[CRIU] [PATCH 0/5] log code update

[Cyrill Gorcunov]

On Sun, Jan 13, 2013 at 12:04:31AM +0400, Cyrill Gorcunov wrote:
> On Sat, Jan 12, 2013 at 11:54:21PM +0400, Andrew Vagin wrote:
> > On Sat, Jan 12, 2013 at 11:37:11PM +0400, Cyrill Gorcunov wrote:
> > > On Sat, Jan 12, 2013 at 11:29:22PM +0400, Andrew Vagin wrote:
> > > > > > 
> > > > > > Where may we need to use pr_err without \n?
> > > > > 
> > > > > Adding implicit \n means you can't print continuations. Actually with
> > > > > timestapm concept introduced this ability has been a bit ruined (which I
> > > > > still have plans to fix one day). But anyway, what we need is to be able
> > > > > to produce correct continuations like
> > > > > 
> > > > > 	pr_info("Dumping something... ");
> > > > > 	... some code here ...
> > > > 
> > > > In this moment another thread prints a message...
> > > > How are you going to synchronise processes?
> > > 
> > > As usual -- with mutex.
> > And all other processes will wait while code is executing? Looks like a
> > big kernel (log) lock...
> 
> If we print per-pid log this won't be a problem, if the log is unified,
> then sure the other writers are to wait a bit but at moment our dumper
> is not threaded and secondly there are a number of ways to eliminate
> waiting problem (say buffered output and such).

In ideal case I guess the "info" output should be like

	Collecting tasks...		OK
	Dumping files...		OK
	Dumping registers		OK
	...
	Killing dumpee			OK

while on debug level there should be all the details we prints now.

	Cyrill