All of lore.kernel.org
 help / color / mirror / Atom feed
From: kernel test robot <lkp@intel.com>
To: Guoqing Jiang <guoqing.jiang@linux.dev>, song@kernel.org
Cc: kbuild-all@lists.01.org, linux-raid@vger.kernel.org,
	jens@chianterastutte.eu
Subject: Re: [PATCH] raid1: ensure bio doesn't have more than BIO_MAX_VECS sectors
Date: Fri, 13 Aug 2021 18:12:36 +0800	[thread overview]
Message-ID: <202108131843.FP48KtPs-lkp@intel.com> (raw)
In-Reply-To: <20210813060510.3545109-1-guoqing.jiang@linux.dev>

[-- Attachment #1: Type: text/plain, Size: 13412 bytes --]

Hi Guoqing,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on song-md/md-next]
[also build test ERROR on v5.14-rc5 next-20210812]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Guoqing-Jiang/raid1-ensure-bio-doesn-t-have-more-than-BIO_MAX_VECS-sectors/20210813-140810
base:   git://git.kernel.org/pub/scm/linux/kernel/git/song/md.git md-next
config: nds32-randconfig-r035-20210813 (attached as .config)
compiler: nds32le-linux-gcc (GCC) 10.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/29b7720a83de1deea0d8ecfafe0db46146636b15
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Guoqing-Jiang/raid1-ensure-bio-doesn-t-have-more-than-BIO_MAX_VECS-sectors/20210813-140810
        git checkout 29b7720a83de1deea0d8ecfafe0db46146636b15
        # save the attached .config to linux build tree
        mkdir build_dir
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-10.3.0 make.cross O=build_dir ARCH=nds32 SHELL=/bin/bash drivers/md/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   In file included from include/linux/kernel.h:15,
                    from include/asm-generic/bug.h:20,
                    from ./arch/nds32/include/generated/asm/bug.h:1,
                    from include/linux/bug.h:5,
                    from include/linux/mmdebug.h:5,
                    from include/linux/gfp.h:5,
                    from include/linux/slab.h:15,
                    from drivers/md/raid1.c:26:
   drivers/md/raid1.c: In function 'raid1_write_request':
>> drivers/md/raid1.c:1459:55: error: 'PAGE_SECTORS' undeclared (first use in this function); did you mean 'PAGE_MEMORY'?
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                                                       ^~~~~~~~~~~~
   include/linux/minmax.h:20:39: note: in definition of macro '__typecheck'
      20 |  (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
         |                                       ^
   include/linux/minmax.h:36:24: note: in expansion of macro '__safe_cmp'
      36 |  __builtin_choose_expr(__safe_cmp(x, y), \
         |                        ^~~~~~~~~~
   include/linux/minmax.h:104:27: note: in expansion of macro '__careful_cmp'
     104 | #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <)
         |                           ^~~~~~~~~~~~~
   drivers/md/raid1.c:1459:16: note: in expansion of macro 'min_t'
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                ^~~~~
   drivers/md/raid1.c:1459:55: note: each undeclared identifier is reported only once for each function it appears in
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                                                       ^~~~~~~~~~~~
   include/linux/minmax.h:20:39: note: in definition of macro '__typecheck'
      20 |  (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
         |                                       ^
   include/linux/minmax.h:36:24: note: in expansion of macro '__safe_cmp'
      36 |  __builtin_choose_expr(__safe_cmp(x, y), \
         |                        ^~~~~~~~~~
   include/linux/minmax.h:104:27: note: in expansion of macro '__careful_cmp'
     104 | #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <)
         |                           ^~~~~~~~~~~~~
   drivers/md/raid1.c:1459:16: note: in expansion of macro 'min_t'
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                ^~~~~
>> include/linux/minmax.h:36:2: error: first argument to '__builtin_choose_expr' not a constant
      36 |  __builtin_choose_expr(__safe_cmp(x, y), \
         |  ^~~~~~~~~~~~~~~~~~~~~
   include/linux/minmax.h:104:27: note: in expansion of macro '__careful_cmp'
