[U-Boot] [PATCH 21/25] fastboot: sunxi: Update fastboot mmc default device

Fri Jul 20 07:39:32 UTC 2018

On Thu, Jul 19, 2018 at 11:41:10PM +0530, Jagan Teki wrote:
> On Thu, Jul 19, 2018 at 6:56 PM, Maxime Ripard
> <maxime.ripard at bootlin.com> wrote:
> > On Wed, Jul 18, 2018 at 08:15:23PM +0100, Alex Kiernan wrote:
> >> On Tue, Jul 17, 2018 at 12:57 PM Maxime Ripard
> >> <maxime.ripard at bootlin.com> wrote:
> >> >
> >> > On Mon, Jul 16, 2018 at 12:11:59PM +0100, Alex Kiernan wrote:
> >> > > On Mon, Jul 16, 2018 at 11:13 AM Jagan Teki <jagannadh.teki at gmail.com> wrote:
> >> > > >
> >> > > > On Mon, Jul 16, 2018 at 3:16 PM, Maxime Ripard
> >> > > > <maxime.ripard at bootlin.com> wrote:
> >> > > > > On Mon, Jul 16, 2018 at 01:49:52PM +0530, Jagan Teki wrote:
> >> > > > >> Usually eMMC is default env fat device for environment,
> >> > > > >> if MMC_SUNXI_SLOT_EXTRA != 1 Sunxi always probed emmc
> >> > > > >> device as 1. but with DM_MMC it can be more possible to
> >> > > > >> probe eMMC as device 2 since for most of the sunxi platforms
> >> > > > >> eMMC is configured mmc2.
> >> > > > >>
> >> > > > >> So update the fastboot mmc default device as 2 if DM_MMC and
> >> > > > >> MMC_SUNXI_SLOT_EXTRA != 1 slot is 2 defined but some boards
> >> > > > >> may not use all possible mmc devices or partly disabled in DT,
> >> > > > >> for those update the device in board specific defconfig.
> >> > > > >>
> >> > > > >> Cc: Olliver Schinagl <oliver at schinagl.nl>
> >> > > > >> Cc: Chen-Yu Tsai <wens at csie.org>
> >> > > > >> Signed-off-by: Jagan Teki <jagan at amarulasolutions.com>
> >> > > > >> ---
> >> > > > >>  configs/A20-OLinuXino-Lime2-eMMC_defconfig | 1 +
> >> > > > >>  configs/Sinlinx_SinA33_defconfig           | 1 +
> >> > > > >>  configs/amarula_a64_relic_defconfig        | 1 +
> >> > > > >>  drivers/fastboot/Kconfig                   | 3 ++-
> >> > > > >>  4 files changed, 5 insertions(+), 1 deletion(-)
> >> > > > >>
> >> > > > >> diff --git a/configs/A20-OLinuXino-Lime2-eMMC_defconfig b/configs/A20-OLinuXino-Lime2-eMMC_defconfig
> >> > > > >> index 5657fc2594..20ea254191 100644
> >> > > > >> --- a/configs/A20-OLinuXino-Lime2-eMMC_defconfig
> >> > > > >> +++ b/configs/A20-OLinuXino-Lime2-eMMC_defconfig
> >> > > > >> @@ -29,4 +29,5 @@ CONFIG_AXP_ALDO4_VOLT=2800
> >> > > > >>  CONFIG_SCSI=y
> >> > > > >>  CONFIG_USB_EHCI_HCD=y
> >> > > > >>  CONFIG_USB_MUSB_GADGET=y
> >> > > > >> +CONFIG_FASTBOOT_FLASH_MMC_DEV=1
> >> > > > >>  CONFIG_SYS_USB_EVENT_POLL_VIA_INT_QUEUE=y
> >> > > > >> diff --git a/configs/Sinlinx_SinA33_defconfig b/configs/Sinlinx_SinA33_defconfig
> >> > > > >> index 394534b8b5..7841219a65 100644
> >> > > > >> --- a/configs/Sinlinx_SinA33_defconfig
> >> > > > >> +++ b/configs/Sinlinx_SinA33_defconfig
> >> > > > >> @@ -21,5 +21,6 @@ CONFIG_DFU_RAM=y
> >> > > > >>  CONFIG_FASTBOOT_CMD_OEM_FORMAT=y
> >> > > > >>  CONFIG_USB_EHCI_HCD=y
> >> > > > >>  CONFIG_USB_MUSB_GADGET=y
> >> > > > >> +CONFIG_FASTBOOT_FLASH_MMC_DEV=1
> >> > > > >
> >> > > > > Your commit doesn't make any sense: the SinaA33 and the Lime2 both
> >> > > > > have the eMMC on MMC2, and you claim you want to update the default to
> >> > > > > point to MMC2, but you're changing both these boards to point to MMC1
> >> > > > > instead?
> >> > > >
> >> > > > If DM_MMC and SLOT != 1 => default device 2 which is updated by
> >> > > > kconfig, this is with all relevant mmc nodes are enabled
