i.MX8MN (ROMv2) fails to boot over USB

Sat Jan 29 03:48:32 CET 2022

On Fri, Jan 28, 2022 at 6:04 PM Adam Ford <aford173 at gmail.com> wrote:
>
> On Fri, Jan 28, 2022 at 7:50 PM Tim Harvey <tharvey at gateworks.com> wrote:
> >
> > On Wed, Jan 26, 2022 at 10:36 AM Adam Ford <aford173 at gmail.com> wrote:
> > >
> > > On Wed, Jan 26, 2022 at 12:31 PM Michael Nazzareno Trimarchi
> > > <michael at amarulasolutions.com> wrote:
> > > >
> > > > HI Adam
> > > >
> > > > On Wed, Jan 26, 2022 at 7:01 PM Adam Ford <aford173 at gmail.com> wrote:
> > > > >
> > > > > On Wed, Jan 26, 2022 at 11:24 AM Michael Nazzareno Trimarchi
> > > > > <michael at amarulasolutions.com> wrote:
> > > > > >
> > > > > > Hi Adam
> > > > > >
> > > > > > On Wed, Jan 26, 2022 at 2:26 AM Tim Harvey <tharvey at gateworks.com> wrote:
> > > > > > >
> > > > > > > On Tue, Jan 25, 2022 at 3:49 PM Adam Ford <aford173 at gmail.com> wrote:
> > > > > > > >
> > > > > > > > I have a flash.bin file that boots over MMC just fine, but when I use
> > > > > > > > that same flash.bin file, it appears to hang after running ATF.  I
> > > > > > > > dumped some debug info for printing the handling of the FIT file, and
> > > > > > > > I noticed the fit read sizes are also not consistent between MMC and
> > > > > > > > USB loads with the identical flash.bin.
> > > > > > > >
> > > > > > > > MMC booting is calling spl_romapi_raw_seekable_read, but the USB
> > > > > > > > booting is not.  Maybe this is normal.
> > > > > > > >
> > > > > > > > I was hoping someone might have some pointers on how I can
> > > > > > > > troubleshoot this.  It's also weird to me that the various nodes are
> > > > > > > > listed twice in the USB boot, but not the MMC boot.
> > > > > > > >
> > > > > > > > I have poked around in the spl_imx_romapi.c file, but I am not seeing
> > > > > > > > anything obvious.
> > > > > > > >
> > > > > > > > SPL and ATF are executing, so it's partially decoding the FIT file and
> > > > > > > > executing those two pieces correctly from what I can see.
> > > > > > > >
> > > > > > > > USB log (no FDT)
> > > > > > > > U-Boot SPL 2022.01-00753-g7f7f25ecb8-dirty (Jan 25 2022 - 15:54:57 -0600)
> > > > > > > > WDT:   Started watchdog at 30280000 with servicing (60s timeout)
> > > > > > > > Trying to boot from BOOTROM
> > > > > > > > Find img info 0x&48027600, size 872
> > > > > > > > Need continue download 1024
> > > > > > > > fit read sector 48027600, sectors=872, dst=42206c40, count=872, size=0x368
> > > > > > > > firmware: 'uboot'
> > > > > > > > External data: dst=40200000, offset=3000, size=a4d98
> > > > > > > > fdt: 'fdt'
> > > > > > > > External data: dst=402a4dc0, offset=b1e60, size=af68
> > > > > > > > loadables: 'atf'
> > > > > > > > External data: dst=960000, offset=a7d98, size=a0c6
> > > > > > > > no string for index 1
> > > > > > > > Download 772040, Total size 773576
> > > > > > > > fit read sector 48027600, sectors=872, dst=42206fb0, count=872, size=0x368
> > > > > > > > firmware: 'uboot'
> > > > > > > > External data: dst=40200000, offset=3000, size=a4d98
> > > > > > > > fdt: 'fdt'
> > > > > > > > External data: dst=402a4dc0, offset=b1e60, size=af68
> > > > > > > > loadables: 'atf'
> > > > > > > > External data: dst=960000, offset=a7d98, size=a0c6
> > > > > > > > no string for index 1
> > > > > > > > NOTICE:  BL31: v2.4(release):lf-5.10.72-2.2.0-0-g5782363f9
> > > > > > > > NOTICE:  BL31: Built : 14:45:34, Jan 25 2022
> > > > > > > >
> > > > > > > > <hang>
> > > > > > > >
> > > > > > > > MMC Booting Log (with DTB):
> > > > > > > >
> > > > > > > > U-Boot SPL 2022.01-00753-g7f7f25ecb8-dirty (Jan 25 2022 - 15:54:57 -0600)
> > > > > > > > WDT:   Started watchdog at 30280000 with servicing (60s timeout)
> > > > > > > > Trying to boot from BOOTROM
> > > > > > > > image offset 0x8000, pagesize 0x200, ivt offset 0x0
> > > > > > > > ROM API load from 0x60000, size 0x400
> > > > > > > > spl_romapi_raw_seekable_read 0x60000, size 0x400
> > > > > > > > fit read sector 300, sectors=2, dst=42206c40, count=2, size=0x368
> > > > > > > > firmware: 'uboot'
> > > > > > > > ROM API load from 0x63000, size 0xa4e00
> > > > > > > > spl_romapi_raw_seekable_read 0x63000, size 0xa4e00
> > > > > > > > External data: dst=40200000, offset=3000, size=a4d98
> > > > > > > > fdt: 'fdt'
