Patchwork patches to quieten the nexus7 boot messages

login
register
mail settings
Submitter Oliver Grawert
Date Jan. 25, 2013, 2:46 p.m.
Message ID <1359125208.3301.6.camel@anubis>
Download mbox | patch
Permalink /patch/215759/
State New
Headers show

Comments

Oliver Grawert - Jan. 25, 2013, 2:46 p.m.
hi,

in the nexus7 kernel a lot of hacked in third party drivers falsely use
pr_err or KERN_ERR for printing out info messages to the console ...

these messages show up before plymouth starts and give a noisy boot
impression even though our kernel logging is already at lowest prio.

to quieten that i have switched the respective functions from pr_err to
pr_info in the attached 8 patches. please apply them to the nexus7 tree.

ciao
	oli
Andy Whitcroft - Jan. 25, 2013, 5 p.m.
On Fri, Jan 25, 2013 at 03:46:48PM +0100, Oliver Grawert wrote:
> hi,
> 
> in the nexus7 kernel a lot of hacked in third party drivers falsely use
> pr_err or KERN_ERR for printing out info messages to the console ...
> 
> these messages show up before plymouth starts and give a noisy boot
> impression even though our kernel logging is already at lowest prio.
> 
> to quieten that i have switched the respective functions from pr_err to
> pr_info in the attached 8 patches. please apply them to the nexus7 tree.

As these are all at KERN_ERR, surley you can set the console down to
KERN_EMERG only and these will all be suppressed ?  Are you already
using loglevel=0 or just quiet.  In our main kernels we have the commit
below pulling the loglevel lower by default:

  commit 8464b5966d404505aa506c33c1099e6e507cdfc7
  Author: Tim Gardner <tim.gardner@canonical.com>
  Date:   Thu Oct 8 11:42:36 2009 -0600

    UBUNTU: SAUCE: (no-up) raise the default console 'quiet' level to 2

I don't see this in the n7 kernel, might that help?  We could even move
it lower if appropriate.

-apw
Oliver Grawert - Jan. 25, 2013, 5:14 p.m.
Am Freitag, den 25.01.2013, 17:00 +0000 schrieb Andy Whitcroft:
> On Fri, Jan 25, 2013 at 03:46:48PM +0100, Oliver Grawert wrote:
> > hi,
> > 
> > in the nexus7 kernel a lot of hacked in third party drivers falsely use
> > pr_err or KERN_ERR for printing out info messages to the console ...
> > 
> > these messages show up before plymouth starts and give a noisy boot
> > impression even though our kernel logging is already at lowest prio.
> > 
> > to quieten that i have switched the respective functions from pr_err to
> > pr_info in the attached 8 patches. please apply them to the nexus7 tree.
> 
> As these are all at KERN_ERR, surley you can set the console down to
> KERN_EMERG only and these will all be suppressed ?  Are you already
> using loglevel=0 or just quiet.  In our main kernels we have the commit
> below pulling the loglevel lower by default:
> 
>   commit 8464b5966d404505aa506c33c1099e6e507cdfc7
>   Author: Tim Gardner <tim.gardner@canonical.com>
>   Date:   Thu Oct 8 11:42:36 2009 -0600
> 
>     UBUNTU: SAUCE: (no-up) raise the default console 'quiet' level to 2
> 
> I don't see this in the n7 kernel, might that help?  We could even move
> it lower if appropriate.
the n7 kernel definitely had the patch that is also used in ubuntu to
set console_loglevel=1 in init/main.c, 

i also know that this didnt prevent the pr_err messages from being
printed to console in our quantal image... 

looking at the code now, it seems the patch was lost during the re-base
the kernel team did, i will try with loglevel=1 again and see which
messages persist to be printed on screen ...


ciao
	oli
Oliver Grawert - Jan. 25, 2013, 5:28 p.m.
hi,
Am Freitag, den 25.01.2013, 18:14 +0100 schrieb Oliver Grawert:
> Am Freitag, den 25.01.2013, 17:00 +0000 schrieb Andy Whitcroft:
> > On Fri, Jan 25, 2013 at 03:46:48PM +0100, Oliver Grawert wrote:
> > > hi,
> > > 
> > > in the nexus7 kernel a lot of hacked in third party drivers falsely use
> > > pr_err or KERN_ERR for printing out info messages to the console ...
> > > 
> > > these messages show up before plymouth starts and give a noisy boot
> > > impression even though our kernel logging is already at lowest prio.
> > > 
> > > to quieten that i have switched the respective functions from pr_err to
> > > pr_info in the attached 8 patches. please apply them to the nexus7 tree.
> > 
> > As these are all at KERN_ERR, surley you can set the console down to
> > KERN_EMERG only and these will all be suppressed ?  Are you already
> > using loglevel=0 or just quiet.  In our main kernels we have the commit
> > below pulling the loglevel lower by default:
> > 
> >   commit 8464b5966d404505aa506c33c1099e6e507cdfc7
> >   Author: Tim Gardner <tim.gardner@canonical.com>
> >   Date:   Thu Oct 8 11:42:36 2009 -0600
> > 
> >     UBUNTU: SAUCE: (no-up) raise the default console 'quiet' level to 2
> > 
> > I don't see this in the n7 kernel, might that help?  We could even move
> > it lower if appropriate.
> the n7 kernel definitely had the patch that is also used in ubuntu to
> set console_loglevel=1 in init/main.c, 
> 
> i also know that this didnt prevent the pr_err messages from being
> printed to console in our quantal image... 
> 
> looking at the code now, it seems the patch was lost during the re-base
> the kernel team did, i will try with loglevel=1 again and see which
> messages persist to be printed on screen ...

just tested, setting console_loglevel=1 in init/main.c gets me a
relatively quiet boot. what still shows up is a bunch of "Powering on
wifi" Powering off wifi" nonsense on teh screen, so please set the
loglevel to 1 and apply quieten_wifi_power_message.patch from the above
set..

ciao
	oli
Andy Whitcroft - Jan. 29, 2013, 9:25 a.m.
I am pulling the missing patch from master and testing the result.

-apw

Patch

diff -ruN linux-nexus7-3.1.10.orig/arch/arm/mach-tegra/board-grouper-sdhci.c linux-nexus7-3.1.10/arch/arm/mach-tegra/board-grouper-sdhci.c

--- linux-nexus7-3.1.10.orig/arch/arm/mach-tegra/board-grouper-sdhci.c	2012-12-14 17:16:54.000000000 +0100

+++ linux-nexus7-3.1.10/arch/arm/mach-tegra/board-grouper-sdhci.c	2013-01-25 14:25:03.111360213 +0100

@@ -192,7 +192,7 @@ 

 
 static int grouper_wifi_power(int on)
 {
-	pr_err("Powering %s wifi\n", (on ? "on" : "off"));

+	pr_info("Powering %s wifi\n", (on ? "on" : "off"));

 
 	gpio_set_value(GROUPER_WLAN_PWR, on);
 	mdelay(100);