From 6a5fa2362b628ee950080bef8895a6fb62f58ab4 Mon Sep 17 00:00:00 2001 From: Steve French Date: Fri, 1 Jan 2010 01:28:43 +0000 Subject: [CIFS] Add support for TCP_NODELAY mount option sockopt=TCP_NODELAY helpful for faster networks boosting performance. Kernel bugzilla bug number 14032. Signed-off-by: Steve French --- fs/cifs/CHANGES | 4 ++++ fs/cifs/cifsfs.h | 2 +- fs/cifs/cifsglob.h | 1 + fs/cifs/connect.c | 30 ++++++++++++++++++++++++++---- 4 files changed, 32 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/cifs/CHANGES b/fs/cifs/CHANGES index 7b2600b380d7..49503d2edc7e 100644 --- a/fs/cifs/CHANGES +++ b/fs/cifs/CHANGES @@ -1,3 +1,7 @@ +Version 1.62 +------------ +Add sockopt=TCP_NODELAY mount option. + Version 1.61 ------------ Fix append problem to Samba servers (files opened with O_APPEND could diff --git a/fs/cifs/cifsfs.h b/fs/cifs/cifsfs.h index ac2b24c192f8..78c1b86d55f6 100644 --- a/fs/cifs/cifsfs.h +++ b/fs/cifs/cifsfs.h @@ -113,5 +113,5 @@ extern long cifs_ioctl(struct file *filep, unsigned int cmd, unsigned long arg); extern const struct export_operations cifs_export_ops; #endif /* EXPERIMENTAL */ -#define CIFS_VERSION "1.61" +#define CIFS_VERSION "1.62" #endif /* _CIFSFS_H */ diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h index 4b35f7ec0583..ed751bb657db 100644 --- a/fs/cifs/cifsglob.h +++ b/fs/cifs/cifsglob.h @@ -149,6 +149,7 @@ struct TCP_Server_Info { bool svlocal:1; /* local server or remote */ bool noblocksnd; /* use blocking sendmsg */ bool noautotune; /* do not autotune send buf sizes */ + bool tcp_nodelay; atomic_t inFlight; /* number of requests on the wire to server */ #ifdef CONFIG_CIFS_STATS2 atomic_t inSend; /* requests trying to send */ diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c index 3bbcaa716b3c..2e9e09ca0e30 100644 --- a/fs/cifs/connect.c +++ b/fs/cifs/connect.c @@ -98,7 +98,7 @@ struct smb_vol { bool nostrictsync:1; /* do not force expensive SMBflush on every sync */ unsigned int rsize; unsigned int wsize; - unsigned int sockopt; + bool sockopt_tcp_nodelay:1; unsigned short int port; char *prepath; }; @@ -1142,9 +1142,11 @@ cifs_parse_mount_options(char *options, const char *devname, simple_strtoul(value, &value, 0); } } else if (strnicmp(data, "sockopt", 5) == 0) { - if (value && *value) { - vol->sockopt = - simple_strtoul(value, &value, 0); + if (!value || !*value) { + cERROR(1, ("no socket option specified")); + continue; + } else if (strnicmp(value, "TCP_NODELAY", 11) == 0) { + vol->sockopt_tcp_nodelay = 1; } } else if (strnicmp(data, "netbiosname", 4) == 0) { if (!value || !*value || (*value == ' ')) { @@ -1514,6 +1516,7 @@ cifs_get_tcp_session(struct smb_vol *volume_info) tcp_ses->noblocksnd = volume_info->noblocksnd; tcp_ses->noautotune = volume_info->noautotune; + tcp_ses->tcp_nodelay = volume_info->sockopt_tcp_nodelay; atomic_set(&tcp_ses->inFlight, 0); init_waitqueue_head(&tcp_ses->response_q); init_waitqueue_head(&tcp_ses->request_q); @@ -1764,6 +1767,7 @@ static int ipv4_connect(struct TCP_Server_Info *server) { int rc = 0; + int val; bool connected = false; __be16 orig_port = 0; struct socket *socket = server->ssocket; @@ -1845,6 +1849,14 @@ ipv4_connect(struct TCP_Server_Info *server) socket->sk->sk_rcvbuf = 140 * 1024; } + if (server->tcp_nodelay) { + val = 1; + rc = kernel_setsockopt(socket, SOL_TCP, TCP_NODELAY, + (char *)&val, sizeof(val)); + if (rc) + cFYI(1, ("set TCP_NODELAY socket option error %d", rc)); + } + cFYI(1, ("sndbuf %d rcvbuf %d rcvtimeo 0x%lx", socket->sk->sk_sndbuf, socket->sk->sk_rcvbuf, socket->sk->sk_rcvtimeo)); @@ -1916,6 +1928,7 @@ static int ipv6_connect(struct TCP_Server_Info *server) { int rc = 0; + int val; bool connected = false; __be16 orig_port = 0; struct socket *socket = server->ssocket; @@ -1987,6 +2000,15 @@ ipv6_connect(struct TCP_Server_Info *server) */ socket->sk->sk_rcvtimeo = 7 * HZ; socket->sk->sk_sndtimeo = 5 * HZ; + + if (server->tcp_nodelay) { + val = 1; + rc = kernel_setsockopt(socket, SOL_TCP, TCP_NODELAY, + (char *)&val, sizeof(val)); + if (rc) + cFYI(1, ("set TCP_NODELAY socket option error %d", rc)); + } + server->ssocket = socket; return rc; -- cgit v1.2.3 From 1dd473fdf1d8a7531e0955480cd129f9c1e8b8a3 Mon Sep 17 00:00:00 2001 From: OGAWA Hirofumi Date: Tue, 12 Jan 2010 03:37:45 +0900 Subject: ocfs2: Fix refcnt leak on ocfs2_fast_follow_link() error path If ->follow_link handler returns an error, it should decrement nd->path refcnt. But ocfs2_fast_follow_link() doesn't decrement. This patch fixes the problem by using nd_set_link() style error handling instead of playing with nd->path. Signed-off-by: OGAWA Hirofumi Signed-off-by: Joel Becker --- fs/ocfs2/symlink.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/symlink.c b/fs/ocfs2/symlink.c index 49b133ccbf11..32499d213fc4 100644 --- a/fs/ocfs2/symlink.c +++ b/fs/ocfs2/symlink.c @@ -137,20 +137,20 @@ static void *ocfs2_fast_follow_link(struct dentry *dentry, } memcpy(link, target, len); - nd_set_link(nd, link); bail: + nd_set_link(nd, status ? ERR_PTR(status) : link); brelse(bh); mlog_exit(status); - return status ? ERR_PTR(status) : link; + return NULL; } static void ocfs2_fast_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie) { - char *link = cookie; - - kfree(link); + char *link = nd_get_link(nd); + if (!IS_ERR(link)) + kfree(link); } const struct inode_operations ocfs2_symlink_inode_operations = { -- cgit v1.2.3 From 0f585f14d4e34ab701283e9237ac7695cd7c9e31 Mon Sep 17 00:00:00 2001 From: OGAWA Hirofumi Date: Tue, 12 Jan 2010 03:36:57 +0900 Subject: GFS2: Fix refcnt leak on gfs2_follow_link() error path If ->follow_link handler return the error, it should decrement nd->path refcnt. This patch fix it. Signed-off-by: OGAWA Hirofumi Signed-off-by: Steven Whitehouse --- fs/gfs2/ops_inode.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/ops_inode.c b/fs/gfs2/ops_inode.c index 78f73ca1ef3e..84350e1be66d 100644 --- a/fs/gfs2/ops_inode.c +++ b/fs/gfs2/ops_inode.c @@ -1088,7 +1088,8 @@ static void *gfs2_follow_link(struct dentry *dentry, struct nameidata *nd) error = vfs_follow_link(nd, buf); if (buf != array) kfree(buf); - } + } else + path_put(&nd->path); return ERR_PTR(error); } -- cgit v1.2.3 From 1db913823c0f8360fccbd24ca67eb073966a5ffd Mon Sep 17 00:00:00 2001 From: "Aneesh Kumar K.V" Date: Fri, 22 Jan 2010 17:06:20 -0500 Subject: ext4: Handle -EDQUOT error on write We need to release the journal before we do a write_inode. Otherwise we could deadlock. Signed-off-by: Aneesh Kumar K.V --- fs/ext4/inode.c | 32 ++++++++++++++++++-------------- 1 file changed, 18 insertions(+), 14 deletions(-) (limited to 'fs') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index c818972c8302..60b3a19e9927 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1835,24 +1835,12 @@ repeat: * later. Real quota accounting is done at pages writeout * time. */ - if (vfs_dq_reserve_block(inode, md_needed + 1)) { - /* - * We tend to badly over-estimate the amount of - * metadata blocks which are needed, so if we have - * reserved any metadata blocks, try to force out the - * inode and see if we have any better luck. - */ - if (md_reserved && retries++ <= 3) - goto retry; + if (vfs_dq_reserve_block(inode, md_needed + 1)) return -EDQUOT; - } if (ext4_claim_free_blocks(sbi, md_needed + 1)) { vfs_dq_release_reservation_block(inode, md_needed + 1); if (ext4_should_retry_alloc(inode->i_sb, &retries)) { - retry: - if (md_reserved) - write_inode_now(inode, (retries == 3)); yield(); goto repeat; } @@ -3032,7 +3020,7 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, struct page **pagep, void **fsdata) { - int ret, retries = 0; + int ret, retries = 0, quota_retries = 0; struct page *page; pgoff_t index; unsigned from, to; @@ -3091,6 +3079,22 @@ retry: if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries)) goto retry; + + if ((ret == -EDQUOT) && + EXT4_I(inode)->i_reserved_meta_blocks && + (quota_retries++ < 3)) { + /* + * Since we often over-estimate the number of meta + * data blocks required, we may sometimes get a + * spurios out of quota error even though there would + * be enough space once we write the data blocks and + * find out how many meta data blocks were _really_ + * required. So try forcing the inode write to see if + * that helps. + */ + write_inode_now(inode, (quota_retries == 3)); + goto retry; + } out: return ret; } -- cgit v1.2.3 From 5f634d064c709ea02c3cdaa850a08323a4a4bf28 Mon Sep 17 00:00:00 2001 From: "Aneesh Kumar K.V" Date: Mon, 25 Jan 2010 04:00:31 -0500 Subject: ext4: Fix quota accounting error with fallocate When we fallocate a region of the file which we had recently written, and which is still in the page cache marked as delayed allocated blocks we need to make sure we don't do the quota update on writepage path. This is because the needed quota updated would have already be done by fallocate. Signed-off-by: Aneesh Kumar K.V --- fs/ext4/ext4.h | 2 ++ fs/ext4/extents.c | 21 +++++++++++++++++++++ fs/ext4/inode.c | 44 +++++++++++++++++++++++++++++++------------- 3 files changed, 54 insertions(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index af7b62699ea9..b98de17e542a 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1443,6 +1443,8 @@ extern int ext4_block_truncate_page(handle_t *handle, extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); extern qsize_t *ext4_get_reserved_space(struct inode *inode); extern int flush_aio_dio_completed_IO(struct inode *inode); +extern void ext4_da_update_reserve_space(struct inode *inode, + int used, int quota_claim); /* ioctl.c */ extern long ext4_ioctl(struct file *, unsigned int, unsigned long); extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long); diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 7d7b74e94687..3b6ff72026f0 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3132,7 +3132,19 @@ out: unmap_underlying_metadata_blocks(inode->i_sb->s_bdev, newblock + max_blocks, allocated - max_blocks); + allocated = max_blocks; } + + /* + * If we have done fallocate with the offset that is already + * delayed allocated, we would have block reservation + * and quota reservation done in the delayed write path. + * But fallocate would have already updated quota and block + * count for this offset. So cancel these reservation + */ + if (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE) + ext4_da_update_reserve_space(inode, allocated, 0); + map_out: set_buffer_mapped(bh_result); out1: @@ -3368,8 +3380,17 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, /* previous routine could use block we allocated */ newblock = ext_pblock(&newex); allocated = ext4_ext_get_actual_len(&newex); + if (allocated > max_blocks) + allocated = max_blocks; set_buffer_new(bh_result); + /* + * Update reserved blocks/metadata blocks after successful + * block allocation which had been deferred till now. + */ + if (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE) + ext4_da_update_reserve_space(inode, allocated, 1); + /* * Cache the extent and update transaction to commit on fdatasync only * when it is _not_ an uninitialized extent. diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 60b3a19e9927..c955f6490b78 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1053,11 +1053,12 @@ static int ext4_calc_metadata_amount(struct inode *inode, sector_t lblock) * Called with i_data_sem down, which is important since we can call * ext4_discard_preallocations() from here. */ -static void ext4_da_update_reserve_space(struct inode *inode, int used) +void ext4_da_update_reserve_space(struct inode *inode, + int used, int quota_claim) { struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); struct ext4_inode_info *ei = EXT4_I(inode); - int mdb_free = 0; + int mdb_free = 0, allocated_meta_blocks = 0; spin_lock(&ei->i_block_reservation_lock); if (unlikely(used > ei->i_reserved_data_blocks)) { @@ -1073,6 +1074,7 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used) ei->i_reserved_data_blocks -= used; used += ei->i_allocated_meta_blocks; ei->i_reserved_meta_blocks -= ei->i_allocated_meta_blocks; + allocated_meta_blocks = ei->i_allocated_meta_blocks; ei->i_allocated_meta_blocks = 0; percpu_counter_sub(&sbi->s_dirtyblocks_counter, used); @@ -1090,9 +1092,23 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used) spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); /* Update quota subsystem */ - vfs_dq_claim_block(inode, used); - if (mdb_free) - vfs_dq_release_reservation_block(inode, mdb_free); + if (quota_claim) { + vfs_dq_claim_block(inode, used); + if (mdb_free) + vfs_dq_release_reservation_block(inode, mdb_free); + } else { + /* + * We did fallocate with an offset that is already delayed + * allocated. So on delayed allocated writeback we should + * not update the quota for allocated blocks. But then + * converting an fallocate region to initialized region would + * have caused a metadata allocation. So claim quota for + * that + */ + if (allocated_meta_blocks) + vfs_dq_claim_block(inode, allocated_meta_blocks); + vfs_dq_release_reservation_block(inode, mdb_free + used); + } /* * If we have done all the pending block allocations and if @@ -1292,18 +1308,20 @@ int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block, */ EXT4_I(inode)->i_state &= ~EXT4_STATE_EXT_MIGRATE; } - } + /* + * Update reserved blocks/metadata blocks after successful + * block allocation which had been deferred till now. We don't + * support fallocate for non extent files. So we can update + * reserve space here. + */ + if ((retval > 0) && + (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE)) + ext4_da_update_reserve_space(inode, retval, 1); + } if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) EXT4_I(inode)->i_delalloc_reserved_flag = 0; - /* - * Update reserved blocks/metadata blocks after successful - * block allocation which had been deferred till now. - */ - if ((retval > 0) && (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE)) - ext4_da_update_reserve_space(inode, retval); - up_write((&EXT4_I(inode)->i_data_sem)); if (retval > 0 && buffer_mapped(bh)) { int ret = check_block_validity(inode, "file system " -- cgit v1.2.3 From 1296cc85c26e94eb865d03f82140f27d598de467 Mon Sep 17 00:00:00 2001 From: "Aneesh Kumar K.V" Date: Fri, 15 Jan 2010 01:27:59 -0500 Subject: ext4: Drop EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE flag We should update reserve space if it is delalloc buffer and that is indicated by EXT4_GET_BLOCKS_DELALLOC_RESERVE flag. So use EXT4_GET_BLOCKS_DELALLOC_RESERVE in place of EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE Signed-off-by: Aneesh Kumar K.V --- fs/ext4/ext4.h | 7 ++----- fs/ext4/extents.c | 4 ++-- fs/ext4/inode.c | 8 ++++---- 3 files changed, 8 insertions(+), 11 deletions(-) (limited to 'fs') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index b98de17e542a..874d169a193e 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -361,14 +361,11 @@ struct ext4_new_group_data { so set the magic i_delalloc_reserve_flag after taking the inode allocation semaphore for */ #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004 - /* Call ext4_da_update_reserve_space() after successfully - allocating the blocks */ -#define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008 /* caller is from the direct IO path, request to creation of an unitialized extents if not allocated, split the uninitialized extent if blocks has been preallocated already*/ -#define EXT4_GET_BLOCKS_DIO 0x0010 -#define EXT4_GET_BLOCKS_CONVERT 0x0020 +#define EXT4_GET_BLOCKS_DIO 0x0008 +#define EXT4_GET_BLOCKS_CONVERT 0x0010 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\ EXT4_GET_BLOCKS_CREATE_UNINIT_EXT) /* Convert extent to initialized after direct IO complete */ diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 3b6ff72026f0..765a4826b118 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -3142,7 +3142,7 @@ out: * But fallocate would have already updated quota and block * count for this offset. So cancel these reservation */ - if (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE) + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) ext4_da_update_reserve_space(inode, allocated, 0); map_out: @@ -3388,7 +3388,7 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode, * Update reserved blocks/metadata blocks after successful * block allocation which had been deferred till now. */ - if (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE) + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) ext4_da_update_reserve_space(inode, allocated, 1); /* diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index c955f6490b78..e11952404e02 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -1316,7 +1316,7 @@ int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block, * reserve space here. */ if ((retval > 0) && - (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE)) + (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)) ext4_da_update_reserve_space(inode, retval, 1); } if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) @@ -2219,10 +2219,10 @@ static int mpage_da_map_blocks(struct mpage_da_data *mpd) * variables are updated after the blocks have been allocated. */ new.b_state = 0; - get_blocks_flags = (EXT4_GET_BLOCKS_CREATE | - EXT4_GET_BLOCKS_DELALLOC_RESERVE); + get_blocks_flags = EXT4_GET_BLOCKS_CREATE; if (mpd->b_state & (1 << BH_Delay)) - get_blocks_flags |= EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE; + get_blocks_flags |= EXT4_GET_BLOCKS_DELALLOC_RESERVE; + blks = ext4_get_blocks(handle, mpd->inode, next, max_blocks, &new, get_blocks_flags); if (blks < 0) { -- cgit v1.2.3 From cb289d6244a37cf932c571d6deb0daa8030f931b Mon Sep 17 00:00:00 2001 From: Davide Libenzi Date: Wed, 13 Jan 2010 09:34:36 -0800 Subject: eventfd - allow atomic read and waitqueue remove KVM needs a wait to atomically remove themselves from the eventfd ->poll() wait queue head, in order to handle correctly their IRQfd deassign operation. This patch introduces such API, plus a way to read an eventfd from its context. Signed-off-by: Davide Libenzi Signed-off-by: Avi Kivity --- fs/eventfd.c | 89 ++++++++++++++++++++++++++++++++++++++++--------- include/linux/eventfd.h | 16 +++++++++ 2 files changed, 90 insertions(+), 15 deletions(-) (limited to 'fs') diff --git a/fs/eventfd.c b/fs/eventfd.c index d26402ff06ea..7758cc382ef0 100644 --- a/fs/eventfd.c +++ b/fs/eventfd.c @@ -135,26 +135,71 @@ static unsigned int eventfd_poll(struct file *file, poll_table *wait) return events; } -static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count, - loff_t *ppos) +static void eventfd_ctx_do_read(struct eventfd_ctx *ctx, __u64 *cnt) +{ + *cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count; + ctx->count -= *cnt; +} + +/** + * eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue. + * @ctx: [in] Pointer to eventfd context. + * @wait: [in] Wait queue to be removed. + * @cnt: [out] Pointer to the 64bit conter value. + * + * Returns zero if successful, or the following error codes: + * + * -EAGAIN : The operation would have blocked. + * + * This is used to atomically remove a wait queue entry from the eventfd wait + * queue head, and read/reset the counter value. + */ +int eventfd_ctx_remove_wait_queue(struct eventfd_ctx *ctx, wait_queue_t *wait, + __u64 *cnt) +{ + unsigned long flags; + + spin_lock_irqsave(&ctx->wqh.lock, flags); + eventfd_ctx_do_read(ctx, cnt); + __remove_wait_queue(&ctx->wqh, wait); + if (*cnt != 0 && waitqueue_active(&ctx->wqh)) + wake_up_locked_poll(&ctx->wqh, POLLOUT); + spin_unlock_irqrestore(&ctx->wqh.lock, flags); + + return *cnt != 0 ? 0 : -EAGAIN; +} +EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue); + +/** + * eventfd_ctx_read - Reads the eventfd counter or wait if it is zero. + * @ctx: [in] Pointer to eventfd context. + * @no_wait: [in] Different from zero if the operation should not block. + * @cnt: [out] Pointer to the 64bit conter value. + * + * Returns zero if successful, or the following error codes: + * + * -EAGAIN : The operation would have blocked but @no_wait was nonzero. + * -ERESTARTSYS : A signal interrupted the wait operation. + * + * If @no_wait is zero, the function might sleep until the eventfd internal + * counter becomes greater than zero. + */ +ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, __u64 *cnt) { - struct eventfd_ctx *ctx = file->private_data; ssize_t res; - __u64 ucnt = 0; DECLARE_WAITQUEUE(wait, current); - if (count < sizeof(ucnt)) - return -EINVAL; spin_lock_irq(&ctx->wqh.lock); + *cnt = 0; res = -EAGAIN; if (ctx->count > 0) - res = sizeof(ucnt); - else if (!(file->f_flags & O_NONBLOCK)) { + res = 0; + else if (!no_wait) { __add_wait_queue(&ctx->wqh, &wait); - for (res = 0;;) { + for (;;) { set_current_state(TASK_INTERRUPTIBLE); if (ctx->count > 0) { - res = sizeof(ucnt); + res = 0; break; } if (signal_pending(current)) { @@ -168,18 +213,32 @@ static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count, __remove_wait_queue(&ctx->wqh, &wait); __set_current_state(TASK_RUNNING); } - if (likely(res > 0)) { - ucnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count; - ctx->count -= ucnt; + if (likely(res == 0)) { + eventfd_ctx_do_read(ctx, cnt); if (waitqueue_active(&ctx->wqh)) wake_up_locked_poll(&ctx->wqh, POLLOUT); } spin_unlock_irq(&ctx->wqh.lock); - if (res > 0 && put_user(ucnt, (__u64 __user *) buf)) - return -EFAULT; return res; } +EXPORT_SYMBOL_GPL(eventfd_ctx_read); + +static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count, + loff_t *ppos) +{ + struct eventfd_ctx *ctx = file->private_data; + ssize_t res; + __u64 cnt; + + if (count < sizeof(cnt)) + return -EINVAL; + res = eventfd_ctx_read(ctx, file->f_flags & O_NONBLOCK, &cnt); + if (res < 0) + return res; + + return put_user(cnt, (__u64 __user *) buf) ? -EFAULT : sizeof(cnt); +} static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t count, loff_t *ppos) diff --git a/include/linux/eventfd.h b/include/linux/eventfd.h index 94dd10366a78..91bb4f27238c 100644 --- a/include/linux/eventfd.h +++ b/include/linux/eventfd.h @@ -10,6 +10,7 @@ #include #include +#include /* * CAREFUL: Check include/asm-generic/fcntl.h when defining @@ -34,6 +35,9 @@ struct file *eventfd_fget(int fd); struct eventfd_ctx *eventfd_ctx_fdget(int fd); struct eventfd_ctx *eventfd_ctx_fileget(struct file *file); int eventfd_signal(struct eventfd_ctx *ctx, int n); +ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, __u64 *cnt); +int eventfd_ctx_remove_wait_queue(struct eventfd_ctx *ctx, wait_queue_t *wait, + __u64 *cnt); #else /* CONFIG_EVENTFD */ @@ -61,6 +65,18 @@ static inline void eventfd_ctx_put(struct eventfd_ctx *ctx) } +static inline ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, + __u64 *cnt) +{ + return -ENOSYS; +} + +static inline int eventfd_ctx_remove_wait_queue(struct eventfd_ctx *ctx, + wait_queue_t *wait, __u64 *cnt) +{ + return -ENOSYS; +} + #endif #endif /* _LINUX_EVENTFD_H */ -- cgit v1.2.3 From 1097df3ffe855eb1476496fa5394816fb197af05 Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Wed, 20 Jan 2010 11:31:11 +0800 Subject: ocfs2: Sync max_inline_data_with_xattr from tools. In ocfs2-tools, we have added ocfs2_max_inline_data_with_xattr, so add it in the kernel's ocfs2_fs.h. Signed-off-by: Tao Ma Signed-off-by: Joel Becker --- fs/ocfs2/ocfs2_fs.h | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h index 1a1a679e51b5..7638a38c32bc 100644 --- a/fs/ocfs2/ocfs2_fs.h +++ b/fs/ocfs2/ocfs2_fs.h @@ -1417,9 +1417,16 @@ static inline int ocfs2_fast_symlink_chars(int blocksize) return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink); } -static inline int ocfs2_max_inline_data(int blocksize) +static inline int ocfs2_max_inline_data_with_xattr(int blocksize, + struct ocfs2_dinode *di) { - return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data); + if (di && (di->i_dyn_features & OCFS2_INLINE_XATTR_FL)) + return blocksize - + offsetof(struct ocfs2_dinode, id2.i_data.id_data) - + di->i_xattr_inline_size; + else + return blocksize - + offsetof(struct ocfs2_dinode, id2.i_data.id_data); } static inline int ocfs2_extent_recs_per_inode(int blocksize) -- cgit v1.2.3 From e5f2cb2b1ad05473fffe6970618997b906f23873 Mon Sep 17 00:00:00 2001 From: Wengang Wang Date: Fri, 22 Jan 2010 21:58:04 +0800 Subject: ocfs2: fix a misleading variable name a local variable "dlm_version" is used as a fs locking version. rename it fs_version. Signed-off-by: Wengang Wang Signed-off-by: Joel Becker --- fs/ocfs2/stack_o2cb.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/stack_o2cb.c b/fs/ocfs2/stack_o2cb.c index e49c41050264..3038c92af493 100644 --- a/fs/ocfs2/stack_o2cb.c +++ b/fs/ocfs2/stack_o2cb.c @@ -277,7 +277,7 @@ static int o2cb_cluster_connect(struct ocfs2_cluster_connection *conn) u32 dlm_key; struct dlm_ctxt *dlm; struct o2dlm_private *priv; - struct dlm_protocol_version dlm_version; + struct dlm_protocol_version fs_version; BUG_ON(conn == NULL); BUG_ON(o2cb_stack.sp_proto == NULL); @@ -304,18 +304,18 @@ static int o2cb_cluster_connect(struct ocfs2_cluster_connection *conn) /* used by the dlm code to make message headers unique, each * node in this domain must agree on this. */ dlm_key = crc32_le(0, conn->cc_name, conn->cc_namelen); - dlm_version.pv_major = conn->cc_version.pv_major; - dlm_version.pv_minor = conn->cc_version.pv_minor; + fs_version.pv_major = conn->cc_version.pv_major; + fs_version.pv_minor = conn->cc_version.pv_minor; - dlm = dlm_register_domain(conn->cc_name, dlm_key, &dlm_version); + dlm = dlm_register_domain(conn->cc_name, dlm_key, &fs_version); if (IS_ERR(dlm)) { rc = PTR_ERR(dlm); mlog_errno(rc); goto out_free; } - conn->cc_version.pv_major = dlm_version.pv_major; - conn->cc_version.pv_minor = dlm_version.pv_minor; + conn->cc_version.pv_major = fs_version.pv_major; + conn->cc_version.pv_minor = fs_version.pv_minor; conn->cc_lockspace = dlm; dlm_register_eviction_cb(dlm, &priv->op_eviction_cb); -- cgit v1.2.3 From 2bd632165c1f783888bd4cbed95f2f304829159b Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 25 Jan 2010 16:57:38 -0800 Subject: ocfs2/trivial: Remove trailing whitespaces Patch removes trailing whitespaces. Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/aops.c | 4 ++-- fs/ocfs2/buffer_head_io.c | 2 +- fs/ocfs2/cluster/heartbeat.c | 6 +++--- fs/ocfs2/cluster/tcp.c | 4 ++-- fs/ocfs2/cluster/tcp_internal.h | 4 ++-- fs/ocfs2/dlm/dlmapi.h | 2 +- fs/ocfs2/dlm/dlmast.c | 2 +- fs/ocfs2/dlm/dlmconvert.c | 2 +- fs/ocfs2/dlm/dlmdomain.c | 2 +- fs/ocfs2/dlm/dlmlock.c | 2 +- fs/ocfs2/dlm/dlmmaster.c | 38 +++++++++++++++++++------------------- fs/ocfs2/dlm/dlmrecovery.c | 38 +++++++++++++++++++------------------- fs/ocfs2/dlm/dlmunlock.c | 8 ++++---- fs/ocfs2/dlmglue.c | 2 +- fs/ocfs2/export.c | 2 +- fs/ocfs2/file.c | 14 +++++++------- fs/ocfs2/inode.c | 4 ++-- fs/ocfs2/journal.c | 2 +- fs/ocfs2/super.c | 2 +- fs/ocfs2/uptodate.c | 4 ++-- 20 files changed, 72 insertions(+), 72 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index 3dae4a13f6e4..7e9df11260f4 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -599,7 +599,7 @@ bail: return ret; } -/* +/* * ocfs2_dio_end_io is called by the dio core when a dio is finished. We're * particularly interested in the aio/dio case. Like the core uses * i_alloc_sem, we use the rw_lock DLM lock to protect io on one node from @@ -670,7 +670,7 @@ static ssize_t ocfs2_direct_IO(int rw, ret = blockdev_direct_IO_no_locking(rw, iocb, inode, inode->i_sb->s_bdev, iov, offset, - nr_segs, + nr_segs, ocfs2_direct_IO_get_blocks, ocfs2_dio_end_io); diff --git a/fs/ocfs2/buffer_head_io.c b/fs/ocfs2/buffer_head_io.c index d43d34a1dd31..21c808f752d8 100644 --- a/fs/ocfs2/buffer_head_io.c +++ b/fs/ocfs2/buffer_head_io.c @@ -368,7 +368,7 @@ int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr, } ocfs2_metadata_cache_io_unlock(ci); - mlog(ML_BH_IO, "block=(%llu), nr=(%d), cached=%s, flags=0x%x\n", + mlog(ML_BH_IO, "block=(%llu), nr=(%d), cached=%s, flags=0x%x\n", (unsigned long long)block, nr, ((flags & OCFS2_BH_IGNORE_CACHE) || ignore_cache) ? "no" : "yes", flags); diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index eda5b8bcddd5..5c9890006708 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -78,7 +78,7 @@ static struct o2hb_callback *hbcall_from_type(enum o2hb_callback_type type); unsigned int o2hb_dead_threshold = O2HB_DEFAULT_DEAD_THRESHOLD; -/* Only sets a new threshold if there are no active regions. +/* Only sets a new threshold if there are no active regions. * * No locking or otherwise interesting code is required for reading * o2hb_dead_threshold as it can't change once regions are active and @@ -170,7 +170,7 @@ static void o2hb_write_timeout(struct work_struct *work) mlog(ML_ERROR, "Heartbeat write timeout to device %s after %u " "milliseconds\n", reg->hr_dev_name, - jiffies_to_msecs(jiffies - reg->hr_last_timeout_start)); + jiffies_to_msecs(jiffies - reg->hr_last_timeout_start)); o2quo_disk_timeout(); } @@ -624,7 +624,7 @@ static int o2hb_check_slot(struct o2hb_region *reg, "seq %llu last %llu changed %u equal %u\n", slot->ds_node_num, (long long)slot->ds_last_generation, le32_to_cpu(hb_block->hb_cksum), - (unsigned long long)le64_to_cpu(hb_block->hb_seq), + (unsigned long long)le64_to_cpu(hb_block->hb_seq), (unsigned long long)slot->ds_last_time, slot->ds_changed_samples, slot->ds_equal_samples); diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index 334f231a422c..938ba181a3d9 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -930,7 +930,7 @@ static void o2net_sendpage(struct o2net_sock_container *sc, cond_resched(); continue; } - mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT + mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT " failed with %zd\n", size, SC_NODEF_ARGS(sc), ret); o2net_ensure_shutdown(nn, sc, 0); break; @@ -1483,7 +1483,7 @@ static void o2net_idle_timer(unsigned long data) mlog(ML_NOTICE, "here are some times that might help debug the " "situation: (tmr %ld.%ld now %ld.%ld dr %ld.%ld adv " "%ld.%ld:%ld.%ld func (%08x:%u) %ld.%ld:%ld.%ld)\n", - sc->sc_tv_timer.tv_sec, (long) sc->sc_tv_timer.tv_usec, + sc->sc_tv_timer.tv_sec, (long) sc->sc_tv_timer.tv_usec, now.tv_sec, (long) now.tv_usec, sc->sc_tv_data_ready.tv_sec, (long) sc->sc_tv_data_ready.tv_usec, sc->sc_tv_advance_start.tv_sec, diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h index 8d58cfe410b1..96fa7ebc530c 100644 --- a/fs/ocfs2/cluster/tcp_internal.h +++ b/fs/ocfs2/cluster/tcp_internal.h @@ -32,10 +32,10 @@ * on their number */ #define O2NET_QUORUM_DELAY_MS ((o2hb_dead_threshold + 2) * O2HB_REGION_TIMEOUT_MS) -/* +/* * This version number represents quite a lot, unfortunately. It not * only represents the raw network message protocol on the wire but also - * locking semantics of the file system using the protocol. It should + * locking semantics of the file system using the protocol. It should * be somewhere else, I'm sure, but right now it isn't. * * With version 11, we separate out the filesystem locking portion. The diff --git a/fs/ocfs2/dlm/dlmapi.h b/fs/ocfs2/dlm/dlmapi.h index b5786a787fab..3cfa114aa391 100644 --- a/fs/ocfs2/dlm/dlmapi.h +++ b/fs/ocfs2/dlm/dlmapi.h @@ -95,7 +95,7 @@ const char *dlm_errname(enum dlm_status err); mlog(ML_ERROR, "dlm status = %s\n", dlm_errname((st))); \ } while (0) -#define DLM_LKSB_UNUSED1 0x01 +#define DLM_LKSB_UNUSED1 0x01 #define DLM_LKSB_PUT_LVB 0x02 #define DLM_LKSB_GET_LVB 0x04 #define DLM_LKSB_UNUSED2 0x08 diff --git a/fs/ocfs2/dlm/dlmast.c b/fs/ocfs2/dlm/dlmast.c index 01cf8cc3d286..dccc439fa087 100644 --- a/fs/ocfs2/dlm/dlmast.c +++ b/fs/ocfs2/dlm/dlmast.c @@ -123,7 +123,7 @@ static void __dlm_queue_ast(struct dlm_ctxt *dlm, struct dlm_lock *lock) dlm_lock_put(lock); /* free up the reserved bast that we are cancelling. * guaranteed that this will not be the last reserved - * ast because *both* an ast and a bast were reserved + * ast because *both* an ast and a bast were reserved * to get to this point. the res->spinlock will not be * taken here */ dlm_lockres_release_ast(dlm, res); diff --git a/fs/ocfs2/dlm/dlmconvert.c b/fs/ocfs2/dlm/dlmconvert.c index ca96bce50e18..f283bce776b4 100644 --- a/fs/ocfs2/dlm/dlmconvert.c +++ b/fs/ocfs2/dlm/dlmconvert.c @@ -396,7 +396,7 @@ static enum dlm_status dlm_send_remote_convert_request(struct dlm_ctxt *dlm, /* instead of logging the same network error over * and over, sleep here and wait for the heartbeat * to notice the node is dead. times out after 5s. */ - dlm_wait_for_node_death(dlm, res->owner, + dlm_wait_for_node_death(dlm, res->owner, DLM_NODE_DEATH_WAIT_MAX); ret = DLM_RECOVERING; mlog(0, "node %u died so returning DLM_RECOVERING " diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c index 0334000676d3..988c9055fd4e 100644 --- a/fs/ocfs2/dlm/dlmdomain.c +++ b/fs/ocfs2/dlm/dlmdomain.c @@ -816,7 +816,7 @@ static int dlm_query_join_handler(struct o2net_msg *msg, u32 len, void *data, } /* Once the dlm ctxt is marked as leaving then we don't want - * to be put in someone's domain map. + * to be put in someone's domain map. * Also, explicitly disallow joining at certain troublesome * times (ie. during recovery). */ if (dlm && dlm->dlm_state != DLM_CTXT_LEAVING) { diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c index 437698e9465f..733337772671 100644 --- a/fs/ocfs2/dlm/dlmlock.c +++ b/fs/ocfs2/dlm/dlmlock.c @@ -269,7 +269,7 @@ static enum dlm_status dlmlock_remote(struct dlm_ctxt *dlm, } dlm_revert_pending_lock(res, lock); dlm_lock_put(lock); - } else if (dlm_is_recovery_lock(res->lockname.name, + } else if (dlm_is_recovery_lock(res->lockname.name, res->lockname.len)) { /* special case for the $RECOVERY lock. * there will never be an AST delivered to put diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c index 03ccf9a7b1f4..a659606dcb95 100644 --- a/fs/ocfs2/dlm/dlmmaster.c +++ b/fs/ocfs2/dlm/dlmmaster.c @@ -366,7 +366,7 @@ void dlm_hb_event_notify_attached(struct dlm_ctxt *dlm, int idx, int node_up) struct dlm_master_list_entry *mle; assert_spin_locked(&dlm->spinlock); - + list_for_each_entry(mle, &dlm->mle_hb_events, hb_events) { if (node_up) dlm_mle_node_up(dlm, mle, NULL, idx); @@ -833,7 +833,7 @@ lookup: __dlm_insert_mle(dlm, mle); /* still holding the dlm spinlock, check the recovery map - * to see if there are any nodes that still need to be + * to see if there are any nodes that still need to be * considered. these will not appear in the mle nodemap * but they might own this lockres. wait on them. */ bit = find_next_bit(dlm->recovery_map, O2NM_MAX_NODES, 0); @@ -883,7 +883,7 @@ redo_request: msleep(500); } continue; - } + } dlm_kick_recovery_thread(dlm); msleep(1000); @@ -939,8 +939,8 @@ wait: res->lockname.name, blocked); if (++tries > 20) { mlog(ML_ERROR, "%s:%.*s: spinning on " - "dlm_wait_for_lock_mastery, blocked=%d\n", - dlm->name, res->lockname.len, + "dlm_wait_for_lock_mastery, blocked=%d\n", + dlm->name, res->lockname.len, res->lockname.name, blocked); dlm_print_one_lock_resource(res); dlm_print_one_mle(mle); @@ -1029,7 +1029,7 @@ recheck: ret = dlm_restart_lock_mastery(dlm, res, mle, *blocked); b = (mle->type == DLM_MLE_BLOCK); if ((*blocked && !b) || (!*blocked && b)) { - mlog(0, "%s:%.*s: status change: old=%d new=%d\n", + mlog(0, "%s:%.*s: status change: old=%d new=%d\n", dlm->name, res->lockname.len, res->lockname.name, *blocked, b); *blocked = b; @@ -1602,7 +1602,7 @@ send_response: } mlog(0, "%u is the owner of %.*s, cleaning everyone else\n", dlm->node_num, res->lockname.len, res->lockname.name); - ret = dlm_dispatch_assert_master(dlm, res, 0, request->node_idx, + ret = dlm_dispatch_assert_master(dlm, res, 0, request->node_idx, DLM_ASSERT_MASTER_MLE_CLEANUP); if (ret < 0) { mlog(ML_ERROR, "failed to dispatch assert master work\n"); @@ -1701,7 +1701,7 @@ again: if (r & DLM_ASSERT_RESPONSE_REASSERT) { mlog(0, "%.*s: node %u create mles on other " - "nodes and requests a re-assert\n", + "nodes and requests a re-assert\n", namelen, lockname, to); reassert = 1; } @@ -1812,7 +1812,7 @@ int dlm_assert_master_handler(struct o2net_msg *msg, u32 len, void *data, spin_unlock(&dlm->master_lock); spin_unlock(&dlm->spinlock); goto done; - } + } } } spin_unlock(&dlm->master_lock); @@ -1883,7 +1883,7 @@ ok: int extra_ref = 0; int nn = -1; int rr, err = 0; - + spin_lock(&mle->spinlock); if (mle->type == DLM_MLE_BLOCK || mle->type == DLM_MLE_MIGRATION) extra_ref = 1; @@ -1891,7 +1891,7 @@ ok: /* MASTER mle: if any bits set in the response map * then the calling node needs to re-assert to clear * up nodes that this node contacted */ - while ((nn = find_next_bit (mle->response_map, O2NM_MAX_NODES, + while ((nn = find_next_bit (mle->response_map, O2NM_MAX_NODES, nn+1)) < O2NM_MAX_NODES) { if (nn != dlm->node_num && nn != assert->node_idx) master_request = 1; @@ -2002,7 +2002,7 @@ kill: __dlm_print_one_lock_resource(res); spin_unlock(&res->spinlock); spin_unlock(&dlm->spinlock); - *ret_data = (void *)res; + *ret_data = (void *)res; dlm_put(dlm); return -EINVAL; } @@ -2040,10 +2040,10 @@ int dlm_dispatch_assert_master(struct dlm_ctxt *dlm, item->u.am.request_from = request_from; item->u.am.flags = flags; - if (ignore_higher) - mlog(0, "IGNORE HIGHER: %.*s\n", res->lockname.len, + if (ignore_higher) + mlog(0, "IGNORE HIGHER: %.*s\n", res->lockname.len, res->lockname.name); - + spin_lock(&dlm->work_lock); list_add_tail(&item->list, &dlm->work_list); spin_unlock(&dlm->work_lock); @@ -2133,7 +2133,7 @@ put: * think that $RECOVERY is currently mastered by a dead node. If so, * we wait a short time to allow that node to get notified by its own * heartbeat stack, then check again. All $RECOVERY lock resources - * mastered by dead nodes are purged when the hearbeat callback is + * mastered by dead nodes are purged when the hearbeat callback is * fired, so we can know for sure that it is safe to continue once * the node returns a live node or no node. */ static int dlm_pre_master_reco_lockres(struct dlm_ctxt *dlm, @@ -2174,7 +2174,7 @@ static int dlm_pre_master_reco_lockres(struct dlm_ctxt *dlm, ret = -EAGAIN; } spin_unlock(&dlm->spinlock); - mlog(0, "%s: reco lock master is %u\n", dlm->name, + mlog(0, "%s: reco lock master is %u\n", dlm->name, master); break; } @@ -2602,7 +2602,7 @@ fail: mlog(0, "%s:%.*s: timed out during migration\n", dlm->name, res->lockname.len, res->lockname.name); - /* avoid hang during shutdown when migrating lockres + /* avoid hang during shutdown when migrating lockres * to a node which also goes down */ if (dlm_is_node_dead(dlm, target)) { mlog(0, "%s:%.*s: expected migration " @@ -2738,7 +2738,7 @@ static int dlm_migration_can_proceed(struct dlm_ctxt *dlm, can_proceed = !!(res->state & DLM_LOCK_RES_MIGRATING); spin_unlock(&res->spinlock); - /* target has died, so make the caller break out of the + /* target has died, so make the caller break out of the * wait_event, but caller must recheck the domain_map */ spin_lock(&dlm->spinlock); if (!test_bit(mig_target, dlm->domain_map)) diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index 2f9e4e19a4f2..57736d3ea7b5 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -1050,7 +1050,7 @@ static void dlm_move_reco_locks_to_list(struct dlm_ctxt *dlm, if (lock->ml.node == dead_node) { mlog(0, "AHA! there was " "a $RECOVERY lock for dead " - "node %u (%s)!\n", + "node %u (%s)!\n", dead_node, dlm->name); list_del_init(&lock->list); dlm_lock_put(lock); @@ -1839,7 +1839,7 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm, * the lvb. */ memcpy(res->lvb, mres->lvb, DLM_LVB_LEN); } else { - /* otherwise, the node is sending its + /* otherwise, the node is sending its * most recent valid lvb info */ BUG_ON(ml->type != LKM_EXMODE && ml->type != LKM_PRMODE); @@ -2114,7 +2114,7 @@ static void dlm_revalidate_lvb(struct dlm_ctxt *dlm, assert_spin_locked(&res->spinlock); if (res->owner == dlm->node_num) - /* if this node owned the lockres, and if the dead node + /* if this node owned the lockres, and if the dead node * had an EX when he died, blank out the lvb */ search_node = dead_node; else { @@ -2152,7 +2152,7 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm, /* this node is the lockres master: * 1) remove any stale locks for the dead node - * 2) if the dead node had an EX when he died, blank out the lvb + * 2) if the dead node had an EX when he died, blank out the lvb */ assert_spin_locked(&dlm->spinlock); assert_spin_locked(&res->spinlock); @@ -2260,7 +2260,7 @@ static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node) } spin_unlock(&res->spinlock); continue; - } + } spin_lock(&res->spinlock); /* zero the lvb if necessary */ dlm_revalidate_lvb(dlm, res, dead_node); @@ -2411,7 +2411,7 @@ static void dlm_reco_unlock_ast(void *astdata, enum dlm_status st) * this function on each node racing to become the recovery * master will not stop attempting this until either: * a) this node gets the EX (and becomes the recovery master), - * or b) dlm->reco.new_master gets set to some nodenum + * or b) dlm->reco.new_master gets set to some nodenum * != O2NM_INVALID_NODE_NUM (another node will do the reco). * so each time a recovery master is needed, the entire cluster * will sync at this point. if the new master dies, that will @@ -2424,7 +2424,7 @@ static int dlm_pick_recovery_master(struct dlm_ctxt *dlm) mlog(0, "starting recovery of %s at %lu, dead=%u, this=%u\n", dlm->name, jiffies, dlm->reco.dead_node, dlm->node_num); -again: +again: memset(&lksb, 0, sizeof(lksb)); ret = dlmlock(dlm, LKM_EXMODE, &lksb, LKM_NOQUEUE|LKM_RECOVERY, @@ -2437,8 +2437,8 @@ again: if (ret == DLM_NORMAL) { mlog(0, "dlm=%s dlmlock says I got it (this=%u)\n", dlm->name, dlm->node_num); - - /* got the EX lock. check to see if another node + + /* got the EX lock. check to see if another node * just became the reco master */ if (dlm_reco_master_ready(dlm)) { mlog(0, "%s: got reco EX lock, but %u will " @@ -2451,12 +2451,12 @@ again: /* see if recovery was already finished elsewhere */ spin_lock(&dlm->spinlock); if (dlm->reco.dead_node == O2NM_INVALID_NODE_NUM) { - status = -EINVAL; + status = -EINVAL; mlog(0, "%s: got reco EX lock, but " "node got recovered already\n", dlm->name); if (dlm->reco.new_master != O2NM_INVALID_NODE_NUM) { mlog(ML_ERROR, "%s: new master is %u " - "but no dead node!\n", + "but no dead node!\n", dlm->name, dlm->reco.new_master); BUG(); } @@ -2468,7 +2468,7 @@ again: * set the master and send the messages to begin recovery */ if (!status) { mlog(0, "%s: dead=%u, this=%u, sending " - "begin_reco now\n", dlm->name, + "begin_reco now\n", dlm->name, dlm->reco.dead_node, dlm->node_num); status = dlm_send_begin_reco_message(dlm, dlm->reco.dead_node); @@ -2501,7 +2501,7 @@ again: mlog(0, "dlm=%s dlmlock says another node got it (this=%u)\n", dlm->name, dlm->node_num); /* another node is master. wait on - * reco.new_master != O2NM_INVALID_NODE_NUM + * reco.new_master != O2NM_INVALID_NODE_NUM * for at most one second */ wait_event_timeout(dlm->dlm_reco_thread_wq, dlm_reco_master_ready(dlm), @@ -2599,7 +2599,7 @@ retry: } if (ret < 0) { struct dlm_lock_resource *res; - /* this is now a serious problem, possibly ENOMEM + /* this is now a serious problem, possibly ENOMEM * in the network stack. must retry */ mlog_errno(ret); mlog(ML_ERROR, "begin reco of dlm %s to node %u " @@ -2612,7 +2612,7 @@ retry: } else { mlog(ML_ERROR, "recovery lock not found\n"); } - /* sleep for a bit in hopes that we can avoid + /* sleep for a bit in hopes that we can avoid * another ENOMEM */ msleep(100); goto retry; @@ -2664,7 +2664,7 @@ int dlm_begin_reco_handler(struct o2net_msg *msg, u32 len, void *data, } if (dlm->reco.dead_node != O2NM_INVALID_NODE_NUM) { mlog(ML_NOTICE, "%s: dead_node previously set to %u, " - "node %u changing it to %u\n", dlm->name, + "node %u changing it to %u\n", dlm->name, dlm->reco.dead_node, br->node_idx, br->dead_node); } dlm_set_reco_master(dlm, br->node_idx); @@ -2730,8 +2730,8 @@ stage2: if (ret < 0) { mlog_errno(ret); if (dlm_is_host_down(ret)) { - /* this has no effect on this recovery - * session, so set the status to zero to + /* this has no effect on this recovery + * session, so set the status to zero to * finish out the last recovery */ mlog(ML_ERROR, "node %u went down after this " "node finished recovery.\n", nodenum); @@ -2768,7 +2768,7 @@ int dlm_finalize_reco_handler(struct o2net_msg *msg, u32 len, void *data, mlog(0, "%s: node %u finalizing recovery stage%d of " "node %u (%u:%u)\n", dlm->name, fr->node_idx, stage, fr->dead_node, dlm->reco.dead_node, dlm->reco.new_master); - + spin_lock(&dlm->spinlock); if (dlm->reco.new_master != fr->node_idx) { diff --git a/fs/ocfs2/dlm/dlmunlock.c b/fs/ocfs2/dlm/dlmunlock.c index 00f53b2aea76..49e29ecd0201 100644 --- a/fs/ocfs2/dlm/dlmunlock.c +++ b/fs/ocfs2/dlm/dlmunlock.c @@ -190,8 +190,8 @@ static enum dlm_status dlmunlock_common(struct dlm_ctxt *dlm, actions &= ~(DLM_UNLOCK_REMOVE_LOCK| DLM_UNLOCK_REGRANT_LOCK| DLM_UNLOCK_CLEAR_CONVERT_TYPE); - } else if (status == DLM_RECOVERING || - status == DLM_MIGRATING || + } else if (status == DLM_RECOVERING || + status == DLM_MIGRATING || status == DLM_FORWARD) { /* must clear the actions because this unlock * is about to be retried. cannot free or do @@ -661,14 +661,14 @@ retry: if (call_ast) { mlog(0, "calling unlockast(%p, %d)\n", data, status); if (is_master) { - /* it is possible that there is one last bast + /* it is possible that there is one last bast * pending. make sure it is flushed, then * call the unlockast. * not an issue if this is a mastered remotely, * since this lock has been removed from the * lockres queues and cannot be found. */ dlm_kick_thread(dlm, NULL); - wait_event(dlm->ast_wq, + wait_event(dlm->ast_wq, dlm_lock_basts_flushed(dlm, lock)); } (*unlockast)(data, status); diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index c5e4a49e3a12..172f4c6ce1be 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -3155,7 +3155,7 @@ out: /* Mark the lockres as being dropped. It will no longer be * queued if blocking, but we still may have to wait on it * being dequeued from the downconvert thread before we can consider - * it safe to drop. + * it safe to drop. * * You can *not* attempt to call cluster_lock on this lockres anymore. */ void ocfs2_mark_lockres_freeing(struct ocfs2_lock_res *lockres) diff --git a/fs/ocfs2/export.c b/fs/ocfs2/export.c index 15713cbb865c..19ad145d2af3 100644 --- a/fs/ocfs2/export.c +++ b/fs/ocfs2/export.c @@ -239,7 +239,7 @@ static int ocfs2_encode_fh(struct dentry *dentry, u32 *fh_in, int *max_len, mlog(0, "Encoding parent: blkno: %llu, generation: %u\n", (unsigned long long)blkno, generation); } - + *max_len = len; bail: diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 06ccf6a86d35..65e9375d2fb3 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -749,7 +749,7 @@ static int ocfs2_write_zero_page(struct inode *inode, int ret; offset = (size & (PAGE_CACHE_SIZE-1)); /* Within page */ - /* ugh. in prepare/commit_write, if from==to==start of block, we + /* ugh. in prepare/commit_write, if from==to==start of block, we ** skip the prepare. make sure we never send an offset for the start ** of a block */ @@ -1779,7 +1779,7 @@ static int ocfs2_prepare_inode_for_write(struct dentry *dentry, struct inode *inode = dentry->d_inode; loff_t saved_pos, end; - /* + /* * We start with a read level meta lock and only jump to an ex * if we need to make modifications here. */ @@ -2033,7 +2033,7 @@ out_dio: pos + count - 1); } - /* + /* * deep in g_f_a_w_n()->ocfs2_direct_IO we pass in a ocfs2_dio_end_io * function pointer which is called when o_direct io completes so that * it can unlock our rw lock. (it's the clustered equivalent of @@ -2198,7 +2198,7 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb, goto bail; } - /* + /* * buffered reads protect themselves in ->readpage(). O_DIRECT reads * need locks to protect pending reads from racing with truncate. */ @@ -2220,10 +2220,10 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb, * We're fine letting folks race truncates and extending * writes with read across the cluster, just like they can * locally. Hence no rw_lock during read. - * + * * Take and drop the meta data lock to update inode fields * like i_size. This allows the checks down below - * generic_file_aio_read() a chance of actually working. + * generic_file_aio_read() a chance of actually working. */ ret = ocfs2_inode_lock_atime(inode, filp->f_vfsmnt, &lock_level); if (ret < 0) { @@ -2248,7 +2248,7 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb, bail: if (have_alloc_sem) up_read(&inode->i_alloc_sem); - if (rw_level != -1) + if (rw_level != -1) ocfs2_rw_unlock(inode, rw_level); mlog_exit(ret); diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c index 0297fb8982b8..88459bdd1ff3 100644 --- a/fs/ocfs2/inode.c +++ b/fs/ocfs2/inode.c @@ -475,7 +475,7 @@ static int ocfs2_read_locked_inode(struct inode *inode, if (args->fi_flags & OCFS2_FI_FLAG_ORPHAN_RECOVERY) { status = ocfs2_try_open_lock(inode, 0); if (status) { - make_bad_inode(inode); + make_bad_inode(inode); return status; } } @@ -684,7 +684,7 @@ bail: return status; } -/* +/* * Serialize with orphan dir recovery. If the process doing * recovery on this orphan dir does an iget() with the dir * i_mutex held, we'll deadlock here. Instead we detect this diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c index bf34c491ae96..9336c60e3a36 100644 --- a/fs/ocfs2/journal.c +++ b/fs/ocfs2/journal.c @@ -2034,7 +2034,7 @@ static int ocfs2_queue_orphans(struct ocfs2_super *osb, status = -ENOENT; mlog_errno(status); return status; - } + } mutex_lock(&orphan_dir_inode->i_mutex); status = ocfs2_inode_lock(orphan_dir_inode, NULL, 0); diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index 26069917a9f5..755cd49a5ef3 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -1062,7 +1062,7 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent) "file system, but write access is " "unavailable.\n"); else - mlog_errno(status); + mlog_errno(status); goto read_super_error; } diff --git a/fs/ocfs2/uptodate.c b/fs/ocfs2/uptodate.c index c61369342a27..a0a120e82b97 100644 --- a/fs/ocfs2/uptodate.c +++ b/fs/ocfs2/uptodate.c @@ -267,8 +267,8 @@ static int ocfs2_buffer_cached(struct ocfs2_caching_info *ci, } /* Warning: even if it returns true, this does *not* guarantee that - * the block is stored in our inode metadata cache. - * + * the block is stored in our inode metadata cache. + * * This can be called under lock_buffer() */ int ocfs2_buffer_uptodate(struct ocfs2_caching_info *ci, -- cgit v1.2.3 From 71656fa6ec10473eb9b646c10a2173fdea2f83c9 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 25 Jan 2010 16:57:39 -0800 Subject: ocfs2/dlm: Ignore LVBs of locks in the Blocked list During lock resource migration, o2dlm fills the packet with a LVB from the first valid lock. For sanity, it ensures that the other valid locks have the same LVB. If not, it BUGs. The valid locks are ones that have granted EX or PR lock levels and are either on the Granted or Converting lists. Locks in the Blocked list cannot have a valid LVB. This patch ensures that we skip the locks in the Blocked list. Fixes oss bugzilla#1202 http://oss.oracle.com/bugzilla/show_bug.cgi?id=1202 Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlm/dlmrecovery.c | 48 ++++++++++++++++++++++++++++++++-------------- 1 file changed, 34 insertions(+), 14 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index 57736d3ea7b5..9d67894cda6d 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -1164,6 +1164,39 @@ static void dlm_init_migratable_lockres(struct dlm_migratable_lockres *mres, mres->master = master; } +static void dlm_prepare_lvb_for_migration(struct dlm_lock *lock, + struct dlm_migratable_lockres *mres, + int queue) +{ + if (!lock->lksb) + return; + + /* Ignore lvb in all locks in the blocked list */ + if (queue == DLM_BLOCKED_LIST) + return; + + /* Only consider lvbs in locks with granted EX or PR lock levels */ + if (lock->ml.type != LKM_EXMODE && lock->ml.type != LKM_PRMODE) + return; + + if (dlm_lvb_is_empty(mres->lvb)) { + memcpy(mres->lvb, lock->lksb->lvb, DLM_LVB_LEN); + return; + } + + /* Ensure the lvb copied for migration matches in other valid locks */ + if (!memcmp(mres->lvb, lock->lksb->lvb, DLM_LVB_LEN)) + return; + + mlog(ML_ERROR, "Mismatched lvb in lock cookie=%u:%llu, name=%.*s, " + "node=%u\n", + dlm_get_lock_cookie_node(be64_to_cpu(lock->ml.cookie)), + dlm_get_lock_cookie_seq(be64_to_cpu(lock->ml.cookie)), + lock->lockres->lockname.len, lock->lockres->lockname.name, + lock->ml.node); + dlm_print_one_lock_resource(lock->lockres); + BUG(); +} /* returns 1 if this lock fills the network structure, * 0 otherwise */ @@ -1181,20 +1214,7 @@ static int dlm_add_lock_to_array(struct dlm_lock *lock, ml->list = queue; if (lock->lksb) { ml->flags = lock->lksb->flags; - /* send our current lvb */ - if (ml->type == LKM_EXMODE || - ml->type == LKM_PRMODE) { - /* if it is already set, this had better be a PR - * and it has to match */ - if (!dlm_lvb_is_empty(mres->lvb) && - (ml->type == LKM_EXMODE || - memcmp(mres->lvb, lock->lksb->lvb, DLM_LVB_LEN))) { - mlog(ML_ERROR, "mismatched lvbs!\n"); - dlm_print_one_lock_resource(lock->lockres); - BUG(); - } - memcpy(mres->lvb, lock->lksb->lvb, DLM_LVB_LEN); - } + dlm_prepare_lvb_for_migration(lock, mres, queue); } ml->node = lock->ml.node; mres->num_locks++; -- cgit v1.2.3 From 26636bf6b2010aa84c54d577231e017ba71493d0 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 25 Jan 2010 16:57:40 -0800 Subject: ocfs2/dlm: Print more messages during lock migration When a lock resource is migrated, the dlm compares the migrated locks with that that was already existing on the new node. If the comparison fails, it BUGs. This patch prints more messages when the comparison fails inorder to help with the root cause analyis. http://oss.oracle.com/bugzilla/show_bug.cgi?id=1206 This does not fix bz1206. However, if we run into it again, we will have more information to chew on. Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlm/dlmrecovery.c | 46 ++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 38 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index 9d67894cda6d..cfb2ae9ab538 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -1750,6 +1750,7 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm, struct dlm_lock *lock = NULL; u8 from = O2NM_MAX_NODES; unsigned int added = 0; + __be64 c; mlog(0, "running %d locks for this lockres\n", mres->num_locks); for (i=0; inum_locks; i++) { @@ -1797,19 +1798,48 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm, /* lock is always created locally first, and * destroyed locally last. it must be on the list */ if (!lock) { - __be64 c = ml->cookie; - mlog(ML_ERROR, "could not find local lock " - "with cookie %u:%llu!