From 88be9f990fe70f0f177ef44a16a477599e91f825 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 5 Jun 2007 10:42:27 -0400 Subject: NFS: Replace vfsmount and dentry in nfs_open_context with struct path Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 0543439a97a..07eea8f64ec 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -31,6 +31,7 @@ #include #include +#include #include #include #include @@ -70,8 +71,7 @@ struct nfs_access_entry { struct nfs4_state; struct nfs_open_context { atomic_t count; - struct vfsmount *vfsmnt; - struct dentry *dentry; + struct path path; struct rpc_cred *cred; struct nfs4_state *state; fl_owner_t lockowner; -- cgit v1.2.3-70-g09d2 From 2aefa104313996d1a9582476cee53d1296c834bf Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 17 Jun 2007 15:40:59 -0400 Subject: NFS: Remove the redundant 'dirty' and 'commit' lists from nfs_inode Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 3 --- fs/nfs/write.c | 3 --- include/linux/nfs_fs.h | 5 +---- include/linux/nfs_page.h | 5 +---- 4 files changed, 2 insertions(+), 14 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 7bcb3dfa617..e7d2bba900b 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1149,14 +1149,11 @@ static void init_once(void * foo, struct kmem_cache * cachep, unsigned long flag inode_init_once(&nfsi->vfs_inode); spin_lock_init(&nfsi->req_lock); - INIT_LIST_HEAD(&nfsi->dirty); - INIT_LIST_HEAD(&nfsi->commit); INIT_LIST_HEAD(&nfsi->open_files); INIT_LIST_HEAD(&nfsi->access_cache_entry_lru); INIT_LIST_HEAD(&nfsi->access_cache_inode_lru); INIT_RADIX_TREE(&nfsi->nfs_page_tree, GFP_ATOMIC); atomic_set(&nfsi->data_updates, 0); - nfsi->ndirty = 0; nfsi->ncommit = 0; nfsi->npages = 0; nfs4_init_once(nfsi); diff --git a/fs/nfs/write.c b/fs/nfs/write.c index 0f779ca12ec..9ef9ec746bf 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -459,7 +459,6 @@ nfs_mark_request_commit(struct nfs_page *req) struct nfs_inode *nfsi = NFS_I(inode); spin_lock(&nfsi->req_lock); - nfs_list_add_request(req, &nfsi->commit); nfsi->ncommit++; set_bit(PG_NEED_COMMIT, &(req)->wb_flags); radix_tree_tag_set(&nfsi->nfs_page_tree, @@ -581,8 +580,6 @@ nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, u res = nfs_scan_list(nfsi, dst, idx_start, npages, NFS_PAGE_TAG_COMMIT); nfsi->ncommit -= res; - if ((nfsi->ncommit == 0) != list_empty(&nfsi->commit)) - printk(KERN_ERR "NFS: desynchronized value of nfs_i.ncommit.\n"); } return res; } diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 07eea8f64ec..a9420547673 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -156,12 +156,9 @@ struct nfs_inode { * This is the list of dirty unwritten pages. */ spinlock_t req_lock; - struct list_head dirty; - struct list_head commit; struct radix_tree_root nfs_page_tree; - unsigned int ndirty, - ncommit, + unsigned int ncommit, npages; /* Open contexts for shared mmap writes */ diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h index 481a42105d6..78e60798d10 100644 --- a/include/linux/nfs_page.h +++ b/include/linux/nfs_page.h @@ -34,8 +34,7 @@ struct nfs_inode; struct nfs_page { - struct list_head wb_list, /* Defines state of page: */ - *wb_list_head; /* read/write/commit */ + struct list_head wb_list; /* Defines state of page: */ struct page *wb_page; /* page to read in/write out */ struct nfs_open_context *wb_context; /* File state context info */ atomic_t wb_complete; /* i/os we're waiting for */ @@ -118,7 +117,6 @@ static inline void nfs_list_add_request(struct nfs_page *req, struct list_head *head) { list_add_tail(&req->wb_list, head); - req->wb_list_head = head; } @@ -132,7 +130,6 @@ nfs_list_remove_request(struct nfs_page *req) if (list_empty(&req->wb_list)) return; list_del_init(&req->wb_list); - req->wb_list_head = NULL; } static inline struct nfs_page * -- cgit v1.2.3-70-g09d2 From dce34ce298d85b81630401f4feb4bd7ac77fe9c7 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 17 Jun 2007 15:47:53 -0400 Subject: NFS: Prevent integer overflow in