     104 | #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <)
         |                           ^~~~~~~~~~~~~
   drivers/md/raid1.c:1459:16: note: in expansion of macro 'min_t'
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                ^~~~~


vim +1459 drivers/md/raid1.c

  1320	
  1321	static void raid1_write_request(struct mddev *mddev, struct bio *bio,
  1322					int max_write_sectors)
  1323	{
  1324		struct r1conf *conf = mddev->private;
  1325		struct r1bio *r1_bio;
  1326		int i, disks;
  1327		struct bitmap *bitmap = mddev->bitmap;
  1328		unsigned long flags;
  1329		struct md_rdev *blocked_rdev;
  1330		struct blk_plug_cb *cb;
  1331		struct raid1_plug_cb *plug = NULL;
  1332		int first_clone;
  1333		int max_sectors;
  1334	
  1335		if (mddev_is_clustered(mddev) &&
  1336		     md_cluster_ops->area_resyncing(mddev, WRITE,
  1337			     bio->bi_iter.bi_sector, bio_end_sector(bio))) {
  1338	
  1339			DEFINE_WAIT(w);
  1340			for (;;) {
  1341				prepare_to_wait(&conf->wait_barrier,
  1342						&w, TASK_IDLE);
  1343				if (!md_cluster_ops->area_resyncing(mddev, WRITE,
  1344								bio->bi_iter.bi_sector,
  1345								bio_end_sector(bio)))
  1346					break;
  1347				schedule();
  1348			}
  1349			finish_wait(&conf->wait_barrier, &w);
  1350		}
  1351	
  1352		/*
  1353		 * Register the new request and wait if the reconstruction
  1354		 * thread has put up a bar for new requests.
  1355		 * Continue immediately if no resync is active currently.
  1356		 */
  1357		wait_barrier(conf, bio->bi_iter.bi_sector);
  1358	
  1359		r1_bio = alloc_r1bio(mddev, bio);
  1360		r1_bio->sectors = max_write_sectors;
  1361	
  1362		if (conf->pending_count >= max_queued_requests) {
  1363			md_wakeup_thread(mddev->thread);
  1364			raid1_log(mddev, "wait queued");
  1365			wait_event(conf->wait_barrier,
  1366				   conf->pending_count < max_queued_requests);
  1367		}
  1368		/* first select target devices under rcu_lock and
  1369		 * inc refcount on their rdev.  Record them by setting
  1370		 * bios[x] to bio
  1371		 * If there are known/acknowledged bad blocks on any device on
  1372		 * which we have seen a write error, we want to avoid writing those
  1373		 * blocks.
  1374		 * This potentially requires several writes to write around
  1375		 * the bad blocks.  Each set of writes gets it's own r1bio
  1376		 * with a set of bios attached.
  1377		 */
  1378	
  1379		disks = conf->raid_disks * 2;
  1380	 retry_write:
  1381		blocked_rdev = NULL;
  1382		rcu_read_lock();
  1383		max_sectors = r1_bio->sectors;
  1384		for (i = 0;  i < disks; i++) {
  1385			struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev);
  1386			if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
  1387				atomic_inc(&rdev->nr_pending);
  1388				blocked_rdev = rdev;
  1389				break;
  1390			}
  1391			r1_bio->bios[i] = NULL;
  1392			if (!rdev || test_bit(Faulty, &rdev->flags)) {
  1393				if (i < conf->raid_disks)
  1394					set_bit(R1BIO_Degraded, &r1_bio->state);
  1395				continue;
  1396			}
  1397	
  1398			atomic_inc(&rdev->nr_pending);
  1399			if (test_bit(WriteErrorSeen, &rdev->flags)) {
  1400				sector_t first_bad;
  1401				int bad_sectors;
  1402				int is_bad;
  1403	
  1404				is_bad = is_badblock(rdev, r1_bio->sector, max_sectors,
  1405						     &first_bad, &bad_sectors);
  1406				if (is_bad < 0) {
  1407					/* mustn't write here until the bad block is
  1408					 * acknowledged*/
  1409					set_bit(BlockedBadBlocks, &rdev->flags);
  1410					blocked_rdev = rdev;
  1411					break;
  1412				}
  1413				if (is_bad && first_bad <= r1_bio->sector) {
  1414					/* Cannot write here at all */
  1415					bad_sectors -= (r1_bio->sector - first_bad);
  1416					if (bad_sectors < max_sectors)
  1417						/* mustn't write more than bad_sectors
  1418						 * to other devices yet
  1419						 */
  1420						max_sectors = bad_sectors;
  1421					rdev_dec_pending(rdev, mddev);
  1422					/* We don't set R1BIO_Degraded as that
  1423					 * only applies if the disk is
  1424					 * missing, so it might be re-added,
  1425					 * and we want to know to recover this
  1426					 * chunk.
