On 18/07/2025 09:18, Varadarajan Narayanan wrote:
> On Thu, Jul 17, 2025 at 03:01:04PM +0200, Casey Connolly wrote:
>>
>>
>> On 16/07/2025 06:57, Varadarajan Narayanan wrote:
>>> On Tue, Jul 15, 2025 at 02:30:09PM +0200, Casey Connolly wrote:
>>>> Hi Varadarajan,
>>>>
>>>> On 7/15/25 12:15, Varadarajan Narayanan wrote:
>>>>> This series enables MMC based fastboot flashing for IPQ9574 boards.
>>>>> For this the following are added/enabled
>>>>> * USB2 & USB3 phys support
>>>>> * USB Gadget support, Fastboot-MMC related configs
>>>>> * Flashing related documentation
>>>>
>>>> This is awesome! Glad to see you have the superspeed phy working. I'm a bit
>>>> confused though since the of_fixup code in mach-snapdragon configures all
>>>> the usb controllers for high speed only and removes the reference to the
>>>> superspeed phy, is there some missing commit to adjust this?
>>>
>>> CONFIG_EVENT is not defined in ipq defconfig. Hence of_live_build() doesn't
>>> notify the EVT_OF_LIVE_BUILT event and qcom_of_fixup_nodes ->
>>> fixup_usb_nodes
>>> is not called.
>>
>> ohh I see, hmm
>>
>>>
>>> => bdinfo
>>> . . .
>>> fdt_blob = 0x000000006fe3b5e0
>>> . . .
>>>
>>> => fdt addr 0x6fe3b5e0
>>> Working FDT set to 6fe3b5e0
>>>
>>> => fdt print /soc@0/usb@8af8800
>>
>> Just fyi, the fdt command won't show changes to the live tree, it does
>> direct FDT parsing whereas the live-tree is an internal representation
>> in U-Boot (which is what makes it faster to access and modify).
>
> ok
>
> [ . . . ]
>
>>> Initially, I manually overrode the 'phys' and 'phy-names' to just have the
>>> usb2
>>> phy and it didn't seem to work. Only after bringing up the USB3 phy did it
>>> work.
>>> Not sure what the relation is between the two.
>>
>> Hmm, you don't even get high speed USB? probably an issue with clocks then.
>
> If 'qcom,select-utmi-as-pipe-clk;' is added to the controller DT node it
> works. Should I post a new version
>
> (1) without SS phy + qcom,select-utmi-as-pipe-clk;
>
> (or)
>
> (2) with SS phy + qcom,select-utmi-as-pipe-clk; (i.e. same as
> this series, with just utmi as pipe clock added)
>
> (or)
>
> (3) leave the series as is
sorry for the delayed reply! It would be awesome to get superspeed
working in U-Boot, if you have it working with this series then I'd like
to merge it and start enabling superspeed on the rest of the SoCs.
We can make the of_fixup code smarter by looking to see if there is a
driver available for the superspeed phy and skip the fixup if so (not
relevant for IPQ though I guess).
So I'd say keep the series as-is and I'll give it a review.
>
> Kindly advice.
>
> Thanks
> VaradaKind regards,
--
// Casey (she/her)
