Re: [PATCH V4 0/7] imx: blk-ctrl: Add interconnect for i.MX8MP

From: Tim Harvey
Date: Wed Oct 26 2022 - 13:22:55 EST

On Mon, Oct 17, 2022 at 11:53 AM Tim Harvey <tharvey@xxxxxxxxxxxxx> wrote:
>
> On Thu, Oct 13, 2022 at 2:49 PM Lucas Stach <l.stach@xxxxxxxxxxxxxx> wrote:
> >
> > Hi Tim,
> >
> > Am Donnerstag, dem 13.10.2022 um 11:23 -0700 schrieb Tim Harvey:
> > > On Wed, Oct 12, 2022 at 11:46 AM Tim Harvey <tharvey@xxxxxxxxxxxxx> wrote:
> > > >
> > > > On Tue, Oct 11, 2022 at 4:13 PM Marek Vasut <marex@xxxxxxx> wrote:
> > > > >
> > > > > On 10/11/22 22:10, Tim Harvey wrote:
> > > > > > On Fri, Jul 8, 2022 at 1:57 AM Peng Fan (OSS) <peng.fan@xxxxxxxxxxx> wrote:
> > > > > > >
> > > > > > > From: Peng Fan <peng.fan@xxxxxxx>
> > > > > > >
> > > > > > > V4:
> > > > > > > Because the header is not included when adding NoC node, the fsl,imx8mp.h
> > > > > > > needs be included in this patchset. So include it in patch 6
> > > > > > >
> > > > > > > V3:
> > > > > > > Move adding NoC node patch to i.MX8MP ICC driver patchset
> > > > > > > Per Lucas's comments, warn once when icc bulk get not return probe defer and continue.
> > > > > > >
> > > > > > > V2:
> > > > > > > Use a low bandwidth value instead INT_MAX
> > > > > > > Minor fix to move fsl,imx8mp.h out to dts patch, not driver patch
> > > > > > > Add A-b tag from DT maintainer
> > > > > > >
> > > > > > > i.MX8MP NoC settings is invalid after related power domain up. So
> > > > > > > need to set valid values after power domain up.
> > > > > > >
> > > > > > > This patchset is to bind interconnect for each entry in blk ctrl.
> > > > > > >
> > > > > > > This patchset is not include DVFS DDRC feature.
> > > > > > >
> > > > > > > Peng Fan (7):
> > > > > > > dt-bindings: soc: imx: add interconnect property for i.MX8MP media blk
> > > > > > > ctrl
> > > > > > > dt-bindings: soc: imx: add interconnect property for i.MX8MP hdmi blk
> > > > > > > ctrl
> > > > > > > dt-bindings: soc: imx: add interconnect property for i.MX8MP hsio blk
> > > > > > > ctrl
> > > > > > > soc: imx: add icc paths for i.MX8MP media blk ctrl
> > > > > > > soc: imx: add icc paths for i.MX8MP hsio/hdmi blk ctrl
> > > > > > > arm64: dts: imx8mp: add interconnects for media blk ctrl
> > > > > > > arm64: dts: imx8mp: add interconnect for hsio blk ctrl
> > > > > > >
> > > > > > > .../soc/imx/fsl,imx8mp-hdmi-blk-ctrl.yaml | 9 +++++
> > > > > > > .../soc/imx/fsl,imx8mp-hsio-blk-ctrl.yaml | 10 +++++
> > > > > > > .../soc/imx/fsl,imx8mp-media-blk-ctrl.yaml | 14 +++++++
> > > > > > > arch/arm64/boot/dts/freescale/imx8mp.dtsi | 18 +++++++++
> > > > > > > drivers/soc/imx/imx8m-blk-ctrl.c | 39 +++++++++++++++++++
> > > > > > > drivers/soc/imx/imx8mp-blk-ctrl.c | 35 +++++++++++++++++
> > > > > > > 6 files changed, 125 insertions(+)
> > > > > > >
> > > > > > > --
> > > > > > > 2.25.1
> > > > > > >
> > > > > >
> > > > > > Hi Peng,
> > > > > >
> > > > > > I built origin/master from commit 041bc24d867a today for an imx8mp
> > > > > > board and am running into errors that appear to be introduced by this
> > > > > > series:
> > > > > > [ 15.177372] platform 381f0040.usb-phy: deferred probe pending
> > > > > > [ 15.183155] platform 382f0040.usb-phy: deferred probe pending
> > > > > > [ 15.188928] platform 33800000.pcie: deferred probe pending
> > > > > > [ 15.194439] platform 32ec0000.blk-ctrl: deferred probe pending
> > > > > > [ 15.200287] platform 38330000.blk-ctrl: deferred probe pending
> > > > > > [ 15.206129] platform 32f10000.blk-ctrl: deferred probe pending
> > > > > > [ 15.211974] platform 32f10100.usb: deferred probe pending
> > > > > > [ 15.217382] platform 32f10108.usb: deferred probe pending
> > > > > > [ 15.222791] platform cpufreq-dt: deferred probe pending
> > > > > > # cat /sys/kernel/debug/devices_deferred
> > > > > > 381f0040.usb-phy platform: supplier 32f10000.blk-ctrl not ready
> > > > > > 382f0040.usb-phy platform: supplier 32f10000.blk-ctrl not ready
> > > > > > 33800000.pcie platform: supplier 32f10000.blk-ctrl not ready
> > > > > > 32ec0000.blk-ctrl imx8m-blk-ctrl: failed to get noc entries
> > > > > > 38330000.blk-ctrl imx8m-blk-ctrl: failed to get noc entries
