linux-kernel - Re: [PATCH] block: keep bdi->io_pages in sync with max_sectors

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

Message-ID: <4ce1ccf2-6a7c-818d-bfe3-b9bf29e8be3e@yandex-team.ru>
Date:   Mon, 2 Mar 2020 13:51:50 +0300
From:   Konstantin Khlebnikov <khlebnikov@...dex-team.ru>
To:     Paul Menzel <pmenzel@...gen.mpg.de>, linux-block@...r.kernel.org,
        Jens Axboe <axboe@...nel.dk>, linux-kernel@...r.kernel.org
Cc:     linux-raid@...r.kernel.org, Song Liu <song@...nel.org>,
        Dmitry Monakhov <dmtrmonakhov@...dex-team.ru>
Subject: Re: [PATCH] block: keep bdi->io_pages in sync with max_sectors_kb for
 stacked devices

On 02/03/2020 13.41, Paul Menzel wrote:
> Dear Konstantin,
> 
> 
> Thank you for the patch.
> 
> 
> On 2020-02-28 15:51, Konstantin Khlebnikov wrote:
>> Field bdi->io_pages added in commit 9491ae4aade6 ("mm: don't cap request
>> size based on read-ahead setting") removes unneeded split of read requests.
>>
>> Stacked drivers do not call blk_queue_max_hw_sectors(). Instead they setup
> 
> The verb is spelled with a space *set up*.
> 
>> limits of their devices by blk_set_stacking_limits() + disk_stack_limits().
>> Field bio->io_pages stays zero until user set max_sectors_kb via sysfs.
>>
>> This patch updates io_pages after merging limits in disk_stack_limits().
>>
>> Commit c6d6e9b0f6b4 ("dm: do not allow readahead to limit IO size") fixed
>> the same problem for device-mapper devices, this one fixes MD RAIDs.
> 
> Add a Fixes: tag?
> 
> Fixes: 9491ae4aade6 ("mm: don't cap request size based on read-ahead setting")

Maybe, but this isn't fatal bug. Just incomplete fix.

Same problem exists for non-disk BDIs. I.e. network filesystem must
set io_pages manually if doesn't want split read requests.

> 
>> Signed-off-by: Konstantin Khlebnikov <khlebnikov@...dex-team.ru>
>> ---
>>   block/blk-settings.c |    2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/block/blk-settings.c b/block/blk-settings.c
>> index c8eda2e7b91e..66c45fd79545 100644
>> --- a/block/blk-settings.c
>> +++ b/block/blk-settings.c
>> @@ -664,6 +664,8 @@ void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
>>   		printk(KERN_NOTICE "%s: Warning: Device %s is misaligned\n",
>>   		       top, bottom);
>>   	}
>> +
>> +	t->backing_dev_info->io_pages = t->limits.max_sectors >> (PAGE_SHIFT-9);
>>   }
>>   EXPORT_SYMBOL(disk_stack_limits);
> 
> Reviewed-by: Paul Menzel <pmenzel@...gen.mpg.de>
> 
> 
> Kind regards,
> 
> Paul
>