Re: [PATCH v3] mmc: block: use single block write in retry

[Jens Axboe <axboe@kernel.dk>]

On 3/25/26 8:21 AM, Bin Liu wrote:
> On Wed, Mar 25, 2026 at 08:12:36AM -0600, Jens Axboe wrote:
>> On 3/25/26 7:49 AM, Bin Liu wrote:
>>> diff --git a/drivers/mmc/core/queue.h b/drivers/mmc/core/queue.h
>>> index 1498840a4ea0..c254e6580afd 100644
>>> --- a/drivers/mmc/core/queue.h
>>> +++ b/drivers/mmc/core/queue.h
>>> @@ -61,6 +61,8 @@ enum mmc_drv_op {
>>>  	MMC_DRV_OP_GET_EXT_CSD,
>>>  };
>>>  
>>> +#define	MQRQ_XFER_SINGLE_BLOCK		BIT(0)
>>> +
>>>  struct mmc_queue_req {
>>>  	struct mmc_blk_request	brq;
>>>  	struct scatterlist	*sg;
>>> @@ -69,6 +71,7 @@ struct mmc_queue_req {
>>>  	void			*drv_op_data;
>>>  	unsigned int		ioc_count;
>>>  	int			retries;
>>> +	u32			flags;
>>>  };
>>>  
>>>  struct mmc_queue {
>>
>> I'm assuming retries is a pretty small number (didn't check), so if you
>> care about layout/space in mmc_queue_req, you could make this:
>>
>> 	u16 retries;
>> 	u16 flags;
> 
> I quickly went throught block.c, it appears retries would be only a
> single digit number. I will update the patch to use u16.

Note that this is just my suggestion, it's down to Ulf. But what I would
do is a prep patch that makes retries u16 and ensures that's sane, then
do your flags patch on top of that. I think that's better than making
that unrelated size change for retries in your patch that isn't related
to that at all. It just sets you up to not grow the struct after your
change.

-- 
Jens Axboe