Skip to contentSkip to conversation

Machine does not find AT keyboard with 2.6.13.1

DISCUSSION Roger Heflin 3 participants 4 messages Sep 15, 2005 – Sep 16, 2005 Input drivers

lore
Share & tools
lore
#0 lore raw

Hello,

I have an older machine that fails to find the AT keyboard. The machine is
based
on an Intel 7501 chipset.

Under the default fedora core 4 kernel, it found the keyboard and that keyboard
worked in UP mode, in SMP mode the machine crashed, but that is a different issue.

Under 2.6.13.1 the machine boots under SMP and does not crash, dmesg does not report
the keyboard being found, and the keyboard fails to work. The .config file does have
CONFIG_KEYBOARD_ATKBD set to Y. The keyboard was being found was seen on
the default
fedora core 4 kernel. There are no extra options on the boot cmdline.

The important messages seem to be:

Fedora Core 4 default UP kernel boot:

Sep 15 19:55:12 node001 kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
Sep 15 19:55:12 node001 kernel: serio: i8042 KBD port at 0x60,0x64 irq 1
Sep 15 19:55:12 node001 kernel: input: AT Translated Set 2 keyboard on

isa0060/serio0

New boot (2.6.13.1 smp boot)

Sep 15 21:12:35 node001 kernel: i8042.c: Can't read CTR while initializing

i8042.

The i8042 is of course missing out of /proc/interrupts on the new boot.

The .config file is copied from the original non-working smp fedora core 4 kernel,
and then "make oldconfig".

Any ideas where to look? Any patches in the git kernels that might address
this
issue?

Roger

#1 lore raw
Quoted 26 lines · Roger Heflin

I have an older machine that fails to find the AT keyboard. The machine is
based
on an Intel 7501 chipset.

Under the default fedora core 4 kernel, it found the keyboard and that keyboard
worked in UP mode, in SMP mode the machine crashed, but that is a different issue.

Under 2.6.13.1 the machine boots under SMP and does not crash, dmesg does not report
the keyboard being found, and the keyboard fails to work. The .config file does have
CONFIG_KEYBOARD_ATKBD set to Y. The keyboard was being found was seen on
the default
fedora core 4 kernel. There are no extra options on the boot cmdline.

The important messages seem to be:

Fedora Core 4 default UP kernel boot:

 Sep 15 19:55:12 node001 kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
 Sep 15 19:55:12 node001 kernel: serio: i8042 KBD port at 0x60,0x64 irq 1
 Sep 15 19:55:12 node001 kernel: input: AT Translated Set 2 keyboard on
 isa0060/serio0
Quoted 5 lines

New boot (2.6.13.1 smp boot)

 Sep 15 21:12:35 node001 kernel: i8042.c: Can't read CTR while initializing
 i8042.

The i8042 is of course missing out of /proc/interrupts on the new boot.

That I8042_CMD_CTL_WCTR write-and-test seems to be new. Can you try taking it out?

drivers/input/serio/i8042.c renamed from drivers/input/serio/i8042.c~a +0 -5
@@ -305,11 +305,6 @@ static int i8042_activate_port(struct i8
 
 	i8042_ctr |= port->irqen;
 
-	if (i8042_command(&i8042_ctr, I8042_CMD_CTL_WCTR)) {
-		i8042_ctr &= ~port->irqen;
-		return -1;
-	}
-
 	return 0;
 }
 

_

#2 lore raw
Quoted 27 lines · Andrew Morton
Quoted 26 lines · Roger Heflin

I have an older machine that fails to find the AT keyboard. The machine is
based
on an Intel 7501 chipset.

Under the default fedora core 4 kernel, it found the keyboard and that keyboard
worked in UP mode, in SMP mode the machine crashed, but that is a different issue.

Under 2.6.13.1 the machine boots under SMP and does not crash, dmesg does not report
the keyboard being found, and the keyboard fails to work. The .config file does have
CONFIG_KEYBOARD_ATKBD set to Y. The keyboard was being found was seen on
the default
fedora core 4 kernel. There are no extra options on the boot cmdline.

The important messages seem to be:

Fedora Core 4 default UP kernel boot:

 Sep 15 19:55:12 node001 kernel: serio: i8042 AUX port at 0x60,0x64 irq 12
 Sep 15 19:55:12 node001 kernel: serio: i8042 KBD port at 0x60,0x64 irq 1
 Sep 15 19:55:12 node001 kernel: input: AT Translated Set 2 keyboard on
 isa0060/serio0
Quoted 3 lines
Quoted 3 lines

New boot (2.6.13.1 smp boot)

 Sep 15 21:12:35 node001 kernel: i8042.c: Can't read CTR while initializing
 i8042.

