Hello,
I haven't really tried a 2.5 kernel for a very long time. I used some of
the earliest, but then I suddenly had problems booting one of them, and
haven't really taken much effort in getting it boot lately.
But now I decided I should try again. I got 2.5.38 booted after some
initial trouble. But, I have a couple of weird problems. First, the
mouse. I have a Logitech Cordless Optical mouse. With kernel 2.4.x I use
MouseManPlusPS/2 as the XFree mouse-driver. Then I can use the wheel and
the fourth button just as expected. But with kernel 2.5.38 neither the
wheel or the fourth button works. I change protocol to IMPS/2 in XFree,
and everything works like expected, but the fourth button works just
like pussing the wheel (third button). This is excactly the same
behavior as with 2.4.20-pre7 (that's why I use MouseManPlusPS/2). Anyone
have a clue why this doesn't work with kernel 2.5.38?
Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
it's dead. I have a Logitech cordless keyboard.
Anyone else experienced this?
Best regards,
Stian Jordet
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Thu, 26 Sep 2002 09:31, Stian Jordet wrote:
> But now I decided I should try again. I got 2.5.38 booted after some
> initial trouble. But, I have a couple of weird problems. First, the
> mouse. I have a Logitech Cordless Optical mouse. With kernel 2.4.x I use
> MouseManPlusPS/2 as the XFree mouse-driver. Then I can use the wheel and
> the fourth button just as expected. But with kernel 2.5.38 neither the
> wheel or the fourth button works. I change protocol to IMPS/2 in XFree,
> and everything works like expected, but the fourth button works just
> like pussing the wheel (third button). This is excactly the same
> behavior as with 2.4.20-pre7 (that's why I use MouseManPlusPS/2). Anyone
> have a clue why this doesn't work with kernel 2.5.38?
Input support was merged, that fundamentally changes the way input handling
works. The new input layer mousedev handler tries to guess which mode you
want. Maybe you want the explorer PS/2 protocol? Or wait for X to get a nice
event input driver.
> Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
> out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
> it's dead. I have a Logitech cordless keyboard.
I'm using a logitech cordless too. However mine is on USB, and I guess you are
using some PS/2 connector? USB is much better tested, so there may be some
bugs.
Brad
- --
http://conf.linux.org.au. 22-25Jan2003. Perth, Aust. Tickets booked.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org
iD8DBQE9kkr2W6pHgIdAuOMRAhibAJ9ZJJoCQCvOExxTxQFYZvfN91mp3QCeJNDC
WT/qGY0Dj7XogoVMJZ4jywo=
=YG6j
-----END PGP SIGNATURE-----
On Thu, Sep 26, 2002 at 01:31:12AM +0200, Stian Jordet wrote:
> I haven't really tried a 2.5 kernel for a very long time. I used some of
> the earliest, but then I suddenly had problems booting one of them, and
> haven't really taken much effort in getting it boot lately.
>
> But now I decided I should try again. I got 2.5.38 booted after some
> initial trouble. But, I have a couple of weird problems. First, the
> mouse. I have a Logitech Cordless Optical mouse. With kernel 2.4.x I use
> MouseManPlusPS/2 as the XFree mouse-driver. Then I can use the wheel and
> the fourth button just as expected. But with kernel 2.5.38 neither the
> wheel or the fourth button works. I change protocol to IMPS/2 in XFree,
> and everything works like expected, but the fourth button works just
> like pussing the wheel (third button). This is excactly the same
> behavior as with 2.4.20-pre7 (that's why I use MouseManPlusPS/2). Anyone
> have a clue why this doesn't work with kernel 2.5.38?
Use ExplorerPS/2 in XFree to get access to all the buttons. The new
input drivers handle the MouseMan protocol right in the kernel, and
independently on the real mouse type (ImPS/2, ImExPS/2, GenPS/2,
Logitech PS2++ USB, Busmouse, whatever) export an ExplorerPS/2-like
virtual mouse to userspace. This is what most applications support
correctly, and is compatible with generic PS/2 and ImPS/2 protocols also.
> Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
> out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
> it's dead. I have a Logitech cordless keyboard.
>
> Anyone else experienced this?
I fixed this in about 2.5.36. Please #define ATKBD_DEBUG in
drivers/input/keyboard/atkbd.c, and send me the kernel output just
before the crash, please. I'll try to reproduce it here meanwhile.
