Subject: Re: [PATCH v2 1/3] md: Fix undefined behaviour in is_mddev_idle
To:     Li Jinlin <lijinlin3@huawei.com>, song@kernel.org,
        philipp.reisner@linbit.com, lars.ellenberg@linbit.com,
        axboe@kernel.dk, jack@suse.cz, ming.lei@redhat.com, tj@kernel.org,
        mcgrof@kernel.org, mcroce@microsoft.com
Cc:     linux-raid@vger.kernel.org, linux-block@vger.kernel.org,
        linux-kernel@vger.kernel.org, drbd-dev@lists.linbit.com,
        linfeilong@huawei.com
References: <20211210051707.2202646-1-lijinlin3@huawei.com>
 <20211210051707.2202646-2-lijinlin3@huawei.com>
From:   Hannes Reinecke <hare@suse.de>
Message-ID: <e9b03c2a-b1f6-e083-3a2b-bf1478628f31@suse.de>
Date:   Fri, 10 Dec 2021 07:45:52 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.12.0
MIME-Version: 1.0
In-Reply-To: <20211210051707.2202646-2-lijinlin3@huawei.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 12/10/21 6:17 AM, Li Jinlin wrote:
> UBSAN reports this problem:
> 
> [ 5984.281385] UBSAN: Undefined behaviour in drivers/md/md.c:8175:15
> [ 5984.281390] signed integer overflow:
> [ 5984.281393] -2147483291 - 2072033152 cannot be represented in type 'int'
> [ 5984.281400] CPU: 25 PID: 1854 Comm: md101_resync Kdump: loaded Not tainted 4.19.90
> [ 5984.281404] Hardware name: Huawei TaiShan 200 (Model 5280)/BC82AMDDA
> [ 5984.281406] Call trace:
> [ 5984.281415]  dump_backtrace+0x0/0x310
> [ 5984.281418]  show_stack+0x28/0x38
> [ 5984.281425]  dump_stack+0xec/0x15c
> [ 5984.281430]  ubsan_epilogue+0x18/0x84
> [ 5984.281434]  handle_overflow+0x14c/0x19c
> [ 5984.281439]  __ubsan_handle_sub_overflow+0x34/0x44
> [ 5984.281445]  is_mddev_idle+0x338/0x3d8
> [ 5984.281449]  md_do_sync+0x1bb8/0x1cf8
> [ 5984.281452]  md_thread+0x220/0x288
> [ 5984.281457]  kthread+0x1d8/0x1e0
> [ 5984.281461]  ret_from_fork+0x10/0x18
> 
> When the stat aacum of the disk is greater than INT_MAX, its
> value becomes negative after casting to 'int', which may lead
> to overflow after subtracting a positive number. In the same
> way, when the value of sync_io is greater than INT_MAX,
> overflow may also occur. These situations will lead to
> undefined behavior.
> 
> Otherwise, if the stat accum of the disk is close to INT_MAX
> when creating raid arrays, the initial value of last_events
> would be set close to INT_MAX when mddev initializes IO
> event counters. 'curr_events - rdev->last_events > 64' will
> always false during synchronization. If all the disks of mddev
> are in this case, is_mddev_idle() will always return 1, which
> may cause non-sync IO is very slow.
> 
> To address these problems, need to use 64bit signed integer
> type for sync_io, last_events, and curr_events.
> 
> In all the drivers that come with the kernel, the sync_io
> variable in struct gendisk is only used for the md driver
> currently. It should be more suitable in struct md_rdev, so
> add the sync_io variable in struct md_rdev, and use it to
> replace. We modify md_sync_acct() and md_sync_acct_bio()
> to fit for this change as well. md_sync_acct_bio() need
> access to the rdev, so we set bio->bi_bdev to rdev before
> calling it, and reset bio->bi_bdev to bdev in this function.
> 
Please make that two patches, one for moving sync_io and one for
fixing the undefined behaviour.

