[U-Boot] [PATCH 3/3] arm64: zynqmp: add support for Avnet UltraZed-EV Starter Kit

Fri May 24 08:01:27 UTC 2019

On 23. 05. 19 15:54, Luca Ceresoli wrote:
> Hi,
> 
> On 22/05/19 12:24, Michal Simek wrote:
>> On 10. 05. 19 10:15, Luca Ceresoli wrote:
>>> Avnet UltraZed-EV Starter Kit is composed by the UltraZed-EV SoM and the
>>> only publicly-available compatible carrier card. The SoM is based on the EV
>>> version of the Xilinx ZynqMP SoC+FPGA.
>>>
>>> The psu_init_gpl.c file has been generated from the board definition files
>>> at [0] using Vivado 2018.3. It has then minimized by
>>> tools/zynqmp_psu_init_minimize.sh and slightly fixed manually.
>>>
>>> [0] https://github.com/Avnet/bdf/tree/3686c9ff7d2f0467fb4fcf39f861b8d6ff183b12/ultrazed_7ev_cc/1.1
>>>
>>> Signed-off-by: Luca Ceresoli <luca at lucaceresoli.net>
>>>
>>> ---
>>>
>>> Whole patchset tested on:
>>>  - current u-boot/master, as is
>>>  - current u-boot-microblaze/master
>>>  - current u-boot-microblaze/master with the addtion of
>>>    commit f89d6133eef2 ("configs: move CONFIG_SPL_TEXT_BASE to Kconfig")
> 
> [...]
> 
>>> diff --git a/arch/arm/dts/avnet-ultrazedev.dts b/arch/arm/dts/avnet-ultrazedev.dts
>>> new file mode 100644
>>> index 000000000000..34d506a28e88
>>> --- /dev/null
>>> +++ b/arch/arm/dts/avnet-ultrazedev.dts
> 
> [...]
> 
>>> +/* I2C peripherals on carrier card */
>>
>> This is the biggest issue I have with this patch. It is SOM which should
>> be described in separate file and then we have carried board.
>> That's why please separate DTS file.
> 
> Absolutely! Will do.
> 
>>> +&i2c_cc {
>>> +	/* Microchip 24AA025E48T-I/OT: 2K I2C Serial EEPROMs with EUI-48 */
>>> +	eeprom: eeprom at 51 {
>>> +		compatible = "at,24c02", "i2c-eeprom";
>>> +		reg = <0x51>;
>>> +	};
>>> +
>>> +	/* Versa Clock 5P49V5935B */
>>> +	vc5: clock-generator at 6a {
>>> +		compatible = "idt,5p49v5935";
>>> +		reg = <0x6a>;
>>> +		#clock-cells = <1>;
>>> +	};
>>> +};
>>> +
>>> +/* Ethernet:
>>> + * - Marvell 88E1512-A0-NNP2I000 PHY on SOM
>>> + * - RJ-45 on Carrier Card
>>> + */
>>
>> This is interesting.
>>
>>> +&gem3 {
>>> +	status = "okay";
>>> +	phy-mode = "rgmii-id";
>>> +};
> 
> Well, the phy-mode is related to the MAC-PHY link, while the existence
> of a connector is carrier-dependent. So I think the right thing to do
> here is:
> 
> In the SOM dtsi:
> 
> &gem3 {
> 	phy-mode = "rgmii-id";
> };
> 
> and in the carrier dts:
> 
> &gem3 {
> 	status = "okay";
> };
> 
> Does it look OK?

I have looked at schematics and SOM dt should have phy-mode , phy
address + phy flags if any.
And yes - you can enable status if there is connector on carrier board.