"usb-handoff" on the kernel command line should solve this. I thought we already had that as a default, or was the patch to make it so dropped?

Quoted 4 lines
Quoted 1 line

The i8042 is of course missing out of /proc/interrupts on the new boot.

That I8042_CMD_CTL_WCTR write-and-test seems to be new. Can you try taking it out?

It's not a test, it's enabling the port interrupt. It won't work without it at all.

Quoted 17 lines
drivers/input/serio/i8042.c renamed from drivers/input/serio/i8042.c~a +0 -5
@@ -305,11 +305,6 @@ static int i8042_activate_port(struct i8
 
 	i8042_ctr |= port->irqen;
 
-	if (i8042_command(&i8042_ctr, I8042_CMD_CTL_WCTR)) {
-		i8042_ctr &= ~port->irqen;
-		return -1;
-	}
-
 	return 0;
 }
 

_

Signature · 2 lines
Vojtech Pavlik
SuSE Labs, SuSE CR
#3 lore raw

Vojtech,

Adding usb-handoff fixed the problem, the keyboard/mouse is found and works correctly.

I never would have guessed that a usb option would fix a ps2 issue.

Thanks
Roger

Quoted 82 lines

-----Original Message-----
From: Vojtech Pavlik [mailto:[email protected]]
Sent: Friday, September 16, 2005 6:46 AM
To: Andrew Morton
Cc: Roger Heflin; [email protected]; Dmitry Torokhov
Subject: Re: Machine does not find AT keyboard with 2.6.13.1

Quoted 3 lines · Andrew Morton
Quoted 2 lines · Roger Heflin

I have an older machine that fails to find the AT

keyboard. The machine is

Quoted 4 lines
Quoted 4 lines

based
on an Intel 7501 chipset.

Under the default fedora core 4 kernel, it found the

keyboard and

Quoted 1 line
Quoted 1 line

that keyboard worked in UP mode, in SMP mode the

machine crashed,

Quoted 4 lines
Quoted 4 lines

but that is a different issue.

Under 2.6.13.1 the machine boots under SMP and does not crash,
dmesg does not report the keyboard being found, and the

keyboard

Quoted 2 lines
Quoted 2 lines

fails to work. The .config file does have
CONFIG_KEYBOARD_ATKBD set to Y. The keyboard was being

found was seen on

Quoted 2 lines
Quoted 2 lines

the default
fedora core 4 kernel. There are no extra options on

the boot cmdline.

Quoted 5 lines
Quoted 5 lines

The important messages seem to be:

Fedora Core 4 default UP kernel boot:

 Sep 15 19:55:12 node001 kernel: serio: i8042 AUX port at 

0x60,0x64

Quoted 7 lines
Quoted 7 lines

irq 12 Sep 15 19:55:12 node001 kernel: serio: i8042 KBD port at 0x60,0x64 irq 1 Sep 15 19:55:12 node001 kernel: input: AT Translated Set 2 keyboard on isa0060/serio0

New boot (2.6.13.1 smp boot)

 Sep 15 21:12:35 node001 kernel: i8042.c: Can't read CTR while 

initializing i8042.

"usb-handoff" on the kernel command line should solve this. I thought we already had that as a default, or was the patch to make it so dropped?

Quoted 1 line
Quoted 1 line

The i8042 is of course missing out of /proc/interrupts

on the new boot.

Quoted 2 lines

That I8042_CMD_CTL_WCTR write-and-test seems to be new.

Can you try

Quoted 1 line

taking it out?

It's not a test, it's enabling the port interrupt. It won't work without it at all.

Quoted 1 line

--- devel/drivers/input/serio/i8042.c~a 2005-09-16

02:45:02.000000000 -0700

Quoted 1 line

+++ devel-akpm/drivers/input/serio/i8042.c 2005-09-16

02:46:51.000000000 -0700

Quoted 12 lines
drivers/input/serio/i8042.c renamed from drivers/input/serio/i8042.c~a +0 -5
@@ -305,11 +305,6 @@ static int i8042_activate_port(struct i8
 
 	i8042_ctr |= port->irqen;
 
-	if (i8042_command(&i8042_ctr, I8042_CMD_CTL_WCTR)) {
-		i8042_ctr &= ~port->irqen;
-		return -1;
-	}
-
 	return 0;
 }
 
Quoted 3 lines

_

--
Vojtech Pavlik
SuSE Labs, SuSE CR