  1427					 * In this case the device is here,
  1428					 * and the fact that this chunk is not
  1429					 * in-sync is recorded in the bad
  1430					 * block log
  1431					 */
  1432					continue;
  1433				}
  1434				if (is_bad) {
  1435					int good_sectors = first_bad - r1_bio->sector;
  1436					if (good_sectors < max_sectors)
  1437						max_sectors = good_sectors;
  1438				}
  1439			}
  1440			r1_bio->bios[i] = bio;
  1441		}
  1442		rcu_read_unlock();
  1443	
  1444		if (unlikely(blocked_rdev)) {
  1445			/* Wait for this device to become unblocked */
  1446			int j;
  1447	
  1448			for (j = 0; j < i; j++)
  1449				if (r1_bio->bios[j])
  1450					rdev_dec_pending(conf->mirrors[j].rdev, mddev);
  1451			r1_bio->state = 0;
  1452			allow_barrier(conf, bio->bi_iter.bi_sector);
  1453			raid1_log(mddev, "wait rdev %d blocked", blocked_rdev->raid_disk);
  1454			md_wait_for_blocked_rdev(blocked_rdev, mddev);
  1455			wait_barrier(conf, bio->bi_iter.bi_sector);
  1456			goto retry_write;
  1457		}
  1458	
> 1459		max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
  1460		if (max_sectors < bio_sectors(bio)) {
  1461			struct bio *split = bio_split(bio, max_sectors,
  1462						      GFP_NOIO, &conf->bio_split);
  1463			bio_chain(split, bio);
  1464			submit_bio_noacct(bio);
  1465			bio = split;
  1466			r1_bio->master_bio = bio;
  1467			r1_bio->sectors = max_sectors;
  1468		}
  1469	
  1470		if (blk_queue_io_stat(bio->bi_bdev->bd_disk->queue))
  1471			r1_bio->start_time = bio_start_io_acct(bio);
  1472		atomic_set(&r1_bio->remaining, 1);
  1473		atomic_set(&r1_bio->behind_remaining, 0);
  1474	
  1475		first_clone = 1;
  1476	
  1477		for (i = 0; i < disks; i++) {
  1478			struct bio *mbio = NULL;
  1479			struct md_rdev *rdev = conf->mirrors[i].rdev;
  1480			if (!r1_bio->bios[i])
  1481				continue;
  1482	
  1483			if (first_clone) {
  1484				/* do behind I/O ?