> 
>>> diff --git a/board/xilinx/zynqmp/avnet-ultrazedev/psu_init_gpl.c b/board/xilinx/zynqmp/avnet-ultrazedev/psu_init_gpl.c
>>> new file mode 100644
>>> index 000000000000..44e683dcee3f
>>> --- /dev/null
>>> +++ b/board/xilinx/zynqmp/avnet-ultrazedev/psu_init_gpl.c
> 
> [...]
> 
>>> +static int init_serdes(void)
>>> +{
>>
>> I can't see any description for serdes in DT. That's why you can likely
>> remove the whole this function if you are not using GTs.
> 
> I see. Actually the board has USB, but I didn't use it, so it's not in
> DT. I'll remove all the serdes code in v2.
> 
>>> diff --git a/configs/avnet_ultrazedev_defconfig b/configs/avnet_ultrazedev_defconfig
>>> new file mode 100644
>>> index 000000000000..932be1cfb07a
>>> --- /dev/null
>>> +++ b/configs/avnet_ultrazedev_defconfig
>>> @@ -0,0 +1,64 @@
>>> +CONFIG_ARM=y
>>> +CONFIG_ARCH_ZYNQMP=y
>>> +CONFIG_SYS_TEXT_BASE=0x8000000
>>> +CONFIG_SYS_MALLOC_F_LEN=0x8000
>>> +CONFIG_SPL=y
>>> +CONFIG_DEBUG_UART_BASE=0xff000000
>>> +CONFIG_DEBUG_UART_CLOCK=100000000
>>> +CONFIG_SPL_SPI_FLASH_SUPPORT=y
>>> +CONFIG_SPL_SPI_SUPPORT=y
>>> +CONFIG_SPL_ZYNQMP_TWO_SDHCI=y
>>> +CONFIG_DEBUG_UART=y
>>> +CONFIG_DISTRO_DEFAULTS=y
>>> +CONFIG_FIT=y
>>> +CONFIG_FIT_VERBOSE=y
>>> +CONFIG_SPL_LOAD_FIT=y
>>> +CONFIG_BOOTDELAY=0
>>> +# CONFIG_DISPLAY_CPUINFO is not set
>>> +CONFIG_SPL_TEXT_BASE=0xfffc0000
>>> +CONFIG_SPL_OS_BOOT=y
>>> +CONFIG_SYS_PROMPT="ZynqMP> "
>>> +CONFIG_CMD_MEMTEST=y
>>> +CONFIG_CMD_FPGA_LOADBP=y
>>> +CONFIG_CMD_FPGA_LOADP=y
>>> +CONFIG_CMD_MMC=y
>>> +CONFIG_CMD_SF=y
>>> +CONFIG_CMD_SPI=y
>>> +CONFIG_CMD_TIME=y
>>> +CONFIG_CMD_TIMER=y
>>> +CONFIG_CMD_EXT4_WRITE=y
>>> +CONFIG_SPL_OF_CONTROL=y
>>> +CONFIG_DEFAULT_DEVICE_TREE="avnet-ultrazedev"
>>> +CONFIG_NET_RANDOM_ETHADDR=y
>>> +CONFIG_SPL_DM=y
>>> +CONFIG_SPL_DM_SEQ_ALIAS=y
>>> +CONFIG_CLK_ZYNQMP=y
>>> +CONFIG_FPGA_XILINX=y
>>> +CONFIG_FPGA_ZYNQMPPL=y
>>> +CONFIG_DM_GPIO=y
>>> +CONFIG_DM_I2C=y
>>> +CONFIG_SYS_I2C_CADENCE=y
>>> +CONFIG_I2C_MUX=y
>>> +CONFIG_I2C_MUX_PCA954x=y
>>> +CONFIG_MISC=y
>>> +CONFIG_I2C_EEPROM=y
>>> +CONFIG_ZYNQ_GEM_I2C_MAC_OFFSET=0xfa
>>> +CONFIG_MMC_SDHCI=y
>>> +CONFIG_MMC_SDHCI_ZYNQ=y
>>> +CONFIG_SPI_FLASH=y
>>> +CONFIG_SPI_FLASH_BAR=y
>>> +CONFIG_SF_DUAL_FLASH=y
>>> +CONFIG_SPI_FLASH_ISSI=y
>>> +CONFIG_SPI_FLASH_MACRONIX=y
>>> +CONFIG_SPI_FLASH_SPANSION=y
>>> +CONFIG_SPI_FLASH_STMICRO=y
>>> +CONFIG_SPI_FLASH_WINBOND=y
>>> +# CONFIG_SPI_FLASH_USE_4K_SECTORS is not set
>>> +CONFIG_ZYNQ_GEM=y
>>> +CONFIG_DEBUG_UART_ZYNQ=y
>>> +CONFIG_DEBUG_UART_ANNOUNCE=y
>>> +CONFIG_ZYNQ_SERIAL=y
>>> +CONFIG_SPI=y
>>> +CONFIG_ZYNQMP_GQSPI=y
>>> +CONFIG_OF_LIBFDT_OVERLAY=y
>>> +CONFIG_EFI_LOADER_BOUNCE_BUFFER=y
>>
>> FYI: I need to take a look how to remove these board stuff and have only
>> one generic defconfig for all these boards.
> 
> Good idea!
> 

Anyway I forget to mentioned one thing. Are you willing to test this
board and keep it updated? I don't want to end up in situation where I
am with microzed and picozed which I am not able to test myself.

Thanks,
Michal