From 2bea038c52e88badbbd5b420c1de918f7f2579e9 Mon Sep 17 00:00:00 2001 From: Boaz Harrosh Date: Thu, 16 Jun 2011 11:35:46 -0400 Subject: pnfs: write: Set mds_offset in the generic layer - it is needed by all LDs In current pnfs tree, all the layouts set mds_offset in their .write_pagelist member. mds_offset is only used by generic layer and should be handled by it. This patch is for upstream. It is needed in this -rc series to fix a bug in objects layout_commit. I'll send patches for objects and blocks to be squashed into current pnfs tree. TODO: It looks like the read path needs the same patch. Signed-off-by: Boaz Harrosh Signed-off-by: Trond Myklebust --- fs/nfs/nfs4filelayout.c | 1 - fs/nfs/write.c | 2 ++ 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c index 0bafcc91c27..f9d03abcd04 100644 --- a/fs/nfs/nfs4filelayout.c +++ b/fs/nfs/nfs4filelayout.c @@ -398,7 +398,6 @@ filelayout_write_pagelist(struct nfs_write_data *data, int sync) * this offset and save the original offset. */ data->args.offset = filelayout_get_dserver_offset(lseg, offset); - data->mds_offset = offset; /* Perform an asynchronous write */ status = nfs_initiate_write(data, ds->ds_clp->cl_rpcclient, diff --git a/fs/nfs/write.c b/fs/nfs/write.c index e268e3b2349..72716805968 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -864,6 +864,8 @@ static int nfs_write_rpcsetup(struct nfs_page *req, data->args.fh = NFS_FH(inode); data->args.offset = req_offset(req) + offset; + /* pnfs_set_layoutcommit needs this */ + data->mds_offset = data->args.offset; data->args.pgbase = req->wb_pgbase + offset; data->args.pages = data->pagevec; data->args.count = count; -- cgit v1.2.3 From e5012d1f3861d18c7f3814e757c1c3ab3741dbcd Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Mon, 11 Jul 2011 17:17:42 -0400 Subject: NFSv4.1: update nfs4_fattr_bitmap_maxsz Attribute IDs assigned in RFC 5661 now require three bitmaps. Fixes hitting a BUG_ON in xdr_shrink_bufhead when getting ACLs. Signed-off-by: Andy Adamson Cc:stable@kernel.org [2.6.39] Signed-off-by: Trond Myklebust --- fs/nfs/nfs4xdr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index 6870bc61cee..e6e8f3b9a1d 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -91,7 +91,7 @@ static int nfs4_stat_to_errno(int); #define encode_getfh_maxsz (op_encode_hdr_maxsz) #define decode_getfh_maxsz (op_decode_hdr_maxsz + 1 + \ ((3+NFS4_FHSIZE) >> 2)) -#define nfs4_fattr_bitmap_maxsz 3 +#define nfs4_fattr_bitmap_maxsz 4 #define encode_getattr_maxsz (op_encode_hdr_maxsz + nfs4_fattr_bitmap_maxsz) #define nfs4_name_maxsz (1 + ((3 + NFS4_MAXNAMLEN) >> 2)) #define nfs4_path_maxsz (1 + ((3 + NFS4_MAXPATHLEN) >> 2)) -- cgit v1.2.3 From 1ce533686c7d40bf900dc346a7279c17a9ee8e0e Mon Sep 17 00:00:00 2001 From: Benjamin Marzinski Date: Mon, 13 Jun 2011 14:27:40 -0500 Subject: GFS2: force a log flush when invalidating the rindex glock Right now, there is nothing that forces the log to get flushed when a node drops its rindex glock so that another node can grow the filesystem. If the log doesn't get flushed, GFS2 can corrupt the sd_log_le_rg list in the following way. A node puts an rgd on the list in rg_lo_add(), and then the rindex glock is dropped so the other node can grow the filesystem. When the node reacquires the rindex glock, that rgd gets deleted in clear_rgrpdi() before ever being removed from the list by gfs2_log_flush(). This code simply forces a log flush when the rindex glock is invalidated, solving the problem. Signed-off-by: Benjamin Marzinski Signed-off-by: Steven Whitehouse --- fs/gfs2/glops.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c index 8ef70f46473..712b722030a 100644 --- a/fs/gfs2/glops.c +++ b/fs/gfs2/glops.c @@ -221,8 +221,10 @@ static void inode_go_inval(struct gfs2_glock *gl, int flags) } } - if (ip == GFS2_I(gl->gl_sbd->sd_rindex)) + if (ip == GFS2_I(gl->gl_sbd->sd_rindex)) { + gfs2_log_flush(gl->gl_sbd, NULL); gl->gl_sbd->sd_rindex_uptodate = 0; + } if (ip && S_ISREG(ip->i_inode.i_mode)) truncate_inode_pages(ip->i_inode.i_mapping, 0); } -- cgit v1.2.3 From 3942ae5319640ced5844b75f44884e4bcb8a2f16 Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Mon, 11 Jul 2011 08:53:30 +0100 Subject: GFS2: Fix race during filesystem mount There is a potential race during filesystem mounting which has recently been reported. It occurs when the userland gfs_controld is able to process requests fast enough that it tries to use the sysfs interface before the lock module is properly initialised. This is a pretty unusual case as normally the lock module initialisation is very quick compared with gfs_controld. This patch adds an interruptible completion which is used to ensure that userland will wait for the initialisation of the lock module to complete. There are other potential solutions to this problem, but this is the quickest at this stage and has been tested both with and without mount.gfs2 present in the system. Signed-off-by: Steven Whitehouse Reported-by: David Booher --- fs/gfs2/incore.h | 2 ++ fs/gfs2/ops_fstype.c | 3 +++ fs/gfs2/sys.c | 7 ++++++- 3 files changed, 11 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/incore.h b/fs/gfs2/incore.h index 0a064e91ac7..81206e70cbf 100644 --- a/fs/gfs2/incore.h +++ b/fs/gfs2/incore.h @@ -17,6 +17,7 @@ #include #include #include +#include #define DIO_WAIT 0x00000010 #define DIO_METADATA 0x00000020 @@ -546,6 +547,7 @@ struct gfs2_sbd { struct gfs2_glock *sd_trans_gl; wait_queue_head_t sd_glock_wait; atomic_t sd_glock_disposal; + struct completion sd_locking_init; /* Inode Stuff */ diff --git a/fs/gfs2/ops_fstype.c b/fs/gfs2/ops_fstype.c index 8ac9ae189b5..2a77071fb7b 100644 --- a/fs/gfs2/ops_fstype.c +++ b/fs/gfs2/ops_fstype.c @@ -72,6 +72,7 @@ static struct gfs2_sbd *init_sbd(struct super_block *sb) init_waitqueue_head(&sdp->sd_glock_wait); atomic_set(&sdp->sd_glock_disposal, 0); + init_completion(&sdp->sd_locking_init); spin_lock_init(&sdp->sd_statfs_spin); spin_lock_init(&sdp->sd_rindex_spin); @@ -1017,11 +1018,13 @@ hostdata_error: fsname++; if (lm->lm_mount == NULL) { fs_info(sdp, "Now mounting FS...