Re: [PATCH 1/3] arm: mediatek: enable SPL fastboot DAs on Genio 510 and 700

[Carlo Caione <ccaione@baylibre.com>]

On Tue, Sep 22, 2026 at 11:27:32 +0100, David Lechner wrote:
> On 9/14/26 8:35 AM, Carlo Caione wrote:
> > From: Julien Masson <jmasson@baylibre.com>
> > 
> > Genio 510 and 700 need a provisioning and recovery path that remains
> > usable when eMMC is blank or its boot firmware is unusable. Add a RAM-
> > based fastboot download agent so boards can be flashed through the
> > MediaTek BootROM download protocol without a working on-device
> > bootloader.
> > 
> > Running the agent in SPL avoids a dependency on BL31 and U-Boot proper
> > during provisioning.
> 
> It would be helpful to have some more explanation here about why the
> different parts of this patch are doing what they are doing.

Yeah, I agree. I'll add the explanation of the why in the commit
message.
 
> > 
> > Signed-off-by: Julien Masson <jmasson@baylibre.com>
> > Signed-off-by: Carlo Caione <ccaione@baylibre.com>
> > ---
> >  arch/arm/dts/mt8188-genio-da-u-boot.dtsi           | 55 ++++++++++++++++++
> >  arch/arm/dts/mt8370-genio-510-evk-u-boot.dtsi      |  4 ++
> >  arch/arm/dts/mt8390-genio-700-evk-u-boot.dtsi      |  4 ++
> >  arch/arm/dts/mtk-genio-da-binman-u-boot.dtsi       | 16 ++++++
> >  .../dts/mtk-genio-loader-spl-binman-u-boot.dtsi    |  6 ++
> >  arch/arm/mach-mediatek/Kconfig                     | 26 +++++++++
> >  arch/arm/mach-mediatek/Makefile                    |  1 +
> >  arch/arm/mach-mediatek/spl_fastboot.c              | 65 ++++++++++++++++++++++
> >  board/mediatek/MAINTAINERS                         |  5 ++
> >  common/spl/Kconfig                                 |  2 +-
> >  configs/mt8188-genio-da-spl-fastboot.config        | 15 +++++
> >  configs/mt8188-genio-da.config                     | 19 +++++++
> 
> I think these should use mt8370 prefix like the other common genio 510/700
> stuff. mt8188 is for chromebooks, so I don't think it would make sense to
> use that number with genio-specific configs.

:( yup, makes sense.

> >  configs/mt8370_genio_510_evk_da_spl_defconfig      |  2 +
> >  configs/mt8390_genio_700_evk_da_spl_defconfig      |  2 +
> >  14 files changed, 221 insertions(+), 1 deletion(-)
> > 
> > diff --git a/arch/arm/dts/mt8188-genio-da-u-boot.dtsi b/arch/arm/dts/mt8188-genio-da-u-boot.dtsi
> > new file mode 100644
> > index 00000000000..e9078f37157
> > --- /dev/null
> > +++ b/arch/arm/dts/mt8188-genio-da-u-boot.dtsi
> > @@ -0,0 +1,55 @@
> > +// SPDX-License-Identifier: (GPL-2.0-only OR MIT)
> > +/*
> > + * Copyright (c) 2026 BayLibre, SAS.
> > + * Author: Carlo Caione <ccaione@baylibre.com>
> > + */
> 
> I would add a comment here explaining that da is download agent and why we
> have this special devicetree that is different from the normal one.

I'll add a comment.

> > +
> > +#include "mtk-genio-binman-u-boot.dtsi"
> > +#include "mtk-genio-da-binman-u-boot.dtsi"
> > +
> > +&ssusb0 {
> > +	dr_mode = "peripheral";
> 
> > +	mediatek,force-vbus;
> 
> This is not an upstream DT property.
> 
> It seems like dr_mode = "peripheral" should already imply this anyway. So
> hopefully we can just modify the USB driver and leave this property out.

Good point. I'll make the change to the MTU3 driver.

> > +};
> > +
> > +&ssusb1 {
> > +	status = "disabled";
> > +};
> > +
> > +&ssusb2 {
> > +	status = "disabled";
> > +};
> > +
> > +#ifdef CONFIG_MTK_BROM_DOWNLOAD_AGENT_SPL_FASTBOOT
> 
> Why do these matter for fastboot but not the ones above?

Because only the fastboot-in-SPL variant uses USB during SPL execution,
in the fastboot-in-U-Boot-proper case these nodes are already in the
regular U-Boot DT (no need for 'bootph-pre-ram').

> > +&pericfg_ao {
> > +	bootph-pre-ram;
> > +};
> > +
> > +&ssusb0 {
> > +	bootph-pre-ram;
> > +};
> > +
> > +&u3phy0 {
> > +	bootph-pre-ram;
> > +};
> > +
> > +&u2port0 {
> > +	bootph-pre-ram;
> > +};
> > +
> > +&usb_default_pins {
> > +	bootph-pre-ram;
> > +
> > +	pins-iddig {
> > +		bootph-pre-ram;
> > +	};
> > +
> > +	pins-valid {
> > +		bootph-pre-ram;
> > +	};
> > +
> > +	pins-vbus {
> > +		bootph-pre-ram;
> > +	};
> > +};
> > +#endif

[...]
> > --- /dev/null
> > +++ b/arch/arm/dts/mtk-genio-da-binman-u-boot.dtsi
> > @@ -0,0 +1,16 @@
> > +// SPDX-License-Identifier: (GPL-2.0-only OR MIT)
> > +/*
> > + * Copyright (c) 2026 BayLibre, SAS.
> > + * Author: Carlo Caione <ccaione@baylibre.com>
> > + */
> > +
> > +&binman {
> > +	da: da {
> > +		filename = "da.bin";
> > +		pad-byte = <0>;
> > +
> > +#define MTK_GENIO_SIMPLE_SPL
> 
> Could we pick a name that describe more precicly what this does? SIMPLE
> is vague.