\n", + c = ml->cookie; + mlog(ML_ERROR, "Could not find local lock " + "with cookie %u:%llu, node %u, " + "list %u, flags 0x%x, type %d, " + "conv %d, highest blocked %d\n", dlm_get_lock_cookie_node(be64_to_cpu(c)), - dlm_get_lock_cookie_seq(be64_to_cpu(c))); + dlm_get_lock_cookie_seq(be64_to_cpu(c)), + ml->node, ml->list, ml->flags, ml->type, + ml->convert_type, ml->highest_blocked); + __dlm_print_one_lock_resource(res); + BUG(); + } + + if (lock->ml.node != ml->node) { + c = lock->ml.cookie; + mlog(ML_ERROR, "Mismatched node# in lock " + "cookie %u:%llu, name %.*s, node %u\n", + dlm_get_lock_cookie_node(be64_to_cpu(c)), + dlm_get_lock_cookie_seq(be64_to_cpu(c)), + res->lockname.len, res->lockname.name, + lock->ml.node); + c = ml->cookie; + mlog(ML_ERROR, "Migrate lock cookie %u:%llu, " + "node %u, list %u, flags 0x%x, type %d, " + "conv %d, highest blocked %d\n", + dlm_get_lock_cookie_node(be64_to_cpu(c)), + dlm_get_lock_cookie_seq(be64_to_cpu(c)), + ml->node, ml->list, ml->flags, ml->type, + ml->convert_type, ml->highest_blocked); __dlm_print_one_lock_resource(res); BUG(); } - BUG_ON(lock->ml.node != ml->node); if (tmpq != queue) { - mlog(0, "lock was on %u instead of %u for %.*s\n", - j, ml->list, res->lockname.len, res->lockname.name); + c = ml->cookie; + mlog(0, "Lock cookie %u:%llu was on list %u " + "instead of list %u for %.*s\n", + dlm_get_lock_cookie_node(be64_to_cpu(c)), + dlm_get_lock_cookie_seq(be64_to_cpu(c)), + j, ml->list, res->lockname.len, + res->lockname.name); + __dlm_print_one_lock_resource(res); spin_unlock(&res->spinlock); continue; } @@ -1906,7 +1936,7 @@ skip_lvb: spin_lock(&res->spinlock); list_for_each_entry(lock, queue, list) { if (lock->ml.cookie == ml->cookie) { - __be64 c = lock->ml.cookie; + c = lock->ml.cookie; mlog(ML_ERROR, "%s:%.*s: %u:%llu: lock already " "exists on this lockres!\n", dlm->name, res->lockname.len, res->lockname.name, -- cgit v1.2.3 From c9edda7140ec6a22accf7f2f86da362dfbfd41fc Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 26 Jan 2010 15:41:34 -0500 Subject: NFS: Fix a reference leak in nfs_wb_cancel_page() Signed-off-by: Trond Myklebust Cc: stable@kernel.org Reviewed-by: Chuck Lever --- fs/nfs/write.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/nfs/write.c b/fs/nfs/write.c index d171696017f4..dac8d7676aff 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1541,6 +1541,7 @@ int nfs_wb_page_cancel(struct inode *inode, struct page *page) break; } ret = nfs_wait_on_request(req); + nfs_release_request(req); if (ret < 0) goto out; } -- cgit v1.2.3 From 82be934a59ff891cac598727e5a862ba2b9d1fac Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 26 Jan 2010 15:41:53 -0500 Subject: NFS: Try to commit unstable writes in nfs_release_page() If someone calls nfs_release_page(), we presumably already know that the page is clean, however it may be holding an unstable write. Signed-off-by: Trond Myklebust Cc: stable@kernel.org Reviewed-by: Chuck Lever --- fs/nfs/file.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/nfs/file.c b/fs/nfs/file.c index 6b891328f332..63f2071d6445 100644 --- a/fs/nfs/file.c +++ b/fs/nfs/file.c @@ -486,6 +486,8 @@ static int nfs_release_page(struct page *page, gfp_t gfp) { dfprintk(PAGECACHE, "NFS: release_page(%p)\n", page); + if (gfp & __GFP_WAIT) + nfs_wb_page(page->mapping->host, page); /* If PagePrivate() is set, then the page is not freeable */ if (PagePrivate(page)) return 0; -- cgit v1.2.3 From 0aa05887af728b058af91197f0ae9b3ae63dd74a Mon Sep 17 00:00:00 2001 From: H Hartley Sweeten Date: Tue, 26 Jan 2010 15:42:03 -0500 Subject: NFS: Make nfs_commitdata_release static The symbol nfs_commitdata_release is only used locally in this file. Make it static to prevent the following sparse warning: warning: symbol 'nfs_commitdata_release' was not declared. Should it be static? Signed-off-by: H Hartley Sweeten Cc: Trond Myklebust Signed-off-by: Trond Myklebust Reviewed-by: Chuck Lever --- fs/nfs/write.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/write.c b/fs/nfs/write.c index dac8d7676aff..7b54b8bb101f 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1233,7 +1233,7 @@ int nfs_writeback_done(struct rpc_task *task, struct nfs_write_data *data) #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4) -void nfs_commitdata_release(void *data) +static void nfs_commitdata_release(void *data) { struct nfs_write_data *wdata = data; -- cgit v1.2.3 From b0706ca415b188ed58788420de4d5c9972b2afb2 Mon Sep 17 00:00:00 2001 From: David Howells Date: Tue, 26 Jan 2010 15:42:11 -0500 Subject: NFS: Avoid warnings when CONFIG_NFS_V4=n Avoid the following warnings when CONFIG_NFS_V4=n: fs/nfs/sysctl.c:19: warning: unused variable `nfs_set_port_max' fs/nfs/sysctl.c:18: warning: unused variable `nfs_set_port_min' by making those variables contingent on NFSv4 being configured. Signed-off-by: David Howells Signed-off-by: Trond Myklebust Reviewed-by: Chuck Lever --- fs/nfs/sysctl.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/nfs/sysctl.c b/fs/nfs/sysctl.c index 70e1fbbaaeab..ad4d2e787b20 100644 --- a/fs/nfs/sysctl.c +++ b/fs/nfs/sysctl.c @@ -15,8 +15,10 @@ #include "callback.h" +#ifdef CONFIG_NFS_V4 static const int nfs_set_port_min = 0; static const int nfs_set_port_max = 65535; +#endif static struct ctl_table_header *nfs_callback_sysctl_table; static ctl_table nfs_cb_sysctls[] = { -- cgit v1.2.3 From 2bee72a6aa1e6d0a4f5da56217f0d0bbbdd0d9a3 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 26 Jan 2010 15:42:21 -0500 Subject: NFSv4: Ensure that the NFSv4 locking can recover from stateid errors In most cases, we just want to mark the lock_stateid sequence id as being uninitialised. Signed-off-by: Trond Myklebust Cc: stable@kernel.org Reviewed-by: Chuck Lever --- fs/nfs/nfs4proc.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 198d51d17c13..0b68238ed0c8 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -4088,6 +4088,22 @@ static const struct rpc_call_ops nfs4_recover_lock_ops = { .rpc_release = nfs4_lock_release, }; +static void nfs4_handle_setlk_error(struct nfs_server *server, struct nfs4_lock_state *lsp, int new_lock_owner, int error) +{ + struct nfs_client *clp = server->nfs_client; + struct nfs4_state *state = lsp->ls_state; + + switch (error) { + case -NFS4ERR_ADMIN_REVOKED: + case -NFS4ERR_BAD_STATEID: + case -NFS4ERR_EXPIRED: + if (new_lock_owner != 0 || + (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0) + nfs4_state_mark_reclaim_nograce(clp, state); + lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED; + }; +} + static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type) { struct nfs4_lockdata *data; @@ -4126,6 +4142,9 @@ static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *f ret = nfs4_wait_for_completion_rpc_task(task); if (ret == 0) { ret = data->rpc_status; + if (ret) + nfs4_handle_setlk_error(data->server, data->lsp, + data->arg.new_lock_owner, ret); } else data->cancelled = 1; rpc_put_task(task); -- cgit v1.2.3 From 8e469ebd6dc32cbaf620e134d79f740bf0ebab79 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 26 Jan 2010 15:42:30 -0500 Subject: NFSv4: Don't allow posix locking against servers that don't support it Signed-off-by: Trond Myklebust Cc: stable@kernel.org Reviewed-by: Chuck Lever --- fs/nfs/nfs4_fs.h | 1 + fs/nfs/nfs4proc.c | 7 ++++++- 2 files changed, 7 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 865265bdca03..ea2f41b26aea 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -146,6 +146,7 @@ enum { NFS_O_RDWR_STATE, /* OPEN stateid has read/write state */ NFS_STATE_RECLAIM_REBOOT, /* OPEN stateid server rebooted */ NFS_STATE_RECLAIM_NOGRACE, /* OPEN stateid needs to recover state */ + NFS_STATE_POSIX_LOCKS, /* Posix locks are supported */ }; struct nfs4_state { diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 0b68238ed0c8..be044b58e811 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1658,6 +1658,8 @@ static int _nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, in status = PTR_ERR(state); if (IS_ERR(state)) goto err_opendata_put; + if ((opendata->o_res.rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) != 0) + set_bit(NFS_STATE_POSIX_LOCKS, &state->flags); nfs4_opendata_put(opendata); nfs4_put_state_owner(sp); *res = state; @@ -4200,8 +4202,11 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock { struct nfs_inode *nfsi = NFS_I(state->inode); unsigned char fl_flags = request->fl_flags; - int status; + int status = -ENOLCK; + if ((fl_flags & FL_POSIX) && + !test_bit(NFS_STATE_POSIX_LOCKS, &state->flags)) + goto out; /* Is this a delegated open? */ status = nfs4_set_lock_state(state, request); if (status != 0) -- cgit v1.2.3 From 03391693a95900875b0973569d2d73ff3aa8972e Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 26 Jan 2010 15:42:38 -0500 Subject: NFSv4.1: Don't call nfs4_schedule_state_recovery() unnecessarily Currently, nfs4_handle_exception() will call it twice if called with an error of -NFS4ERR_STALE_CLIENTID, -NFS4ERR_STALE_STATEID or -NFS4ERR_EXPIRED. Signed-off-by: Trond Myklebust Reviewed-by: Chuck Lever --- fs/nfs/nfs4proc.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index be044b58e811..afbfe673489b 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -256,12 +256,8 @@ static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, ret = nfs4_wait_clnt_recover(clp); if (ret == 0) exception->retry = 1; -#if !defined(CONFIG_NFS_V4_1) break; -#else /* !defined(CONFIG_NFS_V4_1) */ - if (!nfs4_has_session(server->nfs_client)) - break; - /* FALLTHROUGH */ +#if defined(CONFIG_NFS_V4_1) case -NFS4ERR_BADSESSION: case -NFS4ERR_BADSLOT: case -NFS4ERR_BAD_HIGH_SLOT: @@ -274,7 +270,7 @@ static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, nfs4_schedule_state_recovery(clp); exception->retry = 1; break; -#endif /* !defined(CONFIG_NFS_V4_1) */ +#endif /* defined(CONFIG_NFS_V4_1) */ case -NFS4ERR_FILE_OPEN: if (exception->timeout > HZ) { /* We have retried a decent amount, time to -- cgit v1.2.3 From a2c0b9e291208f65221a0ad8a0c80a377707d480 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 26 Jan 2010 15:42:47 -0500 Subject: NFS: Ensure that we handle NFS4ERR_STALE_STATEID correctly Even if the server is crazy, we should be able to mark the stateid as being bad, to ensure it gets recovered. Signed-off-by: Trond Myklebust Reviewed-by: Chuck Lever --- fs/nfs/nfs4_fs.h | 1 + fs/nfs/nfs4proc.c | 44 +++++++++++++++++++++++++++++++------------- fs/nfs/nfs4state.c | 2 +- 3 files changed, 33 insertions(+), 14 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index ea2f41b26aea..0c6fda33d66e 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -278,6 +278,7 @@ extern void nfs4_state_set_mode_locked(struct nfs4_state *, fmode_t); extern void nfs4_schedule_state_recovery(struct nfs_client *); extern void nfs4_schedule_state_manager(struct nfs_client *); extern int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_state *state); +extern int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_state *state); extern void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags); extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp); extern int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl); diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index afbfe673489b..375f0fae2c6a 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -249,14 +249,14 @@ static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, if (state == NULL) break; nfs4_state_mark_reclaim_nograce(clp, state); - case -NFS4ERR_STALE_CLIENTID: + goto do_state_recovery; case -NFS4ERR_STALE_STATEID: + if (state == NULL) + break; + nfs4_state_mark_reclaim_reboot(clp, state); + case -NFS4ERR_STALE_CLIENTID: case -NFS4ERR_EXPIRED: - nfs4_schedule_state_recovery(clp); - ret = nfs4_wait_clnt_recover(clp); - if (ret == 0) - exception->retry = 1; - break; + goto do_state_recovery; #if defined(CONFIG_NFS_V4_1) case -NFS4ERR_BADSESSION: case -NFS4ERR_BADSLOT: @@ -289,6 +289,12 @@ static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, } /* We failed to handle the error */ return nfs4_map_errors(ret); +do_state_recovery: + nfs4_schedule_state_recovery(clp); + ret = nfs4_wait_clnt_recover(clp); + if (ret == 0) + exception->retry = 1; + return ret; } @@ -3420,15 +3426,14 @@ _nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, if (state == NULL) break; nfs4_state_mark_reclaim_nograce(clp, state); - case -NFS4ERR_STALE_CLIENTID: + goto do_state_recovery; case -NFS4ERR_STALE_STATEID: + if (state == NULL) + break; + nfs4_state_mark_reclaim_reboot(clp, state); + case -NFS4ERR_STALE_CLIENTID: case -NFS4ERR_EXPIRED: - rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL); - nfs4_schedule_state_recovery(clp); - if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0) - rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task); - task->tk_status = 0; - return -EAGAIN; + goto do_state_recovery; #if defined(CONFIG_NFS_V4_1) case -NFS4ERR_BADSESSION: case -NFS4ERR_BADSLOT: @@ -3456,6 +3461,13 @@ _nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, } task->tk_status = nfs4_map_errors(task->tk_status); return 0; +do_state_recovery: + rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL); + nfs4_schedule_state_recovery(clp); + if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0) + rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task); + task->tk_status = 0; + return -EAGAIN; } static int @@ -4099,6 +4111,12 @@ static void nfs4_handle_setlk_error(struct nfs_server *server, struct nfs4_lock_ (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0) nfs4_state_mark_reclaim_nograce(clp, state); lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED; + break; + case -NFS4ERR_STALE_STATEID: + if (new_lock_owner != 0 || + (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0) + nfs4_state_mark_reclaim_reboot(clp, state); + lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED; }; } diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 6d263ed79e92..c1e2733f4fa4 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -901,7 +901,7 @@ void nfs4_schedule_state_recovery(struct nfs_client *clp) nfs4_schedule_state_manager(clp); } -static int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_state *state) +int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_state *state) { set_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags); -- cgit v1.2.3 From b04da8bfdfbbd79544cab2fadfdc12e87eb01600 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Tue, 26 Jan 2010 15:04:02 -0800 Subject: fnctl: f_modown should call write_lock_irqsave/restore Commit 703625118069f9f8960d356676662d3db5a9d116 exposed that f_modown() should call write_lock_irqsave instead of just write_lock_irq so that because a caller could have a spinlock held and it would not be good to renable interrupts. Cc: Eric W. Biederman Cc: Al Viro Cc: Alan Cox Cc: Tavis Ormandy Cc: stable Signed-off-by: Greg Kroah-Hartman Signed-off-by: Linus Torvalds --- fs/fcntl.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/fcntl.c b/fs/fcntl.c index 97e01dc0d95f..5ef953e6f908 100644 --- a/fs/fcntl.c +++ b/fs/fcntl.c @@ -199,7 +199,9 @@ static int setfl(int fd, struct file * filp, unsigned long arg) static void f_modown(struct file *filp, struct pid *pid, enum pid_type type, int force) { - write_lock_irq(&filp->f_owner.lock); + unsigned long flags; + + write_lock_irqsave(&filp->f_owner.lock, flags); if (force || !filp->f_owner.pid) { put_pid(filp->f_owner.pid); filp->f_owner.pid = get_pid(pid); @@ -211,7 +213,7 @@ static void f_modown(struct file *filp, struct pid *pid, enum pid_type type, filp->f_owner.euid = cred->euid; } } - write_unlock_irq(&filp->f_owner.lock); + write_unlock_irqrestore(&filp->f_owner.lock, flags); } int __f_setown(struct file *filp, struct pid *pid, enum pid_type type, -- cgit v1.2.3 From afc70ed05a07bfe171f7a5b8fdc80bdb073d314f Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 23 Jan 2010 23:38:27 -0500 Subject: Fix a leak in affs_fill_super() Signed-off-by: Al Viro --- fs/affs/super.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/affs/super.c b/fs/affs/super.c index 104fdcb3a7fc..b2a5958c6191 100644 --- a/fs/affs/super.c +++ b/fs/affs/super.c @@ -316,6 +316,8 @@ static int affs_fill_super(struct super_block *sb, void *data, int silent) &blocksize,&sbi->s_prefix, sbi->s_volume, &mount_flags)) { printk(KERN_ERR "AFFS: Error parsing options\n"); + kfree(sbi->s_prefix); + kfree(sbi); return -EINVAL; } /* N.B. after this point s_prefix must be released */ -- cgit v1.2.3 From 29333920a5a46edcc9b728e2cf0134d5a9b516ee Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 24 Jan 2010 00:04:07 -0500 Subject: Fix remount races with symlink handling in affs A couple of fields in affs_sb_info is used in follow_link() and symlink() for handling AFFS "absolute" symlinks. Need locking against affs_remount() updates. Signed-off-by: Al Viro --- fs/affs/affs.h | 2 +- fs/affs/namei.c | 7 +++++-- fs/affs/super.c | 17 ++++++++++++++--- fs/affs/symlink.c | 7 +++++-- 4 files changed, 25 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/affs/affs.h b/fs/affs/affs.h index e511dc621a2e..0e40caaba456 100644 --- a/fs/affs/affs.h +++ b/fs/affs/affs.h @@ -106,8 +106,8 @@ struct affs_sb_info { u32 s_last_bmap; struct buffer_head *s_bmap_bh; char *s_prefix; /* Prefix for volumes and assigns. */ - int s_prefix_len; /* Length of prefix. */ char s_volume[32]; /* Volume prefix for absolute symlinks. */ + spinlock_t symlink_lock; /* protects the previous two */ }; #define SF_INTL 0x0001 /* International filesystem. */ diff --git a/fs/affs/namei.c b/fs/affs/namei.c index 960d336ec694..d70bbbac6b7b 100644 --- a/fs/affs/namei.c +++ b/fs/affs/namei.c @@ -341,10 +341,13 @@ affs_symlink(struct inode *dir, struct dentry *dentry, const char *symname) p = (char *)AFFS_HEAD(bh)->table; lc = '/'; if (*symname == '/') { + struct affs_sb_info *sbi = AFFS_SB(sb); while (*symname == '/') symname++; - while (AFFS_SB(sb)->s_volume[i]) /* Cannot overflow */ - *p++ = AFFS_SB(sb)->s_volume[i++]; + spin_lock(&sbi->symlink_lock); + while (sbi->s_volume[i]) /* Cannot overflow */ + *p++ = sbi->s_volume[i++]; + spin_unlock(&sbi->symlink_lock); } while (i < maxlen && (c = *symname++)) { if (c == '.' && lc == '/' && *symname == '.' && symname[1] == '/') { diff --git a/fs/affs/super.c b/fs/affs/super.c index b2a5958c6191..be6a6e8ed7d6 100644 --- a/fs/affs/super.c +++ b/fs/affs/super.c @@ -221,8 +221,6 @@ parse_options(char *options, uid_t *uid, gid_t *gid, int *mode, int *reserved, s *mount_opts |= SF_MUFS; break; case Opt_prefix: - /* Free any previous prefix */ - kfree(*prefix); *prefix = match_strdup(&args[0]); if (!*prefix) return 0; @@ -311,6 +309,7 @@ static int affs_fill_super(struct super_block *sb, void *data, int silent) return -ENOMEM; sb->s_fs_info = sbi; mutex_init(&sbi->s_bmlock); + spin_lock_init(&sbi->symlink_lock); if (!parse_options(data,&uid,&gid,&i,&reserved,&root_block, &blocksize,&sbi->s_prefix, @@ -518,14 +517,18 @@ affs_remount(struct super_block *sb, int *flags, char *data) unsigned long mount_flags; int res = 0; char *new_opts = kstrdup(data, GFP_KERNEL); + char volume[32]; + char *prefix = NULL; pr_debug("AFFS: remount(flags=0x%x,opts=\"%s\")\n",*flags,data); *flags |= MS_NODIRATIME; + memcpy(volume, sbi->s_volume, 32); if (!parse_options(data, &uid, &gid, &mode, &reserved, &root_block, - &blocksize, &sbi->s_prefix, sbi->s_volume, + &blocksize, &prefix, volume, &mount_flags)) { + kfree(prefix); kfree(new_opts); return -EINVAL; } @@ -536,6 +539,14 @@ affs_remount(struct super_block *sb, int *flags, char *data) sbi->s_mode = mode; sbi->s_uid = uid; sbi->s_gid = gid; + /* protect against readers */ + spin_lock(&sbi->symlink_lock); + if (prefix) { + kfree(sbi->s_prefix); + sbi->s_prefix = prefix; + } + memcpy(sbi->s_volume, volume, 32); + spin_unlock(&sbi->symlink_lock); if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY)) { unlock_kernel(); diff --git a/fs/affs/symlink.c b/fs/affs/symlink.c index 41782539c907..ee00f08c4f53 100644 --- a/fs/affs/symlink.c +++ b/fs/affs/symlink.c @@ -20,7 +20,6 @@ static int affs_symlink_readpage(struct file *file, struct page *page) int i, j; char c; char lc; - char *pf; pr_debug("AFFS: follow_link(ino=%lu)\n",inode->i_ino); @@ -32,11 +31,15 @@ static int affs_symlink_readpage(struct file *file, struct page *page) j = 0; lf = (struct slink_front *)bh->b_data; lc = 0; - pf = AFFS_SB(inode->i_sb)->s_prefix ? AFFS_SB(inode->i_sb)->s_prefix : "/"; if (strchr(lf->symname,':')) { /* Handle assign or volume name */ + struct affs_sb_info *sbi = AFFS_SB(inode->i_sb); + char *pf; + spin_lock(&sbi->symlink_lock); + pf = sbi->s_prefix ? sbi->s_prefix : "/"; while (i < 1023 && (c = pf[i])) link[i++] = c; + spin_unlock(&sbi->symlink_lock); while (i < 1023 && lf->symname[j] != ':') link[i++] = lf->symname[j++]; if (i < 1023) -- cgit v1.2.3 From 217686e98321a4ff4c1a6cc535e511e37c5d2dbf Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 24 Jan 2010 00:06:22 -0500 Subject: fix affs parse_options() Error handling in that sucker got broken back in 2003. If function returns 0 on failure, it's not nice to add return -EINVAL into it. Adding return 1 on other failure exits is also not a good thing (and yes, original success exits with 1 and some of failure exits with 0 are still there; so's the original logics in callers). Signed-off-by: Al Viro --- fs/affs/super.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/affs/super.c b/fs/affs/super.c index be6a6e8ed7d6..d41e9673cd97 100644 --- a/fs/affs/super.c +++ b/fs/affs/super.c @@ -203,7 +203,7 @@ parse_options(char *options, uid_t *uid, gid_t *gid, int *mode, int *reserved, s switch (token) { case Opt_bs: if (match_int(&args[0], &n)) - return -EINVAL; + return 0; if (n != 512 && n != 1024 && n != 2048 && n != 4096) { printk ("AFFS: Invalid blocksize (512, 1024, 2048, 4096 allowed)\n"); @@ -213,7 +213,7 @@ parse_options(char *options, uid_t *uid, gid_t *gid, int *mode, int *reserved, s break; case Opt_mode: if (match_octal(&args[0], &option)) - return 1; + return 0; *mode = option & 0777; *mount_opts |= SF_SETMODE; break; @@ -231,21 +231,21 @@ parse_options(char *options, uid_t *uid, gid_t *gid, int *mode, int *reserved, s break; case Opt_reserved: if (match_int(&args[0], reserved)) - return 1; + return 0; break; case Opt_root: if (match_int(&args[0], root)) - return 1; + return 0; break; case Opt_setgid: if (match_int(&args[0], &option)) - return 1; + return 0; *gid = option; *mount_opts |= SF_SETGID; break; case Opt_setuid: if (match_int(&args[0], &option)) - return -EINVAL; + return 0; *uid = option; *mount_opts |= SF_SETUID; break; -- cgit v1.2.3 From 5998649f779b7148a8a0c10c46cfa99e27d34dfe Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 24 Jan 2010 00:52:22 -0500 Subject: Fix failure exits in bfs_fill_super() double iput(), leaks... Signed-off-by: Al Viro --- fs/bfs/inode.c | 43 +++++++++++++++++++++---------------------- 1 file changed, 21 insertions(+), 22 deletions(-) (limited to 'fs') diff --git a/fs/bfs/inode.c b/fs/bfs/inode.c index 6f60336c6628..8f3d9fd89604 100644 --- a/fs/bfs/inode.c +++ b/fs/bfs/inode.c @@ -353,35 +353,35 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent) struct inode *inode; unsigned i, imap_len; struct bfs_sb_info *info; - long ret = -EINVAL; + int ret = -EINVAL; unsigned long i_sblock, i_eblock, i_eoff, s_size; info = kzalloc(sizeof(*info), GFP_KERNEL); if (!info) return -ENOMEM; + mutex_init(&info->bfs_lock); s->s_fs_info = info; sb_set_blocksize(s, BFS_BSIZE); - bh = sb_bread(s, 0); - if(!bh) + info->si_sbh = sb_bread(s, 0); + if (!info->si_sbh) goto out; - bfs_sb = (struct bfs_super_block *)bh->b_data; + bfs_sb = (struct bfs_super_block *)info->si_sbh->b_data; if (le32_to_cpu(bfs_sb->s_magic) != BFS_MAGIC) { if (!silent) printf("No BFS filesystem on %s (magic=%08x)\n", s->s_id, le32_to_cpu(bfs_sb->s_magic)); - goto out; + goto out1; } if (BFS_UNCLEAN(bfs_sb, s) && !silent) printf("%s is unclean, continuing\n", s->s_id); s->s_magic = BFS_MAGIC; - info->si_sbh = bh; if (le32_to_cpu(bfs_sb->s_start) > le32_to_cpu(bfs_sb->s_end)) { printf("Superblock is corrupted\n"); - goto out; + goto out1; } info->si_lasti = (le32_to_cpu(bfs_sb->s_start) - BFS_BSIZE) / @@ -390,7 +390,7 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent) imap_len = (info->si_lasti / 8) + 1; info->si_imap = kzalloc(imap_len, GFP_KERNEL); if (!info->si_imap) - goto out; + goto out1; for (i = 0; i < BFS_ROOT_INO; i++) set_bit(i, info->si_imap); @@ -398,15 +398,13 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent) inode = bfs_iget(s, BFS_ROOT_INO); if (IS_ERR(inode)) { ret = PTR_ERR(inode); - kfree(info->si_imap); - goto out; + goto out2; } s->s_root = d_alloc_root(inode); if (!s->s_root) { iput(inode); ret = -ENOMEM; - kfree(info->si_imap); - goto out; + goto out2; } info->si_blocks = (le32_to_cpu(bfs_sb->s_end) + 1) >> BFS_BSIZE_BITS; @@ -419,10 +417,8 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent) bh = sb_bread(s, info->si_blocks - 1); if (!bh) { printf("Last block not available: %lu\n", info->si_blocks - 1); - iput(inode); ret = -EIO; - kfree(info->si_imap); - goto out; + goto out3; } brelse(bh); @@ -459,11 +455,8 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent) printf("Inode 0x%08x corrupted\n", i); brelse(bh); - s->s_root = NULL; - kfree(info->si_imap); - kfree(info); - s->s_fs_info = NULL; - return -EIO; + ret = -EIO; + goto out3; } if (!di->i_ino) { @@ -483,11 +476,17 @@ static int bfs_fill_super(struct super_block *s, void *data, int silent) s->s_dirt = 1; } dump_imap("read_super", s); - mutex_init(&info->bfs_lock); return 0; +out3: + dput(s->s_root); + s->s_root = NULL; +out2: + kfree(info->si_imap); +out1: + brelse(info->si_sbh); out: - brelse(bh); + mutex_destroy(&info->bfs_lock); kfree(info); s->s_fs_info = NULL; return ret; -- cgit v1.2.3 From ef52c75e4b1603f6e6102967ad90118205168eae Mon Sep 17 00:00:00 2001 From: Al Viro Date: Mon, 25 Jan 2010 04:50:43 -0500 Subject: get rid of pointless checks after simple_pin_fs() if we'd just got success from it, vfsmount won't be NULL Signed-off-by: Al Viro --- fs/debugfs/inode.c | 11 ++--------- security/inode.c | 11 ++--------- 2 files changed, 4 insertions(+), 18 deletions(-) (limited to 'fs') diff --git a/fs/debugfs/inode.c b/fs/debugfs/inode.c index b486169f42bf..274ac865bae8 100644 --- a/fs/debugfs/inode.c +++ b/fs/debugfs/inode.c @@ -160,15 +160,8 @@ static int debugfs_create_by_name(const char *name, mode_t mode, * block. A pointer to that is in the struct vfsmount that we * have around. */ - if (!parent) { - if (debugfs_mount && debugfs_mount->mnt_sb) { - parent = debugfs_mount->mnt_sb->s_root; - } - } - if (!parent) { - pr_debug("debugfs: Ah! can not find a parent!