\n"); + complete(&sdp->sd_locking_init); return 0; } ret = lm->lm_mount(sdp, fsname); if (ret == 0) fs_info(sdp, "Joined cluster. Now mounting FS...\n"); + complete(&sdp->sd_locking_init); return ret; } diff --git a/fs/gfs2/sys.c b/fs/gfs2/sys.c index e20eab37bc8..443cabcfcd2 100644 --- a/fs/gfs2/sys.c +++ b/fs/gfs2/sys.c @@ -338,6 +338,9 @@ static ssize_t lkfirst_store(struct gfs2_sbd *sdp, const char *buf, size_t len) rv = sscanf(buf, "%u", &first); if (rv != 1 || first > 1) return -EINVAL; + rv = wait_for_completion_killable(&sdp->sd_locking_init); + if (rv) + return rv; spin_lock(&sdp->sd_jindex_spin); rv = -EBUSY; if (test_bit(SDF_NOJOURNALID, &sdp->sd_flags) == 0) @@ -414,7 +417,9 @@ static ssize_t jid_store(struct gfs2_sbd *sdp, const char *buf, size_t len) rv = sscanf(buf, "%d", &jid); if (rv != 1) return -EINVAL; - + rv = wait_for_completion_killable(&sdp->sd_locking_init); + if (rv) + return rv; spin_lock(&sdp->sd_jindex_spin); rv = -EINVAL; if (sdp->sd_lockstruct.ls_ops->lm_mount == NULL) -- cgit v1.2.3 From 62411ab2fe5f002dff27417630ddf02cc40ca404 Mon Sep 17 00:00:00 2001 From: Shirish Pargaonkar Date: Sun, 10 Jul 2011 06:55:32 -0500 Subject: cifs: Fix signing failure when server mandates signing for NTLMSSP When using NTLMSSP authentication mechanism, if server mandates signing, keep the flags in type 3 messages of the NTLMSSP exchange same as in type 1 messages (i.e. keep the indicated capabilities same). Some of the servers such as Samba, expect the flags such as Negotiate_Key_Exchange in type 3 message of NTLMSSP exchange as well. Some servers like Windows do not. https://bugzilla.samba.org/show_bug.cgi?id=8212 Signed-off-by: Shirish Pargaonkar Acked-by: Jeff Layton Signed-off-by: Steve French --- fs/cifs/sess.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/cifs/sess.c b/fs/cifs/sess.c index 3892ab817a3..d3e619692ee 100644 --- a/fs/cifs/sess.c +++ b/fs/cifs/sess.c @@ -428,8 +428,7 @@ static void build_ntlmssp_negotiate_blob(unsigned char *pbuffer, (SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED)) { flags |= NTLMSSP_NEGOTIATE_SIGN; if (!ses->server->session_estab) - flags |= NTLMSSP_NEGOTIATE_KEY_XCH | - NTLMSSP_NEGOTIATE_EXTENDED_SEC; + flags |= NTLMSSP_NEGOTIATE_KEY_XCH; } sec_blob->NegotiateFlags = cpu_to_le32(flags); @@ -465,10 +464,11 @@ static int build_ntlmssp_auth_blob(unsigned char *pbuffer, NTLMSSP_NEGOTIATE_128 | NTLMSSP_NEGOTIATE_UNICODE | NTLMSSP_NEGOTIATE_NTLM | NTLMSSP_NEGOTIATE_EXTENDED_SEC; if (ses->server->sec_mode & - (SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED)) + (SECMODE_SIGN_REQUIRED | SECMODE_SIGN_ENABLED)) { flags |= NTLMSSP_NEGOTIATE_SIGN; - if (ses->server->sec_mode & SECMODE_SIGN_REQUIRED) - flags |= NTLMSSP_NEGOTIATE_ALWAYS_SIGN; + if (!ses->server->session_estab) + flags |= NTLMSSP_NEGOTIATE_KEY_XCH; + } tmp = pbuffer + sizeof(AUTHENTICATE_MESSAGE); sec_blob->NegotiateFlags = cpu_to_le32(flags); -- cgit v1.2.3 From ea1be1a3c3c4b2bbc32aeb7995e18336c8060e0e Mon Sep 17 00:00:00 2001 From: Steve French Date: Tue, 12 Jul 2011 18:24:43 +0000 Subject: [CIFS] update limit for snprintf in cifs_construct_tcon In 34c87901e113 "Shrink stack space usage in cifs_construct_tcon" we change the size of the username name buffer from MAX_USERNAME_SIZE (256) to 28. This call to snprintf() needs to be updated as well. Reported by Dan Carpenter. Reviewed-by: Dan Carpenter Signed-off-by: Steve French --- fs/cifs/connect.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c index dbd669cc5bc..ccc1afa0bf3 100644 --- a/fs/cifs/connect.c +++ b/fs/cifs/connect.c @@ -3485,7 +3485,7 @@ cifs_construct_tcon(struct cifs_sb_info *cifs_sb, uid_t fsuid) goto out; } - snprintf(username, MAX_USERNAME_SIZE, "krb50x%x", fsuid); + snprintf(username, sizeof(username), "krb50x%x", fsuid); vol_info->username = username; vol_info->local_nls = cifs_sb->local_nls; vol_info->linux_uid = fsuid; -- cgit v1.2.3 From c2ec9471b5b1307429aef1cfaa2b3ae453a61d6f Mon Sep 17 00:00:00 2001 From: Steve French Date: Tue, 12 Jul 2011 19:15:02 +0000 Subject: [CIFS] update cifs to version 1.74 Signed-off-by: Steve French --- fs/cifs/cifsfs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/cifsfs.h b/fs/cifs/cifsfs.h index 0900e1658c9..036ca83e5f4 100644 --- a/fs/cifs/cifsfs.h +++ b/fs/cifs/cifsfs.h @@ -129,5 +129,5 @@ extern long cifs_ioctl(struct file *filep, unsigned int cmd, unsigned long arg); extern const struct export_operations cifs_export_ops; #endif /* CIFS_NFSD_EXPORT */ -#define CIFS_VERSION "1.73" +#define CIFS_VERSION "1.74" #endif /* _CIFSFS_H */ -- cgit v1.2.3 From 94c0d4ecbe7f9fe56e052b26b2ab484e246c07b4 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Tue, 12 Jul 2011 21:40:23 -0400 Subject: Fix ->d_lock locking order in unlazy_walk() Make sure that child is still a child of parent before nested locking of child->d_lock in unlazy_walk(); otherwise we are risking a violation of locking order and deadlocks. Signed-off-by: Al Viro --- fs/namei.