Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <0e60fab5-8a76-9b7e-08cf-fb791e01ae08@huawei.com>
Date:   Fri, 28 Oct 2022 09:01:28 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101
 Thunderbird/91.6.1
Subject: Re: [PATCH RFC v3 2/7] ata: libata-scsi: Add
 ata_internal_queuecommand()
To:     Damien Le Moal <damien.lemoal@opensource.wdc.com>,
        <jejb@linux.ibm.com>, <martin.petersen@oracle.com>, <hare@suse.de>,
        <bvanassche@acm.org>, <hch@lst.de>, <ming.lei@redhat.com>,
        <niklas.cassel@wdc.com>
CC:     <axboe@kernel.dk>, <jinpu.wang@cloud.ionos.com>,
        <linux-block@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <linux-ide@vger.kernel.org>, <linux-scsi@vger.kernel.org>,
        <linuxarm@huawei.com>, <john.garry2@mail.dcu.ie>
References: <1666693976-181094-1-git-send-email-john.garry@huawei.com>
 <1666693976-181094-3-git-send-email-john.garry@huawei.com>
 <08fdb698-0df3-7bc8-e6af-7d13cc96acfa@opensource.wdc.com>
 <83d9dc82-ea37-4a3c-7e67-1c097f777767@huawei.com>
 <9a2f30cc-d0e9-b454-d7cd-1b0bd3cf0bb9@opensource.wdc.com>
From:   John Garry <john.garry@huawei.com>
In-Reply-To: <9a2f30cc-d0e9-b454-d7cd-1b0bd3cf0bb9@opensource.wdc.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 27/10/2022 23:25, Damien Le Moal wrote:
>> So we have this overall flow:
>>
>> ata_exec_internal_sg():
>>    -> alloc request
>>    -> blk_execute_rq_nowait()
>>        ... -> scsi_queue_rq()
>> 		-> sht->reserved_queuecommd()
>> 			-> ata_internal_queuecommand()
>>
>> And then we have ata_internal_queuecommand() -> ata_sas_queuecmd() ->
>> ata_scsi_queue_internal() -> ata_qc_issue().
>>
>> Hope it makes sense.
> OK. Got it.
> However, ata_exec_internal_sg() being used only from EH context with the
> queue quiesced, will blk_execute_rq_nowait() work ? Is there an exception
> for internal reserved tags ?
> 

Well, yeah. So if some error happens and EH kicks in, then full queue 
depth of requests may be allocated. I have seen this for NCQ error. So 
this is why I make in very first patch change allow us to allocate 
reserved request from sdev request queue even when budget is fully 
allocated.

Please also note that for AHCI, I make reserved depth =1, while for SAS 
controllers it is greater. This means that in theory we could alloc > 1x 
reserved command for SATA disk, but I don't think it matters.

Thanks,
John