From daaedb820ad716e00210af8859b194c404202b78 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:09 +0100 Subject: mtd_blkdevs: don't override BLKFLSBUF BLKFLSBUF is not supposed to actually send a flush command to the device, but to tear down buffer cache structures. Remove the mtd_blkdevs implementation and just use the default semantics instead. Signed-off-by: Christoph Hellwig Acked-by: Richard Weinberger Signed-off-by: Jens Axboe --- drivers/mtd/mtd_blkdevs.c | 28 ---------------------------- 1 file changed, 28 deletions(-) diff --git a/drivers/mtd/mtd_blkdevs.c b/drivers/mtd/mtd_blkdevs.c index 0c05f77f9b21..fb8e12d590a1 100644 --- a/drivers/mtd/mtd_blkdevs.c +++ b/drivers/mtd/mtd_blkdevs.c @@ -298,38 +298,10 @@ unlock: return ret; } -static int blktrans_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg) -{ - struct mtd_blktrans_dev *dev = blktrans_dev_get(bdev->bd_disk); - int ret = -ENXIO; - - if (!dev) - return ret; - - mutex_lock(&dev->lock); - - if (!dev->mtd) - goto unlock; - - switch (cmd) { - case BLKFLSBUF: - ret = dev->tr->flush ? dev->tr->flush(dev) : 0; - break; - default: - ret = -ENOTTY; - } -unlock: - mutex_unlock(&dev->lock); - blktrans_dev_put(dev); - return ret; -} - static const struct block_device_operations mtd_block_ops = { .owner = THIS_MODULE, .open = blktrans_open, .release = blktrans_release, - .ioctl = blktrans_ioctl, .getgeo = blktrans_getgeo, }; -- cgit v1.2.3 From 4a9d6d667f0bafed55a9e9f5ae8bceb3680749d7 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:10 +0100 Subject: block: don't call into the driver for BLKFLSBUF BLKFLSBUF is entirely contained in the block core, and there is no good reason to give the driver a hook into processing it. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- block/ioctl.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/block/ioctl.c b/block/ioctl.c index 3fbc382eb926..c6d8863f0409 100644 --- a/block/ioctl.c +++ b/block/ioctl.c @@ -369,15 +369,8 @@ static inline int is_unrecognized_ioctl(int ret) static int blkdev_flushbuf(struct block_device *bdev, fmode_t mode, unsigned cmd, unsigned long arg) { - int ret; - if (!capable(CAP_SYS_ADMIN)) return -EACCES; - - ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg); - if (!is_unrecognized_ioctl(ret)) - return ret; - fsync_bdev(bdev); invalidate_bdev(bdev); return 0; -- cgit v1.2.3 From e00adcadf3af7a8335026d71ab9f0e0a922191ac Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:11 +0100 Subject: block: add a new set_read_only method Add a new method to allow for driver-specific processing when setting or clearing the block device read-only state. This allows to replace the cumbersome and error-prone override of the whole ioctl implementation. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- block/ioctl.c | 5 +++++ include/linux/blkdev.h | 1 + 2 files changed, 6 insertions(+) diff --git a/block/ioctl.c b/block/ioctl.c index c6d8863f0409..a6fa16b97705 100644 --- a/block/ioctl.c +++ b/block/ioctl.c @@ -389,6 +389,11 @@ static int blkdev_roset(struct block_device *bdev, fmode_t mode, return ret; if (get_user(n, (int __user *)arg)) return -EFAULT; + if (bdev->bd_disk->fops->set_read_only) { + ret = bdev->bd_disk->fops->set_read_only(bdev, n); + if (ret) + return ret; + } set_device_ro(bdev, n); return 0; } diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 639cae2c158b..5c1ba8a8d2bc 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -1850,6 +1850,7 @@ struct block_device_operations { void (*unlock_native_capacity) (struct gendisk *); int (*revalidate_disk) (struct gendisk *); int (*getgeo)(struct block_device *, struct hd_geometry *); + int (*set_read_only)(struct block_device *bdev, bool ro); /* this callback is with swap_lock and sometimes page table lock held */ void (*swap_slot_free_notify) (struct block_device *, unsigned long); int (*report_zones)(struct gendisk *, sector_t sector, -- cgit v1.2.3 From 34919e3ba23a106e04f042f19348d9e55739b35a Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:12 +0100 Subject: rbd: implement ->set_read_only to hook into BLKROSET processing Implement the ->set_read_only method instead of parsing the actual ioctl command. Signed-off-by: Christoph Hellwig Acked-by: Ilya Dryomov Signed-off-by: Jens Axboe --- drivers/block/rbd.c | 40 ++++------------------------------------ 1 file changed, 4 insertions(+), 36 deletions(-) diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index f84128abade3..671733e459cf 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -692,12 +692,9 @@ static void rbd_release(struct gendisk *disk, fmode_t mode) put_device(&rbd_dev->dev); } -static int rbd_ioctl_set_ro(struct rbd_device *rbd_dev, unsigned long arg) +static int rbd_set_read_only(struct block_device *bdev, bool ro) { - int ro; - - if (get_user(ro, (int __user *)arg)) - return -EFAULT; + struct rbd_device *rbd_dev = bdev->bd_disk->private_data; /* * Both images mapped read-only and snapshots can't be marked @@ -710,43 +707,14 @@ static int rbd_ioctl_set_ro(struct rbd_device *rbd_dev, unsigned long arg) rbd_assert(!rbd_is_snap(rbd_dev)); } - /* Let blkdev_roset() handle it */ - return -ENOTTY; -} - -static int rbd_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg) -{ - struct rbd_device *rbd_dev = bdev->bd_disk->private_data; - int ret; - - switch (cmd) { - case BLKROSET: - ret = rbd_ioctl_set_ro(rbd_dev, arg); - break; - default: - ret = -ENOTTY; - } - - return ret; -} - -#ifdef CONFIG_COMPAT -static int rbd_compat_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg) -{ - return rbd_ioctl(bdev, mode, cmd, arg); + return 0; } -#endif /* CONFIG_COMPAT */ static const struct block_device_operations rbd_bd_ops = { .owner = THIS_MODULE, .open = rbd_open, .release = rbd_release, - .ioctl = rbd_ioctl, -#ifdef CONFIG_COMPAT - .compat_ioctl = rbd_compat_ioctl, -#endif + .set_read_only = rbd_set_read_only, }; /* -- cgit v1.2.3 From 118cf084adb3964d06e1667cf7d702e56e5cd2c5 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:13 +0100 Subject: md: implement ->set_read_only to hook into BLKROSET processing Implement the ->set_read_only method instead of parsing the actual ioctl command. Signed-off-by: Christoph Hellwig Acked-by: Song Liu Signed-off-by: Jens Axboe --- drivers/md/md.c | 62 ++++++++++++++++++++++++++++----------------------------- 1 file changed, 31 insertions(+), 31 deletions(-) diff --git a/drivers/md/md.c b/drivers/md/md.c index 98bac4f304ae..96d31336ad43 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -7480,7 +7480,6 @@ static inline bool md_ioctl_valid(unsigned int cmd) { switch (cmd) { case ADD_NEW_DISK: - case BLKROSET: case GET_ARRAY_INFO: case GET_BITMAP_FILE: case GET_DISK_INFO: @@ -7507,7 +7506,6 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode, int err = 0; void __user *argp = (void __user *)arg; struct mddev *mddev = NULL; - int ro; bool did_set_md_closing = false; if (!md_ioctl_valid(cmd)) @@ -7687,35 +7685,6 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode, goto unlock; } break; - - case BLKROSET: - if (get_user(ro, (int __user *)(arg))) { - err = -EFAULT; - goto unlock; - } - err = -EINVAL; - - /* if the bdev is going readonly the value of mddev->ro - * does not matter, no writes are coming - */ - if (ro) - goto unlock; - - /* are we are already prepared for writes? */ - if (mddev->ro != 1) - goto unlock; - - /* transitioning to readauto need only happen for - * arrays that call md_write_start - */ - if (mddev->pers) { - err = restart_array(mddev); - if (err == 0) { - mddev->ro = 2; - set_disk_ro(mddev->gendisk, 0); - } - } - goto unlock; } /* @@ -7809,6 +7778,36 @@ static int md_compat_ioctl(struct block_device *bdev, fmode_t mode, } #endif /* CONFIG_COMPAT */ +static int md_set_read_only(struct block_device *bdev, bool ro) +{ + struct mddev *mddev = bdev->bd_disk->private_data; + int err; + + err = mddev_lock(mddev); + if (err) + return err; + + if (!mddev->raid_disks && !mddev->external) { + err = -ENODEV; + goto out_unlock; + } + + /* + * Transitioning to read-auto need only happen for arrays that call + * md_write_start and which are not ready for writes yet. + */ + if (!ro && mddev->ro == 1 && mddev->pers) { + err = restart_array(mddev); + if (err) + goto out_unlock; + mddev->ro = 2; + } + +out_unlock: + mddev_unlock(mddev); + return err; +} + static int md_open(struct block_device *bdev, fmode_t mode) { /* @@ -7886,6 +7885,7 @@ const struct block_device_operations md_fops = #endif .getgeo = md_getgeo, .check_events = md_check_events, + .set_read_only = md_set_read_only, }; static int md_thread(void *arg) -- cgit v1.2.3 From 2c1b6ec170423bb032e825d4d9b2780bac4ce1a9 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:14 +0100 Subject: dasd: implement ->set_read_only to hook into BLKROSET processing Implement the ->set_read_only method instead of parsing the actual ioctl command. Signed-off-by: Christoph Hellwig Reviewed-by: Stefan Haberland Signed-off-by: Jens Axboe --- drivers/s390/block/dasd.c | 1 + drivers/s390/block/dasd_int.h | 3 ++- drivers/s390/block/dasd_ioctl.c | 27 +++++++++------------------ 3 files changed, 12 insertions(+), 19 deletions(-) diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index eb17fea8075c..db24e04ee978 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -3394,6 +3394,7 @@ dasd_device_operations = { .ioctl = dasd_ioctl, .compat_ioctl = dasd_ioctl, .getgeo = dasd_getgeo, + .set_read_only = dasd_set_read_only, }; /******************************************************************************* diff --git a/drivers/s390/block/dasd_int.h b/drivers/s390/block/dasd_int.h index fa552f9f1666..c59a0d63b506 100644 --- a/drivers/s390/block/dasd_int.h +++ b/drivers/s390/block/dasd_int.h @@ -844,7 +844,8 @@ int dasd_scan_partitions(struct dasd_block *); void dasd_destroy_partitions(struct dasd_block *); /* externals in dasd_ioctl.c */ -int dasd_ioctl(struct block_device *, fmode_t, unsigned int, unsigned long); +int dasd_ioctl(struct block_device *, fmode_t, unsigned int, unsigned long); +int dasd_set_read_only(struct block_device *bdev, bool ro); /* externals in dasd_proc.c */ int dasd_proc_init(void); diff --git a/drivers/s390/block/dasd_ioctl.c b/drivers/s390/block/dasd_ioctl.c index cb6427fb9f3d..3359559517bf 100644 --- a/drivers/s390/block/dasd_ioctl.c +++ b/drivers/s390/block/dasd_ioctl.c @@ -532,28 +532,22 @@ static int dasd_ioctl_information(struct dasd_block *block, void __user *argp, /* * Set read only */ -static int -dasd_ioctl_set_ro(struct block_device *bdev, void __user *argp) +int dasd_set_read_only(struct block_device *bdev, bool ro) { struct dasd_device *base; - int intval, rc; + int rc; - if (!capable(CAP_SYS_ADMIN)) - return -EACCES; + /* do not manipulate hardware state for partitions */ if (bdev_is_partition(bdev)) - // ro setting is not allowed for partitions - return -EINVAL; - if (get_user(intval, (int __user *)argp)) - return -EFAULT; + return 0; + base = dasd_device_from_gendisk(bdev->bd_disk); if (!base) return -ENODEV; - if (!intval && test_bit(DASD_FLAG_DEVICE_RO, &base->flags)) { - dasd_put_device(base); - return -EROFS; - } - set_disk_ro(bdev->bd_disk, intval); - rc = dasd_set_feature(base->cdev, DASD_FEATURE_READONLY, intval); + if (!ro && test_bit(DASD_FLAG_DEVICE_RO, &base->flags)) + rc = -EROFS; + else + rc = dasd_set_feature(base->cdev, DASD_FEATURE_READONLY, ro); dasd_put_device(base); return rc; } @@ -633,9 +627,6 @@ int dasd_ioctl(struct block_device *bdev, fmode_t mode, case BIODASDPRRST: rc = dasd_ioctl_reset_profile(block); break; - case BLKROSET: - rc = dasd_ioctl_set_ro(bdev, argp); - break; case DASDAPIVER: rc = dasd_ioctl_api_version(argp); break; -- cgit v1.2.3 From 732e12d805a77f74c907c0a28ece271ef1e81e01 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:15 +0100 Subject: block: don't call into the driver for BLKROSET Now that all drivers that want to hook into setting or clearing the read-only flag use the set_read_only method, this code can be removed. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- block/ioctl.c | 23 ----------------------- 1 file changed, 23 deletions(-) diff --git a/block/ioctl.c b/block/ioctl.c index a6fa16b97705..96cb45447364 100644 --- a/block/ioctl.c +++ b/block/ioctl.c @@ -346,26 +346,6 @@ static int blkdev_pr_clear(struct block_device *bdev, return ops->pr_clear(bdev, c.key); } -/* - * Is it an unrecognized ioctl? The correct returns are either - * ENOTTY (final) or ENOIOCTLCMD ("I don't know this one, try a - * fallback"). ENOIOCTLCMD gets turned into ENOTTY by the ioctl - * code before returning. - * - * Confused drivers sometimes return EINVAL, which is wrong. It - * means "I understood the ioctl command, but the parameters to - * it were wrong". - * - * We should aim to just fix the broken drivers, the EINVAL case - * should go away. - */ -static inline int is_unrecognized_ioctl(int ret) -{ - return ret == -EINVAL || - ret == -ENOTTY || - ret == -ENOIOCTLCMD; -} - static int blkdev_flushbuf(struct block_device *bdev, fmode_t mode, unsigned cmd, unsigned long arg) { @@ -384,9 +364,6 @@ static int blkdev_roset(struct block_device *bdev, fmode_t mode, if (!capable(CAP_SYS_ADMIN)) return -EACCES; - ret = __blkdev_driver_ioctl(bdev, mode, cmd, arg); - if (!is_unrecognized_ioctl(ret)) - return ret; if (get_user(n, (int __user *)arg)) return -EFAULT; if (bdev->bd_disk->fops->set_read_only) { -- cgit v1.2.3 From 7a2f0ce19f2e2ed93b5ace60375b908392ec8afb Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:16 +0100 Subject: loop: use set_disk_ro Use set_disk_ro instead of set_device_ro to match all other block drivers and to ensure all partitions mirror the read-only flag. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/block/loop.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/block/loop.c b/drivers/block/loop.c index a58084c2ed7c..027f5724fe2c 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -1138,7 +1138,7 @@ static int loop_configure(struct loop_device *lo, fmode_t mode, if (error) goto out_unlock; - set_device_ro(bdev, (lo->lo_flags & LO_FLAGS_READ_ONLY) != 0); + set_disk_ro(lo->lo_disk, (lo->lo_flags & LO_FLAGS_READ_ONLY) != 0); lo->use_dio = lo->lo_flags & LO_FLAGS_DIRECT_IO; lo->lo_device = bdev; -- cgit v1.2.3 From 98f49b63e84d4ee1a5c327d0b5f4e8699f6c70fe Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:17 +0100 Subject: block: remove set_device_ro Fold set_device_ro into its only remaining caller. