Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH 2/2] virtio_scsi: implement request batching
To:     Bart Van Assche <bvanassche@acm.org>, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org
Cc:     jejb@linux.ibm.com, martin.petersen@oracle.com,
        linux-scsi@vger.kernel.org, stefanha@redhat.com
References: <20190530112811.3066-1-pbonzini@redhat.com>
 <20190530112811.3066-3-pbonzini@redhat.com>
 <79490df1-0145-5b40-027a-7e8fb96854d4@acm.org>
From:   Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <ff7aaa01-022b-f55a-4bb2-c293cfd86bdd@redhat.com>
Date:   Fri, 31 May 2019 11:03:16 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.6.1
MIME-Version: 1.0
In-Reply-To: <79490df1-0145-5b40-027a-7e8fb96854d4@acm.org>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 30/05/19 19:28, Bart Van Assche wrote:
> On 5/30/19 4:28 AM, Paolo Bonzini wrote:
>> @@ -531,7 +547,8 @@ static int virtscsi_queuecommand(struct Scsi_Host
>> *shost,
>>           req_size = sizeof(cmd->req.cmd);
>>       }
>>   -    ret = virtscsi_kick_cmd(req_vq, cmd, req_size,
>> sizeof(cmd->resp.cmd));
>> +    kick = (sc->flags & SCMD_LAST) != 0;
>> +    ret = virtscsi_add_cmd(req_vq, cmd, req_size,
>> sizeof(cmd->resp.cmd), kick);
> 
> Have you considered to have the SCSI core call commit_rqs() if bd->last
> is true? I think that would avoid that we need to introduce the
> SCMD_LAST flag and that would also avoid that every SCSI LLD that
> supports a commit_rqs callback has to introduce code to test the
> SCMD_LAST flag.

That is slightly worse for performance, as it unlocks and re-locks the
spinlock.

Paolo