Re: [PATCH v6 3/9] mmc: core: Add open-ended Ext memory addressing

[Adrian Hunter <adrian.hunter@intel.com>]

On 5/09/24 09:12, Avri Altman wrote:
> Thanks for having a look.
> 
>>>
>>> +     if (mrq->cmd && mrq->cmd->has_ext_addr)
>>> +             mmc_send_ext_addr(host, mrq->cmd->ext_addr);
>>
>> We should check that this was actually successful, right?
> Actually no, as errors in CMD22 are being carried in the subsequent command.
> 
>>
>>> +
>>>       init_completion(&mrq->cmd_completion);
>>>
>>>       mmc_retune_hold(host);
>>> diff --git a/include/linux/mmc/core.h b/include/linux/mmc/core.h index
>>> f0ac2e469b32..41c21c216584 100644
>>> --- a/include/linux/mmc/core.h
>>> +++ b/include/linux/mmc/core.h
>>> @@ -76,6 +76,11 @@ struct mmc_command {
>>>   */
>>>  #define mmc_cmd_type(cmd)    ((cmd)->flags & MMC_CMD_MASK)
>>>
>>> +     /* for SDUC */
>>> +     u8 has_ext_addr;
>>> +     u8 ext_addr;
>>> +     u16 reserved;
>>
>> Is there a reason for has_ext_addr being u8?
> Theoretically a single bit suffices, and since ext_addr uses only 6 bits, I had that bit to spare in ext_addr,

It could be a bool instead of u8 though.

> but I see no reason to be cheap here - see the reserved u16.
> 
>> What's the reserved for?
> Not to break the packed 4bytes alignment of mmc_command.

Although it isn't "__packed" so compiler should take care of
alignment.

> 
> Thanks,
> Avri
>>
>>> +
>>>       unsigned int            retries;        /* max number of retries */
>>>       int                     error;          /* command error */
>>>
>