From 34ce4e7c23e3da578e459b05c6fb17edecb19e6b Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Tue, 15 Dec 2009 19:34:17 +0200 Subject: exofs: debug print even less * Last debug trimming left in some stupid print, remove them. Fixup some other prints * Shift printing from inode.c to ios.c * Add couple of prints when memory allocation fails. Signed-off-by: Boaz Harrosh --- fs/exofs/inode.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 2afbcebeda71..c88a0c5250cb 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -193,7 +193,7 @@ static int __readpages_done(struct page_collect *pcol, bool do_unlock) else good_bytes = pcol->length - resid; - EXOFS_DBGMSG("readpages_done(0x%lx) good_bytes=0x%llx" + EXOFS_DBGMSG2("readpages_done(0x%lx) good_bytes=0x%llx" " length=0x%lx nr_pages=%u\n", pcol->inode->i_ino, _LLU(good_bytes), pcol->length, pcol->nr_pages); @@ -222,7 +222,7 @@ static int __readpages_done(struct page_collect *pcol, bool do_unlock) } pcol_free(pcol); - EXOFS_DBGMSG("readpages_done END\n"); + EXOFS_DBGMSG2("readpages_done END\n"); return ret; } @@ -290,7 +290,7 @@ static int read_exec(struct page_collect *pcol, bool is_sync) atomic_inc(&pcol->sbi->s_curr_pending); - EXOFS_DBGMSG("read_exec obj=0x%llx start=0x%llx length=0x%lx\n", + EXOFS_DBGMSG2("read_exec obj=0x%llx start=0x%llx length=0x%lx\n", ios->obj.id, _LLU(ios->offset), pcol->length); /* pages ownership was passed to pcol_copy */ @@ -462,7 +462,7 @@ static void writepages_done(struct exofs_io_state *ios, void *p) else good_bytes = pcol->length - resid; - EXOFS_DBGMSG("writepages_done(0x%lx) good_bytes=0x%llx" + EXOFS_DBGMSG2("writepages_done(0x%lx) good_bytes=0x%llx" " length=0x%lx nr_pages=%u\n", pcol->inode->i_ino, _LLU(good_bytes), pcol->length, pcol->nr_pages); @@ -490,7 +490,7 @@ static void writepages_done(struct exofs_io_state *ios, void *p) pcol_free(pcol); kfree(pcol); - EXOFS_DBGMSG("writepages_done END\n"); + EXOFS_DBGMSG2("writepages_done END\n"); } static int write_exec(struct page_collect *pcol) @@ -527,7 +527,7 @@ static int write_exec(struct page_collect *pcol) } atomic_inc(&pcol->sbi->s_curr_pending); - EXOFS_DBGMSG("write_exec(0x%lx, 0x%llx) start=0x%llx length=0x%lx\n", + EXOFS_DBGMSG2("write_exec(0x%lx, 0x%llx) start=0x%llx length=0x%lx\n", pcol->inode->i_ino, pcol->pg_first, _LLU(ios->offset), pcol->length); /* pages ownership was passed to pcol_copy */ @@ -616,7 +616,7 @@ try_again: ret = pcol_add_page(pcol, page, len); if (unlikely(ret)) { - EXOFS_DBGMSG("Failed pcol_add_page " + EXOFS_DBGMSG2("Failed pcol_add_page " "nr_pages=%u total_length=0x%lx\n", pcol->nr_pages, pcol->length); @@ -663,7 +663,7 @@ static int exofs_writepages(struct address_space *mapping, if (expected_pages < 32L) expected_pages = 32L; - EXOFS_DBGMSG("inode(0x%lx) wbc->start=0x%llx wbc->end=0x%llx " + EXOFS_DBGMSG2("inode(0x%lx) wbc->start=0x%llx wbc->end=0x%llx " "nrpages=%lu start=0x%lx end=0x%lx expected_pages=%ld\n", mapping->host->i_ino, wbc->range_start, wbc->range_end, mapping->nrpages, start, end, expected_pages); @@ -1170,8 +1170,10 @@ static int exofs_update_inode(struct inode *inode, int do_sync) int ret; args = kzalloc(sizeof(*args), GFP_KERNEL); - if (!args) + if (!args) { + EXOFS_DBGMSG("Faild kzalloc of args\n"); return -ENOMEM; + } fcb = &args->fcb; @@ -1234,7 +1236,8 @@ static int exofs_update_inode(struct inode *inode, int do_sync) free_args: kfree(args); out: - EXOFS_DBGMSG("ret=>%d\n", ret); + EXOFS_DBGMSG("(0x%lx) do_sync=%d ret=>%d\n", + inode->i_ino, do_sync, ret); return ret; } -- cgit v1.2.3 From 45d3abcb1a7388b2b97582e13bf9dd21784dcaa5 Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Thu, 28 Jan 2010 11:46:16 +0200 Subject: exofs: Move layout related members to a layout structure * Abstract away those members in exofs_sb_info that are related/needed by a layout into a new exofs_layout structure. Embed it in exofs_sb_info. * At exofs_io_state receive/keep a pointer to an exofs_layout. No need for an exofs_sb_info pointer, all we need is at exofs_layout. * Change any usage of above exofs_sb_info members to their new name. Signed-off-by: Boaz Harrosh --- fs/exofs/inode.