Hi all,
A small patch here to reintroduce the dummy keyboard, which seems to have been
lost sometime - refer to my LKML thread:
http://marc.theaimsgroup.com/?l=linux-kernel&m=112885471602308&w=2
I reason that a lot of people have headless boxes (I have 7), and the dummy
keyboard option appears to be unknown, but a good addition...
I found out that CONFIG_DUMMY_KEYB is slightly broken in that although the
header text in 'dummy_keyb.c' states that this should be used with CONFIG_VT,
using this option breaks the kernel build 'as is'. This patch fixes that,
and also allows the use of dummy keyboard with or with out CONFIG_VT which
stops all the timeout warnings etc at boot on a keyboardless box. The option
to use the dummy keyboard is only available if CONFIG_INPUT_KEYBDEV is 'n'.
I apologise sending to list only, but I couldn't find a maintainer of this
area...
Nick
Signed off by Nick Warne <[email protected]>
Documentation/Configure.help | 18 ++++++++++++++++++
arch/i386/kernel/dmi_scan.c | 2 ++
drivers/char/dummy_keyb.c | 1 +
drivers/input/Config.in | 5 +++++
kernel/panic.c | 2 +-
5 files changed, 27 insertions(+), 1 deletion(-)
diff -Naur linux-2.4.31.orig/Documentation/Configure.help
linux-2.4.31n/Documentation/Configure.help
--- linux-2.4.31.orig/Documentation/Configure.help Mon Apr 4 02:42:19 2005
+++ linux-2.4.31n/Documentation/Configure.help Thu Oct 13 09:38:07 2005
@@ -15242,6 +15242,24 @@
The module will be called joydev.o. If you want to compile it as a
module, say M here and read <file:Documentation/modules.txt>.
+Dummy keyboard driver
+CONFIG_DUMMY_KEYB
+ What is this for?
+
+ Not all systems have keyboards. Some don't even have a keyboard
+ port. However, some of those systems have video support and can
+ use the virtual terminal support for display. However, the virtual
+ terminal code expects a keyboard of some kind. This driver keeps
+ the virtual terminal code happy by providing it a "keyboard", albeit
+ a very quiet one.
+
+ If you want to use the virtual terminal support but your system
+ does not support a keyboard, define CONFIG_DUMMY_KEYB along with
+ CONFIG_VT.
+
+ This can also be selected lonesome without any VT support (i.e. no
+ monitor or keyboard attached) - just define CONFIG_DUMMY_KEYB.
+
Event interface support
CONFIG_INPUT_EVDEV
Say Y here if you want your USB or ADB HID device events be
diff -Naur linux-2.4.31.orig/arch/i386/kernel/dmi_scan.c
linux-2.4.31n/arch/i386/kernel/dmi_scan.c
--- linux-2.4.31.orig/arch/i386/kernel/dmi_scan.c Wed Nov 17 11:54:21 2004
+++ linux-2.4.31n/arch/i386/kernel/dmi_scan.c Wed Oct 12 15:57:27 2005
@@ -412,11 +412,13 @@
static __init int broken_ps2_resume(struct dmi_blacklist *d)
{
#ifdef CONFIG_VT
+#ifndef CONFIG_DUMMY_KEYB
if (pm_kbd_request_override == NULL)
{
pm_kbd_request_override = pckbd_pm_resume;
printk(KERN_INFO "%s machine detected. Mousepad Resume Bug workaround
enabled.\n", d->ident);
}
+#endif
#endif
return 0;
}
diff -Naur linux-2.4.31.orig/drivers/char/dummy_keyb.c
linux-2.4.31n/drivers/char/dummy_keyb.c
--- linux-2.4.31.orig/drivers/char/dummy_keyb.c Mon Aug 25 12:44:41 2003
+++ linux-2.4.31n/drivers/char/dummy_keyb.c Wed Oct 12 15:38:48 2005
@@ -29,6 +29,7 @@
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/input.h>
+#include <asm/keyboard.h>
void kbd_leds(unsigned char leds)
{
diff -Naur linux-2.4.31.orig/drivers/input/Config.in
linux-2.4.31n/drivers/input/Config.in
--- linux-2.4.31.orig/drivers/input/Config.in Wed Feb 18 13:36:31 2004
+++ linux-2.4.31n/drivers/input/Config.in Wed Oct 12 15:16:09 2005
@@ -7,6 +7,11 @@
tristate 'Input core support' CONFIG_INPUT
dep_tristate ' Keyboard support' CONFIG_INPUT_KEYBDEV $CONFIG_INPUT
+
+if [ "$CONFIG_INPUT_KEYBDEV" == "n" ]; then
+ bool ' Use dummy keyboard driver' CONFIG_DUMMY_KEYB $CONFIG_INPUT
+fi
+
dep_tristate ' Mouse support' CONFIG_INPUT_MOUSEDEV $CONFIG_INPUT
if [ "$CONFIG_INPUT_MOUSEDEV" != "n" ]; then
int ' Horizontal screen resolution' CONFIG_INPUT_MOUSEDEV_SCREEN_X 1024
diff -Naur linux-2.4.31.orig/kernel/panic.c linux-2.4.31n/kernel/panic.c
--- linux-2.4.31.orig/kernel/panic.c Wed Nov 17 11:54:22 2004
+++ linux-2.4.31n/kernel/panic.c Wed Oct 12 16:07:56 2005
@@ -104,7 +104,7 @@
#endif
sti();
for(;;) {
-#if defined(CONFIG_X86) && defined(CONFIG_VT)
+#if defined(CONFIG_X86) && defined(CONFIG_VT) && !defined(CONFIG_DUMMY_KEYB)
extern void panic_blink(void);
panic_blink();
#endif
--
http://sourceforge.net/projects/quake2plus
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."