> > > > > > 32f10000.blk-ctrl imx8mp-blk-ctrl: failed to get noc entries
> > > > > > 32f10100.usb platform: supplier 32f10000.blk-ctrl not ready
> > > > > > 32f10108.usb platform: supplier 32f10000.blk-ctrl not ready
> > > > > > cpufreq-dt
> > > > > >
> > > > > > Is there a driver I'm perhaps missing that is needed now or are there
> > > > > > some patches that come from a different unmerged tree needed?
> > > > >
> > > > > Do you have these enabled ?
> > > > >
> > > > > CONFIG_INTERCONNECT_IMX8MM=y
> > > > > CONFIG_INTERCONNECT_IMX8MN=y
> > > > > CONFIG_INTERCONNECT_IMX8MQ=y
> > > > > CONFIG_INTERCONNECT_IMX8MP=y
> > > >
> > > > Marek,
> > > >
> > > > Yes, I have those as well as CONFIG_ARCH_NXP which appears new for 6.1 enabled.
> > > >
> > > > Best Regards,
> > > >
> > > > Tim
> > >
> > > I see this issue on origin/master 1440f5760228 using
> > > arch/arm64/configs/defconfig.
> > >
> > > It seems to me that the imx8m*_blk_ctrl_probe will all defer now until
> > > perhaps all the drivers using interconnects can probe, such as
> > > g1/lcdif etc?
> > >
> > > Some added debugging shows me:
> > > [ 14.951371] of_icc_bulk_get path lcdif-rd err=-517
> > > [ 14.956205] devm_of_icc_bulk_get ret=-517
> > > [ 14.960562] imx8m_blk_ctrl_probe failed -517
> > > [ 14.967191] of_icc_bulk_get path g1 err=-517
> > > [ 14.971487] devm_of_icc_bulk_get ret=-517
> > > [ 14.975614] imx8m_blk_ctrl_probe failed -517
> > > [ 14.982200] of_icc_bulk_get path usb1 err=-517
> > > [ 14.986680] devm_of_icc_bulk_get ret=-517
> > > [ 14.990709] imx8mp_blk_ctrl_probe 0:usb1
> > > [ 14.994641] imx8mp_blk_ctrl_probe 1:usb2
> > > [ 15.002086] platform 381f0040.usb-phy: deferred probe pending
> > > [ 15.007875] platform 382f0040.usb-phy: deferred probe pending
> > > [ 15.013636] platform 32f00000.pcie-phy: deferred probe pending
> > > [ 15.019480] platform 33800000.pcie: deferred probe pending
> > > [ 15.024975] platform 32ec0000.blk-ctrl: deferred probe pending
> > > [ 15.030819] platform 38330000.blk-ctrl: deferred probe pending
> > > [ 15.036662] platform 32f10000.blk-ctrl: deferred probe pending
> > > [ 15.042503] platform 32f10100.usb: deferred probe pending
> > > [ 15.047912] platform 32f10108.usb: deferred probe pending
> > >
> > Do you have CONFIG_ARM_IMX_BUS_DEVFREQ enabled? This one will actually
> > instantiate the interconnect devices for the interconnect drivers to
> > hang onto. Once the interconnect is probed the blk-ctrl drivers should
> > probe. There is no dependency into leaf peripheral devices.
> >
> > Regards,
> > Lucas
> >
>
> Lucas,
>
> That's configured as a module in arch/arm64/configs/defconfig - making
> it static still does not help. I am booting a static kernel with no
> modules here so there must be something else required that is
> configured as a module that wasn't needed before this series?
>
> I haven't had much time to look into this. I'm simply using
> arch/arm64/configs/defconfig and CONFIG_INITRAMFS_SOURCE set to a
> rootfs I use for quick testing. If there is indeed now one or more
> modules required before USB and PCIe are available I wonder if they
> should be made static in arch/arm64/configs/defconfig? If not then I
> don't know what the point in having the PCI and USB drivers static is.
>

Lucas,

Sorry, I misspoke - I did not have CONFIG_ARM_IMX_BUS_DEVFREQ and that
was indeed my issue.

Question: Should the intention of arch/arm64/configs/defconfig be to
provide static bus support for PCIe/USB which can be boot devices?

Prior to the interconnect drivers (and even following those drivers
for imx8mm/imx8mn) PCIe/USB are available without module support with
arch/arm64/configs/defconfig but now the following would need to be
changed to make PCIe/USB available with defconfig and no modules:
CONFIG_INTERCONNECT_IMX8MM=y
CONFIG_INTERCONNECT_IMX8MN=y
CONFIG_INTERCONNECT_IMX8MQ=y
CONFIG_INTERCONNECT_IMX8MP=y
CONFIG_ARM_IMX_BUS_DEVFREQ=y

Without this change one would need an initrd to boot off PCIe/NVMe or
USB boot devices. I'm simply not clear what the intended use case is
for the defconfig with regards to modules vs static drivers.

Best Regards,

Tim