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index c88a0c5250cb..03189a958b33 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -65,7 +65,7 @@ static void _pcol_init(struct page_collect *pcol, unsigned expected_pages, /* Create master bios on first Q, later on cloning, each clone will be * allocated on it's destination Q */ - pcol->req_q = osd_request_queue(sbi->s_ods[0]); + pcol->req_q = osd_request_queue(sbi->layout.s_ods[0]); pcol->inode = inode; pcol->expected_pages = expected_pages; @@ -99,7 +99,7 @@ static int pcol_try_alloc(struct page_collect *pcol) BIO_MAX_PAGES_KMALLOC); if (!pcol->ios) { /* First time allocate io_state */ - int ret = exofs_get_io_state(pcol->sbi, &pcol->ios); + int ret = exofs_get_io_state(&pcol->sbi->layout, &pcol->ios); if (ret) return ret; @@ -872,7 +872,7 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, int ret; *obj_size = ~0; - ret = exofs_get_io_state(sbi, &ios); + ret = exofs_get_io_state(&sbi->layout, &ios); if (unlikely(ret)) { EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__); return ret; @@ -1043,7 +1043,7 @@ static void create_done(struct exofs_io_state *ios, void *p) if (unlikely(ret)) { EXOFS_ERR("object=0x%llx creation faild in pid=0x%llx", - _LLU(exofs_oi_objno(oi)), _LLU(sbi->s_pid)); + _LLU(exofs_oi_objno(oi)), _LLU(sbi->layout.s_pid)); /*TODO: When FS is corrupted creation can fail, object already * exist. Get rid of this asynchronous creation, if exist * increment the obj counter and try the next object. Until we @@ -1104,7 +1104,7 @@ struct inode *exofs_new_inode(struct inode *dir, int mode) mark_inode_dirty(inode); - ret = exofs_get_io_state(sbi, &ios); + ret = exofs_get_io_state(&sbi->layout, &ios); if (unlikely(ret)) { EXOFS_ERR("exofs_new_inode: exofs_get_io_state failed\n"); return ERR_PTR(ret); @@ -1202,7 +1202,7 @@ static int exofs_update_inode(struct inode *inode, int do_sync) } else memcpy(fcb->i_data, oi->i_data, sizeof(fcb->i_data)); - ret = exofs_get_io_state(sbi, &ios); + ret = exofs_get_io_state(&sbi->layout, &ios); if (unlikely(ret)) { EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__); goto free_args; @@ -1286,7 +1286,7 @@ void exofs_delete_inode(struct inode *inode) clear_inode(inode); - ret = exofs_get_io_state(sbi, &ios); + ret = exofs_get_io_state(&sbi->layout, &ios); if (unlikely(ret)) { EXOFS_ERR("%s: exofs_get_io_state failed\n", __func__); return; -- cgit v1.2.3 From d9c740d2253e75db8cef8f87a3125c450f3ebd82 Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Thu, 28 Jan 2010 11:58:08 +0200 Subject: exofs: Define on-disk per-inode optional layout attribute * Layouts describe the way a file is spread on multiple devices. The layout information is stored in the objects attribute introduced in this patch. * There can be multiple generating function for the layout. Currently defined: - No attribute present - use below moving-window on global device table, all devices. (This is the only one currently used in exofs) - an obj_id generated moving window - the obj_id is a randomizing factor in the otherwise global map layout. - An explicit layout stored, including a data_map and a device index list. - More might be defined in future ... * There are two attributes defined of the same structure: A-data-files-layout - This layout is used by data-files. If present at a directory, all files of that directory will be created with this layout. A-meta-data-layout - This layout is used by a directory