| Message ID | 20240410142948.2817554-28-yi.zhang@huaweicloud.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | ext4: use iomap for regular file's buffered IO path and enable large folio | expand |
On Wed, Apr 10, 2024 at 10:29:41PM +0800, Zhang Yi wrote: > From: Zhang Yi <yi.zhang@huawei.com> > > Add ext4_iomap_zero_range() for the zero_range iomap path, it zero out > the mapped blocks, all work have been done in iomap_zero_range(), so > call it directly. > > Signed-off-by: Zhang Yi <yi.zhang@huawei.com> > --- > fs/ext4/inode.c | 9 +++++++++ > 1 file changed, 9 insertions(+) > > diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c > index 9d694c780007..5af3b8acf1b9 100644 > --- a/fs/ext4/inode.c > +++ b/fs/ext4/inode.c > @@ -4144,6 +4144,13 @@ static int __ext4_block_zero_page_range(handle_t *handle, > return err; > } > > +static int ext4_iomap_zero_range(struct inode *inode, > + loff_t from, loff_t length) > +{ > + return iomap_zero_range(inode, from, length, NULL, > + &ext4_iomap_buffered_read_ops); > +} Zeroing is a buffered write operation, not a buffered read operation. It runs though iomap_write_begin(), so needs all the stale iomap detection stuff to be set up for correct operation. -Dave.
On 2024/5/1 17:40, Dave Chinner wrote: > On Wed, Apr 10, 2024 at 10:29:41PM +0800, Zhang Yi wrote: >> From: Zhang Yi <yi.zhang@huawei.com> >> >> Add ext4_iomap_zero_range() for the zero_range iomap path, it zero out >> the mapped blocks, all work have been done in iomap_zero_range(), so >> call it directly. >> >> Signed-off-by: Zhang Yi <yi.zhang@huawei.com> >> --- >> fs/ext4/inode.c | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c >> index 9d694c780007..5af3b8acf1b9 100644 >> --- a/fs/ext4/inode.c >> +++ b/fs/ext4/inode.c >> @@ -4144,6 +4144,13 @@ static int __ext4_block_zero_page_range(handle_t *handle, >> return err; >> } >> >> +static int ext4_iomap_zero_range(struct inode *inode, >> + loff_t from, loff_t length) >> +{ >> + return iomap_zero_range(inode, from, length, NULL, >> + &ext4_iomap_buffered_read_ops); >> +} > > Zeroing is a buffered write operation, not a buffered read > operation. It runs though iomap_write_begin(), so needs all the > stale iomap detection stuff to be set up for correct operation. > Yeah, right, thanks for point this out. Although we can guarantee that the zeroing is a partial block overwrite and no need to allocate new blocks on ext4, use ext4_iomap_buffered_read_ops is not appropriate, I'll use write ops instead. Thanks, Yi.
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 9d694c780007..5af3b8acf1b9 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -4144,6 +4144,13 @@ static int __ext4_block_zero_page_range(handle_t *handle, return err; } +static int ext4_iomap_zero_range(struct inode *inode, + loff_t from, loff_t length) +{ + return iomap_zero_range(inode, from, length, NULL, + &ext4_iomap_buffered_read_ops); +} + /* * ext4_block_zero_page_range() zeros out a mapping of length 'length' * starting from file offset 'from'. The range to be zero'd must @@ -4169,6 +4176,8 @@ static int ext4_block_zero_page_range(handle_t *handle, if (IS_DAX(inode)) { return dax_zero_range(inode, from, length, NULL, &ext4_iomap_ops); + } else if (ext4_test_inode_state(inode, EXT4_STATE_BUFFERED_IOMAP)) { + return ext4_iomap_zero_range(inode, from, length); } return __ext4_block_zero_page_range(handle, mapping, from, length); }