linux/fs/xfs
David Chinner 249a8c1124 [XFS] Move AIL pushing into it's own thread
When many hundreds to thousands of threads all try to do simultaneous
transactions and the log is in a tail-pushing situation (i.e. full), we
can get multiple threads walking the AIL list and contending on the AIL
lock.

The AIL push is, in effect, a simple I/O dispatch algorithm complicated by
the ordering constraints placed on it by the transaction subsystem. It
really does not need multiple threads to push on it - even when only a
single CPU is pushing the AIL, it can push the I/O out far faster that
pretty much any disk subsystem can handle.

So, to avoid contention problems stemming from multiple list walkers, move
the list walk off into another thread and simply provide a "target" to
push to. When a thread requires a push, it sets the target and wakes the
push thread, then goes to sleep waiting for the required amount of space
to become available in the log.

This mechanism should also be a lot fairer under heavy load as the waiters
will queue in arrival order, rather than queuing in "who completed a push
first" order.

Also, by moving the pushing to a separate thread we can do more
effectively overload detection and prevention as we can keep context from
loop iteration to loop iteration. That is, we can push only part of the
list each loop and not have to loop back to the start of the list every
time we run. This should also help by reducing the number of items we try
to lock and/or push items that we cannot move.

Note that this patch is not intended to solve the inefficiencies in the
AIL structure and the associated issues with extremely large list
contents. That needs to be addresses separately; parallel access would
cause problems to any new structure as well, so I'm only aiming to isolate
the structure from unbounded parallelism here.