and other meta-data information. Also inherited at creation of subdirectories. * At creation time inodes are created with the layout specified above. A usermode utility may change the creation layout on a give directory or file. Which in the case of directories, will also apply to newly created files/subdirectories, children of that directory. In the simple unaltered case of a newly created exofs, no layout attributes are present, and all layouts adhere to the layout specified at the device-table. * In case of a future file system loaded in an old exofs-driver. At iget(), the generating_function is inspected and if not supported will return an IO error to the application and the inode will not be loaded. So not to damage any data. Note: After this patch we do not yet support any type of layout only the RAID0 patch that enables striping at the super-block level will add support for RAID0 layouts above. This way we are past and future compatible and fully bisectable. * Access to the device table is done by an accessor since it will change according to above information. Signed-off-by: Boaz Harrosh --- fs/exofs/inode.c | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 51 insertions(+), 5 deletions(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 03189a958b33..0163546ba05a 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -859,6 +859,15 @@ int exofs_setattr(struct dentry *dentry, struct iattr *iattr) return error; } +static const struct osd_attr g_attr_inode_file_layout = ATTR_DEF( + EXOFS_APAGE_FS_DATA, + EXOFS_ATTR_INODE_FILE_LAYOUT, + 0); +static const struct osd_attr g_attr_inode_dir_layout = ATTR_DEF( + EXOFS_APAGE_FS_DATA, + EXOFS_ATTR_INODE_DIR_LAYOUT, + 0); + /* * Read an inode from the OSD, and return it as is. We also return the size * attribute in the 'obj_size' argument. @@ -867,11 +876,16 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, struct exofs_fcb *inode, uint64_t *obj_size) { struct exofs_sb_info *sbi = sb->s_fs_info; - struct osd_attr attrs[2]; + struct osd_attr attrs[] = { + [0] = g_attr_inode_data, + [1] = g_attr_inode_file_layout, + [2] = g_attr_inode_dir_layout, + [3] = g_attr_logical_length, + }; struct exofs_io_state *ios; + struct exofs_on_disk_inode_layout *layout; int ret; - *obj_size = ~0; ret = exofs_get_io_state(&sbi->layout, &ios); if (unlikely(ret)) { EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__); @@ -882,8 +896,9 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, exofs_make_credential(oi->i_cred, &ios->obj); ios->cred = oi->i_cred; - attrs[0] = g_attr_inode_data; - attrs[1] = g_attr_logical_length; + attrs[1].len = exofs_on_disk_inode_layout_size(sbi->layout.s_numdevs); + attrs[2].len = exofs_on_disk_inode_layout_size(sbi->layout.s_numdevs); + ios->in_attr = attrs; ios->in_attr_len = ARRAY_SIZE(attrs); @@ -900,12 +915,43 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, memcpy(inode, attrs[0].val_ptr, EXOFS_INO_ATTR_SIZE); ret = extract_attr_from_ios(ios, &attrs[1]); + if (ret) { + EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__); + goto out; + } + if (attrs[1].len) { + layout = attrs[1].val_ptr; + if (layout->gen_func != cpu_to_le16(LAYOUT_MOVING_WINDOW)) { + EXOFS_ERR("%s: unsupported files layout %d\n", + __func__, layout->gen_func); + ret = -ENOTSUPP; + goto out; + } + } + + ret = extract_attr_from_ios(ios, &attrs[2]); + if (ret) { + EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__); + goto out; + } + if (attrs[2].len) { + layout = attrs[2].val_ptr; + if (layout->gen_func != cpu_to_le16(LAYOUT_MOVING_WINDOW)) { + EXOFS_ERR("%s: unsupported meta-data layout %d\n", + __func__, layout->gen_func); + ret = -ENOTSUPP; + goto out; + } + } + + *obj_size = ~0; + ret = extract_attr_from_ios(ios, &attrs[3]); if (ret) { EXOFS_ERR("%s: extract_attr