> > > > > > > > ROM API load from 0x111e00, size 0xb000
> > > > > > > > spl_romapi_raw_seekable_read 0x111e00, size 0xb000
> > > > > > > > External data: dst=402a4dc0, offset=b1e60, size=af68
> > > > > > > > loadables: 'atf'
> > > > > > > > ROM API load from 0x107c00, size 0xa400
> > > > > > > > spl_romapi_raw_seekable_read 0x107c00, size 0xa400
> > > > > > > > External data: dst=960000, offset=a7d98, size=a0c6
> > > > > > > > no string for index 1
> > > > > > > > NOTICE:  BL31: v2.4(release):lf-5.10.72-2.2.0-0-g5782363f9
> > > > > > > > NOTICE:  BL31: Built : 14:45:34, Jan 25 2022
> > > > > > > >
> > > > > > > > <u-boot loads successfully from here>
> > > > > > >
> > > > > > > Adam,
> > > > > > >
> > > > > > > I've previously run into this as well and unfortunately have no
> > > > > > > solution. Note that I had to hack in some support to the SPL for USB
> > > > > > > which I assume you've done as well. If I recall trying to enable
> > > > > > > DM_USB for the SPL took too much space so that hacks I had revolved
> > > > > > > around getting it to work in the non-DM fashion.
> > > > > >
> > > > > > I was having the same problem and I found that was related to
> > > > > > diff --git a/arch/arm/mach-imx/imx8m/soc.c b/arch/arm/mach-imx/imx8m/soc.c
> > > > > > index 863508776d..9d6dab08e6 100644
> > > > > > --- a/arch/arm/mach-imx/imx8m/soc.c
> > > > > > +++ b/arch/arm/mach-imx/imx8m/soc.c
> > > > > > @@ -1333,6 +1333,7 @@ enum env_location env_get_location(enum
> > > > > > env_operation op, int prio)
> > > > > >
> > > > > > You should support USB_BOOT in switch case or change your defconfig
> > > > > > and add CONFIG_ENV_IS_NOWHERE=y
> > > > >
> > > > > Thanks for the tip.  it strikes me as odd that the system would hang
> > > > > if we don't know what to do with the environment.  I would expect it
> > > > > to just load the default and move on.
> > > > >
> > > > > I'll work on a defconfig fix.
> > > > >
> > > >
> > > > Those nxp function should be dropped or fixed anyway
> > >
> > > There was an RFC to move this function to the board files [1].  I
> > > think this is a good idea since some people may want environmental
> > > variables stored in a specific place rather than be dependent on the
> > > boot source.  I know my colleagues at Beacon want that.
> > >
> >
> > Adam,
> >
> > Can you explain what you are doing here and what it means to boot from BOOTROM?
> >
> > I went back and looked at the issue I was running into on the IMX8MN
> > as I thought it was what you were seeing but in my case I was trying
> > to boot via SDP on a board with USB-OTG and an erased eMMC using uuu.
> > The issue I ran into enabling SDP support was that the SPL code
> > overflowed the SRAM size so I disabled SPL_DM_USB and that is where I
> > had to hack in some non-dm code to get around it. That was probably
> > not the right approach and I need to figure out where code can be
> > trimmed down. Are you able to enable SPL_DM_USB and use USB from the
> > SPL or are you doing something different here?
>
> I don't need USB in SPL because the boot ROM can handle all that.  If
> you look at the spl.c file in boards/beacon/imx8mn, you'll see that
> spl_board_boot_device  only returns the boot ROM regardless of the
> actual boot device.  The BootROM handles all the USB traffic in SPL
> for me.  It also handles the SPL booting from MMC, and when using
> NXP's custom U-Boot, it also boots from FlexSPI.   (side note, I am
> going to work on FlexSPI booting upstream soon).  There is special
> code in SPL to use boot ROM API which can do all the loading of ATF,
> U-Boot, etc. from the fit image and SPL can coordinate what data, and
> how much etc.
>
> To make this work, I had to enable CONFIG_ENV_IS_NOWHERE [1] in my
> defconfig because of some platform code for Nano and Plus and how it
> handles the locations of the U-Boot environment.
>
> This should be very similar to how the imx8mn-evk works too.
>
> adam
>
> [1] - https://patchwork.ozlabs.org/project/uboot/patch/20220126182523.58575-1-aford173@gmail.com/

Adam,

Thanks - I will look into that. It would certainly help with code size
issues I'm running into on IMX8MN.

>From what I can tell this support is only available on the Nano and
not the other IMX8M's correct?

Best regards,

Tim