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- block/genhd.c | 7 ------- block/ioctl.c | 2 +- include/linux/genhd.h | 1 - 3 files changed, 1 insertion(+), 9 deletions(-) diff --git a/block/genhd.c b/block/genhd.c index 9387f050c248..b85db1f2233c 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -1846,13 +1846,6 @@ static void set_disk_ro_uevent(struct gendisk *gd, int ro) kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp); } -void set_device_ro(struct block_device *bdev, int flag) -{ - bdev->bd_part->policy = flag; -} - -EXPORT_SYMBOL(set_device_ro); - void set_disk_ro(struct gendisk *disk, int flag) { struct disk_part_iter piter; diff --git a/block/ioctl.c b/block/ioctl.c index 96cb45447364..04255dc5f3bf 100644 --- a/block/ioctl.c +++ b/block/ioctl.c @@ -371,7 +371,7 @@ static int blkdev_roset(struct block_device *bdev, fmode_t mode, if (ret) return ret; } - set_device_ro(bdev, n); + bdev->bd_part->policy = n; return 0; } diff --git a/include/linux/genhd.h b/include/linux/genhd.h index 03da3f603d30..52eb592c874a 100644 --- a/include/linux/genhd.h +++ b/include/linux/genhd.h @@ -304,7 +304,6 @@ extern void del_gendisk(struct gendisk *gp); extern struct gendisk *get_gendisk(dev_t dev, int *partno); extern struct block_device *bdget_disk(struct gendisk *disk, int partno); -extern void set_device_ro(struct block_device *bdev, int flag); extern void set_disk_ro(struct gendisk *disk, int flag); static inline int get_disk_ro(struct gendisk *disk) -- cgit v1.2.3 From a7cb3d2f09c8405aed59d97a7d02cebea43cd3c7 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 3 Nov 2020 11:00:18 +0100 Subject: block: remove __blkdev_driver_ioctl Just open code it in the few callers. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- block/ioctl.c | 25 +++++-------------------- drivers/block/pktcdvd.c | 6 ++++-- drivers/md/bcache/request.c | 5 +++-- drivers/md/dm.c | 5 ++++- include/linux/blkdev.h | 2 -- 5 files changed, 16 insertions(+), 27 deletions(-) diff --git a/block/ioctl.c b/block/ioctl.c index 04255dc5f3bf..6b785181344f 100644 --- a/block/ioctl.c +++ b/block/ioctl.c @@ -219,23 +219,6 @@ static int compat_put_ulong(compat_ulong_t __user *argp, compat_ulong_t val) } #endif -int __blkdev_driver_ioctl(struct block_device *bdev, fmode_t mode, - unsigned cmd, unsigned long arg) -{ - struct gendisk *disk = bdev->bd_disk; - - if (disk->fops->ioctl) - return disk->fops->ioctl(bdev, mode, cmd, arg); - - return -ENOTTY; -} -/* - * For the record: _GPL here is only because somebody decided to slap it - * on the previous export. Sheer idiocy, since it wasn't copyrightable - * at all and could be open-coded without any exports by anybody who cares. - */ -EXPORT_SYMBOL_GPL(__blkdev_driver_ioctl); - #ifdef CONFIG_COMPAT /* * This is the equivalent of compat_ptr_ioctl(), to be used by block @@ -594,10 +577,12 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd, } ret = blkdev_common_ioctl(bdev, mode, cmd, arg, argp); - if (ret == -ENOIOCTLCMD) - return __blkdev_driver_ioctl(bdev, mode, cmd, arg); + if (ret != -ENOIOCTLCMD) + return ret; - return ret; + if (!bdev->bd_disk->fops->ioctl) + return -ENOTTY; + return bdev->bd_disk->fops->ioctl(bdev, mode, cmd, arg); } EXPORT_SYMBOL_GPL(blkdev_ioctl); /* for /dev/raw */ diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c index 467dbd06b7cd..ef1c1f094ea4 100644 --- a/drivers/block/pktcdvd.c +++ b/drivers/block/pktcdvd.c @@ -2584,9 +2584,11 @@ static int pkt_ioctl(struct block_device *bdev, fmode_t mode, unsigned int cmd, case CDROM_LAST_WRITTEN: case CDROM_SEND_PACKET: case SCSI_IOCTL_SEND_COMMAND: - ret = __blkdev_driver_ioctl(pd->bdev, mode, cmd, arg); + if (!bdev->bd_disk->fops->ioctl) + ret = -ENOTTY; + else + ret = bdev->bd_disk->fops->ioctl(bdev, mode, cmd, arg); break; - default: pkt_dbg(2, pd, "Unknown ioctl (%x)\n", cmd); ret = -ENOTTY; diff --git a/drivers/md/bcache/request.c b/drivers/md/bcache/request.c index 214326383145..afac8d07c1bd 100644 --- a/drivers/md/bcache/request.c +++ b/drivers/md/bcache/request.c @@ -1230,8 +1230,9 @@ static int cached_dev_ioctl(struct bcache_device *d, fmode_t mode, if (dc->io_disable) return -EIO; - - return __blkdev_driver_ioctl(dc->bdev, mode, cmd, arg); + if (!dc->bdev->bd_disk->fops->ioctl) + return -ENOTTY; + return dc->bdev->bd_disk->fops->ioctl(dc->bdev, mode, cmd, arg); } void bch_cached_dev_request_init(struct cached_dev *dc) diff --git a/drivers/md/dm.c b/drivers/md/dm.c index c18fc2548518..6db395c3d28b 100644 --- a/drivers/md/dm.c +++ b/drivers/md/dm.c @@ -570,7 +570,10 @@ static int dm_blk_ioctl(struct block_device *bdev, fmode_t mode, } } - r = __blkdev_driver_ioctl(bdev, mode, cmd, arg); + if (!bdev->bd_disk->fops->ioctl) + r = -ENOTTY; + else + r = bdev->bd_disk->fops->ioctl(bdev, mode, cmd, arg); out: dm_unprepare_ioctl(md, srcu_idx); return r; diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 5c1ba8a8d2bc..05b346a68c2e 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -1867,8 +1867,6 @@ extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t, #define blkdev_compat_ptr_ioctl NULL #endif -extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int, - unsigned long); extern int bdev_read_page(struct block_device *, sector_t, struct page *); extern int bdev_write_page(struct block_device *, sector_t, struct page *, struct writeback_control *); -- cgit v1.2.3 From 6b3ba9762f9f9f651873af34481ca20e4a6791e7 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:24 +0100 Subject: block: cleanup del_gendisk a bit Merge three hidden gendisk checks into one. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- block/genhd.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/block/genhd.c b/block/genhd.c index b85db1f2233c..d41176eb1f36 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -895,6 +895,9 @@ void del_gendisk(struct gendisk *disk) might_sleep(); + if (WARN_ON_ONCE(!disk->queue)) + return; + blk_integrity_del(disk); disk_del_events(disk); @@ -917,20 +920,18 @@ void del_gendisk(struct gendisk *disk) disk->flags &= ~GENHD_FL_UP; up_write(&disk->lookup_sem); - if (!(disk->flags & GENHD_FL_HIDDEN)) + if (!(disk->flags & GENHD_FL_HIDDEN)) { sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); - if (disk->queue) { + /* * Unregister bdi before releasing device numbers (as they can * get reused and we'd get clashes in sysfs). */ - if (!(disk->flags & GENHD_FL_HIDDEN)) - bdi_unregister(disk->queue->backing_dev_info); - blk_unregister_queue(disk); - } else { - WARN_ON(1); + bdi_unregister(disk->queue->backing_dev_info); } + blk_unregister_queue(disk); + if (!(disk->flags & GENHD_FL_HIDDEN)) blk_unregister_region(disk_devt(disk), disk->minors); /* -- cgit v1.2.3 From 62b508f8b6b1b52843cd90f0b2068ed963f25bd3 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:25 +0100 Subject: block: open code kobj_map into in block/genhd.c Copy and paste the kobj_map functionality in the block code in preparation for completely rewriting it. Signed-off-by: Christoph Hellwig Reviewed-by: Greg Kroah-Hartman Signed-off-by: Jens Axboe --- block/genhd.c | 130 ++++++++++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 117 insertions(+), 13 deletions(-) diff --git a/block/genhd.c b/block/genhd.c index d41176eb1f36..667d1d6fd70a 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -17,7 +17,6 @@ #include #include #include -#include #include #include #include @@ -29,6 +28,16 @@ static DEFINE_MUTEX(block_class_lock); static struct kobject *block_depr; +struct bdev_map { + struct bdev_map *next; + dev_t dev; + unsigned long range; + struct module *owner; + struct kobject *(*probe)(dev_t, int *, void *); + int (*lock)(dev_t, void *); + void *data; +} *bdev_map[255]; + /* for extended dynamic devt allocation, currently only one major is used */ #define NR_EXT_DEVT (1 << MINORBITS) @@ -520,8 +529,6 @@ void unregister_blkdev(unsigned int major, const char *name) EXPORT_SYMBOL(unregister_blkdev); -static struct kobj_map *bdev_map; - /** * blk_mangle_minor - scatter minor numbers apart * @minor: minor number to mangle @@ -648,16 +655,60 @@ void blk_register_region(dev_t devt, unsigned long range, struct module *module, struct kobject *(*probe)(dev_t, int *, void *), int (*lock)(dev_t, void *), void *data) { - kobj_map(bdev_map, devt, range, module, probe, lock, data); -} + unsigned n = MAJOR(devt + range - 1) - MAJOR(devt) + 1; + unsigned index = MAJOR(devt); + unsigned i; + struct bdev_map *p; + + n = min(n, 255u); + p = kmalloc_array(n, sizeof(struct bdev_map), GFP_KERNEL); + if (p == NULL) + return; + for (i = 0; i < n; i++, p++) { + p->owner = module; + p->probe = probe; + p->lock = lock; + p->dev = devt; + p->range = range; + p->data = data; + } + + mutex_lock(&block_class_lock); + for (i = 0, p -= n; i < n; i++, p++, index++) { + struct bdev_map **s = &bdev_map[index % 255]; + while (*s && (*s)->range < range) + s = &(*s)->next; + p->next = *s; + *s = p; + } + mutex_unlock(&block_class_lock); +} EXPORT_SYMBOL(blk_register_region); void blk_unregister_region(dev_t devt, unsigned long range) { - kobj_unmap(bdev_map, devt, range); -} + unsigned n = MAJOR(devt + range - 1) - MAJOR(devt) + 1; + unsigned index = MAJOR(devt); + unsigned i; + struct bdev_map *found = NULL; + mutex_lock(&block_class_lock); + for (i = 0; i < min(n, 255u); i++, index++) { + struct bdev_map **s; + for (s = &bdev_map[index % 255]; *s; s = &(*s)->next) { + struct bdev_map *p = *s; + if (p->dev == devt && p->range == range) { + *s = p->next; + if (!found) + found = p; + break; + } + } + } + mutex_unlock(&block_class_lock); + kfree(found); +} EXPORT_SYMBOL(blk_unregister_region); static struct kobject *exact_match(dev_t devt, int *partno, void *data) @@ -979,6 +1030,47 @@ static ssize_t disk_badblocks_store(struct device *dev, return badblocks_store(disk->bb, page, len, 0); } +static struct gendisk *lookup_gendisk(dev_t dev, int *partno) +{ + struct kobject *kobj; + struct bdev_map *p; + unsigned long best = ~0UL; + +retry: + mutex_lock(&block_class_lock); + for (p = bdev_map[MAJOR(dev) % 255]; p; p = p->next) { + struct kobject *(*probe)(dev_t, int *, void *); + struct module *owner; + void *data; + + if (p->dev > dev || p->dev + p->range - 1 < dev) + continue; + if (p->range - 1 >= best) + break; + if (!try_module_get(p->owner)) + continue; + owner = p->owner; + data = p->data; + probe = p->probe; + best = p->range - 1; + *partno = dev - p->dev; + if (p->lock && p->lock(dev, data) < 0) { + module_put(owner); + continue; + } + mutex_unlock(&block_class_lock); + kobj = probe(dev, partno, data); + /* Currently ->owner protects _only_ ->probe() itself. */ + module_put(owner); + if (kobj) + return dev_to_disk(kobj_to_dev(kobj)); + goto retry; + } + mutex_unlock(&block_class_lock); + return NULL; +} + + /** * get_gendisk - get partitioning information for a given device * @devt: device to get partitioning information for @@ -996,11 +1088,7 @@ struct gendisk *get_gendisk(dev_t devt, int *partno) might_sleep(); if (MAJOR(devt) != BLOCK_EXT_MAJOR) { - struct kobject *kobj; - - kobj = kobj_lookup(bdev_map, devt, partno); - if (kobj) - disk = dev_to_disk(kobj_to_dev(kobj)); + disk = lookup_gendisk(devt, partno); } else { struct hd_struct *part; @@ -1213,6 +1301,22 @@ static struct kobject *base_probe(dev_t devt, int *partno, void *data) return NULL; } +static void bdev_map_init(void) +{ + struct bdev_map *base; + int i; + + base = kzalloc(sizeof(*base), GFP_KERNEL); + if (!base) + panic("cannot allocate bdev_map"); + + base->dev = 1; + base->range = ~0 ; + base->probe = base_probe; + for (i = 0; i < 255; i++) + bdev_map[i] = base; +} + static int __init genhd_device_init(void) { int error; @@ -1221,7 +1325,7 @@ static int __init genhd_device_init(void) error = class_register(&block_class); if (unlikely(error)) return error; - bdev_map = kobj_map_init(base_probe, &block_class_lock); + bdev_map_init(); blk_dev_init(); register_blkdev(BLOCK_EXT_MAJOR, "blkext"); -- cgit v1.2.3 From e49fbbbf0aa14f011ab037086f37f58bd058a6ee Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:26 +0100 Subject: block: split block_class_lock Split the block_class_lock mutex into one each to protect bdev_map and major_names. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- block/genhd.c | 29 +++++++++++++++-------------- 1 file changed, 15 insertions(+), 14 deletions(-) diff --git a/block/genhd.c b/block/genhd.c index 667d1d6fd70a..8226add353be 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -25,7 +25,6 @@ #include "blk.h" -static DEFINE_MUTEX(block_class_lock); static struct kobject *block_depr; struct bdev_map { @@ -37,6 +36,7 @@ struct bdev_map { int (*lock)(dev_t, void *); void *data; } *bdev_map[255]; +static DEFINE_MUTEX(bdev_map_lock); /* for extended dynamic devt allocation, currently only one major is used */ #define NR_EXT_DEVT (1 << MINORBITS) @@ -403,6 +403,7 @@ static struct blk_major_name { int major; char name[16]; } *major_names[BLKDEV_MAJOR_HASH_SIZE]; +static DEFINE_MUTEX(major_names_lock); /* index in the above - for now: assume no multimajor ranges */ static inline int major_to_index(unsigned major) @@ -415,11 +416,11 @@ void blkdev_show(struct seq_file *seqf, off_t offset) { struct blk_major_name *dp; - mutex_lock(&block_class_lock); + mutex_lock(&major_names_lock); for (dp = major_names[major_to_index(offset)]; dp; dp = dp->next) if (dp->major == offset) seq_printf(seqf, "%3d %s\n", dp->major, dp->name); - mutex_unlock(&block_class_lock); + mutex_unlock(&major_names_lock); } #endif /* CONFIG_PROC_FS */ @@ -448,7 +449,7 @@ int register_blkdev(unsigned int major, const char *name) struct blk_major_name **n, *p; int index, ret = 0; - mutex_lock(&block_class_lock); + mutex_lock(&major_names_lock); /* temporary */ if (major == 0) { @@ -501,7 +502,7 @@ int register_blkdev(unsigned int major, const char *name) kfree(p); } out: - mutex_unlock(&block_class_lock); + mutex_unlock(&major_names_lock); return ret; } @@ -513,7 +514,7 @@ void unregister_blkdev(unsigned int major, const char *name) struct blk_major_name *p = NULL; int index = major_to_index(major); - mutex_lock(&block_class_lock); + mutex_lock(&major_names_lock); for (n = &major_names[index]; *n; n = &(*n)->next) if ((*n)->major == major) break; @@ -523,7 +524,7 @@ void unregister_blkdev(unsigned int major, const char *name) p = *n; *n = p->next; } - mutex_unlock(&block_class_lock); + mutex_unlock(&major_names_lock); kfree(p); } @@ -674,7 +675,7 @@ void blk_register_region(dev_t devt, unsigned long range, struct module *module, p->data = data; } - mutex_lock(&block_class_lock); + mutex_lock(&bdev_map_lock); for (i = 0, p -= n; i < n; i++, p++, index++) { struct bdev_map **s = &bdev_map[index % 255]; while (*s && (*s)->range < range) @@ -682,7 +683,7 @@ void blk_register_region(dev_t devt, unsigned long range, struct module *module, p->next = *s; *s = p; } - mutex_unlock(&block_class_lock); + mutex_unlock(&bdev_map_lock); } EXPORT_SYMBOL(blk_register_region); @@ -693,7 +694,7 @@ void blk_unregister_region(dev_t devt, unsigned long range) unsigned i; struct bdev_map *found = NULL; - mutex_lock(&block_class_lock); + mutex_lock(&bdev_map_lock); for (i = 0; i < min(n, 255u); i++, index++) { struct bdev_map **s; for (s = &bdev_map[index % 255]; *s; s = &(*s)->next) { @@ -706,7 +707,7 @@ void blk_unregister_region(dev_t devt, unsigned long range) } } } - mutex_unlock(&block_class_lock); + mutex_unlock(&bdev_map_lock); kfree(found); } EXPORT_SYMBOL(blk_unregister_region); @@ -1037,7 +1038,7 @@ static struct gendisk *lookup_gendisk(dev_t dev, int *partno) unsigned long best = ~0UL; retry: - mutex_lock(&block_class_lock); + mutex_lock(&bdev_map_lock); for (p = bdev_map[MAJOR(dev) % 255]; p; p = p->next) { struct kobject *(*probe)(dev_t, int *, void *); struct module *owner; @@ -1058,7 +1059,7 @@ retry: module_put(owner); continue; } - mutex_unlock(&block_class_lock); + mutex_unlock(&bdev_map_lock); kobj = probe(dev, partno, data); /* Currently ->owner protects _only_ ->probe() itself. */ module_put(owner); @@ -1066,7 +1067,7 @@ retry: return dev_to_disk(kobj_to_dev(kobj)); goto retry; } - mutex_unlock(&block_class_lock); + mutex_unlock(&bdev_map_lock); return NULL; } -- cgit v1.2.3 From bd8eff3ba2caca53ea72cf3cc87a7797771dd7d2 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:27 +0100 Subject: block: rework requesting modules for unclaimed devices Instead of reusing the ranges in bdev_map, add a new helper that is called if no ranges was found. This is a first step to unpeel and eventually remove the complex ranges structure. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- block/genhd.c | 25 +++++++++++++++---------- 1 file changed, 15 insertions(+), 10 deletions(-) diff --git a/block/genhd.c b/block/genhd.c index 8226add353be..81017bd3b333 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -1031,6 +1031,13 @@ static ssize_t disk_badblocks_store(struct device *dev, return badblocks_store(disk->bb, page, len, 0); } +static void request_gendisk_module(dev_t devt) +{ + if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) + /* Make old-style 2.4 aliases work */ + request_module("block-major-%d", MAJOR(devt)); +} + static struct gendisk *lookup_gendisk(dev_t dev, int *partno) { struct kobject *kobj; @@ -1055,6 +1062,14 @@ retry: probe = p->probe; best = p->range - 1; *partno = dev - p->dev; + + if (!probe) { + mutex_unlock(&bdev_map_lock); + module_put(owner); + request_gendisk_module(dev); + goto retry; + } + if (p->lock && p->lock(dev, data) < 0) { module_put(owner); continue; @@ -1293,15 +1308,6 @@ static const struct seq_operations partitions_op = { }; #endif - -static struct kobject *base_probe(dev_t devt, int *partno, void *data) -{ - if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) - /* Make old-style 2.4 aliases work */ - request_module("block-major-%d", MAJOR(devt)); - return NULL; -} - static void bdev_map_init(void) { struct bdev_map *base; @@ -1313,7 +1319,6 @@ static void bdev_map_init(void) base->dev = 1; base->range = ~0 ; - base->probe = base_probe; for (i = 0; i < 255; i++) bdev_map[i] = base; } -- cgit v1.2.3 From a160c6159d4a0cf82f28bc1658a958e278ec3688 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:28 +0100 Subject: block: add an optional probe callback to major_names Add a callback to the major_names array that allows a driver to override how to probe for dev_t that doesn't currently have a gendisk registered. This will help separating the lookup of the gendisk by dev_t vs probe action for a not currently registered dev_t. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- block/genhd.c | 21 ++++++++++++++++++--- include/linux/genhd.h | 5 ++++- 2 files changed, 22 insertions(+), 4 deletions(-) diff --git a/block/genhd.c b/block/genhd.c index 81017bd3b333..20521163fd06 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -402,6 +402,7 @@ static struct blk_major_name { struct blk_major_name *next; int major; char name[16]; + void (*probe)(dev_t devt); } *major_names[BLKDEV_MAJOR_HASH_SIZE]; static DEFINE_MUTEX(major_names_lock); @@ -444,7 +445,8 @@ void blkdev_show(struct seq_file *seqf, off_t offset) * See Documentation/admin-guide/devices.txt for the list of allocated * major numbers. */ -int register_blkdev(unsigned int major, const char *name) +int __register_blkdev(unsigned int major, const char *name, + void (*probe)(dev_t devt)) { struct blk_major_name **n, *p; int index, ret = 0; @@ -483,6 +485,7 @@ int register_blkdev(unsigned int major, const char *name) } p->major = major; + p->probe = probe; strlcpy(p->name, name, sizeof(p->name)); p->next = NULL; index = major_to_index(major); @@ -505,8 +508,7 @@ out: mutex_unlock(&major_names_lock); return ret; } - -EXPORT_SYMBOL(register_blkdev); +EXPORT_SYMBOL(__register_blkdev); void unregister_blkdev(unsigned int major, const char *name) { @@ -1033,6 +1035,19 @@ static ssize_t disk_badblocks_store(struct device *dev, static void request_gendisk_module(dev_t devt) { + unsigned int major = MAJOR(devt); + struct blk_major_name **n; + + mutex_lock(&major_names_lock); + for (n = &major_names[major_to_index(major)]; *n; n = &(*n)->next) { + if ((*n)->major == major && (*n)->probe) { + (*n)->probe(devt); + mutex_unlock(&major_names_lock); + return; + } + } + mutex_unlock(&major_names_lock); + if (request_module("block-major-%d-%d", MAJOR(devt), MINOR(devt)) > 0) /* Make old-style 2.4 aliases work */ request_module("block-major-%d", MAJOR(devt)); diff --git a/include/linux/genhd.h b/include/linux/genhd.h index 52eb592c874a..811d0f83c4cf 100644 --- a/include/linux/genhd.h +++ b/include/linux/genhd.h @@ -367,7 +367,10 @@ extern void blk_unregister_region(dev_t devt, unsigned long range); #define alloc_disk(minors) alloc_disk_node(minors, NUMA_NO_NODE) -int register_blkdev(unsigned int major, const char *name); +int __register_blkdev(unsigned int major, const char *name, + void (*probe)(dev_t devt)); +#define register_blkdev(major, name) \ + __register_blkdev(major, name, NULL) void unregister_blkdev(unsigned int major, const char *name); void revalidate_disk_size(struct gendisk *disk, bool verbose); -- cgit v1.2.3 From d18e8b1bf9e2ee814a7f886a156bf762d52e178b Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:29 +0100 Subject: ide: remove ide_{,un}register_region There is no need to ever register the fake gendisk used for ide-tape. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- drivers/ide/ide-probe.c | 32 -------------------------------- drivers/ide/ide-tape.c | 2 -- include/linux/ide.h | 3 --- 3 files changed, 37 deletions(-) diff --git a/drivers/ide/ide-probe.c b/drivers/ide/ide-probe.c index 1ddc45a04418..076d34b38172 100644 --- a/drivers/ide/ide-probe.c +++ b/drivers/ide/ide-probe.c @@ -929,38 +929,6 @@ static struct kobject *ata_probe(dev_t dev, int *part, void *data) return NULL; } -static struct kobject *exact_match(dev_t dev, int *part, void *data) -{ - struct gendisk *p = data; - *part &= (1 << PARTN_BITS) - 1; - return &disk_to_dev(p)->kobj; -} - -static int exact_lock(dev_t dev, void *data) -{ - struct gendisk *p = data; - - if (!get_disk_and_module(p)) - return -1; - return 0; -} - -void ide_register_region(struct gendisk *disk) -{ - blk_register_region(MKDEV(disk->major, disk->first_minor), - disk->minors, NULL, exact_match, exact_lock, disk); -} - -EXPORT_SYMBOL_GPL(ide_register_region); - -void ide_unregister_region(struct gendisk *disk) -{ - blk_unregister_region(MKDEV(disk->major, disk->first_minor), - disk->minors); -} - -EXPORT_SYMBOL_GPL(ide_unregister_region); - void ide_init_disk(struct gendisk *disk, ide_drive_t *drive) { ide_hwif_t *hwif = drive->hwif; diff --git a/drivers/ide/ide-tape.c b/drivers/ide/ide-tape.c index 6f26634b22bb..88b96437b22e 100644 --- a/drivers/ide/ide-tape.c +++ b/drivers/ide/ide-tape.c @@ -1822,7 +1822,6 @@ static void ide_tape_remove(ide_drive_t *drive) ide_proc_unregister_driver(drive, tape->driver); device_del(&tape->dev); - ide_unregister_region(tape->disk); mutex_lock(&idetape_ref_mutex); put_device(&tape->dev); @@ -2026,7 +2025,6 @@ static int ide_tape_probe(ide_drive_t *drive) "n%s", tape->name); g->fops = &idetape_block_ops; - ide_register_region(g); return 0; diff --git a/include/linux/ide.h b/include/linux/ide.h index 62653769509f..2c300689a51a 100644 --- a/include/linux/ide.h +++ b/include/linux/ide.h @@ -1493,9 +1493,6 @@ static inline void ide_acpi_port_init_devices(ide_hwif_t *hwif) { ; } static inline void ide_acpi_set_state(ide_hwif_t *hwif, int on) {} #endif -void ide_register_region(struct gendisk *); -void ide_unregister_region(struct gendisk *); - void ide_check_nien_quirk_list(ide_drive_t *); void ide_undecoded_slave(ide_drive_t *); -- cgit v1.2.3 From f9550f1b3981755b0ba9bd385c3c6efdfadd8fe4 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:30 +0100 Subject: swim: don't call blk_register_region The swim driver (unlike various other floppy drivers) doesn't have magic device nodes for certain modes, and already registers a gendisk for each of the floppies supported by a device. Thus the region registered is a no-op and can be removed. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- drivers/block/swim.c | 17 ----------------- 1 file changed, 17 deletions(-) diff --git a/drivers/block/swim.c b/drivers/block/swim.c index 52dd1efa00f9..cc6a0bc6c005 100644 --- a/drivers/block/swim.c +++ b/drivers/block/swim.c @@ -745,18 +745,6 @@ static const struct block_device_operations floppy_fops = { .check_events = floppy_check_events, }; -static struct kobject *floppy_find(dev_t dev, int *part, void *data) -{ - struct swim_priv *swd = data; - int drive = (*part & 3); - - if (drive >= swd->floppy_count) - return NULL; - - *part = 0; - return get_disk_and_module(swd->unit[drive].disk); -} - static int swim_add_floppy(struct swim_priv *swd, enum drive_location location) { struct floppy_state *fs = &swd->unit[swd->floppy_count]; @@ -846,9 +834,6 @@ static int swim_floppy_init(struct swim_priv *swd) add_disk(swd->unit[drive].disk); } - blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE, - floppy_find, NULL, swd); - return 0; exit_put_disks: @@ -932,8 +917,6 @@ static int swim_remove(struct platform_device *dev) int drive; struct resource *res; - blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); - for (drive = 0; drive < swd->floppy_count; drive++) { del_gendisk(swd->unit[drive].disk); blk_cleanup_queue(swd->unit[drive].disk->queue); -- cgit v1.2.3 From 996e509bbc956d60f761dc955f5c5131e9a8fb13 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:31 +0100 Subject: sd: use __register_blkdev to avoid a modprobe for an unregistered dev_t Switch from using blk_register_region to the probe callback passed to __register_blkdev to disable the request_module call for an unclaimed dev_t in the SD majors. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Acked-by: Martin K. Petersen Signed-off-by: Jens Axboe --- drivers/scsi/sd.c | 19 +++++-------------- 1 file changed, 5 insertions(+), 14 deletions(-) diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c index 656bcf4940d6..106a9cda0eb7 100644 --- a/drivers/scsi/sd.c +++ b/drivers/scsi/sd.c @@ -630,13 +630,11 @@ static struct scsi_driver sd_template = { }; /* - * Dummy kobj_map->probe function. - * The default ->probe function will call modprobe, which is - * pointless as this module is already loaded. + * Don't request a new module, as that could deadlock in multipath + * environment. */ -static struct kobject *sd_default_probe(dev_t devt, int *partno, void *data) +static void sd_default_probe(dev_t devt) { - return NULL; } /* @@ -3528,9 +3526,6 @@ static int sd_remove(struct device *dev) free_opal_dev(sdkp->opal_dev); - blk_register_region(devt, SD_MINORS, NULL, - sd_default_probe, NULL, NULL); - mutex_lock(&sd_ref_mutex); dev_set_drvdata(dev, NULL); put_device(&sdkp->dev); @@ -3720,11 +3715,9 @@ static int __init init_sd(void) SCSI_LOG_HLQUEUE(3, printk("init_sd: sd driver entry point\n")); for (i = 0; i < SD_MAJORS; i++) { - if (register_blkdev(sd_major(i), "sd") != 0) + if (__register_blkdev(sd_major(i), "sd", sd_default_probe)) continue; majors++; - blk_register_region(sd_major(i), SD_MINORS, NULL, - sd_default_probe, NULL, NULL); } if (!majors) @@ -3797,10 +3790,8 @@ static void __exit exit_sd(void) class_unregister(&sd_disk_class); - for (i = 0; i < SD_MAJORS; i++) { - blk_unregister_region(sd_major(i), SD_MINORS); + for (i = 0; i < SD_MAJORS; i++) unregister_blkdev(sd_major(i), "sd"); - } } module_init(init_sd); -- cgit v1.2.3 From 7cc178a6b994b7c994f1811c4b9fedc015ee0c9e Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:32 +0100 Subject: brd: use __register_blkdev to allocate devices on demand Use the simpler mechanism attached to major_name to allocate a brd device when a currently unregistered minor is accessed. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- drivers/block/brd.c | 39 +++++++++++---------------------------- 1 file changed, 11 insertions(+), 28 deletions(-) diff --git a/drivers/block/brd.c b/drivers/block/brd.c index cc49a921339f..c43a6ab4b1f3 100644 --- a/drivers/block/brd.c +++ b/drivers/block/brd.c @@ -426,14 +426,15 @@ static void brd_free(struct brd_device *brd) kfree(brd); } -static struct brd_device *brd_init_one(int i, bool *new) +static void brd_probe(dev_t dev) { struct brd_device *brd; + int i = MINOR(dev) / max_part; - *new = false; + mutex_lock(&brd_devices_mutex); list_for_each_entry(brd, &brd_devices, brd_list) { if (brd->brd_number == i) - goto out; + goto out_unlock; } brd = brd_alloc(i); @@ -442,9 +443,9 @@ static struct brd_device *brd_init_one(int i, bool *new) add_disk(brd->brd_disk); list_add_tail(&brd->brd_list, &brd_devices); } - *new = true; -out: - return brd; + +out_unlock: + mutex_unlock(&brd_devices_mutex); } static void brd_del_one(struct brd_device *brd) @@ -454,23 +455,6 @@ static void brd_del_one(struct brd_device *brd) brd_free(brd); } -static struct kobject *brd_probe(dev_t dev, int *part, void *data) -{ - struct brd_device *brd; - struct kobject *kobj; - bool new; - - mutex_lock(&brd_devices_mutex); - brd = brd_init_one(MINOR(dev) / max_part, &new); - kobj = brd ? get_disk_and_module(brd->brd_disk) : NULL; - mutex_unlock(&brd_devices_mutex); - - if (new) - *part = 0; - - return kobj; -} - static inline void brd_check_and_reset_par(void) { if (unlikely(!max_part)) @@ -510,11 +494,12 @@ static int __init brd_init(void) * dynamically. */ - if (register_blkdev(RAMDISK_MAJOR, "ramdisk")) + if (__register_blkdev(RAMDISK_MAJOR, "ramdisk", brd_probe)) return -EIO; brd_check_and_reset_par(); + mutex_lock(&brd_devices_mutex); for (i = 0; i < rd_nr; i++) { brd = brd_alloc(i); if (!brd) @@ -532,9 +517,7 @@ static int __init brd_init(void) brd->brd_disk->queue = brd->brd_queue; add_disk(brd->brd_disk); } - - blk_register_region(MKDEV(RAMDISK_MAJOR, 0), 1UL << MINORBITS, - THIS_MODULE, brd_probe, NULL, NULL); + mutex_unlock(&brd_devices_mutex); pr_info("brd: module loaded\n"); return 0; @@ -544,6 +527,7 @@ out_free: list_del(&brd->brd_list); brd_free(brd); } + mutex_unlock(&brd_devices_mutex); unregister_blkdev(RAMDISK_MAJOR, "ramdisk"); pr_info("brd: module NOT loaded !!!