[v2,02/11] md: add infra for nr_pending changes

Message ID	065afbfd34f44eada6575fd48114e782@kioxia.com (mailing list archive)
State	Changes Requested, archived
Headers	show Received: from mo-csw-fb.securemx.jp (mo-csw-fb1120.securemx.jp [210.130.202.128]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 947C910A0E; Thu, 18 Apr 2024 07:21:40 +0000 (UTC) From: tada keisuke <keisuke1.tada@kioxia.com> To: "song@kernel.org" <song@kernel.org>, "yukuai3@huawei.com" <yukuai3@huawei.com> CC: "linux-raid@vger.kernel.org" <linux-raid@vger.kernel.org>, "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org> Subject: [PATCH v2 02/11] md: add infra for nr_pending changes Thread-Topic: [PATCH v2 02/11] md: add infra for nr_pending changes Thread-Index: AdqLx9x1MkCN7uTlRyqnTzQSCFkVuA== Date: Thu, 18 Apr 2024 05:44:01 +0000 Message-ID: <065afbfd34f44eada6575fd48114e782@kioxia.com> Accept-Language: ja-JP, en-US Content-Language: ja-JP Content-Type: text/plain; charset="iso-2022-jp" Content-Transfer-Encoding: quoted-printable Precedence: bulk MIME-Version: 1.0 X-CrossPremisesHeadersFilteredBySendConnector: CNN1EXMB04.r1.kioxia.com
Series	[v2,01/11] md: add infra for active_aligned_reads changes \| expand [v2,01/11] md: add infra for active_aligned_reads changes [v2,02/11] md: add infra for nr_pending changes [v2,03/11] md: workaround for inconsistency of config state in takeover [v2,04/11] md: minimize execution of zero check for nr_pending [v2,05/11] md: match the type of variables to percpu_ref [v2,06/11] md: avoid conflicts in active_aligned_reads operations [v2,07/11] md: change the type of nr_pending from atomic_t to percpu_ref [v2,08/11] md: add atomic mode switching in RAID 1/10 [v2,09/11] md: add atomic mode switching when removing disk [v2,10/11] md: add atomic mode switching when I/O completion [v2,11/11] md: avoid conflicts in nr_pending operations

diff --git a/drivers/md/md-bitmap.c b/drivers/md/md-bitmap.c index 059afc24c08b..6d49a6e4cd2f 100644 --- a/drivers/md/md-bitmap.c +++ b/drivers/md/md-bitmap.c @@ -190,7 +190,7 @@ static struct md_rdev *next_active_rdev(struct md_rdev *rdev, struct mddev *mdde if (rdev->raid_disk >= 0 && !test_bit(Faulty, &rdev->flags)) { /* this is a usable devices */ - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); rcu_read_unlock(); return rdev; } diff --git a/drivers/md/md.c b/drivers/md/md.c index 00bbafcd27bb..d53ca23bacf7 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -575,7 +575,7 @@ static void submit_flushes(struct work_struct *ws) !test_bit(Faulty, &rdev->flags)) { struct bio *bi; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); rcu_read_unlock(); bi = bio_alloc_bioset(rdev->bdev, 0, REQ_OP_WRITE | REQ_PREFLUSH, @@ -1054,7 +1054,7 @@ void md_super_write(struct mddev *mddev, struct md_rdev *rdev, | REQ_PREFLUSH | REQ_FUA, GFP_NOIO, &mddev->sync_set); - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); bio->bi_iter.bi_sector = sector; __bio_add_page(bio, page, size, 0); @@ -9269,7 +9269,7 @@ static bool rdev_removeable(struct md_rdev *rdev) return false; /* There are still inflight io, don't remove this rdev. */ - if (atomic_read(&rdev->nr_pending)) + if (nr_pending_is_not_zero(rdev)) return false; /* diff --git a/drivers/md/md.h b/drivers/md/md.h index 029dd0491a36..f57a894b924b 100644 --- a/drivers/md/md.h +++ b/drivers/md/md.h @@ -211,6 +211,36 @@ enum flag_bits { Nonrot, /* non-rotational device (SSD) */ }; +static inline void nr_pending_inc(struct md_rdev *rdev) +{ + atomic_inc(&rdev->nr_pending); +} + +static inline void nr_pending_dec(struct md_rdev *rdev) +{ + atomic_dec(&rdev->nr_pending); +} + +static inline bool nr_pending_is_zero(struct md_rdev *rdev) +{ + return atomic_read(&rdev->nr_pending) == 0; +} + +static inline bool nr_pending_is_not_zero(struct md_rdev *rdev) +{ + return atomic_read(&rdev->nr_pending) != 0; +} + +static inline unsigned int nr_pending_read(struct md_rdev *rdev) +{ + return atomic_read(&rdev->nr_pending); +} + +static inline bool nr_pending_dec_and_test(struct md_rdev *rdev) +{ + return atomic_dec_and_test(&rdev->nr_pending); +} + static inline int is_badblock(struct md_rdev *rdev, sector_t s, int sectors, sector_t *first_bad, int *bad_sectors) { @@ -846,7 +876,7 @@ static inline bool is_rdev_broken(struct md_rdev *rdev) static inline void rdev_dec_pending(struct md_rdev *rdev, struct mddev *mddev) { int faulty = test_bit(Faulty, &rdev->flags); - if (atomic_dec_and_test(&rdev->nr_pending) && faulty) { + if (nr_pending_dec_and_test(rdev) && faulty) { set_bit(MD_RECOVERY_NEEDED, &mddev->recovery); md_wakeup_thread(mddev->thread); } diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index be8ac24f50b6..962cebbba6cd 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -581,7 +581,7 @@ static void update_read_sectors(struct r1conf *conf, int disk, { struct raid1_info *info = &conf->mirrors[disk]; - atomic_inc(&info->rdev->nr_pending); + nr_pending_inc(info->rdev); if (info->next_seq_sect != this_sector) info->seq_start = this_sector; info->next_seq_sect = this_sector + len; @@ -784,7 +784,7 @@ static int choose_best_rdev(struct r1conf *conf, struct r1bio *r1_bio) if (ctl.readable_disks++ == 1) set_bit(R1BIO_FailFast, &r1_bio->state); - pending = atomic_read(&rdev->nr_pending); + pending = nr_pending_read(rdev); dist = abs(r1_bio->sector - conf->mirrors[disk].head_position); /* Don't change to another disk for sequential reads */ @@ -1495,7 +1495,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio, write_behind = true; if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); blocked_rdev = rdev; break; } @@ -1506,7 +1506,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio, continue; } - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); if (test_bit(WriteErrorSeen, &rdev->flags)) { sector_t first_bad; int bad_sectors; @@ -1879,7 +1879,7 @@ static bool raid1_remove_conf(struct r1conf *conf, int disk) struct md_rdev *rdev = info->rdev; if (!rdev || test_bit(In_sync, &rdev->flags) || - atomic_read(&rdev->nr_pending)) + nr_pending_is_not_zero(rdev)) return false; /* Only remove non-faulty devices if recovery is not possible. */ @@ -1987,7 +1987,7 @@ static int raid1_remove_disk(struct mddev *mddev, struct md_rdev *rdev) struct md_rdev *repl = conf->mirrors[conf->raid_disks + number].rdev; freeze_array(conf, 0); - if (atomic_read(&repl->nr_pending)) { + if (nr_pending_is_not_zero(repl)) { /* It means that some queued IO of retry_list * hold repl. Thus, we cannot set replacement * as NULL, avoiding rdev NULL pointer @@ -2403,7 +2403,7 @@ static void fix_read_error(struct r1conf *conf, struct r1bio *r1_bio) (!test_bit(Faulty, &rdev->flags) && rdev->recovery_offset >= sect + s)) && rdev_has_badblock(rdev, sect, s) == 0) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); if (sync_page_io(rdev, sect, s<<9, conf->tmppage, REQ_OP_READ, false)) success = 1; @@ -2433,7 +2433,7 @@ static void fix_read_error(struct r1conf *conf, struct r1bio *r1_bio) rdev = conf->mirrors[d].rdev; if (rdev && !test_bit(Faulty, &rdev->flags)) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); r1_sync_page_io(rdev, sect, s, conf->tmppage, REQ_OP_WRITE); rdev_dec_pending(rdev, mddev); @@ -2447,7 +2447,7 @@ static void fix_read_error(struct r1conf *conf, struct r1bio *r1_bio) rdev = conf->mirrors[d].rdev; if (rdev && !test_bit(Faulty, &rdev->flags)) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); if (r1_sync_page_io(rdev, sect, s, conf->tmppage, REQ_OP_READ)) { atomic_add(s, &rdev->corrected_errors); @@ -2909,7 +2909,7 @@ static sector_t raid1_sync_request(struct mddev *mddev, sector_t sector_nr, } } if (rdev && bio->bi_end_io) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); bio->bi_iter.bi_sector = sector_nr + rdev->data_offset; bio_set_dev(bio, rdev->bdev); if (test_bit(FailFast, &rdev->flags)) diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index a4556d2e46bf..2b1238befd8a 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -808,7 +808,7 @@ static struct md_rdev *read_balance(struct r10conf *conf, nonrot = bdev_nonrot(rdev->bdev); has_nonrot_disk |= nonrot; - pending = atomic_read(&rdev->nr_pending); + pending = nr_pending_read(rdev); if (min_pending > pending && nonrot) { min_pending = pending; best_pending_slot = slot; @@ -849,7 +849,7 @@ static struct md_rdev *read_balance(struct r10conf *conf, } if (slot >= 0) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); r10_bio->read_slot = slot; } else rdev = NULL; @@ -1296,12 +1296,12 @@ static void wait_blocked_dev(struct mddev *mddev, struct r10bio *r10_bio) rdev = conf->mirrors[i].rdev; rrdev = conf->mirrors[i].replacement; if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); blocked_rdev = rdev; break; } if (rrdev && unlikely(test_bit(Blocked, &rrdev->flags))) { - atomic_inc(&rrdev->nr_pending); + nr_pending_inc(rrdev); blocked_rdev = rrdev; break; } @@ -1322,7 +1322,7 @@ static void wait_blocked_dev(struct mddev *mddev, struct r10bio *r10_bio) * Mustn't write here until the bad block * is acknowledged */ - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); set_bit(BlockedBadBlocks, &rdev->flags); blocked_rdev = rdev; break; @@ -1467,11 +1467,11 @@ static void raid10_write_request(struct mddev *mddev, struct bio *bio, } if (rdev) { r10_bio->devs[i].bio = bio; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); } if (rrdev) { r10_bio->devs[i].repl_bio = bio; - atomic_inc(&rrdev->nr_pending); + nr_pending_inc(rrdev); } } @@ -1731,11 +1731,11 @@ static int raid10_handle_discard(struct mddev *mddev, struct bio *bio) if (rdev) { r10_bio->devs[disk].bio = bio; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); } if (rrdev) { r10_bio->devs[disk].repl_bio = bio; - atomic_inc(&rrdev->nr_pending); + nr_pending_inc(rrdev); } } @@ -2155,7 +2155,7 @@ static int raid10_remove_disk(struct mddev *mddev, struct md_rdev *rdev) return 0; if (test_bit(In_sync, &rdev->flags) || - atomic_read(&rdev->nr_pending)) { + nr_pending_is_not_zero(rdev)) { err = -EBUSY; goto abort; } @@ -2394,7 +2394,7 @@ static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio) bio_copy_data(tbio, fbio); - atomic_inc(&conf->mirrors[d].rdev->nr_pending); + nr_pending_inc(conf->mirrors[d].rdev); atomic_inc(&r10_bio->remaining); md_sync_acct(conf->mirrors[d].rdev->bdev, bio_sectors(tbio)); @@ -2552,12 +2552,12 @@ static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio) */ d = r10_bio->devs[1].devnum; if (wbio->bi_end_io) { - atomic_inc(&conf->mirrors[d].rdev->nr_pending); + nr_pending_inc(conf->mirrors[d].rdev); md_sync_acct(conf->mirrors[d].rdev->bdev, bio_sectors(wbio)); submit_bio_noacct(wbio); } if (wbio2) { - atomic_inc(&conf->mirrors[d].replacement->nr_pending); + nr_pending_inc(conf->mirrors[d].replacement); md_sync_acct(conf->mirrors[d].replacement->bdev, bio_sectors(wbio2)); submit_bio_noacct(wbio2); @@ -2633,7 +2633,7 @@ static void fix_read_error(struct r10conf *conf, struct mddev *mddev, struct r10 rdev_has_badblock(rdev, r10_bio->devs[sl].addr + sect, s) == 0) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); success = sync_page_io(rdev, r10_bio->devs[sl].addr + sect, @@ -2682,7 +2682,7 @@ static void fix_read_error(struct r10conf *conf, struct mddev *mddev, struct r10 !test_bit(In_sync, &rdev->flags)) continue; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); if (r10_sync_page_io(rdev, r10_bio->devs[sl].addr + sect, @@ -2714,7 +2714,7 @@ static void fix_read_error(struct r10conf *conf, struct mddev *mddev, struct r10 !test_bit(In_sync, &rdev->flags)) continue; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); switch (r10_sync_page_io(rdev, r10_bio->devs[sl].addr + sect, @@ -3342,9 +3342,9 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr, continue; } if (mrdev) - atomic_inc(&mrdev->nr_pending); + nr_pending_inc(mrdev); if (mreplace) - atomic_inc(&mreplace->nr_pending); + nr_pending_inc(mreplace); r10_bio = raid10_alloc_init_r10buf(conf); r10_bio->state = 0; @@ -3413,7 +3413,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr, bio->bi_iter.bi_sector = from_addr + rdev->data_offset; bio_set_dev(bio, rdev->bdev); - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); /* and we write to 'i' (if not in_sync) */ for (k=0; k<conf->copies; k++) @@ -3601,7 +3601,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr, continue; } } - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); atomic_inc(&r10_bio->remaining); bio->bi_next = biolist; biolist = bio; @@ -3617,7 +3617,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr, if (rdev == NULL || test_bit(Faulty, &rdev->flags)) continue; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); /* Need to set up for writing to the replacement */ bio = r10_bio->devs[i].repl_bio; @@ -4906,7 +4906,7 @@ static void reshape_request_write(struct mddev *mddev, struct r10bio *r10_bio) if (!rdev || test_bit(Faulty, &rdev->flags)) continue; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); md_sync_acct_bio(b, r10_bio->sectors); atomic_inc(&r10_bio->remaining); b->bi_next = NULL; @@ -4986,7 +4986,7 @@ static int handle_reshape_read_error(struct mddev *mddev, goto failed; addr = r10b->devs[slot].addr + idx * PAGE_SIZE; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); success = sync_page_io(rdev, addr, s << 9, diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c index 874874fe4fa1..1253466666e4 100644 --- a/drivers/md/raid5-cache.c +++ b/drivers/md/raid5-cache.c @@ -1892,7 +1892,7 @@ r5l_recovery_replay_one_stripe(struct r5conf *conf, /* in case device is broken */ rdev = conf->disks[disk_index].rdev; if (rdev) { - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); sync_page_io(rdev, sh->sector, PAGE_SIZE, sh->dev[disk_index].page, REQ_OP_WRITE, false); @@ -1900,7 +1900,7 @@ r5l_recovery_replay_one_stripe(struct r5conf *conf, } rrdev = conf->disks[disk_index].replacement; if (rrdev) { - atomic_inc(&rrdev->nr_pending); + nr_pending_inc(rrdev); sync_page_io(rrdev, sh->sector, PAGE_SIZE, sh->dev[disk_index].page, REQ_OP_WRITE, false); diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 3b04d8b526b1..25da83eaa2ef 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -1199,11 +1199,11 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) if (rdev && test_bit(Faulty, &rdev->flags)) rdev = NULL; if (rdev) - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); if (rrdev && test_bit(Faulty, &rrdev->flags)) rrdev = NULL; if (rrdev) - atomic_inc(&rrdev->nr_pending); + nr_pending_inc(rrdev); /* We have already checked bad blocks for reads. Now * need to check for writes. We never accept write errors @@ -1231,7 +1231,7 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s) * will dec nr_pending, we must * increment it first. */ - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); md_wait_for_blocked_rdev(rdev, conf->mddev); } else { /* Acknowledged bad block - skip the write */ @@ -3628,7 +3628,7 @@ handle_failed_stripe(struct r5conf *conf, struct stripe_head *sh, if (rdev && test_bit(In_sync, &rdev->flags) && !test_bit(Faulty, &rdev->flags)) - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); else rdev = NULL; if (rdev) { @@ -4729,7 +4729,7 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) set_bit(BlockedBadBlocks, &rdev->flags); s->blocked_rdev = rdev; - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); } } clear_bit(R5_Insync, &dev->flags); @@ -4767,7 +4767,7 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) clear_bit(R5_Insync, &dev->flags); if (rdev2 && !test_bit(Faulty, &rdev2->flags)) { s->handle_bad_blocks = 1; - atomic_inc(&rdev2->nr_pending); + nr_pending_inc(rdev2); } else clear_bit(R5_WriteError, &dev->flags); } @@ -4778,7 +4778,7 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) if (rdev2 && !test_bit(Faulty, &rdev2->flags)) { s->handle_bad_blocks = 1; - atomic_inc(&rdev2->nr_pending); + nr_pending_inc(rdev2); } else clear_bit(R5_MadeGood, &dev->flags); } @@ -4787,7 +4787,7 @@ static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s) if (rdev2 && !test_bit(Faulty, &rdev2->flags)) { s->handle_bad_blocks = 1; - atomic_inc(&rdev2->nr_pending); + nr_pending_inc(rdev2); } else clear_bit(R5_MadeGoodRepl, &dev->flags); } @@ -5478,7 +5478,7 @@ static int raid5_read_one_chunk(struct mddev *mddev, struct bio *raid_bio) return 0; } - atomic_inc(&rdev->nr_pending); + nr_pending_inc(rdev); if (rdev_has_badblock(rdev, sector, bio_sectors(raid_bio))) { rdev_dec_pending(rdev, mddev); @@ -8157,7 +8157,7 @@ static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev) clear_bit(In_sync, &rdev->flags); if (test_bit(In_sync, &rdev->flags) || - atomic_read(&rdev->nr_pending)) { + nr_pending_is_not_zero(rdev)) { err = -EBUSY; goto abort; }

[v2,02/11] md: add infra for nr_pending changes

Commit Message

Patch