SGI-PV: 972759
SGI-Modid: xfs-linux-melb:xfs-kern:30371a

Signed-off-by: David Chinner <dgc@sgi.com>
Signed-off-by: Lachlan McIlroy <lachlan@sgi.com>
2008-02-07 18:22:51 +11:00
..
linux-2.6 [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
quota [XFS] Remove the BPCSHIFT and NB* based macros from XFS. 2008-02-07 18:17:58 +11:00
support [XFS] lose xfs_hex_dump in favor of print_hex_dump 2008-02-07 18:13:05 +11:00
Kbuild kbuild/xfs: introduce fs/xfs/Kbuild 2006-01-09 20:48:03 +01:00
Kconfig [PATCH] BLOCK: Make it possible to disable the block layer [try #6] 2006-09-30 20:52:31 +02:00
Makefile [XFS] Sort out some cosmetic differences between XFS trees. 2005-09-05 11:47:01 +10:00
Makefile-linux-2.6 [XFS] kill xfs_iocore_t 2008-02-07 16:48:58 +11:00
xfs.h [XFS] clean up vnode/inode tracing 2008-02-07 16:42:19 +11:00
xfs_acl.c [XFS] use generic_permission 2008-02-07 18:22:38 +11:00
xfs_acl.h [XFS] use generic_permission 2008-02-07 18:22:38 +11:00
xfs_ag.h [XFS] Unwrap pagb_lock. 2008-02-07 16:46:39 +11:00
xfs_alloc.c [XFS] Remove spin.h 2008-02-07 16:47:45 +11:00
xfs_alloc.h [XFS] Lazy Superblock Counters 2007-07-14 15:28:50 +10:00
xfs_alloc_btree.c [XFS] Lazy Superblock Counters 2007-07-14 15:28:50 +10:00
xfs_alloc_btree.h [XFS] Remove unused arguments from the XFS_BTREE_*_ADDR macros. 2007-02-10 18:37:33 +11:00
xfs_arch.h [XFS] Merge in trivial changes, sync up headers with userspace 2006-01-12 10:29:53 +11:00
xfs_attr.c [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_attr.h [XFS] kill struct bhv_vnode 2007-10-16 11:40:24 +10:00
xfs_attr_leaf.c [XFS] Unwrap XFS_SB_LOCK. 2008-02-07 16:47:15 +11:00
xfs_attr_leaf.h [XFS] Add EA list callbacks for xfs kernel use. Cleanup some namespace 2006-09-28 11:01:37 +10:00
xfs_attr_sf.h [XFS] endianess annotations for xfs_attr_shortform_t 2006-03-17 17:29:25 +11:00
xfs_bit.c [XFS] Use the generic bitops rather than implementing them ourselves. 2008-02-07 18:14:22 +11:00
xfs_bit.h [XFS] Fix xfs_lowbit64 2008-02-07 18:19:41 +11:00
xfs_bmap.c [XFS] Remove the BPCSHIFT and NB* based macros from XFS. 2008-02-07 18:17:58 +11:00
xfs_bmap.h [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_bmap_btree.c [XFS] optimize XFS_IS_REALTIME_INODE w/o realtime config 2008-02-07 18:16:43 +11:00
xfs_bmap_btree.h [XFS] endianess annotations for xfs_bmbt_rec_t 2007-10-15 16:26:44 +10:00
xfs_btree.c [XFS] endianess annotations for xfs_bmbt_key Trivial as there are no 2006-09-28 10:58:17 +10:00
xfs_btree.h [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_buf_item.c [XFS] Unwrap AIL_LOCK 2008-02-07 16:44:23 +11:00
xfs_buf_item.h [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_clnt.h [XFS] Radix tree based inode caching 2007-10-15 16:50:50 +10:00
xfs_da_btree.c [XFS] Unwrap xfs_dabuf_global_lock 2008-02-07 16:46:48 +11:00
xfs_da_btree.h [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_dfrag.c [XFS] stop re-checking permissions in xfs_swapext 2008-02-07 18:22:24 +11:00
xfs_dfrag.h [XFS] Add parameters to xfs_bmapi() and xfs_bunmapi() to have them report 2006-06-09 14:48:12 +10:00
xfs_dinode.h [XFS] Remove CFORK macros and use code directly in IFORK and DFORK macros. 2008-02-07 18:19:24 +11:00
xfs_dir2.c [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_dir2.h [XFS] decontaminate vnode operations from behavior details 2007-10-15 16:54:29 +10:00
xfs_dir2_block.c [XFS] Put the correct offset in dirent d_off 2007-12-18 17:16:23 +11:00
xfs_dir2_block.h [XFS] use filldir internally 2007-10-15 16:49:49 +10:00
xfs_dir2_data.c [XFS] Radix tree based inode caching 2007-10-15 16:50:50 +10:00
xfs_dir2_data.h [XFS] Reduce shouting by removing unnecessary macros from dir2 code. 2007-07-14 15:37:02 +10:00
xfs_dir2_leaf.c [XFS] Put the correct offset in dirent d_off 2007-12-18 17:16:23 +11:00
xfs_dir2_leaf.h [XFS] use filldir internally 2007-10-15 16:49:49 +10:00
xfs_dir2_node.c [XFS] Radix tree based inode caching 2007-10-15 16:50:50 +10:00
xfs_dir2_node.h [XFS] Reduce shouting by removing unnecessary macros from dir2 code. 2007-07-14 15:37:02 +10:00
xfs_dir2_sf.c [XFS] Put the correct offset in dirent d_off 2007-12-18 17:16:23 +11:00
xfs_dir2_sf.h [XFS] use filldir internally 2007-10-15 16:49:49 +10:00
xfs_dir2_trace.c [XFS] Remove version 1 directory code. Never functioned on Linux, just 2006-06-20 13:04:51 +10:00
xfs_dir2_trace.h [XFS] Update license/copyright notices to match the prefered SGI 2005-11-02 14:58:39 +11:00
xfs_dmapi.h [XFS] kill the vfs_flags member in struct bhv_vfs 2007-10-16 11:45:57 +10:00
xfs_dmops.c [XFS] fixups after behavior removal merge into mainline git 2007-10-19 17:14:45 +10:00
xfs_error.c [XFS] lose xfs_hex_dump in favor of print_hex_dump 2008-02-07 18:13:05 +11:00
xfs_error.h [XFS] lose xfs_hex_dump in favor of print_hex_dump 2008-02-07 18:13:05 +11:00