\n"); - return -EFAULT; - } + if (!parent) + parent = debugfs_mount->mnt_sb->s_root; *dentry = NULL; mutex_lock(&parent->d_inode->i_mutex); diff --git a/security/inode.c b/security/inode.c index f7496c6a022b..c3a793881d04 100644 --- a/security/inode.c +++ b/security/inode.c @@ -156,15 +156,8 @@ static int create_by_name(const char *name, mode_t mode, * block. A pointer to that is in the struct vfsmount that we * have around. */ - if (!parent ) { - if (mount && mount->mnt_sb) { - parent = mount->mnt_sb->s_root; - } - } - if (!parent) { - pr_debug("securityfs: Ah! can not find a parent!\n"); - return -EFAULT; - } + if (!parent) + parent = mount->mnt_sb->s_root; mutex_lock(&parent->d_inode->i_mutex); *dentry = lookup_one_len(name, parent, strlen(name)); -- cgit v1.2.3 From 7e32b7bb734047c5e3cecf2e896b9cf8fc35d1e8 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Mon, 25 Jan 2010 06:05:54 -0500 Subject: fix leak in romfs_fill_super() Signed-off-by: Al Viro --- fs/romfs/super.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/romfs/super.c b/fs/romfs/super.c index c117fa80d1e9..42d213546894 100644 --- a/fs/romfs/super.c +++ b/fs/romfs/super.c @@ -544,6 +544,7 @@ error: error_rsb_inval: ret = -EINVAL; error_rsb: + kfree(rsb); return ret; } -- cgit v1.2.3 From 083c73c253c23c20359a344dfe1198ea628e6259 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Mon, 25 Jan 2010 06:16:19 -0500 Subject: fix oops in fs/9p late mount failure if 9P ->get_sb() fails late (at root inode or root dentry allocation), we'll hit its ->kill_sb() with NULL ->s_root Signed-off-by: Al Viro --- fs/9p/vfs_super.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/9p/vfs_super.c b/fs/9p/vfs_super.c index 14a86448572c..69357c0d9899 100644 --- a/fs/9p/vfs_super.c +++ b/fs/9p/vfs_super.c @@ -188,7 +188,8 @@ static void v9fs_kill_super(struct super_block *s) P9_DPRINTK(P9_DEBUG_VFS, " %p\n", s); - v9fs_dentry_release(s->s_root); /* clunk root */ + if (s->s_root) + v9fs_dentry_release(s->s_root); /* clunk root */ kill_anon_super(s); -- cgit v1.2.3 From bbec919150037b8a2e58e32d3ba642ba3b6582a5 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 28 Jan 2010 13:43:50 +0100 Subject: reiserfs: Fix vmalloc call under reiserfs lock Vmalloc is called to allocate journal->j_cnode_free_list but we hold the reiserfs lock at this time, which raises a {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} lock inversion. Just drop the reiserfs lock at this time, as it's not even needed but kept for paranoid reasons. This fixes: [ INFO: inconsistent lock state ] 2.6.33-rc5 #1 --------------------------------- inconsistent {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} usage. kswapd0/313 [HC0[0]:SC0[0]:HE1:SE1] takes: (&REISERFS_SB(s)->lock){+.+.?.}, at: [] reiserfs_write_lock_once+0x28/0x50 {RECLAIM_FS-ON-W} state was registered at: [] mark_held_locks+0x62/0x90 [] lockdep_trace_alloc+0x9a/0xc0 [] kmem_cache_alloc+0x26/0xf0 [] __get_vm_area_node+0x6c/0xf0 [] __vmalloc_node+0x7e/0xa0 [] vmalloc+0x2b/0x30 [] journal_init+0x6cb/0xa10 [] reiserfs_fill_super+0x342/0xb80 [] get_sb_bdev+0x145/0x180 [] get_super_block+0x21/0x30 [] vfs_kern_mount+0x40/0xd0 [] do_kern_mount+0x39/0xd0 [] do_mount+0x2c7/0x6d0 [] sys_mount+0x66/0xa0 [] mount_block_root+0xc4/0x245 [] mount_root+0x59/0x5f [] prepare_namespace+0x111/0x14b [] kernel_init+0xcf/0xdb [] kernel_thread_helper+0x6/0x1c irq event stamp: 63236801 hardirqs last enabled at (63236801): [] __mutex_unlock_slowpath+0x9a/0x120 hardirqs last disabled at (63236800): [] __mutex_unlock_slowpath+0x39/0x120 softirqs last enabled at (63218800): [] __do_softirq+0xc1/0x110 softirqs last disabled at (63218789): [] do_softirq+0x4d/0x60 other info that might help us debug this: 2 locks held by kswapd0/313: #0: (shrinker_rwsem){++++..}, at: [] shrink_slab+0x24/0x170 #1: (&type->s_umount_key#19){++++..}, at: [] shrink_dcache_memory+0xfd/0x1a0 stack backtrace: Pid: 313, comm: kswapd0 Not tainted 2.6.33-rc5 #1 Call Trace: [] ? printk+0x18/0x1c [] print_usage_bug+0x15f/0x1a0 [] mark_lock+0x39f/0x5a0 [] ? trace_hardirqs_off+0xb/0x10 [] ? check_usage_forwards+0x0/0xf0 [] __lock_acquire+0x214/0xa70 [] ? sched_clock_cpu+0x95/0x110 [] lock_acquire+0x7a/0xa0 [] ? reiserfs_write_lock_once+0x28/0x50 [] mutex_lock_nested+0x5f/0x2b0 [] ? reiserfs_write_lock_once+0x28/0x50 [] ? reiserfs_write_lock_once+0x28/0x50 [] reiserfs_write_lock_once+0x28/0x50 [] reiserfs_delete_inode+0x50/0x140 [] ? generic_delete_inode+0x5f/0x150 [] ? reiserfs_delete_inode+0x0/0x140 [] generic_delete_inode+0x9c/0x150 [] generic_drop_inode+0x3d/0x60 [] iput+0x47/0x50 [] dentry_iput+0x6f/0xf0 [] d_kill+0x24/0x50 [] __shrink_dcache_sb+0x21d/0x2b0 [] shrink_dcache_memory+0x12f/0x1a0 [] shrink_slab+0x10e/0x170 [] kswapd+0x477/0x6a0 [] ? isolate_pages_global+0x0/0x1b0 [] ? autoremove_wake_function+0x0/0x40 [] ? kswapd+0x0/0x6a0 [] kthread+0x6c/0x80 [] ? kthread+0x0/0x80 [] kernel_thread_helper+0x6/0x1c Reported-by: Alexander Beregalov Signed-off-by: Frederic Weisbecker Cc: Christian Kujau Cc: Chris Mason --- fs/reiserfs/journal.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c index 83ac4d3b3cb0..ba98546fabbd 100644 --- a/fs/reiserfs/journal.c +++ b/fs/reiserfs/journal.c @@ -2913,7 +2913,9 @@ int journal_init(struct super_block *sb, const char *j_dev_name, journal->j_mount_id = 10; journal->j_state = 0; atomic_set(&(journal->j_jlock), 0); + reiserfs_write_unlock(sb); journal->j_cnode_free_list = allocate_cnodes(num_cnodes); + reiserfs_write_lock(sb); journal->j_cnode_free_orig = journal->j_cnode_free_list; journal->j_cnode_free = journal->j_cnode_free_list ? num_cnodes : 0; journal->j_cnode_used = 0; -- cgit v1.2.3 From 1d6165851cd8e3f919d446cd6da35dee44e8837e Mon Sep 17 00:00:00 2001 From: Dmitry Monakhov Date: Wed, 27 Jan 2010 22:44:36 +0300 Subject: block: fix bio_add_page for non trivial merge_bvec_fn case We have to properly decrease bi_size in order to merge_bvec_fn return right result. Otherwise this result in false merge rejects for two absolutely valid bio_vecs. This may cause significant performance penalty for example fs_block_size == 1k and block device is raid0 with small chunk_size = 8k. Then it is impossible to merge 7-th fs-block in to bio which already has 6 fs-blocks. Cc: Signed-off-by: Dmitry Monakhov Signed-off-by: Jens Axboe --- fs/bio.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/bio.c b/fs/bio.c index 12429c9553eb..88094afc29ea 100644 --- a/fs/bio.c +++ b/fs/bio.c @@ -542,13 +542,18 @@ static int __bio_add_page(struct request_queue *q, struct bio *bio, struct page if (page == prev->bv_page && offset == prev->bv_offset + prev->bv_len) { + unsigned int prev_bv_len = prev->bv_len; prev->bv_len += len; if (q->merge_bvec_fn) { struct bvec_merge_data bvm = { + /* prev_bvec is already charged in + bi_size, discharge it in order to + simulate merging updated prev_bvec + as new bvec. */ .bi_bdev = bio->bi_bdev, .bi_sector = bio->bi_sector, - .bi_size = bio->bi_size, + .bi_size = bio->bi_size - prev_bv_len, .bi_rw = bio->bi_rw, }; -- cgit v1.2.3 From a555f810af6d63ea5960abaed88e150ad95c3011 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 28 Jan 2010 16:18:15 -0500 Subject: Btrfs: Add mount -o compress-force The default btrfs mount -o compress mode will quickly back off compressing a file if it notices that compression does not reduce the size of the data being written. This can save considerable CPU because all future writes to the file go through uncompressed. But some files are both very large and have mixed data stored in them. In that case, we want to add the ability to always try compressing data before writing it. This commit adds mount -o compress-force. A later commit will add a new inode flag that does the same thing. Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 1 + fs/btrfs/inode.c | 3 ++- fs/btrfs/super.c | 9 ++++++++- 3 files changed, 11 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 9f806dd04c27..2aa8ec6a0981 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1161,6 +1161,7 @@ struct btrfs_root { #define BTRFS_MOUNT_SSD_SPREAD (1 << 8) #define BTRFS_MOUNT_NOSSD (1 << 9) #define BTRFS_MOUNT_DISCARD (1 << 10) +#define BTRFS_MOUNT_FORCE_COMPRESS (1 << 11) #define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt) #define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index b330e27c2d8b..f46c57276844 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -483,7 +483,8 @@ again: nr_pages_ret = 0; /* flag the file so we don't compress in the future */ - BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS; + if (!btrfs_test_opt(root, FORCE_COMPRESS)) + BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS; } if (will_compress) { *num_added += 1; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 3f9b45704fcd..8a1ea6e64575 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -66,7 +66,8 @@ enum { Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow, Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, - Opt_compress, Opt_notreelog, Opt_ratio, Opt_flushoncommit, + Opt_compress, Opt_compress_force, Opt_notreelog, Opt_ratio, + Opt_flushoncommit, Opt_discard, Opt_err, }; @@ -82,6 +83,7 @@ static match_table_t tokens = { {Opt_alloc_start, "alloc_start=%s"}, {Opt_thread_pool, "thread_pool=%d"}, {Opt_compress, "compress"}, + {Opt_compress_force, "compress-force"}, {Opt_ssd, "ssd"}, {Opt_ssd_spread, "ssd_spread"}, {Opt_nossd, "nossd"}, @@ -173,6 +175,11 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) printk(KERN_INFO "btrfs: use compression\n"); btrfs_set_opt(info->mount_opt, COMPRESS); break; + case Opt_compress_force: + printk(KERN_INFO "btrfs: forcing compression\n"); + btrfs_set_opt(info->mount_opt, FORCE_COMPRESS); + btrfs_set_opt(info->mount_opt, COMPRESS); + break; case Opt_ssd: printk(KERN_INFO "btrfs: use ssd allocation scheme\n"); btrfs_set_opt(info->mount_opt, SSD); -- cgit v1.2.3 From b8d9bfeb18f9af794020d96e9bee984d18a8d737 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 15 Dec 2009 06:54:17 +0000 Subject: Btrfs: remove tree_search() in extent_map.c This patch removes tree_search() in extent_map.c because it is not called by anything. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/extent_map.c | 14 -------------- 1 file changed, 14 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c index ccbdcb54ec5d..5a4f73b79b75 100644 --- a/fs/btrfs/extent_map.c +++ b/fs/btrfs/extent_map.c @@ -155,20 +155,6 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 offset, return NULL; } -/* - * look for an offset in the tree, and if it can't be found, return - * the first offset we can find smaller than 'offset'. - */ -static inline struct rb_node *tree_search(struct rb_root *root, u64 offset) -{ - struct rb_node *prev; - struct rb_node *ret; - ret = __tree_search(root, offset, &prev, NULL); - if (!ret) - return prev; - return ret; -} - /* check to see if two extent_map structs are adjacent and safe to merge */ static int mergable_maps(struct extent_map *prev, struct extent_map *next) { -- cgit v1.2.3 From d1ea6a61454e7d7ff0873d0ad1ae27d5807da0d3 Mon Sep 17 00:00:00 2001 From: "Aneesh Kumar K.V" Date: Wed, 20 Jan 2010 07:28:54 +0000 Subject: Btrfs: Use correct values when updating inode i_size on fallocate commit f2bc9dd07e3424c4ec5f3949961fe053d47bc825 Author: Aneesh Kumar K.V Date: Wed Jan 20 12:57:53 2010 +0530 Btrfs: Use correct values when updating inode i_size on fallocate Even though we allocate more, we should be updating inode i_size as per the arguments passed Signed-off-by: Aneesh Kumar K.V Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index f46c57276844..5606361b5f04 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -5800,7 +5800,7 @@ out_fail: } static int prealloc_file_range(struct inode *inode, u64 start, u64 end, - u64 alloc_hint, int mode) + u64 alloc_hint, int mode, loff_t actual_len) { struct btrfs_trans_handle *trans; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -5809,6 +5809,7 @@ static int prealloc_file_range(struct inode *inode, u64 start, u64 end, u64 cur_offset = start; u64 num_bytes = end - start; int ret = 0; + u64 i_size; while (num_bytes > 0) { alloc_size = min(num_bytes, root->fs_info->max_extent); @@ -5847,8 +5848,12 @@ static int prealloc_file_range(struct inode *inode, u64 start, u64 end, BTRFS_I(inode)->flags |= BTRFS_INODE_PREALLOC; if (!(mode & FALLOC_FL_KEEP_SIZE) && cur_offset > inode->i_size) { - i_size_write(inode, cur_offset); - btrfs_ordered_update_i_size(inode, cur_offset, NULL); + if (cur_offset > actual_len) + i_size = actual_len; + else + i_size = cur_offset; + i_size_write(inode, i_size); + btrfs_ordered_update_i_size(inode, i_size, NULL); } ret = btrfs_update_inode(trans, root, inode); @@ -5941,7 +5946,7 @@ static long btrfs_fallocate(struct inode *inode, int mode, !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { ret = prealloc_file_range(inode, cur_offset, last_byte, - alloc_hint, mode); + alloc_hint, mode, offset+len); if (ret < 0) { free_extent_map(em); break; -- cgit v1.2.3 From f858153c367a397235d3e81136741e40e44faf1d Mon Sep 17 00:00:00 2001 From: Yang Hongyang Date: Tue, 26 Jan 2010 00:48:23 +0000 Subject: Btrfs: fix a memory leak in btrfs_init_acl In btrfs_init_acl() cloned acl is not released Signed-off-by: Yang Hongyang Signed-off-by: Chris Mason --- fs/btrfs/acl.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c index fa44e92e9b8f..da3133c69830 100644 --- a/fs/btrfs/acl.c +++ b/fs/btrfs/acl.c @@ -269,6 +269,7 @@ int btrfs_init_acl(struct btrfs_trans_handle *trans, ACL_TYPE_ACCESS); } } + posix_acl_release(clone); } failed: posix_acl_release(acl); -- cgit v1.2.3 From e3acc2a6850efff647f1c5458524eb3a8bcba20a Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 26 Jan 2010 14:30:53 +0000 Subject: Btrfs: run orphan cleanup on default fs root This patch revert's commit 6c090a11e1c403b727a6a8eff0b97d5fb9e95cb5 Since it introduces this problem where we can run orphan cleanup on a volume that can have orphan entries re-added. Instead of my original fix, Yan Zheng pointed out that we can just revert my original fix and then run the orphan cleanup in open_ctree after we look up the fs_root. I have tested this with all the tests that gave me problems and this patch fixes both problems. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 6 ++++++ fs/btrfs/inode.c | 6 ------ 2 files changed, 6 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 009e3bd18f23..87b25543d7d1 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1993,6 +1993,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, if (!fs_info->fs_root) goto fail_trans_kthread; + if (!(sb->s_flags & MS_RDONLY)) { + down_read(&fs_info->cleanup_work_sem); + btrfs_orphan_cleanup(fs_info->fs_root); + up_read(&fs_info->cleanup_work_sem); + } + return tree_root; fail_trans_kthread: diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 5606361b5f04..8cd109972fa6 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -3797,12 +3797,6 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) if (location.type == BTRFS_INODE_ITEM_KEY) { inode = btrfs_iget(dir->i_sb, &location, root); - if (unlikely(root->clean_orphans) && - !(inode->i_sb->s_flags & MS_RDONLY)) { - down_read(&root->fs_info->cleanup_work_sem); - btrfs_orphan_cleanup(root); - up_read(&root->fs_info->cleanup_work_sem); - } return inode; } -- cgit v1.2.3 From f48b90756bd834dda852ff514f2690d3175b1f44 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 27 Jan 2010 02:07:59 +0000 Subject: Btrfs: do not mark the chunk as readonly if in degraded mode If a RAID setup has chunks that span multiple disks, and one of those disks has failed, btrfs_chunk_readonly will return 1 since one of the disks in that chunk's stripes is dead and therefore not writeable. So instead if we are in degraded mode, return 0 so we can go ahead and allocate stuff. Without this patch all of the block groups in a RAID1 setup will end up read-only, which will mean we can't add new disks to the array since we won't be able to make allocations. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 220dad5db017..66122bdf8bbf 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -2538,6 +2538,11 @@ int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset) if (!em) return 1; + if (btrfs_test_opt(root, DEGRADED)) { + free_extent_map(em); + return 0; + } + map = (struct map_lookup *)em->bdev; for (i = 0; i < map->num_stripes; i++) { if (!map->stripes[i].dev->writeable) { -- cgit v1.2.3 From 7f59203abeaf18bf3497b308891f95a4489810ad Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 27 Jan 2010 02:09:00 +0000 Subject: Btrfs: check return value of open_bdev_exclusive properly Hit this problem while testing RAID1 failure stuff. open_bdev_exclusive returns ERR_PTR(), not NULL. So change the return value properly. This is important if you accidently specify a device that doesn't exist when trying to add a new device to an array, you will panic the box dereferencing bdev. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 66122bdf8bbf..5eb7459e3782 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1434,8 +1434,8 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path) return -EINVAL; bdev = open_bdev_exclusive(device_path, 0, root->fs_info->bdev_holder); - if (!bdev) - return -EIO; + if (IS_ERR(bdev)) + return PTR_ERR(bdev); if (root->fs_info->fs_devices->seeding) { seeding_dev = 1; -- cgit v1.2.3 From 035fe03a7ad56982b30ab3a522b7b08d58feccd0 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 27 Jan 2010 02:09:38 +0000 Subject: Btrfs: check total number of devices when removing missing If you have a disk failure in RAID1 and then add a new disk to the array, and then try to remove the missing volume, it will fail. The reason is the sanity check only looks at the total number of rw devices, which is just 2 because we have 2 good disks and 1 bad one. Instead check the total number of devices in the array to make sure we can actually remove the device. Tested this with a failed disk setup and with this test we can now run btrfs-vol -r missing /mount/point and it works fine. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 5eb7459e3782..41ecbb2347f2 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1135,7 +1135,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) root->fs_info->avail_metadata_alloc_bits; if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && - root->fs_info->fs_devices->rw_devices <= 4) { + root->fs_info->fs_devices->num_devices <= 4) { printk(KERN_ERR "btrfs: unable to go below four devices " "on raid10\n"); ret = -EINVAL; @@ -1143,7 +1143,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) } if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) && - root->fs_info->fs_devices->rw_devices <= 2) { + root->fs_info->fs_devices->num_devices <= 2) { printk(KERN_ERR "btrfs: unable to go below two " "devices on raid1\n"); ret = -EINVAL; -- cgit v1.2.3 From 221af7f87b97431e3ee21ce4b0e77d5411cf1549 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Thu, 28 Jan 2010 22:14:42 -0800 Subject: Split 'flush_old_exec' into two functions 'flush_old_exec()' is the point of no return when doing an execve(), and it is pretty badly misnamed. It doesn't just flush the old executable environment, it also starts up the new one. Which is very inconvenient for things like setting up the new personality, because we want the new personality to affect the starting of the new environment, but at the same time we do _not_ want the new personality to take effect if flushing the old one fails. As a result, the x86-64 '32-bit' personality is actually done using this insane "I'm going to change the ABI, but I haven't done it yet" bit (TIF_ABI_PENDING), with SET_PERSONALITY() not actually setting the personality, but just the "pending" bit, so that "flush_thread()" can do the actual personality magic. This patch in no way changes any of that insanity, but it does split the 'flush_old_exec()' function up into a preparatory part that can fail (still called flush_old_exec()), and a new part that will actually set up the new exec environment (setup_new_exec()). All callers are changed to trivially comply with the new world order. Signed-off-by: H. Peter Anvin Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- arch/sh/kernel/process_64.c | 2 +- arch/x86/ia32/ia32_aout.c | 10 ++++++---- fs/binfmt_aout.c | 1 + fs/binfmt_elf.c | 27 ++------------------------- fs/binfmt_elf_fdpic.c | 3 +++ fs/binfmt_flat.c | 1 + fs/binfmt_som.c | 1 + fs/exec.c | 26 ++++++++++++++++---------- include/linux/binfmts.h | 1 + include/linux/sched.h | 2 +- 10 files changed, 33 insertions(+), 41 deletions(-) (limited to 'fs') diff --git a/arch/sh/kernel/process_64.c b/arch/sh/kernel/process_64.c index 31f80c61b031..ec79faf6f021 100644 --- a/arch/sh/kernel/process_64.c +++ b/arch/sh/kernel/process_64.c @@ -368,7 +368,7 @@ void exit_thread(void) void flush_thread(void) { - /* Called by fs/exec.c (flush_old_exec) to remove traces of a + /* Called by fs/exec.c (setup_new_exec) to remove traces of a * previously running executable. */ #ifdef CONFIG_SH_FPU if (last_task_used_math == current) { diff --git a/arch/x86/ia32/ia32_aout.c b/arch/x86/ia32/ia32_aout.c index 2a4d073d2cf1..435d2a5323da 100644 --- a/arch/x86/ia32/ia32_aout.c +++ b/arch/x86/ia32/ia32_aout.c @@ -308,15 +308,17 @@ static int load_aout_binary(struct linux_binprm *bprm, struct pt_regs *regs) if (retval) return retval; - regs->cs = __USER32_CS; - regs->r8 = regs->r9 = regs->r10 = regs->r11 = regs->r12 = - regs->r13 = regs->r14 = regs->r15 = 0; - /* OK, This is the point of no return */ set_personality(PER_LINUX); set_thread_flag(TIF_IA32); clear_thread_flag(TIF_ABI_PENDING); + setup_new_exec(bprm); + + regs->cs = __USER32_CS; + regs->r8 = regs->r9 = regs->r10 = regs->r11 = regs->r12 = + regs->r13 = regs->r14 = regs->r15 = 0; + current->mm->end_code = ex.a_text + (current->mm->start_code = N_TXTADDR(ex)); current->mm->end_data = ex.a_data + diff --git a/fs/binfmt_aout.c b/fs/binfmt_aout.c index 346b69405363..fdd397099172 100644 --- a/fs/binfmt_aout.c +++ b/fs/binfmt_aout.c @@ -264,6 +264,7 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs) #else set_personality(PER_LINUX); #endif + setup_new_exec(bprm); current->mm->end_code = ex.a_text + (current->mm->start_code = N_TXTADDR(ex)); diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c index edd90c49003c..fd5b2ea5d299 100644 --- a/fs/binfmt_elf.c +++ b/fs/binfmt_elf.c @@ -662,27 +662,6 @@ static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs) if (elf_interpreter[elf_ppnt->p_filesz - 1] != '\0') goto out_free_interp; - /* - * The early SET_PERSONALITY here is so that the lookup - * for the interpreter happens in the namespace of the - * to-be-execed image. SET_PERSONALITY can select an - * alternate root. - * - * However, SET_PERSONALITY is NOT allowed to switch - * this task into the new images's memory mapping - * policy - that is, TASK_SIZE must still evaluate to - * that which is appropriate to the execing application. - * This is because exit_mmap() needs to have TASK_SIZE - * evaluate to the size of the old image. - * - * So if (say) a 64-bit application is execing a 32-bit - * application it is the architecture's responsibility - * to defer changing the value of TASK_SIZE until the - * switch really is going to happen - do this in - * flush_thread(). - akpm - */ - SET_PERSONALITY(loc->elf_ex); - interpreter = open_exec(elf_interpreter); retval = PTR_ERR(interpreter); if (IS_ERR(interpreter)) @@ -730,9 +709,6 @@ static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs) /* Verify the interpreter has a valid arch */ if (!elf_check_arch(&loc->interp_elf_ex)) goto out_free_dentry; - } else { - /* Executables without an interpreter also need a personality */ - SET_PERSONALITY(loc->elf_ex); } /* Flush all traces of the currently running executable */ @@ -752,7 +728,8 @@ static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs) if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space) current->flags |= PF_RANDOMIZE; - arch_pick_mmap_layout(current->mm); + + setup_new_exec(bprm); /* Do this so that we can load the interpreter, if need be. We will change some of these later */ diff --git a/fs/binfmt_elf_fdpic.c b/fs/binfmt_elf_fdpic.c index c57d9ce5ff7e..18d77297ccc8 100644 --- a/fs/binfmt_elf_fdpic.c +++ b/fs/binfmt_elf_fdpic.c @@ -321,6 +321,9 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm, set_personality(PER_LINUX_FDPIC); if (elf_read_implies_exec(&exec_params.hdr, executable_stack)) current->personality |= READ_IMPLIES_EXEC; + + setup_new_exec(bprm); + set_binfmt(&elf_fdpic_format); current->mm->start_code = 0; diff --git a/fs/binfmt_flat.c b/fs/binfmt_flat.c index d4a00ea1054c..42c6b4a54445 100644 --- a/fs/binfmt_flat.c +++ b/fs/binfmt_flat.c @@ -519,6 +519,7 @@ static int load_flat_file(struct linux_binprm * bprm, /* OK, This is the point of no return */ set_personality(PER_LINUX_32BIT); + setup_new_exec(bprm); } /* diff --git a/fs/binfmt_som.c b/fs/binfmt_som.c index 2a9b5330cc5e..cc8560f6c9b0 100644 --- a/fs/binfmt_som.c +++ b/fs/binfmt_som.c @@ -227,6 +227,7 @@ load_som_binary(struct linux_binprm * bprm, struct pt_regs * regs) /* OK, This is the point of no return */ current->flags &= ~PF_FORKNOEXEC; current->personality = PER_HPUX; + setup_new_exec(bprm); /* Set the task size for HP-UX processes such that * the gateway page is outside the address space. diff --git a/fs/exec.c b/fs/exec.c index 632b02e34ec7..675c3f44c2ea 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -941,9 +941,7 @@ void set_task_comm(struct task_struct *tsk, char *buf) int flush_old_exec(struct linux_binprm * bprm) { - char * name; - int i, ch, retval; - char tcomm[sizeof(current->comm)]; + int retval; /* * Make sure we have a private signal table and that @@ -963,6 +961,20 @@ int flush_old_exec(struct linux_binprm * bprm) goto out; bprm->mm = NULL; /* We're using it now */ + return 0; + +out: + return retval; +} +EXPORT_SYMBOL(flush_old_exec); + +void setup_new_exec(struct linux_binprm * bprm) +{ + int i, ch; + char * name; + char tcomm[sizeof(current->comm)]; + + arch_pick_mmap_layout(current->mm); /* This is the point of no return */ current->sas_ss_sp = current->sas_ss_size = 0; @@ -1019,14 +1031,8 @@ int flush_old_exec(struct linux_binprm * bprm) flush_signal_handlers(current, 0); flush_old_files(current->files); - - return 0; - -out: - return retval; } - -EXPORT_SYMBOL(flush_old_exec); +EXPORT_SYMBOL(setup_new_exec); /* * Prepare credentials and lock ->cred_guard_mutex. diff --git a/include/linux/binfmts.h b/include/linux/binfmts.h index cd4349bdc34e..89c6249fc561 100644 --- a/include/linux/binfmts.h +++ b/include/linux/binfmts.h @@ -109,6 +109,7 @@ extern int prepare_binprm(struct linux_binprm *); extern int __must_check remove_arg_zero(struct linux_binprm *); extern int search_binary_handler(struct linux_binprm *,struct pt_regs *); extern int flush_old_exec(struct linux_binprm * bprm); +extern void setup_new_exec(struct linux_binprm * bprm); extern int suid_dumpable; #define SUID_DUMP_DISABLE 0 /* No setuid dumping */ diff --git a/include/linux/sched.h b/include/linux/sched.h index 6f7bba93929b..abdfacc58653 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1369,7 +1369,7 @@ struct task_struct { char comm[TASK_COMM_LEN]; /* executable name excluding path - access with [gs]et_task_comm (which lock it with task_lock()) - - initialized normally by flush_old_exec */ + - initialized normally by setup_new_exec */ /* file system info */ int link_count, total_link_count; #ifdef CONFIG_SYSVIPC -- cgit v1.2.3 From 9e9432c267e4047db98b9d4fba95099c6effcef9 Mon Sep 17 00:00:00 2001 From: Chuck Ebbert Date: Sat, 30 Jan 2010 20:28:19 +0100 Subject: block: fix bugs in bio-integrity mempool usage Fix two bugs in the bio integrity code: use_bip_pool() always returns 0 because it checks against the wrong limit, causing the mempool to be used only when regular allocation fails. When the mempool is used as a fallback we don't free the data properly. Signed-Off-By: Chuck Ebbert Acked-by: Martin K. Petersen Signed-off-by: Jens Axboe --- fs/bio-integrity.