of logical_length failed\n", __func__); goto out; } - *obj_size = get_unaligned_be64(attrs[1].val_ptr); + *obj_size = get_unaligned_be64(attrs[3].val_ptr); out: exofs_put_io_state(ios); -- cgit v1.2.3 From 5d952b8391692553c31e620a92d6e09262a9a307 Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Mon, 1 Feb 2010 13:35:51 +0200 Subject: exofs: RAID0 support We now support striping over mirror devices. Including variable sized stripe_unit. Some limits: * stripe_unit must be a multiple of PAGE_SIZE * stripe_unit * stripe_count is maximum upto 32-bit (4Gb) Tested RAID0 over mirrors, RAID0 only, mirrors only. All check. Design notes: * I'm not using a vectored raid-engine mechanism yet. Following the pnfs-objects-layout data-map structure, "Mirror" is just a private case of "group_width" == 1, and RAID0 is a private case of "Mirrors" == 1. The performance lose of the general case over the particular special case optimization is totally negligible, also considering the extra code size. * In general I added a prepare_stripes() stage that divides the to-be-io pages to the participating devices, the previous exofs_ios_write/read, now becomes _write/read_mirrors and a new write/read upper layer loops on all devices calling _write/read_mirrors. Effectively the prepare_stripes stage is the all secret. Also truncate need fixing to accommodate for striping. * In a RAID0 arrangement, in a regular usage scenario, if all inode layouts will start at the same device, the small files fill up the first device and the later devices stay empty, the farther the device the emptier it is. To fix that, each inode will start at a different stripe_unit, according to it's obj_id modulus number-of-stripe-units. And will then span all stripe-units in the same incrementing order wrapping back to the beginning of the device table. We call it a stripe-units moving window. Special consideration was taken to keep all devices in a mirror arrangement identical. So a broken osd-device could just be cloned from one of the mirrors and no FS scrubbing is needed. (We do that by rotating stripe-unit at a time and not a single device at a time.) TODO: We no longer verify object_length == inode->i_size in exofs_iget. (since i_size is stripped on multiple objects now). I should introduce a multiple-device attribute reading, and use it in exofs_iget. Signed-off-by: Boaz Harrosh --- fs/exofs/inode.c | 26 ++++---------------------- 1 file changed, 4 insertions(+), 22 deletions(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 0163546ba05a..2b3163ea56eb 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -869,18 +869,17 @@ static const struct osd_attr g_attr_inode_dir_layout = ATTR_DEF( 0); /* - * Read an inode from the OSD, and return it as is. We also return the size - * attribute in the 'obj_size' argument. + * Read the Linux inode info from the OSD, and return it as is. In exofs the + * inode info is in an application specific page/attribute of the osd-object. */ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, - struct exofs_fcb *inode, uint64_t *obj_size) + struct exofs_fcb *inode) { struct exofs_sb_info *sbi = sb->s_fs_info; struct osd_attr attrs[] = { [0] = g_attr_inode_data, [1] = g_attr_inode_file_layout, [2] = g_attr_inode_dir_layout, - [3] = g_attr_logical_length, }; struct exofs_io_state *ios; struct exofs_on_disk_inode_layout *layout; @@ -944,15 +943,6 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, } } - *obj_size = ~0; - ret = extract_attr_from_ios(ios, &attrs[3]); - if (ret) { - EXOFS_ERR("%s: extract_attr of logical_length failed\n", - __func__); - goto out; - } - *obj_size = get_unaligned_be64(attrs[3].val_ptr); - out: exofs_put_io_state(ios); return ret; @@ -971,7 +961,6 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino) struct