2024-03-29 06:43:07

by Wu Bo

Subject: [PATCH] f2fs: fix 0 addr of multi devices when dio mapping

Consider of a f2fs system with 2 devices:
Info: Device[0] : /dev/block/dm-46 blkaddr = 0--3fffff
Info: Device[1] : /dev/block/dm-47 blkaddr = 400000--67fffff

f2fs_map_blocks will return logical addr of fs when doing buffered io:
f2fs_map_blocks: dev = (254,46), ino = 40020, file offset = 462865, start blkaddr = 0x400000

While dio mapping will return the addr of device:
f2fs_map_blocks: dev = (254,47), ino = 40020, file offset = 462865, start blkaddr = 0x0

So the addr=0 is valid when the map.m_bdev is not s_bdev.

Fixes: 8d3c1fa3fa5e ("f2fs: don't rely on F2FS_MAP_* in f2fs_iomap_begin")
Signed-off-by: Wu Bo <[email protected]>
---
fs/f2fs/data.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index d9494b5fc7c1..dca3628932c7 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -4185,7 +4185,8 @@ static int f2fs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
if (WARN_ON_ONCE(map.m_pblk == COMPRESS_ADDR))
return -EINVAL;

- if (map.m_pblk != NULL_ADDR) {
+ if (map.m_pblk != NULL_ADDR ||
+ (map.m_multidev_dio && map.m_bdev != inode->i_sb->s_bdev)) {
iomap->length = blks_to_bytes(inode, map.m_len);
iomap->type = IOMAP_MAPPED;
iomap->flags |= IOMAP_F_MERGED;
--
2.25.1

2024-03-29 10:27:36

by Chao Yu

Subject: Re: [PATCH] f2fs: fix 0 addr of multi devices when dio mapping

On 2024/3/29 14:55, Wu Bo wrote:
> Consider of a f2fs system with 2 devices:
> Info: Device[0] : /dev/block/dm-46 blkaddr = 0--3fffff
> Info: Device[1] : /dev/block/dm-47 blkaddr = 400000--67fffff
>
> f2fs_map_blocks will return logical addr of fs when doing buffered io:
> f2fs_map_blocks: dev = (254,46), ino = 40020, file offset = 462865, start blkaddr = 0x400000
>
> While dio mapping will return the addr of device:
> f2fs_map_blocks: dev = (254,47), ino = 40020, file offset = 462865, start blkaddr = 0x0
>
> So the addr=0 is valid when the map.m_bdev is not s_bdev.

Thanks for catching this, there is already a formal patch as below link:

https://lore.kernel.org/linux-f2fs-devel/[email protected]

Thanks,

>
> Fixes: 8d3c1fa3fa5e ("f2fs: don't rely on F2FS_MAP_* in f2fs_iomap_begin")
> Signed-off-by: Wu Bo <[email protected]>
> ---
> fs/f2fs/data.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> index d9494b5fc7c1..dca3628932c7 100644
> --- a/fs/f2fs/data.c
> +++ b/fs/f2fs/data.c
> @@ -4185,7 +4185,8 @@ static int f2fs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> if (WARN_ON_ONCE(map.m_pblk == COMPRESS_ADDR))
> return -EINVAL;
>
> - if (map.m_pblk != NULL_ADDR) {
> + if (map.m_pblk != NULL_ADDR ||
> + (map.m_multidev_dio && map.m_bdev != inode->i_sb->s_bdev)) {
> iomap->length = blks_to_bytes(inode, map.m_len);
> iomap->type = IOMAP_MAPPED;
> iomap->flags |= IOMAP_F_MERGED;