Re: [PATCH] printk/kdb: Redirect printk messages into kdb in any context

From: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
To: Daniel Thompson <daniel.thompson@linaro.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>,
	Petr Mladek <pmladek@suse.com>,
	Sumit Garg <sumit.garg@linaro.org>,
	Jason Wessel <jason.wessel@windriver.com>,
	Douglas Anderson <dianders@chromium.org>,
	kgdb-bugreport@lists.sourceforge.net,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Arnd Bergmann <arnd@arndb.de>,
	Andrew Morton <akpm@linux-foundation.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Steven Rostedt <rostedt@goodmis.org>,
	John Ogness <john.ogness@linutronix.de>
Subject: Re: [PATCH] printk/kdb: Redirect printk messages into kdb in any context
Date: Sat, 16 May 2020 01:24:11 +0900	[thread overview]
Message-ID: <20200515162411.GG42471@jagdpanzerIV.localdomain> (raw)
In-Reply-To: <20200515134806.5cw4xxnxw7k3223l@holly.lan>

On (20/05/15 14:48), Daniel Thompson wrote:
> On Fri, May 15, 2020 at 07:33:08PM +0900, Sergey Senozhatsky wrote:
> > On (20/05/15 10:50), Petr Mladek wrote:

[..]

> > Is this guaranteed that we never execute this path from NMI?
> 
> Absolutely not.
> 
> The execution context for kdb is pretty much unique...

OK, that was what I expected.

> we are running a debug mode with all CPUs parked in a holding loop with
> interrupts disabled. One CPU is at an unknown exception state and the
> others are either handling an IRQ or NMI depending on architecture[1].

Can a CPU be parked while holding the console driver port lock?

Hmm, a side note - this also means that we cannot handle it from
poll-ing console drivers and need to switch to direct console writes.

> However there are a number of factors that IMHO weigh in favour of
> allowing kdb to intercept here.
> 
> 1. kgdb/kdb are designed to work from NMI, modulo the bugs that are
>    undoubtedly present.
> 
> 2. A synchronous breakpoint (including an implicit breakpoint-on-oops)
>    from any code that executes with irqs disabled will exhibit most of
>    the same problems as an NMI but without waking up all the NMI logic.
> 
> 3. kdb_trap_printk is only set for *very* narrow time intervals by the
>    debug master (the single CPU in the system that is *not* in a
>    holding loop). Thus in all cases the system has already successfully
>    executed kdb_printf() several times before we ever call the printk()
>    interception code.
> 
>    Or put another way, even if we did tickle a bug speculated about in
>    #1, it won't be the call to printk() that triggers it; we'd never
>    get that far!

OK. I would appreciate a more detailed commit message:
- what do we fix, and what risks do we take. Just for the record.

+ a small nit: looking at for_each_console() loop -- not all consoles
can be invoked at any time and not all consoles are enabled at any time.
You _probably_ might want to do what printk does in call_console_drivers()
loop. printk also had problems with console callbacks being placed in
sections that get discarded, but that's way too niche.

	-ss