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/namei.c b/fs/namei.c index 0223c41fb11..5c867dd1c0b 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -433,6 +433,8 @@ static int unlazy_walk(struct nameidata *nd, struct dentry *dentry) goto err_parent; BUG_ON(nd->inode != parent->d_inode); } else { + if (dentry->d_parent != parent) + goto err_parent; spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED); if (!__d_rcu_to_refcount(dentry, nd->seq)) goto err_child; -- cgit v1.2.3 From 380f7c65a7eb3288e4b6812acf3474a1de230707 Mon Sep 17 00:00:00 2001 From: Steven Whitehouse Date: Thu, 14 Jul 2011 08:59:44 +0100 Subject: GFS2: Resolve inode eviction and ail list interaction bug This patch contains a few misc fixes which resolve a recently reported issue. This patch has been a real team effort and has received a lot of testing. The first issue is that the ail lock needs to be held over a few more operations. The lock thats added into gfs2_releasepage() may possibly be a candidate for replacing with RCU at some future point, but at this stage we've gone for the obvious fix. The second issue is that gfs2_write_inode() can end up calling a glock recursively when called from gfs2_evict_inode() via the syncing code, so it needs a guard added. The third issue is that we either need to not truncate the metadata pages of inodes which have zero link count, but which we cannot deallocate due to them still being in use by other nodes, or we need to ensure that those pages have all made it through the journal and ail lists first. This patch takes the former approach, but the latter has also been tested and there is nothing to choose between them performance-wise. So again, we could revise that decision in the future. Also, the inode eviction process is now better documented. Signed-off-by: Steven Whitehouse Tested-by: Bob Peterson Tested-by: Abhijith Das Reported-by: Barry J. Marson Reported-by: David Teigland --- fs/gfs2/aops.c | 3 +++ fs/gfs2/glops.c | 4 ++-- fs/gfs2/log.c | 1 + fs/gfs2/super.c | 36 ++++++++++++++++++++++++++++++------ 4 files changed, 36 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c index 802ac5eeba2..f9fbbe96c22 100644 --- a/fs/gfs2/aops.c +++ b/fs/gfs2/aops.c @@ -1069,6 +1069,7 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask) return 0; gfs2_log_lock(sdp); + spin_lock(&sdp->sd_ail_lock); head = bh = page_buffers(page); do { if (atomic_read(&bh->b_count)) @@ -1080,6 +1081,7 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask) goto not_possible; bh = bh->b_this_page; } while(bh != head); + spin_unlock(&sdp->sd_ail_lock); gfs2_log_unlock(sdp); head = bh = page_buffers(page); @@ -1112,6 +1114,7 @@ not_possible: /* Should never happen */ WARN_ON(buffer_dirty(bh)); WARN_ON(buffer_pinned(bh)); cannot_release: + spin_unlock(&sdp->sd_ail_lock); gfs2_log_unlock(sdp); return 0; } diff --git a/fs/gfs2/glops.c b/fs/gfs2/glops.c index 712b722030a..2cca29316bd 100644 --- a/fs/gfs2/glops.c +++ b/fs/gfs2/glops.c @@ -47,10 +47,10 @@ static void __gfs2_ail_flush(struct gfs2_glock *gl) bd_ail_gl_list); bh = bd->bd_bh; gfs2_remove_from_ail(bd); - spin_unlock(&sdp->sd_ail_lock); - bd->bd_bh = NULL; bh->b_private = NULL; + spin_unlock(&sdp->sd_ail_lock); + bd->bd_blkno = bh->b_blocknr; gfs2_log_lock(sdp); gfs2_assert_withdraw(sdp, !buffer_busy(bh)); diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c index 903115f2bb3..85c62923ee2 100644 --- a/fs/gfs2/log.c +++ b/fs/gfs2/log.c @@ -903,6 +903,7 @@ void gfs2_meta_syncfs(struct gfs2_sbd *sdp) if (gfs2_ail1_empty(sdp)) break; } + gfs2_log_flush(sdp, NULL); } static inline int gfs2_jrnl_flush_reqd(struct gfs2_sbd *sdp) diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index ed540e7018b..fb0edf73548 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -757,13 +757,17 @@ static int gfs2_write_inode(struct inode *inode, struct writeback_control *wbc) struct timespec atime; struct gfs2_dinode *di; int ret = -EAGAIN; + int unlock_required = 0; /* Skip timestamp update, if this is from a memalloc */ if (current->flags & PF_MEMALLOC) goto do_flush; - ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh); - if (ret) - goto do_flush; + if (!gfs2_glock_is_locked_by_me(ip->i_gl)) { + ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh); + if (ret) + goto do_flush; + unlock_required = 1; + } ret = gfs2_trans_begin(sdp, RES_DINODE, 0); if (ret) goto do_unlock; @@ -780,7 +784,8 @@ static int gfs2_write_inode(struct inode *inode, struct writeback_control *wbc) } gfs2_trans_end(sdp); do_unlock: - gfs2_glock_dq_uninit(&gh); + if (unlock_required) + gfs2_glock_dq_uninit(&gh); do_flush: if (wbc->sync_mode == WB_SYNC_ALL) gfs2_log_flush(GFS2_SB(inode), ip->i_gl); @@ -1427,7 +1432,20 @@ out: return error; } -/* +/** + * gfs2_evict_inode - Remove an inode from cache + * @inode: The inode to evict + * + * There are three cases to consider: + * 1. i_nlink == 0, we are final opener (and must deallocate) + * 2. i_nlink == 0, we are not the final opener (and cannot deallocate) + * 3. i_nlink > 0 + * + * If the fs is read only, then we have to treat all cases as per #3 + * since we are unable to do any deallocation. The inode