linux-kernel - Re: [PATCH v11 3/3] scsi: set max_bio

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:   Thu,  3 Jun 2021 18:28:30 +0900
From:   Changheun Lee <nanich.lee@...sung.com>
To:     bvanassche@....org
Cc:     Johannes.Thumshirn@....com, alex_y_xu@...oo.ca,
        asml.silence@...il.com, axboe@...nel.dk, bgoncalv@...hat.com,
        damien.lemoal@....com, gregkh@...uxfoundation.org,
        hch@...radead.org, jaegeuk@...nel.org, jisoo2146.oh@...sung.com,
        junho89.kim@...sung.com, linux-block@...r.kernel.org,
        linux-kernel@...r.kernel.org, ming.lei@...hat.com,
        mj0123.lee@...sung.com, nanich.lee@...sung.com, osandov@...com,
        patchwork-bot@...nel.org, seunghwan.hyun@...sung.com,
        sookwan7.kim@...sung.com, tj@...nel.org, tom.leiming@...il.com,
        woosung2.lee@...sung.com, yi.zhang@...hat.com,
        yt0928.kim@...sung.com
Subject: Re: [PATCH v11 3/3] scsi: set max_bio_bytes with queue max sectors

> On 6/2/21 5:10 AM, Changheun Lee wrote:
> > Set max_bio_bytes same with queue max sectors. It will lead to fast bio
> > submit when bio size is over than queue max sectors. And it might be helpful
> > to align submit bio size in some case.
> > 
> > Signed-off-by: Changheun Lee <nanich.lee@...sung.com>
> > ---
> >  drivers/scsi/scsi_lib.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> > index 532304d42f00..f6269268b0e0 100644
> > --- a/drivers/scsi/scsi_lib.c
> > +++ b/drivers/scsi/scsi_lib.c
> > @@ -1837,6 +1837,8 @@ void __scsi_init_queue(struct Scsi_Host *shost, struct request_queue *q)
> >  	blk_queue_virt_boundary(q, shost->virt_boundary_mask);
> >  	dma_set_max_seg_size(dev, queue_max_segment_size(q));
> >  
> > +	blk_queue_max_bio_bytes(q, queue_max_sectors(q));
> > +
> >  	/*
> >  	 * Set a reasonable default alignment:  The larger of 32-byte (dword),
> >  	 * which is a common minimum for HBAs, and the minimum DMA alignment,
> 
> Has this patch been tested with dm-crypt on top of a SCSI device? I'm
> concerned that this patch will trigger data corruption with dm-crypt on
> top because the above change will make the following dm-crypt code fail
> for a sufficiently large bio:
> 
> bio_add_page(clone, page, len, 0);
> 
> When testing dm-crypt on top of this patch series, please change the
> above dm-crypt code into the following before running any tests:
> 
> WARN_ON(bio_add_page(clone, page, len, 0) < len);
> 
> Thanks,
> 
> Bart.

I think it will be OK if nr_iovecs is not over BIO_MAX_VECS in
crypt_alloc_buffer(). Because page is added to bio in PAGE_SIZE as maximum,
and the minimum of max_bio_bytes is "BIO_MAX_VECS * PAGE_SIZE".

for (i = 0; i < nr_iovecs; i++) {
	page = mempool_alloc(&cc->page_pool, gfp_mask);
	if (!page) {
		crypt_free_buffer_pages(cc, clone);
		bio_put(clone);
		gfp_mask |= __GFP_DIRECT_RECLAIM;
		goto retry;
	}

	len = (remaining_size > PAGE_SIZE) ? PAGE_SIZE : remaining_size;

	bio_add_page(clone, page, len, 0);

	remaining_size -= len;
}

Anyway as your advise, I'm checking it in my test environment. There are no
problem by now yet. I'm testing with 1MB max_bio_bytes of SCSI, and UINT_MAX
max_bio_bytes of dm-crypt.

Thank you,
Changheun Lee.