svn commit: r257815 - in user/ed/newcons/sys/dev: fb vt vt/hw/fb

[Doug Ambrisko]

On Tue, Nov 12, 2013 at 11:56:55PM +0200, Aleksandr Rybalko wrote:
| On Tue, 12 Nov 2013 08:59:48 -0800
| Doug Ambrisko <ambrisko at ambrisko.com> wrote:
| 
| > On Tue, Nov 12, 2013 at 12:12:16AM +0200, Aleksandr Rybalko wrote:
| > | On Mon, 11 Nov 2013 15:05:26 -0500
| > | John Baldwin <jhb at freebsd.org> wrote:
| > | 
| > | > On Thursday, November 07, 2013 4:08:53 pm Aleksandr Rybalko wrote:
| > | > > Author: ray
| > | > > Date: Thu Nov  7 21:08:52 2013
| > | > > New Revision: 257815
| > | > > URL: http://svnweb.freebsd.org/changeset/base/257815
| > | > > 
| > | > > Log:
| > | > >   Handle suspend/resume. Switch to console window before
| > | > > suspend, switch back on resume. That fix issue with broken Xorg
| > | > > image after resume. Fix some style whilst here.
| > | > 
| > | > Could you add a runtime sysctl to control this?  I haven't needed
| > | > it on my x220.  syscons had a sysctl for this.
| > | 
| > | No problem John.
| > | Committed r258023.
| > | kern.vt.suspendswitch=0 will do the trick.
| > 
| > This feature should work well for me.  I added this sysctl:
| > 
| > Index: sys/dev/vt/vt_core.c
| > ===================================================================
| > --- sys/dev/vt/vt_core.c	(revision 258052)
| > +++ sys/dev/vt/vt_core.c	(working copy)
| > @@ -108,6 +108,7 @@
| >  VT_SYSCTL_INT(debug, 0, "Newcons debug level");
| >  VT_SYSCTL_INT(deadtimer, 15, "Time to wait busy process in
| > VT_PROCESS mode"); VT_SYSCTL_INT(suspendswitch, 1, "Switch to VT0
| > before suspend"); +VT_SYSCTL_INT(splash, 1, "Display splash screen");
| >  
| >  static unsigned int vt_unit = 0;
| >  static MALLOC_DEFINE(M_VT, "vt", "vt device");
| > @@ -759,7 +760,8 @@
| >  	vt_axis_t top, left;
| >  
| >  	/* Display a nice boot splash. */
| > -	if (!(vd->vd_flags & VDF_TEXTMODE) && (boothowto & RB_MUTE))
| > {
| > +	if (!vt_splash && !(vd->vd_flags & VDF_TEXTMODE) && \
| > +	    (boothowto & RB_MUTE)) {
| >  
| >  		top = (vd->vd_height - vt_logo_height) / 2;
| >  		left = (vd->vd_width - vt_logo_width) / 2;
| > 
| > so I can see the FreeBSD boot messages when I set kern.vt.splash=0.
| > 
| > Thanks,
| > 
| > Doug A.
| 
| Hello Doug,
| 
| it looks weird for me. Splash not used now until you enable MUTE flag.
| Is it still shown for you?

Okay, maybe this is just strange.  I assumed the RB_MUTE was used
as designed but maybe it isn't.  Julian added RB_MUTE so that the
console was muted and didn't display anything at all.  So RB_MUTE
shouldn't toggle the splash screen.  I could see the muted console
be used with a splash screen.  I haven't used the syscons splash
screen in a long time (prior company in which we added muted support).
A muted console with a splash screen should display the splash screen
and no console messages until getty is started.

I have to admit that the RB_MUTE code was added to newcons after I added
my sysctl to disable the splash screen.  I should probably look at
what syscons is doing.  I was actually going to look into it
since it would be nice to enable the splash screen were I work
now in VMware offering.

FWIW, I tried to enable hw.vga.textmode and that was a bad idea.
If I remember correctly it gave me the issues when I wasn't running
newscon.
 
| It is absolutely no problem to add such sysctl, but I want to done it
| much close to how it works for syscons now, to not confuse users as
| much as possible. Just not done it yet :)

That's a good idea to make the transition smoother.  I can take a look
at syscon as well to see how that works.  I'm not sure how many people
use a muted console but I have used it at a few companies.  Those are
mostly with serial consoles.  I also have patches to the muted support
to unmute it when the machine drops into the kernel debugger or displays
a backtrace etc.  A muted console isn't very good when you are in the
debugger.

Thanks,

Doug A.