  1485				 * Not if there are too many, or cannot
  1486				 * allocate memory, or a reader on WriteMostly
  1487				 * is waiting for behind writes to flush */
  1488				if (bitmap &&
  1489				    (atomic_read(&bitmap->behind_writes)
  1490				     < mddev->bitmap_info.max_write_behind) &&
  1491				    !waitqueue_active(&bitmap->behind_wait)) {
  1492					alloc_behind_master_bio(r1_bio, bio);
  1493				}
  1494	
  1495				md_bitmap_startwrite(bitmap, r1_bio->sector, r1_bio->sectors,
  1496						     test_bit(R1BIO_BehindIO, &r1_bio->state));
  1497				first_clone = 0;
  1498			}
  1499	
  1500			if (r1_bio->behind_master_bio)
  1501				mbio = bio_clone_fast(r1_bio->behind_master_bio,
  1502						      GFP_NOIO, &mddev->bio_set);
  1503			else
  1504				mbio = bio_clone_fast(bio, GFP_NOIO, &mddev->bio_set);
  1505	
  1506			if (r1_bio->behind_master_bio) {
  1507				if (test_bit(CollisionCheck, &rdev->flags))
  1508					wait_for_serialization(rdev, r1_bio);
  1509				if (test_bit(WriteMostly, &rdev->flags))
  1510					atomic_inc(&r1_bio->behind_remaining);
  1511			} else if (mddev->serialize_policy)
  1512				wait_for_serialization(rdev, r1_bio);
  1513	
  1514			r1_bio->bios[i] = mbio;
  1515	
  1516			mbio->bi_iter.bi_sector	= (r1_bio->sector +
  1517					   conf->mirrors[i].rdev->data_offset);
  1518			bio_set_dev(mbio, conf->mirrors[i].rdev->bdev);
  1519			mbio->bi_end_io	= raid1_end_write_request;
  1520			mbio->bi_opf = bio_op(bio) | (bio->bi_opf & (REQ_SYNC | REQ_FUA));
  1521			if (test_bit(FailFast, &conf->mirrors[i].rdev->flags) &&
  1522			    !test_bit(WriteMostly, &conf->mirrors[i].rdev->flags) &&
  1523			    conf->raid_disks - mddev->degraded > 1)
  1524				mbio->bi_opf |= MD_FAILFAST;
  1525			mbio->bi_private = r1_bio;
  1526	
  1527			atomic_inc(&r1_bio->remaining);
  1528	
  1529			if (mddev->gendisk)
  1530				trace_block_bio_remap(mbio, disk_devt(mddev->gendisk),
  1531						      r1_bio->sector);
  1532			/* flush_pending_writes() needs access to the rdev so...*/
  1533			mbio->bi_bdev = (void *)conf->mirrors[i].rdev;
  1534	
  1535			cb = blk_check_plugged(raid1_unplug, mddev, sizeof(*plug));
  1536			if (cb)
  1537				plug = container_of(cb, struct raid1_plug_cb, cb);
  1538			else
  1539				plug = NULL;
  1540			if (plug) {
  1541				bio_list_add(&plug->pending, mbio);
  1542				plug->pending_cnt++;
  1543			} else {
  1544				spin_lock_irqsave(&conf->device_lock, flags);
  1545				bio_list_add(&conf->pending_bio_list, mbio);
  1546				conf->pending_count++;
  1547				spin_unlock_irqrestore(&conf->device_lock, flags);
  1548				md_wakeup_thread(mddev->thread);
  1549			}
  1550		}
  1551	
  1552		r1_bio_write_done(r1_bio);
  1553	
  1554		/* In case raid1d snuck in to freeze_array */
  1555		wake_up(&conf->wait_barrier);
  1556	}
  1557	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 30071 bytes --]

WARNING: multiple messages have this Message-ID (diff)
From: kernel test robot <lkp@intel.com>
To: kbuild-all@lists.01.org
Subject: Re: [PATCH] raid1: ensure bio doesn't have more than BIO_MAX_VECS sectors
Date: Fri, 13 Aug 2021 18:12:36 +0800	[thread overview]
Message-ID: <202108131843.FP48KtPs-lkp@intel.com> (raw)
In-Reply-To: <20210813060510.3545109-1-guoqing.jiang@linux.dev>

[-- Attachment #1: Type: text/plain, Size: 13738 bytes --]

Hi Guoqing,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on song-md/md-next]
[also build test ERROR on v5.14-rc5 next-20210812]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Guoqing-Jiang/raid1-ensure-bio-doesn-t-have-more-than-BIO_MAX_VECS-sectors/20210813-140810
base:   git://git.kernel.org/pub/scm/linux/kernel/git/song/md.git md-next
config: nds32-randconfig-r035-20210813 (attached as .config)
compiler: nds32le-linux-gcc (GCC) 10.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/29b7720a83de1deea0d8ecfafe0db46146636b15
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Guoqing-Jiang/raid1-ensure-bio-doesn-t-have-more-than-BIO_MAX_VECS-sectors/20210813-140810
        git checkout 29b7720a83de1deea0d8ecfafe0db46146636b15
        # save the attached .config to linux build tree
        mkdir build_dir
