writeback: pay attention to wbc->nr_to_write in write_cache_pages
authorDave Chinner <dchinner@redhat.com>
Wed, 9 Jun 2010 00:37:18 +0000 (10:37 +1000)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 9 Jun 2010 01:12:44 +0000 (18:12 -0700)
commit0b5649278e39a068aaf91399941bab1b4a4a3cc2
tree3fd2c782385137f5b135c07149de772e207fdaf8
parent8d7458daea2a6809d32418bf489b949d23de99ea
writeback: pay attention to wbc->nr_to_write in write_cache_pages

If a filesystem writes more than one page in ->writepage, write_cache_pages
fails to notice this and continues to attempt writeback when wbc->nr_to_write
has gone negative - this trace was captured from XFS:

    wbc_writeback_start: towrt=1024
    wbc_writepage: towrt=1024
    wbc_writepage: towrt=0
    wbc_writepage: towrt=-1
    wbc_writepage: towrt=-5
    wbc_writepage: towrt=-21
    wbc_writepage: towrt=-85

This has adverse effects on filesystem writeback behaviour. write_cache_pages()
needs to terminate after a certain number of pages are written, not after a
certain number of calls to ->writepage are made.  This is a regression
introduced by 17bc6c30cf6bfffd816bdc53682dd46fc34a2cf4 ("vfs: Add
no_nrwrite_index_update writeback control flag"), but cannot be reverted
directly due to subsequent bug fixes that have gone in on top of it.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/writeback.h
include/trace/events/ext4.h
mm/page-writeback.c