dm-bufio: align write boundary on physical block size

There may be devices with physical block size larger than 4k.

If dm-bufio sends I/O that is not aligned on physical block size,
performance is degraded.

The 4k minimum alignment limit is there because some SSDs report logical
and physical block size 512 despite having 4k internally - so dm-bufio
shouldn't send I/Os not aligned on 4k boundary, because they perform
badly (the SSD does read-modify-write for them).

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Reported-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: stable@vger.kernel.org
pull/1354/merge
Mikulas Patocka 2025-10-20 14:48:13 +02:00
parent ce51c6963a
commit d0ac06ae53
1 changed files with 6 additions and 4 deletions

View File

@ -1374,7 +1374,7 @@ static void submit_io(struct dm_buffer *b, enum req_op op, unsigned short ioprio
{ {
unsigned int n_sectors; unsigned int n_sectors;
sector_t sector; sector_t sector;
unsigned int offset, end; unsigned int offset, end, align;
b->end_io = end_io; b->end_io = end_io;
@ -1388,9 +1388,11 @@ static void submit_io(struct dm_buffer *b, enum req_op op, unsigned short ioprio
b->c->write_callback(b); b->c->write_callback(b);
offset = b->write_start; offset = b->write_start;
end = b->write_end; end = b->write_end;
offset &= -DM_BUFIO_WRITE_ALIGN; align = max(DM_BUFIO_WRITE_ALIGN,
end += DM_BUFIO_WRITE_ALIGN - 1; bdev_physical_block_size(b->c->bdev));
end &= -DM_BUFIO_WRITE_ALIGN; offset &= -align;
end += align - 1;
end &= -align;
if (unlikely(end > b->c->block_size)) if (unlikely(end > b->c->block_size))
end = b->c->block_size; end = b->c->block_size;