The following patches update the below patches in the patch queue.
ext4_truncate_block_allocated_on_a_failed_ext4_write_begin.patch
ext4_i_disksize_lock_race_fix.patch
The change is to move some parts of the patch from ext4_i_disksize_lock_race_fix.patch
to ext4_truncate_block_allocated_on_a_failed_ext4_write_begin.patch
From: Aneesh Kumar K.V <[email protected]>
For blocksize < pagesize we need to remove blocks that
got allocte in block_wirte_begin if we fail with ENOSPC
for later blocks. block_write_begin internally does
this if it allocated page locally. This make sure
we don't have blocks outisde inode.i_size during
ENOSPC
Signed-off-by: Aneesh Kumar K.V <[email protected]>
---
fs/ext4/inode.c | 14 ++++++++++++++
1 files changed, 14 insertions(+), 0 deletions(-)
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 48534be..e7c84e4 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1391,6 +1391,13 @@ static int ext4_write_begin(struct file *file, struct address_space *mapping,
unlock_page(page);
ext4_journal_stop(handle);
page_cache_release(page);
+ /*
+ * block_write_begin may have instantiated a few blocks
+ * outside i_size. Trim these off again. Don't need
+ * i_size_read because we hold i_mutex.
+ */
+ if (pos + len > inode->i_size)
+ vmtruncate(inode, inode->i_size);
}
if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
@@ -2560,6 +2567,13 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
unlock_page(page);
ext4_journal_stop(handle);
page_cache_release(page);
+ /*
+ * block_write_begin may have instantiated a few blocks
+ * outside i_size. Trim these off again. Don't need
+ * i_size_read because we hold i_mutex.
+ */
+ if (pos + len > inode->i_size)
+ vmtruncate(inode, inode->i_size);
}
if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
From: Aneesh Kumar K.V <[email protected]>
With delayed allocation we use i_data_sem to update
i_disksize. We need to update i_disksize only
if the new size specified is greater than the
current value and we need to make sure we don't
race with other i_disksize update. With delayed
allocation we will switch to nondelalloc write_begin
if we are low on free blocks. That means nondelalloc
write_begin also need to use the same locking.
We also need to check and update i_disksize even
if the new size is less that inode.i_size because
of delayed allocation.
Signed-off-by: Aneesh Kumar K.V <[email protected]>
---
fs/ext4/ext4.h | 11 +++++++++++
fs/ext4/extents.c | 9 +++++----
fs/ext4/inode.c | 54 +++++++++++++++++++++++++++++------------------------
3 files changed, 46 insertions(+), 28 deletions(-)
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 79bdcd8..9fb7796 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1219,6 +1219,17 @@ do { \
#define EXT4_FREEBLOCKS_WATERMARK 0
#endif
+static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
+{
+ /*
+ * XXX: replace with spinlock if seen contended -bzzz
+ */
+ down_write(&EXT4_I(inode)->i_data_sem);
+ if (newsize > EXT4_I(inode)->i_disksize)
+ EXT4_I(inode)->i_disksize = newsize;
+ up_write(&EXT4_I(inode)->i_data_sem);
+ return ;
+}
/*
* Inodes and files operations
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index b24d3c5..cbf388a 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -2877,10 +2877,11 @@ static void ext4_falloc_update_inode(struct inode *inode,
* Update only when preallocation was requested beyond
* the file size.
*/
- if (!(mode & FALLOC_FL_KEEP_SIZE) &&
- new_size > i_size_read(inode)) {
- i_size_write(inode, new_size);
- EXT4_I(inode)->i_disksize = new_size;
+ if (!(mode & FALLOC_FL_KEEP_SIZE)) {
+ if (new_size > i_size_read(inode))
+ i_size_write(inode, new_size);
+ if (new_size > EXT4_I(inode)->i_disksize)
+ ext4_update_i_disksize(inode, new_size);
}
}
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index e7c84e4..a06a4a8 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1434,16 +1434,18 @@ static int ext4_ordered_write_end(struct file *file,
ret = ext4_jbd2_file_inode(handle, inode);
if (ret == 0) {
- /*
- * generic_write_end() will run mark_inode_dirty() if i_size
- * changes. So let's piggyback the i_disksize mark_inode_dirty
- * into that.