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-10.3.0 make.cross O=build_dir ARCH=nds32 SHELL=/bin/bash drivers/md/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   In file included from include/linux/kernel.h:15,
                    from include/asm-generic/bug.h:20,
                    from ./arch/nds32/include/generated/asm/bug.h:1,
                    from include/linux/bug.h:5,
                    from include/linux/mmdebug.h:5,
                    from include/linux/gfp.h:5,
                    from include/linux/slab.h:15,
                    from drivers/md/raid1.c:26:
   drivers/md/raid1.c: In function 'raid1_write_request':
>> drivers/md/raid1.c:1459:55: error: 'PAGE_SECTORS' undeclared (first use in this function); did you mean 'PAGE_MEMORY'?
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                                                       ^~~~~~~~~~~~
   include/linux/minmax.h:20:39: note: in definition of macro '__typecheck'
      20 |  (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
         |                                       ^
   include/linux/minmax.h:36:24: note: in expansion of macro '__safe_cmp'
      36 |  __builtin_choose_expr(__safe_cmp(x, y), \
         |                        ^~~~~~~~~~
   include/linux/minmax.h:104:27: note: in expansion of macro '__careful_cmp'
     104 | #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <)
         |                           ^~~~~~~~~~~~~
   drivers/md/raid1.c:1459:16: note: in expansion of macro 'min_t'
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                ^~~~~
   drivers/md/raid1.c:1459:55: note: each undeclared identifier is reported only once for each function it appears in
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                                                       ^~~~~~~~~~~~
   include/linux/minmax.h:20:39: note: in definition of macro '__typecheck'
      20 |  (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
         |                                       ^
   include/linux/minmax.h:36:24: note: in expansion of macro '__safe_cmp'
      36 |  __builtin_choose_expr(__safe_cmp(x, y), \
         |                        ^~~~~~~~~~
   include/linux/minmax.h:104:27: note: in expansion of macro '__careful_cmp'
     104 | #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <)
         |                           ^~~~~~~~~~~~~
   drivers/md/raid1.c:1459:16: note: in expansion of macro 'min_t'
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                ^~~~~
>> include/linux/minmax.h:36:2: error: first argument to '__builtin_choose_expr' not a constant
      36 |  __builtin_choose_expr(__safe_cmp(x, y), \
         |  ^~~~~~~~~~~~~~~~~~~~~
   include/linux/minmax.h:104:27: note: in expansion of macro '__careful_cmp'
     104 | #define min_t(type, x, y) __careful_cmp((type)(x), (type)(y), <)
         |                           ^~~~~~~~~~~~~
   drivers/md/raid1.c:1459:16: note: in expansion of macro 'min_t'
    1459 |  max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
         |                ^~~~~


vim +1459 drivers/md/raid1.c

  1320	
  1321	static void raid1_write_request(struct mddev *mddev, struct bio *bio,
  1322					int max_write_sectors)
  1323	{
  1324		struct r1conf *conf = mddev->private;
  1325		struct r1bio *r1_bio;
  1326		int i, disks;
  1327		struct bitmap *bitmap = mddev->bitmap;
  1328		unsigned long flags;
  1329		struct md_rdev *blocked_rdev;
  1330		struct blk_plug_cb *cb;
  1331		struct raid1_plug_cb *plug = NULL;
  1332		int first_clone;
  1333		int max_sectors;
  1334	
  1335		if (mddev_is_clustered(mddev) &&
  1336		     md_cluster_ops->area_resyncing(mddev, WRITE,
  1337			     bio->bi_iter.bi_sector, bio_end_sector(bio))) {
  1338	
  1339			DEFINE_WAIT(w);
  1340			for (;;) {
  1341				prepare_to_wait(&conf->wait_barrier,
  1342						&w, TASK_IDLE);
  1343				if (!md_cluster_ops->area_resyncing(mddev, WRITE,
  1344								bio->bi_iter.bi_sector,
  1345								bio_end_sector(bio)))
  1346					break;
  1347				schedule();
  1348			}
  1349			finish_wait(&conf->wait_barrier, &w);
  1350		}
  1351	
  1352		/*
  1353		 * Register the new request and wait if the reconstruction
  1354		 * thread has put up a bar for new requests.