xfs_extfree_item.c [XFS] Unwrap AIL_LOCK 2008-02-07 16:44:23 +11:00
xfs_extfree_item.h [XFS] cleanup the field types of some item format structures 2006-09-28 10:55:43 +10:00
xfs_filestream.c [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_filestream.h [XFS] Concurrent Multi-File Data Streams 2007-07-14 15:40:53 +10:00
xfs_fs.h [XFS] fix 32-bit compat ioctls for GETXFLAGS, SETXFLAGS, GETVERSION 2008-02-07 18:13:17 +11:00
xfs_fsops.c [XFS] Unwrap XFS_SB_LOCK. 2008-02-07 16:47:15 +11:00
xfs_fsops.h [XFS] Write log dummy record when freezing filesystem 2006-01-11 15:30:08 +11:00
xfs_ialloc.c [XFS] dinode endianess annotations 2007-10-15 16:48:30 +10:00
xfs_ialloc.h [XFS] Pick a single default inode cluster size. 2007-10-15 16:39:35 +10:00
xfs_ialloc_btree.c [XFS] use NULL for pointer initialisation instead of zero-cast-to-ptr 2006-09-28 10:58:40 +10:00
xfs_ialloc_btree.h [XFS] kill XFS_INOBT_IS_FREE_DISK 2008-02-07 18:12:41 +11:00
xfs_iget.c [XFS] Fix inode allocation latency 2008-02-07 18:16:07 +11:00
xfs_imap.h [XFS] Update license/copyright notices to match the prefered SGI 2005-11-02 14:58:39 +11:00
xfs_inode.c [XFS] use generic_permission 2008-02-07 18:22:38 +11:00
xfs_inode.h [XFS] use generic_permission 2008-02-07 18:22:38 +11:00
xfs_inode_item.c [XFS] Fix inode allocation latency 2008-02-07 18:16:07 +11:00
xfs_inode_item.h [XFS] cleanup the field types of some item format structures 2006-09-28 10:55:43 +10:00
xfs_inum.h [XFS] Update license/copyright notices to match the prefered SGI 2005-11-02 14:58:39 +11:00
xfs_iomap.c [XFS] optimize XFS_IS_REALTIME_INODE w/o realtime config 2008-02-07 18:16:43 +11:00
xfs_iomap.h [XFS] kill unnessecary ioops indirection 2008-02-07 16:44:14 +11:00
xfs_itable.c [XFS] Remove CFORK macros and use code directly in IFORK and DFORK macros. 2008-02-07 18:19:24 +11:00
xfs_itable.h [XFS] Fix XFS_IOC_FSBULKSTAT{,_SINGLE} & XFS_IOC_FSINUMBERS in compat mode 2007-07-14 15:42:50 +10:00
xfs_log.c [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
xfs_log.h [XFS] xlog_rec_header/xlog_rec_ext_header endianess annotations 2008-02-07 18:11:47 +11:00
xfs_log_priv.h [XFS] xlog_rec_header/xlog_rec_ext_header endianess annotations 2008-02-07 18:11:47 +11:00
xfs_log_recover.c [XFS] prevent panic during log recovery due to bogus op_hdr length 2008-02-07 18:20:58 +11:00
xfs_log_recover.h [XFS] Update license/copyright notices to match the prefered SGI 2005-11-02 14:58:39 +11:00
xfs_mount.c [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
xfs_mount.h [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
xfs_mru_cache.c [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_mru_cache.h [XFS] On-demand reaping of the MRU cache 2007-09-17 16:42:02 +10:00
xfs_qmops.c [XFS] Unwrap XFS_SB_LOCK. 2008-02-07 16:47:15 +11:00
xfs_quota.h [XFS] remove dependency of the quota module on behaviors 2007-10-16 11:43:26 +10:00
xfs_refcache.h [XFS] Cleanup cosmetic differences between source trees. 2005-11-03 16:14:31 +11:00
xfs_rename.c [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_rtalloc.c [XFS] Use the generic bitops rather than implementing them ourselves. 2008-02-07 18:14:22 +11:00
xfs_rtalloc.h [XFS] optimize XFS_IS_REALTIME_INODE w/o realtime config 2008-02-07 18:16:43 +11:00
xfs_rw.c [XFS] decontaminate vfs operations from behavior details 2007-10-16 11:43:55 +10:00
xfs_rw.h [XFS] optimize XFS_IS_REALTIME_INODE w/o realtime config 2008-02-07 18:16:43 +11:00
xfs_sb.h [XFS] superblock endianess annotations 2007-10-15 16:49:09 +10:00
xfs_trans.c [XFS] Unwrap AIL_LOCK 2008-02-07 16:44:23 +11:00
xfs_trans.h [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
xfs_trans_ail.c [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
xfs_trans_buf.c Revert "[XFS] Avoid replaying inode buffer initialisation log items if on-disk version is newer." 2007-10-01 07:59:03 -07:00
xfs_trans_extfree.c [XFS] Radix tree based inode caching 2007-10-15 16:50:50 +10:00
xfs_trans_inode.c [XFS] Remove version 1 directory code. Never functioned on Linux, just 2006-06-20 13:04:51 +10:00
xfs_trans_item.c [XFS] Fix up sparse warnings. 2008-02-07 18:14:38 +11:00
xfs_trans_priv.h [XFS] Move AIL pushing into it's own thread 2008-02-07 18:22:51 +11:00
xfs_trans_space.h [XFS] Remove version 1 directory code. Never functioned on Linux, just 2006-06-20 13:04:51 +10:00
xfs_types.h [XFS] use filldir internally 2007-10-15 16:49:49 +10:00
xfs_utils.c [XFS] Unwrap XFS_SB_LOCK. 2008-02-07 16:47:15 +11:00
xfs_utils.h [XFS] clean up vnode/inode tracing 2008-02-07 16:42:19 +11:00
xfs_vfsops.c [XFS] Cleanup various fid related bits: 2008-02-07 18:20:11 +11:00
xfs_vfsops.h [XFS] Cleanup various fid related bits: 2008-02-07 18:20:11 +11:00
xfs_vnodeops.c [XFS] use generic_permission 2008-02-07 18:22:38 +11:00
xfs_vnodeops.h [XFS] use generic_permission 2008-02-07 18:22:38 +11:00