exofs_i_info *oi; struct exofs_fcb fcb; struct inode *inode; - uint64_t obj_size; int ret; inode = iget_locked(sb, ino); @@ -983,7 +972,7 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino) __oi_init(oi); /* read the inode from the osd */ - ret = exofs_get_inode(sb, oi, &fcb, &obj_size); + ret = exofs_get_inode(sb, oi, &fcb); if (ret) goto bad_inode; @@ -1004,13 +993,6 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino) inode->i_blkbits = EXOFS_BLKSHIFT; inode->i_generation = le32_to_cpu(fcb.i_generation); - if ((inode->i_size != obj_size) && - (!exofs_inode_is_fast_symlink(inode))) { - EXOFS_ERR("WARNING: Size of inode=%llu != object=%llu\n", - inode->i_size, _LLU(obj_size)); - /* FIXME: call exofs_inode_recovery() */ - } - oi->i_dir_start_lookup = 0; if ((inode->i_nlink == 0) && (inode->i_mode == 0)) { -- cgit v1.2.3 From 86093aaff5be5b214613eb60553e236bdb389c84 Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Thu, 28 Jan 2010 18:24:06 +0200 Subject: exofs: convert io_state to use pages array instead of bio at input * inode.c operations are full-pages based, and not actually true scatter-gather * Lets us use more pages at once upto 512 (from 249) in 64 bit * Brings us much much closer to be able to use exofs's io_state engine from objlayout driver. (Once I decide where to put the common code) After RAID0 patch the outer (input) bio was never used as a bio, but was simply a page carrier into the raid engine. Even in the simple mirror/single-dev arrangement pages info was copied into a second bio. It is now easer to just pass a pages array into the io_state and prepare bio(s) once. Signed-off-by: Boaz Harrosh --- fs/exofs/inode.c | 81 ++++++++++++++++++++++++++++---------------------------- 1 file changed, 41 insertions(+), 40 deletions(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 2b3163ea56eb..6ca0b0117f04 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -41,16 +41,18 @@ enum { BIO_MAX_PAGES_KMALLOC = (PAGE_SIZE - sizeof(struct bio)) / sizeof(struct bio_vec), + MAX_PAGES_KMALLOC = + PAGE_SIZE / sizeof(struct page *), }; struct page_collect { struct exofs_sb_info *sbi; - struct request_queue *req_q; struct inode *inode; unsigned expected_pages; struct exofs_io_state *ios; - struct bio *bio; + struct page **pages; + unsigned alloc_pages; unsigned nr_pages; unsigned long length; loff_t pg_first; /* keep 64bit also in 32-arches */ @@ -62,15 +64,12 @@ static void _pcol_init(struct page_collect *pcol, unsigned expected_pages, struct exofs_sb_info *sbi = inode->i_sb->s_fs_info; pcol->sbi = sbi; - /* Create master bios on first Q, later on cloning, each clone will be - * allocated on it's destination Q - */ - pcol->req_q = osd_request_queue(sbi->layout.s_ods[0]); pcol->inode = inode; pcol->expected_pages = expected_pages; pcol->ios = NULL; - pcol->bio = NULL; + pcol->pages = NULL; + pcol->alloc_pages = 0; pcol->nr_pages = 0; pcol->length = 0; pcol->pg_first = -1; @@ -80,7 +79,8 @@ static void _pcol_reset(struct page_collect *pcol) { pcol->expected_pages -= min(pcol->nr_pages, pcol->expected_pages); - pcol->bio = NULL; + pcol->pages = NULL; + pcol->alloc_pages = 0; pcol->nr_pages = 0; pcol->length = 0; pcol->pg_first = -1; @@ -90,13 +90,13 @@ static void _pcol_reset(struct page_collect *pcol) * it might not end here. don't be left with nothing */ if (!pcol->expected_pages) - pcol->expected_pages = BIO_MAX_PAGES_KMALLOC; + pcol->expected_pages = MAX_PAGES_KMALLOC; } static int pcol_try_alloc(struct page_collect *pcol) { - int pages = min_t(unsigned, pcol->expected_pages, - BIO_MAX_PAGES_KMALLOC); + unsigned pages = min_t(unsigned, pcol->expected_pages, + MAX_PAGES_KMALLOC); if (!pcol->ios) { /* First time allocate io_state */ int ret = exofs_get_io_state(&pcol->sbi->layout, &pcol->ios); @@ -105,23 +105,28 @@ static int pcol_try_alloc(struct page_collect *pcol) return ret; } + /* TODO: easily support bio chaining */ + pages = min_t(unsigned, pages, + pcol->sbi->layout.group_width * BIO_MAX_PAGES_KMALLOC); + for (; pages; pages >>= 1) { - pcol->bio = bio_kmalloc(GFP_KERNEL, pages); - if (likely(pcol->bio)) + pcol->pages = kmalloc(pages * sizeof(struct page *), + GFP_KERNEL); + if (likely(pcol->pages)) { + pcol->alloc_pages = pages; return 0; + } } - EXOFS_ERR("Failed to bio_kmalloc expected_pages=%u\n", + EXOFS_ERR("Failed to kmalloc expected_pages=%u\n", pcol->expected_pages); return -ENOMEM; } static void pcol_free(struct page_collect *pcol) { - if (pcol->bio) { - bio_put(pcol->bio); - pcol->bio = NULL; - } + kfree(pcol->pages); + pcol->pages = NULL; if (pcol->ios) { exofs_put_io_state(pcol->ios); @@ -132,11 +137,10 @@ static void pcol_free(struct page_collect *pcol) static int pcol_add_page(struct page_collect *pcol, struct page *page, unsigned len) { - int added_len = bio_add_pc_page(pcol->req_q, pcol->bio, page, len, 0); - if (unlikely(len != added_len)) + if (unlikely(pcol->nr_pages >= pcol->alloc_pages)) return -ENOMEM; - ++pcol->nr_pages; + pcol->pages[pcol->nr_pages++] = page; pcol->length += len; return 0; } @@ -181,7 +185,6 @@ static void update_write_page(struct page *page, int ret) */ static int __readpages_done(struct page_collect *pcol, bool do_unlock) { - struct bio_vec *bvec; int i; u64 resid; u64 good_bytes; @@ -198,8 +201,8 @@ static int __readpages_done(struct page_collect *pcol, bool do_unlock) pcol->inode->i_ino, _LLU(good_bytes), pcol->length, pcol->nr_pages); - __bio_for_each_segment(bvec, pcol->bio, i, 0) { - struct page *page = bvec->bv_page; + for (i = 0; i < pcol->nr_pages; i++) { + struct page *page = pcol->pages[i]; struct inode *inode = page->mapping->host; int page_stat; @@ -218,7 +221,7 @@ static int __readpages_done(struct page_collect *pcol, bool do_unlock) ret = update_read_page(page, page_stat); if (do_unlock) unlock_page(page); - length += bvec->bv_len; + length += PAGE_SIZE; } pcol_free(pcol); @@ -238,11 +241,10 @@ static void readpages_done(struct exofs_io_state *ios, void *p) static void _unlock_pcol_pages(struct page_collect *pcol, int ret, int rw) { - struct bio_vec *bvec; int i; - __bio_for_each_segment(bvec, pcol->bio, i, 0) { - struct page *page = bvec->bv_page; + for (i = 0; i < pcol->nr_pages; i++) { + struct page *page = pcol->pages[i]; if (rw == READ) update_read_page(page, ret); @@ -260,13 +262,14 @@ static int read_exec(struct page_collect *pcol, bool is_sync) struct page_collect *pcol_copy = NULL; int ret; - if (!pcol->bio) + if (!pcol->pages) return 0; /* see comment in _readpage() about sync reads */ WARN_ON(is_sync && (pcol->nr_pages != 1)); - ios->bio = pcol->bio; + ios->pages = pcol->pages; + ios->nr_pages = pcol->nr_pages; ios->length = pcol->length; ios->offset = pcol->pg_first << PAGE_CACHE_SHIFT; @@ -366,7 +369,7 @@ try_again: goto try_again; } - if (!pcol->bio) { + if (!pcol->pages) { ret = pcol_try_alloc(pcol); if (unlikely(ret)) goto fail; @@ -448,7 +451,6 @@ static int exofs_readpage(struct file *file, struct page *page) static void writepages_done(struct exofs_io_state *ios, void *p) { struct page_collect *pcol = p; - struct bio_vec *bvec; int i; u64 resid; u64 good_bytes; @@ -467,8 +469,8 @@ static void writepages_done(struct exofs_io_state *ios, void *p) pcol->inode->i_ino, _LLU(good_bytes), pcol->length, pcol->nr_pages); - __bio_for_each_segment(bvec, pcol->bio, i, 0) { - struct page *page = bvec->bv_page; + for (i = 0; i < pcol->nr_pages; i++) { + struct page *page = pcol->pages[i]; struct inode *inode = page->mapping->host; int page_stat; @@ -485,7 +487,7 @@ static void writepages_done(struct exofs_io_state *ios, void *p) EXOFS_DBGMSG2(" writepages_done(0x%lx, 