nfs_scan_list() Also ensure that nfs_inode ncommit and npages are large enough to represent all possible values for the number of pages. Signed-off-by: Trond Myklebust --- fs/nfs/pagelist.c | 2 ++ include/linux/nfs_fs.h | 2 +- 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'include/linux/nfs_fs.h') diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index ad90cbe7670..68f6bf12200 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -425,6 +425,8 @@ int nfs_scan_list(struct nfs_inode *nfsi, req->wb_index, tag); nfs_list_add_request(req, dst); res++; + if (res == INT_MAX) + goto out; } } diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index a9420547673..750708ccd70 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -158,7 +158,7 @@ struct nfs_inode { spinlock_t req_lock; struct radix_tree_root nfs_page_tree; - unsigned int ncommit, + unsigned long ncommit, npages; /* Open contexts for shared mmap writes */ -- cgit v1.2.3-70-g09d2 From 3bec63db55463365110d00721ed60a31e4614cb6 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 17 Jun 2007 16:02:44 -0400 Subject: NFS: Convert struct nfs_open_context to use a kref Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 40 +++++++++++++++++++++++----------------- include/linux/nfs_fs.h | 3 ++- 2 files changed, 25 insertions(+), 18 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index e7d2bba900b..01fc8ab0c56 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -461,7 +461,6 @@ static struct nfs_open_context *alloc_nfs_open_context(struct vfsmount *mnt, str ctx = kmalloc(sizeof(*ctx), GFP_KERNEL); if (ctx != NULL) { - atomic_set(&ctx->count, 1); ctx->path.dentry = dget(dentry); ctx->path.mnt = mntget(mnt); ctx->cred = get_rpccred(cred); @@ -469,6 +468,7 @@ static struct nfs_open_context *alloc_nfs_open_context(struct vfsmount *mnt, str ctx->lockowner = current->files; ctx->error = 0; ctx->dir_cookie = 0; + kref_init(&ctx->kref); } return ctx; } @@ -476,27 +476,33 @@ static struct nfs_open_context *alloc_nfs_open_context(struct vfsmount *mnt, str struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx) { if (ctx != NULL) - atomic_inc(&ctx->count); + kref_get(&ctx->kref); return ctx; } -void put_nfs_open_context(struct nfs_open_context *ctx) +static void nfs_free_open_context(struct kref *kref) { - if (atomic_dec_and_test(&ctx->count)) { - if (!list_empty(&ctx->list)) { - struct inode *inode = ctx->path.dentry->d_inode; - spin_lock(&inode->i_lock); - list_del(&ctx->list); - spin_unlock(&inode->i_lock); - } - if (ctx->state != NULL) - nfs4_close_state(&ctx->path, ctx->state, ctx->mode); - if (ctx->cred != NULL) - put_rpccred(ctx->cred); - dput(ctx->path.dentry); - mntput(ctx->path.mnt); - kfree(ctx); + struct nfs_open_context *ctx = container_of(kref, + struct nfs_open_context, kref); + + if (!list_empty(&ctx->list)) { + struct inode *inode = ctx->path.dentry->d_inode; + spin_lock(&inode->i_lock); + list_del(&ctx->list); + spin_unlock(&inode->i_lock); } + if (ctx->state != NULL) + nfs4_close_state(&ctx->path, ctx->state, ctx->mode); + if (ctx->cred != NULL) + put_rpccred(ctx->cred); + dput(ctx->path.dentry); + mntput(ctx->path.mnt); + kfree(ctx); +} + +void put_nfs_open_context(struct nfs_open_context *ctx) +{ + kref_put(&ctx->kref, nfs_free_open_context); } /* diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 750708ccd70..bf24151d63b 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -30,6 +30,7 @@ #ifdef __KERNEL__ #include +#include #include #include #include @@ -70,7 +71,7 @@ struct nfs_access_entry { struct nfs4_state; struct nfs_open_context { - atomic_t count; + struct kref kref; struct path path; struct rpc_cred *cred; struct nfs4_state *state; -- cgit v1.2.3-70-g09d2 From 587142f85f796cf0b823dd3080e815f02ff6b952 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 2 Jul 2007 09:57:54 -0400 Subject: NFS: Replace NFS_I(inode)->req_lock with inode->i_lock There is no justification for keeping a special spinlock for the exclusive use of the NFS writeback code. Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 1 - fs/nfs/pagelist.c | 11 ++++--- fs/nfs/write.c | 84 ++++++++++++++++++++++++-------------------------- include/linux/nfs_fs.h | 1 - 4 files changed, 46 insertions(+), 51 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 01fc8ab0c56..9d5124166d2 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1154,7 +1154,6 @@ static void init_once(void * foo, struct kmem_cache * cachep, unsigned long flag struct nfs_inode *nfsi = (struct nfs_inode *) foo; inode_init_once(&nfsi->vfs_inode); - spin_lock_init(&nfsi->req_lock); INIT_LIST_HEAD(&nfsi->open_files); INIT_LIST_HEAD(&nfsi->access_cache_entry_lru); INIT_LIST_HEAD(&nfsi->access_cache_inode_lru); diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index 8d2642f24b8..f56dae5216f 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -126,12 +126,13 @@ static int nfs_set_page_tag_locked(struct nfs_page *req) */ void nfs_clear_page_tag_locked(struct nfs_page *req) { - struct nfs_inode *nfsi = NFS_I(req->wb_context->path.dentry->d_inode); + struct inode *inode = req->wb_context->path.dentry->d_inode; + struct nfs_inode *nfsi = NFS_I(inode); if (req->wb_page != NULL) { - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); radix_tree_tag_clear(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_LOCKED); - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); } nfs_unlock_request(req); } @@ -390,7 +391,7 @@ void nfs_pageio_cond_complete(struct nfs_pageio_descriptor *desc, pgoff_t index) * If the number of requests is set to 0, the entire address_space * starting at index idx_start, is scanned. * The requests are *not* checked to ensure that they form a contiguous set. - * You must be holding the inode's req_lock when calling this function + * You must be holding the inode's i_lock when calling this function */ int nfs_scan_list(struct nfs_inode *nfsi, struct list_head *dst, pgoff_t idx_start, @@ -430,7 +431,7 @@ int nfs_scan_list(struct nfs_inode *nfsi, } } /* for latency reduction */ - cond_resched_lock(&nfsi->req_lock); + cond_resched_lock(&nfsi->vfs_inode.i_lock); } out: return res; diff --git a/fs/nfs/write.c b/fs/nfs/write.c index 9ef9ec746bf..73ac992ece8 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -124,12 +124,12 @@ static struct nfs_page *nfs_page_find_request_locked(struct page *page) static struct nfs_page *nfs_page_find_request(struct page *page) { + struct inode *inode = page->mapping->host; struct nfs_page *req = NULL; - spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock; - spin_lock(req_lock); + spin_lock(&inode->i_lock); req = nfs_page_find_request_locked(page); - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); return req; } @@ -251,16 +251,16 @@ static void nfs_end_page_writeback(struct page *page) static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio, struct page *page) { + struct inode *inode = page->mapping->host; + struct nfs_inode *nfsi = NFS_I(inode); struct nfs_page *req; - struct nfs_inode *nfsi = NFS_I(page->mapping->host); - spinlock_t *req_lock = &nfsi->req_lock; int ret; - spin_lock(req_lock); + spin_lock(&inode->i_lock); for(;;) { req = nfs_page_find_request_locked(page); if (req == NULL) { - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); return 1; } if (nfs_lock_request_dontget(req)) @@ -270,28 +270,28 @@ static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio, * succeed provided that someone hasn't already marked the * request as dirty (in which case we don't care). */ - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); ret = nfs_wait_on_request(req); nfs_release_request(req); if (ret != 0) return ret; - spin_lock(req_lock); + spin_lock(&inode->i_lock); } if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) { /* This request is marked for commit */ - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); nfs_unlock_request(req); nfs_pageio_complete(pgio); return 1; } if (nfs_set_page_writeback(page) != 0) { - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); BUG(); } radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_LOCKED); ret = test_bit(PG_NEED_FLUSH, &req->wb_flags); - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); nfs_pageio_add_request(pgio, req); return ret; } @@ -412,7 +412,7 @@ static void nfs_inode_remove_request(struct nfs_page *req) BUG_ON (!NFS_WBACK_BUSY(req)); - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); set_page_private(req->wb_page, 0); ClearPagePrivate(req->wb_page); radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index); @@ -420,11 +420,11 @@ static void nfs_inode_remove_request(struct nfs_page *req) __set_page_dirty_nobuffers(req->wb_page); nfsi->npages--; if (!nfsi->npages) { - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); nfs_end_data_update(inode); iput(inode); } else - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); nfs_clear_request(req); nfs_release_request(req); } @@ -458,13 +458,13 @@ nfs_mark_request_commit(struct nfs_page *req) struct inode *inode = req->wb_context->path.dentry->d_inode; struct nfs_inode *nfsi = NFS_I(inode); - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); nfsi->ncommit++; set_bit(PG_NEED_COMMIT, &(req)->wb_flags); radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_COMMIT); - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS); __mark_inode_dirty(inode, I_DIRTY_DATASYNC); } @@ -534,10 +534,10 @@ static int nfs_wait_on_requests_locked(struct inode *inode, pgoff_t idx_start, u BUG_ON(!NFS_WBACK_BUSY(req)); kref_get(&req->wb_kref); - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); error = nfs_wait_on_request(req); nfs_release_request(req); - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); if (error < 0) return error; res++; @@ -602,7 +602,6 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx, { struct address_space *mapping = page->mapping; struct inode *inode = mapping->host; - struct nfs_inode *nfsi = NFS_I(inode); struct nfs_page *req, *new = NULL; pgoff_t rqend, end; @@ -612,13 +611,13 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx, /* Loop over all inode entries and see if we find * A request for the page we wish to update */ - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); req = nfs_page_find_request_locked(page); if (req) { if (!nfs_lock_request_dontget(req)) { int error; - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); error = nfs_wait_on_request(req); nfs_release_request(req); if (error < 0) { @@ -628,7 +627,7 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx, } continue; } - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); if (new) nfs_release_request(new); break; @@ -639,14 +638,14 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx, nfs_lock_request_dontget(new); error = nfs_inode_add_request(inode, new); if (error) { - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); nfs_unlock_request(new); return ERR_PTR(error); } - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); return new; } - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); new = nfs_create_request(ctx, inode, page, offset, bytes); if (IS_ERR(new)) @@ -974,9 +973,9 @@ static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata) } if (nfs_write_need_commit(data)) { - spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock; + struct inode *inode = page->mapping->host; - spin_lock(req_lock); + spin_lock(&inode->i_lock); if (test_bit(PG_NEED_RESCHED, &req->wb_flags)) { /* Do nothing we need to resend the writes */ } else if (!test_and_set_bit(PG_NEED_COMMIT, &req->wb_flags)) { @@ -987,7 +986,7 @@ static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata) clear_bit(PG_NEED_COMMIT, &req->wb_flags); dprintk(" server reboot detected\n"); } - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); } else dprintk(" OK\n"); @@ -1277,13 +1276,12 @@ static const struct rpc_call_ops nfs_commit_ops = { int nfs_commit_inode(struct inode *inode, int how) { - struct nfs_inode *nfsi = NFS_I(inode); LIST_HEAD(head); int res; - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); res = nfs_scan_commit(inode, &head, 0, 0); - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); if (res) { int error = nfs_commit_list(inode, &head, how); if (error < 0) @@ -1301,7 +1299,6 @@ static inline int nfs_commit_list(struct inode *inode, struct list_head *head, i long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_control *wbc, int how) { struct inode *inode = mapping->host; - struct nfs_inode *nfsi = NFS_I(inode); pgoff_t idx_start, idx_end; unsigned int npages = 0; LIST_HEAD(head); @@ -1323,7 +1320,7 @@ long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_contr } } how &= ~FLUSH_NOCOMMIT; - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); do { ret = nfs_wait_on_requests_locked(inode, idx_start, npages); if (ret != 0) @@ -1334,18 +1331,19 @@ long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_contr if (pages == 0) break; if (how & FLUSH_INVALIDATE) { - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); nfs_cancel_commit_list(&head); ret = pages; - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); continue; } pages += nfs_scan_commit(inode, &head, 0, 0); - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); ret = nfs_commit_list(inode, &head, how); - spin_lock(&nfsi->req_lock); + spin_lock(&inode->i_lock); + } while (ret >= 0); - spin_unlock(&nfsi->req_lock); + spin_unlock(&inode->i_lock); return ret; } @@ -1439,7 +1437,6 @@ int nfs_set_page_dirty(struct page *page) { struct address_space *mapping = page->mapping; struct inode *inode; - spinlock_t *req_lock; struct nfs_page *req; int ret; @@ -1448,18 +1445,17 @@ int nfs_set_page_dirty(struct page *page) inode = mapping->host; if (!inode) goto out_raced; - req_lock = &NFS_I(inode)->req_lock; - spin_lock(req_lock); + spin_lock(&inode->i_lock); req = nfs_page_find_request_locked(page); if (req != NULL) { /* Mark any existing write requests for flushing */ ret = !test_and_set_bit(PG_NEED_FLUSH, &req->wb_flags); - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); nfs_release_request(req); return ret; } ret = __set_page_dirty_nobuffers(page); - spin_unlock(req_lock); + spin_unlock(&inode->i_lock); return ret; out_raced: return !TestSetPageDirty(page); diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index bf24151d63b..cf395351cdd 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -156,7 +156,6 @@ struct nfs_inode { /* * This is the list of dirty unwritten pages. */ - spinlock_t req_lock; struct radix_tree_root nfs_page_tree; unsigned long ncommit, -- cgit v1.2.3-70-g09d2 From 412c77cee6d6e73fbe1dc3d67f52163efed33fc4 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 3 Jul 2007 16:10:55 -0400 Subject: NFSv4: Defer inode revalidation when setting up a delegation Currently we force a synchronous call to __nfs_revalidate_inode() in nfs_inode_set_delegation(). This not only ensures that we cannot call nfs_inode_set_delegation from an asynchronous context, but it also slows down any call to open(). Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 10 ++++++---- fs/nfs/inode.c | 4 +++- include/linux/nfs_fs.h | 1 + 3 files changed, 10 insertions(+), 5 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index 56f4f6a99d4..20ac403469a 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -124,10 +124,6 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct struct nfs_delegation *delegation; int status = 0; - /* Ensure we first revalidate the attributes and page cache! */ - if ((nfsi->cache_validity & (NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_ATTR))) - __nfs_revalidate_inode(NFS_SERVER(inode), inode); - delegation = kmalloc(sizeof(*delegation), GFP_KERNEL); if (delegation == NULL) return -ENOMEM; @@ -154,6 +150,12 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct status = -EIO; } } + + /* Ensure we revalidate the attributes and page cache! */ + spin_lock(&inode->i_lock); + nfsi->cache_validity |= NFS_INO_REVAL_FORCED; + spin_unlock(&inode->i_lock); + spin_unlock(&clp->cl_lock); kfree(delegation); return status; diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 9d5124166d2..3d9fccf4ef9 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1072,8 +1072,10 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr) invalid &= ~NFS_INO_INVALID_DATA; if (data_stable) invalid &= ~(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ATIME|NFS_INO_REVAL_PAGECACHE); - if (!nfs_have_delegation(inode, FMODE_READ)) + if (!nfs_have_delegation(inode, FMODE_READ) || + (nfsi->cache_validity & NFS_INO_REVAL_FORCED)) nfsi->cache_validity |= invalid; + nfsi->cache_validity &= ~NFS_INO_REVAL_FORCED; return 0; out_changed: diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index cf395351cdd..e94971040de 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -184,6 +184,7 @@ struct nfs_inode { #define NFS_INO_INVALID_ACCESS 0x0008 /* cached access cred invalid */ #define NFS_INO_INVALID_ACL 0x0010 /* cached acls are invalid */ #define NFS_INO_REVAL_PAGECACHE 0x0020 /* must revalidate pagecache */ +#define NFS_INO_REVAL_FORCED 0x0040 /* force revalidation ignoring a delegation */ /* * Bit offsets in flags field -- cgit v1.2.3-70-g09d2 From 433c92379d9c2c59c2ebc7628fe4fb02cfc2daf8 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Sun, 1 Jul 2007 12:12:14 -0400 Subject: NFS: Clean up nfs_size_to_loff_t() Use the same file size limit that lockd uses. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- include/linux/nfs_fs.h | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index e94971040de..7deb5b0347f 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -503,12 +503,10 @@ extern int nfsroot_mount(struct sockaddr_in *, char *, struct nfs_fh *, * inline functions */ -static inline loff_t -nfs_size_to_loff_t(__u64 size) +static inline loff_t nfs_size_to_loff_t(__u64 size) { - loff_t maxsz = (((loff_t) ULONG_MAX) << PAGE_CACHE_SHIFT) + PAGE_CACHE_SIZE - 1; - if (size > maxsz) - return maxsz; + if (size > (__u64) OFFSET_MAX - 1) + return OFFSET_MAX - 1; return (loff_t) size; } -- cgit v1.2.3-70-g09d2 From f18289931d705f9c4634b361341a1677bea97aca Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Sun, 1 Jul 2007 12:12:51 -0400 Subject: NFS: Add a new NFS debugging flag just for mount processing Note to self: fix up /usr/sbin/rpcdebug too 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/nfs_fs.h') diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 7deb5b0347f..04f659f1e56 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -553,6 +553,7 @@ extern void * nfs_root_data(void); #define NFSDBG_ROOT 0x0080 #define NFSDBG_CALLBACK 0x0100 #define NFSDBG_CLIENT 0x0200 +#define NFSDBG_MOUNT 0x0400 #define NFSDBG_ALL 0xFFFF #ifdef __KERNEL__ -- cgit v1.2.3-70-g09d2 From 3ea97309e6b18bce200211b3f9188e8023321adc Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Sun, 1 Jul 2007 12:13:27 -0400 Subject: NFS: Remake nfsroot_mount as a permanent part of NFS client In preparation for supporting NFSv2 and NFSv3 mount option handling in the kernel NFS client, convert mount_clnt.c to be a permanent part of the NFS client, instead of built only when CONFIG_ROOT_NFS is enabled. In addition, we also replace the "struct sockaddr_in *" argument with something more generic, to help support IPv6 at some later point. Signed-off-by: Chuck Lever Signed-off-by: Trond Myklebust --- fs/nfs/Makefile | 4 ++-- fs/nfs/mount_clnt.c | 55 +++++++++++++++++++++++++------------------------- fs/nfs/nfsroot.c | 3 ++- include/linux/nfs_fs.h | 5 ++--- 4 files changed, 34 insertions(+), 33 deletions(-) (limited to 'include/linux/nfs_fs.h') diff --git a/fs/nfs/Makefile b/fs/nfs/Makefile index f4580b44eef..b55cb236cf7 100644 --- a/fs/nfs/Makefile +++ b/fs/nfs/Makefile @@ -6,8 +6,8 @@ obj-$(CONFIG_NFS_FS) += nfs.o nfs-y := client.o dir.o file.o getroot.o inode.o super.o nfs2xdr.o \ pagelist.o proc.o read.o symlink.o unlink.o \ - write.o namespace.o -nfs-$(CONFIG_ROOT_NFS) += nfsroot.o mount_clnt.o + write.o namespace.o mount_clnt.o +nfs-$(CONFIG_ROOT_NFS) += nfsroot.o nfs-$(CONFIG_NFS_V3) += nfs3proc.o nfs3xdr.o nfs-$(CONFIG_NFS_V3_ACL) += nfs3acl.o nfs-$(CONFIG_NFS_V4) += nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o \ diff --git a/fs/nfs/mount_clnt.c b/fs/nfs/mount_clnt.c index 2892ec84306..ee4899d9662 100644 --- a/fs/nfs/mount_clnt.c +++ b/fs/nfs/mount_clnt.c @@ -18,7 +18,7 @@ #include #ifdef RPC_DEBUG -# define NFSDBG_FACILITY NFSDBG_ROOT +# define NFSDBG_FACILITY NFSDBG_MOUNT #endif /* @@ -28,7 +28,6 @@ #define MOUNT_UMNT 3 */ -static struct rpc_clnt * mnt_create(struct sockaddr_in *, int, int); static struct rpc_program mnt_program; struct mnt_fhstatus { @@ -36,14 +35,21 @@ struct mnt_fhstatus { struct nfs_fh * fh; }; -/* - * Obtain an NFS file handle for the given host and path +/** + * nfs_mount - Obtain an NFS file handle for the given host and path + * @addr: pointer to server's address + * @len: size of server's address + * @hostname: name of server host, or NULL + * @path: pointer to string containing export path to mount + * @version: mount version to use for this request + * @protocol: transport protocol to use for thie request + * @fh: pointer to location to place returned file handle + * + * Uses default timeout parameters specified by underlying transport. */ -int -nfsroot_mount(struct sockaddr_in *addr, char *path, struct nfs_fh *fh, - int version, int protocol) +int nfs_mount(struct sockaddr *addr, size_t len, char *hostname, char *path, + int version, int protocol, struct nfs_fh *fh) { - struct rpc_clnt *mnt_clnt; struct mnt_fhstatus result = { .fh = fh }; @@ -51,12 +57,23 @@ nfsroot_mount(struct sockaddr_in *addr, char *path, struct nfs_fh *fh, .rpc_argp = path, .rpc_resp = &result, }; + struct rpc_create_args args = { + .protocol = protocol, + .address = addr, + .addrsize = len, + .servername = hostname, + .program = &mnt_program, + .version = version, + .authflavor = RPC_AUTH_UNIX, + .flags = RPC_CLNT_CREATE_INTR, + }; + struct rpc_clnt *mnt_clnt; int status; - dprintk("NFS: nfs_mount(%08x:%s)\n", - (unsigned)ntohl(addr->sin_addr.s_addr), path); + dprintk("NFS: sending MNT request for %s:%s\n", + (hostname ? hostname : "server"), path); - mnt_clnt = mnt_create(addr, version, protocol); + mnt_clnt = rpc_create(&args); if (IS_ERR(mnt_clnt)) return PTR_ERR(mnt_clnt); @@ -70,22 +87,6 @@ nfsroot_mount(struct sockaddr_in *addr, char *path, struct nfs_fh *fh, return status < 0? status : (result.status? -EACCES : 0); } -static struct rpc_clnt *mnt_create(struct sockaddr_in *srvaddr, int version, - int protocol) -{ - struct rpc_create_args args = { - .protocol = protocol, - .address = (struct sockaddr *)srvaddr, - .addrsize = sizeof(*srvaddr), - .program = &mnt_program, - .version = version, - .authflavor = RPC_AUTH_UNIX, - .flags = RPC_CLNT_CREATE_INTR, - }; - - return rpc_create(&args); -} - /* * XDR encode/decode functions for MOUNT */ diff --git a/fs/nfs/nfsroot.c b/fs/nfs/nfsroot.c index f0db4703b1c..3490322d114 100644 --- a/fs/nfs/nfsroot.c +++ b/fs/nfs/nfsroot.c @@ -496,7 +496,8 @@ static int __init root_nfs_get_handle(void) NFS_MNT3_VERSION : NFS_MNT_VERSION; set_sockaddr(&sin, servaddr, htons(mount_port)); - status = nfsroot_mount(&sin, nfs_path, &fh, version, protocol); + status = nfs_mount((struct sockaddr *) &sin, sizeof(sin), NULL, + nfs_path, version, protocol, &fh); if (status < 0) printk(KERN_ERR "Root-NFS: Server returned error %d " "while mounting %s\n", status, nfs_path); diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 04f659f1e56..c098ae194f7 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -494,10 +494,9 @@ static inline void nfs3_forget_cached_acls(struct inode *inode) /* * linux/fs/mount_clnt.c - * (Used only by nfsroot module) */ -extern int nfsroot_mount(struct sockaddr_in *, char *, struct nfs_fh *, - int, int); +extern int nfs_mount(struct sockaddr *, size_t, char *, char *, + int, int, struct nfs_fh *); /* * inline functions -- cgit v1.2.3-70-g09d2