  1355		 * Continue immediately if no resync is active currently.
  1356		 */
  1357		wait_barrier(conf, bio->bi_iter.bi_sector);
  1358	
  1359		r1_bio = alloc_r1bio(mddev, bio);
  1360		r1_bio->sectors = max_write_sectors;
  1361	
  1362		if (conf->pending_count >= max_queued_requests) {
  1363			md_wakeup_thread(mddev->thread);
  1364			raid1_log(mddev, "wait queued");
  1365			wait_event(conf->wait_barrier,
  1366				   conf->pending_count < max_queued_requests);
  1367		}
  1368		/* first select target devices under rcu_lock and
  1369		 * inc refcount on their rdev.  Record them by setting
  1370		 * bios[x] to bio
  1371		 * If there are known/acknowledged bad blocks on any device on
  1372		 * which we have seen a write error, we want to avoid writing those
  1373		 * blocks.
  1374		 * This potentially requires several writes to write around
  1375		 * the bad blocks.  Each set of writes gets it's own r1bio
  1376		 * with a set of bios attached.
  1377		 */
  1378	
  1379		disks = conf->raid_disks * 2;
  1380	 retry_write:
  1381		blocked_rdev = NULL;
  1382		rcu_read_lock();
  1383		max_sectors = r1_bio->sectors;
  1384		for (i = 0;  i < disks; i++) {
  1385			struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev);
  1386			if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
  1387				atomic_inc(&rdev->nr_pending);
  1388				blocked_rdev = rdev;
  1389				break;
  1390			}
  1391			r1_bio->bios[i] = NULL;
  1392			if (!rdev || test_bit(Faulty, &rdev->flags)) {
  1393				if (i < conf->raid_disks)
  1394					set_bit(R1BIO_Degraded, &r1_bio->state);
  1395				continue;
  1396			}
  1397	
  1398			atomic_inc(&rdev->nr_pending);
  1399			if (test_bit(WriteErrorSeen, &rdev->flags)) {
  1400				sector_t first_bad;
  1401				int bad_sectors;
  1402				int is_bad;
  1403	
  1404				is_bad = is_badblock(rdev, r1_bio->sector, max_sectors,
  1405						     &first_bad, &bad_sectors);
  1406				if (is_bad < 0) {
  1407					/* mustn't write here until the bad block is
  1408					 * acknowledged*/
  1409					set_bit(BlockedBadBlocks, &rdev->flags);
  1410					blocked_rdev = rdev;
  1411					break;
  1412				}
  1413				if (is_bad && first_bad <= r1_bio->sector) {
  1414					/* Cannot write here at all */
  1415					bad_sectors -= (r1_bio->sector - first_bad);
  1416					if (bad_sectors < max_sectors)
  1417						/* mustn't write more than bad_sectors
  1418						 * to other devices yet
  1419						 */
  1420						max_sectors = bad_sectors;
  1421					rdev_dec_pending(rdev, mddev);
  1422					/* We don't set R1BIO_Degraded as that
  1423					 * only applies if the disk is
  1424					 * missing, so it might be re-added,
  1425					 * and we want to know to recover this
  1426					 * chunk.
