From c4d6d8dbf335c7fa47341654a37c53a512b519bb Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 20 Dec 2012 21:52:32 +0000 Subject: CacheFiles: Fix the marking of cached pages Under some circumstances CacheFiles defers the marking of pages with PG_fscache so that it can take advantage of pagevecs to reduce the number of calls to fscache_mark_pages_cached() and the netfs's hook to keep track of this. There are, however, two problems with this: (1) It can lead to the PG_fscache mark being applied _after_ the page is set PG_uptodate and unlocked (by the call to fscache_end_io()). (2) CacheFiles's ref on the page is dropped immediately following fscache_end_io() - and so may not still be held when the mark is applied. This can lead to the page being passed back to the allocator before the mark is applied. Fix this by, where appropriate, marking the page before calling fscache_end_io() and releasing the page. This means that we can't take advantage of pagevecs and have to make a separate call for each page to the marking routines. The symptoms of this are Bad Page state errors cropping up under memory pressure, for example: BUG: Bad page state in process tar pfn:002da page:ffffea0000009fb0 count:0 mapcount:0 mapping: (null) index:0x1447 page flags: 0x1000(private_2) Pid: 4574, comm: tar Tainted: G W 3.1.0-rc4-fsdevel+ #1064 Call Trace: [] ? dump_page+0xb9/0xbe [] bad_page+0xd5/0xea [] get_page_from_freelist+0x35b/0x46a [] __alloc_pages_nodemask+0x362/0x662 [] __do_page_cache_readahead+0x13a/0x267 [] ? __do_page_cache_readahead+0xa2/0x267 [] ra_submit+0x1c/0x20 [] ondemand_readahead+0x28b/0x29a [] ? ondemand_readahead+0x163/0x29a [] page_cache_sync_readahead+0x38/0x3a [] generic_file_aio_read+0x2ab/0x67e [] nfs_file_read+0xa4/0xc9 [nfs] [] do_sync_read+0xba/0xfa [] ? security_file_permission+0x7b/0x84 [] ? rw_verify_area+0xab/0xc8 [] vfs_read+0xaa/0x13a [] sys_read+0x45/0x6c [] system_call_fastpath+0x16/0x1b As can be seen, PG_private_2 (== PG_fscache) is set in the page flags. Instrumenting fscache_mark_pages_cached() to verify whether page->mapping was set appropriately showed that sometimes it wasn't. This led to the discovery that sometimes the page has apparently been reclaimed by the time the marker got to see it. Reported-by: M. Stevens Signed-off-by: David Howells Reviewed-by: Jeff Layton --- fs/cachefiles/rdwr.c | 34 +++++++++++----------------------- 1 file changed, 11 insertions(+), 23 deletions(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index c994691d9445..3367abdcdac4 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -176,9 +176,8 @@ static void cachefiles_read_copier(struct fscache_operation *_op) recheck: if (PageUptodate(monitor->back_page)) { copy_highpage(monitor->netfs_page, monitor->back_page); - - pagevec_add(&pagevec, monitor->netfs_page); - fscache_mark_pages_cached(monitor->op, &pagevec); + fscache_mark_page_cached(monitor->op, + monitor->netfs_page); error = 0; } else if (!PageError(monitor->back_page)) { /* the page has probably been truncated */ @@ -335,8 +334,7 @@ backing_page_already_present: backing_page_already_uptodate: _debug("- uptodate"); - pagevec_add(pagevec, netpage); - fscache_mark_pages_cached(op, pagevec); + fscache_mark_page_cached(op, netpage); copy_highpage(netpage, backpage); fscache_end_io(op, netpage, 0); @@ -448,8 +446,7 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, &pagevec); } else if (cachefiles_has_space(cache, 0, 1) == 0) { /* there's space in the cache we can use */ - pagevec_add(&pagevec, page); - fscache_mark_pages_cached(op, &pagevec); + fscache_mark_page_cached(op, page); ret = -ENODATA; } else { ret = -ENOBUFS; @@ -465,8 +462,7 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, */ static int cachefiles_read_backing_file(struct cachefiles_object *object, struct fscache_retrieval *op, - struct list_head *list, - struct pagevec *mark_pvec) + struct list_head *list) { struct cachefiles_one_read *monitor = NULL; struct address_space *bmapping = object->backer->d_inode->i_mapping; @@ -626,13 +622,13 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, page_cache_release(backpage); backpage = NULL; - if (!pagevec_add(mark_pvec, netpage)) - fscache_mark_pages_cached(op, mark_pvec); + fscache_mark_page_cached(op, netpage); page_cache_get(netpage); if (!pagevec_add(&lru_pvec, netpage)) __pagevec_lru_add_file(&lru_pvec); + /* the netpage is unlocked and marked up to date here */ fscache_end_io(op, netpage, 0); page_cache_release(netpage); netpage = NULL; @@ -775,15 +771,11 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op, /* submit the apparently valid pages to the backing fs to be read from * disk */ if (nrbackpages > 0) { - ret2 = cachefiles_read_backing_file(object, op, &backpages, - &pagevec); + ret2 = cachefiles_read_backing_file(object, op, &backpages); if (ret2 == -ENOMEM || ret2 == -EINTR) ret = ret2; } - if (pagevec_count(&pagevec) > 0) - fscache_mark_pages_cached(op, &pagevec); - _leave(" = %d [nr=%u%s]", ret, *nr_pages, list_empty(pages) ? " empty" : ""); return ret; @@ -806,7 +798,6 @@ int cachefiles_allocate_page(struct fscache_retrieval *op, { struct cachefiles_object *object; struct cachefiles_cache *cache; - struct pagevec pagevec; int ret; object = container_of(op->op.object, @@ -817,13 +808,10 @@ int cachefiles_allocate_page(struct fscache_retrieval *op, _enter("%p,{%lx},", object, page->index); ret = cachefiles_has_space(cache, 0, 1); - if (ret == 0) { - pagevec_init(&pagevec, 0); - pagevec_add(&pagevec, page); - fscache_mark_pages_cached(op, &pagevec); - } else { + if (ret == 0) + fscache_mark_page_cached(op, page); + else ret = -ENOBUFS; - } _leave(" = %d", ret); return ret; -- cgit v1.2.3 From 5f4f9f4af185d5e76c966d2d3420a61870c856e7 Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 20 Dec 2012 21:52:33 +0000 Subject: CacheFiles: Downgrade the requirements passed to the allocator Downgrade the requirements passed to the allocator in the gfp flags parameter. FS-Cache/CacheFiles can handle OOM conditions simply by aborting the attempt to store an object or a page in the cache. Signed-off-by: David Howells --- fs/cachefiles/interface.c | 8 ++++---- fs/cachefiles/internal.h | 2 ++ fs/cachefiles/key.c | 2 +- fs/cachefiles/rdwr.c | 18 ++++++++++-------- fs/cachefiles/xattr.c | 2 +- fs/fscache/page.c | 2 +- 6 files changed, 19 insertions(+), 15 deletions(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c index 67bef6d01484..9bff0f878cfd 100644 --- a/fs/cachefiles/interface.c +++ b/fs/cachefiles/interface.c @@ -41,12 +41,12 @@ static struct fscache_object *cachefiles_alloc_object( _enter("{%s},%p,", cache->cache.identifier, cookie); - lookup_data = kmalloc(sizeof(*lookup_data), GFP_KERNEL); + lookup_data = kmalloc(sizeof(*lookup_data), cachefiles_gfp); if (!lookup_data) goto nomem_lookup_data; /* create a new object record and a temporary leaf image */ - object = kmem_cache_alloc(cachefiles_object_jar, GFP_KERNEL); + object = kmem_cache_alloc(cachefiles_object_jar, cachefiles_gfp); if (!object) goto nomem_object; @@ -63,7 +63,7 @@ static struct fscache_object *cachefiles_alloc_object( * - stick the length on the front and leave space on the back for the * encoder */ - buffer = kmalloc((2 + 512) + 3, GFP_KERNEL); + buffer = kmalloc((2 + 512) + 3, cachefiles_gfp); if (!buffer) goto nomem_buffer; @@ -219,7 +219,7 @@ static void cachefiles_update_object(struct fscache_object *_object) return; } - auxdata = kmalloc(2 + 512 + 3, GFP_KERNEL); + auxdata = kmalloc(2 + 512 + 3, cachefiles_gfp); if (!auxdata) { _leave(" [nomem]"); return; diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h index bd6bc1bde2d7..49382519907a 100644 --- a/fs/cachefiles/internal.h +++ b/fs/cachefiles/internal.h @@ -23,6 +23,8 @@ extern unsigned cachefiles_debug; #define CACHEFILES_DEBUG_KLEAVE 2 #define CACHEFILES_DEBUG_KDEBUG 4 +#define cachefiles_gfp (__GFP_WAIT | __GFP_NORETRY | __GFP_NOMEMALLOC) + /* * node records */ diff --git a/fs/cachefiles/key.c b/fs/cachefiles/key.c index 81b8b2b3a674..33b58c60f2d1 100644 --- a/fs/cachefiles/key.c +++ b/fs/cachefiles/key.c @@ -78,7 +78,7 @@ char *cachefiles_cook_key(const u8 *raw, int keylen, uint8_t type) _debug("max: %d", max); - key = kmalloc(max, GFP_KERNEL); + key = kmalloc(max, cachefiles_gfp); if (!key) return NULL; diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index 3367abdcdac4..9108b8ea505a 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -238,7 +238,7 @@ static int cachefiles_read_backing_file_one(struct cachefiles_object *object, _debug("read back %p{%lu,%d}", netpage, netpage->index, page_count(netpage)); - monitor = kzalloc(sizeof(*monitor), GFP_KERNEL); + monitor = kzalloc(sizeof(*monitor), cachefiles_gfp); if (!monitor) goto nomem; @@ -257,13 +257,14 @@ static int cachefiles_read_backing_file_one(struct cachefiles_object *object, goto backing_page_already_present; if (!newpage) { - newpage = page_cache_alloc_cold(bmapping); + newpage = __page_cache_alloc(cachefiles_gfp | + __GFP_COLD); if (!newpage) goto nomem_monitor; } ret = add_to_page_cache(newpage, bmapping, - netpage->index, GFP_KERNEL); + netpage->index, cachefiles_gfp); if (ret == 0) goto installed_new_backing_page; if (ret != -EEXIST) @@ -481,7 +482,7 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, netpage, netpage->index, page_count(netpage)); if (!monitor) { - monitor = kzalloc(sizeof(*monitor), GFP_KERNEL); + monitor = kzalloc(sizeof(*monitor), cachefiles_gfp); if (!monitor) goto nomem; @@ -496,13 +497,14 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, goto backing_page_already_present; if (!newpage) { - newpage = page_cache_alloc_cold(bmapping); + newpage = __page_cache_alloc(cachefiles_gfp | + __GFP_COLD); if (!newpage) goto nomem; } ret = add_to_page_cache(newpage, bmapping, - netpage->index, GFP_KERNEL); + netpage->index, cachefiles_gfp); if (ret == 0) goto installed_new_backing_page; if (ret != -EEXIST) @@ -532,7 +534,7 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, _debug("- monitor add"); ret = add_to_page_cache(netpage, op->mapping, netpage->index, - GFP_KERNEL); + cachefiles_gfp); if (ret < 0) { if (ret == -EEXIST) { page_cache_release(netpage); @@ -608,7 +610,7 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, _debug("- uptodate"); ret = add_to_page_cache(netpage, op->mapping, netpage->index, - GFP_KERNEL); + cachefiles_gfp); if (ret < 0) { if (ret == -EEXIST) { page_cache_release(netpage); diff --git a/fs/cachefiles/xattr.c b/fs/cachefiles/xattr.c index e18b183b47e1..73b46288b54b 100644 --- a/fs/cachefiles/xattr.c +++ b/fs/cachefiles/xattr.c @@ -174,7 +174,7 @@ int cachefiles_check_object_xattr(struct cachefiles_object *object, ASSERT(dentry); ASSERT(dentry->d_inode); - auxbuf = kmalloc(sizeof(struct cachefiles_xattr) + 512, GFP_KERNEL); + auxbuf = kmalloc(sizeof(struct cachefiles_xattr) + 512, cachefiles_gfp); if (!auxbuf) { _leave(" = -ENOMEM"); return -ENOMEM; diff --git a/fs/fscache/page.c b/fs/fscache/page.c index d7c663cfc923..248a12e22532 100644 --- a/fs/fscache/page.c +++ b/fs/fscache/page.c @@ -759,7 +759,7 @@ int __fscache_write_page(struct fscache_cookie *cookie, fscache_stat(&fscache_n_stores); - op = kzalloc(sizeof(*op), GFP_NOIO); + op = kzalloc(sizeof(*op), GFP_NOIO | __GFP_NOMEMALLOC | __GFP_NORETRY); if (!op) goto nomem; -- cgit v1.2.3 From 37491a1339df26259b06dfa33f30e574e9e52034 Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 20 Dec 2012 21:52:34 +0000 Subject: CacheFiles: Make some debugging statements conditional Downgrade some debugging statements to not unconditionally print stuff, but rather be conditional on the appropriate module parameter setting. Signed-off-by: David Howells --- fs/cachefiles/rdwr.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index 9108b8ea505a..bf123d9c3206 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -77,25 +77,25 @@ static int cachefiles_read_reissue(struct cachefiles_object *object, struct page *backpage = monitor->back_page, *backpage2; int ret; - kenter("{ino=%lx},{%lx,%lx}", + _enter("{ino=%lx},{%lx,%lx}", object->backer->d_inode->i_ino, backpage->index, backpage->flags); /* skip if the page was truncated away completely */ if (backpage->mapping != bmapping) { - kleave(" = -ENODATA [mapping]"); + _leave(" = -ENODATA [mapping]"); return -ENODATA; } backpage2 = find_get_page(bmapping, backpage->index); if (!backpage2) { - kleave(" = -ENODATA [gone]"); + _leave(" = -ENODATA [gone]"); return -ENODATA; } if (backpage != backpage2) { put_page(backpage2); - kleave(" = -ENODATA [different]"); + _leave(" = -ENODATA [different]"); return -ENODATA; } @@ -114,7 +114,7 @@ static int cachefiles_read_reissue(struct cachefiles_object *object, if (PageUptodate(backpage)) goto unlock_discard; - kdebug("reissue read"); + _debug("reissue read"); ret = bmapping->a_ops->readpage(NULL, backpage); if (ret < 0) goto unlock_discard; @@ -129,7 +129,7 @@ static int cachefiles_read_reissue(struct cachefiles_object *object, } /* it'll reappear on the todo list */ - kleave(" = -EINPROGRESS"); + _leave(" = -EINPROGRESS"); return -EINPROGRESS; unlock_discard: @@ -137,7 +137,7 @@ unlock_discard: spin_lock_irq(&object->work_lock); list_del(&monitor->op_link); spin_unlock_irq(&object->work_lock); - kleave(" = %d", ret); + _leave(" = %d", ret); return ret; } -- cgit v1.2.3 From 9f10523f891928330b7529da54c1a3cc65180b1a Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 20 Dec 2012 21:52:35 +0000 Subject: FS-Cache: Fix operation state management and accounting Fix the state management of internal fscache operations and the accounting of what operations are in what states. This is done by: (1) Give struct fscache_operation a enum variable that directly represents the state it's currently in, rather than spreading this knowledge over a bunch of flags, who's processing the operation at the moment and whether it is queued or not. This makes it easier to write assertions to check the state at various points and to prevent invalid state transitions. (2) Add an 'operation complete' state and supply a function to indicate the completion of an operation (fscache_op_complete()) and make things call it. The final call to fscache_put_operation() can then check that an op in the appropriate state (complete or cancelled). (3) Adjust the use of object->n_ops, ->n_in_progress, ->n_exclusive to better govern the state of an object: (a) The ->n_ops is now the number of extant operations on the object and is now decremented by fscache_put_operation() only. (b) The ->n_in_progress is simply the number of objects that have been taken off of the object's pending queue for the purposes of being run. This is decremented by fscache_op_complete() only. (c) The ->n_exclusive is the number of exclusive ops that have been submitted and queued or are in progress. It is decremented by fscache_op_complete() and by fscache_cancel_op(). fscache_put_operation() and fscache_operation_gc() now no longer try to clean up ->n_exclusive and ->n_in_progress. That was leading to double decrements against fscache_cancel_op(). fscache_cancel_op() now no longer decrements ->n_ops. That was leading to double decrements against fscache_put_operation(). fscache_submit_exclusive_op() now decides whether it has to queue an op based on ->n_in_progress being > 0 rather than ->n_ops > 0 as the latter will persist in being true even after all preceding operations have been cancelled or completed. Furthermore, if an object is active and there are runnable ops against it, there must be at least one op running. (4) Add a remaining-pages counter (n_pages) to struct fscache_retrieval and provide a function to record completion of the pages as they complete. When n_pages reaches 0, the operation is deemed to be complete and fscache_op_complete() is called. Add calls to fscache_retrieval_complete() anywhere we've finished with a page we've been given to read or allocate for. This includes places where we just return pages to the netfs for reading from the server and where accessing the cache fails and we discard the proposed netfs page. The bugs in the unfixed state management manifest themselves as oopses like the following where the operation completion gets out of sync with return of the cookie by the netfs. This is possible because the cache unlocks and returns all the netfs pages before recording its completion - which means that there's nothing to stop the netfs discarding them and returning the cookie. FS-Cache: Cookie 'NFS.fh' still has outstanding reads ------------[ cut here ]------------ kernel BUG at fs/fscache/cookie.c:519! invalid opcode: 0000 [#1] SMP CPU 1 Modules linked in: cachefiles nfs fscache auth_rpcgss nfs_acl lockd sunrpc Pid: 400, comm: kswapd0 Not tainted 3.1.0-rc7-fsdevel+ #1090 /DG965RY RIP: 0010:[] [] __fscache_relinquish_cookie+0x170/0x343 [fscache] RSP: 0018:ffff8800368cfb00 EFLAGS: 00010282 RAX: 000000000000003c RBX: ffff880023cc8790 RCX: 0000000000000000 RDX: 0000000000002f2e RSI: 0000000000000001 RDI: ffffffff813ab86c RBP: ffff8800368cfb50 R08: 0000000000000002 R09: 0000000000000000 R10: ffff88003a1b7890 R11: ffff88001df6e488 R12: ffff880023d8ed98 R13: ffff880023cc8798 R14: 0000000000000004 R15: ffff88003b8bf370 FS: 0000000000000000(0000) GS:ffff88003bd00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 00000000008ba008 CR3: 0000000023d93000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process kswapd0 (pid: 400, threadinfo ffff8800368ce000, task ffff88003b8bf040) Stack: ffff88003b8bf040 ffff88001df6e528 ffff88001df6e528 ffffffffa00b46b0 ffff88003b8bf040 ffff88001df6e488 ffff88001df6e620 ffffffffa00b46b0 ffff88001ebd04c8 0000000000000004 ffff8800368cfb70 ffffffffa00b2c91 Call Trace: [] nfs_fscache_release_inode_cookie+0x3b/0x47 [nfs] [] nfs_clear_inode+0x3c/0x41 [nfs] [] nfs4_evict_inode+0x2f/0x33 [nfs] [] evict+0xa1/0x15c [] dispose_list+0x2c/0x38 [] prune_icache_sb+0x28c/0x29b [] prune_super+0xd5/0x140 [] shrink_slab+0x102/0x1ab [] balance_pgdat+0x2f2/0x595 [] ? process_timeout+0xb/0xb [] kswapd+0x270/0x289 [] ? __init_waitqueue_head+0x46/0x46 [] ? balance_pgdat+0x595/0x595 [] kthread+0x7f/0x87 [] kernel_thread_helper+0x4/0x10 [] ? finish_task_switch+0x45/0xc0 [] ? retint_restore_args+0xe/0xe [] ? __init_kthread_worker+0x53/0x53 [] ? gs_change+0xb/0xb Signed-off-by: David Howells --- Documentation/filesystems/caching/backend-api.txt | 26 ++++++- Documentation/filesystems/caching/operations.txt | 2 +- fs/cachefiles/rdwr.c | 31 ++++++-- fs/fscache/object.c | 2 - fs/fscache/operation.c | 91 +++++++++++++++-------- fs/fscache/page.c | 25 ++++++- include/linux/fscache-cache.h | 37 +++++++-- 7 files changed, 164 insertions(+), 50 deletions(-) (limited to 'fs/cachefiles') diff --git a/Documentation/filesystems/caching/backend-api.txt b/Documentation/filesystems/caching/backend-api.txt index 382d52cdaf2d..f4769b9399df 100644 --- a/Documentation/filesystems/caching/backend-api.txt +++ b/Documentation/filesystems/caching/backend-api.txt @@ -419,7 +419,10 @@ performed on the denizens of the cache. These are held in a structure of type: If an I/O error occurs, fscache_io_error() should be called and -ENOBUFS returned if possible or fscache_end_io() called with a suitable error - code.. + code. + + fscache_put_retrieval() should be called after a page or pages are dealt + with. This will complete the operation when all pages are dealt with. (*) Request pages be read from cache [mandatory]: @@ -526,6 +529,27 @@ FS-Cache provides some utilities that a cache backend may make use of: error value should be 0 if successful and an error otherwise. + (*) Record that one or more pages being retrieved or allocated have been dealt + with: + + void fscache_retrieval_complete(struct fscache_retrieval *op, + int n_pages); + + This is called to record the fact that one or more