4_CLNT_FS_LOCATIONS, NFSPROC4_CLNT_RELEASE_LOCKOWNER, + NFSPROC4_CLNT_SECINFO, /* nfs41 */ NFSPROC4_CLNT_EXCHANGE_ID, diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 71ee6799db9b..3f32bf175840 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -3,6 +3,7 @@ #include #include +#include /* * To change the maximum rsize and wsize supported by the NFS client, adjust @@ -14,6 +15,9 @@ #define NFS_DEF_FILE_IO_SIZE (4096U) #define NFS_MIN_FILE_IO_SIZE (1024U) +/* Forward declaration for NFS v3 */ +struct nfs4_secinfo_flavors; + struct nfs_fsid { uint64_t major; uint64_t minor; @@ -936,6 +940,38 @@ struct nfs4_fs_locations_res { struct nfs4_sequence_res seq_res; }; +struct nfs4_secinfo_oid { + unsigned int len; + char data[GSS_OID_MAX_LEN]; +}; + +struct nfs4_secinfo_gss { + struct nfs4_secinfo_oid sec_oid4; + unsigned int qop4; + unsigned int service; +}; + +struct nfs4_secinfo_flavor { + unsigned int flavor; + struct nfs4_secinfo_gss gss; +}; + +struct nfs4_secinfo_flavors { + unsigned int num_flavors; + struct nfs4_secinfo_flavor flavors[0]; +}; + +struct nfs4_secinfo_arg { + const struct nfs_fh *dir_fh; + const struct qstr *name; + struct nfs4_sequence_args seq_args; +}; + +struct nfs4_secinfo_res { + struct nfs4_secinfo_flavors *flavors; + struct nfs4_sequence_res seq_res; +}; + #endif /* CONFIG_NFS_V4 */ struct nfstime4 { @@ -1118,6 +1154,7 @@ 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 7ebb931598cd95cccea10d4bc4c0123a464ea565 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Thu, 24 Mar 2011 17:12:30 +0000 Subject: NFS: use secinfo when crossing mountpoints A submount may use different security than the parent mount does. We should figure out what sec flavor the submount uses at mount time. Signed-off-by: Bryan Schumaker 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 3f32bf175840..fa1ba78ca2c8 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -82,6 +82,7 @@ struct nfs_fattr { #define NFS_ATTR_FATTR_CHANGE (1U << 17) #define NFS_ATTR_FATTR_PRECHANGE (1U << 18) #define NFS_ATTR_FATTR_V4_REFERRAL (1U << 19) /* NFSv4 referral */ +#define NFS_ATTR_FATTR_MOUNTPOINT (1U << 20) /* Treat as mountpoint */ #define NFS_ATTR_FATTR (NFS_ATTR_FATTR_TYPE \ | NFS_ATTR_FATTR_MODE \ -- cgit v1.2.3 From 8f70e95f9f4159184f557a1db60c909d7c1bd2e3 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Thu, 24 Mar 2011 17:12:31 +0000 Subject: NFS: Determine initial mount security When sec= is not presented as a mount option, we should attempt to determine what security flavor the server is using. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- include/linux/sunrpc/gss_api.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/sunrpc/gss_api.h b/include/linux/sunrpc/gss_api.h index 5d8048beb051..332da61cf8b7 100644 --- a/include/linux/sunrpc/gss_api.h +++ b/include/linux/sunrpc/gss_api.h @@ -126,6 +126,9 @@ struct gss_api_mech *gss_mech_get_by_name(const char *); /* Similar, but get by pseudoflavor. */ struct gss_api_mech *gss_mech_get_by_pseudoflavor(u32); +/* Fill in an array with a list of supported pseudoflavors */ +int gss_mech_list_pseudoflavors(u32 *); + /* Just increments the mechanism's reference count and returns its input: */ struct gss_api_mech * gss_mech_get(struct gss_api_mech *); -- cgit v1.2.3 From 35124a0994fc02545b14b9fa3aad000b3331f1c0 Mon Sep 17 00:00:00 2001 From: Weston Andros Adamson Date: Thu, 24 Mar 2011 16:48:21 -0400 Subject: Cleanup XDR parsing for LAYOUTGET, GETDEVICEINFO changes LAYOUTGET and GETDEVICEINFO XDR parsing to: - not use vmap, which doesn't work on incoherent archs - use xdr_stream parsing for all xdr Signed-off-by: Weston Andros Adamson Signed-off-by: Trond Myklebust --- include/linux/nfs_xdr.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 84f3585c5728..a6e21b10f43d 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -190,8 +190,9 @@ struct nfs4_get_lease_time_res { #define PNFS_LAYOUT_MAXSIZE 4096 struct nfs4_layoutdriver_data { + struct page **pages; + __u32 pglen; __u32 len; - void *buf; }; struct pnfs_layout_range { @@ -209,6 +210,7 @@ struct nfs4_layoutget_args { struct nfs_open_context *ctx; struct nfs4_sequence_args seq_args; nfs4_stateid stateid; + struct nfs4_layoutdriver_data layout; }; struct nfs4_layoutget_res { @@ -216,8 +218,8 @@ struct nfs4_layoutget_res { struct pnfs_layout_range range; __u32 type; nfs4_stateid stateid; - struct nfs4_layoutdriver_data layout; struct nfs4_sequence_res seq_res; + struct nfs4_layoutdriver_data *layoutp; }; struct nfs4_layoutget { -- cgit v1.2.3