linux-kernel - Re: [resend][bug] low-probability console lockups since 5.19

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <YzYMqr5kEsKJ3cwn@spud>
Date:   Thu, 29 Sep 2022 22:22:50 +0100
From:   Conor Dooley <conor@...nel.org>
To:     John Ogness <john.ogness@...utronix.de>
Cc:     Conor Dooley <conor.dooley@...rochip.com>,
        Petr Mladek <pmladek@...e.com>,
        Thorsten Leemhuis <regressions@...mhuis.info>,
        senozhatsky@...omium.org, rostedt@...dmis.org,
        linux-kernel@...r.kernel.org, regressions@...ts.linux.dev
Subject: Re: [resend][bug] low-probability console lockups since 5.19

On Thu, Sep 29, 2022 at 04:19:03PM +0206, John Ogness wrote:
> On 2022-09-29, Conor Dooley <conor.dooley@...rochip.com> wrote:
> > I will, possibly tonight but probably not, run the bisection again
> > with the threaded printer merge reverted. Hopefully it is not filled
> > with conflicts if I go that way...
> 
> It might be easier for you to just apply the following patch to run your
> tests . This simple patch disables kthread usage.

Thanks a million John, restarted bisection from master prior to the
merge of the reverts & have ended up finding the actual issue thanks
to your hack - a RISC-V specific clocksource change. I'll close the
regzbot issue & open a new one against the actual source of my issues.

#regzbot invalid: bisection lead me up the garden path

It looks like I have an issue that's caused by the PCI controller driver
on this SoC* which triggers RCU stalls - but thanks to the clocksource
issue the RCU stalls never get reported & the system just locks up. The
system locked up in the same place on the kthreads branch so I guess I
just got unlucky with my bisection because of that.

Thanks again for the hack to bisect with,
Conor.

* likely due to running the wrong FPGA bitstream, the root port's
connection to the cpus is configured by the FPGA fabric.

> 
> John Ogness
> 
> 
> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> index b095fb5f5f61..9679887c0805 100644
> --- a/kernel/printk/printk.c
> +++ b/kernel/printk/printk.c
> @@ -3605,6 +3605,10 @@ static int __init printk_activate_kthreads(void)
>  {
>  	struct console *con;
>  
> +	/* hack: never activate kthreads */
> +	if (!printk_kthreads_available)
> +		return 0;
> +
>  	console_lock();
>  	printk_kthreads_available = true;
>  	for_each_console(con)