- */
loff_t new_i_size;
new_i_size = pos + copied;
- if (new_i_size > EXT4_I(inode)->i_disksize)
- EXT4_I(inode)->i_disksize = new_i_size;
+ if (new_i_size > EXT4_I(inode)->i_disksize) {
+ ext4_update_i_disksize(inode, new_i_size);
+ /* We need to mark inode dirty even if
+ * new_i_size is less that inode->i_size
+ * bu greater than i_disksize.(hint delalloc)
+ */
+ ext4_mark_inode_dirty(handle, inode);
+ }
+
ret2 = generic_write_end(file, mapping, pos, len, copied,
page, fsdata);
copied = ret2;
@@ -1468,8 +1470,14 @@ static int ext4_writeback_write_end(struct file *file,
loff_t new_i_size;
new_i_size = pos + copied;
- if (new_i_size > EXT4_I(inode)->i_disksize)
- EXT4_I(inode)->i_disksize = new_i_size;
+ if (new_i_size > EXT4_I(inode)->i_disksize) {
+ ext4_update_i_disksize(inode, new_i_size);
+ /* We need to mark inode dirty even if
+ * new_i_size is less that inode->i_size
+ * bu greater than i_disksize.(hint delalloc)
+ */
+ ext4_mark_inode_dirty(handle, inode);
+ }
ret2 = generic_write_end(file, mapping, pos, len, copied,
page, fsdata);
@@ -1494,6 +1502,7 @@ static int ext4_journalled_write_end(struct file *file,
int ret = 0, ret2;
int partial = 0;
unsigned from, to;
+ loff_t new_i_size;
from = pos & (PAGE_CACHE_SIZE - 1);
to = from + len;
@@ -1508,11 +1517,12 @@ static int ext4_journalled_write_end(struct file *file,
to, &partial, write_end_fn);
if (!partial)
SetPageUptodate(page);
- if (pos+copied > inode->i_size)
+ new_i_size = pos + copied;
+ if (new_i_size > inode->i_size)
i_size_write(inode, pos+copied);
EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
- if (inode->i_size > EXT4_I(inode)->i_disksize) {
- EXT4_I(inode)->i_disksize = inode->i_size;
+ if (new_i_size > EXT4_I(inode)->i_disksize) {
+ ext4_update_i_disksize(inode, new_i_size);
ret2 = ext4_mark_inode_dirty(handle, inode);
if (!ret)
ret = ret2;
@@ -2227,18 +2237,9 @@ static int ext4_da_get_block_write(struct inode *inode, sector_t iblock,
if (disksize > i_size_read(inode))
disksize = i_size_read(inode);
if (disksize > EXT4_I(inode)->i_disksize) {
- /*
- * XXX: replace with spinlock if seen contended -bzzz
- */
- down_write(&EXT4_I(inode)->i_data_sem);
- if (disksize > EXT4_I(inode)->i_disksize)
- EXT4_I(inode)->i_disksize = disksize;
- up_write(&EXT4_I(inode)->i_data_sem);
Hi,
I guess git send-email did something wrong. I had these two patches as
followup for the compose mail with this subject. Any how i am attaching
the patches below. The patches update
ext4_truncate_block_allocated_on_a_failed_ext4_write_begin.patch and
ext4_i_disksize_lock_race_fix.patch by moving some changes from
ext4_i_disksize_lock_race_fix.patch to the other.
-aneesh
On Fri, Sep 12, 2008 at 02:37:23PM +0530, Aneesh Kumar K.V wrote:
>
> I guess git send-email did something wrong. I had these two patches as
> followup for the compose mail with this subject. Any how i am attaching
> the patches below. The patches update
> ext4_truncate_block_allocated_on_a_failed_ext4_write_begin.patch and
> ext4_i_disksize_lock_race_fix.patch by moving some changes from
> ext4_i_disksize_lock_race_fix.patch to the other.
Thanks, applied. I merged back in the patch description improvements
I had made, and made a few more to improve the English description of
these two patches.
- Ted