Bug 112572 - PS/2 problems, keyboard lights constantly flickering
Summary: PS/2 problems, keyboard lights constantly flickering
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Red Hat Raw Hide
Classification: Retired
Component: kernel
Version: 1.0
Hardware: All
OS: Linux
medium
high
Target Milestone: ---
Assignee: Dave Jones
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2003-12-23 13:10 UTC by Tim Waugh
Modified: 2015-01-04 22:04 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2004-10-30 03:36:02 UTC
Embargoed:


Attachments (Terms of Use)
linux-kbd.patch (355 bytes, patch)
2004-03-11 13:20 UTC, Tim Waugh
no flags Details | Diff

Description Tim Waugh 2003-12-23 13:10:46 UTC
Description of problem:
When I boot a 2.6 kernel, either by installing the package on a FC1
system and rebooting or by trying to boot a boot.iso from a
development tree, no keyboard or mouse input works.

In addition, the keyboard lights are constantly flickering.  They are
mostly on, but flicker off.

Version-Release number of selected component (if applicable):
kernel-smp-2.6.0-0.1.14.i686.rpm

How reproducible:
100%

This is a dual-processor Dell Precision 420.

Comment 1 Cokey de Percin 2003-12-24 20:53:24 UTC
I'm having problems using a ps2 keyboard with the just released 2.6 
kernel and Core 1.0.  I downloaded, configured, compiled & installed 
the released 2.6.0 kernel (kernel-source-2.6.0-0.1.14.i386.rpm) and 
everything went smoothly until I tried to login from the 1st console 
(X is turned off locally) and the keyboard didn't work.  Thinking I 
had done something stupid when I rolled by own kernel, I pulled down 
a prebuilt kernel (kernel-2.6.0-0.1.14.i686.rpm) and installed it.  
No change.  The situation is I can telnet in or use remote X and 
everything works as it should but when I try to use the local ps2 
keyboard in the default text console, the system doesn't seem to 
recognize it.  I can boot with the original kernel (2.4.22-1.2115.
nptl) and the keyboard works correctly in the text consoles.

The hardware is an elderly UP ASUS P3B-F with 512M & 300Mhz 
processor.  It has USB, but no devices.  No ATA, SCSI disk & CDROM 
hung off an Adaptec2940 Ultra2 SCSI adapter.  100% reproducible.

Comment 2 Sammy 2003-12-31 14:27:10 UTC
I have the same problem. I have an usb mouse and a ps/2 
keyboard on a relatively new system. I'll keep investigating 
to see if there is a solution. 

Comment 3 Cokey de Percin 2004-01-01 14:24:27 UTC
Just tried an unmodified 2.6.1-rc1 (full kernel, not patched), no joy.
 Console still not resposive.

Comment 4 Tim Waugh 2004-01-08 15:21:45 UTC
I tried playing with these kernel options:

acpi
apm
atkbd_reset
i8042_noaux
i8042_reset
i8042_unlock
i8042_direct
i8042_dumbkbd

None seemed to have any effect.

Comment 5 Dave Jones 2004-01-08 16:26:33 UTC
Still a problem with -30 kernel in the development tree ?

Comment 6 Tim Waugh 2004-01-08 17:13:34 UTC
I'm afraid so.

Comment 7 Sammy 2004-01-14 21:05:00 UTC
Same struggle here....I get a message in /var/log/messages that say: 
 
i8042.c: Can't read CTR while initializing i8042 
 
In the mean time I have recompiled 2.6.1 arjanv latest with few 
unrelated patches (mpt fusion, alsa 101 etc.) with everything compiled 
into the kernel and this works without a glitch. I can' tell the difference 
since 8042 stuff seems to be compiled in for the -41 kernel too. 
 
I just saw this patch in the -mm3 kernel: 
 
input-01-i8042-suspend.patch 
 
I'll compile -41 with this and see if it helps. 
 

Comment 8 Sammy 2004-01-15 15:28:08 UTC
Still there with -43 and the one that has the above patch 
compiled in. 
 
Since I am running 2.6.1+patches perfectly fine with 
the same SERIO options compiled in the problem 
seems to be not what is turned on but what is turned 
OFF. In my custom compilation I turned of everything 
that I do not use (acpi etc). 

Comment 9 Cokey de Percin 2004-01-17 02:28:07 UTC
Just tried 2.6.1-mm4; no joy

Comment 10 Cokey de Percin 2004-02-26 00:59:27 UTC
2.6.3-1.96 Fedora kernel source w/custom options/compile seems to have
solved my problems.  Keyboard/mouse working as they should.

