From c25da5b7baf1d243e6612ba2b97e2a2c4a1376f6 Mon Sep 17 00:00:00 2001 From: Nathan Huckleberry Date: Wed, 1 Feb 2023 17:23:48 -0800 Subject: [PATCH] dm verity: stop using WQ_UNBOUND for verify_wq Setting WQ_UNBOUND increases scheduler latency on ARM64. This is likely due to the asymmetric architecture of ARM64 processors. I've been unable to reproduce the results that claim WQ_UNBOUND gives a performance boost on x86-64. This flag is causing performance issues for multiple subsystems within Android. Notably, the same slowdown exists for decompression with EROFS. | open-prebuilt-camera | WQ_UNBOUND | ~WQ_UNBOUND | |-----------------------|------------|---------------| | verity wait time (us) | 11746 | 119 (-98%) | | erofs wait time (us) | 357805 | 174205 (-51%) | | sha256 ramdisk random read | WQ_UNBOUND | ~WQ_UNBOUND | |----------------------------|-----------=---|-------------| | arm64 (accelerated) | bw=42.4MiB/s | bw=212MiB/s | | arm64 (generic) | bw=16.5MiB/s | bw=48MiB/s | | x86_64 (generic) | bw=233MiB/s | bw=230MiB/s | Using a alloc_workqueue() @max_active arg of num_online_cpus() only made sense with WQ_UNBOUND. Switch the @max_active arg to 0 (aka default, which is 256 per-cpu). Also, eliminate 'wq_flags' since it really doesn't serve a purpose. Cc: Sami Tolvanen Cc: Eric Biggers Signed-off-by: Nathan Huckleberry Reviewed-by: Mikulas Patocka Signed-off-by: Mike Snitzer --- drivers/md/dm-verity-target.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/drivers/md/dm-verity-target.c b/drivers/md/dm-verity-target.c index ccf5b85..4c52b05 100644 --- a/drivers/md/dm-verity-target.c +++ b/drivers/md/dm-verity-target.c @@ -1162,7 +1162,6 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv) struct dm_verity_sig_opts verify_args = {0}; struct dm_arg_set as; unsigned int num; - unsigned int wq_flags; unsigned long long num_ll; int r; int i; @@ -1399,8 +1398,6 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv) goto bad; } - /* WQ_UNBOUND greatly improves performance when running on ramdisk */ - wq_flags = WQ_MEM_RECLAIM | WQ_UNBOUND; /* * Using WQ_HIGHPRI improves throughput and completion latency by * reducing wait times when reading from a dm-verity device. @@ -1410,8 +1407,7 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv) * will fall-back to using it for error handling (or if the bufio cache * doesn't have required hashes). */ - wq_flags |= WQ_HIGHPRI; - v->verify_wq = alloc_workqueue("kverityd", wq_flags, num_online_cpus()); + v->verify_wq = alloc_workqueue("kverityd", WQ_MEM_RECLAIM | WQ_HIGHPRI, 0); if (!v->verify_wq) { ti->error = "Cannot allocate workqueue"; r = -ENOMEM; -- 2.7.4