Re: [PATCH 2/2] Subject: printk: Don't trap random context in infinite log_buf flush

[Sergey Senozhatsky]

Hi Tejun,

On (11/06/17 16:22), Tejun Heo wrote:
> Hello,
> 
> On Sat, Nov 04, 2017 at 01:24:08PM +0900, Sergey Senozhatsky wrote:
> > thanks for the patch set. we are currently looking at another approach:
> > lkml.kernel.org/r/20171102134515.6eef16de@xxxxxxxxxxxxxxxxxx
> > 
> > would you be interested in taking a look?
>  
> Hmm... It took some tweaking but the code at the end locks up the
> machine w/ Steven's patch applied and it's not that contrived a case
> (e.g. out of memory messages from packet tx/rx paths while OOM is in
> progress).

thanks!

just to make sure. there is a typo in Steven's patch:

	while (!READ_ONCE(console_waiter))

should be

	while (READ_ONCE(console_waiter))

is this the "tweaking" you are talking about?

> > there are some concerns, like a huge number of printk-s happening while
> > console_sem is locked. e.g. console_lock()/console_unlock() on one of the
> > CPUs, or console_lock(); printk(); ... printk(); console_unlock();
> 
> Unless we make all messages fully synchronous, I don't think there's a
> good solution for that and I don't think we wanna make everything
> fully synchronous.

this is where it becomes complicated. offloading logic is not binary,
unfortunately. we normally want to offload; but not always. things
like sysrq or late PM warnings, or kexec, etc. want to stay fully sync,
regardless the consequences. some of sysrq prints out even do
touch_nmi_watchdog() and touch_all_softlockup_watchdogs(). current
printk-kthread patch set tries to consider those cases and to avoid
any offloading.

	-ss