Re: [PATCH 06/15] scsi: core: Convert to scsi_execute_cmd

From: Bart Van Assche <bvanassche@acm.org>
To: Mike Christie <michael.christie@oracle.com>,
	Christoph Hellwig <hch@lst.de>
Cc: mwilck@suse.com, martin.petersen@oracle.com,
	linux-scsi@vger.kernel.org,
	james.bottomley@hansenpartnership.com
Subject: Re: [PATCH 06/15] scsi: core: Convert to scsi_execute_cmd
Date: Tue, 22 Nov 2022 10:46:10 -0800	[thread overview]
Message-ID: <92d8899c-7a73-ebde-87f6-8defab44d551@acm.org> (raw)
In-Reply-To: <33b7d1ca-4421-2f90-071b-0661f3893865@oracle.com>

On 11/22/22 08:13, Mike Christie wrote:
> On 11/22/22 12:38 AM, Christoph Hellwig wrote:
>> On Mon, Nov 21, 2022 at 09:39:25PM -0600, Mike Christie wrote:
>>> +	result = scsi_execute_cmd(sdev, cmd, REQ_OP_DRV_IN, buffer,
>>> +				  request_len, 30 * HZ, 3,
>>> +				  ((struct scsi_exec_args) { .sshdr = &sshdr }));
>>
>> Maybe it's me, but I hate the syntax to declare structs inside argument
> 
> I'll change it to setup the scsi_exec_args struct like normal. I've got this
> comment several times now.
> 
> With the current design we know all the args when we declare the variables so
> I can do it then like normal.

Will struct scsi_exec_args be retained? I'm asking this because I'd like 
to add another argument to the (already too large) argument list of 
scsi_execute().

Thanks,

Bart.