will be + * deallocated by the next read/write node to attempt an allocation + * in the same resource group + * * We have to (at the moment) hold the inodes main lock to cover * the gap between unlocking the shared lock on the iopen lock and * taking the exclusive lock. I'd rather do a shared -> exclusive @@ -1470,6 +1488,8 @@ static void gfs2_evict_inode(struct inode *inode) if (error) goto out_truncate; + /* Case 1 starts here */ + if (S_ISDIR(inode->i_mode) && (ip->i_diskflags & GFS2_DIF_EXHASH)) { error = gfs2_dir_exhash_dealloc(ip); @@ -1493,13 +1513,16 @@ static void gfs2_evict_inode(struct inode *inode) goto out_unlock; out_truncate: + /* Case 2 starts here */ error = gfs2_trans_begin(sdp, 0, sdp->sd_jdesc->jd_blocks); if (error) goto out_unlock; - gfs2_final_release_pages(ip); + /* Needs to be done before glock release & also in a transaction */ + truncate_inode_pages(&inode->i_data, 0); gfs2_trans_end(sdp); out_unlock: + /* Error path for case 1 */ if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) gfs2_glock_dq(&ip->i_iopen_gh); gfs2_holder_uninit(&ip->i_iopen_gh); @@ -1507,6 +1530,7 @@ out_unlock: if (error && error != GLR_TRYFAILED && error != -EROFS) fs_warn(sdp, "gfs2_evict_inode: %d\n", error); out: + /* Case 3 starts here */ truncate_inode_pages(&inode->i_data, 0); end_writeback(inode); -- cgit v1.2.3 From 1836750115f20b774e55c032a3893e8c5bdf41ed Mon Sep 17 00:00:00 2001 From: Al Viro Date: Tue, 12 Jul 2011 21:42:24 -0400 Subject: fix loop checks in d_materialise_unique() Both __d_unalias() and __d_materialise_dentry() need loop prevention. Grab rename_lock in caller, check for loops there... As a side benefit, we have dentry_lock_for_move() called only under rename_lock, which seriously reduces deadlock potential of the execrable "locking order" used for ->d_lock. Signed-off-by: Al Viro --- fs/dcache.c | 51 ++++++++++++++++++++++++++++++++++----------------- 1 file changed, 34 insertions(+), 17 deletions(-) (limited to 'fs') diff --git a/fs/dcache.c b/fs/dcache.c index 37f72ee5bf7..6e4ea6d8777 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -2213,14 +2213,15 @@ static void dentry_unlock_parents_for_move(struct dentry *dentry, * The hash value has to match the hash queue that the dentry is on.. */ /* - * d_move - move a dentry + * __d_move - move a dentry * @dentry: entry to move * @target: new dentry * * Update the dcache to reflect the move of a file name. Negative - * dcache entries should not be moved in this way. + * dcache entries should not be moved in this way. Caller hold + * rename_lock. */ -void d_move(struct dentry * dentry, struct dentry * target) +static void __d_move(struct dentry * dentry, struct dentry * target) { if (!dentry->d_inode) printk(KERN_WARNING "VFS: moving negative dcache entry\n"); @@ -2228,8 +2229,6 @@ void d_move(struct dentry * dentry, struct dentry * target) BUG_ON(d_ancestor(dentry, target)); BUG_ON(d_ancestor(target, dentry)); - write_seqlock(&rename_lock); - dentry_lock_for_move(dentry, target); write_seqcount_begin(&dentry->d_seq); @@ -2275,6 +2274,20 @@ void d_move(struct dentry * dentry, struct dentry * target) spin_unlock(&target->d_lock); fsnotify_d_move(dentry); spin_unlock(&dentry->d_lock); +} + +/* + * d_move - move a dentry + * @dentry: entry to move + * @target: new dentry + * + * Update the dcache to reflect the move of a file name. Negative + * dcache entries should not be moved in this way. + */ +void d_move(struct dentry *dentry, struct dentry *target) +{ + write_seqlock(&rename_lock); + __d_move(dentry, target); write_sequnlock(&rename_lock); } EXPORT_SYMBOL(d_move); @@ -2302,7 +2315,7 @@ struct dentry *d_ancestor(struct dentry *p1, struct dentry *p2) * This helper attempts to cope with remotely renamed directories * * It assumes that the caller is already holding - * dentry->d_parent->d_inode->i_mutex and the inode->i_lock + * dentry->d_parent->d_inode->i_mutex, inode->i_lock and rename_lock * * Note: If ever the locking in lock_rename() changes, then please * remember to update this too... @@ -2317,11 +2330,6 @@ static struct dentry *__d_unalias(struct inode *inode, if (alias->d_parent == dentry->d_parent) goto out_unalias; - /* Check for loops */ - ret = ERR_PTR(-ELOOP); - if (d_ancestor(alias, dentry)) - goto out_err; - /* See lock_rename() */ ret = ERR_PTR(-EBUSY); if (!mutex_trylock(&dentry->d_sb->s_vfs_rename_mutex)) @@ -2331,7 +2339,7 @@ static struct dentry *__d_unalias(struct inode *inode, goto out_err; m2 = &alias->d_parent->d_inode->i_mutex; out_unalias: - d_move(alias, dentry); + __d_move(alias, dentry); ret = alias; out_err: spin_unlock(&inode->i_lock); @@ -2416,15 +2424,24 @@ struct dentry *d_materialise_unique(struct dentry *dentry, struct inode *inode) alias = __d_find_alias(inode, 0); if (alias) { actual = alias; - /* Is this an anonymous mountpoint that we could splice - * into our tree? */ - if (IS_ROOT(alias)) { + write_seqlock(&rename_lock); + + if (d_ancestor(alias, dentry)) { + /* Check for loops */ + actual = ERR_PTR(-ELOOP); + } else if (IS_ROOT(alias)) { + /* Is this an anonymous mountpoint that we + * could splice into our tree? */ __d_materialise_dentry(dentry, alias); + write_sequnlock(&rename_lock); __d_drop(alias); goto found; + } else { + /* Nope, but we must(!) avoid directory + * aliasing */ + actual = __d_unalias(inode, dentry, alias); } - /* Nope, but we must(!) avoid directory aliasing */ - actual = __d_unalias(inode, dentry, alias); + write_sequnlock(&rename_lock); if (IS_ERR(actual)) dput(alias); goto out_nolock; -- cgit v1.2.3 From dc137bf553dbb6855bd7efc34fedcd03102455f7 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 16 Jul 2011 23:37:20 -0400 Subject: cifs: build_path_from_dentry() race fix deal with d_move() races properly; rename_lock read-retry loop, rcu_read_lock() held while walking to root, d_lock held over subtraction from namelen and copying the component to stabilize ->d_name. Signed-off-by: Al Viro --- fs/cifs/dir.