linux-kernel - Re: [PATCH] loop: Fix missing discard support when using LOOP

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <8445d1b6-0118-37d4-c927-3b7e6ff480ea@klausen.dk>
Date:   Fri, 18 Jun 2021 01:26:19 +0200
From:   Kristian Klausen <kristian@...usen.dk>
To:     Ming Lei <ming.lei@...hat.com>
Cc:     linux-block@...r.kernel.org, stable@...r.kernel.org,
        Jens Axboe <axboe@...nel.dk>,
        Martijn Coenen <maco@...roid.com>,
        open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] loop: Fix missing discard support when using
 LOOP_CONFIGURE

On 18.06.2021 01.03, Ming Lei wrote:
> On Fri, Jun 18, 2021 at 12:11:57AM +0200, Kristian Klausen wrote:
>
> Commit log?

I will try to come up with a commit message. AFAIU the missing call to 
loop_config_discard() causes the block device to report "no discard 
support" (best-case) or stale/incorrect discard support (ex: if the loop 
device was configured with LOOP_SET_STATUS(64) and later LOOP_CONFIGURE).

>
>> Cc: <stable@...r.kernel.org> # 5.8.x-
>> Fixes: 3448914e8cc5 ("loop: Add LOOP_CONFIGURE ioctl")
>> Signed-off-by: Kristian Klausen <kristian@...usen.dk>
>> ---
>> Tested like so (without the patch):
>> losetup 2.37<= uses LOOP_CONFIGURE instead of LOOP_SET_STATUS64[1]
>>
>> # fallocate -l100M disk.img
>> # rmmod loop
>> # losetup --version
>> losetup from util-linux 2.36.2
>> # losetup --find --show disk.img
>> /dev/loop0
>> # grep '' /sys/devices/virtual/block/loop0/queue/*discard*
>> /sys/devices/virtual/block/loop0/queue/discard_granularity:4096
>> /sys/devices/virtual/block/loop0/queue/discard_max_bytes:4294966784
>> /sys/devices/virtual/block/loop0/queue/discard_max_hw_bytes:4294966784
>> /sys/devices/virtual/block/loop0/queue/discard_zeroes_data:0
>> /sys/devices/virtual/block/loop0/queue/max_discard_segments:1
>> # losetup -d /dev/loop0
>> # [update util-linux]
>> # losetup --version
>> losetup from util-linux 2.37
>> # rmmod loop
>> # losetup --find --show disk.img
>> /dev/loop0
>> # grep '' /sys/devices/virtual/block/loop0/queue/*discard*
>> /sys/devices/virtual/block/loop0/queue/discard_granularity:0
>> /sys/devices/virtual/block/loop0/queue/discard_max_bytes:0
>> /sys/devices/virtual/block/loop0/queue/discard_max_hw_bytes:0
>> /sys/devices/virtual/block/loop0/queue/discard_zeroes_data:0
>> /sys/devices/virtual/block/loop0/queue/max_discard_segments:1
>>
>>
>> With the patch applied:
>>
>> # losetup --version
>> losetup from util-linux 2.37
>> # rmmod loop
>> # losetup --find --show disk.img
>> /dev/loop0
>> # grep '' /sys/devices/virtual/block/loop0/queue/*discard*
>> /sys/devices/virtual/block/loop0/queue/discard_granularity:4096
>> /sys/devices/virtual/block/loop0/queue/discard_max_bytes:4294966784
>> /sys/devices/virtual/block/loop0/queue/discard_max_hw_bytes:4294966784
>> /sys/devices/virtual/block/loop0/queue/discard_zeroes_data:0
>> /sys/devices/virtual/block/loop0/queue/max_discard_segments:1
>>
>> [1] https://github.com/karelzak/util-linux/pull/1152
>>
>>   drivers/block/loop.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
>> index 76e12f3482a9..ec957f6d8a49 100644
>> --- a/drivers/block/loop.c
>> +++ b/drivers/block/loop.c
>> @@ -1168,6 +1168,8 @@ static int loop_configure(struct loop_device *lo, fmode_t mode,
>>   	if (partscan)
>>   		lo->lo_disk->flags &= ~GENHD_FL_NO_PART_SCAN;
>>   
>> +	loop_config_discard(lo);
>> +
> It could be better to move loop_config_discard() around
> loop_update_rotational/loop_update_dio(), then we setup everything
> before updating loop as Lo_bound.

I will move it before loop_update_rotational().

>
> Otherwise, this patch looks fine.
>
>
> Thanks,
> Ming
>