diff options
| author | Eric Whitney <enwlinux@gmail.com> | 2021-01-13 17:14:03 -0500 |
|---|---|---|
| committer | Theodore Ts'o <tytso@mit.edu> | 2021-02-08 18:03:56 -0500 |
| commit | 3258386aba670e3406a499d2d0b7395e14c8d097 (patch) | |
| tree | 5e87e98a11a72b4a7443a273bd4034a782793488 /fs/ext4/extents.c | |
| parent | b5776e7524afbd4569978ff790864755c438bba7 (diff) | |
| download | linux-3258386aba670e3406a499d2d0b7395e14c8d097.tar.gz | |
ext4: reset retry counter when ext4_alloc_file_blocks() makes progress
Change the retry policy in ext4_alloc_file_blocks() to allow for a full
retry cycle whenever a portion of an allocation request has been
fulfilled. A large allocation request often results in multiple calls
to ext4_map_blocks(), each of which is potentially subject to a
temporary ENOSPC condition and retry cycle. The current code only
allows for a single retry cycle.
This patch does not address a known bug or reported complaint.
However, it should make block allocation for fallocate and zero range
more robust.
In addition, simplify the conditional controlling the allocation while
loop, where testing len alone is sufficient. Remove the assignment to
ret2 in the error path after the call to ext4_map_blocks() since its
value isn't subsequently used.
Signed-off-by: Eric Whitney <enwlinux@gmail.com>
Link: https://lore.kernel.org/r/20210113221403.18258-1-enwlinux@gmail.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Diffstat (limited to 'fs/ext4/extents.c')
| -rw-r--r-- | fs/ext4/extents.c | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index 3960b7ec3ab770..77c7c8a54da737 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -4382,8 +4382,7 @@ static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset, { struct inode *inode = file_inode(file); handle_t *handle; - int ret = 0; - int ret2 = 0, ret3 = 0; + int ret, ret2 = 0, ret3 = 0; int retries = 0; int depth = 0; struct ext4_map_blocks map; @@ -4408,7 +4407,7 @@ static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset, depth = ext_depth(inode); retry: - while (ret >= 0 && len) { + while (len) { /* * Recalculate credits when extent tree depth changes. */ @@ -4430,9 +4429,13 @@ retry: inode->i_ino, map.m_lblk, map.m_len, ret); ext4_mark_inode_dirty(handle, inode); - ret2 = ext4_journal_stop(handle); + ext4_journal_stop(handle); break; } + /* + * allow a full retry cycle for any remaining allocations + */ + retries = 0; map.m_lblk += ret; map.m_len = len = len - ret; epos = (loff_t)map.m_lblk << inode->i_blkbits; @@ -4450,11 +4453,8 @@ retry: if (unlikely(ret2)) break; } - if (ret == -ENOSPC && - ext4_should_retry_alloc(inode->i_sb, &retries)) { - ret = 0; + if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries)) goto retry; - } return ret > 0 ? ret2 : ret; } |