[PATCH v1 0/9] Fix Allwinner D1 boot regression

[Emil Renner Berthing]

Thomas Gleixner wrote:
> On Thu, Aug 15 2024 at 03:29, Emil Renner Berthing wrote:
> > Thomas Gleixner wrote:
> > Thanks for looking at this! Unfortunately the above patch isn't enough to fix
> > the issue:
> >
> > https://termbin.com/7sgc
> >
> > It still hangs after the "[    0.176451] cpuidle: using governor teo" message
> > until the watchdog reboots the systems.
>
> So what's puzzling is that there is a timer installed early on:
>
> [    0.000000] clocksource: riscv_clocksource: ....
>
> That same init function installs the per cpu riscv clockevent, so there
> should be a timer available.
>
> The deffered probing of the PLIC driver delays obviously the probing of
> the sun4i timer, but that should not matter when another timer is
> available. So the sun4i driver might be a red herring.
>
> Can you please add "ignore_loglevel initcall_debug" to the command line
> and provide the output of a booting and a failing kernel?

6.11-rc3 + these reverts:  https://termbin.com/q6wk
6.11-rc3 + Samuel's patch: https://termbin.com/7cgs

> And on the booting kernel please provide the output from:
>
> # cat /sys/devices/system/clockevents/clockevent0/current_device
> # cat /sys/devices/system/clockevents/broadcast/current_device
> # cat /sys/devices/system/clocksource/clocksource0/current_clocksource

On both a 6.8.6 kernel and 6.11-rc3 + reverts I get:

  # cat /sys/devices/system/clockevents/clockevent0/current_device
  sun4i_tick
  # cat /sys/devices/system/clockevents/broadcast/current_device
  riscv_timer_clockevent
  # cat /sys/devices/system/clocksource/clocksource0/current_clocksource
  riscv_clocksource

/Emil