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/dir.c b/fs/cifs/dir.c index 81914df47ef..fa8c21d913b 100644 --- a/fs/cifs/dir.c +++ b/fs/cifs/dir.c @@ -55,6 +55,7 @@ build_path_from_dentry(struct dentry *direntry) char dirsep; struct cifs_sb_info *cifs_sb = CIFS_SB(direntry->d_sb); struct cifs_tcon *tcon = cifs_sb_master_tcon(cifs_sb); + unsigned seq; if (direntry == NULL) return NULL; /* not much we can do if dentry is freed and @@ -68,22 +69,29 @@ build_path_from_dentry(struct dentry *direntry) dfsplen = 0; cifs_bp_rename_retry: namelen = dfsplen; + seq = read_seqbegin(&rename_lock); + rcu_read_lock(); for (temp = direntry; !IS_ROOT(temp);) { namelen += (1 + temp->d_name.len); temp = temp->d_parent; if (temp == NULL) { cERROR(1, "corrupt dentry"); + rcu_read_unlock(); return NULL; } } + rcu_read_unlock(); full_path = kmalloc(namelen+1, GFP_KERNEL); if (full_path == NULL) return full_path; full_path[namelen] = 0; /* trailing null */ + rcu_read_lock(); for (temp = direntry; !IS_ROOT(temp);) { + spin_lock(&temp->d_lock); namelen -= 1 + temp->d_name.len; if (namelen < 0) { + spin_unlock(&temp->d_lock); break; } else { full_path[namelen] = dirsep; @@ -91,14 +99,17 @@ cifs_bp_rename_retry: temp->d_name.len); cFYI(0, "name: %s", full_path + namelen); } + spin_unlock(&temp->d_lock); temp = temp->d_parent; if (temp == NULL) { cERROR(1, "corrupt dentry"); + rcu_read_unlock(); kfree(full_path); return NULL; } } - if (namelen != dfsplen) { + rcu_read_unlock(); + if (namelen != dfsplen || read_seqretry(&rename_lock, seq)) { cERROR(1, "did not end path lookup where expected namelen is %d", namelen); /* presumably this is only possible if racing with a rename -- cgit v1.2.3 From 1b71fe2efa31cd18c865db474a4cd473b6ab5281 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 16 Jul 2011 23:43:58 -0400 Subject: ceph analog of cifs build_path_from_dentry() race fix ... unfortunately, cifs bug got copied. Fix is essentially the same. Signed-off-by: Al Viro --- fs/ceph/mds_client.c | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c index 79743d146be..0c1d9175652 100644 --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -1438,12 +1438,15 @@ char *ceph_mdsc_build_path(struct dentry *dentry, int *plen, u64 *base, struct dentry *temp; char *path; int len, pos; + unsigned seq; if (dentry == NULL) return ERR_PTR(-EINVAL); retry: len = 0; + seq = read_seqbegin(&rename_lock); + rcu_read_lock(); for (temp = dentry; !IS_ROOT(temp);) { struct inode *inode = temp->d_inode; if (inode && ceph_snap(inode) == CEPH_SNAPDIR) @@ -1455,10 +1458,12 @@ retry: len += 1 + temp->d_name.len; temp = temp->d_parent; if (temp == NULL) { + rcu_read_unlock(); pr_err("build_path corrupt dentry %p\n", dentry); return ERR_PTR(-EINVAL); } } + rcu_read_unlock(); if (len) len--; /* no leading '/' */ @@ -1467,9 +1472,12 @@ retry: return ERR_PTR(-ENOMEM); pos = len; path[pos] = 0; /* trailing null */ + rcu_read_lock(); for (temp = dentry; !IS_ROOT(temp) && pos != 0; ) { - struct inode *inode = temp->d_inode; + struct inode *inode; + spin_lock(&temp->d_lock); + inode = temp->d_inode; if (inode && ceph_snap(inode) == CEPH_SNAPDIR) { dout("build_path path+%d: %p SNAPDIR\n", pos, temp); @@ -1478,21 +1486,26 @@ retry: break; } else { pos -= temp->d_name.len; - if (pos < 0) + if (pos < 0) { + spin_unlock(&temp->d_lock); break; + } strncpy(path + pos, temp->d_name.name, temp->d_name.len); } + spin_unlock(&temp->d_lock); if (pos) path[--pos] = '/'; temp = temp->d_parent; if (temp == NULL) { + rcu_read_unlock(); pr_err("build_path corrupt dentry\n"); kfree(path); return ERR_PTR(-EINVAL); } } - if (pos != 0) { + rcu_read_unlock(); + if (pos != 0 || read_seqretry(&rename_lock, seq)) { pr_err("build_path did not end path lookup where " "expected, namelen is %d, pos is %d\n", len, pos); /* presumably this is only possible if racing with a -- cgit v1.2.3 From a803b8067e317832d6a251c5b0486e36a4f81922 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Fri, 8 Jul 2011 20:56:55 -0400 Subject: fix exofs ->get_parent() NULL is not a possible return value for that method, TYVM... Signed-off-by: Al Viro --- fs/exofs/super.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/exofs/super.c b/fs/exofs/super.c index 06065bd37fc..c57beddcc21 100644 --- a/fs/exofs/super.c +++ b/fs/exofs/super.c @@ -913,7 +913,7 @@ struct dentry *exofs_get_parent(struct dentry *child) unsigned long ino = exofs_parent_ino(child); if (!ino) - return NULL; + return ERR_PTR(-ESTALE); return d_obtain_alias(exofs_iget(child->d_inode->i_sb, ino)); } -- cgit v1.2.3 From 642c937b4ed2e51d2f2e4c46ab7cd8b5bddf268b Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 17 Jul 2011 10:07:34 -0400 Subject: ufs should use d_splice_alias() it's NFS-exportable, so... Signed-off-by: Al Viro --- fs/ufs/namei.