Re: stop_machine() soft lockup

[Peter Zijlstra]

On Wed, Sep 05, 2018 at 03:45:53PM +0200, Niklas Cassel wrote:

> However, I just managed to get this lockdep splat when booting without
> earlycon and ftrace=irqsoff
> I'm not sure that it is related, but I'm planning on looking into it anyway:

That looks like your generic printk is crap splat.

I tend to use this patch-set:

  https://lkml.kernel.org/r/20170928121823.430053219@xxxxxxxxxxxxx

and have (on x86):

  earlyprintk=serial,ttyS0,115200 force_early_printk debug ignore_loglevel

That results in a printk that does pure UART bit-banging and works from
any context without issue.