diff options
| author | Arnd Bergmann <arnd@arndb.de> | 2012-11-12 22:18:43 +0100 |
|---|---|---|
| committer | Arnd Bergmann <arnd@arndb.de> | 2012-11-13 10:14:38 +0100 |
| commit | b5932cc839d809a07a43b93555ccc3fbabc2a766 (patch) | |
| tree | 528d7a0b3353c08ec7461dee80d7d3151061dc07 /fs/xfs/xfs_log.c | |
| parent | a5434fc9e2ad9bac545f058a24b284c051a1faae (diff) | |
| parent | 16cf8a80a8f0f4757427b17cdfb6c4897674db68 (diff) | |
Merge branch 'cleanups/dma' into next/cleanup
Separate patches from Marek Szyprowski <m.szyprowski@samsung.com>:
Commit e9da6e9905e639b0 ("ARM: dma-mapping: remove custom consistent dma
region") replaced custom consistent memory handling, so setting
consistent dma memory size is not longer required. This patch series
cleans sub-architecture platform code to remove all calls to the
obsolated init_consistent_dma_size() function and finally removes the
init_consistent_dma_size() stub itself.
* cleanups/dma:
ARM: at91: remove obsoleted init_consistent_dma_size()
ARM: u300: remove obsoleted init_consistent_dma_size()
ARM: dma-mapping: remove init_consistent_dma_size() stub
ARM: shmobile: remove obsoleted init_consistent_dma_size()
ARM: davinci: remove obsoleted init_consistent_dma_size()
ARM: samsung: remove obsoleted init_consistent_dma_size()
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Diffstat (limited to 'fs/xfs/xfs_log.c')
| -rw-r--r-- | fs/xfs/xfs_log.c | 19 |
1 files changed, 16 insertions, 3 deletions
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c index 7f4f9370d0e..4dad756962d 100644 --- a/fs/xfs/xfs_log.c +++ b/fs/xfs/xfs_log.c @@ -2387,14 +2387,27 @@ xlog_state_do_callback( /* - * update the last_sync_lsn before we drop the + * Completion of a iclog IO does not imply that + * a transaction has completed, as transactions + * can be large enough to span many iclogs. We + * cannot change the tail of the log half way + * through a transaction as this may be the only + * transaction in the log and moving th etail to + * point to the middle of it will prevent + * recovery from finding the start of the + * transaction. Hence we should only update the + * last_sync_lsn if this iclog contains + * transaction completion callbacks on it. + * + * We have to do this before we drop the * icloglock to ensure we are the only one that * can update it. */ ASSERT(XFS_LSN_CMP(atomic64_read(&log->l_last_sync_lsn), be64_to_cpu(iclog->ic_header.h_lsn)) <= 0); - atomic64_set(&log->l_last_sync_lsn, - be64_to_cpu(iclog->ic_header.h_lsn)); + if (iclog->ic_callback) + atomic64_set(&log->l_last_sync_lsn, + be64_to_cpu(iclog->ic_header.h_lsn)); } else ioerrors++; |