Fair enough.

[...]
> >  obj-$(CONFIG_TARGET_MT7622) += mt7622/
> >  obj-$(CONFIG_TARGET_MT7623) += mt7623/
> > diff --git a/arch/arm/mach-mediatek/spl_fastboot.c b/arch/arm/mach-mediatek/spl_fastboot.c
> > new file mode 100644
> > index 00000000000..0ba76d47dfb
> > --- /dev/null
> > +++ b/arch/arm/mach-mediatek/spl_fastboot.c
> > @@ -0,0 +1,65 @@
> > +// SPDX-License-Identifier: GPL-2.0-only
> > +/*
> > + * Copyright (c) 2024 BayLibre, SAS.
> 
> 2026

uh, time flies.

> > + * Author: Julien Masson <jmasson@baylibre.com>
> > + */
> > +
> > +#include <dm/uclass.h>
> > +#include <fastboot.h>
> > +#include <mmc.h>
> 
> Suggest also <stdio.h> for printf and <linux/errno.h> for ENODEV.

Ok

> > +
> > +static int mtk_mmc_boot_prepare(void)
> > +{
> > +	struct udevice *dev;
> > +	struct mmc *mmc;
> > +	int ret;
> > +
> > +	ret = uclass_get_device_by_seq(UCLASS_MMC,
> > +				       CONFIG_FASTBOOT_FLASH_MMC_DEV, &dev);
> 
> Should this be find_mmc_device(CONFIG_VAL(FASTBOOT_FLASH_MMC_DEV))?

No, we need this because we need to actually probe the device.

[...]
> > diff --git a/common/spl/Kconfig b/common/spl/Kconfig
> > index f9b453df766..77da2f4c8aa 100644
> > --- a/common/spl/Kconfig
> > +++ b/common/spl/Kconfig
> > @@ -194,7 +194,7 @@ config SPL_SHOW_ERRORS
> >  config SPL_BINMAN_SYMBOLS
> >  	bool "Declare binman symbols in SPL"
> >  	depends on SPL_FRAMEWORK && BINMAN
> > -	default n if MTK_GENIO_BOOT_IMAGES
> > +	default n if MTK_GENIO_IMAGE_CONTRACT
> 
> Should we do something to make sure we don't select MTK_GENIO_BOOT_IMAGES
> and MTK_BROM_DOWNLOAD_AGENT at the same time?

I'll add a depends on !MTK_GENIO_BOOT_IMAGES

> >  	default y
> >  	help
> >  	  This enables use of symbols in SPL which refer to other entries in
> > diff --git a/configs/mt8188-genio-da-spl-fastboot.config b/configs/mt8188-genio-da-spl-fastboot.config
> > new file mode 100644
> > index 00000000000..01b35f1144e
> > --- /dev/null
> > +++ b/configs/mt8188-genio-da-spl-fastboot.config
> > @@ -0,0 +1,15 @@
> > +#include <configs/mt8188-genio-da.config>
> > +
> > +CONFIG_MTK_BROM_DOWNLOAD_AGENT_SPL_FASTBOOT=y
> > +# Sparse writes need an 8 MiB DMA buffer in addition to the fastboot state.
> > +CONFIG_SPL_SYS_MALLOC_SIZE=0x1000000
> > +CONFIG_SPL_MMC_WRITE=y
> > +CONFIG_SPL_PHY=y
> > +CONFIG_SPL_DM_USB_GADGET=y
> > +CONFIG_SPL_USB_GADGET=y
> > +CONFIG_SPL_USB_MTU3=y
> > +CONFIG_SPL_FASTBOOT=y
> > +CONFIG_SPL_FASTBOOT_REBOOT=y
> > +CONFIG_SPL_FASTBOOT_FLASH_MMC=y
> > +CONFIG_SPL_FASTBOOT_MMC_BOOT_SUPPORT=y
> > +CONFIG_SPL_FASTBOOT_MMC_USER_SUPPORT=y
> > diff --git a/configs/mt8188-genio-da.config b/configs/mt8188-genio-da.config
> 
> Do these really need to be two separate files? As mentioned above, there
> isn't really such thing as a mt8188-genio seprate from a mt8370-genio-common.
> mt8188.config already contains things that really are genio-specific that
> would be more "correct" to move to a separate file. But that is a bigger
> cleanup that affects all MediaTek Genio boards that we can look at later.

ok, I can slightly make this better by splitting:

- mt8370-genio-da.config: settings shared by both DA variants
- mt8370-genio-da-spl-fastboot.config: SPL fastboot settings
- mt8370-genio-da-uboot-fastboot.config: U-Boot-proper fastboot settings

Thanks!

--
Carlo Caione