From: Dave Chinner Date: Tue, 20 Jul 2010 07:53:44 +0000 (+1000) Subject: xfs: fix xfs_trans_add_item() lockdep warnings X-Git-Url: https://git.stricted.de/?a=commitdiff_plain;h=438697064aaa2f64e0fcc6586582a3e7ec36005b;p=GitHub%2FLineageOS%2Fandroid_kernel_motorola_exynos9610.git xfs: fix xfs_trans_add_item() lockdep warnings xfs_trans_add_item() is called with ip->i_ilock held, which means it is unsafe for memory reclaim to recurse back into the filesystem (ilock is required in writeback). Hence the allocation needs to be KM_NOFS to avoid recursion. Lockdep report indicating memory allocation being called with the ip->i_ilock held is as follows: [ 1749.866796] ================================= [ 1749.867788] [ INFO: inconsistent lock state ] [ 1749.868327] 2.6.35-rc3-dgc+ #25 [ 1749.868741] --------------------------------- [ 1749.868741] inconsistent {IN-RECLAIM_FS-W} -> {RECLAIM_FS-ON-W} usage. [ 1749.868741] dd/2835 [HC0[0]:SC0[0]:HE1:SE1] takes: [ 1749.868741] (&(&ip->i_lock)->mr_lock){++++?.}, at: [] xfs_ilock+0x10b/0x190 [ 1749.868741] {IN-RECLAIM_FS-W} state was registered at: [ 1749.868741] [] __lock_acquire+0x437/0x1450 [ 1749.868741] [] lock_acquire+0xa6/0x160 [ 1749.868741] [] down_write_nested+0x65/0xb0 [ 1749.868741] [] xfs_ilock+0x10b/0x190 [ 1749.868741] [] xfs_reclaim_inode+0x99/0x310 [ 1749.868741] [] xfs_inode_ag_walk+0x8b/0x150 [ 1749.868741] [] xfs_inode_ag_iterator+0x8b/0xf0 [ 1749.868741] [] xfs_reclaim_inode_shrink+0x88/0x90 [ 1749.868741] [] shrink_slab+0x137/0x1a0 [ 1749.868741] [] balance_pgdat+0x421/0x6a0 [ 1749.868741] [] kswapd+0x11d/0x320 [ 1749.868741] [] kthread+0x96/0xa0 [ 1749.868741] [] kernel_thread_helper+0x4/0x10 [ 1749.868741] irq event stamp: 4234335 [ 1749.868741] hardirqs last enabled at (4234335): [] kmem_cache_free+0x115/0x220 [ 1749.868741] hardirqs last disabled at (4234334): [] kmem_cache_free+0x3d/0x220 [ 1749.868741] softirqs last enabled at (4233112): [] __do_softirq+0x142/0x260 [ 1749.868741] softirqs last disabled at (4233095): [] call_softirq+0x1c/0x50 [ 1749.868741] [ 1749.868741] other info that might help us debug this: [ 1749.868741] 2 locks held by dd/2835: [ 1749.868741] #0: (&(&ip->i_iolock)->mr_lock#2){+.+.+.}, at: [] xfs_ilock_nowait+0xed/0x200 [ 1749.868741] #1: (&(&ip->i_lock)->mr_lock){++++?.}, at: [] xfs_ilock+0x10b/0x190 [ 1749.868741] [ 1749.868741] stack backtrace: [ 1749.868741] Pid: 2835, comm: dd Not tainted 2.6.35-rc3-dgc+ #25 [ 1749.868741] Call Trace: [ 1749.868741] [] print_usage_bug+0x18a/0x190 [ 1749.868741] [] ? save_stack_trace+0x2f/0x50 [ 1749.868741] [] ? check_usage_backwards+0x0/0xf0 [ 1749.868741] [] mark_lock+0x331/0x400 [ 1749.868741] [] mark_held_locks+0x67/0x90 [ 1749.868741] [] lockdep_trace_alloc+0xa1/0xe0 [ 1749.868741] [] kmem_cache_alloc+0x39/0x1e0 [ 1749.868741] [] kmem_zone_alloc+0x94/0xe0 [ 1749.868741] [] kmem_zone_zalloc+0x1e/0x50 [ 1749.868741] [] xfs_trans_add_item+0x72/0xb0 [ 1749.868741] [] xfs_trans_ijoin+0xa1/0xd0 [ 1749.868741] [] xfs_itruncate_finish+0x312/0x5d0 [ 1749.868741] [] xfs_free_eofblocks+0x227/0x280 [ 1749.868741] [] xfs_release+0x138/0x190 [ 1749.868741] [] xfs_file_release+0x15/0x20 [ 1749.868741] [] fput+0x13f/0x260 [ 1749.868741] [] filp_close+0x52/0x80 [ 1749.868741] [] sys_close+0xb9/0x120 [ 1749.868741] [] system_call_fastpath+0x16/0x1b Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Alex Elder Signed-off-by: Dave Chinner --- diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c index f2065ccb6c2d..fdca7416c754 100644 --- a/fs/xfs/xfs_trans.c +++ b/fs/xfs/xfs_trans.c @@ -1134,7 +1134,7 @@ xfs_trans_add_item( ASSERT(lip->li_mountp = tp->t_mountp); ASSERT(lip->li_ailp = tp->t_mountp->m_ail); - lidp = kmem_zone_zalloc(xfs_log_item_desc_zone, KM_SLEEP); + lidp = kmem_zone_zalloc(xfs_log_item_desc_zone, KM_SLEEP | KM_NOFS); lidp->lid_item = lip; lidp->lid_flags = 0;