[U-Boot] [PATCH v2 00/31] dm: tegra: Move nyan-big and beaver to livetree

Mon Jun 12 03:49:17 UTC 2017

Hi Marcel,

On 9 June 2017 at 10:30, Simon Glass <sjg at chromium.org> wrote:
>
> Hi Marcel,
>
> On 9 June 2017 at 06:39, Marcel Ziswiler <marcel.ziswiler at toradex.com> wrote:
> >
> > Hi Simon
> >
> > On Thu, 2017-06-08 at 22:04 -0600, Simon Glass wrote:
> > > Hi,
> > >
> > > On 8 June 2017 at 21:06, Simon Glass <sjg at chromium.org> wrote:
> > > > Hi Tom (Warren) / Marcel,
> > > >
> > > > On 2 June 2017 at 21:03, Simon Glass <sjg at chromium.org> wrote:
> > > > >
> > > > > This moves an entire board to use a live device tree as an
> > > > > example of the
> > > > > impact.
> > > > >
> > > > > Nyan-big was chosen because I can easily and boot U-Boot without
> > > > > any
> > > > > media swapping, etc. Beaver is enabled as well since it failed to
> > > > > boot
> > > > > with serial v1 due to a disabled console node.
> > > > >
> > > >
> > > > Have you been able to test this v2 series? Some problems were
> > > > reported
> > > > with v1 and I have found a few more since I discovered my jetson-
> > > > tk1
> > > > board. I'm planning to send a v3 soon, but please let me know if
> > > > you
> > > > see any issues.
> > >
> > > One more note - I pushed a v3 series to u-boot-dm/livet-working but
> > > have not sent the patches. Let me know if you find anything.
> >
> > While Apalis TK1 now at least boots without any GPIO driver issues I
> > still saw the following few glitches:
> >
> > Apalis TK1 # gpio status -a
> > Bank A:
> > A0: func
> > A1: input: 1 [ ]
> > A2: func
> > ...
> > FF7: func
> >
> > Bank as3722_:
> > as3722_0: unknown
> > as3722_1: unknown
> > as3722_2: unknown
> > as3722_3: unknown
> > as3722_4: unknown
> > as3722_5: unknown
> > as3722_6: unknown
> > as3722_7: unknown
> >
> > Apalis TK1 # mmc dev 0
> > tegra_mmc_send_cmd_bounced: MMC Timeout
> >     Interrupt status        0x00000001
> >     Interrupt status enable 0xffff003b
> >     Interrupt signal enable 0xffff0002
> >     Present status          0x01fb02f6
> > mmc_init: -1, time 8066
> >
> > While PCIe enumeration does not error out it seems not to detect the
> > Ethernet chip:
> >
> > Apalis TK1 # pci enum
> > Apalis TK1 # dm tree
> >  Class       Probed   Name
> > ----------------------------------------
> >  root        [ + ]    root_driver
> >  pci         [ + ]    |-- pcie-controller at 01003000
> > ...
> >
> > That's how it should look (e.g. with older working U-Boot):
> >
> > Apalis TK1 # ver
> >
> > U-Boot 2016.11-2.7.2+g60021a4 (Apr 10 2017 - 07:55:49 +0200)
> > arm-angstrom-linux-gnueabi-gcc (Linaro GCC 6.2-2016.11) 6.2.1 20161016
> > GNU ld (GNU Binutils) 2.27.0.20160806
> > Apalis TK1 # pci enum
> > Apalis TK1 # dm tree
> >  Class       Probed   Name
> > ----------------------------------------
> >  root        [ + ]    root_driver
> >  pci         [ + ]    |-- pcie-controller at 01003000
> >  pci         [ + ]    |   `-- pci_0:2.0
> >  eth         [   ]    |       `-- e1000#0
> >
> > After enabling CONFIG_OF_LIVE I noticed the following additional issue:
> >
> > Apalis TK1 # pci enum
> > ERROR: tegra-pcie: resource "pads" not found
> > at u-boot.git/drivers/pci/pci_tegra.c:499/tegra_pcie_parse_dt()
>
> OK that sounds like a new problem, I'll take a look.
>
> >
> > Trying to use a vanilla mainline Linux kernel device tree also did not
> > help:
> >
> > U-Boot 2017.07-rc1-00150-gf325a89-dirty (Jun 09 2017 - 14:33:21 +0200)
> >
> > TEGRA124
> > DRAM:  2 GiB
> > ERROR: tegra-xusb-padctl: failed to parse "nvidia,lanes" property
> > at /run/media/zim/Build/Sources/u-boot.git/arch/arm/mach-
> > tegra/tegra124/../xusb-padctl-
> > common.c:87/tegra_xusb_padctl_group_parse_dt()
> > ERROR: tegra-xusb-padctl: failed to parse group usb2
> > at /run/media/zim/Build/Sources/u-boot.git/arch/arm/mach-
> > tegra/tegra124/../xusb-padctl-
> > common.c:237/tegra_xusb_padctl_config_parse_dt()
> > ERROR: tegra-xusb-padctl: failed to parse entry pads: -22
> > at /run/media/zim/Build/Sources/u-boot.git/arch/arm/mach-
> > tegra/tegra124/../xusb-padctl-common.c:270/tegra_xusb_padctl_parse_dt()
> > ERROR: tegra-xusb-padctl: failed to parse "nvidia,lanes" property
> > at /run/media/zim/Build/Sources/u-boot.git/arch/arm/mach-
> > tegra/tegra124/../xusb-padctl-
> > common.c:87/tegra_xusb_padctl_group_parse_dt()
> > ERROR: tegra-xusb-padctl: failed to parse group usb2-0
> > at /run/media/zim/Build/Sources/u-boot.git/arch/arm/mach-
> > tegra/tegra124/../xusb-padctl-
> > common.c:237/tegra_xusb_padctl_config_parse_dt()
> > ERROR: tegra-xusb-padctl: failed to parse entry ports: -22
> > at /run/media/zim/Build/Sources/u-boot.git/arch/arm/mach-
> > tegra/tegra124/../xusb-padctl-common.c:270/tegra_xusb_padctl_parse_dt()
>
> This should be fixed by u-boot-dm/livet-working  - can you please
> check you have the latest version?
>
> >
> > Just the order of the SDHCI controllers changed:
> >
> > Apalis TK1 # mmc dev 2; mmc info
> > tegra_mmc_send_cmd_bounced: MMC Timeout
> >     Interrupt status        0x00000001
> >     Interrupt status enable 0xffff003b
> >     Interrupt signal enable 0xffff0002
> >     Present status          0x01fb02f6
> > mmc_init: -1, time 8066
>
> It could be this:
>
> http://patchwork.ozlabs.org/patch/772771/
>
> So the ordering has changed?
>
> >
> > But PCIe enumeration still fails:
> >
> > Apalis TK1 # pci enum
> > ERROR: tegra-pcie: resource "pads" not found
> > at /run/media/zim/Build/Sources/u-
> > boot.git/drivers/pci/pci_tegra.c:499/tegra_pcie_parse_dt()
> >
> > Let me know should you require any additional information.
>
> Much appreciated.
>
> As above can you please confirm the git hash you tested?

I found the tegra's PCI was not ported to livetree and also there was
a bug in the PCI uclass (which suggests that the PCI test coverage is
not so great).

I've pushed a new version to u-boot-dm/livet-working and will send v3
patches if they pass buildman.

My tegra-TK1 PCI does not work. I think it is because I accidentally
corrupted the PMIC register values at one point. Is there a way to
restore it?

Regards,
Simon