--
Vojtech Pavlik
SuSE Labs
tor, 2002-09-26 kl. 10:58 skrev Vojtech Pavlik:
> On Thu, Sep 26, 2002 at 01:31:12AM +0200, Stian Jordet wrote:
>
> > I haven't really tried a 2.5 kernel for a very long time. I used some of
> > the earliest, but then I suddenly had problems booting one of them, and
> > haven't really taken much effort in getting it boot lately.
> >
> > But now I decided I should try again. I got 2.5.38 booted after some
> > initial trouble. But, I have a couple of weird problems. First, the
> > mouse. I have a Logitech Cordless Optical mouse. With kernel 2.4.x I use
> > MouseManPlusPS/2 as the XFree mouse-driver. Then I can use the wheel and
> > the fourth button just as expected. But with kernel 2.5.38 neither the
> > wheel or the fourth button works. I change protocol to IMPS/2 in XFree,
> > and everything works like expected, but the fourth button works just
> > like pussing the wheel (third button). This is excactly the same
> > behavior as with 2.4.20-pre7 (that's why I use MouseManPlusPS/2). Anyone
> > have a clue why this doesn't work with kernel 2.5.38?
>
> Use ExplorerPS/2 in XFree to get access to all the buttons. The new
> input drivers handle the MouseMan protocol right in the kernel, and
> independently on the real mouse type (ImPS/2, ImExPS/2, GenPS/2,
> Logitech PS2++ USB, Busmouse, whatever) export an ExplorerPS/2-like
> virtual mouse to userspace. This is what most applications support
> correctly, and is compatible with generic PS/2 and ImPS/2 protocols also.
That did the trick, thank you very much :) And I see ExplorerPS/2 works
like a charm with 2.4.x as well, so everything is just fine :) I guess
this actually was a stupid question, and I'm sorry.
>
> > Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
> > out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
> > it's dead. I have a Logitech cordless keyboard.
> >
> > Anyone else experienced this?
>
> I fixed this in about 2.5.36. Please #define ATKBD_DEBUG in
> drivers/input/keyboard/atkbd.c, and send me the kernel output just
> before the crash, please. I'll try to reproduce it here meanwhile.
I'm added #define ATKBD_DEBUG right below all the other define's in
atkbd.c, did a make clean; make dep; make bzImage and tried again. I
can't see any difference. It still just prints "input: AT Set 2 keyboard
on isa0060/serio0". I did, however, find out that if I press SHIFT+what
ever of the buttons arrows, insert, home, page up/down, delete and end,
I get just the same behaviour. It does not happen with CTRL or ALT.
Thank you very much.
Best regards,
Stian Jordet
On Thu, Sep 26, 2002 at 01:32:59PM +0200, Stian Jordet wrote:
> > > Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
> > > out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
> > > it's dead. I have a Logitech cordless keyboard.
> > >
> > > Anyone else experienced this?
> >
> > I fixed this in about 2.5.36. Please #define ATKBD_DEBUG in
> > drivers/input/keyboard/atkbd.c, and send me the kernel output just
> > before the crash, please. I'll try to reproduce it here meanwhile.
>
> I'm added #define ATKBD_DEBUG right below all the other define's in
> atkbd.c, did a make clean; make dep; make bzImage and tried again. I
> can't see any difference. It still just prints "input: AT Set 2 keyboard
> on isa0060/serio0".
Hmm, have you looked into 'dmesg'? It prints the information with
KERN_DEBUG priority, which often won't make it on the screen or into the
logs ...
> I did, however, find out that if I press SHIFT+what
> ever of the buttons arrows, insert, home, page up/down, delete and end,
> I get just the same behaviour. It does not happen with CTRL or ALT.
Can you try passing 'i8042_direct' on the kernel command line to see if
it cures the problem? It looks like your keyboard is doing some very
strange 84-key-at-emulation, stranger than others do ...
> Thank you very much.
--
Vojtech Pavlik
SuSE Labs
tor, 2002-09-26 kl. 13:37 skrev Vojtech Pavlik:
> Hmm, have you looked into 'dmesg'? It prints the information with
> KERN_DEBUG priority, which often won't make it on the screen or into the
> logs ...
>
> > I did, however, find out that if I press SHIFT+what
> > ever of the buttons arrows, insert, home, page up/down, delete and end,
> > I get just the same behaviour. It does not happen with CTRL or ALT.
>
> Can you try passing 'i8042_direct' on the kernel command line to see if
> it cures the problem? It looks like your keyboard is doing some very
> strange 84-key-at-emulation, stranger than others do ...
