Re: [PATCH v2] block: Improve limiting the bio size

From: Changheun Lee <nanich.lee@samsung.com>
To: bvanassche@acm.org
Cc: axboe@kernel.dk, bgoncalv@redhat.com, hch@lst.de,
	jaegeuk@kernel.org, linux-block@vger.kernel.org,
	ming.lei@redhat.com, nanich.lee@samsung.com, yi.zhang@redhat.com
Subject: Re: [PATCH v2] block: Improve limiting the bio size
Date: Tue, 27 Apr 2021 09:28:57 +0900	[thread overview]
Message-ID: <20210427002857.8038-1-nanich.lee@samsung.com> (raw)
In-Reply-To: <34286266-1c03-35bc-94e8-08bd0ac3400a@acm.org>

> On 4/26/21 1:34 AM, Changheun Lee wrote:
> > Should we check queue point in bio_max_size()?
> > __device_add_disk() can be called with "register_queue=false" like as
> > device_add_disk_no_queue_reg(). How about below?
> > 
> > unsigned int bio_max_size(struct bio *bio)
> > {
> > 	struct request_queue *q;
> > 
> > 	q = (bio->bi_bdev) ? bio->bi_bdev->bd_disk->queue : NULL;
> > 	return q ? q->limits.bio_max_bytes : UINT_MAX;
> > }
> 
> How could bio_max_size() get called from inside __device_add_disk() if
> no request queue is registered? Did I perhaps miss something?
> 
> Thanks,
> 
> Bart.

__device_add_disk() do not call bio_max_size(). I just imagined bio
operation on disk without request queue. Disk can be added without queue via
device_add_disk_no_queue_reg(). It might be my miss-understood about it.
I didn't check bio operation is possible on disk without request queue yet. 

Thanks,

Changheun Lee.