> Signed-off-by: Li Jinlin <lijinlin3@huawei.com>
> ---
>   drivers/md/md.c       |  6 +++---
>   drivers/md/md.h       | 13 +++++++++----
>   drivers/md/raid1.c    |  4 ++--
>   drivers/md/raid10.c   | 24 ++++++++++++------------
>   drivers/md/raid5.c    |  4 ++--
>   include/linux/genhd.h |  1 -
>   6 files changed, 28 insertions(+), 24 deletions(-)
> 
> diff --git a/drivers/md/md.c b/drivers/md/md.c
> index 5111ed966947..f1b71a92801e 100644
> --- a/drivers/md/md.c
> +++ b/drivers/md/md.c
> @@ -8429,14 +8429,14 @@ static int is_mddev_idle(struct mddev *mddev, int init)
>   {
>   	struct md_rdev *rdev;
>   	int idle;
> -	int curr_events;
> +	s64 curr_events;
>   
>   	idle = 1;
>   	rcu_read_lock();
>   	rdev_for_each_rcu(rdev, mddev) {
>   		struct gendisk *disk = rdev->bdev->bd_disk;
> -		curr_events = (int)part_stat_read_accum(disk->part0, sectors) -
> -			      atomic_read(&disk->sync_io);
> +		curr_events = (s64)part_stat_read_accum(disk->part0, sectors) -
> +			      atomic64_read(&rdev->sync_io);

So you are replacing a 'signed integer' (ie 32bit) calculation with a 
'signed 64-bit integer' calculation.
IE you just shifted the overflow from INT_MAX to LONG_MAX, without 
actually fixing it, or?

>   		/* sync IO will cause sync_io to increase before the disk_stats
>   		 * as sync_io is counted when a request starts, and
>   		 * disk_stats is counted when it completes.
> diff --git a/drivers/md/md.h b/drivers/md/md.h
> index 53ea7a6961de..584e357e0940 100644
> --- a/drivers/md/md.h
> +++ b/drivers/md/md.h
> @@ -50,7 +50,7 @@ struct md_rdev {
>   
>   	sector_t sectors;		/* Device size (in 512bytes sectors) */
>   	struct mddev *mddev;		/* RAID array if running */
> -	int last_events;		/* IO event timestamp */
> +	s64 last_events;		/* IO event timestamp */
>   
>   	/*
>   	 * If meta_bdev is non-NULL, it means that a separate device is
> @@ -138,6 +138,8 @@ struct md_rdev {
>   		unsigned int size;	/* Size in sectors of the PPL space */
>   		sector_t sector;	/* First sector of the PPL space */
>   	} ppl;
> +
> +	atomic64_t sync_io;		/* counter of sync IO (unit sectors) */
>   };
>   enum flag_bits {
>   	Faulty,			/* device is known to have a fault */
> @@ -549,14 +551,17 @@ static inline int mddev_trylock(struct mddev *mddev)
>   }
>   extern void mddev_unlock(struct mddev *mddev);
>   
> -static inline void md_sync_acct(struct block_device *bdev, unsigned long nr_sectors)
> +static inline void md_sync_acct(struct md_rdev *rdev, unsigned long nr_sectors)
>   {
> -	atomic_add(nr_sectors, &bdev->bd_disk->sync_io);
> +	atomic64_add(nr_sectors, &rdev->sync_io);
>   }
>   
>   static inline void md_sync_acct_bio(struct bio *bio, unsigned long nr_sectors)
>   {
> -	md_sync_acct(bio->bi_bdev, nr_sectors);
> +	struct md_rdev *rdev = (void *)bio->bi_bdev;

That look weird. bio->bi_bdev should be a 'struct gendisk', not a MD 
internal data structure.