pages have been dealt + with and are no longer the concern of this operation. When the number of + pages remaining in the operation reaches 0, the operation will be + completed. + + + (*) Record operation completion: + + void fscache_op_complete(struct fscache_operation *op); + + This is called to record the completion of an operation. This deducts + this operation from the parent object's run state, potentially permitting + one or more pending operations to start running. + + (*) Set highest store limit: void fscache_set_store_limit(struct fscache_object *object, diff --git a/Documentation/filesystems/caching/operations.txt b/Documentation/filesystems/caching/operations.txt index b6b070c57cbf..bee2a5f93d60 100644 --- a/Documentation/filesystems/caching/operations.txt +++ b/Documentation/filesystems/caching/operations.txt @@ -174,7 +174,7 @@ Operations are used through the following procedure: necessary (the object might have died whilst the thread was waiting). When it has finished doing its processing, it should call - fscache_put_operation() on it. + fscache_op_complete() and fscache_put_operation() on it. (4) The operation holds an effective lock upon the object, preventing other exclusive ops conflicting until it is released. The operation can be diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index bf123d9c3206..93a0815e0498 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -197,6 +197,7 @@ static void cachefiles_read_copier(struct fscache_operation *_op) fscache_end_io(op, monitor->netfs_page, error); page_cache_release(monitor->netfs_page); + fscache_retrieval_complete(op, 1); fscache_put_retrieval(op); kfree(monitor); @@ -339,6 +340,7 @@ backing_page_already_uptodate: copy_highpage(netpage, backpage); fscache_end_io(op, netpage, 0); + fscache_retrieval_complete(op, 1); success: _debug("success"); @@ -360,6 +362,7 @@ read_error: goto out; io_error: cachefiles_io_error_obj(object, "Page read error on backing file"); + fscache_retrieval_complete(op, 1); ret = -ENOBUFS; goto out; @@ -369,6 +372,7 @@ nomem_monitor: fscache_put_retrieval(monitor->op); kfree(monitor); nomem: + fscache_retrieval_complete(op, 1); _leave(" = -ENOMEM"); return -ENOMEM; } @@ -407,7 +411,7 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, _enter("{%p},{%lx},,,", object, page->index); if (!object->backer) - return -ENOBUFS; + goto enobufs; inode = object->backer->d_inode; ASSERT(S_ISREG(inode->i_mode)); @@ -416,7 +420,7 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, /* calculate the shift required to use bmap */ if (inode->i_sb->s_blocksize > PAGE_SIZE) - return -ENOBUFS; + goto enobufs; shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits; @@ -448,13 +452,19 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, } else if (cachefiles_has_space(cache, 0, 1) == 0) { /* there's space in the cache we can use */ fscache_mark_page_cached(op, page); + fscache_retrieval_complete(op, 1); ret = -ENODATA; } else { - ret = -ENOBUFS; + goto enobufs; } _leave(" = %d", ret); return ret; + +enobufs: + fscache_retrieval_complete(op, 1); + _leave(" = -ENOBUFS"); + return -ENOBUFS; } /* @@ -632,6 +642,7 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, /* the netpage is unlocked and marked up to date here */ fscache_end_io(op, netpage, 0); + fscache_retrieval_complete(op, 1); page_cache_release(netpage); netpage = NULL; continue; @@ -659,6 +670,7 @@ out: list_for_each_entry_safe(netpage, _n, list, lru) { list_del(&netpage->lru); page_cache_release(netpage); + fscache_retrieval_complete(op, 1); } _leave(" = %d", ret); @@ -707,7 +719,7 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op, *nr_pages); if (!object->backer) - return -ENOBUFS; + goto all_enobufs; space = 1; if (cachefiles_has_space(cache, 0, *nr_pages) < 0) @@ -720,7 +732,7 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op, /* calculate the shift required to use bmap */ if (inode->i_sb->s_blocksize > PAGE_SIZE) - return -ENOBUFS; + goto all_enobufs; shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits; @@ -760,7 +772,10 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op, nrbackpages++; } else if (space && pagevec_add(&pagevec, page) == 0) { fscache_mark_pages_cached(op, &pagevec); + fscache_retrieval_complete(op, 1); ret = -ENODATA; + } else { + fscache_retrieval_complete(op, 1); } } @@ -781,6 +796,10 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op, _leave(" = %d [nr=%u%s]", ret, *nr_pages, list_empty(pages) ? " empty" : ""); return ret; + +all_enobufs: + fscache_retrieval_complete(op, *nr_pages); + return -ENOBUFS; } /* @@ -815,6 +834,7 @@ int cachefiles_allocate_page(struct fscache_retrieval *op, else ret = -ENOBUFS; + fscache_retrieval_complete(op, 1); _leave(" = %d", ret); return ret; } @@ -864,6 +884,7 @@ int cachefiles_allocate_pages(struct fscache_retrieval *op, ret = -ENOBUFS; } + fscache_retrieval_complete(op, *nr_pages); _leave(" = %d", ret); return ret; } diff --git a/fs/fscache/object.c b/fs/fscache/object.c index b6b897c550ac..773bc798a416 100644 --- a/fs/fscache/object.c +++ b/fs/fscache/object.c @@ -587,8 +587,6 @@ static void fscache_object_available(struct fscache_object *object) if (object->n_in_progress == 0) { if (object->n_ops > 0) { ASSERTCMP(object->n_ops, >=, object->n_obj_ops); - ASSERTIF(object->n_ops > object->n_obj_ops, - !list_empty(&object->pending_ops)); fscache_start_operations(object); } else { ASSERT(list_empty(&object->pending_ops)); diff --git a/fs/fscache/operation.c b/fs/fscache/operation.c index c857ab824d6e..748f9553c2cb 100644 --- a/fs/fscache/operation.c +++ b/fs/fscache/operation.c @@ -37,6 +37,7 @@ void fscache_enqueue_operation(struct fscache_operation *op) ASSERT(op->processor != NULL); ASSERTCMP(op->object->state, >=, FSCACHE_OBJECT_AVAILABLE); ASSERTCMP(atomic_read(&op->usage), >, 0); + ASSERTCMP(op->state, ==, FSCACHE_OP_ST_IN_PROGRESS); fscache_stat(&fscache_n_op_enqueue); switch (op->flags & FSCACHE_OP_TYPE) { @@ -64,6 +65,9 @@ EXPORT_SYMBOL(fscache_enqueue_operation); static void fscache_run_op(struct fscache_object *object, struct fscache_operation *op) { + ASSERTCMP(op->state, ==, FSCACHE_OP_ST_PENDING); + + op->state = FSCACHE_OP_ST_IN_PROGRESS; object->n_in_progress++; if (test_and_clear_bit(FSCACHE_OP_WAITING, &op->flags)) wake_up_bit(&op->flags, FSCACHE_OP_WAITING); @@ -80,22 +84,23 @@ static void fscache_run_op(struct fscache_object *object, int fscache_submit_exclusive_op(struct fscache_object *object, struct fscache_operation *op) { - int ret; - _enter("{OBJ%x OP%x},", object->debug_id, op->debug_id); + ASSERTCMP(op->state, ==, FSCACHE_OP_ST_INITIALISED); + ASSERTCMP(atomic_read(&op->usage), >, 