Re: [PATCH 2/2] configs: rock5b-rk3588: add rtl8169 driver

[Eugen Hristev <eugen.hristev@collabora.com>]

On 5/15/23 16:16, Tom Rini wrote:
> On Mon, May 15, 2023 at 03:57:14PM +0300, Eugen Hristev wrote:
>> On 5/15/23 15:52, Tom Rini wrote:
>>> On Mon, May 15, 2023 at 03:36:24PM +0300, Eugen Hristev wrote:
>>>> On 5/6/23 01:03, Tom Rini wrote:
>>>>> On Tue, Apr 25, 2023 at 04:06:59PM +0300, Eugen Hristev wrote:
>>>>>> Add the rtl8169 driver, which supports the rtl8125b device, which is
>>>>>> connected on the pciE bus on this board.
>>>>>> Enable also CONFIG_SYS_HAS_NONCACHED_MEMORY to have the descriptors stored.
>>>>>>
>>>>>> Signed-off-by: Eugen Hristev <eugen.hristev@collabora.com>
>>>>>> ---
>>>>>>     configs/rock5b-rk3588_defconfig | 2 ++
>>>>>>     1 file changed, 2 insertions(+)
>>>>>>
>>>>>> diff --git a/configs/rock5b-rk3588_defconfig b/configs/rock5b-rk3588_defconfig
>>>>>> index a14fcd2ee924..bfa48227aee2 100644
>>>>>> --- a/configs/rock5b-rk3588_defconfig
>>>>>> +++ b/configs/rock5b-rk3588_defconfig
>>>>>> @@ -1,5 +1,6 @@
>>>>>>     CONFIG_ARM=y
>>>>>>     CONFIG_SKIP_LOWLEVEL_INIT=y
>>>>>> +CONFIG_SYS_HAS_NONCACHED_MEMORY=y
>>>>>>     CONFIG_COUNTER_FREQUENCY=24000000
>>>>>>     CONFIG_ARCH_ROCKCHIP=y
>>>>>>     CONFIG_TEXT_BASE=0x00a00000
>>>>>> @@ -71,6 +72,7 @@ CONFIG_MMC_SDHCI_SDMA=y
>>>>>>     CONFIG_MMC_SDHCI_ROCKCHIP=y
>>>>>>     CONFIG_SPI_FLASH_MACRONIX=y
>>>>>>     CONFIG_ETH_DESIGNWARE=y
>>>>>> +CONFIG_RTL8169=y
>>>>>>     CONFIG_GMAC_ROCKCHIP=y
>>>>>>     CONFIG_PCI=y
>>>>>>     CONFIG_PCIE_DW_ROCKCHIP=y
>>>>>
>>>>> Does this depend on some series I missed where PCI is enabled on this
>>>>> config?
>>>>>
>>>>
>>>> Hi Tom,
>>>>
>>>> the network device is connected on PCI express, that is right. I have a
>>>> separate series for that.
>>>> However, the network driver itself and the enabling of the network driver is
>>>> somewhat independent, so I sent it as a separate series (this one )
>>>> I think the config patch can go through rockchip custodian tree.
>>>
>>> With 1/2 now applied this results in failure to build due to DM_PCI not
>>> being enabled.
>>>
>>
>> This is most strange. 1/2 are just changes in the driver. How does that
>> cause a build failure ?
>> The driver is not even selected for this config in 2/2, before the actual
>> patch.
>> Do you have the log of the failure so I can have a look ?
> 
> I don't have the log handy right now, please re-test the config changes
> on top of master and repost.
> 


Hi Tom,

I figured out the issue.
My patch is indeed based on another patch that selects CONFIG_PCI.

However, the rtl8169 driver depends on PCI, even if it does not have 
'depends on PCI' in the Kconfig.
So selecting it without PCI, well, the error.

So I sent another patch :

https://patchwork.ozlabs.org/project/uboot/patch/20230517104124.111075-1-eugen.hristev@collabora.com/

... that should have the rtl8169 depend on PCI now.

That being said, this current patch can be applied on kever's rockchip 
master branch on top of PCI series, but if it's to be applied on master, 
it should work on top of the 'rtl8169 depends on PCI' patch now.

I hope it makes sense.
Thanks