> +
> +	bio_set_dev(bio, rdev->bdev);
> +	md_sync_acct(rdev, nr_sectors);
>   }
>   
>   struct md_personality
> diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
> index 7dc8026cf6ee..74c42dabe57c 100644
> --- a/drivers/md/raid1.c
> +++ b/drivers/md/raid1.c
> @@ -2232,7 +2232,7 @@ static void sync_request_write(struct mddev *mddev, struct r1bio *r1_bio)
>   
>   		wbio->bi_end_io = end_sync_write;
>   		atomic_inc(&r1_bio->remaining);
> -		md_sync_acct(conf->mirrors[i].rdev->bdev, bio_sectors(wbio));
> +		md_sync_acct(conf->mirrors[i].rdev, bio_sectors(wbio));
>   
>   		submit_bio_noacct(wbio);
>   	}
> @@ -2791,7 +2791,7 @@ static sector_t raid1_sync_request(struct mddev *mddev, sector_t sector_nr,
>   		if (rdev && bio->bi_end_io) {
>   			atomic_inc(&rdev->nr_pending);
>   			bio->bi_iter.bi_sector = sector_nr + rdev->data_offset;
> -			bio_set_dev(bio, rdev->bdev);
> +			bio->bi_bdev = (void *)rdev;
>   			if (test_bit(FailFast, &rdev->flags))
>   				bio->bi_opf |= MD_FAILFAST;
>   		}
> diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
> index dde98f65bd04..fc1e6c0996de 100644
> --- a/drivers/md/raid10.c
> +++ b/drivers/md/raid10.c
> @@ -2407,7 +2407,7 @@ static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio)
>   
>   		atomic_inc(&conf->mirrors[d].rdev->nr_pending);
>   		atomic_inc(&r10_bio->remaining);
> -		md_sync_acct(conf->mirrors[d].rdev->bdev, bio_sectors(tbio));
> +		md_sync_acct(conf->mirrors[d].rdev, bio_sectors(tbio));
>   
>   		if (test_bit(FailFast, &conf->mirrors[d].rdev->flags))
>   			tbio->bi_opf |= MD_FAILFAST;
> @@ -2430,7 +2430,7 @@ static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio)
>   			bio_copy_data(tbio, fbio);
>   		d = r10_bio->devs[i].devnum;
>   		atomic_inc(&r10_bio->remaining);
> -		md_sync_acct(conf->mirrors[d].replacement->bdev,
> +		md_sync_acct(conf->mirrors[d].replacement,
>   			     bio_sectors(tbio));
>   		submit_bio_noacct(tbio);
>   	}
> @@ -2562,12 +2562,12 @@ static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio)
>   		wbio2 = NULL;
>   	if (wbio->bi_end_io) {
>   		atomic_inc(&conf->mirrors[d].rdev->nr_pending);
> -		md_sync_acct(conf->mirrors[d].rdev->bdev, bio_sectors(wbio));
> +		md_sync_acct(conf->mirrors[d].rdev, bio_sectors(wbio));
>   		submit_bio_noacct(wbio);
>   	}
>   	if (wbio2) {
>   		atomic_inc(&conf->mirrors[d].replacement->nr_pending);
> -		md_sync_acct(conf->mirrors[d].replacement->bdev,
> +		md_sync_acct(conf->mirrors[d].replacement,
>   			     bio_sectors(wbio2));
>   		submit_bio_noacct(wbio2);
>   	}
> @@ -3486,7 +3486,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
>   				from_addr = r10_bio->devs[j].addr;
>   				bio->bi_iter.bi_sector = from_addr +
>   					rdev->data_offset;
> -				bio_set_dev(bio, rdev->bdev);
> +				bio->bi_bdev = (void *)rdev;
>   				atomic_inc(&rdev->nr_pending);
>   				/* and we write to 'i' (if not in_sync) */
>   
> @@ -3508,7 +3508,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
>   					bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
>   					bio->bi_iter.bi_sector = to_addr
>   						+ mrdev->data_offset;
> -					bio_set_dev(bio, mrdev->bdev);
> +					bio->bi_bdev = (void *)mrdev;
>   					atomic_inc(&r10_bio->remaining);
>   				} else
>   					r10_bio->devs[1].bio->bi_end_io = NULL;
> @@ -3529,7 +3529,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
>   				bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
>   				bio->bi_iter.bi_sector = to_addr +