0); + spin_lock(&object->lock); ASSERTCMP(object->n_ops, >=, object->n_in_progress); ASSERTCMP(object->n_ops, >=, object->n_exclusive); ASSERT(list_empty(&op->pend_link)); - ret = -ENOBUFS; + op->state = FSCACHE_OP_ST_PENDING; if (fscache_object_is_active(object)) { op->object = object; object->n_ops++; object->n_exclusive++; /* reads and writes must wait */ - if (object->n_ops > 1) { + if (object->n_in_progress > 0) { atomic_inc(&op->usage); list_add_tail(&op->pend_link, &object->pending_ops); fscache_stat(&fscache_n_op_pend); @@ -111,7 +116,6 @@ int fscache_submit_exclusive_op(struct fscache_object *object, /* need to issue a new write op after this */ clear_bit(FSCACHE_OBJECT_PENDING_WRITE, &object->flags); - ret = 0; } else if (object->state == FSCACHE_OBJECT_CREATING) { op->object = object; object->n_ops++; @@ -119,14 +123,13 @@ int fscache_submit_exclusive_op(struct fscache_object *object, atomic_inc(&op->usage); list_add_tail(&op->pend_link, &object->pending_ops); fscache_stat(&fscache_n_op_pend); - ret = 0; } else { /* not allowed to submit ops in any other state */ BUG(); } spin_unlock(&object->lock); - return ret; + return 0; } /* @@ -186,6 +189,7 @@ int fscache_submit_op(struct fscache_object *object, _enter("{OBJ%x OP%x},{%u}", object->debug_id, op->debug_id, atomic_read(&op->usage)); + ASSERTCMP(op->state, ==, FSCACHE_OP_ST_INITIALISED); ASSERTCMP(atomic_read(&op->usage), >, 0); spin_lock(&object->lock); @@ -196,6 +200,7 @@ int fscache_submit_op(struct fscache_object *object, ostate = object->state; smp_rmb(); + op->state = FSCACHE_OP_ST_PENDING; if (fscache_object_is_active(object)) { op->object = object; object->n_ops++; @@ -225,12 +230,15 @@ int fscache_submit_op(struct fscache_object *object, object->state == FSCACHE_OBJECT_LC_DYING || object->state == FSCACHE_OBJECT_WITHDRAWING) { fscache_stat(&fscache_n_op_rejected); + op->state = FSCACHE_OP_ST_CANCELLED; ret = -ENOBUFS; } else if (!test_bit(FSCACHE_IOERROR, &object->cache->flags)) { fscache_report_unexpected_submission(object, op, ostate); ASSERT(!fscache_object_is_active(object)); + op->state = FSCACHE_OP_ST_CANCELLED; ret = -ENOBUFS; } else { + op->state = FSCACHE_OP_ST_CANCELLED; ret = -ENOBUFS; } @@ -290,13 +298,18 @@ int fscache_cancel_op(struct fscache_operation *op) _enter("OBJ%x OP%x}", op->object->debug_id, op->debug_id); + ASSERTCMP(op->state, >=, FSCACHE_OP_ST_PENDING); + ASSERTCMP(op->state, !=, FSCACHE_OP_ST_CANCELLED); + ASSERTCMP(atomic_read(&op->usage), >, 0); + spin_lock(&object->lock); ret = -EBUSY; - if (!list_empty(&op->pend_link)) { + if (op->state == FSCACHE_OP_ST_PENDING) { + ASSERT(!list_empty(&op->pend_link)); fscache_stat(&fscache_n_op_cancelled); list_del_init(&op->pend_link); - object->n_ops--; + op->state = FSCACHE_OP_ST_CANCELLED; if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags)) object->n_exclusive--; if (test_and_clear_bit(FSCACHE_OP_WAITING, &op->flags)) @@ -310,6 +323,37 @@ int fscache_cancel_op(struct fscache_operation *op) return ret; } +/* + * Record the completion of an in-progress operation. + */ +void fscache_op_complete(struct fscache_operation *op) +{ + struct fscache_object *object = op->object; + + _enter("OBJ%x", object->debug_id); + + ASSERTCMP(op->state, ==, FSCACHE_OP_ST_IN_PROGRESS); + ASSERTCMP(object->n_in_progress, >, 0); + ASSERTIFCMP(test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags), + object->n_exclusive, >, 0); + ASSERTIFCMP(test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags), + object->n_in_progress, ==, 1); + + spin_lock(&object->lock); + + op->state = FSCACHE_OP_ST_COMPLETE; + + if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags)) + object->n_exclusive--; + object->n_in_progress--; + if (object->n_in_progress == 0) + fscache_start_operations(object); + + spin_unlock(&object->lock); + _leave(""); +} +EXPORT_SYMBOL(fscache_op_complete); + /* * release an operation * - queues pending ops if this is the last in-progress op @@ -328,8 +372,9 @@ void fscache_put_operation(struct fscache_operation *op) return; _debug("PUT OP"); - if (test_and_set_bit(FSCACHE_OP_DEAD, &op->flags)) - BUG(); + ASSERTIFCMP(op->state != FSCACHE_OP_ST_COMPLETE, + op->state, ==, FSCACHE_OP_ST_CANCELLED); + op->state = FSCACHE_OP_ST_DEAD; fscache_stat(&fscache_n_op_release); @@ -365,16 +410,6 @@ void fscache_put_operation(struct fscache_operation *op) return; } - if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags)) { - ASSERTCMP(object->n_exclusive, >, 0); - object->n_exclusive--; - } - - ASSERTCMP(object->n_in_progress, >, 0); - object->n_in_progress--; - if (object->n_in_progress == 0) - fscache_start_operations(object); - ASSERTCMP(object->n_ops, >, 0); object->n_ops--; if (object->n_ops == 0) @@ -413,23 +448,14 @@ void fscache_operation_gc(struct work_struct *work) spin_unlock(&cache->op_gc_list_lock); object = op->object; + spin_lock(&object->lock); _debug("GC DEFERRED REL OBJ%x OP%x", object->debug_id, op->debug_id); fscache_stat(&fscache_n_op_gc); ASSERTCMP(atomic_read(&op->usage), ==, 0); - - spin_lock(&object->lock); - if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags)) { - ASSERTCMP(object->n_exclusive, >, 0); - object->n_exclusive--; - } - - ASSERTCMP(object->n_in_progress, >, 0); - object->n_in_progress--; - if (object->n_in_progress == 0) - fscache_start_operations(object); + ASSERTCMP(op->state, ==, FSCACHE_OP_ST_DEAD); ASSERTCMP(object->n_ops, >, 0); object->n_ops--; @@ -437,6 +463,7 @@ void fscache_operation_gc(struct work_struct *work) fscache_raise_event(object, FSCACHE_OBJECT_EV_CLEARED); spin_unlock(&object->lock); + kfree(op); } while (count++ < 20); diff --git a/fs/fscache/page.c b/fs/fscache/page.c index 248a12e22532..b38b13d2a555 100644 --- a/fs/fscache/page.c +++ b/fs/fscache/page.c @@ -162,6 +162,7 @@ static void fscache_attr_changed_op(struct fscache_operation *op) fscache_abort_object(object); } + fscache_op_complete(op); _leave(""); } @@ -223,6 +224,8 @@ static void fscache_release_retrieval_op(struct fscache_operation *_op) _enter("{OP%x}", op->op.debug_id); + ASSERTCMP(op->n_pages, ==, 0); + fscache_hist(fscache_retrieval_histogram, op->start_time); if (op->context) fscache_put_context(op->op.object->cookie, op->context); @@ -320,6 +323,11 @@ static int fscache_wait_for_retrieval_activation(struct fscache_object *object, _debug("<<< GO"); check_if_dead: + if (op->op.state == FSCACHE_OP_ST_CANCELLED) { + fscache_stat(stat_object_dead); + _leave(" = -ENOBUFS [cancelled]"); + return -ENOBUFS; + } if (unlikely(fscache_object_is_dead(object))) { fscache_stat(stat_object_dead); return -ENOBUFS; @@ -364,6 +372,7 @@ int __fscache_read_or_alloc_page(struct fscache_cookie *cookie, _leave(" = -ENOMEM"); return -ENOMEM; } + op->n_pages = 1; spin_lock(&cookie->lock); @@ -375,10 +384,10 @@ int __fscache_read_or_alloc_page(struct fscache_cookie *cookie, ASSERTCMP(object->state, >, FSCACHE_OBJECT_LOOKING_UP); atomic_inc(&object->n_reads); - set_bit(FSCACHE_OP_DEC_READ_CNT, &op->op.flags); + __set_bit(FSCACHE_OP_DEC_READ_CNT, &op->op.flags); if (fscache_submit_op(object, &op->op) < 0) - goto nobufs_unlock; + goto nobufs_unlock_dec; spin_unlock(&cookie->lock); fscache_stat(&fscache_n_retrieval_ops); @@ -425,6 +434,8 @@ error: _leave(" = %d", ret); return ret; +nobufs_unlock_dec: + atomic_dec(&object->n_reads); nobufs_unlock: spin_unlock(&cookie->lock); kfree(op); @@ -482,6 +493,7 @@ int __fscache_read_or_alloc_pages(struct fscache_cookie *cookie, op = fscache_alloc_retrieval(mapping, end_io_func, context); if (!op) return -ENOMEM; + op->n_pages = *nr_pages; spin_lock(&cookie->lock); @@ -491,10 +503,10 @@ int __fscache_read_or_alloc_pages(struct fscache_cookie *cookie, struct fscache_object, cookie_link); atomic_inc(&object->n_reads); - set_bit(FSCACHE_OP_DEC_READ_CNT, &op->op.flags); + __set_bit(FSCACHE_OP_DEC_READ_CNT, &op->op.flags); if (fscache_submit_op(object, &op->op) < 0) - goto nobufs_unlock; + goto nobufs_unlock_dec; spin_unlock(&cookie->lock); fscache_stat(&fscache_n_retrieval_ops); @@ -541,6 +553,8 @@ error: _leave(" = %d", ret); return ret; +nobufs_unlock_dec: + atomic_dec(&object->n_reads); nobufs_unlock: spin_unlock(&cookie->lock); kfree(op); @@ -583,6 +597,7 @@ int __fscache_alloc_page(struct fscache_cookie *cookie, op = fscache_alloc_retrieval(page->mapping, NULL, NULL); if (!op) return -ENOMEM; + op->n_pages = 1; spin_lock(&cookie->lock); @@ -696,6 +711,7 @@ static void fscache_write_op(struct fscache_operation *_op) fscache_end_page_write(object, page); if (ret < 0) { fscache_abort_object(object); + fscache_op_complete(&op->op); } else { fscache_enqueue_operation(&op->op); } @@ -710,6 +726,7 @@ superseded: spin_unlock(&cookie->stores_lock); clear_bit(FSCACHE_OBJECT_PENDING_WRITE, &object->flags); spin_unlock(&object->lock); + fscache_op_complete(&op->op); _leave(""); } diff --git a/include/linux/fscache-cache.h b/include/linux/fscache-cache.h index e3d6d939d959..f5facd1d333f 100644 --- a/include/linux/fscache-cache.h +++ b/include/linux/fscache-cache.h @@ -75,6 +75,16 @@ extern wait_queue_head_t fscache_cache_cleared_wq; typedef void (*fscache_operation_release_t)(struct fscache_operation *op); typedef void (*fscache_operation_processor_t)(struct fscache_operation *op); +enum fscache_operation_state { + FSCACHE_OP_ST_BLANK, /* Op is not yet submitted */ + FSCACHE_OP_ST_INITIALISED, /* Op is initialised */ + FSCACHE_OP_ST_PENDING, /* Op is blocked from running */ + FSCACHE_OP_ST_IN_PROGRESS, /* Op is in progress */ + FSCACHE_OP_ST_COMPLETE, /* Op is complete */ + FSCACHE_OP_ST_CANCELLED, /* Op has been cancelled */ + FSCACHE_OP_ST_DEAD /* Op is now dead */ +}; + struct fscache_operation { struct work_struct work; /* record for async ops */ struct list_head pend_link; /* link in object->pending_ops */ @@ -86,10 +96,10 @@ struct fscache_operation { #define FSCACHE_OP_MYTHREAD 0x0002 /* - processing is done be issuing thread, not pool */ #define FSCACHE_OP_WAITING 4 /* cleared when op is woken */ #define FSCACHE_OP_EXCLUSIVE 5 /* exclusive op, other ops must wait */ -#define FSCACHE_OP_DEAD 6 /* op is now dead */ -#define FSCACHE_OP_DEC_READ_CNT 7 /* decrement object->n_reads on destruction */ -#define FSCACHE_OP_KEEP_FLAGS 0xc0 /* flags to keep when repurposing an op */ +#define FSCACHE_OP_DEC_READ_CNT 6 /* decrement object->n_reads on destruction */ +#define FSCACHE_OP_KEEP_FLAGS 0x0070 /* flags to keep when repurposing an op */ + enum fscache_operation_state state; atomic_t usage; unsigned debug_id; /* debugging ID */ @@ -106,6 +116,7 @@ extern atomic_t fscache_op_debug_id; extern void fscache_op_work_func(struct work_struct *work); extern void fscache_enqueue_operation(struct fscache_operation *); +extern void fscache_op_complete(struct fscache_operation *); extern void fscache_put_operation(struct fscache_operation *); /** @@ -122,6 +133,7 @@ static inline void fscache_operation_init(struct fscache_operation *op, { INIT_WORK(&op->work, fscache_op_work_func); atomic_set(&op->usage, 1); + op->state = FSCACHE_OP_ST_INITIALISED; op->debug_id = atomic_inc_return(&fscache_op_debug_id); op->processor = processor; op->release = release; @@ -138,6 +150,7 @@ struct fscache_retrieval { void *context; /* netfs read context (pinned) */ struct list_head to_do; /* list of things to be done by the backend */ unsigned long start_time; /* time at which retrieval started */ + unsigned n_pages; /* number of pages to be retrieved */ }; typedef int (*fscache_page_retrieval_func_t)(struct fscache_retrieval *op, @@ -173,9 +186,23 @@ static inline void fscache_enqueue_retrieval(struct fscache_retrieval *op) fscache_enqueue_operation(&op->op); } +/** + * fscache_retrieval_complete - Record (partial) completion of a retrieval + * @op: The retrieval operation affected + * @n_pages: The number of pages to account for + */ +static inline void fscache_retrieval_complete(struct fscache_retrieval *op, + int n_pages) +{ + op->n_pages -= n_pages; + if (op->n_pages <= 0) + fscache_op_complete(&op->op); +} + /** * fscache_put_retrieval - Drop a reference to a retrieval operation * @op: The retrieval operation affected + * @n_pages: The number of pages to account for * * Drop a reference to a retrieval operation. */ @@ -333,10 +360,10 @@ struct fscache_object { int debug_id; /* debugging ID */ int n_children; /* number of child objects */ - int n_ops; /* number of ops outstanding on object */ + int n_ops; /* number of extant ops on object */ int n_obj_ops; /* number of object ops outstanding on object */ int n_in_progress; /* number of ops in progress */ - int n_exclusive; /* number of exclusive ops queued */ + int n_exclusive; /* number of exclusive ops queued or in progress */ atomic_t n_reads; /* number of read ops in progress */ spinlock_t lock; /* state and operations lock */ -- cgit v1.2.3 From 9dc8d9bfe4415efb61a5e9390706b8a3bffef329 Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 20 Dec 2012 21:52:36 +0000 Subject: CacheFiles: Implement invalidation Implement invalidation for CacheFiles. This is in two parts: (1) Provide an invalidation method (which just truncates the backing file). (2) Abort attempts to copy anything read from the backing file whilst invalidation is in progress. Question: CacheFiles uses truncation in a couple of places. It has been using notify_change() rather than sys_truncate() or something similar. This means it bypasses a bunch of checks and suchlike that it possibly should be making (security, file locking, lease breaking, vfsmount write). Should it be using vfs_truncate() as added by a preceding patch or should it use notify_write() and assume that anyone poking around in the cache files on disk gets everything they deserve? Signed-off-by: David Howells --- fs/cachefiles/interface.