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c index 49a34e7f7306..a16f29e888cd 100644 --- a/fs/bio-integrity.c +++ b/fs/bio-integrity.c @@ -61,7 +61,7 @@ static inline unsigned int vecs_to_idx(unsigned int nr) static inline int use_bip_pool(unsigned int idx) { - if (idx == BIOVEC_NR_POOLS) + if (idx == BIOVEC_MAX_IDX) return 1; return 0; @@ -95,6 +95,7 @@ struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *bio, /* Use mempool if lower order alloc failed or max vecs were requested */ if (bip == NULL) { + idx = BIOVEC_MAX_IDX; /* so we free the payload properly later */ bip = mempool_alloc(bs->bio_integrity_pool, gfp_mask); if (unlikely(bip == NULL)) { -- cgit v1.2.3 From 3256a05531b1164a9c138da701b922a113bddf82 Mon Sep 17 00:00:00 2001 From: Ryusuke Konishi Date: Sun, 31 Jan 2010 12:39:50 +0900 Subject: nilfs2: fix potential leak of dirty data on umount This fixes incorrect usage of nilfs_segctor_confirm() test function in nilfs_segctor_destroy(); nilfs_segctor_confirm() returns zero if the filesystem is not clean, so its use in nilfs_segctor_destroy() needs inversion. Signed-off-by: Ryusuke Konishi --- fs/nilfs2/segment.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c index 17584c524486..105b508b47a8 100644 --- a/fs/nilfs2/segment.c +++ b/fs/nilfs2/segment.c @@ -2829,7 +2829,7 @@ static void nilfs_segctor_destroy(struct nilfs_sc_info *sci) || sci->sc_seq_request != sci->sc_seq_done); spin_unlock(&sci->sc_state_lock); - if (flag || nilfs_segctor_confirm(sci)) + if (flag || !nilfs_segctor_confirm(sci)) nilfs_segctor_write_out(sci); WARN_ON(!list_empty(&sci->sc_copied_buffers)); -- cgit v1.2.3 From 55f0b4c546d1c87cccba63dc0fc5eb70e2b41733 Mon Sep 17 00:00:00 2001 From: Benjamin Marzinski Date: Mon, 25 Jan 2010 11:23:24 -0600 Subject: GFS2: Don't withdraw on partial rindex entries ince gfs2 writes the rindex file a block at a time, and releases the exclusive lock after each block, it is possible that another process will grab the lock in the middle of the write. Since rindex entries are not an even divisor of blocks, that other process may see partial entries. On grows, this is fine. The process can simply ignore the the partial entires. Previously, the code withdrew when it saw partial entries. Now it simply ignores them. Signed-off-by: Benjamin Marzinski Signed-off-by: Steven Whitehouse --- fs/gfs2/rgrp.c | 5 ----- 1 file changed, 5 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c index 0608f490c295..6702b82b375c 100644 --- a/fs/gfs2/rgrp.c +++ b/fs/gfs2/rgrp.c @@ -591,11 +591,6 @@ static int gfs2_ri_update(struct gfs2_inode *ip) u64 rgrp_count = ip->i_disksize; int error; - if (do_div(rgrp_count, sizeof(struct gfs2_rindex))) { - gfs2_consist_inode(ip); - return -EIO; - } - clear_rgrpdi(sdp); file_ra_state_init(&ra_state, inode->i_mapping); -- cgit v1.2.3 From 7fe3ec6fe58d2bfe97fe7a5d731c29299a8ffd35 Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Fri, 29 Jan 2010 15:20:34 +0000 Subject: GFS2: Fix previous patch The do_div() call needs to remain. Signed-off-by: Steven Whitehouse --- fs/gfs2/rgrp.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c index 6702b82b375c..46534a554cce 100644 --- a/fs/gfs2/rgrp.c +++ b/fs/gfs2/rgrp.c @@ -591,6 +591,7 @@ static int gfs2_ri_update(struct gfs2_inode *ip) u64 rgrp_count = ip->i_disksize; int error; + do_div(rgrp_count, sizeof(struct gfs2_rindex)); clear_rgrpdi(sdp); file_ra_state_init(&ra_state, inode->i_mapping); -- cgit v1.2.3 From ea8d62dadd0217334fb2c5d60e7f89e14076ca10 Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Fri, 29 Jan 2010 15:48:57 +0000 Subject: GFS2: Use GFP_NOFS for alloc structure This is called under a glock, so its a good plan to use GFP_NOFS Signed-off-by: Steven Whitehouse --- fs/gfs2/rgrp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c index 46534a554cce..503b842f3ba2 100644 --- a/fs/gfs2/rgrp.c +++ b/fs/gfs2/rgrp.c @@ -911,7 +911,7 @@ void gfs2_rgrp_repolish_clones(struct gfs2_rgrpd *rgd) struct gfs2_alloc *gfs2_alloc_get(struct gfs2_inode *ip) { BUG_ON(ip->i_alloc != NULL); - ip->i_alloc = kzalloc(sizeof(struct gfs2_alloc), GFP_KERNEL); + ip->i_alloc = kzalloc(sizeof(struct gfs2_alloc), GFP_NOFS); return ip->i_alloc; } -- cgit v1.2.3 From 7ab02af428c2d312c0cf8fb0b01cc1eb21131a3d Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Tue, 2 Feb 2010 12:37:44 -0800 Subject: Fix 'flush_old_exec()/setup_new_exec()' split Commit 221af7f87b9 ("Split 'flush_old_exec' into two functions") split the function at the point of no return - ie right where there were no more error cases to check. That made sense from a technical standpoint, but when we then also combined it with the actual personality setting going in between flush_old_exec() and setup_new_exec(), it needs to be a bit more careful. In particular, we need to make sure that we really flush the old personality bits in the 'flush' stage, rather than later in the 'setup' stage, since otherwise we might be flushing the _new_ personality state that we're just setting up. So this moves the flags and personality flushing (and 'flush_thread()', which is the arch-specific function that generally resets lazy FP state etc) of the old process into flush_old_exec(), so that it doesn't affect any state that execve() is setting up for the new process environment. This was reported by Michal Simek as breaking his Microblaze qemu environment. Reported-and-tested-by: Michal Simek Cc: Peter Anvin Signed-off-by: Linus Torvalds --- fs/exec.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/exec.c b/fs/exec.c index 675c3f44c2ea..0790a107ff7e 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -961,6 +961,11 @@ int flush_old_exec(struct linux_binprm * bprm) goto out; bprm->mm = NULL; /* We're using it now */ + + current->flags &= ~PF_RANDOMIZE; + flush_thread(); + current->personality &= ~bprm->per_clear; + return 0; out: @@ -997,9 +1002,6 @@ void setup_new_exec(struct linux_binprm * bprm) tcomm[i] = '\0'; set_task_comm(current, tcomm); - current->flags &= ~PF_RANDOMIZE; - flush_thread(); - /* Set the new mm task size. We have to do that late because it may * depend on TIF_32BIT which is only updated in flush_thread() on * some architectures like powerpc @@ -1015,8 +1017,6 @@ void setup_new_exec(struct linux_binprm * bprm) set_dumpable(current->mm, suid_dumpable); } - current->personality &= ~bprm->per_clear; - /* * Flush performance counters when crossing a * security domain: -- cgit v1.2.3 From 931e80e4b3263db75c8e34f078d22f11bbabd3a3 Mon Sep 17 00:00:00 2001 From: anfei zhou Date: Tue, 2 Feb 2010 13:44:02 -0800 Subject: mm: flush dcache before writing into page to avoid alias The cache alias problem will happen if the changes of user shared mapping is not flushed before copying, then user and kernel mapping may be mapped into two different cache line, it is impossible to guarantee the coherence after iov_iter_copy_from_user_atomic. So the right steps should be: flush_dcache_page(page); kmap_atomic(page); write to page; kunmap_atomic(page); flush_dcache_page(page); More precisely, we might create two new APIs flush_dcache_user_page and flush_dcache_kern_page to replace the two flush_dcache_page accordingly. Here is a snippet tested on omap2430 with VIPT cache, and I think it is not ARM-specific: int val = 0x11111111; fd = open("abc", O_RDWR); addr = mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0); *(addr+0) = 0x44444444; tmp = *(addr+0); *(addr+1) = 0x77777777; write(fd, &val, sizeof(int)); close(fd); The results are not always 0x11111111 0x77777777 at the beginning as expected. Sometimes we see 0x44444444 0x77777777. Signed-off-by: Anfei Cc: Russell King Cc: Miklos Szeredi Cc: Nick Piggin Cc: Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/fuse/file.c | 3 +++ mm/filemap.c | 3 +++ 2 files changed, 6 insertions(+) (limited to 'fs') diff --git a/fs/fuse/file.c b/fs/fuse/file.c index c18913a777ae..a9f5e137f1d3 100644 --- a/fs/fuse/file.c +++ b/fs/fuse/file.c @@ -828,6 +828,9 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req, if (!page) break; + if (mapping_writably_mapped(mapping)) + flush_dcache_page(page); + pagefault_disable(); tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes); pagefault_enable(); diff --git a/mm/filemap.c b/mm/filemap.c index e3736923220e..698ea80f2102 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -2232,6 +2232,9 @@ again: if (unlikely(status)) break; + if (mapping_writably_mapped(mapping)) + flush_dcache_page(page); + pagefault_disable(); copied = iov_iter_copy_from_user_atomic(page, i, offset, bytes); pagefault_enable(); -- cgit v1.2.3 From d622b89a2f58613a9c1407b22b02aecdd2187a7c Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Sat, 30 Jan 2010 23:32:19 +0800 Subject: ocfs2: Fix memory overflow in cow_by_page. In ocfs2_duplicate_clusters_by_page, we calculate map_end by shifting page_index. But actually in case we meet with a large offset(say in a i686 box, poff_t is only 32 bits and page_index=2056240), we will overflow. So change the type of page_index to loff_t. Signed-off-by: Tao Ma Signed-off-by: Joel Becker --- fs/ocfs2/refcounttree.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c index 74db2be75dd6..5b64468de0b0 100644 --- a/fs/ocfs2/refcounttree.c +++ b/fs/ocfs2/refcounttree.c @@ -2945,7 +2945,7 @@ static int ocfs2_duplicate_clusters_by_page(handle_t *handle, while (offset < end) { page_index = offset >> PAGE_CACHE_SHIFT; - map_end = (page_index + 1) << PAGE_CACHE_SHIFT; + map_end = ((loff_t)page_index + 1) << PAGE_CACHE_SHIFT; if (map_end > end) map_end = end; @@ -3170,7 +3170,7 @@ static int ocfs2_cow_sync_writeback(struct super_block *sb, while (offset < end) { page_index = offset >> PAGE_CACHE_SHIFT; - map_end = (page_index + 1) << PAGE_CACHE_SHIFT; + map_end = ((loff_t)page_index + 1) << PAGE_CACHE_SHIFT; if (map_end > end) map_end = end; -- cgit v1.2.3 From 0a1ea437d87af830786605813972e8e277992917 Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Mon, 1 Feb 2010 17:05:33 +0800 Subject: ocfs2: Only bug out when page size is larger than cluster size. In CoW, we have to make sure that the page is already written out to the disk. So we have a BUG_ON(PageDirty(page)). In ppc platform we have pagesize=64K, so if the cs=4K, if the file have fragmented clusters, we will map the page many times. See this file as an example. Tree Depth: 0 Count: 19 Next Free Rec: 14 ## Offset Clusters Block# Flags 0 0 4 2164864 0x2 Refcounted 1 4 2 9302792 0x2 Refcounted ... We have to replace the extent recs one by one, so the page with index 0 will be mapped and dirtied twice. I'd like to leave the BUG_ON there while adding a check so that in case we meet with an error in other platforms, we can find it easily. Signed-off-by: Tao Ma Signed-off-by: Joel Becker --- fs/ocfs2/refcounttree.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c index 5b64468de0b0..8ae65c9c020c 100644 --- a/fs/ocfs2/refcounttree.c +++ b/fs/ocfs2/refcounttree.c @@ -2957,8 +2957,12 @@ static int ocfs2_duplicate_clusters_by_page(handle_t *handle, page = grab_cache_page(mapping, page_index); - /* This page can't be dirtied before we CoW it out. */ - BUG_ON(PageDirty(page)); + /* + * In case PAGE_CACHE_SIZE <= CLUSTER_SIZE, This page + * can't be dirtied before we CoW it out. + */ + if (PAGE_CACHE_SIZE <= OCFS2_SB(sb)->s_clustersize) + BUG_ON(PageDirty(page)); if (!PageUptodate(page)) { ret = block_read_full_page(page, ocfs2_get_block); -- cgit v1.2.3 From 60c486744c9a30ea60fa863e9587242dde2fe4bd Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Wed, 3 Feb 2010 09:56:04 +0800 Subject: ocfs2: Add parenthesis to wrap the check for O_DIRECT. Add parenthesis to wrap the check for O_DIRECT. Signed-off-by: Tao Ma Signed-off-by: Joel Becker --- fs/ocfs2/file.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 65e9375d2fb3..558ce0312421 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -2013,8 +2013,8 @@ out_dio: /* buffered aio wouldn't have proper lock coverage today */ BUG_ON(ret == -EIOCBQUEUED && !(file->f_flags & O_DIRECT)); - if ((file->f_flags & O_DSYNC && !direct_io) || IS_SYNC(inode) || - (file->f_flags & O_DIRECT && has_refcount)) { + if (((file->f_flags & O_DSYNC) && !direct_io) || IS_SYNC(inode) || + ((file->f_flags & O_DIRECT) && has_refcount)) { ret = filemap_fdatawrite_range(file->f_mapping, pos, pos + count - 1); if (ret < 0) -- cgit v1.2.3 From cd34edd8cf80b507bb84b3f0c2988fe05099ffb5 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 25 Jan 2010 17:58:30 -0800 Subject: ocfs2/dlm: Handle EAGAIN for compatibility - v2 Mainline commit aad1b15310b9bcd59fa81ab8f2b1513b59553ea8 made the dlm_begin_reco_handler() return -EAGAIN instead of EAGAIN. As this error is transmitted over the wire, we want the receiver, dlm_send_begin_reco_message(), to understand both the older EAGAIN and the newer -EAGAIN, to allow rolling upgrade of the cluster nodes. Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlm/dlmrecovery.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index cfb2ae9ab538..ad712211d4ea 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -2639,7 +2639,13 @@ retry: "begin reco msg (%d)\n", dlm->name, nodenum, ret); ret = 0; } - if (ret == -EAGAIN) { + + /* + * Prior to commit aad1b15310b9bcd59fa81ab8f2b1513b59553ea8, + * dlm_begin_reco_handler() returned EAGAIN and not -EAGAIN. + * We are handling both for compatibility reasons. + */ + if (ret == -EAGAIN || ret == EAGAIN) { mlog(0, "%s: trying to start recovery of node " "%u, but node %u is waiting for last recovery " "to complete, backoff for a bit\n", dlm->name, -- cgit v1.2.3 From 34e6c59af06cbca07b1490ec0015ea2d303470d3 Mon Sep 17 00:00:00 2001 From: Tao Ma Date: Wed, 27 Jan 2010 10:21:52 +0800 Subject: ocfs2: Use compat_ptr in reflink_arguments. Although we use u64 to pass userspace pointers to the kernel to avoid compat_ioctl, it doesn't work in some ppc platform. So wrap them with compat_ptr and add compat_ioctl. The detailed discussion about compat_ptr can be found in thread http://lkml.org/lkml/2009/10/27/423. We indeed met with a bug when testing on ppc(-EFAULT is returned when using old_path). This patch try to fix this. I have tested in ppc64(with 32 bit reflink) and x86_64(with i686 reflink), both works. Signed-off-by: Tao Ma Signed-off-by: Joel Becker --- fs/ocfs2/ioctl.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c index 31fbb0619510..7d9d9c132cef 100644 --- a/fs/ocfs2/ioctl.c +++ b/fs/ocfs2/ioctl.c @@ -7,6 +7,7 @@ #include #include +#include #define MLOG_MASK_PREFIX ML_INODE #include @@ -181,6 +182,10 @@ long ocfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) #ifdef CONFIG_COMPAT long ocfs2_compat_ioctl(struct file *file, unsigned cmd, unsigned long arg) { + bool preserve; + struct reflink_arguments args; + struct inode *inode = file->f_path.dentry->d_inode; + switch (cmd) { case OCFS2_IOC32_GETFLAGS: cmd = OCFS2_IOC_GETFLAGS; @@ -195,8 +200,15 @@ long ocfs2_compat_ioctl(struct file *file, unsigned cmd, unsigned long arg) case OCFS2_IOC_GROUP_EXTEND: case OCFS2_IOC_GROUP_ADD: case OCFS2_IOC_GROUP_ADD64: - case OCFS2_IOC_REFLINK: break; + case OCFS2_IOC_REFLINK: + if (copy_from_user(&args, (struct reflink_arguments *)arg, + sizeof(args))) + return -EFAULT; + preserve = (args.preserve != 0); + + return ocfs2_reflink_ioctl(inode, compat_ptr(args.old_path), + compat_ptr(args.new_path), preserve); default: return -ENOIOCTLCMD; } -- cgit v1.2.3 From 0b94a909eb2e2f6990d05fd486a0cb4902ef1ae7 Mon Sep 17 00:00:00 2001 From: Wengang Wang Date: Thu, 21 Jan 2010 10:50:02 -0800 Subject: ocfs2: Fix setting of OCFS2_LOCK_BLOCKED during bast During bast, set the OCFS2_LOCK_BLOCKED flag only if the lock needs to downconverted. Signed-off-by: Wengang Wang Acked-by: Sunil Mushran Acked-by: Mark Fasheh Signed-off-by: Joel Becker --- fs/ocfs2/dlmglue.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index 172f4c6ce1be..0cdf63042b76 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -907,8 +907,6 @@ static int ocfs2_generic_handle_bast(struct ocfs2_lock_res *lockres, assert_spin_locked(&lockres->l_lock); - lockres_or_flags(lockres, OCFS2_LOCK_BLOCKED); - if (level > lockres->l_blocking) { /* only schedule a downconvert if we haven't already scheduled * one that goes low enough to satisfy the level we're @@ -921,6 +919,9 @@ static int ocfs2_generic_handle_bast(struct ocfs2_lock_res *lockres, lockres->l_blocking = level; } + if (needs_downconvert) + lockres_or_flags(lockres, OCFS2_LOCK_BLOCKED); + mlog_exit(needs_downconvert); return needs_downconvert; } -- cgit v1.2.3 From a19128260107f951d1b4c421cf98b92f8092b069 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Thu, 21 Jan 2010 10:50:03 -0800 Subject: ocfs2: Prevent a livelock in dlmglue There is possibility of a livelock in __ocfs2_cluster_lock(). If a node were to get an ast for an upconvert request, followed immediately by a bast, there is a small window where the fs may downconvert the lock before the process requesting the upconvert is able to take the lock. This patch adds a new flag to indicate that the upconvert is still in progress and that the dc thread should not downconvert it right now. Wengang Wang and Joel Becker contributed heavily to this patch. Reported-by: David Teigland Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlmglue.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++--- fs/ocfs2/ocfs2.h | 4 ++++ 2 files changed, 50 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index 0cdf63042b76..85d7c490755b 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -875,6 +875,14 @@ static inline void ocfs2_generic_handle_convert_action(struct ocfs2_lock_res *lo lockres_or_flags(lockres, OCFS2_LOCK_NEEDS_REFRESH); lockres->l_level = lockres->l_requested; + + /* + * We set the OCFS2_LOCK_UPCONVERT_FINISHING flag before clearing + * the OCFS2_LOCK_BUSY flag to prevent the dc thread from + * downconverting the lock before the upconvert has fully completed. + */ + lockres_or_flags(lockres, OCFS2_LOCK_UPCONVERT_FINISHING); + lockres_clear_flags(lockres, OCFS2_LOCK_BUSY); mlog_exit_void(); @@ -1134,6 +1142,7 @@ static inline void ocfs2_recover_from_dlm_error(struct ocfs2_lock_res *lockres, mlog_entry_void(); spin_lock_irqsave(&lockres->l_lock, flags); lockres_clear_flags(lockres, OCFS2_LOCK_BUSY); + lockres_clear_flags(lockres, OCFS2_LOCK_UPCONVERT_FINISHING); if (convert) lockres->l_action = OCFS2_AST_INVALID; else @@ -1324,13 +1333,13 @@ static int __ocfs2_cluster_lock(struct ocfs2_super *osb, again: wait = 0; + spin_lock_irqsave(&lockres->l_lock, flags); + if (catch_signals && signal_pending(current)) { ret = -ERESTARTSYS; - goto out; + goto unlock; } - spin_lock_irqsave(&lockres->l_lock, flags); - mlog_bug_on_msg(lockres->l_flags & OCFS2_LOCK_FREEING, "Cluster lock called on freeing lockres %s! flags " "0x%lx\n", lockres->l_name, lockres->l_flags); @@ -1347,6 +1356,25 @@ again: goto unlock; } + if (lockres->l_flags & OCFS2_LOCK_UPCONVERT_FINISHING) { + /* + * We've upconverted. If the lock now has a level we can + * work with, we take it. If, however, the lock is not at the + * required level, we go thru the full cycle. One way this could + * happen is if a process requesting an upconvert to PR is + * closely followed by another requesting upconvert to an EX. + * If the process requesting EX lands here, we want it to + * continue attempting to upconvert and let the process + * requesting PR take the lock. + * If multiple processes request upconvert to PR, the first one + * here will take the lock. The others will have to go thru the + * OCFS2_LOCK_BLOCKED check to ensure that there is no pending + * downconvert request. + */ + if (level <= lockres->l_level) + goto update_holders; + } + if (lockres->l_flags & OCFS2_LOCK_BLOCKED && !ocfs2_may_continue_on_blocked_lock(lockres, level)) { /* is the lock is currently blocked on behalf of @@ -1417,11 +1445,14 @@ again: goto again; } +update_holders: /* Ok, if we get here then we're good to go. */ ocfs2_inc_holders(lockres, level); ret = 0; unlock: + lockres_clear_flags(lockres, OCFS2_LOCK_UPCONVERT_FINISHING); + spin_unlock_irqrestore(&lockres->l_lock, flags); out: /* @@ -3402,6 +3433,18 @@ recheck: goto leave; } + /* + * This prevents livelocks. OCFS2_LOCK_UPCONVERT_FINISHING flag is + * set when the ast is received for an upconvert just before the + * OCFS2_LOCK_BUSY flag is cleared. Now if the fs received a bast + * on the heels of the ast, we want to delay the downconvert just + * enough to allow the up requestor to do its task. Because this + * lock is in the blocked queue, the lock will be downconverted + * as soon as the requestor is done with the lock. + */ + if (lockres->l_flags & OCFS2_LOCK_UPCONVERT_FINISHING) + goto leave_requeue; + /* if we're blocking an exclusive and we have *any* holders, * then requeue. */ if ((lockres->l_blocking == DLM_LOCK_EX) diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h index 9362eea7424b..740f448041e2 100644 --- a/fs/ocfs2/ocfs2.h +++ b/fs/ocfs2/ocfs2.h @@ -136,6 +136,10 @@ enum ocfs2_unlock_action { #define OCFS2_LOCK_PENDING (0x00000400) /* This lockres is pending a call to dlm_lock. Only exists with BUSY set. */ +#define OCFS2_LOCK_UPCONVERT_FINISHING (0x00000800) /* blocks the dc thread + * from downconverting + * before the upconvert + * has completed */ struct ocfs2_lock_res_ops; -- cgit v1.2.3 From 0d74125a6a68d4f1969ecaf0b3543f315916ccdc Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Fri, 29 Jan 2010 09:44:11 -0800 Subject: ocfs2: Do not downconvert if the lock level is already compatible During upconvert, if the master were to send a BAST, dlmglue will detect the upconversion in process and send a cancel convert to the master. Upon receiving the AST for the cancel convert, it will re-process the lock resource to determine whether it needs downconverting. Say, the up was from PR to EX and the BAST was for EX. After the cancel convert, it will need to downconvert to NL. However, if the node was originally upconverting from NL to EX, then there would be no reason to downconvert (assuming the same message sequence). This patch makes dlmglue consider the possibility that the current lock level is already compatible and that downconverting is not required. Joel Becker assisted in fixing this issue. Fixes ossbz#1178 http://oss.oracle.com/bugzilla/show_bug.cgi?id=1178 Reported-by: Coly Li Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlmglue.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'fs') diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index 85d7c490755b..ac24f49ae2fb 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -3445,6 +3445,19 @@ recheck: if (lockres->l_flags & OCFS2_LOCK_UPCONVERT_FINISHING) goto leave_requeue; + /* + * How can we block and yet be at NL? We were trying to upconvert + * from NL and got canceled. The code comes back here, and now + * we notice and clear BLOCKING. + */ + if (lockres->l_level == DLM_LOCK_NL) { + BUG_ON(lockres->l_ex_holders || lockres->l_ro_holders); + lockres->l_blocking = DLM_LOCK_NL; + lockres_clear_flags(lockres, OCFS2_LOCK_BLOCKED); + spin_unlock_irqrestore(&lockres->l_lock, flags); + goto leave; + } + /* if we're blocking an exclusive and we have *any* holders, * then requeue. */ if ((lockres->l_blocking == DLM_LOCK_EX) -- cgit v1.2.3 From db0f6ce69776370232431eb8be85a5b18b0019c0 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 1 Feb 2010 16:55:50 -0800 Subject: ocfs2: Remove overzealous BUG_ON during blocked lock processing During blocked lock processing, we should consider the possibility that the lock is no longer blocking. Joel Becker assisted in fixing this issue. Reported-by: David Teigland Signed-off-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlmglue.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index ac24f49ae2fb..ce8e061c9a22 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -3392,9 +3392,17 @@ static int ocfs2_unblock_lock(struct ocfs2_super *osb, spin_lock_irqsave(&lockres->l_lock, flags); - BUG_ON(!(lockres->l_flags & OCFS2_LOCK_BLOCKED)); - recheck: + /* + * Is it still blocking? If not, we have no more work to do. + */ + if (!