>   					mreplace->data_offset;
> -				bio_set_dev(bio, mreplace->bdev);
> +				bio->bi_bdev = (void *)mreplace;
>   				atomic_inc(&r10_bio->remaining);
>   				break;
>   			}
> @@ -3684,7 +3684,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
>   			if (test_bit(FailFast, &rdev->flags))
>   				bio->bi_opf |= MD_FAILFAST;
>   			bio->bi_iter.bi_sector = sector + rdev->data_offset;
> -			bio_set_dev(bio, rdev->bdev);
> +			bio->bi_bdev = (void *)rdev;
>   			count++;
>   
>   			rdev = rcu_dereference(conf->mirrors[d].replacement);
> @@ -3706,7 +3706,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
>   			if (test_bit(FailFast, &rdev->flags))
>   				bio->bi_opf |= MD_FAILFAST;
>   			bio->bi_iter.bi_sector = sector + rdev->data_offset;
> -			bio_set_dev(bio, rdev->bdev);
> +			bio->bi_bdev = (void *)rdev;
>   			count++;
>   			rcu_read_unlock();
>   		}
> @@ -4865,7 +4865,7 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr,
>   
>   	read_bio = bio_alloc_bioset(GFP_KERNEL, RESYNC_PAGES, &mddev->bio_set);
>   
> -	bio_set_dev(read_bio, rdev->bdev);
> +	read_bio->bi_bdev = (void *)rdev;
>   	read_bio->bi_iter.bi_sector = (r10_bio->devs[r10_bio->read_slot].addr
>   			       + rdev->data_offset);
>   	read_bio->bi_private = r10_bio;
> @@ -4921,7 +4921,7 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr,
>   		if (!rdev2 || test_bit(Faulty, &rdev2->flags))
>   			continue;
>   
> -		bio_set_dev(b, rdev2->bdev);
> +		b->bi_bdev = (void *)rdev2;
>   		b->bi_iter.bi_sector = r10_bio->devs[s/2].addr +
>   			rdev2->new_data_offset;
>   		b->bi_end_io = end_reshape_write;
> @@ -5016,7 +5016,7 @@ static void reshape_request_write(struct mddev *mddev, struct r10bio *r10_bio)
>   		}
>   		atomic_inc(&rdev->nr_pending);
>   		rcu_read_unlock();
> -		md_sync_acct_bio(b, r10_bio->sectors);
> +		md_sync_acct(rdev, r10_bio->sectors);
>   		atomic_inc(&r10_bio->remaining);
>   		b->bi_next = NULL;
>   		submit_bio_noacct(b);
> diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
> index 9c1a5877cf9f..b932282ff50a 100644
> --- a/drivers/md/raid5.c
> +++ b/drivers/md/raid5.c
> @@ -1167,7 +1167,7 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
>   		if (rdev) {
>   			if (s->syncing || s->expanding || s->expanded
>   			    || s->replacing)
> -				md_sync_acct(rdev->bdev, RAID5_STRIPE_SECTORS(conf));
> +				md_sync_acct(rdev, RAID5_STRIPE_SECTORS(conf));
>   
>   			set_bit(STRIPE_IO_STARTED, &sh->state);
>   
> @@ -1234,7 +1234,7 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
>   		if (rrdev) {
>   			if (s->syncing || s->expanding || s->expanded
>   			    || s->replacing)
> -				md_sync_acct(rrdev->bdev, RAID5_STRIPE_SECTORS(conf));
> +				md_sync_acct(rrdev, RAID5_STRIPE_SECTORS(conf));
>   
>   			set_bit(STRIPE_IO_STARTED, &sh->state);
>   
> diff --git a/include/linux/genhd.h b/include/linux/genhd.h
> index 74c410263113..6b84444111e4 100644
> --- a/include/linux/genhd.h
> +++ b/include/linux/genhd.h
> @@ -150,7 +150,6 @@ struct gendisk {
>   	struct list_head slave_bdevs;
>   #endif
>   	struct timer_rand_state *random;
> -	atomic_t sync_io;		/* RAID */
>   	struct disk_events *ev;
>   #ifdef  CONFIG_BLK_DEV_INTEGRITY
>   	struct kobject integrity_kobj;
> 

Cheers,

Hannes
-- 
Dr. Hannes Reinecke                Kernel Storage Architect
hare@suse.de                              +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), Geschäftsführer: Felix Imendörffer