0x%lx) status=%d\n", inode->i_ino, page->index, page_stat); - length += bvec->bv_len; + length += PAGE_SIZE; } pcol_free(pcol); @@ -500,7 +502,7 @@ static int write_exec(struct page_collect *pcol) struct page_collect *pcol_copy = NULL; int ret; - if (!pcol->bio) + if (!pcol->pages) return 0; pcol_copy = kmalloc(sizeof(*pcol_copy), GFP_KERNEL); @@ -512,9 +514,8 @@ static int write_exec(struct page_collect *pcol) *pcol_copy = *pcol; - pcol_copy->bio->bi_rw |= (1 << BIO_RW); /* FIXME: bio_set_dir() */ - - ios->bio = pcol_copy->bio; + ios->pages = pcol_copy->pages; + ios->nr_pages = pcol_copy->nr_pages; ios->offset = pcol_copy->pg_first << PAGE_CACHE_SHIFT; ios->length = pcol_copy->length; ios->done = writepages_done; @@ -605,7 +606,7 @@ try_again: goto try_again; } - if (!pcol->bio) { + if (!pcol->pages) { ret = pcol_try_alloc(pcol); if (unlikely(ret)) goto fail; -- cgit v1.2.3 From 96391e2bae0f8882b6f44809202a68be66e91dce Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Tue, 9 Feb 2010 11:43:21 +0200 Subject: exofs: Error recovery if object is missing from storage If an object is referenced by a directory but does not exist on a target, it is a very serious corruption that means: 1. Either a power failure with very slim chance of it happening. Because the directory update is always submitted much after object creation, but if a directory is written to one device and the object creation to another it might theoretically happen. 2. It only ever happened to me while developing with BUGs causing file corruption. Crashes could also cause it but they are more like case 1. In any way the object does not exist, so data is surely lost. If there is a mix-up in the obj-id or data-map, then lost objects can be salvaged by off-line fsck. The only recoverable information is the directory name. By letting it appear as a regular empty file, with date==0 (1970 Jan 1st) ownership to root, we enable recovery of the only useful information. And also enable deletion or over-write. I can see how this can hurt. Signed-off-by: Boaz Harrosh --- fs/exofs/inode.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 6ca0b0117f04..5514f3c2c2f4 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -903,8 +903,18 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi, ios->in_attr_len = ARRAY_SIZE(attrs); ret = exofs_sbi_read(ios); - if (ret) + if (unlikely(ret)) { + EXOFS_ERR("object(0x%llx) corrupted, return empty file=>%d\n", + _LLU(ios->obj.id), ret); + memset(inode, 0, sizeof(*inode)); + inode->i_mode = 0040000 | (0777 & ~022); + /* If object is lost on target we might as well enable it's + * delete. + */ + if ((ret == -ENOENT) || (ret == -EINVAL)) + ret = 0; goto out; + } ret = extract_attr_from_ios(ios, &attrs[0]); if (ret) { -- cgit v1.2.3 From a9185b41a4f84971b930c519f0c63bd450c4810d Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Fri, 5 Mar 2010 09:21:37 +0100 Subject: pass writeback_control to ->write_inode This gives the filesystem more information about the writeback that is happening. Trond requested this for the NFS unstable write handling, and other filesystems might benefit from this too by beeing able to distinguish between the different callers in more detail. Signed-off-by: Christoph Hellwig Signed-off-by: Al Viro --- fs/exofs/inode.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/exofs/inode.c') diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index 5514f3c2c2f4..a17e4b733e35 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -1280,9 +1280,9 @@ out: return ret; } -int exofs_write_inode(struct inode *inode, int wait) +int exofs_write_inode(struct inode *inode, struct writeback_control *wbc) { - return exofs_update_inode(inode, wait); + return exofs_update_inode(inode, wbc->sync_mode == WB_SYNC_ALL); } /* -- cgit v1.2.3