From 486aff5e04823bb1e60c4045226eb244024b8420 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Thu, 24 Aug 2017 15:12:50 -0700 Subject: xfs: perform dax_device lookup at mount The ->iomap_begin() operation is a hot path, so cache the fs_dax_get_by_host() result at mount time to avoid the incurring the hash lookup overhead on a per-i/o basis. Reported-by: Christoph Hellwig Reviewed-by: Christoph Hellwig Reviewed-by: Darrick J. Wong Signed-off-by: Dan Williams --- fs/xfs/xfs_aops.c | 13 +++++++++++++ fs/xfs/xfs_aops.h | 1 + fs/xfs/xfs_buf.c | 4 +++- fs/xfs/xfs_buf.h | 3 ++- fs/xfs/xfs_iomap.c | 10 +--------- fs/xfs/xfs_super.c | 25 +++++++++++++++++++++---- 6 files changed, 41 insertions(+), 15 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c index 6bf120bb1a17..78185f3b10b2 100644 --- a/fs/xfs/xfs_aops.c +++ b/fs/xfs/xfs_aops.c @@ -80,6 +80,19 @@ xfs_find_bdev_for_inode( return mp->m_ddev_targp->bt_bdev; } +struct dax_device * +xfs_find_daxdev_for_inode( + struct inode *inode) +{ + struct xfs_inode *ip = XFS_I(inode); + struct xfs_mount *mp = ip->i_mount; + + if (XFS_IS_REALTIME_INODE(ip)) + return mp->m_rtdev_targp->bt_daxdev; + else + return mp->m_ddev_targp->bt_daxdev; +} + /* * We're now finished for good with this page. Update the page state via the * associated buffer_heads, paying attention to the start and end offsets that diff --git a/fs/xfs/xfs_aops.h b/fs/xfs/xfs_aops.h index cc174ec6c2fd..88c85ea63da0 100644 --- a/fs/xfs/xfs_aops.h +++ b/fs/xfs/xfs_aops.h @@ -59,5 +59,6 @@ int xfs_setfilesize(struct xfs_inode *ip, xfs_off_t offset, size_t size); extern void xfs_count_page_state(struct page *, int *, int *); extern struct block_device *xfs_find_bdev_for_inode(struct inode *); +extern struct dax_device *xfs_find_daxdev_for_inode(struct inode *); #endif /* __XFS_AOPS_H__ */ diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c index 72f038492ba8..6deb86c845d1 100644 --- a/fs/xfs/xfs_buf.c +++ b/fs/xfs/xfs_buf.c @@ -1802,7 +1802,8 @@ xfs_setsize_buftarg_early( xfs_buftarg_t * xfs_alloc_buftarg( struct xfs_mount *mp, - struct block_device *bdev) + struct block_device *bdev, + struct dax_device *dax_dev) { xfs_buftarg_t *btp; @@ -1811,6 +1812,7 @@ xfs_alloc_buftarg( btp->bt_mount = mp; btp->bt_dev = bdev->bd_dev; btp->bt_bdev = bdev; + btp->bt_daxdev = dax_dev; if (xfs_setsize_buftarg_early(btp, bdev)) goto error; diff --git a/fs/xfs/xfs_buf.h b/fs/xfs/xfs_buf.h index 20721261dae5..bf71507ddb16 100644 --- a/fs/xfs/xfs_buf.h +++ b/fs/xfs/xfs_buf.h @@ -108,6 +108,7 @@ typedef unsigned int xfs_buf_flags_t; typedef struct xfs_buftarg { dev_t bt_dev; struct block_device *bt_bdev; + struct dax_device *bt_daxdev; struct xfs_mount *bt_mount; unsigned int bt_meta_sectorsize; size_t bt_meta_sectormask; @@ -385,7 +386,7 @@ xfs_buf_update_cksum(struct xfs_buf *bp, unsigned long cksum_offset) * Handling of buftargs. */ extern xfs_buftarg_t *xfs_alloc_buftarg(struct xfs_mount *, - struct block_device *); + struct block_device *, struct dax_device *); extern void xfs_free_buftarg(struct xfs_mount *, struct xfs_buftarg *); extern void xfs_wait_buftarg(xfs_buftarg_t *); extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int); diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c index 813394c62849..7c934e407332 100644 --- a/fs/xfs/xfs_iomap.c +++ b/fs/xfs/xfs_iomap.c @@ -69,6 +69,7 @@ xfs_bmbt_to_iomap( iomap->offset = XFS_FSB_TO_B(mp, imap->br_startoff); iomap->length = XFS_FSB_TO_B(mp, imap->br_blockcount); iomap->bdev = xfs_find_bdev_for_inode(VFS_I(ip)); + iomap->dax_dev = xfs_find_daxdev_for_inode(VFS_I(ip)); } xfs_extlen_t @@ -976,7 +977,6 @@ xfs_file_iomap_begin( int nimaps = 1, error = 0; bool shared = false, trimmed = false; unsigned lockmode; - struct block_device *bdev; if (XFS_FORCED_SHUTDOWN(mp)) return -EIO; @@ -1087,13 +1087,6 @@ xfs_file_iomap_begin( xfs_bmbt_to_iomap(ip, iomap, &imap); - /* optionally associate a dax device with the iomap bdev */ - bdev = iomap->bdev; - if (blk_queue_dax(bdev->bd_queue)) - iomap->dax_dev = fs_dax_get_by_host(bdev->bd_disk->disk_name); - else - iomap->dax_dev = NULL; - if (shared) iomap->flags |= IOMAP_F_SHARED; return 0; @@ -1171,7 +1164,6 @@ xfs_file_iomap_end( unsigned flags, struct iomap *iomap) { - fs_put_dax(iomap->dax_dev); if ((flags & IOMAP_WRITE) && iomap->type == IOMAP_DELALLOC) return xfs_file_iomap_end_delalloc(XFS_I(inode), offset, length, written, iomap); diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c index 38aaacdbb8b3..ee4225c65f0c 100644 --- a/fs/xfs/xfs_super.c +++ b/fs/xfs/xfs_super.c @@ -714,17 +714,26 @@ STATIC void xfs_close_devices( struct xfs_mount *mp) { + struct dax_device *dax_ddev = mp->m_ddev_targp->bt_daxdev; + if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) { struct block_device *logdev = mp->m_logdev_targp->bt_bdev; + struct dax_device *dax_logdev = mp->m_logdev_targp->bt_daxdev; + xfs_free_buftarg(mp, mp->m_logdev_targp); xfs_blkdev_put(logdev); + fs_put_dax(dax_logdev); } if (mp->m_rtdev_targp) { struct block_device *rtdev = mp->m_rtdev_targp->bt_bdev; + struct dax_device *dax_rtdev = mp->m_rtdev_targp->bt_daxdev; + xfs_free_buftarg(mp, mp->m_rtdev_targp); xfs_blkdev_put(rtdev); + fs_put_dax(dax_rtdev); } xfs_free_buftarg(mp, mp->m_ddev_targp); + fs_put_dax(dax_ddev); } /* @@ -742,6 +751,8 @@ xfs_open_devices( struct xfs_mount *mp) { struct block_device *ddev = mp->m_super->s_bdev; + struct dax_device *dax_ddev = fs_dax_get_by_bdev(ddev); + struct dax_device *dax_logdev = NULL, *dax_rtdev = NULL; struct block_device *logdev = NULL, *rtdev = NULL; int error; @@ -752,6 +763,7 @@ xfs_open_devices( error = xfs_blkdev_get(mp, mp->m_logname, &logdev); if (error) goto out; + dax_logdev = fs_dax_get_by_bdev(logdev); } if (mp->m_rtname) { @@ -765,24 +777,25 @@ xfs_open_devices( error = -EINVAL; goto out_close_rtdev; } + dax_rtdev = fs_dax_get_by_bdev(rtdev); } /* * Setup xfs_mount buffer target pointers */ error = -ENOMEM; - mp->m_ddev_targp = xfs_alloc_buftarg(mp, ddev); + mp->m_ddev_targp = xfs_alloc_buftarg(mp, ddev, dax_ddev); if (!mp->m_ddev_targp) goto out_close_rtdev; if (rtdev) { - mp->m_rtdev_targp = xfs_alloc_buftarg(mp, rtdev); + mp->m_rtdev_targp = xfs_alloc_buftarg(mp, rtdev, dax_rtdev); if (!mp->m_rtdev_targp) goto out_free_ddev_targ; } if (logdev && logdev != ddev) { - mp->m_logdev_targp = xfs_alloc_buftarg(mp, logdev); + mp->m_logdev_targp = xfs_alloc_buftarg(mp, logdev, dax_logdev); if (!mp->m_logdev_targp) goto out_free_rtdev_targ; } else { @@ -798,10 +811,14 @@ xfs_open_devices( xfs_free_buftarg(mp, mp->m_ddev_targp); out_close_rtdev: xfs_blkdev_put(rtdev); + fs_put_dax(dax_rtdev); out_close_logdev: - if (logdev && logdev != ddev) + if (logdev && logdev != ddev) { xfs_blkdev_put(logdev); + fs_put_dax(dax_logdev); + } out: + fs_put_dax(dax_ddev); return error; } -- cgit v1.2.3 From 8cf037a8b22f71c6659bd0246bfaba9e79766f64 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Wed, 30 Aug 2017 09:43:34 -0700 Subject: ext2: perform dax_device lookup at mount The ->iomap_begin() operation is a hot path, so cache the fs_dax_get_by_host() result at mount time to avoid the incurring the