diff options
| author | Dave Chinner <dchinner@redhat.com> | 2012-04-23 05:58:56 (GMT) |
|---|---|---|
| committer | Ben Myers <bpm@sgi.com> | 2012-05-14 21:20:52 (GMT) |
| commit | aa5c158ec97bd4014f47a2bc0150fb6b20e6c48b (patch) | |
| tree | 8c79791167a5d659949952ad11a08ce325eb025c /fs/xfs/xfs_trans_buf.c | |
| parent | 7ca790a507a9288ebedab90a8e40b9afa8e4e949 (diff) | |
| download | linux-aa5c158ec97bd4014f47a2bc0150fb6b20e6c48b.tar.xz | |
xfs: kill XBF_DONTBLOCK
Just about all callers of xfs_buf_read() and xfs_buf_get() use XBF_DONTBLOCK.
This is used to make memory allocation use GFP_NOFS rather than GFP_KERNEL to
avoid recursion through memory reclaim back into the filesystem.
All the blocking get calls in growfs occur inside a transaction, even though
they are no part of the transaction, so all allocation will be GFP_NOFS due to
the task flag PF_TRANS being set. The blocking read calls occur during log
recovery, so they will probably be unaffected by converting to GFP_NOFS
allocations.
Hence make XBF_DONTBLOCK behaviour always occur for buffers and kill the flag.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
Diffstat (limited to 'fs/xfs/xfs_trans_buf.c')
| -rw-r--r-- | fs/xfs/xfs_trans_buf.c | 25 |
1 files changed, 4 insertions, 21 deletions
diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c index 5e4cf61..ccc6da1 100644 --- a/fs/xfs/xfs_trans_buf.c +++ b/fs/xfs/xfs_trans_buf.c @@ -148,8 +148,7 @@ xfs_trans_get_buf(xfs_trans_t *tp, * Default to a normal get_buf() call if the tp is NULL. */ if (tp == NULL) - return xfs_buf_get(target_dev, blkno, len, - flags | XBF_DONT_BLOCK); + return xfs_buf_get(target_dev, blkno, len, flags); /* * If we find the buffer in the cache with this transaction @@ -174,15 +173,7 @@ xfs_trans_get_buf(xfs_trans_t *tp, return (bp); } - /* - * We always specify the XBF_DONT_BLOCK flag within a transaction - * so that get_buf does not try to push out a delayed write buffer - * which might cause another transaction to take place (if the - * buffer was delayed alloc). Such recursive transactions can - * easily deadlock with our current transaction as well as cause - * us to run out of stack space. - */ - bp = xfs_buf_get(target_dev, blkno, len, flags | XBF_DONT_BLOCK); + bp = xfs_buf_get(target_dev, blkno, len, flags); if (bp == NULL) { return NULL; } @@ -283,7 +274,7 @@ xfs_trans_read_buf( * Default to a normal get_buf() call if the tp is NULL. */ if (tp == NULL) { - bp = xfs_buf_read(target, blkno, len, flags | XBF_DONT_BLOCK); + bp = xfs_buf_read(target, blkno, len, flags); if (!bp) return (flags & XBF_TRYLOCK) ? EAGAIN : XFS_ERROR(ENOMEM); @@ -367,15 +358,7 @@ xfs_trans_read_buf( return 0; } - /* - * We always specify the XBF_DONT_BLOCK flag within a transaction - * so that get_buf does not try to push out a delayed write buffer - * which might cause another transaction to take place (if the - * buffer was delayed alloc). Such recursive transactions can - * easily deadlock with our current transaction as well as cause - * us to run out of stack space. - */ - bp = xfs_buf_read(target, blkno, len, flags | XBF_DONT_BLOCK); + bp = xfs_buf_read(target, blkno, len, flags); if (bp == NULL) { *bpp = NULL; return (flags & XBF_TRYLOCK) ? |