linux-kernel - Re: [PATCH 01/11] blk-mq: Add blk_mq_init_queue

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]

Date:   Wed, 23 Mar 2022 09:01:33 +0000
From:   John Garry <john.garry@...wei.com>
To:     Bart Van Assche <bvanassche@....org>, <axboe@...nel.dk>,
        <damien.lemoal@...nsource.wdc.com>, <jejb@...ux.ibm.com>,
        <martin.petersen@...cle.com>, <hch@....de>, <ming.lei@...hat.com>,
        <hare@...e.de>
CC:     <chenxiang66@...ilicon.com>, <linux-block@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <linux-ide@...r.kernel.org>,
        <linux-scsi@...r.kernel.org>, <dm-devel@...hat.com>,
        <beanhuo@...ron.com>
Subject: Re: [PATCH 01/11] blk-mq: Add blk_mq_init_queue_ops()

On 23/03/2022 02:57, Bart Van Assche wrote:
> On 3/22/22 03:39, John Garry wrote:
>> Add an API to allocate a request queue which accepts a custom set of
>> blk_mq_ops for that request queue.
>>
>> The reason which we may want custom ops is for queuing requests which we
>> don't want to go through the normal queuing path.
> 

Hi Bart,

 > Custom ops shouldn't be required for this. See e.g. how tmf_queue
 > is used in the UFS driver for an example of a queue implementation
 > with custom operations and that does not require changes of the block
 > layer core.

The UFS code uses a private tagset (in ufs_hba.tmf_tag_set) for only 
management of TMF tags/memories. This tagset does not really have any 
custom operations. All it has is a stub of .queue_rq CB in 
ufshcd_queue_tmf() and that is because this CB is compulsory.

As for the idea of having multiple tagsets per shost with real custom 
operations, this idea was mentioned before, but I think managing 
multiple tagsets could be trouble. For a start, it would mean that we 
need a distinct allocation of reserved and regular tags, and sometimes 
we don't want this - as Hannes mentioned earlier, many HBAs have low 
queue depth and cannot afford to permanently carve out a bunch of 
reserved tags.

Thanks,
John