  1427					 * In this case the device is here,
  1428					 * and the fact that this chunk is not
  1429					 * in-sync is recorded in the bad
  1430					 * block log
  1431					 */
  1432					continue;
  1433				}
  1434				if (is_bad) {
  1435					int good_sectors = first_bad - r1_bio->sector;
  1436					if (good_sectors < max_sectors)
  1437						max_sectors = good_sectors;
  1438				}
  1439			}
  1440			r1_bio->bios[i] = bio;
  1441		}
  1442		rcu_read_unlock();
  1443	
  1444		if (unlikely(blocked_rdev)) {
  1445			/* Wait for this device to become unblocked */
  1446			int j;
  1447	
  1448			for (j = 0; j < i; j++)
  1449				if (r1_bio->bios[j])
  1450					rdev_dec_pending(conf->mirrors[j].rdev, mddev);
  1451			r1_bio->state = 0;
  1452			allow_barrier(conf, bio->bi_iter.bi_sector);
  1453			raid1_log(mddev, "wait rdev %d blocked", blocked_rdev->raid_disk);
  1454			md_wait_for_blocked_rdev(blocked_rdev, mddev);
  1455			wait_barrier(conf, bio->bi_iter.bi_sector);
  1456			goto retry_write;
  1457		}
  1458	
> 1459		max_sectors = min_t(int, max_sectors, BIO_MAX_VECS * PAGE_SECTORS);
  1460		if (max_sectors < bio_sectors(bio)) {
  1461			struct bio *split = bio_split(bio, max_sectors,
  1462						      GFP_NOIO, &conf->bio_split);
  1463			bio_chain(split, bio);
  1464			submit_bio_noacct(bio);
  1465			bio = split;
  1466			r1_bio->master_bio = bio;
  1467			r1_bio->sectors = max_sectors;
  1468		}
  1469	
  1470		if (blk_queue_io_stat(bio->bi_bdev->bd_disk->queue))
  1471			r1_bio->start_time = bio_start_io_acct(bio);
  1472		atomic_set(&r1_bio->remaining, 1);
  1473		atomic_set(&r1_bio->behind_remaining, 0);
  1474	
  1475		first_clone = 1;
  1476	
  1477		for (i = 0; i < disks; i++) {
  1478			struct bio *mbio = NULL;
  1479			struct md_rdev *rdev = conf->mirrors[i].rdev;
  1480			if (!r1_bio->bios[i])
  1481				continue;
  1482	
  1483			if (first_clone) {
  1484				/* do behind I/O ?
  1485				 * Not if there are too many, or cannot
  1486				 * allocate memory, or a reader on WriteMostly
  1487				 * is waiting for behind writes to flush */
  1488				if (bitmap &&
  1489				    (atomic_read(&bitmap->behind_writes)
  1490				     < mddev->bitmap_info.max_write_behind) &&
  1491				    !waitqueue_active(&bitmap->behind_wait)) {
  1492					alloc_behind_master_bio(r1_bio, bio);
  1493				}
  1494	
  1495				md_bitmap_startwrite(bitmap, r1_bio->sector, r1_bio->sectors,
  1496						     test_bit(R1BIO_BehindIO, &r1_bio->state));
  1497				first_clone = 0;
  1498			}
  1499	
  1500			if (r1_bio->behind_master_bio)
  1501				mbio = bio_clone_fast(r1_bio->behind_master_bio,
  1502						      GFP_NOIO, &mddev->bio_set);
  1503			else
  1504				mbio = bio_clone_fast(bio, GFP_NOIO, &mddev->bio_set);
  1505	
  1506			if (r1_bio->behind_master_bio) {
  1507				if (test_bit(CollisionCheck, &rdev->flags))
  1508					wait_for_serialization(rdev, r1_bio);
  1509				if (test_bit(WriteMostly, &rdev->flags))
  1510					atomic_inc(&r1_bio->behind_remaining);
  1511			} else if (mddev->serialize_policy)
