Re: [PATCH RFC 03/15] mmc: sunxi: Block signal voltage switching (CMD11)

[Chen-Yu Tsai]

On Fri, Jan 29, 2016 at 6:42 PM, Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> On 21 January 2016 at 06:26, Chen-Yu Tsai <wens@xxxxxxxx> wrote:
>> Allwinner's mmc controller supports signal voltage switching. This is
>> supported in code in Allwinner's kernel. However, publicly available
>> boards all tie it to a fixed 3.0/3.3V regulator, with options to tie
>> it to 1.8V for eMMC on some.
>>
>> Since Allwinner's kernel is an ancient 3.4, it is hard to say whether
>> adapting it's code to a modern mainline kernel would work. Block signal
>> voltage switching until someone has proper hardware to implement and
>> test this.
>>
>> This only affects SD UHS-1 modes, as eMMC switches the voltage directly
>> without any signaling.
>>
>> Signed-off-by: Chen-Yu Tsai <wens@xxxxxxxx>
>> ---
>>  drivers/mmc/host/sunxi-mmc.c | 14 ++++++++++++++
>>  1 file changed, 14 insertions(+)
>>
>> diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c
>> index 790f01662b4e..0495ae7da6d6 100644
>> --- a/drivers/mmc/host/sunxi-mmc.c
>> +++ b/drivers/mmc/host/sunxi-mmc.c
>> @@ -816,6 +816,20 @@ static void sunxi_mmc_request(struct mmc_host *mmc, struct mmc_request *mrq)
>>                 }
>>         }
>>
>> +       /*
>> +        * TODO Support signal voltage switching
>> +        *
>> +        * Compared to Allwinner's kernel, recent updates in the mmc core
>> +        * mean this should be as easy as setting the flags in cmd_val and
>> +        * imask, and waiting for it to finish. However no boards support
>> +        * this so this cannot be tested. Block it for now.
>> +        */
>> +       if (cmd->opcode == SD_SWITCH_VOLTAGE) {
>> +               mrq->cmd->error = -EPERM;
>> +               mmc_request_done(mmc, mrq);
>> +               return;
>> +       }
>
> Unless some of the MMC_CAP_UHS* mode is set, this command shouldn't be sent.
>
> So, if you *really* want to protect from this, I think it's better to
> clear these caps in the ->probe() function, after mmc_of_parse() has
> been called.

OK. If we need to block these, then we need to block all the other
unsupported caps, which seems excessive and unnecessary. I'll drop
this patch.

Thanks
ChenYu

>
>> +
>>         if (cmd->opcode == MMC_GO_IDLE_STATE) {
>>                 cmd_val |= SDXC_SEND_INIT_SEQUENCE;
>>                 imask |= SDXC_COMMAND_DONE;
>> --
>> 2.7.0.rc3
>>
>
> Kind regards
> Uffe