[PATCH v5 4/4] printk: use the lockless ringbuffer

Andy Shevchenko andy.shevchenko at gmail.com
Mon Jul 20 04:22:51 EDT 2020


On Mon, Jul 20, 2020 at 9:45 AM Marco Elver <elver at google.com> wrote:
> On Sun, Jul 19, 2020 at 12:43PM +0900, Sergey Senozhatsky wrote:
> > On (20/07/18 14:10), Marco Elver wrote:
> > >
> > > It seems this causes a regression observed at least with newline-only
> > > printks. I noticed this during -next testing because various debugging
> > > tools (K*SAN, lockdep, etc.) use e.g. pr_{err,warn,info}("\n") to format
> > > reports.
> > >
> > > Without wanting to wait for a report from one of these debugging tools,
> > > a simple reproducer is below. Without this patch, the expected newline
> > > is printed.
> >
> > Empty/blank lines carry no valuable payload, could you please explain
> > why do you consider this to be a regression?
>
> Empty/blank lines are visually valuable.
>
> Did I miss a discussion somewhere that this change is acceptable?
> Unfortunately, I can't find it mentioned in the commit message, and
> therefore assumed it's a regression.
>
> As I said, a number of debugging tools use them to format reports to be
> more readable (visually separate title and report body, and separate
> parts of the report).

While I can find it useful in some cases, though messages can be
interleaved, ...

> Also, such reports are often parsed by CI systems,
> and by changing the reports, these CI systems may break. But those are
> just the usecases I'm acutely aware of -- please see a full list of
> newline-print users below.

...but this is a weak argument. If your CI relies on message rather on
the ABI, you earn the breakage.
Go and fix your CI to do sane things instead.

> Breaking the observable behaviour of a widely used interface such as
> printk doesn't seem right. Where the newline-print is inappropriate,
> wouldn't removing that newline-print be more appropriate (instead of
> forcing this behaviour on everyone)?

-- 
With Best Regards,
Andy Shevchenko



More information about the kexec mailing list