From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-scsi-owner@vger.kernel.org>
Date: Sat, 30 Jun 2018 09:12:19 +0800
From: Ming Lei <ming.lei@redhat.com>
To: Guenter Roeck <linux@roeck-us.net>
Cc: linux-scsi@vger.kernel.org, Jens Axboe <axboe@fb.com>,
        linux-block@vger.kernel.org, Omar Sandoval <osandov@fb.com>,
        "Martin K. Petersen" <martin.petersen@oracle.com>,
        James Bottomley <james.bottomley@hansenpartnership.com>,
        Christoph Hellwig <hch@lst.de>,
        Don Brace <don.brace@microsemi.com>,
        Kashyap Desai <kashyap.desai@broadcom.com>,
        Mike Snitzer <snitzer@redhat.com>,
        Hannes Reinecke <hare@suse.de>,
        Laurence Oberman <loberman@redhat.com>,
        Bart Van Assche <bart.vanassche@wdc.com>
Subject: Re: [V2, 2/2] scsi: avoid to hold host-wide counter of host_busy for
 scsi_mq
Message-ID: <20180630011213.GA12152@ming.t460p>
References: <20180624140327.28146-3-ming.lei@redhat.com>
 <20180629162054.GA19036@roeck-us.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <20180629162054.GA19036@roeck-us.net>
Sender: linux-scsi-owner@vger.kernel.org
List-ID: <linux-block@vger.kernel.org>

On Fri, Jun 29, 2018 at 09:20:54AM -0700, Guenter Roeck wrote:
> Hi,
> 
> On Sun, Jun 24, 2018 at 10:03:27PM +0800, Ming Lei wrote:
> > It isn't necessary to check the host depth in scsi_queue_rq() any more
> > since it has been respected by blk-mq before calling scsi_queue_rq() via
> > getting driver tag.
> > 
> > Lots of LUNs may attach to same host, and per-host IOPS may reach millions
> > level, so we should avoid to this expensive atomic operations on the
> > hostwide counter in IO path.
> > 
> > This patch implemens scsi_host_busy() via blk_mq_tagset_busy_iter() for
> > reading the count of busy IOs for scsi_mq.
> > 
> > It is observed that IOPS is increased by 15% in IO test on scsi_debug
> > (32 LUNs, 32 submit queues, 1024 can_queue, libaio/dio) in one
> > dual-socket system.
> > 
> 
> This patch breaks two of my qemu test builds in -next: parisc:defconfig
> and arm:versatilepb-scsi:versatile_defconfig (which is versatilepb booting
> from scsi disk). The symptom is the same for both: Boot stalls after scsi
> bus initialization.
> 
> arm:
> 
> sym53c8xx 0000:00:0c.0: enabling device (0100 -> 0103)
> sym0: <895a> rev 0x0 at pci 0000:00:0c.0 irq 66
> sym0: No NVRAM, ID 7, Fast-40, LVD, parity checking
> sym0: SCSI BUS has been reset.
> scsi host0: sym-2.2.3
> random: fast init done
> [stalls]
> 
> parisc:
> 
> sym53c8xx 0000:00:00.0: enabling SERR and PARITY (0107 -> 0147)
> sym0: <895a> rev 0x0 at pci 0000:00:00.0 irq 17
> sym0: PA-RISC Firmware, ID 7, Fast-40, LVD, parity checking
> sym0: SCSI BUS has been reset.
> scsi host0: sym-2.2.3
> random: fast init done
> [stalls]
> 
> Reverting the patch fixes the problem. Bisect log is attached.

Hi Guenter,

Thanks for your test & report!

This looks a bit weird, I need to take a close look given this
patch supposes to be nop for non-blk-mq, which is exactly your
case.

Thanks,
Ming