[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20191226171731.GE3158@mit.edu>
Date: Thu, 26 Dec 2019 12:17:31 -0500
From: "Theodore Y. Ts'o" <tytso@....edu>
To: Jan Kara <jack@...e.cz>
Cc: Ritesh Harjani <riteshh@...ux.ibm.com>, linux-ext4@...r.kernel.org,
Dan Williams <dan.j.williams@...el.com>,
"Berrocal, Eduardo" <eduardo.berrocal@...el.com>
Subject: Re: [PATCH] ext4: Optimize ext4 DIO overwrites
On Thu, Dec 19, 2019 at 08:28:23PM +0100, Jan Kara wrote:
> > However depending on which patch lands first one may need a
> > re-basing. Will conflict with this-
> > https://marc.info/?l=linux-ext4&m=157613016931238&w=2
>
> Yes, but the conflict is minor and trivial to resolve.
>
Is this the correct resolution?
--- a/fs/ext4/file.c
+++ b/fs/ext4/file.c
@@ -447,6 +447,7 @@ static ssize_t ext4_dio_write_iter(struct kiocb *iocb, struct iov_iter *from)
struct inode *inode = file_inode(iocb->ki_filp);
loff_t offset = iocb->ki_pos;
size_t count = iov_iter_count(from);
+ const struct iomap_ops *iomap_ops = &ext4_iomap_ops;
bool extend = false, unaligned_io = false;
bool ilock_shared = true;
@@ -526,7 +527,9 @@ static ssize_t ext4_dio_write_iter(struct kiocb *iocb, struct iov_iter *from)
ext4_journal_stop(handle);
}
- ret = iomap_dio_rw(iocb, from, &ext4_iomap_ops, &ext4_dio_write_ops,
+ if (ilock_shared)
+ iomap_ops = &ext4_iomap_overwrite_ops;
+ ret = iomap_dio_rw(iocb, from, iomap_ops, &ext4_dio_write_ops,
is_sync_kiocb(iocb) || unaligned_io || extend);
if (extend)
- Ted
Powered by blists - more mailing lists