> >> > > > but these two boards mmc1 is not enabled so emmc will detected in device 1
> >> > > >
> >> > > > >
> >> > > > >>  CONFIG_SYS_USB_EVENT_POLL_VIA_INT_QUEUE=y
> >> > > > >>  CONFIG_USB_FUNCTION_MASS_STORAGE=y
> >> > > > >> diff --git a/configs/amarula_a64_relic_defconfig b/configs/amarula_a64_relic_defconfig
> >> > > > >> index b72cbfabc6..caeb3f6008 100644
> >> > > > >> --- a/configs/amarula_a64_relic_defconfig
> >> > > > >> +++ b/configs/amarula_a64_relic_defconfig
> >> > > > >> @@ -12,4 +12,5 @@ CONFIG_DEFAULT_DEVICE_TREE="sun50i-a64-amarula-relic"
> >> > > > >>  # CONFIG_SPL_DOS_PARTITION is not set
> >> > > > >>  # CONFIG_SPL_EFI_PARTITION is not set
> >> > > > >>  CONFIG_USB_MUSB_GADGET=y
> >> > > > >> +CONFIG_FASTBOOT_FLASH_MMC_DEV=0
> >> > > > >>  CONFIG_SYS_USB_EVENT_POLL_VIA_INT_QUEUE=y
> >> > > > >> diff --git a/drivers/fastboot/Kconfig b/drivers/fastboot/Kconfig
> >> > > > >> index bc25ea1d9c..4a1bfd119c 100644
> >> > > > >> --- a/drivers/fastboot/Kconfig
> >> > > > >> +++ b/drivers/fastboot/Kconfig
> >> > > > >> @@ -88,7 +88,8 @@ config FASTBOOT_FLASH_MMC_DEV
> >> > > > >>       int "Define FASTBOOT MMC FLASH default device"
> >> > > > >>       depends on FASTBOOT_FLASH_MMC
> >> > > > >>       default 0 if ARCH_SUNXI && MMC_SUNXI_SLOT_EXTRA = -1
> >> > > > >> -     default 1 if ARCH_SUNXI && MMC_SUNXI_SLOT_EXTRA != -1
> >> > > > >> +     default 1 if ARCH_SUNXI && !DM_MMC && MMC_SUNXI_SLOT_EXTRA != -1
> >> > > > >> +     default 2 if ARCH_SUNXI && DM_MMC && MMC_SUNXI_SLOT_EXTRA != -1
> >> > > > >
> >> > > > > It'd be better to be fixed properly, instead of just relying on a
> >> > > > > broken index.
> >> > > >
> >> > > > I don't think we can't do anything with this now, since this INDEX
> >> > > > more rely on SPL for pinctrl enablement. if you have any suggestion
> >> > > > please share.
> >> > >
> >> > > Would another answer (at least for this specific case) to change the
> >> > > fastboot code so it doesn't need the device number in advance? Given
> >> > > we get device names along the lines of 'mmcsda1', we could parse out
> >> > > the 'a' to figure out the device number (and then use the alias code
> >> > > so your board can expose a portable name).
> >> >
> >> > That sounds like a pretty good solution. I guess we would have some
> >> > corner cases when we have a NAND and an MMC card for example that can
> >> > be flashed using fastboot, but I'm not sure we have boards in that
> >> > setup at the moment (for Allwinner SoCs at least).
> >> >
> >>
> >> It's not something I need right now, but having spent so much time
> >> refactoring the fastboot code I'd like to get to this! Support for
> >> `getvar all` is also on my todo list.
> >>
> >> The NAND stuff is a good call, I'm not sure how that fits in (not
> >> least not having a board which has a NAND part).
> >
> > Maybe we can hack something at the partition level though: instead of
> > having the partition name itself, what about prefixing it by the
> > device name? ie having something like mmc1:rootfs and nand0:rootfs, as
> > partition name.
> 
> How about giving an input during command 'fastboot <controller>
> [dev_type] <dev[:part]>' it can be fastboot 0 mmc 1 similar like ums
> did 'ums 0 mmc 1'

The dev ID doesn't have much sense with the DM either though.

Maxime

-- 
Maxime Ripard, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
https://bootlin.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180720/4e6919ec/attachment.sig>