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/ufs/namei.c b/fs/ufs/namei.c index 29309e25417..b57aab9a118 100644 --- a/fs/ufs/namei.c +++ b/fs/ufs/namei.c @@ -56,16 +56,12 @@ static struct dentry *ufs_lookup(struct inode * dir, struct dentry *dentry, stru lock_ufs(dir->i_sb); ino = ufs_inode_by_name(dir, &dentry->d_name); - if (ino) { + if (ino) inode = ufs_iget(dir->i_sb, ino); - if (IS_ERR(inode)) { - unlock_ufs(dir->i_sb); - return ERR_CAST(inode); - } - } unlock_ufs(dir->i_sb); - d_add(dentry, inode); - return NULL; + if (IS_ERR(inode)) + return ERR_CAST(inode); + return d_splice_alias(inode, dentry); } /* -- cgit v1.2.3 From 0577d1ba411f9c40693b8b3e4aa7e0892cd03091 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 17 Jul 2011 19:04:14 -0400 Subject: cramfs: get_cramfs_inode() returns ERR_PTR() on failure ... and we want to report these failures in ->lookup() anyway. Signed-off-by: Al Viro --- fs/cramfs/inode.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) (limited to 'fs') diff --git a/fs/cramfs/inode.c b/fs/cramfs/inode.c index e141939080f..739fb59bcdc 100644 --- a/fs/cramfs/inode.c +++ b/fs/cramfs/inode.c @@ -37,7 +37,7 @@ static DEFINE_MUTEX(read_mutex); /* These macros may change in future, to provide better st_ino semantics. */ #define OFFSET(x) ((x)->i_ino) -static unsigned long cramino(struct cramfs_inode *cino, unsigned int offset) +static unsigned long cramino(const struct cramfs_inode *cino, unsigned int offset) { if (!cino->offset) return offset + 1; @@ -61,7 +61,7 @@ static unsigned long cramino(struct cramfs_inode *cino, unsigned int offset) } static struct inode *get_cramfs_inode(struct super_block *sb, - struct cramfs_inode *cramfs_inode, unsigned int offset) + const struct cramfs_inode *cramfs_inode, unsigned int offset) { struct inode *inode; static struct timespec zerotime; @@ -317,7 +317,7 @@ static int cramfs_fill_super(struct super_block *sb, void *data, int silent) /* Set it all up.. */ sb->s_op = &cramfs_ops; root = get_cramfs_inode(sb, &super.root, 0); - if (!root) + if (IS_ERR(root)) goto out; sb->s_root = d_alloc_root(root); if (!sb->s_root) { @@ -423,6 +423,7 @@ static int cramfs_readdir(struct file *filp, void *dirent, filldir_t filldir) static struct dentry * cramfs_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd) { unsigned int offset = 0; + struct inode *inode = NULL; int sorted; mutex_lock(&read_mutex); @@ -449,8 +450,8 @@ static struct dentry * cramfs_lookup(struct inode *dir, struct dentry *dentry, s for (;;) { if (!namelen) { - mutex_unlock(&read_mutex); - return ERR_PTR(-EIO); + inode = ERR_PTR(-EIO); + goto out; } if (name[namelen-1]) break; @@ -462,17 +463,18 @@ static struct dentry * cramfs_lookup(struct inode *dir, struct dentry *dentry, s if (retval > 0) continue; if (!retval) { - struct cramfs_inode entry = *de; - mutex_unlock(&read_mutex); - d_add(dentry, get_cramfs_inode(dir->i_sb, &entry, dir_off)); - return NULL; + inode = get_cramfs_inode(dir->i_sb, de, dir_off); + break; } /* else (retval < 0) */ if (sorted) break; } +out: mutex_unlock(&read_mutex); - d_add(dentry, NULL); + if (IS_ERR(inode)) + return ERR_CAST(inode); + d_add(dentry, inode); return NULL; } -- cgit v1.2.3 From 3cc0658e35124ace881f6942839dcae877c3eaed Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 17 Jul 2011 22:24:15 -0400 Subject: hppfs: fix dentry leak Signed-off-by: Al Viro --- fs/hppfs/hppfs.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/hppfs/hppfs.c b/fs/hppfs/hppfs.c index 87ed48e0343..7d6a0e92bcf 100644 --- a/fs/hppfs/hppfs.c +++ b/fs/hppfs/hppfs.c @@ -174,13 +174,11 @@ static struct dentry *hppfs_lookup(struct inode *ino, struct dentry *dentry, err = -ENOMEM; inode = get_inode(ino->i_sb, proc_dentry); if (!inode) - goto out_dput; + goto out; d_add(dentry, inode); return NULL; - out_dput: - dput(proc_dentry); out: return ERR_PTR(err); } @@ -690,8 +688,10 @@ static struct inode *get_inode(struct super_block *sb, struct dentry *dentry) struct inode *proc_ino = dentry->d_inode; struct inode *inode = new_inode(sb); - if (!inode) + if (!inode) { + dput(dentry); return ERR_PTR(-ENOMEM); + } if (S_ISDIR(dentry->d_inode->i_mode)) { inode->i_op = &hppfs_dir_iops; @@ -704,7 +704,7 @@ static struct inode *get_inode(struct super_block *sb, struct dentry *dentry) inode->i_fop = &hppfs_file_fops; } - HPPFS_I(inode)->proc_dentry = dget(dentry); + HPPFS_I(inode)->proc_dentry = dentry; inode->i_uid = proc_ino->i_uid; inode->i_gid = proc_ino->i_gid; @@ -737,7 +737,7 @@ static int hppfs_fill_super(struct super_block *sb, void *d, int silent) sb->s_fs_info = proc_mnt; err = -ENOMEM; - root_inode = get_inode(sb, proc_mnt->mnt_sb->s_root); + root_inode = get_inode(sb, dget(proc_mnt->mnt_sb->s_root)); if (!root_inode) goto out_mntput; -- cgit v1.2.3 From 0916a5e45fbd2604a303c8cc18e6b2b7c815e4c9 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 17 Jul 2011 22:27:22 -0400 Subject: hppfs_lookup(): don't open-code lookup_one_len() ... and it's getting it wrong, too - missing ->d_revalidate() calls when it's dealing with filesystem (procfs) that has non-trivial ->d_revalidate()... Signed-off-by: Al Viro --- fs/hppfs/hppfs.c | 19 +++---------------- 1 file changed, 3 insertions(+), 16 deletions(-) (limited to 'fs') diff --git a/fs/hppfs/hppfs.c b/fs/hppfs/hppfs.c index 7d6a0e92bcf..85c098a499f 100644 --- a/fs/hppfs/hppfs.c +++ b/fs/hppfs/hppfs.c @@ -139,7 +139,8 @@ static int file_removed(struct dentry *dentry, const char *file) static struct dentry *hppfs_lookup(struct inode *ino, struct dentry *dentry, struct nameidata *nd) { - struct dentry *proc_dentry, *new, *parent; + struct dentry *proc_dentry, *parent; + struct qstr *name = &dentry->d_name; struct inode *inode; int err, deleted; @@ -149,23 +150,9 @@ static struct dentry *hppfs_lookup(struct inode *ino, struct dentry *dentry, else if (deleted) return ERR_PTR(-ENOENT); - err = -ENOMEM; parent = HPPFS_I(ino)->proc_dentry; mutex_lock(&parent->d_inode->i_mutex); - proc_dentry = d_lookup(parent, &dentry->d_name); - if (proc_dentry == NULL) { - proc_dentry = d_alloc(parent, &dentry->d_name); - if (proc_dentry == NULL) { - mutex_unlock(&parent->d_inode->i_mutex); - goto out; - } - new = (*parent->d_inode->i_op->lookup)(parent->d_inode, - proc_dentry, NULL); - if (new) { - dput(proc_dentry); - proc_dentry = new; - } - } + proc_dentry = lookup_one_len(name->name, parent, name->len); mutex_unlock(&parent->d_inode->i_mutex); if (IS_ERR(proc_dentry)) -- cgit v1.2.3 From fec11dd9a0109fe52fd631e5c510778d6cbff6cc Mon Sep 17 00:00:00 2001 From: Al Viro Date: Mon, 18 Jul 2011 13:50:40 -0400 Subject: Fix cifs_get_root() Add missing ->i_mutex, convert to lookup_one_len() instead of (broken) open-coded analog, cope with getting something like a//b as relative pathname. Simplify the hell out of it, while we are there... Signed-off-by: Al Viro Reviewed-by: Jeff Layton --- fs/cifs/cifsfs.c | 100 ++++++++++++++++--------------------------------------- 1 file changed, 29 insertions(+), 71 deletions(-) (limited to 'fs') diff --git a/fs/cifs/cifsfs.c b/fs/cifs/cifsfs.c index 3e298997629..bc4b12ca537 100644 --- a/fs/cifs/cifsfs.c +++ b/fs/cifs/cifsfs.c @@ -35,6 +35,7 @@ #include #include #include +#include #include #include "cifsfs.h" #include "cifspdu.h" @@ -542,14 +543,12 @@ static const struct super_operations cifs_super_ops = { static struct dentry * cifs_get_root(struct smb_vol *vol, struct super_block *sb) { - int xid, rc; - struct inode *inode; - struct qstr name; - struct dentry *dparent = NULL, *dchild = NULL, *alias; + struct dentry *dentry; struct cifs_sb_info *cifs_sb = CIFS_SB(sb); - unsigned int i, full_len, len; - char *full_path = NULL, *pstart; + char *full_path = NULL; + char *s, *p; char sep; + int xid; full_path = cifs_build_path_to_root(vol, cifs_sb, cifs_sb_master_tcon(cifs_sb)); @@ -560,73 +559,32 @@ cifs_get_root(struct smb_vol *vol, struct super_block *sb) xid = GetXid(); sep = CIFS_DIR_SEP(cifs_sb); - dparent = dget(sb->s_root); - full_len = strlen(full_path); - full_path[full_len] = sep; - pstart = full_path + 1; - - for (i = 1, len = 0; i <= full_len; i++) { - if (full_path[i] != sep || !len) { - len++; - continue; - } - - full_path[i] = 0; - cFYI(1, "get dentry for %s", pstart); - - name.name = pstart; - name.len = len; - name.hash = full_name_hash(pstart, len); - dchild = d_lookup(dparent, &name); - if (dchild == NULL) { - cFYI(1, "not exists"); - dchild = d_alloc(dparent, &name); - if (dchild == NULL) { - dput(dparent); - dparent = ERR_PTR(-ENOMEM); - goto out; - } - } - - cFYI(1, "get inode"); - if (dchild->d_inode == NULL) { - cFYI(1, "not exists"); - inode = NULL; - if (cifs_sb_master_tcon(CIFS_SB(sb))->unix_ext) - rc = cifs_get_inode_info_unix(&inode, full_path, - sb, xid); - else - rc = cifs_get_inode_info(&inode, full_path, - NULL, sb, xid, NULL); - if (rc) { - dput(dchild); - dput(dparent); - dparent = ERR_PTR(rc); - goto out; - } - alias = d_materialise_unique(dchild, inode); - if (alias != NULL) { - dput(dchild); - if (IS_ERR(alias)) { - dput(dparent); - dparent = ERR_PTR(-EINVAL); /* XXX */ - goto out; - } - dchild = alias; - } - } - cFYI(1, "parent %p, child %p", dparent, dchild); - - dput(dparent); - dparent = dchild; - len = 0; - pstart = full_path + i + 1; - full_path[i] = sep; - } -out: + dentry = dget(sb->s_root); + p = s = full_path; + + do { + struct inode *dir = dentry->d_inode; + struct dentry *child; + + /* skip separators */ + while (*s == sep) + s++; + if (!*s) + break; + p = s++; + /* next separator */ + while (*s && *s != sep) + s++; + + mutex_lock(&dir->i_mutex); + child = lookup_one_len(p, dentry, s - p); + mutex_unlock(&dir->i_mutex); + dput(dentry); + dentry = child; + } while (!IS_ERR(dentry)); _FreeXid(xid); kfree(full_path); - return dparent; + return dentry; } static int cifs_set_super(struct super_block *sb, void *data) -- cgit v1.2.3 From 59430262401bec02d415179c43dbe5b8819c09ce Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Mon, 18 Jul 2011 15:43:29 -0700 Subject: vfs: fix race in rcu lookup of pruned dentry Don't update *inode in __follow_mount_rcu() until we'd verified that there is mountpoint there. Kudos to Hugh Dickins for catching that one in the first place and eventually figuring out the solution (and catching a braino in the earlier version of patch). Signed-off-by: Linus Torvalds Signed-off-by: Al Viro --- fs/namei.