This is an alrady known issue that dm-thin volume cannot be used as swap, otherwise a deadlock may happen when dm-thin internal memory demond triggers swap I/O on the dm-thin volume itself.
Thanks to Mikulas Patocka for commit a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device"), this method can also be used for dm-thin to avoid the recursive I/O when it is used as swap.
This patch just simply sets ti->limit_swap_bios by tree in pool_ctr() and thin_ctr(), other important stuffs are already done by Patocka in the above mentioned commit.
In my test, I create a dm-thin volume /dev/vg/swap and use it as swap device. Then I run fio on another dm-thin volume /dev/vg/main and use large --blocksize to trigger swap I/O onto /dev/vg/swap.
The following fio command line is used in my test, fio --name recursive-swap-io --lockmem 1 --iodepth 128 \ --ioengine libaio --filename /dev/vg/main --rw randrw \ --blocksize 1M --numjobs 32 --time_based --runtime=12h
Without the patch, the whole system can be locked up within 15 seconds.
With this patch, there is no any deadlock or hang task observed after 2 hours fio running.
Further more, I change --blocksize from 1M to 128M, around 30 seconds after fio running, no I/O rate displayed by fio, and the out-of-memory killer message shows up in kernel message. After around 20 minutes all fio processes are killed and the whole system backs to be alive.
This is exactly what is expected when recursive I/O happens on dm-thin volume when it is used as swap.
Note: this change depends on commit a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device")
Signed-off-by: Coly Li colyli@suse.de Cc: Mikulas Patocka mpatocka@redhat.com Cc: Mike Snitzer snitzer@kernel.org Cc: stable@vger.kernel.org --- Changelog, v2: fix typo in commit log. v1: initial version.
drivers/md/dm-thin.c | 2 ++ 1 file changed, 2 insertions(+)
--- a/drivers/md/dm-thin.c +++ b/drivers/md/dm-thin.c @@ -3355,6 +3355,7 @@ static int pool_ctr(struct dm_target *ti pt->low_water_blocks = low_water_blocks; pt->adjusted_pf = pt->requested_pf = pf; ti->num_flush_bios = 1; + ti->limit_swap_bios = true;
/* * Only need to enable discards if the pool should pass @@ -4233,6 +4234,7 @@ static int thin_ctr(struct dm_target *ti goto bad;
ti->num_flush_bios = 1; + ti->limit_swap_bios = true; ti->flush_supported = true; ti->accounts_remapped_io = true; ti->per_io_data_size = sizeof(struct dm_thin_endio_hook);