>
You had, ofcourse right, it was in my syslog. But the keystroke that
make my computer freeze isn't there. This is the last line:
kernel: atkbd.c: Received 1d flags 00
But I find this line several places, so it's obviously not the one
causing the crash. Altough, when passing i8042_direct to the kernel,
everything works just as expected. My keyboard is a Logitech Cordless
Desktop. The keyboard and mouse shares the same receiver, which both is
connected via ps/2.
Best regards,
Stian Jordet
On Thu, Sep 26, 2002 at 05:30:09PM +0200, Stian Jordet wrote:
> tor, 2002-09-26 kl. 13:37 skrev Vojtech Pavlik:
> > Hmm, have you looked into 'dmesg'? It prints the information with
> > KERN_DEBUG priority, which often won't make it on the screen or into the
> > logs ...
> >
> > > I did, however, find out that if I press SHIFT+what
> > > ever of the buttons arrows, insert, home, page up/down, delete and end,
> > > I get just the same behaviour. It does not happen with CTRL or ALT.
> >
> > Can you try passing 'i8042_direct' on the kernel command line to see if
> > it cures the problem? It looks like your keyboard is doing some very
> > strange 84-key-at-emulation, stranger than others do ...
> >
> You had, ofcourse right, it was in my syslog. But the keystroke that
> make my computer freeze isn't there. This is the last line:
>
> kernel: atkbd.c: Received 1d flags 00
>
> But I find this line several places, so it's obviously not the one
> causing the crash. Altough, when passing i8042_direct to the kernel,
> everything works just as expected. My keyboard is a Logitech Cordless
> Desktop. The keyboard and mouse shares the same receiver, which both is
> connected via ps/2.
Great. So, the problem is in i8042.c untranslating the keycodes. Please
also enable #define I8042_DEBUG_IO in drivers/input/serio/i8042.h, don't
start X, enable maximum console loglevel by "echo 16 16 16 16 >
/proc/sys/kernel/printk", and press the killing key combination.
(without i8042_direct, of course). Then send me the ten last or so lines
printed. This should allow me to fix the problem. Thanks.
--
Vojtech Pavlik
SuSE Labs
tor, 2002-09-26 kl. 18:57 skrev Vojtech Pavlik:
> Great. So, the problem is in i8042.c untranslating the keycodes. Please
> also enable #define I8042_DEBUG_IO in drivers/input/serio/i8042.h, don't
> start X, enable maximum console loglevel by "echo 16 16 16 16 >
> /proc/sys/kernel/printk", and press the killing key combination.
> (without i8042_direct, of course). Then send me the ten last or so lines
> printed. This should allow me to fix the problem. Thanks.
Ok, I did as prescribed. I did press SHIFT+PAGEUP, and it came quite a
lot of output, more than one page. I'm not sure if this is enough, but
it was all I could see. When I booted, I looked in the syslog, but none
of what I wrote off the screen was there. I guess that's because it
froze. Ok, here come. This is writeoff, so might be some small errors.
I don't know, but I guess I must have touched the mouse to get the four
last lines? I guess so, but I'm including them to be sure.
i8042.c: fa <- i8042 (interrupt, kbd, 1) [60519]
atkbd.c: Received fa flags 00
atkbd.c: Sent: 02
i8042.c: 02 -> i8042 (kbd-data) [60519]
i8042.c: fa <- i8042 (interrupt, kbd, 1) [60523]
atkbd.c: Received fa flags 00
atkbd.c: Sent: f0
i8042.c: f0 -> i8042 (kbd-data) [60523]
i8042.c: fa <- i8042 (interrupt, kbd, 1) [60527]
atkbd.c: Received fa flags 00
atkbd.c: Sent: 00
i8042.c: 00 -> i8042 (kbd-data) [60527]
i8042.c: fa <- i8042 (interrupt, kbd, 1) [60531]
atkbd.c: Received fa flags 00
i8042.c: 41 <- i8042 (interrupt, kbd, 1) [60532]
atkbd.c: Received 02 flags 00
input: AT Set 2 keyboard on isa0060/serio0
i8042.c: b6 <- i8042 (interrupt, kbd, 1) [60599]
atkbd.c: Received f0 flags 00
atkbd.c: Received 59 flags 00
i8042.c: 38 <- i8042 (interrupt, aux, 12) [104472]
i8042.c: fd <- i8042 (interrupt, aux, 12) [104473]
i8042.c: fe <- i8042 (interrupt, aux, 12) [104474]
i8042.c: 00 <- i8042 (interrupt, aux, 12) [104475]
I hope you get something out of this, I don't, for sure. The input:...
line is the same as I get without any debug info when it locks.