\n"); @@ -557,7 +541,6 @@ static void __exit brd_exit(void) list_for_each_entry_safe(brd, next, &brd_devices, brd_list) brd_del_one(brd); - blk_unregister_region(MKDEV(RAMDISK_MAJOR, 0), 1UL << MINORBITS); unregister_blkdev(RAMDISK_MAJOR, "ramdisk"); pr_info("brd: module unloaded\n"); -- cgit v1.2.3 From 8410d38c255200f71b67ddb37021c36273e7e78c Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:33 +0100 Subject: loop: use __register_blkdev to allocate devices on demand Use the simpler mechanism attached to major_name to allocate a brd device when a currently unregistered minor is accessed. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- drivers/block/loop.c | 30 ++++++++---------------------- 1 file changed, 8 insertions(+), 22 deletions(-) diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 027f5724fe2c..22e59410b971 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -2235,24 +2235,18 @@ out: return ret; } -static struct kobject *loop_probe(dev_t dev, int *part, void *data) +static void loop_probe(dev_t dev) { + int idx = MINOR(dev) >> part_shift; struct loop_device *lo; - struct kobject *kobj; - int err; + + if (max_loop && idx >= max_loop) + return; mutex_lock(&loop_ctl_mutex); - err = loop_lookup(&lo, MINOR(dev) >> part_shift); - if (err < 0) - err = loop_add(&lo, MINOR(dev) >> part_shift); - if (err < 0) - kobj = NULL; - else - kobj = get_disk_and_module(lo->lo_disk); + if (loop_lookup(&lo, idx) < 0) + loop_add(&lo, idx); mutex_unlock(&loop_ctl_mutex); - - *part = 0; - return kobj; } static long loop_control_ioctl(struct file *file, unsigned int cmd, @@ -2372,14 +2366,11 @@ static int __init loop_init(void) goto err_out; - if (register_blkdev(LOOP_MAJOR, "loop")) { + if (__register_blkdev(LOOP_MAJOR, "loop", loop_probe)) { err = -EIO; goto misc_out; } - blk_register_region(MKDEV(LOOP_MAJOR, 0), range, - THIS_MODULE, loop_probe, NULL, NULL); - /* pre-create number of devices given by config or max_loop */ mutex_lock(&loop_ctl_mutex); for (i = 0; i < nr; i++) @@ -2405,16 +2396,11 @@ static int loop_exit_cb(int id, void *ptr, void *data) static void __exit loop_exit(void) { - unsigned long range; - - range = max_loop ? max_loop << part_shift : 1UL << MINORBITS; - mutex_lock(&loop_ctl_mutex); idr_for_each(&loop_index_idr, &loop_exit_cb, NULL); idr_destroy(&loop_index_idr); - blk_unregister_region(MKDEV(LOOP_MAJOR, 0), range); unregister_blkdev(LOOP_MAJOR, "loop"); misc_deregister(&loop_misc); -- cgit v1.2.3 From 28144f9998e047a9bac31421914335c6bc6eaa67 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:34 +0100 Subject: md: use __register_blkdev to allocate devices on demand Use the simpler mechanism attached to major_name to allocate a md device when a currently unregistered minor is accessed. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Acked-by: Song Liu Signed-off-by: Jens Axboe --- drivers/md/md.c | 21 ++++++++------------- 1 file changed, 8 insertions(+), 13 deletions(-) diff --git a/drivers/md/md.c b/drivers/md/md.c index 96d31336ad43..3c79243e9d07 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -5765,11 +5765,12 @@ static int md_alloc(dev_t dev, char *name) return error; } -static struct kobject *md_probe(dev_t dev, int *part, void *data) +static void md_probe(dev_t dev) { + if (MAJOR(dev) == MD_MAJOR && MINOR(dev) >= 512) + return; if (create_on_open) md_alloc(dev, NULL); - return NULL; } static int add_named_array(const char *val, const struct kernel_param *kp) @@ -6535,7 +6536,7 @@ static void autorun_devices(int part) break; } - md_probe(dev, NULL, NULL); + md_probe(dev); mddev = mddev_find(dev); if (!mddev || !mddev->gendisk) { if (mddev) @@ -9567,18 +9568,15 @@ static int __init md_init(void) if (!md_rdev_misc_wq) goto err_rdev_misc_wq; - if ((ret = register_blkdev(MD_MAJOR, "md")) < 0) + ret = __register_blkdev(MD_MAJOR, "md", md_probe); + if (ret < 0) goto err_md; - if ((ret = register_blkdev(0, "mdp")) < 0) + ret = __register_blkdev(0, "mdp", md_probe); + if (ret < 0) goto err_mdp; mdp_major = ret; - blk_register_region(MKDEV(MD_MAJOR, 0), 512, THIS_MODULE, - md_probe, NULL, NULL); - blk_register_region(MKDEV(mdp_major, 0), 1UL< Date: Thu, 29 Oct 2020 15:58:35 +0100 Subject: ide: switch to __register_blkdev for command set probing ide is the last user of the blk_register_region framework except for the tracking of allocated gendisk. Switch to __register_blkdev, even if that doesn't allow us to trivially find out which command set to probe for. That means we now always request all modules when a user tries to access an unclaimed ide device node, but except for a few potentially loaded modules for a fringe use case of a deprecated and soon to be removed driver that doesn't make a difference. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Signed-off-by: Jens Axboe --- drivers/ide/ide-probe.c | 34 ++++++---------------------------- 1 file changed, 6 insertions(+), 28 deletions(-) diff --git a/drivers/ide/ide-probe.c b/drivers/ide/ide-probe.c index 076d34b38172..1c1567bb5194 100644 --- a/drivers/ide/ide-probe.c +++ b/drivers/ide/ide-probe.c @@ -902,31 +902,12 @@ out_up: return 1; } -static int ata_lock(dev_t dev, void *data) +static void ata_probe(dev_t dev) { - /* FIXME: we want to pin hwif down */ - return 0; -} - -static struct kobject *ata_probe(dev_t dev, int *part, void *data) -{ - ide_hwif_t *hwif = data; - int unit = *part >> PARTN_BITS; - ide_drive_t *drive = hwif->devices[unit]; - - if ((drive->dev_flags & IDE_DFLAG_PRESENT) == 0) - return NULL; - - if (drive->media == ide_disk) - request_module("ide-disk"); - if (drive->media == ide_cdrom || drive->media == ide_optical) - request_module("ide-cd"); - if (drive->media == ide_tape) - request_module("ide-tape"); - if (drive->media == ide_floppy) - request_module("ide-floppy"); - - return NULL; + request_module("ide-disk"); + request_module("ide-cd"); + request_module("ide-tape"); + request_module("ide-floppy"); } void ide_init_disk(struct gendisk *disk, ide_drive_t *drive) @@ -967,7 +948,7 @@ static int hwif_init(ide_hwif_t *hwif) return 0; } - if (register_blkdev(hwif->major, hwif->name)) + if (__register_blkdev(hwif->major, hwif->name, ata_probe)) return 0; if (!hwif->sg_max_nents) @@ -989,8 +970,6 @@ static int hwif_init(ide_hwif_t *hwif) goto out; } - blk_register_region(MKDEV(hwif->major, 0), MAX_DRIVES << PARTN_BITS, - THIS_MODULE, ata_probe, ata_lock, hwif); return 1; out: @@ -1582,7 +1561,6 @@ static void ide_unregister(ide_hwif_t *hwif) /* * Remove us from the kernel's knowledge */ - blk_unregister_region(MKDEV(hwif->major, 0), MAX_DRIVES<sg_table); unregister_blkdev(hwif->major, hwif->name); -- cgit v1.2.3 From 302cfee150291c6cd85b1ca197d062d0b423d09c Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 29 Oct 2020 15:58:36 +0100 Subject: floppy: use a separate gendisk for each media format The floppy driver usually autodetects the media when used with the normal /dev/fd? devices, which also are the only nodes created by udev. But it also supports various aliases that force a given media format. That is currently supported using the blk_register_region framework which finds the floppy gendisk even for a 'mismatched' dev_t. The problem with this (besides the code complexity) is that it creates multiple struct block_device instances for the whole device of a single gendisk, which can lead to interesting issues in code not aware of that fact. To fix this just create a separate gendisk for each of the aliases if they are accessed. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- drivers/block/floppy.c | 154 +++++++++++++++++++++++++++++++------------------ 1 file changed, 97 insertions(+), 57 deletions(-) diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c index 7df79ae6b0a1..dfe1dfc901cc 100644 --- a/drivers/block/floppy.c +++ b/drivers/block/floppy.c @@ -402,7 +402,6 @@ static struct floppy_drive_params drive_params[N_DRIVE]; static struct floppy_drive_struct drive_state[N_DRIVE]; static struct floppy_write_errors write_errors[N_DRIVE]; static struct timer_list motor_off_timer[N_DRIVE]; -static struct gendisk *disks[N_DRIVE]; static struct blk_mq_tag_set tag_sets[N_DRIVE]; static struct block_device *opened_bdev[N_DRIVE]; static DEFINE_MUTEX(open_lock); @@ -477,6 +476,8 @@ static struct floppy_struct floppy_type[32] = { { 3200,20,2,80,0,0x1C,0x00,0xCF,0x2C,"H1600" }, /* 31 1.6MB 3.5" */ }; +static struct gendisk *disks[N_DRIVE][ARRAY_SIZE(floppy_type)]; + #define SECTSIZE (_FD_SECTSIZE(*floppy)) /* Auto-detection: Disk type used until the next media change occurs. */ @@ -4111,7 +4112,7 @@ static int floppy_open(struct block_device *bdev, fmode_t mode) new_dev = MINOR(bdev->bd_dev); drive_state[drive].fd_device = new_dev; - set_capacity(disks[drive], floppy_sizes[new_dev]); + set_capacity(disks[drive][ITYPE(new_dev)], floppy_sizes[new_dev]); if (old_dev != -1 && old_dev != new_dev) { if (buffer_drive == drive) buffer_track = -1; @@ -4579,15 +4580,58 @@ static bool floppy_available(int drive) return true; } -static struct kobject *floppy_find(dev_t dev, int *part, void *data) +static int floppy_alloc_disk(unsigned int drive, unsigned int type) { - int drive = (*part & 3) | ((*part & 0x80) >> 5); - if (drive >= N_DRIVE || !floppy_available(drive)) - return NULL; - if (((*part >> 2) & 0x1f) >= ARRAY_SIZE(floppy_type)) - return NULL; - *part = 0; - return get_disk_and_module(disks[drive]); + struct gendisk *disk; + int err; + + disk = alloc_disk(1); + if (!disk) + return -ENOMEM; + + disk->queue = blk_mq_init_queue(&tag_sets[drive]); + if (IS_ERR(disk->queue)) { + err = PTR_ERR(disk->queue); + disk->queue = NULL; + put_disk(disk); + return err; + } + + blk_queue_bounce_limit(disk->queue, BLK_BOUNCE_HIGH); + blk_queue_max_hw_sectors(disk->queue, 64); + disk->major = FLOPPY_MAJOR; + disk->first_minor = TOMINOR(drive) | (type << 2); + disk->fops = &floppy_fops; + disk->events = DISK_EVENT_MEDIA_CHANGE; + if (type) + sprintf(disk->disk_name, "fd%d_type%d", drive, type); + else + sprintf(disk->disk_name, "fd%d", drive); + /* to be cleaned up... */ + disk->private_data = (void *)(long)drive; + disk->flags |= GENHD_FL_REMOVABLE; + + disks[drive][type] = disk; + return 0; +} + +static DEFINE_MUTEX(floppy_probe_lock); + +static void floppy_probe(dev_t dev) +{ + unsigned int drive = (MINOR(dev) & 3) | ((MINOR(dev) & 0x80) >> 5); + unsigned int type = (MINOR(dev) >> 2) & 0x1f; + + if (drive >= N_DRIVE || !floppy_available(drive) || + type >= ARRAY_SIZE(floppy_type)) + return; + + mutex_lock(&floppy_probe_lock); + if (!disks[drive][type]) { + if (floppy_alloc_disk(drive, type) == 0) + add_disk(disks[drive][type]); + } + mutex_unlock(&floppy_probe_lock); } static int __init do_floppy_init(void) @@ -4609,33 +4653,25 @@ static int __init do_floppy_init(void) return -ENOMEM; for (drive = 0; drive < N_DRIVE; drive++) { - disks[drive] = alloc_disk(1); - if (!disks[drive]) { - err = -ENOMEM; + memset(&tag_sets[drive], 0, sizeof(tag_sets[drive])); + tag_sets[drive].ops = &floppy_mq_ops; + tag_sets[drive].nr_hw_queues = 1; + tag_sets[drive].nr_maps = 1; + tag_sets[drive].queue_depth = 2; + tag_sets[drive].numa_node = NUMA_NO_NODE; + tag_sets[drive].flags = BLK_MQ_F_SHOULD_MERGE; + err = blk_mq_alloc_tag_set(&tag_sets[drive]); + if (err) goto out_put_disk; - } - disks[drive]->queue = blk_mq_init_sq_queue(&tag_sets[drive], - &floppy_mq_ops, 2, - BLK_MQ_F_SHOULD_MERGE); - if (IS_ERR(disks[drive]->queue)) { - err = PTR_ERR(disks[drive]->queue); - disks[drive]->queue = NULL; + err = floppy_alloc_disk(drive, 0); + if (err) goto out_put_disk; - } - - blk_queue_bounce_limit(disks[drive]->queue, BLK_BOUNCE_HIGH); - blk_queue_max_hw_sectors(disks[drive]->queue, 64); - disks[drive]->major = FLOPPY_MAJOR; - disks[drive]->first_minor = TOMINOR(drive); - disks[drive]->fops = &floppy_fops; - disks[drive]->events = DISK_EVENT_MEDIA_CHANGE; - sprintf(disks[drive]->disk_name, "fd%d", drive); timer_setup(&motor_off_timer[drive], motor_off_callback, 0); } - err = register_blkdev(FLOPPY_MAJOR, "fd"); + err = __register_blkdev(FLOPPY_MAJOR, "fd", floppy_probe); if (err) goto out_put_disk; @@ -4643,9 +4679,6 @@ static int __init do_floppy_init(void) if (err) goto out_unreg_blkdev; - blk_register_region(MKDEV(FLOPPY_MAJOR, 0), 256, THIS_MODULE, - floppy_find, NULL, NULL); - for (i = 0; i < 256; i++) if (ITYPE(i)) floppy_sizes[i] = floppy_type[ITYPE(i)].size; @@ -4673,7 +4706,7 @@ static int __init do_floppy_init(void) if (fdc_state[0].address == -1) { cancel_delayed_work(&fd_timeout); err = -ENODEV; - goto out_unreg_region; + goto out_unreg_driver; } #if N_FDC > 1 fdc_state[1].address = FDC2; @@ -4684,7 +4717,7 @@ static int __init do_floppy_init(void) if (err) { cancel_delayed_work(&fd_timeout); err = -EBUSY; - goto out_unreg_region; + goto out_unreg_driver; } /* initialise drive state */ @@ -4761,10 +4794,8 @@ static int __init do_floppy_init(void) if (err) goto out_remove_drives; - /* to be cleaned up... */ - disks[drive]->private_data = (void *)(long)drive; - disks[drive]->flags |= GENHD_FL_REMOVABLE; - device_add_disk(&floppy_device[drive].dev, disks[drive], NULL); + device_add_disk(&floppy_device[drive].dev, disks[drive][0], + NULL); } return 0; @@ -4772,30 +4803,27 @@ static int __init do_floppy_init(void) out_remove_drives: while (drive--) { if (floppy_available(drive)) { - del_gendisk(disks[drive]); + del_gendisk(disks[drive][0]); platform_device_unregister(&floppy_device[drive]); } } out_release_dma: if (atomic_read(&usage_count)) floppy_release_irq_and_dma(); -out_unreg_region: - blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); +out_unreg_driver: platform_driver_unregister(&floppy_driver); out_unreg_blkdev: unregister_blkdev(FLOPPY_MAJOR, "fd"); out_put_disk: destroy_workqueue(floppy_wq); for (drive = 0; drive < N_DRIVE; drive++) { - if (!disks[drive]) + if (!disks[drive][0]) break; - if (disks[drive]->queue) { - del_timer_sync(&motor_off_timer[drive]); - blk_cleanup_queue(disks[drive]->queue); - disks[drive]->queue = NULL; - blk_mq_free_tag_set(&tag_sets[drive]); - } - put_disk(disks[drive]); + del_timer_sync(&motor_off_timer[drive]); + blk_cleanup_queue(disks[drive][0]->queue); + disks[drive][0]->queue = NULL; + blk_mq_free_tag_set(&tag_sets[drive]); + put_disk(disks[drive][0]); } return err; } @@ -5006,9 +5034,8 @@ module_init(floppy_module_init); static void __exit floppy_module_exit(void) { - int drive; + int drive, i; - blk_unregister_region(MKDEV(FLOPPY_MAJOR, 0), 256); unregister_blkdev(FLOPPY_MAJOR, "fd"); platform_driver_unregister(&floppy_driver); @@ -5018,10 +5045,16 @@ static void __exit floppy_module_exit(void) del_timer_sync(&motor_off_timer[drive]); if (floppy_available(drive)) { - del_gendisk(disks[drive]); + for (i = 0; i < ARRAY_SIZE(floppy_type); i++) { + if (disks[drive][i]) + del_gendisk(disks[drive][i]); + } platform_device_unregister(&floppy_device[drive]); } - blk_cleanup_queue(disks[drive]->queue); + for (i = 0; i < ARRAY_SIZE(floppy_type); i++) { + if (disks[drive][i]) + blk_cleanup_queue(disks[drive][i]->queue); + } blk_mq_free_tag_set(&tag_sets[drive]); /* @@ -5029,10 +5062,17 @@ static void __exit floppy_module_exit(void) * queue reference in put_disk(). */ if (!(allowed_drive_mask & (1 << drive)) || -