Comment 11 Tim Waugh 2004-02-26 10:45:10 UTC
Which options?  The 2.6.3-1.96smp i686 package does not fix the
problem for me; neither does 2.6.3-1.109smp.

Comment 12 Cokey de Percin 2004-02-27 00:21:45 UTC
I should have been clearer.  I downloaded and installed the kernel
source (kernel-source-2.6.3-1.96.i386.rpm) and then configured from
scratch and compiled.  Note that my machine is not SMP; don't see why
that should make a difference but there's no way to tell.  Note also
that I'm using gcc32.  We may have totally different problems with
similar symptoms.

Comment 13 Mike A. Harris 2004-02-27 03:29:06 UTC
Just saying you recompiled the kernel "with custom options" and the
problem went away, doesn't help us to know which specific options
you changed which may have affected the problem however.  Can you
be more specific?

Comment 14 Mike A. Harris 2004-02-27 07:46:48 UTC
Be sure to upgrade to XFree86 4.3.0-60 in rawhide also if you haven't
already.  It fixes the atkbd.c related bug.

Comment 15 Tim Waugh 2004-02-27 09:23:22 UTC
Mike, I can guarantee this is nothing to do with X.  The keyboard is
unresponsive before init even starts -- in fact, right at the point
where the 'serio' lines appear.

Comment 16 Mike A. Harris 2004-02-27 17:46:07 UTC
Ok, thanks Tim

Comment 17 Tim Waugh 2004-03-08 15:32:04 UTC
fdepercin.com: could you at least attach your config file so
that we can look at the differences between what you use (which works)
and what we use (which doesn't)?

Comment 18 Tim Waugh 2004-03-10 22:24:34 UTC
So switching keyboards to the one that actually came with this machine
makes the problem go away.

What make of keyboard do other people see this problem with?  The one
I'm using is a "Microsoft Natural Keyboard Elite".

Comment 19 Tim Waugh 2004-03-11 09:42:28 UTC
I rebuilt atkbd with debugging on (and some extra printks for
atkbd_connect/reconnect/disconnect), and here is the output from a
Dell keyboard (works fine):

serio: i8042 AUX port at 0x60,0x64 irq 12
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fe flags 01
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
serio: i8042 KBD port at 0x60,0x64 irq 1
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Received ab flags 00
atkbd.c: Received 41 flags 00
atkbd.c: Sent: ed
atkbd.c: Received fa flags 00
atkbd.c: Sent: 00
atkbd.c: Received fa flags 00
atkbd.c: Sent: f3
atkbd.c: Received fa flags 00
atkbd.c: Sent: 00
atkbd.c: Received fa flags 00
atkbd.c: Sent: f4
atkbd.c: Received fa flags 00
input: AT Translated Set 2 keyboard on isa0060/serio0



But here is the output using the Microsoft keyboard (doesn't work):

serio: i8042 AUX port at 0x60,0x64 irq 12
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fe flags 01
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
serio: i8042 KBD port at 0x60,0x64 irq 1
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
[...]
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
[...]
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Received ab flags 00
atkbd.c: Received ff flags 01
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
input: AT Translated Set 2 keyboard on isa0060/serio0
[...]
atkbd.c: Received aa flags 00
atkbd: disconnected
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Received ff flags 01
atkbd: connecting
atkbd.c: Sent: f2
atkbd.c: Received fa flags 00
atkbd.c: Sent: ed
atkbd.c: Received fe flags 01
(sequence continues)

Comment 20 Tim Waugh 2004-03-11 13:20:50 UTC
Created attachment 98457 [details]
linux-kbd.patch

This patch works around the problem for me.

Comment 21 Tim Waugh 2004-03-27 18:20:13 UTC
So can we add this patch?

Comment 22 Dave Jones 2004-03-29 00:30:25 UTC
Please report this to the input maintainer vojtech.  Touching
this code makes me nervous as a lot of the time its a case of "fixes
for person A" whilst "breaks person B's setup".

This patch doesn't strike me as a real fix, but a workaround.


Comment 23 Vojtech Pavlik 2004-03-29 11:48:56 UTC
Can I get the whole sequence (without bits snipped out) for both
the working and non-working case?

It seems to me that the Microsoft keyboard doesn't accept the Set LED
command, which would be highly unusual.

Comment 24 Tim Waugh 2004-03-30 10:19:02 UTC
As I recall I only cut out irrelevant printks (from other parts of the
kernel).

The working trace appears in full, as it was in dmesg.  The
non-working case is unending.

Comment 25 Tim Waugh 2004-07-26 15:48:15 UTC
I no longer have access to this hardware.


Note You need to log in before you can comment on or make changes to this bug.