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/namei.c b/fs/namei.c index 5c867dd1c0b..14ab8d3f2f0 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -942,7 +942,6 @@ static bool __follow_mount_rcu(struct nameidata *nd, struct path *path, * Don't forget we might have a non-mountpoint managed dentry * that wants to block transit. */ - *inode = path->dentry->d_inode; if (unlikely(managed_dentry_might_block(path->dentry))) return false; @@ -955,6 +954,12 @@ static bool __follow_mount_rcu(struct nameidata *nd, struct path *path, path->mnt = mounted; path->dentry = mounted->mnt_root; nd->seq = read_seqcount_begin(&path->dentry->d_seq); + /* + * Update the inode too. We don't need to re-check the + * dentry sequence number here after this d_inode read, + * because a mount-point is always pinned. + */ + *inode = path->dentry->d_inode; } return true; } -- cgit v1.2.3 From f7b88631a89757d70192044c9d9f2e8d2fc02f2c Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Tue, 19 Jul 2011 08:49:25 -0700 Subject: fs/libfs.c: fix simple_attr_write() on 32bit machines Assume that /sys/kernel/debug/dummy64 is debugfs file created by debugfs_create_x64(). # cd /sys/kernel/debug # echo 0x1234567812345678 > dummy64 # cat dummy64 0x0000000012345678 # echo 0x80000000 > dummy64 # cat dummy64 0xffffffff80000000 A value larger than INT_MAX cannot be written to the debugfs file created by debugfs_create_u64 or debugfs_create_x64 on 32bit machine. Because simple_attr_write() uses simple_strtol() for the conversion. To fix this, use simple_strtoll() instead. Signed-off-by: Akinobu Mita Cc: Greg Kroah-Hartman Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/libfs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/libfs.c b/fs/libfs.c index c88eab55aec..275ca4749a2 100644 --- a/fs/libfs.c +++ b/fs/libfs.c @@ -822,7 +822,7 @@ ssize_t simple_attr_write(struct file *file, const char __user *buf, goto out; attr->set_buf[size] = '\0'; - val = simple_strtol(attr->set_buf, NULL, 0); + val = simple_strtoll(attr->set_buf, NULL, 0); ret = attr->set(attr->data, val); if (ret == 0) ret = len; /* on success, claim we got the whole input */ -- cgit v1.2.3 From 2cebaa58b7de775386732bbd6cd11c3f5b73faf0 Mon Sep 17 00:00:00 2001 From: Pavel Shilovsky Date: Wed, 20 Jul 2011 18:24:09 +0400 Subject: CIFS: Fix wrong length in cifs_iovec_read Signed-off-by: Pavel Shilovsky Reviewed-by: Jeff Layton Signed-off-by: Steve French --- fs/cifs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/file.c b/fs/cifs/file.c index bb71471a4d9..a9b4a24f2a1 100644 --- a/fs/cifs/file.c +++ b/fs/cifs/file.c @@ -1737,7 +1737,7 @@ cifs_iovec_read(struct file *file, const struct iovec *iov, io_parms.pid = pid; io_parms.tcon = pTcon; io_parms.offset = *poffset; - io_parms.length = len; + io_parms.length = cur_len; rc = CIFSSMBRead(xid, &io_parms, &bytes_read, &read_data, &buf_type); pSMBr = (struct smb_com_read_rsp *)read_data; -- cgit v1.2.3 From b307d4655a71749ac3f91c6dbe33d28cc026ceeb Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Thu, 21 Jul 2011 15:02:43 +0100 Subject: FS-Cache: Fix __fscache_uncache_all_inode_pages()'s outer loop The compiler, at least for ix86 and m68k, validly warns that the comparison: next <= (loff_t)-1 is always true (and it's always true also for x86-64 and probably all other arches - as long as pgoff_t isn't wider than loff_t). The intention appears to be to avoid wrapping of "next", so rather than eliminating the pointless comparison, fix the loop to indeed get exited when "next" would otherwise wrap. On m68k the following warning is observed: fs/fscache/page.c: In function '__fscache_uncache_all_inode_pages': fs/fscache/page.c:979: warning: comparison is always false due to limited range of data type Reported-by: Geert Uytterhoeven Reported-by: Jan Beulich Signed-off-by: Jan Beulich Signed-off-by: David Howells Cc: Suresh Jayaraman Cc: Geert Uytterhoeven Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- fs/fscache/page.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-) (limited to 'fs') diff --git a/fs/fscache/page.c b/fs/fscache/page.c index 2f343b4d7a7..3f7a59bfa7a 100644 --- a/fs/fscache/page.c +++ b/fs/fscache/page.c @@ -976,16 +976,12 @@ void __fscache_uncache_all_inode_pages(struct fscache_cookie *cookie, pagevec_init(&pvec, 0); next = 0; - while (next <= (loff_t)-1 && - pagevec_lookup(&pvec, mapping, next, PAGEVEC_SIZE) - ) { + do { + if (!pagevec_lookup(&pvec, mapping, next, PAGEVEC_SIZE)) + break; for (i = 0; i < pagevec_count(&pvec); i++) { struct page *page = pvec.pages[i]; - pgoff_t page_index = page->index; - - ASSERTCMP(page_index, >=, next); - next = page_index + 1; - + next = page->index; if (PageFsCache(page)) { __fscache_wait_on_page_write(cookie, page); __fscache_uncache_page(cookie, page); @@ -993,7 +989,7 @@ void __fscache_uncache_all_inode_pages(struct fscache_cookie *cookie, } pagevec_release(&pvec); cond_resched(); - } + } while (++next); _leave(""); } -- cgit v1.2.3 From b91da88fed84843313a1b6fd1b1c834a24bbcf9e Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Thu, 21 Jul 2011 11:01:42 -0700 Subject: vfs: drop conditional inode prefetch in __do_lookup_rcu It seems to hurt performance in real life. Yes, the inode will be used later, but the conditional doesn't seem to predict all that well (negative dentries are not uncommon) and it looks like the cost of prefetching is simply higher than depending on the cache doing the right thing. As usual. Signed-off-by: Linus Torvalds --- fs/dcache.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs') diff --git a/fs/dcache.c b/fs/dcache.c index 6e4ea6d8777..fbdcbca4072 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -1813,8 +1813,6 @@ seqretry: tname = dentry->d_name.name; i = dentry->d_inode; prefetch(tname); - if (i) - prefetch(i); /* * This seqcount check is required to ensure name and * len are loaded atomically, so as not to walk off the -- cgit v1.2.3