Re: [PATCH v2] serial/pmac_zilog: Remove flawed mitigation for rx irq flood

From: Michael Ellerman
Date: Thu Apr 04 2024 - 23:10:24 EST


Finn Thain <fthain@xxxxxxxxxxxxxx> writes:
> The mitigation was intended to stop the irq completely. That may be
> better than a hard lock-up but it turns out that you get a crash anyway
> if you're using pmac_zilog as a serial console:
>
> ttyPZ0: pmz: rx irq flood !
> BUG: spinlock recursion on CPU#0, swapper/0
>
> That's because the pr_err() call in pmz_receive_chars() results in
> pmz_console_write() attempting to lock a spinlock already locked in
> pmz_interrupt(). With CONFIG_DEBUG_SPINLOCK=y, this produces a fatal
> BUG splat. The spinlock in question is the one in struct uart_port.
>
> Even when it's not fatal, the serial port rx function ceases to work.
> Also, the iteration limit doesn't play nicely with QEMU, as can be
> seen in the bug report linked below.
>
> A web search for other reports of the error message "pmz: rx irq flood"
> didn't produce anything. So I don't think this code is needed any more.
> Remove it.

Yeah I think you're probably right.

I assume you have tested this on an actual pmac, as well as qemu?

cheers