(lockres->l_flags & OCFS2_LOCK_BLOCKED)) { + BUG_ON(lockres->l_blocking != DLM_LOCK_NL); + spin_unlock_irqrestore(&lockres->l_lock, flags); + ret = 0; + goto leave; + } + if (lockres->l_flags & OCFS2_LOCK_BUSY) { /* XXX * This is a *big* race. The OCFS2_LOCK_PENDING flag -- cgit v1.2.3 From e402746a945ceb9d0486a8e3d5917c9228fa4404 Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Mon, 25 Jan 2010 11:20:19 +0000 Subject: GFS2: Wait for unlock completion on umount This patch adds a wait on umount between the point at which we dispose of all glocks and the point at which we unmount the lock protocol. This ensures that we've received all the replies to our unlock requests before we stop the locking. Signed-off-by: Steven Whitehouse Reported-by: Fabio M. Di Nitto --- fs/gfs2/incore.h | 2 ++ fs/gfs2/lock_dlm.c | 7 ++++++- fs/gfs2/ops_fstype.c | 2 ++ fs/gfs2/super.c | 3 +++ 4 files changed, 13 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h index 4792200978c8..bc0ad158e6b4 100644 --- a/fs/gfs2/incore.h +++ b/fs/gfs2/incore.h @@ -544,6 +544,8 @@ struct gfs2_sbd { struct gfs2_holder sd_live_gh; struct gfs2_glock *sd_rename_gl; struct gfs2_glock *sd_trans_gl; + wait_queue_head_t sd_glock_wait; + atomic_t sd_glock_disposal; /* Inode Stuff */ diff --git a/fs/gfs2/lock_dlm.c b/fs/gfs2/lock_dlm.c index 46df988323bc..cdd0755d7823 100644 --- a/fs/gfs2/lock_dlm.c +++ b/fs/gfs2/lock_dlm.c @@ -21,6 +21,7 @@ static void gdlm_ast(void *arg) { struct gfs2_glock *gl = arg; unsigned ret = gl->gl_state; + struct gfs2_sbd *sdp = gl->gl_sbd; BUG_ON(gl->gl_lksb.sb_flags & DLM_SBF_DEMOTED); @@ -30,6 +31,8 @@ static void gdlm_ast(void *arg) switch (gl->gl_lksb.sb_status) { case -DLM_EUNLOCK: /* Unlocked, so glock can be freed */ kmem_cache_free(gfs2_glock_cachep, gl); + if (atomic_dec_and_test(&sdp->sd_glock_disposal)) + wake_up(&sdp->sd_glock_wait); return; case -DLM_ECANCEL: /* Cancel while getting lock */ ret |= LM_OUT_CANCELED; @@ -167,7 +170,8 @@ static unsigned int gdlm_lock(struct gfs2_glock *gl, static void gdlm_put_lock(struct kmem_cache *cachep, void *ptr) { struct gfs2_glock *gl = ptr; - struct lm_lockstruct *ls = &gl->gl_sbd->sd_lockstruct; + struct gfs2_sbd *sdp = gl->gl_sbd; + struct lm_lockstruct *ls = &sdp->sd_lockstruct; int error; if (gl->gl_lksb.sb_lkid == 0) { @@ -183,6 +187,7 @@ static void gdlm_put_lock(struct kmem_cache *cachep, void *ptr) (unsigned long long)gl->gl_name.ln_number, error); return; } + atomic_inc(&sdp->sd_glock_disposal); } static void gdlm_cancel(struct gfs2_glock *gl) diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c index edfee24f3636..9390fc7d8d40 100644 --- a/fs/gfs2/ops_fstype.c +++ b/fs/gfs2/ops_fstype.c @@ -82,6 +82,8 @@ static struct gfs2_sbd *init_sbd(struct super_block *sb) gfs2_tune_init(&sdp->sd_tune); + init_waitqueue_head(&sdp->sd_glock_wait); + atomic_set(&sdp->sd_glock_disposal, 0); spin_lock_init(&sdp->sd_statfs_spin); spin_lock_init(&sdp->sd_rindex_spin); diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index c282ad41f3d1..66242b32db5b 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -21,6 +21,7 @@ #include #include #include +#include #include "gfs2.h" #include "incore.h" @@ -860,6 +861,8 @@ restart: gfs2_jindex_free(sdp); /* Take apart glock structures and buffer lists */ gfs2_gl_hash_clear(sdp); + /* Wait for dlm to reply to all our unlock requests */ + wait_event(sdp->sd_glock_wait, atomic_read(&sdp->sd_glock_disposal) == 0); /* Unmount the locking protocol */ gfs2_lm_unmount(sdp); -- cgit v1.2.3 From 8f05228ee7c8f409ae3c6f9c3e13d7ccb9c18360 Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Fri, 29 Jan 2010 15:21:27 +0000 Subject: GFS2: Extend umount wait coverage to full glock lifetime Although all glocks are, by the time of the umount glock wait, scheduled for demotion, some of them haven't made it far enough through the process for the original set of waiting code to wait for them. This extends the ref count to the whole glock lifetime in order to ensure that the waiting does catch all glocks. It does make it a bit more invasive, but it seems the only sensible solution at the moment. Signed-off-by: Steven Whitehouse --- fs/gfs2/glock.c | 4 ++++ fs/gfs2/glock.h | 2 +- fs/gfs2/lock_dlm.c | 6 +++--- fs/gfs2/ops_fstype.c | 10 +++++++++- fs/gfs2/super.c | 2 -- 5 files changed, 17 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index f455a03a09e2..f42663325931 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -769,6 +769,7 @@ int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number, if (!gl) return -ENOMEM; + atomic_inc(&sdp->sd_glock_disposal); gl->gl_flags = 0; gl->gl_name = name; atomic_set(&gl->gl_ref, 1); @@ -1538,6 +1539,9 @@ void gfs2_gl_hash_clear(struct gfs2_sbd *sdp) up_write(&gfs2_umount_flush_sem); msleep(10); } + flush_workqueue(glock_workqueue); + wait_event(sdp->sd_glock_wait, atomic_read(&sdp->sd_glock_disposal) == 0); + gfs2_dump_lockstate(sdp); } void gfs2_glock_finish_truncate(struct gfs2_inode *ip) diff --git a/fs/gfs2/glock.h b/fs/gfs2/glock.h index 13f0bd228132..c0262faf4725 100644 --- a/fs/gfs2/glock.h +++ b/fs/gfs2/glock.h @@ -123,7 +123,7 @@ struct lm_lockops { int (*lm_mount) (struct gfs2_sbd *sdp, const char *fsname); void (*lm_unmount) (struct gfs2_sbd *sdp); void (*lm_withdraw) (struct gfs2_sbd *sdp); - void (*lm_put_lock) (struct kmem_cache *cachep, void *gl); + void (*lm_put_lock) (struct kmem_cache *cachep, struct gfs2_glock *gl); unsigned int (*lm_lock) (struct gfs2_glock *gl, unsigned int req_state, unsigned int flags); void (*lm_cancel) (struct gfs2_glock *gl); diff --git a/fs/gfs2/lock_dlm.c b/fs/gfs2/lock_dlm.c index cdd0755d7823..0e5e0e7022e5 100644 --- a/fs/gfs2/lock_dlm.c +++ b/fs/gfs2/lock_dlm.c @@ -167,15 +167,16 @@ static unsigned int gdlm_lock(struct gfs2_glock *gl, return LM_OUT_ASYNC; } -static void gdlm_put_lock(struct kmem_cache *cachep, void *ptr) +static void gdlm_put_lock(struct kmem_cache *cachep, struct gfs2_glock *gl) { - struct gfs2_glock *gl = ptr; struct gfs2_sbd *sdp = gl->gl_sbd; struct lm_lockstruct *ls = &sdp->sd_lockstruct; int error; if (gl->gl_lksb.sb_lkid == 0) { kmem_cache_free(cachep, gl); + if (atomic_dec_and_test(&sdp->sd_glock_disposal)) + wake_up(&sdp->sd_glock_wait); return; } @@ -187,7 +188,6 @@ static void gdlm_put_lock(struct kmem_cache *cachep, void *ptr) (unsigned long long)gl->gl_name.ln_number, error); return; } - atomic_inc(&sdp->sd_glock_disposal); } static void gdlm_cancel(struct gfs2_glock *gl) diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c index 9390fc7d8d40..8a102f731003 100644 --- a/fs/gfs2/ops_fstype.c +++ b/fs/gfs2/ops_fstype.c @@ -985,9 +985,17 @@ static const match_table_t nolock_tokens = { { Opt_err, NULL }, }; +static void nolock_put_lock(struct kmem_cache *cachep, struct gfs2_glock *gl) +{ + struct gfs2_sbd *sdp = gl->gl_sbd; + kmem_cache_free(cachep, gl); + if (atomic_dec_and_test(&sdp->sd_glock_disposal)) + wake_up(&sdp->sd_glock_wait); +} + static const struct lm_lockops nolock_ops = { .lm_proto_name = "lock_nolock", - .lm_put_lock = kmem_cache_free, + .lm_put_lock = nolock_put_lock, .lm_tokens = &nolock_tokens, }; diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index 66242b32db5b..b9dd3da22c0a 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -861,8 +861,6 @@ restart: gfs2_jindex_free(sdp); /* Take apart glock structures and buffer lists */ gfs2_gl_hash_clear(sdp); - /* Wait for dlm to reply to all our unlock requests */ - wait_event(sdp->sd_glock_wait, atomic_read(&sdp->sd_glock_disposal) == 0); /* Unmount the locking protocol */ gfs2_lm_unmount(sdp); -- cgit v1.2.3 From 9f557cd8073104b39528794d44e129331ded649f Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Feb 2010 08:27:22 -0500 Subject: NFS: Fix an Oops when truncating a file The VM/VFS does not allow mapping->a_ops->invalidatepage() to fail. Unfortunately, nfs_wb_page_cancel() may fail if a fatal signal occurs. Since the NFS code assumes that the page stays mapped for as long as the writeback is active, we can end up Oopsing (among other things). The only safe fix here is to convert nfs_wait_on_request(), so as to make it uninterruptible (as is already the case with wait_on_page_writeback()). Signed-off-by: Trond Myklebust Cc: stable@kernel.org --- fs/nfs/pagelist.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index e2975939126a..a12c45b65dd4 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -176,6 +176,12 @@ void nfs_release_request(struct nfs_page *req) kref_put(&req->wb_kref, nfs_free_request); } +static int nfs_wait_bit_uninterruptible(void *word) +{ + io_schedule(); + return 0; +} + /** * nfs_wait_on_request - Wait for a request to complete. * @req: request to wait upon. @@ -186,14 +192,9 @@ void nfs_release_request(struct nfs_page *req) int nfs_wait_on_request(struct nfs_page *req) { - int ret = 0; - - if (!test_bit(PG_BUSY, &req->wb_flags)) - goto out; - ret = out_of_line_wait_on_bit(&req->wb_flags, PG_BUSY, - nfs_wait_bit_killable, TASK_KILLABLE); -out: - return ret; + return wait_on_bit(&req->wb_flags, PG_BUSY, + nfs_wait_bit_uninterruptible, + TASK_UNINTERRUPTIBLE); } /** -- cgit v1.2.3 From 387c149b54b4321cbc790dadbd4f8eedb5a90468 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Feb 2010 08:27:35 -0500 Subject: NFS: Fix a umount race Ensure that we unregister the bdi before kill_anon_super() calls ida_remove() on our device name. Signed-off-by: Trond Myklebust Cc: stable@kernel.org --- fs/nfs/super.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/super.c b/fs/nfs/super.c index ce907efc5508..f1afee4eea77 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -243,6 +243,7 @@ static int nfs_show_stats(struct seq_file *, struct vfsmount *); static int nfs_get_sb(struct file_system_type *, int, const char *, void *, struct vfsmount *); static int nfs_xdev_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name, void *raw_data, struct vfsmount *mnt); +static void nfs_put_super(struct super_block *); static void nfs_kill_super(struct super_block *); static int nfs_remount(struct super_block *sb, int *flags, char *raw_data); @@ -266,6 +267,7 @@ static const struct super_operations nfs_sops = { .alloc_inode = nfs_alloc_inode, .destroy_inode = nfs_destroy_inode, .write_inode = nfs_write_inode, + .put_super = nfs_put_super, .statfs = nfs_statfs, .clear_inode = nfs_clear_inode, .umount_begin = nfs_umount_begin, @@ -335,6 +337,7 @@ static const struct super_operations nfs4_sops = { .alloc_inode = nfs_alloc_inode, .destroy_inode = nfs_destroy_inode, .write_inode = nfs_write_inode, + .put_super = nfs_put_super, .statfs = nfs_statfs, .clear_inode = nfs4_clear_inode, .umount_begin = nfs_umount_begin, @@ -2257,6 +2260,17 @@ error_splat_super: goto out; } +/* + * Ensure that we unregister the bdi before kill_anon_super + * releases the device name + */ +static void nfs_put_super(struct super_block *s) +{ + struct nfs_server *server = NFS_SB(s); + + bdi_unregister(&server->backing_dev_info); +} + /* * Destroy an NFS2/3 superblock */ @@ -2265,7 +2279,6 @@ static void nfs_kill_super(struct super_block *s) struct nfs_server *server = NFS_SB(s); kill_anon_super(s); - bdi_unregister(&server->backing_dev_info); nfs_fscache_release_super_cookie(s); nfs_free_server(server); } -- cgit v1.2.3 From 9b4b351346b41d923d69adec865814fdaac4dba9 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Feb 2010 08:27:35 -0500 Subject: NFS: Don't clobber the attribute type in nfs_update_inode() If the NFS_ATTR_FATTR_TYPE field isn't set in fattr->valid, then we should not set the S_IFMT part of inode->i_mode. Reported-by: Al Viro Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index faa091865ad0..f141bde7756a 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -1261,8 +1261,10 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr) if (fattr->valid & NFS_ATTR_FATTR_MODE) { if ((inode->i_mode & S_IALLUGO) != (fattr->mode & S_IALLUGO)) { + umode_t newmode = inode->i_mode & S_IFMT; + newmode |= fattr->mode & S_IALLUGO; + inode->i_mode = newmode; invalid |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ACCESS|NFS_INO_INVALID_ACL; - inode->i_mode = fattr->mode; } } else if (server->caps & NFS_CAP_MODE) invalid |= save_cache_validity & (NFS_INO_INVALID_ATTR -- cgit v1.2.3 From 079b805782f94f4b278132286a8c9bc4655d1c51 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Wed, 3 Feb 2010 10:16:54 -0800 Subject: ocfs2: Plugs race between the dc thread and an unlock ast message This patch plugs a race between the downconvert thread and an unlock ast message. Specifically, after the downconvert worker has done its task, the dc thread needs to check whether an unlock ast made the downconvert moot. Reported-by: David Teigland Signed-off-by: Sunil Mushran Acked-by: Mark Fasheh Signed-off-by: Joel Becker --- fs/ocfs2/dlmglue.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index ce8e061c9a22..e044019cb3b1 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -3384,6 +3384,7 @@ static int ocfs2_unblock_lock(struct ocfs2_super *osb, unsigned long flags; int blocking; int new_level; + int level; int ret = 0; int set_lvb = 0; unsigned int gen; @@ -3503,6 +3504,7 @@ recheck: * may sleep, so we save off a copy of what we're blocking as * it may change while we're not holding the spin lock. */ blocking = lockres->l_blocking; + level = lockres->l_level; spin_unlock_irqrestore(&lockres->l_lock, flags); ctl->unblock_action = lockres->l_ops->downconvert_worker(lockres, blocking); @@ -3511,7 +3513,7 @@ recheck: goto leave; spin_lock_irqsave(&lockres->l_lock, flags); - if (blocking != lockres->l_blocking) { + if ((blocking != lockres->l_blocking) || (level != lockres->l_level)) { /* If this changed underneath us, then we can't drop * it just yet. */ goto recheck; -- cgit v1.2.3 From cda70ba8c05a8661f882862c4699a31d215ab151 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 1 Feb 2010 17:34:58 -0800 Subject: ocfs2/dlm: Remove BUG_ON in dlm recovery when freeing locks of a dead node During recovery, the dlm frees the locks for the dead node. If it finds a lock in a resource for the dead node, it expects that node to also have a ref in that lock resource. If not, it BUGs. ossbz#1175 was filed with the above BUG. Now, while it is correct that we should be expecting the ref, I see no reason why we have to BUG. After all, we are freeing up the lock and clearing the ref. This patch replaces the BUG_ON with a printk(). Hopefully, that will give us more clues next time this happens. http://oss.oracle.com/bugzilla/show_bug.cgi?id=1175 Signed-off-by: Sunil Mushran Acked-by: Mark Fasheh Signed-off-by: Joel Becker --- fs/ocfs2/dlm/dlmrecovery.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index ad712211d4ea..344bcf90cbf4 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -2243,7 +2243,12 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm, mlog(0, "%s:%.*s: freed %u locks for dead node %u, " "dropping ref from lockres\n", dlm->name, res->lockname.len, res->lockname.name, freed, dead_node); - BUG_ON(!test_bit(dead_node, res->refmap)); + if(!test_bit(dead_node, res->refmap)) { + mlog(ML_ERROR, "%s:%.*s: freed %u locks for dead node %u, " + "but ref was not set\n", dlm->name, + res->lockname.len, res->lockname.name, freed, dead_node); + __dlm_print_one_lock_resource(res); + } dlm_lockres_clear_refmap_bit(dead_node, res); } else if (test_bit(dead_node, res->refmap)) { mlog(0, "%s:%.*s: dead node %u had a ref, but had " -- cgit v1.2.3 From f044ba7835b84e69c68b620ca8fa27e5ef67759d Mon Sep 17 00:00:00 2001 From: "Yan, Zheng" Date: Thu, 4 Feb 2010 08:46:56 +0000 Subject: Btrfs: fix race between allocate and release extent buffer. Increase extent buffer's reference count while holding the lock. Otherwise it can race with try_release_extent_buffer. Signed-off-by: Yan Zheng Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 96577e8bf9fd..b177ed319612 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3165,10 +3165,9 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, spin_unlock(&tree->buffer_lock); goto free_eb; } - spin_unlock(&tree->buffer_lock); - /* add one reference for the tree */ atomic_inc(&eb->refs); + spin_unlock(&tree->buffer_lock); return eb; free_eb: -- cgit v1.2.3 From 014e4ac4f7d9c981750491fa40ea35efadc9ed49 Mon Sep 17 00:00:00 2001 From: Roel Kluin Date: Fri, 29 Jan 2010 10:42:11 +0000 Subject: Btrfs: make error return negative in btrfs_sync_file() It appears the error return should be negative Signed-off-by: Roel Kluin Signed-off-by: Chris Mason --- fs/btrfs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index ae96fdae1f7d..413a30dafcda 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1133,7 +1133,7 @@ int btrfs_sync_file(struct file *file, struct dentry *dentry, int datasync) } mutex_lock(&dentry->d_inode->i_mutex); out: - return ret > 0 ? EIO : ret; + return ret > 0 ? -EIO : ret; } static const struct vm_operations_struct btrfs_file_vm_ops = { -- cgit v1.2.3 From d7ce5843bb28ada6845ab2ae8510ba3f12d33154 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 2 Feb 2010 08:46:44 +0000 Subject: Btrfs: remove BUG_ON() due to mounting bad filesystem Mounting a bad filesystem caused a BUG_ON(). The following is steps to reproduce it. # mkfs.btrfs /dev/sda2 # mount /dev/sda2 /mnt # mkfs.btrfs /dev/sda1 /dev/sda2 (the program says that /dev/sda2 was mounted, and then exits. ) # umount /mnt # mount /dev/sda1 /mnt At the third step, mkfs.btrfs exited in the way of make filesystem. So the initialization of the filesystem didn't finish. So the filesystem was bad, and it caused BUG_ON() when mounting it. But BUG_ON() should be called by the wrong code, not user's operation, so I think it is a bug of btrfs. This patch fixes it. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 7 ++++++- fs/btrfs/relocation.c | 3 ++- 2 files changed, 8 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 87b25543d7d1..2b59201b955c 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1982,7 +1982,12 @@ struct btrfs_root *open_ctree(struct super_block *sb, if (!(sb->s_flags & MS_RDONLY)) { ret = btrfs_recover_relocation(tree_root); - BUG_ON(ret); + if (ret < 0) { + printk(KERN_WARNING + "btrfs: failed to recover relocation\n"); + err = -EINVAL; + goto fail_trans_kthread; + } } location.objectid = BTRFS_FS_TREE_OBJECTID; diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index ed3e4a2ec2c8..ab7ab5318745 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -3764,7 +3764,8 @@ out: BTRFS_DATA_RELOC_TREE_OBJECTID); if (IS_ERR(fs_root)) err = PTR_ERR(fs_root); - btrfs_orphan_cleanup(fs_root); + else + btrfs_orphan_cleanup(fs_root); } return err; } -- cgit v1.2.3 From 7a7965f83e89f0be506a96769938a721e4e5ae50 Mon Sep 17 00:00:00 2001 From: "Yan, Zheng" Date: Mon, 1 Feb 2010 02:41:17 +0000 Subject: Btrfs: Fix oopsen when dropping empty tree. When dropping a empty tree, walk_down_tree() skips checking extent information for the tree root. This will triggers a BUG_ON in walk_up_proc(). Signed-off-by: Yan Zheng Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 432a2da4641e..559f72489b3b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5402,10 +5402,6 @@ static noinline int walk_down_tree(struct btrfs_trans_handle *trans, int ret; while (level >= 0) { - if (path->slots[level] >= - btrfs_header_nritems(path->nodes[level])) - break; - ret = walk_down_proc(trans, root, path, wc, lookup_info); if (ret > 0) break; @@ -5413,6 +5409,10 @@ static noinline int walk_down_tree(struct btrfs_trans_handle *trans, if (level == 0) break; + if (path->slots[level] >= + btrfs_header_nritems(path->nodes[level])) + break; + ret = do_walk_down(trans, root, path, wc, &lookup_info); if (ret > 0) { path->slots[level]++; -- cgit v1.2.3 From efd049fb26a162c3830fd3cb1001fdc09b147f3b Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 2 Feb 2010 20:50:10 +0000 Subject: Btrfs: do not try and lookup the file extent when finishing ordered io When running the following fio job [torrent] filename=torrent-test rw=randwrite size=4g filesize=4g bs=4k ioengine=sync you would see long stalls where no work was being done. That is because we were doing all this extra work to read in the file extent outside of the transaction, however in the random io case this ends up hurting us because the file extents are not there to begin with. So axe this logic, since we end up reading in the file extent when we go to update it anyway. This took the fio job from 11 mb/s with several ~10 second stalls to 24 mb/s to a couple of 1-2 second stalls. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 46 ++-------------------------------------------- 1 file changed, 2 insertions(+), 44 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 8cd109972fa6..6782aa19130d 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -1681,24 +1681,6 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans, * before we start the transaction. It limits the amount of btree * reads required while inside the transaction. */ -static noinline void reada_csum(struct btrfs_root *root, - struct btrfs_path *path, - struct btrfs_ordered_extent *ordered_extent) -{ - struct btrfs_ordered_sum *sum; - u64 bytenr; - - sum = list_entry(ordered_extent->list.next, struct btrfs_ordered_sum, - list); - bytenr = sum->sums[0].bytenr; - - /* - * we don't care about the results, the point of this search is - * just to get the btree leaves into ram - */ - btrfs_lookup_csum(NULL, root->fs_info->csum_root, path, bytenr, 0); -} - /* as ordered data IO finishes, this gets called so we can finish * an ordered extent if the range of bytes in the file it covers are * fully written. @@ -1709,7 +1691,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) struct btrfs_trans_handle *trans; struct btrfs_ordered_extent *ordered_extent = NULL; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; - struct btrfs_path *path; int compressed = 0; int ret; @@ -1717,32 +1698,9 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) if (!ret) return 0; - /* - * before we join the transaction, try to do some of our IO. - * This will limit the amount of IO that we have to do with - * the transaction running. We're unlikely to need to do any - * IO if the file extents are new, the disk_i_size checks - * covers the most common case. - */ - if (start < BTRFS_I(inode)->disk_i_size) { - path = btrfs_alloc_path(); - if (path) { - ret = btrfs_lookup_file_extent(NULL, root, path, - inode->i_ino, - start, 0); - ordered_extent = btrfs_lookup_ordered_extent(inode, - start); - if (!list_empty(&ordered_extent->list)) { - btrfs_release_path(root, path); - reada_csum(root, path, ordered_extent); - } - btrfs_free_path(path); - } - } - - if (!ordered_extent) - ordered_extent = btrfs_lookup_ordered_extent(inode, start); + ordered_extent = btrfs_lookup_ordered_extent(inode, start); BUG_ON(!ordered_extent); + if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { BUG_ON(!list_empty(&ordered_extent->list)); ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent); -- cgit v1.2.3 From 23b5c50945f2294add0137799400329c0ebba290 Mon Sep 17 00:00:00 2001 From: "Aneesh Kumar K.V" Date: Thu, 4 Feb 2010 11:33:03 -0500 Subject: Btrfs: apply updated fallocate i_size fix This version of the i_size fix for fallocate makes sure we only update the i_size when the current fallocate is really operating outside of i_size. Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 6782aa19130d..4deb280f8969 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -5799,7 +5799,9 @@ static int prealloc_file_range(struct inode *inode, u64 start, u64 end, inode->i_ctime = CURRENT_TIME; BTRFS_I(inode)->flags |= BTRFS_INODE_PREALLOC; if (!(mode & FALLOC_FL_KEEP_SIZE) && - cur_offset > inode->i_size) { + (actual_len > inode->i_size) && + (cur_offset > inode->i_size)) { + if (cur_offset > actual_len) i_size = actual_len; else -- cgit v1.2.3 From bd6b0bf87d8cf3d9cfeadeb12dbf5449e3e50765 Mon Sep 17 00:00:00 2001 From: Roel Kluin Date: Fri, 5 Feb 2010 10:26:27 +0100 Subject: ocfs2: Fix contiguousness check in ocfs2_try_to_merge_extent_map() The wrong member was compared in the continguousness check. Acked-by: Tao Ma Signed-off-by: Roel Kluin Signed-off-by: Joel Becker --- fs/ocfs2/extent_map.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c index d35a27f4523e..5328529e7fd2 100644 --- a/fs/ocfs2/extent_map.c +++ b/fs/ocfs2/extent_map.c @@ -192,7 +192,7 @@ static int ocfs2_try_to_merge_extent_map(struct ocfs2_extent_map_item *emi, emi->ei_clusters += ins->ei_clusters; return 1; } else if ((ins->ei_phys + ins->ei_clusters) == emi->ei_phys && - (ins->ei_cpos + ins->ei_clusters) == emi->ei_phys && + (ins->ei_cpos + ins->ei_clusters) == emi->ei_cpos && ins->ei_flags == emi->ei_flags) { emi->ei_phys = ins->ei_phys; emi->ei_cpos = ins->ei_cpos; -- cgit v1.2.3 From f12f98dba6ea1517cd7fbb912208893b9c014c15 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 5 Feb 2010 13:14:00 -0500 Subject: cifs: fix length calculation for converted unicode readdir names cifs_from_ucs2 returns the length of the converted name, including the length of the NULL terminator. We don't want to include the NULL terminator in the dentry name length however since that'll throw off the hash calculation for the dentry cache. I believe that this is the root cause of several problems that have cropped up recently that seem to be papered over with the "noserverino" mount option. More confirmation of that would be good, but this is clearly a bug and it fixes at least one reproducible problem that was reported. This patch fixes at least this reproducer in this kernel.org bug: http://bugzilla.kernel.org/show_bug.cgi?id=15088#c12 Reported-by: Bjorn Tore Sund Acked-by: Dave Kleikamp Signed-off-by: Jeff Layton Cc: stable@kernel.org Signed-off-by: Steve French --- fs/cifs/readdir.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c index f84062f9a985..f5618f8cc462 100644 --- a/fs/cifs/readdir.c +++ b/fs/cifs/readdir.c @@ -666,6 +666,7 @@ static int cifs_get_name_from_search_buf(struct qstr *pqst, min(len, max_len), nlt, cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SPECIAL_CHR); + pqst->len -= nls_nullsize(nlt); } else { pqst->name = filename; pqst->len = len; -- cgit v1.2.3 From 301a6a317797ca362951ea21da397c05236f0070 Mon Sep 17 00:00:00 2001 From: Steve French Date: Sat, 6 Feb 2010 07:08:53 +0000 Subject: [CIFS] Maximum username length check in session setup does not match Fix length check reported by D. Binderman (see below) d binderman wrote: > > I just ran the sourceforge tool cppcheck over the source code of the > new Linux kernel 2.6.33-rc6 > > It said > > [./cifs/sess.c:250]: (error) Buffer access out-of-bounds May turn out to be harmless, but best to be safe. Note max username length is defined to 32 due to Linux (Windows maximum is 20). Signed-off-by: Steve French --- fs/cifs/sess.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/cifs/sess.c b/fs/cifs/sess.c index 7085a6275c4c..aaa9c1c5a5bd 100644 --- a/fs/cifs/sess.c +++ b/fs/cifs/sess.c @@ -223,9 +223,9 @@ static void unicode_ssetup_strings(char **pbcc_area, struct cifsSesInfo *ses, /* null user mount */ *bcc_ptr = 0; *(bcc_ptr+1) = 0; - } else { /* 300 should be long enough for any conceivable user name */ + } else { bytes_ret = cifs_strtoUCS((__le16 *) bcc_ptr, ses->userName, - 300, nls_cp); + MAX_USERNAME_SIZE, nls_cp); } bcc_ptr += 2 * bytes_ret; bcc_ptr += 2; /* account for null termination */ @@ -246,11 +246,10 @@ static void ascii_ssetup_strings(char **pbcc_area, struct cifsSesInfo *ses, /* copy user */ if (ses->userName == NULL) { /* BB what about null user mounts - check that we do this BB */ - } else { /* 300 should be long enough for any conceivable user name */ - strncpy(bcc_ptr, ses->userName, 300); + } else { + strncpy(bcc_ptr, ses->userName, MAX_USERNAME_SIZE); } - /* BB improve check for overflow */ - bcc_ptr += strnlen(ses->userName, 300); + bcc_ptr += strnlen(ses->userName, MAX_USERNAME_SIZE); *bcc_ptr = 0; bcc_ptr++; /* account for null termination */ -- cgit v1.2.3 From 8dd5ca532c2d2c2b85f16bc038ebfff05b8853e1 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Thu, 28 Jan 2010 22:11:38 -0500 Subject: befs: fix leak Signed-off-by: Al Viro --- fs/befs/linuxvfs.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/befs/linuxvfs.c b/fs/befs/linuxvfs.c index 33baf27fac78..34ddda888e63 100644 --- a/fs/befs/linuxvfs.c +++ b/fs/befs/linuxvfs.c @@ -873,6 +873,7 @@ befs_fill_super(struct super_block *sb, void *data, int silent) brelse(bh); unacquire_priv_sbp: + kfree(befs_sb->mount_opts.iocharset); kfree(sb->s_fs_info); unacquire_none: -- cgit v1.2.3 From 4b06e5b9ad8abb20105b2b25e42c509ebe9b2d76 Mon Sep 17 00:00:00 2001 From: Jun'ichi Nomura Date: Fri, 29 Jan 2010 09:56:22 +0900 Subject: freeze_bdev: don't deactivate successfully frozen MS_RDONLY sb Thanks Thomas and Christoph for testing and review. I removed 'smp_wmb()' before up_write from the previous patch, since up_write() should have necessary ordering constraints. (I.e. the change of s_frozen is visible to others after up_write) I'm quite sure the change is harmless but if you are uncomfortable with Tested-by/Reviewed-by on the modified patch, please remove them. If MS_RDONLY, freeze_bdev should just up_write(s_umount) instead of deactivate_locked_super(). Also, keep sb->s_frozen consistent so that remount can check the frozen state. Otherwise a crash reported here can happen: http://lkml.org/lkml/2010/1/16/37 http://lkml.org/lkml/2010/1/28/53 This patch should be applied for 2.6.32 stable series, too. Reviewed-by: Christoph Hellwig Tested-by: Thomas Backlund Signed-off-by: Jun'ichi Nomura Cc: stable@kernel.org Signed-off-by: Al Viro --- fs/block_dev.