Re: HiKey boot hangs with Linus' HEAD due to coresight dts changes?

[John Stultz]

On Thu, Nov 16, 2017 at 6:06 PM, Leo Yan <leo.yan@xxxxxxxxxx> wrote:
> Hi John,
>
> On Thu, Nov 16, 2017 at 05:58:54PM -0800, John Stultz wrote:
>> On Thu, Nov 16, 2017 at 5:54 PM, John Stultz <john.stultz@xxxxxxxxxx> wrote:
>> > Testing with Linus' HEAD today on the HiKey board, I started seeing boot hangs:
>> > ...
>> >
>> > [    0.677971] Serial: AMBA PL011 UART driver
>> > [    0.694299] pstore: using zlib compression
>> > [    0.698830] console [pstore-1] enabled
>> > [    0.702623] pstore: Registered ramoops as persistent store backend
>> > [    0.708880] ramoops: attached 0x100000@0x21f00000, ecc: 0/0
>> > [    0.715847] f8015000.uart: ttyAMA0 at MMIO 0xf8015000 (irq = 7,
>> > base_baud = 0) is a PL011 rev2
>> > [    0.725011] uart-pl011 f7111000.uart: could not find pctldev for
>> > node /soc/pinmux@f7010000/uart1_pmx_func, deferring probe
>> > [    0.736375] uart-pl011 f7112000.uart: could not find pctldev for
>> > node /soc/pinmux@f7010000/uart2_pmx_func, deferring probe
>> > <hangs here>
>> >
>> > Bisecting this down, points to:
>> >   0b79842775fa ("arm64: dts: hi6220: add coresight dt nodes")
>> >
>> > Reverting this patch makes things work again.
>> >
>> > Any ideas as to what might be wrong?
>> >
>> > Please let me know if I can test any fixes or provide any additional info.
>>
>> If its helpful, the full tree I'm using (with the problematic patch
>> reverted - as well as a few other needed fixups for recent
>> regressions) is here:
>>    https://git.linaro.org/people/john.stultz/android-dev.git
>> dev/hikey-mainline-WIP
>>
>> I'm using the "hikey_defconfig" added in the patchset to build with.
>
> Could you check if below patch has been merged into your code base?
>
> From f71a1da256962fbaac48ce0ec01600a1a43f42a1 Mon Sep 17 00:00:00 2001
> From: Leo Yan <leo.yan@xxxxxxxxxx>
> Date: Thu, 31 Aug 2017 20:36:47 +0800
> Subject: [PATCH] clk: hi6220: mark clock cs_atb_syspll as critical
>
> Clock cs_atb_syspll is pll used for coresight trace bus; when clock
> cs_atb_syspll is disabled and operates its child clock node cs_atb
> results in system hang. So mark clock cs_atb_syspll as critical to
> keep it enabled.

No I don't have this in yet. Hopefully its queued up already, and it
just landed out of order?

Adding this patch does seem to resolve the issue.

thanks
-john