hash lookup overhead on a per-i/o basis. Cc: "Theodore Ts'o" Cc: Andreas Dilger Reviewed-by: Jan Kara Reported-by: Christoph Hellwig Signed-off-by: Dan Williams --- fs/ext2/ext2.h | 1 + fs/ext2/inode.c | 11 +++-------- fs/ext2/super.c | 4 ++++ 3 files changed, 8 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/ext2/ext2.h b/fs/ext2/ext2.h index 23ebb92484c6..28de3edd4f4d 100644 --- a/fs/ext2/ext2.h +++ b/fs/ext2/ext2.h @@ -114,6 +114,7 @@ struct ext2_sb_info { */ spinlock_t s_lock; struct mb_cache *s_ea_block_cache; + struct dax_device *s_daxdev; }; static inline spinlock_t * diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c index 30163d007b2f..4dca6f348714 100644 --- a/fs/ext2/inode.c +++ b/fs/ext2/inode.c @@ -800,10 +800,10 @@ int ext2_get_block(struct inode *inode, sector_t iblock, static int ext2_iomap_begin(struct inode *inode, loff_t offset, loff_t length, unsigned flags, struct iomap *iomap) { - struct block_device *bdev; unsigned int blkbits = inode->i_blkbits; unsigned long first_block = offset >> blkbits; unsigned long max_blocks = (length + (1 << blkbits) - 1) >> blkbits; + struct ext2_sb_info *sbi = EXT2_SB(inode->i_sb); bool new = false, boundary = false; u32 bno; int ret; @@ -814,13 +814,9 @@ static int ext2_iomap_begin(struct inode *inode, loff_t offset, loff_t length, return ret; iomap->flags = 0; - bdev = inode->i_sb->s_bdev; - iomap->bdev = bdev; + iomap->bdev = inode->i_sb->s_bdev; iomap->offset = (u64)first_block << blkbits; - if (blk_queue_dax(bdev->bd_queue)) - iomap->dax_dev = fs_dax_get_by_host(bdev->bd_disk->disk_name); - else - iomap->dax_dev = NULL; + iomap->dax_dev = sbi->s_daxdev; if (ret == 0) { iomap->type = IOMAP_HOLE; @@ -842,7 +838,6 @@ static int ext2_iomap_end(struct inode *inode, loff_t offset, loff_t length, ssize_t written, unsigned flags, struct iomap *iomap) { - fs_put_dax(iomap->dax_dev); if (iomap->type == IOMAP_MAPPED && written < length && (flags & IOMAP_WRITE)) diff --git a/fs/ext2/super.c b/fs/ext2/super.c index 7b1bc9059863..fc18edd81815 100644 --- a/fs/ext2/super.c +++ b/fs/ext2/super.c @@ -171,6 +171,7 @@ static void ext2_put_super (struct super_block * sb) brelse (sbi->s_sbh); sb->s_fs_info = NULL; kfree(sbi->s_blockgroup_lock); + fs_put_dax(sbi->s_daxdev); kfree(sbi); } @@ -813,6 +814,7 @@ static unsigned long descriptor_loc(struct super_block *sb, static int ext2_fill_super(struct super_block *sb, void *data, int silent) { + struct dax_device *dax_dev = fs_dax_get_by_bdev(sb->s_bdev); struct buffer_head * bh; struct ext2_sb_info * sbi; struct ext2_super_block * es; @@ -842,6 +844,7 @@ static int ext2_fill_super(struct super_block *sb, void *data, int silent) } sb->s_fs_info = sbi; sbi->s_sb_block = sb_block; + sbi->s_daxdev = dax_dev; spin_lock_init(&sbi->s_lock); @@ -1200,6 +1203,7 @@ failed_sbi: kfree(sbi->s_blockgroup_lock); kfree(sbi); failed: + fs_put_dax(dax_dev); return ret; } -- cgit v1.2.3 From 5e405595e5bf4c09fab9ca1e7dbe5b62872757b5 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Thu, 24 Aug 2017 16:42:48 -0700 Subject: ext4: perform dax_device lookup at mount The ->iomap_begin() operation is a hot path, so cache the fs_dax_get_by_host() result at mount time to avoid the incurring the hash lookup overhead on a per-i/o basis. Cc: "Theodore Ts'o" Cc: Andreas Dilger Reviewed-by: Jan Kara Reported-by: Christoph Hellwig Signed-off-by: Dan Williams --- fs/ext4/ext4.h | 1 + fs/ext4/inode.c | 11 +++-------- fs/ext4/super.c | 4 ++++ 3 files changed, 8 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index a2bb7d2870e4..194e622dc3dd 