[PATCH] mmc: sd: Handle UHS-I voltage signaling without power cycle

[Judith Mendez]

Hi,

On 5/14/26 1:50 PM, Kathpalia, Tanmay wrote:
> 
> On 5/14/2026 5:09 AM, Judith Mendez wrote:
>> Hi Tanmay, all,
>>
>> On 10/21/25 3:45 PM, Tanmay Kathpalia wrote:
>>> Some boards have SD card connectors where the power rail cannot be 
>>> switched
>>> off by the driver. However there are various circumstances when a card
>>> might be re-initialized, such as after system resume, warm re-boot, or
>>> error handling. However, a UHS card will continue to use 1.8V signaling
>>> unless it is power cycled.
>>>
>>> If the card has not been power cycled, it may still be using 1.8V
>>> signaling. According to the SD spec., the Bus Speed Mode (function 
>>> group 1)
>>> bits 2 to 4 are zero if the card is initialized at 3.3V signal level. 
>>> Thus
>>> they can be used to determine if the card has already switched to 1.8V
>>> signaling. Detect that situation and try to initialize a UHS-I (1.8V)
>>> transfer mode.
>>
>> This implementation broke am65 IDK board SD card boot, I have still
>> to check why this breaks only one board, potentially there might
>> be a quirk only for this board... But for now sending the question
>> in case anyone might have an idea what is going on & save me some
>> time.
>>
>>
> 
> Thanks for reporting this. I had a deeper look at the code and found
> an issue that is likely the root cause of the failure on am65 IDK.
> 
> During a normal cold boot with a UHS-capable host and UHS card, the
> voltage switch to 1.8V is already performed inside sd_send_op_cond()
> as part of the ACMD41 handshake, and this is recorded in mmc->ocr via
> the S18R bit. By the time mmc_startup() calls sd_get_capabilities(),
> the card is already operating at 1.8V and reports UHS bus speed modes,
> causing mmc_sd_card_using_v18() to return true. The warm-reboot
> recovery path then fires unconditionally and switches the host voltage
> a second time, which on boards like am65 IDK can leave the controller
> in a bad state.
> 
> The fix is to guard the recovery path so it only activates when the
> voltage switch was not already performed during the current session:
> 
> -        if (mmc_sd_card_using_v18(mmc)) {
> +        if (!(mmc->ocr & OCR_S18R) && mmc_sd_card_using_v18(mmc)) {
> 
> Could you please try this change and let me know if it resolves the
> failure on am65 IDK?
> 
> I will send the patch for review on the mailing list.
> 

Seems like that did not work for me [0] :(

[0] https://gist.github.com/jmenti/bb2dfc93630219a4d137edf2e20ce66c

I might not have a chance to debug/respond here soon since ill be out
until may 25, so until then, happy hacking :D

~ Judith