Nick Warne wrote:
>Hi all,
>
>A small patch here to reintroduce the dummy keyboard, which seems to have been
>lost sometime - refer to my LKML thread:
>
>http://marc.theaimsgroup.com/?l=linux-kernel&m=112885471602308&w=2
>
>
>
>
>
A serial based keyboard driver for general purpose debuggers and test
harnesses would be a whole lot more useful. Ever consider it? Snatch
the one out of kdb and submit it as a patch. The AIMS test. I remember
that from my Memorex Telex days on OS/2.
Jeff
Hi Nick,
On Thu, Oct 13, 2005 at 06:38:44PM +0100, Nick Warne wrote:
(...)
> diff -Naur linux-2.4.31.orig/arch/i386/kernel/dmi_scan.c
> linux-2.4.31n/arch/i386/kernel/dmi_scan.c
> --- linux-2.4.31.orig/arch/i386/kernel/dmi_scan.c Wed Nov 17 11:54:21 2004
> +++ linux-2.4.31n/arch/i386/kernel/dmi_scan.c Wed Oct 12 15:57:27 2005
> @@ -412,11 +412,13 @@
> static __init int broken_ps2_resume(struct dmi_blacklist *d)
> {
> #ifdef CONFIG_VT
> +#ifndef CONFIG_DUMMY_KEYB
Please could you change this to #if defined(CONFIG_VT) && !defined(CONFIG_DUMMY_KEYB) ?
Marcelo, I'd like this one to be merged, as it is useful to many of us, and
is already fixed in 2.6 (where you're not forced to have a keyboard). It's
not intrusive and at most a build fix. Would you please accept Nick's patch
after the little clean-up above ?
Thanks in advance,
Willy
>> #ifdef CONFIG_VT
>> +#ifndef CONFIG_DUMMY_KEYB
> Please could you change this to #if defined(CONFIG_VT) &&
> !defined(CONFIG_DUMMY_KEYB) ?
> Marcelo, I'd like this one to be merged, as it is useful to many of us,
> and is already fixed in 2.6 (where you're not forced to have a
> keyboard). It's not intrusive and at most a build fix. Would you please
> accept Nick's patch after the little clean-up above ?
OK, the adjusted patch attached. I did look at the ifdefs in that file
for a few minutes; but I see now I should have done it like this in the
first place.
Nick
On Fri, Oct 14, 2005 at 10:19:44AM +0100, Nick Warne wrote:
>
> >> #ifdef CONFIG_VT
> >> +#ifndef CONFIG_DUMMY_KEYB
>
> > Please could you change this to #if defined(CONFIG_VT) &&
> > !defined(CONFIG_DUMMY_KEYB) ?
>
> > Marcelo, I'd like this one to be merged, as it is useful to many of us,
> > and is already fixed in 2.6 (where you're not forced to have a
> > keyboard). It's not intrusive and at most a build fix. Would you please
> > accept Nick's patch after the little clean-up above ?
>
>
> OK, the adjusted patch attached. I did look at the ifdefs in that file
> for a few minutes; but I see now I should have done it like this in the
> first place.
Looks fine - Willy can you keep it in your tree and add it later on to the
2.4.33-pre tree?
On Tuesday 18 October 2005 09:02, Marcelo Tosatti wrote:
> > OK, the adjusted patch attached. I did look at the ifdefs in that file
> > for a few minutes; but I see now I should have done it like this in the
> > first place.
>
> Looks fine - Willy can you keep it in your tree and add it later on to the
> 2.4.33-pre tree?
Hi Marcelo,
I just noticed something - I maybe buggered up:
+
+if [ "$CONFIG_INPUT_KEYBDEV" == "n" ]; then
I think I done wrong there - should that be:
[ "$CONFIG_INPUT_KEYBDEV" = "n" ]
menuconfig all works correctly, though?
If so I will rectify the patch again.
Nick
--
http://sourceforge.net/projects/quake2plus
"Person who say it cannot be done should not interrupt person doing it."
-Chinese Proverb
On Tue, Oct 18, 2005 at 06:02:26AM -0200, Marcelo Tosatti wrote:
> On Fri, Oct 14, 2005 at 10:19:44AM +0100, Nick Warne wrote:
> >
> > >> #ifdef CONFIG_VT
> > >> +#ifndef CONFIG_DUMMY_KEYB
> >
> > > Please could you change this to #if defined(CONFIG_VT) &&
> > > !defined(CONFIG_DUMMY_KEYB) ?
> >
> > > Marcelo, I'd like this one to be merged, as it is useful to many of us,
> > > and is already fixed in 2.6 (where you're not forced to have a
> > > keyboard). It's not intrusive and at most a build fix. Would you please
> > > accept Nick's patch after the little clean-up above ?
> >
> >
> > OK, the adjusted patch attached. I did look at the ifdefs in that file
> > for a few minutes; but I see now I should have done it like this in the
> > first place.
>
> Looks fine - Willy can you keep it in your tree and add it later on to the
> 2.4.33-pre tree?
OK, fine.
Willy