Oh well, time for some schoolwork.
Thanks.
Best regards,
Stian Jordet
On Fri, Sep 27, 2002 at 12:50:40AM +0200, Stian Jordet wrote:
> tor, 2002-09-26 kl. 18:57 skrev Vojtech Pavlik:
> > Great. So, the problem is in i8042.c untranslating the keycodes. Please
> > also enable #define I8042_DEBUG_IO in drivers/input/serio/i8042.h, don't
> > start X, enable maximum console loglevel by "echo 16 16 16 16 >
> > /proc/sys/kernel/printk", and press the killing key combination.
> > (without i8042_direct, of course). Then send me the ten last or so lines
> > printed. This should allow me to fix the problem. Thanks.
>
> Ok, I did as prescribed. I did press SHIFT+PAGEUP, and it came quite a
> lot of output, more than one page. I'm not sure if this is enough, but
> it was all I could see. When I booted, I looked in the syslog, but none
> of what I wrote off the screen was there. I guess that's because it
> froze. Ok, here come. This is writeoff, so might be some small errors.
>
> I don't know, but I guess I must have touched the mouse to get the four
> last lines? I guess so, but I'm including them to be sure.
That's right. :)
>
> i8042.c: fa <- i8042 (interrupt, kbd, 1) [60519]
> atkbd.c: Received fa flags 00
> atkbd.c: Sent: 02
> i8042.c: 02 -> i8042 (kbd-data) [60519]
> i8042.c: fa <- i8042 (interrupt, kbd, 1) [60523]
> atkbd.c: Received fa flags 00
> atkbd.c: Sent: f0
> i8042.c: f0 -> i8042 (kbd-data) [60523]
> i8042.c: fa <- i8042 (interrupt, kbd, 1) [60527]
> atkbd.c: Received fa flags 00
> atkbd.c: Sent: 00
> i8042.c: 00 -> i8042 (kbd-data) [60527]
> i8042.c: fa <- i8042 (interrupt, kbd, 1) [60531]
> atkbd.c: Received fa flags 00
> i8042.c: 41 <- i8042 (interrupt, kbd, 1) [60532]
> atkbd.c: Received 02 flags 00
> input: AT Set 2 keyboard on isa0060/serio0
> i8042.c: b6 <- i8042 (interrupt, kbd, 1) [60599]
> atkbd.c: Received f0 flags 00
> atkbd.c: Received 59 flags 00
> i8042.c: 38 <- i8042 (interrupt, aux, 12) [104472]
> i8042.c: fd <- i8042 (interrupt, aux, 12) [104473]
> i8042.c: fe <- i8042 (interrupt, aux, 12) [104474]
> i8042.c: 00 <- i8042 (interrupt, aux, 12) [104475]
>
>
> I hope you get something out of this, I don't, for sure. The input:...
> line is the same as I get without any debug info when it locks.
>
> Oh well, time for some schoolwork.
Unfortunately the little bit of information I needed scrolled away
already. Can you try with the other shift (right?)? That won't
probably crash your machine, but will most likely generate an "Unknown
scancode" message. Again, send me the log lines. This time they should
make it in the syslog well.
Thanks.
--
Vojtech Pavlik
SuSE Labs
fre, 2002-09-27 kl. 09:10 skrev Vojtech Pavlik:
> Unfortunately the little bit of information I needed scrolled away
> already. Can you try with the other shift (right?)? That won't
> probably crash your machine, but will most likely generate an "Unknown
> scancode" message. Again, send me the log lines. This time they should
> make it in the syslog well.
Ok, the combination which freezes the computer is right SHIFT, and
pageup/down, etc. Left SHIFT works just like expected. This time I first
wrote 'cp /var/log/syslog /tmp/syslog', then 'echo cut-here >
/var/log/syslog' Left-SHIFT+PAGEUP, arrow up two times, to get the cp..,
enter. Then I edited /tmp/syslog and copied only what was after
"cut-here". So the keystrokes included here should be Left-SHIFT+PAGEUP,
ARROW-UP, ARROW-UP, ENTER. If this works like I think it should. As you
can see, it did not generate an "Unknown scancode"...
