[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87h7cqg0xk.fsf@jogness.linutronix.de>
Date: Fri, 05 Nov 2021 15:03:27 +0106
From: John Ogness <john.ogness@...utronix.de>
To: Nicholas Piggin <npiggin@...il.com>
Cc: Laurent Dufour <ldufour@...ux.ibm.com>,
linux-kernel@...r.kernel.org, Petr Mladek <pmladek@...e.com>
Subject: Re: Removal of printk safe buffers delays NMI context printk
On 2021-11-05, Nicholas Piggin <npiggin@...il.com> wrote:
>> What was removed from 93d102f094b was irq_work triggering on all
>> CPUs.
>
> No, it was the caller executing the flush for all remote CPUs itself.
> irq work was not involved (and irq work can't be raised in a remote
> CPU from NMI context).
Maybe I am missing something. In 93d102f094b~1 I see:
watchdog_smp_panic
printk_safe_flush
__printk_safe_flush
printk_safe_flush_buffer
printk_safe_flush_line
printk_deferred
vprintk_deferred
vprintk_emit (but no direct printing)
defer_console_output
irq_work_queue
AFAICT, using defer_console_output() instead of your new printk_flush()
should cause the exact behavior as before.
> but we do need that printk flush capability back there and for
> nmi_backtrace.
Agreed. I had not considered this necessary side-effect when I removed
the NMI safe buffers.
I am just wondering if we should fix the regression by going back to
using irq_work (such as defer_console_output()) or if we want to
introduce something new that introduces direct printing.
John Ogness
Powered by blists - more mailing lists