  1512				wait_for_serialization(rdev, r1_bio);
  1513	
  1514			r1_bio->bios[i] = mbio;
  1515	
  1516			mbio->bi_iter.bi_sector	= (r1_bio->sector +
  1517					   conf->mirrors[i].rdev->data_offset);
  1518			bio_set_dev(mbio, conf->mirrors[i].rdev->bdev);
  1519			mbio->bi_end_io	= raid1_end_write_request;
  1520			mbio->bi_opf = bio_op(bio) | (bio->bi_opf & (REQ_SYNC | REQ_FUA));
  1521			if (test_bit(FailFast, &conf->mirrors[i].rdev->flags) &&
  1522			    !test_bit(WriteMostly, &conf->mirrors[i].rdev->flags) &&
  1523			    conf->raid_disks - mddev->degraded > 1)
  1524				mbio->bi_opf |= MD_FAILFAST;
  1525			mbio->bi_private = r1_bio;
  1526	
  1527			atomic_inc(&r1_bio->remaining);
  1528	
  1529			if (mddev->gendisk)
  1530				trace_block_bio_remap(mbio, disk_devt(mddev->gendisk),
  1531						      r1_bio->sector);
  1532			/* flush_pending_writes() needs access to the rdev so...*/
  1533			mbio->bi_bdev = (void *)conf->mirrors[i].rdev;
  1534	
  1535			cb = blk_check_plugged(raid1_unplug, mddev, sizeof(*plug));
  1536			if (cb)
  1537				plug = container_of(cb, struct raid1_plug_cb, cb);
  1538			else
  1539				plug = NULL;
  1540			if (plug) {
  1541				bio_list_add(&plug->pending, mbio);
  1542				plug->pending_cnt++;
  1543			} else {
  1544				spin_lock_irqsave(&conf->device_lock, flags);
  1545				bio_list_add(&conf->pending_bio_list, mbio);
  1546				conf->pending_count++;
  1547				spin_unlock_irqrestore(&conf->device_lock, flags);
  1548				md_wakeup_thread(mddev->thread);
  1549			}
  1550		}
  1551	
  1552		r1_bio_write_done(r1_bio);
  1553	
  1554		/* In case raid1d snuck in to freeze_array */
  1555		wake_up(&conf->wait_barrier);
  1556	}
  1557	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

[-- Attachment #2: config.gz --]
[-- Type: application/gzip, Size: 30071 bytes --]

  parent reply	other threads:[~2021-08-13 10:13 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-13  6:05 [PATCH] raid1: ensure bio doesn't have more than BIO_MAX_VECS sectors Guoqing Jiang
2021-08-13  7:49 ` Christoph Hellwig
2021-08-13  8:38   ` Guoqing Jiang
2021-08-14  7:55     ` Christoph Hellwig
2021-08-14  8:57       ` Ming Lei
2021-08-16  6:27         ` Guoqing Jiang
2021-08-16  7:13           ` Ming Lei
2021-08-16  9:37         ` Christoph Hellwig
2021-08-16 11:40           ` Ming Lei
2021-08-17  5:06             ` Christoph Hellwig
2021-08-17 12:32               ` Ming Lei
2021-09-24 15:34                 ` Jens Stutte (Archiv)
2021-09-25 23:02                   ` Guoqing Jiang
2021-08-13  9:27 ` kernel test robot
2021-08-13  9:27   ` kernel test robot
2021-08-13 10:12 ` kernel test robot [this message]
2021-08-13 10:12   ` kernel test robot

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=202108131843.FP48KtPs-lkp@intel.com \
    --to=lkp@intel.com \
    --cc=guoqing.jiang@linux.dev \
    --cc=jens@chianterastutte.eu \
    --cc=kbuild-all@lists.01.org \
    --cc=linux-raid@vger.kernel.org \
    --cc=song@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.