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -1528,6 +1528,7 @@ struct ext4_sb_info { /* Barrier between changing inodes' journal flags and writepages ops. */ struct percpu_rw_semaphore s_journal_flag_rwsem; + struct dax_device *s_daxdev; }; static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb) diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index c774bdc22759..16424b5c4e88 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3404,7 +3404,7 @@ static int ext4_releasepage(struct page *page, gfp_t wait) static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length, unsigned flags, struct iomap *iomap) { - struct block_device *bdev; + struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb); unsigned int blkbits = inode->i_blkbits; unsigned long first_block = offset >> blkbits; unsigned long last_block = (offset + length - 1) >> blkbits; @@ -3473,12 +3473,8 @@ retry: } iomap->flags = 0; - bdev = inode->i_sb->s_bdev; - iomap->bdev = bdev; - if (blk_queue_dax(bdev->bd_queue)) - iomap->dax_dev = fs_dax_get_by_host(bdev->bd_disk->disk_name); - else - iomap->dax_dev = NULL; + iomap->bdev = inode->i_sb->s_bdev; + iomap->dax_dev = sbi->s_daxdev; iomap->offset = first_block << blkbits; if (ret == 0) { @@ -3511,7 +3507,6 @@ static int ext4_iomap_end(struct inode *inode, loff_t offset, loff_t length, int blkbits = inode->i_blkbits; bool truncate = false; - fs_put_dax(iomap->dax_dev); if (!(flags & IOMAP_WRITE) || (flags & IOMAP_FAULT)) return 0; diff --git a/fs/ext4/super.c b/fs/ext4/super.c index d61a70e2193a..3e58f952eddc 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -951,6 +951,7 @@ static void ext4_put_super(struct super_block *sb) if (sbi->s_chksum_driver) crypto_free_shash(sbi->s_chksum_driver); kfree(sbi->s_blockgroup_lock); + fs_put_dax(sbi->s_daxdev); kfree(sbi); } @@ -3377,6 +3378,7 @@ static void ext4_set_resv_clusters(struct super_block *sb) static int ext4_fill_super(struct super_block *sb, void *data, int silent) { + struct dax_device *dax_dev = fs_dax_get_by_bdev(sb->s_bdev); char *orig_data = kstrdup(data, GFP_KERNEL); struct buffer_head *bh; struct ext4_super_block *es = NULL; @@ -3399,6 +3401,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO; ext4_group_t first_not_zeroed; + sbi->s_daxdev = dax_dev; if ((data && !orig_data) || !sbi) goto out_free_base; @@ -4378,6 +4381,7 @@ out_fail: out_free_base: kfree(sbi); kfree(orig_data); + fs_put_dax(dax_dev); return err ? err : ret; } -- cgit v1.2.3 From aed9eb1b21e85a846c805bc299bbb9d039d4a95b Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Tue, 5 Sep 2017 17:51:23 +0100 Subject: ext4: fix null pointer dereference on sbi In the case of a kzalloc failure when allocating sbi we end up with a null pointer dereference on sbi when assigning sbi->s_daxdev. Fix this by moving the assignment of sbi->s_daxdev to after the null pointer check of sbi. Detected by CoverityScan CID#1455379 ("Dereference before null check") Fixes: 5e405595e5bf ("ext4: perform dax_device lookup at mount") Signed-off-by: Colin Ian King Signed-off-by: Dan Williams --- fs/ext4/super.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 3e58f952eddc..55772b2d05ee 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -3401,10 +3401,10 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO; ext4_group_t first_not_zeroed; - sbi->s_daxdev = dax_dev; if ((data && !orig_data) || !sbi) goto out_free_base; + sbi->s_daxdev = dax_dev; sbi->s_blockgroup_lock = kzalloc(sizeof(struct blockgroup_lock), GFP_KERNEL); if (!sbi->s_blockgroup_lock) -- cgit v1.2.3