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/block_dev.c b/fs/block_dev.c index 73d6a735b8f3..d11d0289f3d2 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -246,7 +246,8 @@ struct super_block *freeze_bdev(struct block_device *bdev) if (!sb) goto out; if (sb->s_flags & MS_RDONLY) { - deactivate_locked_super(sb); + sb->s_frozen = SB_FREEZE_TRANS; + up_write(&sb->s_umount); mutex_unlock(&bdev->bd_fsfreeze_mutex); return sb; } @@ -307,7 +308,7 @@ int thaw_bdev(struct block_device *bdev, struct super_block *sb) BUG_ON(sb->s_bdev != bdev); down_write(&sb->s_umount); if (sb->s_flags & MS_RDONLY) - goto out_deactivate; + goto out_unfrozen; if (sb->s_op->unfreeze_fs) { error = sb->s_op->unfreeze_fs(sb); @@ -321,11 +322,11 @@ int thaw_bdev(struct block_device *bdev, struct super_block *sb) } } +out_unfrozen: sb->s_frozen = SB_UNFROZEN; smp_wmb(); wake_up(&sb->s_wait_unfrozen); -out_deactivate: if (sb) deactivate_locked_super(sb); out_unlock: -- cgit v1.2.3 From 1e41568d7378d1ba8c64ba137b9ddd00b59f893a Mon Sep 17 00:00:00 2001 From: Al Viro Date: Tue, 26 Jan 2010 05:43:08 -0500 Subject: Take ima_path_check() in nfsd past dentry_open() in nfsd_open() Signed-off-by: Al Viro --- fs/nfsd/vfs.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c index c194793b642b..325959e264ce 100644 --- a/fs/nfsd/vfs.c +++ b/fs/nfsd/vfs.c @@ -752,6 +752,8 @@ nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, flags, current_cred()); if (IS_ERR(*filp)) host_err = PTR_ERR(*filp); + host_err = ima_path_check(&(*filp)->f_path, + access & (MAY_READ | MAY_WRITE | MAY_EXEC)); out_nfserr: err = nfserrno(host_err); out: @@ -2127,7 +2129,6 @@ nfsd_permission(struct svc_rqst *rqstp, struct svc_export *exp, */ path.mnt = exp->ex_path.mnt; path.dentry = dentry; - err = ima_path_check(&path, acc & (MAY_READ | MAY_WRITE | MAY_EXEC)); nfsd_out: return err? nfserrno(err) : 0; } -- cgit v1.2.3 From 8eb988c70e7709b7bd1a69f0ec53d19ac20dea84 Mon Sep 17 00:00:00 2001 From: Mimi Zohar Date: Wed, 20 Jan 2010 15:35:41 -0500 Subject: fix ima breakage The "Untangling ima mess, part 2 with counters" patch messed up the counters. Based on conversations with Al Viro, this patch streamlines ima_path_check() by removing the counter maintaince. The counters are now updated independently, from measuring the file, in __dentry_open() and alloc_file() by calling ima_counts_get(). ima_path_check() is called from nfsd and do_filp_open(). It also did not measure all files that should have been measured. Reason: ima_path_check() got bogus value passed as mask. [AV: mea culpa] [AV: add missing nfsd bits] Signed-off-by: Mimi Zohar Signed-off-by: Al Viro --- fs/namei.c | 6 +- fs/nfsd/vfs.c | 3 +- include/linux/ima.h | 4 +- security/integrity/ima/ima_main.c | 236 +++++++++++++++----------------------- 4 files changed, 97 insertions(+), 152 deletions(-) (limited to 'fs') diff --git a/fs/namei.c b/fs/namei.c index 94a5e60779f9..cd77b6375efd 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -1736,8 +1736,7 @@ do_last: if (nd.root.mnt) path_put(&nd.root); if (!IS_ERR(filp)) { - error = ima_path_check(&filp->f_path, filp->f_mode & - (MAY_READ | MAY_WRITE | MAY_EXEC)); + error = ima_path_check(filp, acc_mode); if (error) { fput(filp); filp = ERR_PTR(error); @@ -1797,8 +1796,7 @@ ok: } filp = nameidata_to_filp(&nd); if (!IS_ERR(filp)) { - error = ima_path_check(&filp->f_path, filp->f_mode & - (MAY_READ | MAY_WRITE | MAY_EXEC)); + error = ima_path_check(filp, acc_mode); if (error) { fput(filp); filp = ERR_PTR(error); diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c index 325959e264ce..32477e3a645c 100644 --- a/fs/nfsd/vfs.c +++ b/fs/nfsd/vfs.c @@ -752,8 +752,7 @@ nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, flags, current_cred()); if (IS_ERR(*filp)) host_err = PTR_ERR(*filp); - host_err = ima_path_check(&(*filp)->f_path, - access & (MAY_READ | MAY_WRITE | MAY_EXEC)); + host_err = ima_path_check(*filp, access); out_nfserr: err = nfserrno(host_err); out: diff --git a/include/linux/ima.h b/include/linux/ima.h index 99dc6d5cf7e5..aa55a8f1f5b9 100644 --- a/include/linux/ima.h +++ b/include/linux/ima.h @@ -17,7 +17,7 @@ struct linux_binprm; extern int ima_bprm_check(struct linux_binprm *bprm); extern int ima_inode_alloc(struct inode *inode); extern void ima_inode_free(struct inode *inode); -extern int ima_path_check(struct path *path, int mask); +extern int ima_path_check(struct file *file, int mask); extern void ima_file_free(struct file *file); extern int ima_file_mmap(struct file *file, unsigned long prot); extern void ima_counts_get(struct file *file); @@ -38,7 +38,7 @@ static inline void ima_inode_free(struct inode *inode) return; } -static inline int ima_path_check(struct path *path, int mask) +static inline int ima_path_check(struct file *file, int mask) { return 0; } diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c index a89f44d5e030..75aee18f6163 100644 --- a/security/integrity/ima/ima_main.c +++ b/security/integrity/ima/ima_main.c @@ -84,6 +84,36 @@ out: return found; } +/* ima_read_write_check - reflect possible reading/writing errors in the PCR. + * + * When opening a file for read, if the file is already open for write, + * the file could change, resulting in a file measurement error. + * + * Opening a file for write, if the file is already open for read, results + * in a time of measure, time of use (ToMToU) error. + * + * In either case invalidate the PCR. + */ +enum iint_pcr_error { TOMTOU, OPEN_WRITERS }; +static void ima_read_write_check(enum iint_pcr_error error, + struct ima_iint_cache *iint, + struct inode *inode, + const unsigned char *filename) +{ + switch (error) { + case TOMTOU: + if (iint->readcount > 0) + ima_add_violation(inode, filename, "invalid_pcr", + "ToMToU"); + break; + case OPEN_WRITERS: + if (iint->writecount > 0) + ima_add_violation(inode, filename, "invalid_pcr", + "open_writers"); + break; + } +} + /* * Update the counts given an fmode_t */ @@ -98,6 +128,47 @@ static void ima_inc_counts(struct ima_iint_cache *iint, fmode_t mode) iint->writecount++; } +/* + * ima_counts_get - increment file counts + * + * Maintain read/write counters for all files, but only + * invalidate the PCR for measured files: + * - Opening a file for write when already open for read, + * results in a time of measure, time of use (ToMToU) error. + * - Opening a file for read when already open for write, + * could result in a file measurement error. + * + */ +void ima_counts_get(struct file *file) +{ + struct dentry *dentry = file->f_path.dentry; + struct inode *inode = dentry->d_inode; + fmode_t mode = file->f_mode; + struct ima_iint_cache *iint; + int rc; + + if (!ima_initialized || !S_ISREG(inode->i_mode)) + return; + iint = ima_iint_find_get(inode); + if (!iint) + return; + mutex_lock(&iint->mutex); + rc = ima_must_measure(iint, inode, MAY_READ, PATH_CHECK); + if (rc < 0) + goto out; + + if (mode & FMODE_WRITE) { + ima_read_write_check(TOMTOU, iint, inode, dentry->d_name.name); + goto out; + } + ima_read_write_check(OPEN_WRITERS, iint, inode, dentry->d_name.name); +out: + ima_inc_counts(iint, file->f_mode); + mutex_unlock(&iint->mutex); + + kref_put(&iint->refcount, iint_free); +} + /* * Decrement ima counts */ @@ -153,123 +224,6 @@ void ima_file_free(struct file *file) kref_put(&iint->refcount, iint_free); } -/* ima_read_write_check - reflect possible reading/writing errors in the PCR. - * - * When opening a file for read, if the file is already open for write, - * the file could change, resulting in a file measurement error. - * - * Opening a file for write, if the file is already open for read, results - * in a time of measure, time of use (ToMToU) error. - * - * In either case invalidate the PCR. - */ -enum iint_pcr_error { TOMTOU, OPEN_WRITERS }; -static void ima_read_write_check(enum iint_pcr_error error, - struct ima_iint_cache *iint, - struct inode *inode, - const unsigned char *filename) -{ - switch (error) { - case TOMTOU: - if (iint->readcount > 0) - ima_add_violation(inode, filename, "invalid_pcr", - "ToMToU"); - break; - case OPEN_WRITERS: - if (iint->writecount > 0) - ima_add_violation(inode, filename, "invalid_pcr", - "open_writers"); - break; - } -} - -static int get_path_measurement(struct ima_iint_cache *iint, struct file *file, - const unsigned char *filename) -{ - int rc = 0; - - ima_inc_counts(iint, file->f_mode); - - rc = ima_collect_measurement(iint, file); - if (!rc) - ima_store_measurement(iint, file, filename); - return rc; -} - -/** - * ima_path_check - based on policy, collect/store measurement. - * @path: contains a pointer to the path to be measured - * @mask: contains MAY_READ, MAY_WRITE or MAY_EXECUTE - * - * Measure the file being open for readonly, based on the - * ima_must_measure() policy decision. - * - * Keep read/write counters for all files, but only - * invalidate the PCR for measured files: - * - Opening a file for write when already open for read, - * results in a time of measure, time of use (ToMToU) error. - * - Opening a file for read when already open for write, - * could result in a file measurement error. - * - * Always return 0 and audit dentry_open failures. - * (Return code will be based upon measurement appraisal.) - */ -int ima_path_check(struct path *path, int mask) -{ - struct inode *inode = path->dentry->d_inode; - struct ima_iint_cache *iint; - struct file *file = NULL; - int rc; - - if (!ima_initialized || !S_ISREG(inode->i_mode)) - return 0; - iint = ima_iint_find_get(inode); - if (!iint) - return 0; - - mutex_lock(&iint->mutex); - - rc = ima_must_measure(iint, inode, MAY_READ, PATH_CHECK); - if (rc < 0) - goto out; - - if ((mask & MAY_WRITE) || (mask == 0)) - ima_read_write_check(TOMTOU, iint, inode, - path->dentry->d_name.name); - - if ((mask & (MAY_WRITE | MAY_READ | MAY_EXEC)) != MAY_READ) - goto out; - - ima_read_write_check(OPEN_WRITERS, iint, inode, - path->dentry->d_name.name); - if (!(iint->flags & IMA_MEASURED)) { - struct dentry *dentry = dget(path->dentry); - struct vfsmount *mnt = mntget(path->mnt); - - file = dentry_open(dentry, mnt, O_RDONLY | O_LARGEFILE, - current_cred()); - if (IS_ERR(file)) { - int audit_info = 0; - - integrity_audit_msg(AUDIT_INTEGRITY_PCR, inode, - dentry->d_name.name, - "add_measurement", - "dentry_open failed", - 1, audit_info); - file = NULL; - goto out; - } - rc = get_path_measurement(iint, file, dentry->d_name.name); - } -out: - mutex_unlock(&iint->mutex); - if (file) - fput(file); - kref_put(&iint->refcount, iint_free); - return 0; -} -EXPORT_SYMBOL_GPL(ima_path_check); - static int process_measurement(struct file *file, const unsigned char *filename, int mask, int function) { @@ -297,33 +251,6 @@ out: return rc; } -/* - * ima_counts_get - increment file counts - * - * - for IPC shm and shmat file. - * - for nfsd exported files. - * - * Increment the counts for these files to prevent unnecessary - * imbalance messages. - */ -void ima_counts_get(struct file *file) -{ - struct inode *inode = file->f_dentry->d_inode; - struct ima_iint_cache *iint; - - if (!ima_initialized || !S_ISREG(inode->i_mode)) - return; - iint = ima_iint_find_get(inode); - if (!iint) - return; - mutex_lock(&iint->mutex); - ima_inc_counts(iint, file->f_mode); - mutex_unlock(&iint->mutex); - - kref_put(&iint->refcount, iint_free); -} -EXPORT_SYMBOL_GPL(ima_counts_get); - /** * ima_file_mmap - based on policy, collect/store measurement. * @file: pointer to the file to be measured (May be NULL) @@ -369,6 +296,27 @@ int ima_bprm_check(struct linux_binprm *bprm) return 0; } +/** + * ima_path_check - based on policy, collect/store measurement. + * @file: pointer to the file to be measured + * @mask: contains MAY_READ, MAY_WRITE or MAY_EXECUTE + * + * Measure files based on the ima_must_measure() policy decision. + * + * Always return 0 and audit dentry_open failures. + * (Return code will be based upon measurement appraisal.) + */ +int ima_path_check(struct file *file, int mask) +{ + int rc; + + rc = process_measurement(file, file->f_dentry->d_name.name, + mask & (MAY_READ | MAY_WRITE | MAY_EXEC), + PATH_CHECK); + return 0; +} +EXPORT_SYMBOL_GPL(ima_path_check); + static int __init init_ima(void) { int error; -- cgit v1.2.3 From 9bbb6cad0173e6220f3ac609e26beb48dab3b7cd Mon Sep 17 00:00:00 2001 From: Mimi Zohar Date: Tue, 26 Jan 2010 17:02:40 -0500 Subject: ima: rename ima_path_check to ima_file_check ima_path_check actually deals with files! call it ima_file_check instead. Signed-off-by: Eric Paris Acked-by: Mimi Zohar Signed-off-by: Al Viro --- fs/namei.c | 4 ++-- fs/nfsd/vfs.c | 2 +- include/linux/ima.h | 4 ++-- security/integrity/ima/ima_main.c | 6 +++--- 4 files changed, 8 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/namei.c b/fs/namei.c index cd77b6375efd..d62fdc875f22 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -1736,7 +1736,7 @@ do_last: if (nd.root.mnt) path_put(&nd.root); if (!IS_ERR(filp)) { - error = ima_path_check(filp, acc_mode); + error = ima_file_check(filp, acc_mode); if (error) { fput(filp); filp = ERR_PTR(error); @@ -1796,7 +1796,7 @@ ok: } filp = nameidata_to_filp(&nd); if (!IS_ERR(filp)) { - error = ima_path_check(filp, acc_mode); + error = ima_file_check(filp, acc_mode); if (error) { fput(filp); filp = ERR_PTR(error); diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c index 32477e3a645c..97d79eff6b7f 100644 --- a/fs/nfsd/vfs.c +++ b/fs/nfsd/vfs.c @@ -752,7 +752,7 @@ nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, int type, flags, current_cred()); if (IS_ERR(*filp)) host_err = PTR_ERR(*filp); - host_err = ima_path_check(*filp, access); + host_err = ima_file_check(*filp, access); out_nfserr: err = nfserrno(host_err); out: diff --git a/include/linux/ima.h b/include/linux/ima.h index aa55a8f1f5b9..975837e7d6c0 100644 --- a/include/linux/ima.h +++ b/include/linux/ima.h @@ -17,7 +17,7 @@ struct linux_binprm; extern int ima_bprm_check(struct linux_binprm *bprm); extern int ima_inode_alloc(struct inode *inode); extern void ima_inode_free(struct inode *inode); -extern int ima_path_check(struct file *file, int mask); +extern int ima_file_check(struct file *file, int mask); extern void ima_file_free(struct file *file); extern int ima_file_mmap(struct file *file, unsigned long prot); extern void ima_counts_get(struct file *file); @@ -38,7 +38,7 @@ static inline void ima_inode_free(struct inode *inode) return; } -static inline int ima_path_check(struct file *file, int mask) +static inline int ima_file_check(struct file *file, int mask) { return 0; } diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c index eb1cf6498cc9..b76e1f03ea2b 100644 --- a/security/integrity/ima/ima_main.c +++ b/security/integrity/ima/ima_main.c @@ -14,7 +14,7 @@ * * File: ima_main.c * implements the IMA hooks: ima_bprm_check, ima_file_mmap, - * and ima_path_check. + * and ima_file_check. */ #include #include @@ -306,7 +306,7 @@ int ima_bprm_check(struct linux_binprm *bprm) * Always return 0 and audit dentry_open failures. * (Return code will be based upon measurement appraisal.) */ -int ima_path_check(struct file *file, int mask) +int ima_file_check(struct file *file, int mask) { int rc; @@ -315,7 +315,7 @@ int ima_path_check(struct file *file, int mask) PATH_CHECK); return 0; } -EXPORT_SYMBOL_GPL(ima_path_check); +EXPORT_SYMBOL_GPL(ima_file_check); static int __init init_ima(void) { -- cgit v1.2.3 From 89068c576bf324ef6fbd50dfc745148f7def202c Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 7 Feb 2010 03:07:29 -0500 Subject: Take ima_file_free() to proper place. Hooks: Just Say No. Signed-off-by: Al Viro --- fs/file_table.c | 1 + security/security.c | 2 -- 2 files changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/file_table.c b/fs/file_table.c index 69652c5bd5f0..b98404b54383 100644 --- a/fs/file_table.c +++ b/fs/file_table.c @@ -253,6 +253,7 @@ void __fput(struct file *file) if (file->f_op && file->f_op->release) file->f_op->release(inode, file); security_file_free(file); + ima_file_free(file); if (unlikely(S_ISCHR(inode->i_mode) && inode->i_cdev != NULL)) cdev_put(inode->i_cdev); fops_put(file->f_op); diff --git a/security/security.c b/security/security.c index 24e060be9fa5..122b748d0f4c 100644 --- a/security/security.c +++ b/security/security.c @@ -666,8 +666,6 @@ int security_file_alloc(struct file *file) void security_file_free(struct file *file) { security_ops->file_free_security(file); - if (file->f_dentry) - ima_file_free(file); } int security_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) -- cgit v1.2.3 From 80e1e823989ec44d8e35bdfddadbddcffec90424 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 7 Feb 2010 10:11:23 -0800 Subject: Fix race in tty_fasync() properly MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This reverts commit 703625118069 ("tty: fix race in tty_fasync") and commit b04da8bfdfbb ("fnctl: f_modown should call write_lock_irqsave/ restore") that tried to fix up some of the fallout but was incomplete. It turns out that we really cannot hold 'tty->ctrl_lock' over calling __f_setown, because not only did that cause problems with interrupt disables (which the second commit fixed), it also causes a potential ABBA deadlock due to lock ordering. Thanks to Tetsuo Handa for following up on the issue, and running lockdep to show the problem. It goes roughly like this: - f_getown gets filp->f_owner.lock for reading without interrupts disabled, so an interrupt that happens while that lock is held can cause a lockdep chain from f_owner.lock -> sighand->siglock. - at the same time, the tty->ctrl_lock -> f_owner.lock chain that commit 703625118069 introduced, together with the pre-existing sighand->siglock -> tty->ctrl_lock chain means that we have a lock dependency the other way too. So instead of extending tty->ctrl_lock over the whole __f_setown() call, we now just take a reference to the 'pid' structure while holding the lock, and then release it after having done the __f_setown. That still guarantees that 'struct pid' won't go away from under us, which is all we really ever needed. Reported-and-tested-by: Tetsuo Handa Acked-by: Greg Kroah-Hartman Acked-by: Américo Wang Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- drivers/char/tty_io.c | 4 +++- fs/fcntl.c | 6 ++---- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c index c6f3b48be9dd..dcb9083ecde0 100644 --- a/drivers/char/tty_io.c +++ b/drivers/char/tty_io.c @@ -1951,8 +1951,10 @@ static int tty_fasync(int fd, struct file *filp, int on) pid = task_pid(current); type = PIDTYPE_PID; } - retval = __f_setown(filp, pid, type, 0); + get_pid(pid); spin_unlock_irqrestore(&tty->ctrl_lock, flags); + retval = __f_setown(filp, pid, type, 0); + put_pid(pid); if (retval) goto out; } else { diff --git a/fs/fcntl.c b/fs/fcntl.c index 5ef953e6f908..97e01dc0d95f 100644 --- a/fs/fcntl.c +++ b/fs/fcntl.c @@ -199,9 +199,7 @@ static int setfl(int fd, struct file * filp, unsigned long arg) static void f_modown(struct file *filp, struct pid *pid, enum pid_type type, int force) { - unsigned long flags; - - write_lock_irqsave(&filp->f_owner.lock, flags); + write_lock_irq(&filp->f_owner.lock); if (force || !filp->f_owner.pid) { put_pid(filp->f_owner.pid); filp->f_owner.pid = get_pid(pid); @@ -213,7 +211,7 @@ static void f_modown(struct file *filp, struct pid *pid, enum pid_type type, filp->f_owner.euid = cred->euid; } } - write_unlock_irqrestore(&filp->f_owner.lock, flags); + write_unlock_irq(&filp->f_owner.lock); } int __f_setown(struct file *filp, struct pid *pid, enum pid_type type, -- cgit v1.2.3 From ccd4bb1beb3316de4611de24d223ad761b5a7e95 Mon Sep 17 00:00:00 2001 From: Steve French Date: Mon, 8 Feb 2010 17:39:58 +0000 Subject: [CIFS] Don't cache timestamps on utimes due to coarse granularity force revalidate of the file when any of the timestamps are set since some filesytem types do not have finer granularity timestamps and we can not always detect which file systems round timestamps down to determine whether we can cache the mtime on setattr samba bugzilla 3775 Acked-by: Shirish Pargaonkar Signed-off-by: Steve French --- fs/cifs/inode.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c index cf18ee765590..e3fda978f481 100644 --- a/fs/cifs/inode.c +++ b/fs/cifs/inode.c @@ -1762,8 +1762,18 @@ cifs_setattr_unix(struct dentry *direntry, struct iattr *attrs) CIFS_MOUNT_MAP_SPECIAL_CHR); } - if (!rc) + if (!rc) { rc = inode_setattr(inode, attrs); + + /* force revalidate when any of these times are set since some + of the fs types (eg ext3, fat) do not have fine enough + time granularity to match protocol, and we do not have a + a way (yet) to query the server fs's time granularity (and + whether it rounds times down). + */ + if (!rc && (attrs->ia_valid & (ATTR_MTIME | ATTR_CTIME))) + cifsInode->time = 0; + } out: kfree(args); kfree(full_path); -- cgit v1.2.3 From 05507fa2ac8d5e503bcf33ee43329449027d9060 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 5 Feb 2010 13:30:36 -0500 Subject: cifs: fix dentry hash calculation for case-insensitive mounts case-insensitive mounts shouldn't use full_name_hash(). Make sure we use the parent dentry's d_hash routine when one is set. Reported-by: Dave Kleikamp Signed-off-by: Jeff Layton Signed-off-by: Steve French --- fs/cifs/readdir.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c index f5618f8cc462..c343b14ba2d3 100644 --- a/fs/cifs/readdir.c +++ b/fs/cifs/readdir.c @@ -77,6 +77,11 @@ cifs_readdir_lookup(struct dentry *parent, struct qstr *name, cFYI(1, ("For %s", name->name)); + if (parent->d_op && parent->d_op->d_hash) + parent->d_op->d_hash(parent, name); + else + name->hash = full_name_hash(name->name, name->len); + dentry = d_lookup(parent, name); if (dentry) { /* FIXME: check for inode number changes? */ @@ -671,8 +676,6 @@ static int cifs_get_name_from_search_buf(struct qstr *pqst, pqst->name = filename; pqst->len = len; } - pqst->hash = full_name_hash(pqst->name, pqst->len); -/* cFYI(1, ("filldir on %s",pqst->name)); */ return rc; } -- cgit v1.2.3 From 84eb8fb42c120ff32b201c1cdd910033c888f699 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Tue, 5 Jan 2010 19:41:44 +0900 Subject: [SCSI] compat_ioct: fix bsg SG_IO bsg's SG_IO doesn't work on 32-bit userspace and 64-bit kernelspace. The problem is that both sg and bsg drivers use SG_IO ioctl. sg_ioctl_trans() does 32/64-bit conversion even against bsg header. It messes up bsg header. bsg driver gets garbage. This patch fixes sg_ioctl_trans to handle only sg header (struct sg_io_hdr). Reported-by: Giridhar Malavali Signed-off-by: FUJITA Tomonori Signed-off-by: James Bottomley --- fs/compat_ioctl.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs') diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c index c5c45de1a2ee..7cbbc7ab4b50 100644 --- a/fs/compat_ioctl.c +++ b/fs/compat_ioctl.c @@ -301,6 +301,12 @@ static int sg_ioctl_trans(unsigned int fd, unsigned int cmd, u32 data; void __user *dxferp; int err; + int interface_id; + + if (get_user(interface_id, &sgio32->interface_id)) + return -EFAULT; + if (interface_id != 'S') + return sys_ioctl(fd, cmd, (unsigned long)sgio32); if (get_user(iovec_count, &sgio32->iovec_count)) return -EFAULT; -- cgit v1.2.3 From 260c64d23532caf19abb77e696971da05c388489 Mon Sep 17 00:00:00 2001 From: "J. Bruce Fields" Date: Mon, 8 Feb 2010 13:42:26 -0500 Subject: Revert "nfsd4: fix error return when pseudoroot missing" Commit f39bde24b275ddc45d fixed the error return from PUTROOTFH in the case where there is no pseudofilesystem. This is really a case we shouldn't hit on a correctly configured server: in the absence of a root filehandle, there's no point accepting version 4 NFS rpc calls at all. But the shared responsibility between kernel and userspace here means the kernel on its own can't eliminate the possiblity of this happening. And we have indeed gotten this wrong in distro's, so new client-side mount code that attempts to negotiate v4 by default first has to work around this case. Therefore when commit f39bde24b275ddc45d arrived at roughly the same time as the new v4-default mount code, which explicitly checked only for the previous error, the result was previously fine mounts suddenly failing. We'll fix both sides for now: revert the error change, and make the client-side mount workaround more robust. Signed-off-by: J. Bruce Fields --- fs/nfsd/export.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) (limited to 'fs') diff --git a/fs/nfsd/export.c b/fs/nfsd/export.c index c487810a2366..a0c4016413f1 100644 --- a/fs/nfsd/export.c +++ b/fs/nfsd/export.c @@ -1316,19 +1316,11 @@ rqst_exp_parent(struct svc_rqst *rqstp, struct path *path) static struct svc_export *find_fsidzero_export(struct svc_rqst *rqstp) { - struct svc_export *exp; u32 fsidv[2]; mk_fsid(FSID_NUM, fsidv, 0, 0, 0, NULL); - exp = rqst_exp_find(rqstp, FSID_NUM, fsidv); - /* - * We shouldn't have accepting an nfsv4 request at all if we - * don't have a pseudoexport!: - */ - if (IS_ERR(exp) && PTR_ERR(exp) == -ENOENT) - exp = ERR_PTR(-ESERVERFAULT); - return exp; + return rqst_exp_find(rqstp, FSID_NUM, fsidv); } /* -- cgit v1.2.3 From 86a06abab0ffbb9d8ce2b7f6b6652412ce2d2c36 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Fri, 5 Feb 2010 17:55:56 -0800 Subject: ocfs2/dlm: Fix printing of lockname The debug call printing the name of the lock resource was chopping off the last character. This patch fixes the problem. Signed-off-by: Sunil Mushran Acked-by: Mark Fasheh Signed-off-by: Joel Becker --- fs/ocfs2/dlm/dlmdebug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmdebug.c b/fs/ocfs2/dlm/dlmdebug.c index 42b0bad7a612..0cd24cf54396 100644 --- a/fs/ocfs2/dlm/dlmdebug.c +++ b/fs/ocfs2/dlm/dlmdebug.c @@ -102,7 +102,7 @@ void __dlm_print_one_lock_resource(struct dlm_lock_resource *res) assert_spin_locked(&res->spinlock); stringify_lockname(res->lockname.name, res->lockname.len, - buf, sizeof(buf) - 1); + buf, sizeof(buf)); printk("lockres: %s, owner=%u, state=%u\n", buf, res->owner, res->state); printk(" last used: %lu, refcnt: %u, on purge list: %s\n", -- cgit v1.2.3 From 6efd806634f7526f723f3aa7ceffd3887a932d9c Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Fri, 5 Feb 2010 15:41:23 -0800 Subject: ocfs2/cluster: Make o2net connect messages KERN_NOTICE Connect and disconnect messages are more than informational as they are required during root cause analysis for failures. This patch changes them from KERN_INFO to KERN_NOTICE. Signed-off-by: Sunil Mushran Acked-by: Mark Faseh Signed-off-by: Joel Becker --- fs/ocfs2/cluster/tcp.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index 938ba181a3d9..d8d0c65ac03c 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -485,7 +485,7 @@ static void o2net_set_nn_state(struct o2net_node *nn, } if (was_valid && !valid) { - printk(KERN_INFO "o2net: no longer connected to " + printk(KERN_NOTICE "o2net: no longer connected to " SC_NODEF_FMT "\n", SC_NODEF_ARGS(old_sc)); o2net_complete_nodes_nsw(nn); } @@ -493,7 +493,7 @@ static void o2net_set_nn_state(struct o2net_node *nn, if (!was_valid && valid) { o2quo_conn_up(o2net_num_from_nn(nn)); cancel_delayed_work(&nn->nn_connect_expired); - printk(KERN_INFO "o2net: %s " SC_NODEF_FMT "\n", + printk(KERN_NOTICE "o2net: %s " SC_NODEF_FMT "\n", o2nm_this_node() > sc->sc_node->nd_num ? "connected to" : "accepted connection from", SC_NODEF_ARGS(sc)); @@ -1476,7 +1476,7 @@ static void o2net_idle_timer(unsigned long data) do_gettimeofday(&now); - printk(KERN_INFO "o2net: connection to " SC_NODEF_FMT " has been idle for %u.