[Qemu-devel] Re: [PATCH] savevm: print migration failure to stderr rather than monitor

[Alex Williamson <alex.williamson@redhat.com>]

On Fri, 2011-01-07 at 16:46 +0100, Jan Kiszka wrote:
> Am 07.01.2011 16:39, Alex Williamson wrote:
> > On Fri, 2011-01-07 at 09:51 +0100, Jan Kiszka wrote:
> >> Am 07.01.2011 08:18, Alex Williamson wrote:
> >>> monitor_print only does anything for foreground commands, so we
> >>> don't ever see this error message in the case of a 'migrate -d'.
> >>
> >> Your change needlessly steals the error from the monitor console where
> >> it belongs if migrate is used without -d. IIRC, mon is NULL in detached
> >> mode, so only print to stderr if there is no alternative. Otherwise
> >> stick with the monitor for interactive use.
> > 
> > Indeed, mon is NULL.  That makes this an easy
> > 
> > if (mon) {
> >     monitor_printf()
> > } else {
> >     fprintf()
> > }
> > 
> > But I wonder if we should put the fprintf in the monitor_printf() path
> > so we're not just special casing this one user.  Should all
> > monitor_printfs go to stderr if there's no monitor?  Thanks,
> 
> IIRC, there are valid cased where you want to suppress status updates of
> some subsystem by handing out a NULL monitor.
> 
> If this error is critical (likely), then user error_report instead. It
> does the right thing.

Thanks, that works well.  Follow-up to come.

Alex