Re: [RFC PATCH] printk: Change timestamp to triplet as mono, boot and real

[Petr Mladek]

On Thu 2020-08-13 10:55:00, Sergey Senozhatsky wrote:
> On (20/08/11 15:02), Petr Mladek wrote:
> > On Tue 2020-08-11 14:05:12, Thomas Gleixner wrote:
> > > Petr Mladek <pmladek@xxxxxxxx> writes:
> > > > At least "crash" tool would need an update anyway. AFAIK, it checks
> > > > the size of struct printk_log and refuses to read it when it changes.
> > > >
> > > > It means that the hack with VMCOREINFO_FIELD_OFFSET probably is not
> > > > needed because we would need to update the crashdump-related tools anyway.
> > > >
> > > > Well, the timing is good. We are about to switch the printk ring
> > > > buffer into a lockless one. It requires updating the crashdump tools
> > > > as well. We could do this at the same time. The lockless ring buffer
> > > > already is in linux-next. It is aimed for 5.10 or 5.11.
> > > ...
> > > > It would be great to synchronize all these changes changes of the
> > > > printk log buffer structures.
> > > 
> > > I agree that having one update is a good thing, but pretty please can we
> > > finally make progress with this and not create yet another dependency?
> > 
> > To make it clear. I definitely do not want to block lockless printk by
> > this.
> > 
> > BTW: I am not 100% convinced that storing all three timestamps is
> > worth it. It increases the code complexity, metadata size. It needs
> > an interface with the userspace that has to stay backward compatible.
> 
> Can we, perhaps, store those various "alternative" timestamps in dict so
> then whoever wants to read them can just parse the dict key:value pairs
> attach to each printk message?

Interesting idea. It might be a way how to add optional metadata
without breaking compatibility with crashdump tools.

Well, I have bad feeling about it. Some of the reasons might be:

  + would take more space (prefix + text vs. binary representation)
  + not reliable because dict is currently dropped when no space
  + it would make the controversial dictionary feature more important


I would prefer to solve this by storing the timestamps in the
structure with metadata.

Best Regards,
Petr