%u " + printk(KERN_NOTICE "o2net: connection to " SC_NODEF_FMT " has been idle for %u.%u " "seconds, shutting it down.\n", SC_NODEF_ARGS(sc), o2net_idle_timeout() / 1000, o2net_idle_timeout() % 1000); -- cgit v1.2.3 From 7a4439c406c21b1e900ed497cec1a79d05b38c07 Mon Sep 17 00:00:00 2001 From: "M. Mohan Kumar" Date: Mon, 8 Feb 2010 15:36:48 -0600 Subject: 9p: Include fsync support for 9p client Implement the fsync in the client side by marking stat field values to 'don't touch' so that server may interpret it as a request to guarantee that the contents of the associated file are committed to stable storage before the Rwstat message is returned. Without this patch, calling fsync on a 9p file results in "Invalid argument" error. Please check the attached C program. Signed-off-by: Aneesh Kumar K.V Signed-off-by: M. Mohan Kumar Acked-by: Venkateswararao Jujjuri (JV) Signed-off-by: Eric Van Hensbergen --- fs/9p/v9fs_vfs.h | 1 + fs/9p/vfs_file.c | 19 +++++++++++++++++++ fs/9p/vfs_inode.c | 2 +- 3 files changed, 21 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/9p/v9fs_vfs.h b/fs/9p/v9fs_vfs.h index 3a7560e35865..ed835836e0dc 100644 --- a/fs/9p/v9fs_vfs.h +++ b/fs/9p/v9fs_vfs.h @@ -60,3 +60,4 @@ void v9fs_dentry_release(struct dentry *); int v9fs_uflags2omode(int uflags, int extended); ssize_t v9fs_file_readn(struct file *, char *, char __user *, u32, u64); +void v9fs_blank_wstat(struct p9_wstat *wstat); diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c index 3902bf43a088..74a0461a9ac0 100644 --- a/fs/9p/vfs_file.c +++ b/fs/9p/vfs_file.c @@ -257,6 +257,23 @@ v9fs_file_write(struct file *filp, const char __user * data, return total; } +static int v9fs_file_fsync(struct file *filp, struct dentry *dentry, + int datasync) +{ + struct p9_fid *fid; + struct p9_wstat wstat; + int retval; + + P9_DPRINTK(P9_DEBUG_VFS, "filp %p dentry %p datasync %x\n", filp, + dentry, datasync); + + fid = filp->private_data; + v9fs_blank_wstat(&wstat); + + retval = p9_client_wstat(fid, &wstat); + return retval; +} + static const struct file_operations v9fs_cached_file_operations = { .llseek = generic_file_llseek, .read = do_sync_read, @@ -266,6 +283,7 @@ static const struct file_operations v9fs_cached_file_operations = { .release = v9fs_dir_release, .lock = v9fs_file_lock, .mmap = generic_file_readonly_mmap, + .fsync = v9fs_file_fsync, }; const struct file_operations v9fs_file_operations = { @@ -276,4 +294,5 @@ const struct file_operations v9fs_file_operations = { .release = v9fs_dir_release, .lock = v9fs_file_lock, .mmap = generic_file_readonly_mmap, + .fsync = v9fs_file_fsync, }; diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c index 9d03d1ebca6f..a407fa3388c0 100644 --- a/fs/9p/vfs_inode.c +++ b/fs/9p/vfs_inode.c @@ -176,7 +176,7 @@ int v9fs_uflags2omode(int uflags, int extended) * */ -static void +void v9fs_blank_wstat(struct p9_wstat *wstat) { wstat->type = ~0; -- cgit v1.2.3 From d8c8a9e36560e9ff4c99279d64ce5dd0e1a33fa6 Mon Sep 17 00:00:00 2001 From: Eric Van Hensbergen Date: Mon, 8 Feb 2010 16:23:23 -0600 Subject: 9p: fix option parsing Options pointer is being moved before calling kfree() which seems to cause problems. This uses a separate pointer to track and free original allocation. Signed-off-by: Venkateswararao Jujjuri Signed-off-by: Eric Van Hensbergen w --- fs/9p/v9fs.c | 10 ++++++---- net/9p/client.c | 9 +++++---- net/9p/trans_fd.c | 10 ++++++---- net/9p/trans_rdma.c | 9 +++++---- 4 files changed, 22 insertions(+), 16 deletions(-) (limited to 'fs') diff --git a/fs/9p/v9fs.c b/fs/9p/v9fs.c index cf62b05e296a..6848788a13db 100644 --- a/fs/9p/v9fs.c +++ b/fs/9p/v9fs.c @@ -84,7 +84,7 @@ static const match_table_t tokens = { static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) { - char *options; + char *options, *tmp_options; substring_t args[MAX_OPT_ARGS]; char *p; int option = 0; @@ -102,9 +102,10 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) if (!opts) return 0; - options = kstrdup(opts, GFP_KERNEL); - if (!options) + tmp_options = kstrdup(opts, GFP_KERNEL); + if (!tmp_options) goto fail_option_alloc; + options = tmp_options; while ((p = strsep(&options, ",")) != NULL) { int token; @@ -194,7 +195,8 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) continue; } } - kfree(options); + + kfree(tmp_options); return ret; fail_option_alloc: diff --git a/net/9p/client.c b/net/9p/client.c index a2e2d61b903b..cbe066966b3c 100644 --- a/net/9p/client.c +++ b/net/9p/client.c @@ -69,7 +69,7 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...); static int parse_opts(char *opts, struct p9_client *clnt) { - char *options; + char *options, *tmp_options; char *p; substring_t args[MAX_OPT_ARGS]; int option; @@ -81,12 +81,13 @@ static int parse_opts(char *opts, struct p9_client *clnt) if (!opts) return 0; - options = kstrdup(opts, GFP_KERNEL); - if (!options) { + tmp_options = kstrdup(opts, GFP_KERNEL); + if (!tmp_options) { P9_DPRINTK(P9_DEBUG_ERROR, "failed to allocate copy of option string\n"); return -ENOMEM; } + options = tmp_options; while ((p = strsep(&options, ",")) != NULL) { int token; @@ -125,7 +126,7 @@ static int parse_opts(char *opts, struct p9_client *clnt) } free_and_return: - kfree(options); + kfree(tmp_options); return ret; } diff --git a/net/9p/trans_fd.c b/net/9p/trans_fd.c index be1cb909d8c0..31d0b05582a9 100644 --- a/net/9p/trans_fd.c +++ b/net/9p/trans_fd.c @@ -714,7 +714,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts) char *p; substring_t args[MAX_OPT_ARGS]; int option; - char *options; + char *options, *tmp_options; int ret; opts->port = P9_PORT; @@ -724,12 +724,13 @@ static int parse_opts(char *params, struct p9_fd_opts *opts) if (!params) return 0; - options = kstrdup(params, GFP_KERNEL); - if (!options) { + tmp_options = kstrdup(params, GFP_KERNEL); + if (!tmp_options) { P9_DPRINTK(P9_DEBUG_ERROR, "failed to allocate copy of option string\n"); return -ENOMEM; } + options = tmp_options; while ((p = strsep(&options, ",")) != NULL) { int token; @@ -760,7 +761,8 @@ static int parse_opts(char *params, struct p9_fd_opts *opts) continue; } } - kfree(options); + + kfree(tmp_options); return 0; } diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c index 65cb29db03f8..2c95a89c0f46 100644 --- a/net/9p/trans_rdma.c +++ b/net/9p/trans_rdma.c @@ -166,7 +166,7 @@ static int parse_opts(char *params, struct p9_rdma_opts *opts) char *p; substring_t args[MAX_OPT_ARGS]; int option; - char *options; + char *options, *tmp_options; int ret; opts->port = P9_PORT; @@ -177,12 +177,13 @@ static int parse_opts(char *params, struct p9_rdma_opts *opts) if (!params) return 0; - options = kstrdup(params, GFP_KERNEL); - if (!options) { + tmp_options = kstrdup(params, GFP_KERNEL); + if (!tmp_options) { P9_DPRINTK(P9_DEBUG_ERROR, "failed to allocate copy of option string\n"); return -ENOMEM; } + options = tmp_options; while ((p = strsep(&options, ",")) != NULL) { int token; @@ -216,7 +217,7 @@ static int parse_opts(char *params, struct p9_rdma_opts *opts) } /* RQ must be at least as large as the SQ */ opts->rq_depth = max(opts->rq_depth, opts->sq_depth); - kfree(options); + kfree(tmp_options); return 0; } -- cgit v1.2.3 From bf2d29c64dd777e9a40bc4533e721944a590250f Mon Sep 17 00:00:00 2001 From: Eric Van Hensbergen Date: Mon, 8 Feb 2010 17:59:34 -0600 Subject: 9p: fix memory leak in v9fs_parse_options() If match_strdup() fail this function exits without freeing the options string. Signed-off-by: Venkateswararao Jujjuri Sigend-off-by: Eric Van Hensbergen --- fs/9p/v9fs.c | 27 +++++++++++++++++---------- 1 file changed, 17 insertions(+), 10 deletions(-) (limited to 'fs') diff --git a/fs/9p/v9fs.c b/fs/9p/v9fs.c index 6848788a13db..7d6c2139891d 100644 --- a/fs/9p/v9fs.c +++ b/fs/9p/v9fs.c @@ -103,8 +103,10 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) return 0; tmp_options = kstrdup(opts, GFP_KERNEL); - if (!tmp_options) + if (!tmp_options) { + ret = -ENOMEM; goto fail_option_alloc; + } options = tmp_options; while ((p = strsep(&options, ",")) != NULL) { @@ -160,8 +162,12 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) break; case Opt_cache: s = match_strdup(&args[0]); - if (!s) - goto fail_option_alloc; + if (!s) { + ret = -ENOMEM; + P9_DPRINTK(P9_DEBUG_ERROR, + "problem allocating copy of cache arg\n"); + goto free_and_return; + } if (strcmp(s, "loose") == 0) v9ses->cache = CACHE_LOOSE; @@ -174,8 +180,12 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) case Opt_access: s = match_strdup(&args[0]); - if (!s) - goto fail_option_alloc; + if (!s) { + ret = -ENOMEM; + P9_DPRINTK(P9_DEBUG_ERROR, + "problem allocating copy of access arg\n"); + goto free_and_return; + } v9ses->flags &= ~V9FS_ACCESS_MASK; if (strcmp(s, "user") == 0) @@ -196,13 +206,10 @@ static int v9fs_parse_options(struct v9fs_session_info *v9ses, char *opts) } } +free_and_return: kfree(tmp_options); - return ret; - fail_option_alloc: - P9_DPRINTK(P9_DEBUG_ERROR, - "failed to allocate copy of option argument\n"); - return -ENOMEM; + return ret; } /** -- cgit v1.2.3 From 2c1740098c708b465e87637b237feb2fd98f129a Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 8 Feb 2010 09:32:27 -0500 Subject: NFS: Fix a bug in nfs_fscache_release_page() Not having an fscache cookie is perfectly valid if the user didn't mount with the fscache option. This patch fixes http://bugzilla.kernel.org/show_bug.cgi?id=15234 Signed-off-by: Trond Myklebust Acked-by: David Howells Cc: stable@kernel.org --- fs/nfs/fscache.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c index fa588006588d..237874f1af23 100644 --- a/fs/nfs/fscache.c +++ b/fs/nfs/fscache.c @@ -354,12 +354,11 @@ void nfs_fscache_reset_inode_cookie(struct inode *inode) */ int nfs_fscache_release_page(struct page *page, gfp_t gfp) { - struct nfs_inode *nfsi = NFS_I(page->mapping->host); - struct fscache_cookie *cookie = nfsi->fscache; - - BUG_ON(!cookie); - if (PageFsCache(page)) { + struct nfs_inode *nfsi = NFS_I(page->mapping->host); + struct fscache_cookie *cookie = nfsi->fscache; + + BUG_ON(!cookie); dfprintk(FSCACHE, "NFS: fscache releasepage (0x%p/0x%p/0x%p)\n", cookie, page, nfsi); -- cgit v1.2.3 From 7549ad5f9b6eda49bbac4b14c5b8f37bf464f922 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 8 Feb 2010 09:32:34 -0500 Subject: NFS: Remove a redundant check for PageFsCache in nfs_migrate_page() Signed-off-by: Trond Myklebust Acked-by: David Howells --- fs/nfs/write.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/write.c b/fs/nfs/write.c index 7b54b8bb101f..d63d964a0392 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1598,8 +1598,7 @@ int nfs_migrate_page(struct address_space *mapping, struct page *newpage, struct nfs_page *req; int ret; - if (PageFsCache(page)) - nfs_fscache_release_page(page, GFP_KERNEL); + nfs_fscache_release_page(page, GFP_KERNEL); req = nfs_find_and_lock_request(page); ret = PTR_ERR(req); -- cgit v1.2.3 From fdcb45777a3d1689c5541e1f85ee3ebbd197d2c1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 8 Feb 2010 09:32:40 -0500 Subject: NFS: Fix the mapping of the NFSERR_SERVERFAULT error It was recently pointed out that the NFSERR_SERVERFAULT error, which is designed to inform the user of a serious internal error on the server, was being mapped to an error value that is internal to the kernel. This patch maps it to the error EREMOTEIO, which is exported to userland through errno.h. Signed-off-by: Trond Myklebust Cc: stable@kernel.org --- fs/nfs/mount_clnt.c | 2 +- fs/nfs/nfs2xdr.c | 2 +- fs/nfs/nfs4xdr.c | 6 +++--- 3 files changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfs/mount_clnt.c b/fs/nfs/mount_clnt.c index 0adefc40cc89..59047f8d7d72 100644 --- a/fs/nfs/mount_clnt.c +++ b/fs/nfs/mount_clnt.c @@ -120,7 +120,7 @@ static struct { { .status = MNT3ERR_INVAL, .errno = -EINVAL, }, { .status = MNT3ERR_NAMETOOLONG, .errno = -ENAMETOOLONG, }, { .status = MNT3ERR_NOTSUPP, .errno = -ENOTSUPP, }, - { .status = MNT3ERR_SERVERFAULT, .errno = -ESERVERFAULT, }, + { .status = MNT3ERR_SERVERFAULT, .errno = -EREMOTEIO, }, }; struct mountres { diff --git a/fs/nfs/nfs2xdr.c b/fs/nfs/nfs2xdr.c index 5e078b222b4e..7bc2da8efd4a 100644 --- a/fs/nfs/nfs2xdr.c +++ b/fs/nfs/nfs2xdr.c @@ -699,7 +699,7 @@ static struct { { NFSERR_BAD_COOKIE, -EBADCOOKIE }, { NFSERR_NOTSUPP, -ENOTSUPP }, { NFSERR_TOOSMALL, -ETOOSMALL }, - { NFSERR_SERVERFAULT, -ESERVERFAULT }, + { NFSERR_SERVERFAULT, -EREMOTEIO }, { NFSERR_BADTYPE, -EBADTYPE }, { NFSERR_JUKEBOX, -EJUKEBOX }, { -1, -EIO } diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index e437fd6a819f..5cd5184b56db 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -4631,7 +4631,7 @@ static int decode_sequence(struct xdr_stream *xdr, * If the server returns different values for sessionID, slotID or * sequence number, the server is looney tunes. */ - status = -ESERVERFAULT; + status = -EREMOTEIO; if (memcmp(id.data, res->sr_session->sess_id.data, NFS4_MAX_SESSIONID_LEN)) { @@ -5774,7 +5774,7 @@ static struct { { NFS4ERR_BAD_COOKIE, -EBADCOOKIE }, { NFS4ERR_NOTSUPP, -ENOTSUPP }, { NFS4ERR_TOOSMALL, -ETOOSMALL }, - { NFS4ERR_SERVERFAULT, -ESERVERFAULT }, + { NFS4ERR_SERVERFAULT, -EREMOTEIO }, { NFS4ERR_BADTYPE, -EBADTYPE }, { NFS4ERR_LOCKED, -EAGAIN }, { NFS4ERR_SYMLINK, -ELOOP }, @@ -5801,7 +5801,7 @@ nfs4_stat_to_errno(int stat) } if (stat <= 10000 || stat > 10100) { /* The server is looney tunes. */ - return -ESERVERFAULT; + return -EREMOTEIO; } /* If we cannot translate the error, the recovery routines should * handle it. -- cgit v1.2.3 From f79f11852831ba8837e82b73364e6f1cd0145499 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Wed, 10 Feb 2010 16:14:04 +0100 Subject: compat_ioctl: ignore RAID_VERSION ioctl md ioctls are now handled by the md driver itself, but mdadm may call RAID_VERSION on other devices as well. Mark the command as IGNORE_IOCTL so this fails silently rather than printing an annoying message. Reported-by: "Michael S. Tsirkin" Cc: "Rafael J. Wysocki" Signed-off-by: Arnd Bergmann Signed-off-by: Linus Torvalds --- fs/compat_ioctl.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c index c5c45de1a2ee..b6f23b25370e 100644 --- a/fs/compat_ioctl.c +++ b/fs/compat_ioctl.c @@ -1038,6 +1038,8 @@ COMPATIBLE_IOCTL(FIOQSIZE) #ifdef CONFIG_BLOCK /* loop */ IGNORE_IOCTL(LOOP_CLR_FD) +/* md calls this on random blockdevs */ +IGNORE_IOCTL(RAID_VERSION) /* SG stuff */ COMPATIBLE_IOCTL(SG_SET_TIMEOUT) COMPATIBLE_IOCTL(SG_GET_TIMEOUT) -- cgit v1.2.3 From 4cfbafd33f5ae99688ab82525a1d449c1c1b198f Mon Sep 17 00:00:00 2001 From: Andreas Schwab Date: Wed, 10 Feb 2010 13:56:40 -0800 Subject: compat_ioctl: add compat handler for TIOCGSID ioctl This is used by tcgetsid(3). Signed-off-by: Andreas Schwab Cc: Alan Cox Acked-by: Arnd Bergmann Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/compat_ioctl.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c index b6f23b25370e..30698a13fb22 100644 --- a/fs/compat_ioctl.c +++ b/fs/compat_ioctl.c @@ -936,6 +936,7 @@ COMPATIBLE_IOCTL(TCSETSF) COMPATIBLE_IOCTL(TIOCLINUX) COMPATIBLE_IOCTL(TIOCSBRK) COMPATIBLE_IOCTL(TIOCCBRK) +COMPATIBLE_IOCTL(TIOCGSID) COMPATIBLE_IOCTL(TIOCGICOUNT) /* Little t */ COMPATIBLE_IOCTL(TIOCGETD) -- cgit v1.2.3 From 803bf5ec259941936262d10ecc84511b76a20921 Mon Sep 17 00:00:00 2001 From: Michael Neuling Date: Wed, 10 Feb 2010 13:56:42 -0800 Subject: fs/exec.c: restrict initial stack space expansion to rlimit When reserving stack space for a new process, make sure we're not attempting to expand the stack by more than rlimit allows. This fixes a bug caused by b6a2fea39318e43fee84fa7b0b90d68bed92d2ba ("mm: variable length argument support") and unmasked by fc63cf237078c86214abcb2ee9926d8ad289da9b ("exec: setup_arg_pages() fails to return errors"). This bug means that when limiting the stack to less the 20*PAGE_SIZE (eg. 80K on 4K pages or 'ulimit -s 79') all processes will be killed before they start. This is particularly bad with 64K pages, where a ulimit below 1280K will kill every process. To test, do: 'ulimit -s 15; ls' before and after the patch is applied. Before it's applied, 'ls' should be killed. After the patch is applied, 'ls' should no longer be killed. A stack limit of 15KB since it's small enough to trigger 20*PAGE_SIZE. Also 15KB not a multiple of PAGE_SIZE, which is a trickier case to handle correctly with this code. 4K pages should be fine to test with. [kosaki.motohiro@jp.fujitsu.com: cleanup] [akpm@linux-foundation.org: cleanup cleanup] Signed-off-by: Michael Neuling Signed-off-by: KOSAKI Motohiro Cc: Americo Wang Cc: Anton Blanchard Cc: Oleg Nesterov Cc: James Morris Cc: Ingo Molnar Cc: Serge Hallyn Cc: Benjamin Herrenschmidt Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/exec.c | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/exec.c b/fs/exec.c index 0790a107ff7e..e95c692ef0e4 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -571,6 +571,9 @@ int setup_arg_pages(struct linux_binprm *bprm, struct vm_area_struct *prev = NULL; unsigned long vm_flags; unsigned long stack_base; + unsigned long stack_size; + unsigned long stack_expand; + unsigned long rlim_stack; #ifdef CONFIG_STACK_GROWSUP /* Limit stack size to 1GB */ @@ -627,10 +630,24 @@ int setup_arg_pages(struct linux_binprm *bprm, goto out_unlock; } + stack_expand = EXTRA_STACK_VM_PAGES * PAGE_SIZE; + stack_size = vma->vm_end - vma->vm_start; + /* + * Align this down to a page boundary as expand_stack + * will align it up. + */ + rlim_stack = rlimit(RLIMIT_STACK) & PAGE_MASK; + rlim_stack = min(rlim_stack, stack_size); #ifdef CONFIG_STACK_GROWSUP - stack_base = vma->vm_end + EXTRA_STACK_VM_PAGES * PAGE_SIZE; + if (stack_size + stack_expand > rlim_stack) + stack_base = vma->vm_start + rlim_stack; + else + stack_base = vma->vm_end + stack_expand; #else - stack_base = vma->vm_start - EXTRA_STACK_VM_PAGES * PAGE_SIZE; + if (stack_size + stack_expand > rlim_stack) + stack_base = vma->vm_end - rlim_stack; + else + stack_base = vma->vm_start - stack_expand; #endif ret = expand_stack(vma, stack_base); if (ret) -- cgit v1.2.3 From 0e5a9fb0426108d750c97c25b1ab04d3768b5aff Mon Sep 17 00:00:00 2001 From: Abhijith Das Date: Fri, 5 Feb 2010 18:25:41 -0500 Subject: GFS2: Fix error code We need this one-liner to signal the mount helper of the 'insufficient journals' condition. Signed-off-by: Abhijith Das Signed-off-by: Steven Whitehouse --- fs/gfs2/ops_fstype.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c index 8a102f731003..a86ed6381566 100644 --- a/fs/gfs2/ops_fstype.c +++ b/fs/gfs2/ops_fstype.c @@ -725,7 +725,7 @@ static int init_journal(struct gfs2_sbd *sdp, int undo) goto fail; } - error = -EINVAL; + error = -EUSERS; if (!gfs2_jindex_size(sdp)) { fs_err(sdp, "no journals!\n"); goto fail_jindex; -- cgit v1.2.3 From 07ccb7bf2c928fef4fea2cda69ba2e23479578db Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Fri, 12 Feb 2010 10:10:55 +0000 Subject: GFS2: Fix bmap allocation corner-case bug This patch solves a corner case during allocation which occurs if both metadata (indirect) and data blocks are required but there is an obstacle in the filesystem (e.g. a resource group header or another allocated block) such that when the allocation is requested only enough blocks for the metadata are returned. By changing the exit condition of this loop, we ensure that a minimum of one data block will always be returned. Signed-off-by: Steven Whitehouse --- fs/gfs2/bmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c index 6d47379e794b..583e823307ae 100644 --- a/fs/gfs2/bmap.c +++ b/fs/gfs2/bmap.c @@ -541,7 +541,7 @@ static int gfs2_bmap_alloc(struct inode *inode, const sector_t lblock, *ptr++ = cpu_to_be64(bn++); break; } - } while (state != ALLOC_DATA); + } while ((state != ALLOC_DATA) || !dblock); ip->i_height = height; gfs2_add_inode_blocks(&ip->i_inode, alloced); -- cgit v1.2.3 From 3f6fae9559225741c91f1320090b285da1413290 Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Thu, 11 Feb 2010 07:43:00 +0000 Subject: Btrfs: btrfs_mark_extent_written uses the wrong slot My test do: fallocate a big file and do write. The file is 512M, but after file write is done btrfs-debug-tree shows: item 6 key (257 EXTENT_DATA 0) itemoff 3516 itemsize 53 extent data disk byte 1103101952 nr 536870912 extent data offset 0 nr 399634432 ram 536870912 extent compression 0 Looks like a regression introducted by 6c7d54ac87f338c479d9729e8392eca3f76e11e1, where we set wrong slot. Signed-off-by: Shaohua Li Acked-by: Yan Zheng Signed-off-by: Chris Mason --- fs/btrfs/file.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 413a30dafcda..a7fd9f3a750a 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -720,13 +720,15 @@ again: inode->i_ino, orig_offset); BUG_ON(ret); } - fi = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_file_extent_item); if (del_nr == 0) { + fi = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); btrfs_set_file_extent_type(leaf, fi, BTRFS_FILE_EXTENT_REG); btrfs_mark_buffer_dirty(leaf); } else { + fi = btrfs_item_ptr(leaf, del_slot - 1, + struct btrfs_file_extent_item); btrfs_set_file_extent_type(leaf, fi, BTRFS_FILE_EXTENT_REG); btrfs_set_file_extent_num_bytes(leaf, fi, -- cgit v1.2.3 From 175359f89df39f4faed663c8cfd6ee0222d2fa1e Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 11 Feb 2010 13:13:10 +0100 Subject: reiserfs: Fix softlockup while waiting on an inode When we wait for an inode through reiserfs_iget(), we hold the reiserfs lock. And waiting for an inode may imply waiting for its writeback. But the inode writeback path may also require the reiserfs lock, which leads to a deadlock. We just need to release the reiserfs lock from reiserfs_iget() to fix this. Reported-by: Alexander Beregalov Signed-off-by: Frederic Weisbecker Tested-by: Christian Kujau Cc: Chris Mason --- fs/reiserfs/inode.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c index 9087b10209e6..2df0f5c7c60b 100644 --- a/fs/reiserfs/inode.c +++ b/fs/reiserfs/inode.c @@ -1497,9 +1497,11 @@ struct inode *reiserfs_iget(struct super_block *s, const struct cpu_key *key) args.objectid = key->on_disk_key.k_objectid; args.dirid = key->on_disk_key.k_dir_id; + reiserfs_write_unlock(s); inode = iget5_locked(s, key->on_disk_key.k_objectid, reiserfs_find_actor, reiserfs_init_locked_inode, (void *)(&args)); + reiserfs_write_lock(s); if (!inode) return ERR_PTR(-ENOMEM); -- cgit v1.2.3 From 65d269538a1129495ac45a14a777cd11cfe881d8 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Mon, 15 Feb 2010 12:19:53 -0500 Subject: NFS: Too many GETATTR and ACCESS calls after direct I/O The cached read and write paths initialize fattr->time_start in their setup procedures. The value of fattr->time_start is propagated to read_cache_jiffies by nfs_update_inode(). Subsequent calls to nfs_attribute_timeout() will then use a good time stamp when computing the attribute cache timeout, and squelch unneeded GETATTR calls. Since the direct I/O paths erroneously leave the inode's fattr->time_start field set to zero, read_cache_jiffies for that inode is set to zero after any direct read or write operation. This triggers an otw GETATTR or ACCESS call to update the file's attribute and access caches properly, even when the NFS READ or WRITE replies have usable post-op attributes. Make sure the direct read and write setup code performs the same fattr initialization as the cached I/O paths to prevent unnecessary GETATTR calls. This was likely introduced by commit 0e574af1 in 2.6.15, which appears to add new nfs_fattr_init() call sites in the cached read and write paths, but not in the equivalent places in fs/nfs/direct.c. A subsequent commit in the same series, 33801147, introduces the fattr->time_start field. Interestingly, the direct write reschedule path already has a call to nfs_fattr_init() in the right place. Reported-by: Quentin Barnes Signed-off-by: Chuck Lever Cc: stable@kernel.org Signed-off-by: Trond Myklebust Signed-off-by: Linus Torvalds --- fs/nfs/direct.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c index e1d415e97849..0d289823e856 100644 --- a/fs/nfs/direct.c +++ b/fs/nfs/direct.c @@ -342,6 +342,7 @@ static ssize_t nfs_direct_read_schedule_segment(struct nfs_direct_req *dreq, data->res.fattr = &data->fattr; data->res.eof = 0; data->res.count = bytes; + nfs_fattr_init(&data->fattr); msg.rpc_argp = &data->args; msg.rpc_resp = &data->res; @@ -575,6 +576,7 @@ static void nfs_direct_commit_schedule(struct nfs_direct_req *dreq) data->res.count = 0; data->res.fattr = &data->fattr; data->res.verf = &data->verf; + nfs_fattr_init(&data->fattr); NFS_PROTO(data->inode)->commit_setup(data, &msg); @@ -766,6 +768,7 @@ static ssize_t nfs_direct_write_schedule_segment(struct nfs_direct_req *dreq, data->res.fattr = &data->fattr; data->res.count = bytes; data->res.verf = &data->verf; + nfs_fattr_init(&data->fattr); task_setup_data.task = &data->task; task_setup_data.callback_data = data; -- cgit v1.2.3 From 7c0ff870d1ed287504a61ed865f3d728c757436b Mon Sep 17 00:00:00 2001 From: "Eric W. Biederman" Date: Wed, 3 Feb 2010 23:13:24 -0800 Subject: sysfs: sysfs_sd_setattr set iattrs unconditionally There is currently a bug in sysfs_sd_setattr inherited from sysfs_setattr in 2.6.32 where the first time we set the attributes on a sysfs file we allocate backing store but do not set the backing store attributes. Resulting in overly restrictive permissions on sysfs files. The fix is to simply modify the code so that it always executes when we update the sysfs attributes, as we did in 2.6.31 and earlier. Signed-off-by: Eric W. Biederman Tested-by: Jean Delvare Cc: stable Signed-off-by: Greg Kroah-Hartman --- fs/sysfs/inode.c | 35 +++++++++++++++++------------------ 1 file changed, 17 insertions(+), 18 deletions(-) (limited to 'fs') diff --git a/fs/sysfs/inode.c b/fs/sysfs/inode.c index 220b758523ae..6a06a1d1ea7b 100644 --- a/fs/sysfs/inode.c +++ b/fs/sysfs/inode.c @@ -81,24 +81,23 @@ int sysfs_sd_setattr(struct sysfs_dirent *sd, struct iattr * iattr) if (!sd_attrs) return -ENOMEM; sd->s_iattr = sd_attrs; - } else { - /* attributes were changed at least once in past */ - iattrs = &sd_attrs->ia_iattr; - - if (ia_valid & ATTR_UID) - iattrs->ia_uid = iattr->ia_uid; - if (ia_valid & ATTR_GID) - iattrs->ia_gid = iattr->ia_gid; - if (ia_valid & ATTR_ATIME) - iattrs->ia_atime = iattr->ia_atime; - if (ia_valid & ATTR_MTIME) - iattrs->ia_mtime = iattr->ia_mtime; - if (ia_valid & ATTR_CTIME) - iattrs->ia_ctime = iattr->ia_ctime; - if (ia_valid & ATTR_MODE) { - umode_t mode = iattr->ia_mode; - iattrs->ia_mode = sd->s_mode = mode; - } + } + /* attributes were changed at least once in past */ + iattrs = &sd_attrs->ia_iattr; + + if (ia_valid & ATTR_UID) + iattrs->ia_uid = iattr->ia_uid; + if (ia_valid & ATTR_GID) + iattrs->ia_gid = iattr->ia_gid; + if (ia_valid & ATTR_ATIME) + iattrs->ia_atime = iattr->ia_atime; + if (ia_valid & ATTR_MTIME) + iattrs->ia_mtime = iattr->ia_mtime; + if (ia_valid & ATTR_CTIME) + iattrs->ia_ctime = iattr->ia_ctime; + if (ia_valid & ATTR_MODE) { + umode_t mode = iattr->ia_mode; + iattrs->ia_mode = sd->s_mode = mode; } return 0; } -- cgit v1.2.3