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++ fs/cachefiles/rdwr.c | 5 ++++- 2 files changed, 53 insertions(+), 1 deletion(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c index 9bff0f878cfd..7a9d574b961c 100644 --- a/fs/cachefiles/interface.c +++ b/fs/cachefiles/interface.c @@ -440,6 +440,54 @@ truncate_failed: return ret; } +/* + * Invalidate an object + */ +static void cachefiles_invalidate_object(struct fscache_operation *op) +{ + struct cachefiles_object *object; + struct cachefiles_cache *cache; + const struct cred *saved_cred; + struct path path; + uint64_t ni_size; + int ret; + + object = container_of(op->object, struct cachefiles_object, fscache); + cache = container_of(object->fscache.cache, + struct cachefiles_cache, cache); + + op->object->cookie->def->get_attr(op->object->cookie->netfs_data, + &ni_size); + + _enter("{OBJ%x},[%llu]", + op->object->debug_id, (unsigned long long)ni_size); + + if (object->backer) { + ASSERT(S_ISREG(object->backer->d_inode->i_mode)); + + fscache_set_store_limit(&object->fscache, ni_size); + + path.dentry = object->backer; + path.mnt = cache->mnt; + + cachefiles_begin_secure(cache, &saved_cred); + ret = vfs_truncate(&path, 0); + if (ret == 0) + ret = vfs_truncate(&path, ni_size); + cachefiles_end_secure(cache, saved_cred); + + if (ret != 0) { + fscache_set_store_limit(&object->fscache, 0); + if (ret == -EIO) + cachefiles_io_error_obj(object, + "Invalidate failed"); + } + } + + fscache_op_complete(op); + _leave(""); +} + /* * dissociate a cache from all the pages it was backing */ @@ -455,6 +503,7 @@ const struct fscache_cache_ops cachefiles_cache_ops = { .lookup_complete = cachefiles_lookup_complete, .grab_object = cachefiles_grab_object, .update_object = cachefiles_update_object, + .invalidate_object = cachefiles_invalidate_object, .drop_object = cachefiles_drop_object, .put_object = cachefiles_put_object, .sync_cache = cachefiles_sync_cache, diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index 93a0815e0498..2c994885520a 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -174,7 +174,10 @@ static void cachefiles_read_copier(struct fscache_operation *_op) _debug("- copy {%lu}", monitor->back_page->index); recheck: - if (PageUptodate(monitor->back_page)) { + if (test_bit(FSCACHE_COOKIE_INVALIDATING, + &object->fscache.cookie->flags)) { + error = -ESTALE; + } else if (PageUptodate(monitor->back_page)) { copy_highpage(monitor->netfs_page, monitor->back_page); fscache_mark_page_cached(monitor->op, monitor->netfs_page); -- cgit v1.2.3 From b4cf1e08c8ac95eff65faa53904f7f13ac78194b Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 5 Dec 2012 13:34:45 +0000 Subject: CacheFiles: Add missing retrieval completions CacheFiles is missing some calls to fscache_retrieval_complete() in the error handling/collision paths of its reader functions. This can be seen by the following assertion tripping in fscache_put_operation() whereby the operation being destroyed is still in the in-progress state and has not been cancelled or completed: FS-Cache: Assertion failed 3 == 5 is false ------------[ cut here ]------------ kernel BUG at fs/fscache/operation.c:408! invalid opcode: 0000 [#1] SMP CPU 2 Modules linked in: xfs ioatdma dca loop joydev evdev psmouse dcdbas pcspkr serio_raw i5000_edac edac_core i5k_amb shpchp pci_hotplug sg sr_mod] Pid: 8062, comm: httpd Not tainted 3.1.0-rc8 #1 Dell Inc. PowerEdge 1950/0DT097 RIP: 0010:[] [] fscache_put_operation+0x304/0x330 RSP: 0018:ffff880062f739d8 EFLAGS: 00010296 RAX: 0000000000000025 RBX: ffff8800c5122e84 RCX: ffffffff81ddf040 RDX: 00000000ffffffff RSI: 0000000000000082 RDI: ffffffff81ddef30 RBP: ffff880062f739f8 R08: 0000000000000005 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000003 R12: ffff8800c5122e40 R13: ffff880037a2cd20 R14: ffff880087c7a058 R15: ffff880087c7a000 FS: 00007f63dcf636e0(0000) GS:ffff88022fc80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f0c0a91f000 CR3: 0000000062ec2000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process httpd (pid: 8062, threadinfo ffff880062f72000, task ffff880087e58000) Stack: ffff880062f73bf8 0000000000000000 ffff880062f73bf8 ffff880037a2cd20 ffff880062f73a68 ffffffff8119aa7e ffff88006540e000 ffff880062f73ad4 ffff88008e9a4308 ffff880037a2cd20 ffff880062f73a48 ffff8800c5122e40 Call Trace: [] __fscache_read_or_alloc_pages+0x1fe/0x530 [] __nfs_readpages_from_fscache+0x70/0x1c0 [] nfs_readpages+0xca/0x1e0 [] ? rpc_do_put_task+0x36/0x50 [] ? alloc_nfs_open_context+0x4b/0x110 [] ? rpc_call_sync+0x5a/0x70 [] __do_page_cache_readahead+0x1ca/0x270 [] ra_submit+0x21/0x30 [] ondemand_readahead+0x11d/0x250 [] page_cache_sync_readahead+0x36/0x60 [] generic_file_aio_read+0x454/0x770 [] nfs_file_read+0xe1/0x130 [] do_sync_read+0xd9/0x120 [] ? mntput+0x1f/0x40 [] ? fput+0x1cb/0x260 [] vfs_read+0xc8/0x180 [] sys_read+0x55/0x90 Reported-by: Mark Moseley Signed-off-by: David Howells --- fs/cachefiles/rdwr.c | 14 ++++++++++---- fs/fscache/page.c | 2 ++ 2 files changed, 12 insertions(+), 4 deletions(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index 2c994885520a..480992259707 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -361,8 +361,10 @@ out: read_error: _debug("read error %d", ret); - if (ret == -ENOMEM) + if (ret == -ENOMEM) { + fscache_retrieval_complete(op, 1); goto out; + } io_error: cachefiles_io_error_obj(object, "Page read error on backing file"); fscache_retrieval_complete(op, 1); @@ -551,6 +553,7 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, if (ret < 0) { if (ret == -EEXIST) { page_cache_release(netpage); + fscache_retrieval_complete(op, 1); continue; } goto nomem; @@ -627,6 +630,7 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, if (ret < 0) { if (ret == -EEXIST) { page_cache_release(netpage); + fscache_retrieval_complete(op, 1); continue; } goto nomem; @@ -645,9 +649,9 @@ static int cachefiles_read_backing_file(struct cachefiles_object *object, /* the netpage is unlocked and marked up to date here */ fscache_end_io(op, netpage, 0); - fscache_retrieval_complete(op, 1); page_cache_release(netpage); netpage = NULL; + fscache_retrieval_complete(op, 1); continue; } @@ -682,15 +686,17 @@ out: nomem: _debug("nomem"); ret = -ENOMEM; - goto out; + goto record_page_complete; read_error: _debug("read error %d", ret); if (ret == -ENOMEM) - goto out; + goto record_page_complete; io_error: cachefiles_io_error_obj(object, "Page read error on backing file"); ret = -ENOBUFS; +record_page_complete: + fscache_retrieval_complete(op, 1); goto out; } diff --git a/fs/fscache/page.c b/fs/fscache/page.c index 7bf9d2557052..4dbbca162620 100644 --- a/fs/fscache/page.c +++ b/fs/fscache/page.c @@ -329,6 +329,8 @@ check_if_dead: return -ENOBUFS; } if (unlikely(fscache_object_is_dead(object))) { + pr_err("%s() = -ENOBUFS [obj dead %d]", __func__, op->op.state); + fscache_cancel_op(&op->op); fscache_stat(stat_object_dead); return -ENOBUFS; } -- cgit v1.2.3 From c2d35bfe4b508451b75b5b6bc60a08dbdc44f952 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 5 Dec 2012 13:34:47 +0000 Subject: FS-Cache: Don't mask off the object event mask when printing it Don't mask off the object event mask when printing it. That way it can be seen if threre are bits set that shouldn't be. Signed-off-by: David Howells --- fs/cachefiles/namei.c | 3 +-- fs/fscache/object-list.c | 2 +- 2 files changed, 2 insertions(+), 3 deletions(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c index b0b5f7cdfffa..8c01c5fcdf75 100644 --- a/fs/cachefiles/namei.c +++ b/fs/cachefiles/namei.c @@ -40,8 +40,7 @@ void __cachefiles_printk_object(struct cachefiles_object *object, printk(KERN_ERR "%sobjstate=%s fl=%lx wbusy=%x ev=%lx[%lx]\n", prefix, fscache_object_states[object->fscache.state], object->fscache.flags, work_busy(&object->fscache.work), - object->fscache.events, - object->fscache.event_mask & FSCACHE_OBJECT_EVENTS_MASK); + object->fscache.events, object->fscache.event_mask); printk(KERN_ERR "%sops=%u inp=%u exc=%u\n", prefix, object->fscache.n_ops, object->fscache.n_in_progress, object->fscache.n_exclusive); diff --git a/fs/fscache/object-list.c b/fs/fscache/object-list.c index ebe29c581380..f27c89d17885 100644 --- a/fs/fscache/object-list.c +++ b/fs/fscache/object-list.c @@ -245,7 +245,7 @@ static int fscache_objlist_show(struct seq_file *m, void *v) obj->n_in_progress, obj->n_exclusive, atomic_read(&obj->n_reads), - obj->event_mask & FSCACHE_OBJECT_EVENTS_MASK, + obj->event_mask, obj->events, obj->flags, work_busy(&obj->work)); -- cgit v1.2.3 From 1f372dff1da37e2b36ae9085368fa46896398598 Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 13 Dec 2012 20:03:13 +0000 Subject: FS-Cache: Mark cancellation of in-progress operation Mark as cancelled an operation that is in progress rather than pending at the time it is cancelled, and call fscache_complete_op() to cancel an operation so that blocked ops can be started. Signed-off-by: David Howells --- fs/cachefiles/interface.c | 2 +- fs/fscache/operation.c | 7 ++++--- fs/fscache/page.c | 10 +++++----- include/linux/fscache-cache.h | 4 ++-- 4 files changed, 12 insertions(+), 11 deletions(-) (limited to 'fs/cachefiles') diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c index 7a9d574b961c..746ce532e130 100644 --- a/fs/cachefiles/interface.c +++ b/fs/cachefiles/interface.c @@ -484,7 +484,7 @@ static void cachefiles_invalidate_object(struct fscache_operation *op) } } - fscache_op_complete(op); + fscache_op_complete(op, true); _leave(""); } diff --git a/fs/fscache/operation.c b/fs/fscache/operation.c index 9e6b7d232bb1..36c59604130d 100644 --- a/fs/fscache/operation.c +++ b/fs/fscache/operation.c @@ -363,9 +363,9 @@ void fscache_cancel_all_ops(struct fscache_object *object) } /* - * Record the completion of an in-progress operation. + * Record the completion or cancellation of an in-progress operation. */ -void fscache_op_complete(struct fscache_operation *op) +void fscache_op_complete(struct fscache_operation *op, bool cancelled) { struct fscache_object *object = op->object; @@ -380,7 +380,8 @@ void fscache_op_complete(struct fscache_operation *op) spin_lock(&object->lock); - op->state = FSCACHE_OP_ST_COMPLETE; + op->state = cancelled ? + FSCACHE_OP_ST_CANCELLED : FSCACHE_OP_ST_COMPLETE; if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags)) object->n_exclusive--; diff --git a/fs/fscache/page.c b/fs/fscache/page.c index ef0218f5080d..8a92b9fabe83 100644 --- a/fs/fscache/page.c +++ b/fs/fscache/page.c @@ -171,7 +171,7 @@ static void fscache_attr_changed_op(struct fscache_operation *op) fscache_abort_object(object); } - fscache_op_complete(op); + fscache_op_complete(op, true); _leave(""); } @@ -704,7 +704,7 @@ static void fscache_write_op(struct fscache_operation *_op) * exists, so we should just cancel this write operation. */ spin_unlock(&object->lock); - op->op.state = FSCACHE_OP_ST_CANCELLED; + fscache_op_complete(&op->op, false); _leave(" [inactive]"); return; } @@ -717,7 +717,7 @@ static void fscache_write_op(struct fscache_operation *_op) * cancel this write operation. */ spin_unlock(&object->lock); - op->op.state = FSCACHE_OP_ST_CANCELLED; + fscache_op_complete(&op->op, false); _leave(" [cancel] op{f=%lx s=%u} obj{s=%u f=%lx}", _op->flags, _op->state, object->state, object->flags); return; @@ -755,7 +755,7 @@ static void fscache_write_op(struct fscache_operation *_op) fscache_end_page_write(object, page); if (ret < 0) { fscache_abort_object(object); - fscache_op_complete(&op->op); + fscache_op_complete(&op->op, true); } else { fscache_enqueue_operation(&op->op); } @@ -770,7 +770,7 @@ superseded: spin_unlock(&cookie->stores_lock); clear_bit(FSCACHE_OBJECT_PENDING_WRITE, &object->flags); spin_unlock(&object->lock); - fscache_op_complete(&op->op); + fscache_op_complete(&op->op, true); _leave(""); } diff --git a/include/linux/fscache-cache.h b/include/linux/fscache-cache.h index 73e68c8d5df4..5dfa0aa216b6 100644 --- a/include/linux/fscache-cache.h +++ b/include/linux/fscache-cache.h @@ -116,7 +116,7 @@ extern atomic_t fscache_op_debug_id; extern void fscache_op_work_func(struct work_struct *work); extern void fscache_enqueue_operation(struct fscache_operation *); -extern void fscache_op_complete(struct fscache_operation *); +extern void fscache_op_complete(struct fscache_operation *, bool); extern void fscache_put_operation(struct fscache_operation *); /** @@ -196,7 +196,7 @@ static inline void fscache_retrieval_complete(struct fscache_retrieval *op, { op->n_pages -= n_pages; if (op->n_pages <= 0) - fscache_op_complete(&op->op); + fscache_op_complete(&op->op, true); } /** -- cgit v1.2.3