Sep 27 13:37:37 chevrolet kernel: i8042.c: 9c <- i8042 (interrupt, kbd, 1) [317387]
Sep 27 13:37:37 chevrolet kernel: atkbd.c: Received f0 flags 00
Sep 27 13:37:37 chevrolet kernel: atkbd.c: Received 5a flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: 2a <- i8042 (interrupt, kbd, 1) [319722]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received 12 flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [319872]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: aa <- i8042 (interrupt, kbd, 1) [319879]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received f0 flags 00
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received 12 flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [319882]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: 49 <- i8042 (interrupt, kbd, 1) [319885]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received 7d flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [319948]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: c9 <- i8042 (interrupt, kbd, 1) [319955]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received f0 flags 00
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received 7d flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [319958]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:39 chevrolet kernel: i8042.c: 2a <- i8042 (interrupt, kbd, 1) [319961]
Sep 27 13:37:39 chevrolet kernel: atkbd.c: Received 12 flags 00
Sep 27 13:37:40 chevrolet kernel: i8042.c: aa <- i8042 (interrupt, kbd, 1) [320061]
Sep 27 13:37:40 chevrolet kernel: atkbd.c: Received f0 flags 00
Sep 27 13:37:40 chevrolet kernel: atkbd.c: Received 12 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [322157]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: 48 <- i8042 (interrupt, kbd, 1) [322160]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received 75 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [322274]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: c8 <- i8042 (interrupt, kbd, 1) [322280]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received f0 flags 00
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received 75 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [322649]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: 48 <- i8042 (interrupt, kbd, 1) [322653]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received 75 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: e0 <- i8042 (interrupt, kbd, 1) [322756]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received e0 flags 00
Sep 27 13:37:42 chevrolet kernel: i8042.c: c8 <- i8042 (interrupt, kbd, 1) [322762]
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received f0 flags 00
Sep 27 13:37:42 chevrolet kernel: atkbd.c: Received 75 flags 00
Sep 27 13:37:45 chevrolet kernel: i8042.c: 1c <- i8042 (interrupt, kbd, 1) [325706]
Sep 27 13:37:45 chevrolet kernel: atkbd.c: Received 5a flags 00
Hope this helps, even though I doubt. It's only the right-SHIFT causing troubles.
Thanks.
Best regards,
Stian Jordet
fre, 2002-09-27 kl. 09:10 skrev Vojtech Pavlik:
> Unfortunately the little bit of information I needed scrolled away
> already. Can you try with the other shift (right?)? That won't
> probably crash your machine, but will most likely generate an "Unknown
> scancode" message. Again, send me the log lines. This time they should
> make it in the syslog well.
I guess this didn't help you either. But this night I got a bright idea,
or so I think, atleast. I'm actually using a KVM-switch, which I
ofcourse should have told you earlier, but it wasn't in my mind at all.
When not using the KVM switch, I get this message:
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
Will this help?
Thanks.
Best regards,
Stian Jordet
On Fri, Sep 27, 2002 at 01:51:43PM +0200, Stian Jordet wrote:
> fre, 2002-09-27 kl. 09:10 skrev Vojtech Pavlik:
> > Unfortunately the little bit of information I needed scrolled away
> > already. Can you try with the other shift (right?)? That won't
> > probably crash your machine, but will most likely generate an "Unknown
> > scancode" message. Again, send me the log lines. This time they should
> > make it in the syslog well.
> Ok, the combination which freezes the computer is right SHIFT, and
> pageup/down, etc. Left SHIFT works just like expected. This time I first
> wrote 'cp /var/log/syslog /tmp/syslog', then 'echo cut-here >
> /var/log/syslog' Left-SHIFT+PAGEUP, arrow up two times, to get the cp..,
> enter. Then I edited /tmp/syslog and copied only what was after
> "cut-here". So the keystrokes included here should be Left-SHIFT+PAGEUP,
> ARROW-UP, ARROW-UP, ENTER. If this works like I think it should. As you
> can see, it did not generate an "Unknown scancode"...
Can you test whether the attached patch helps?
--
Vojtech Pavlik
SuSE Labs
man, 2002-10-07 kl. 15:00 skrev Vojtech Pavlik:
> On Fri, Sep 27, 2002 at 01:51:43PM +0200, Stian Jordet wrote:
> > fre, 2002-09-27 kl. 09:10 skrev Vojtech Pavlik:
> > > Unfortunately the little bit of information I needed scrolled away
> > > already. Can you try with the other shift (right?)? That won't
> > > probably crash your machine, but will most likely generate an "Unknown
> > > scancode" message. Again, send me the log lines. This time they should
> > > make it in the syslog well.
> > Ok, the combination which freezes the computer is right SHIFT, and
> > pageup/down, etc. Left SHIFT works just like expected. This time I first
> > wrote 'cp /var/log/syslog /tmp/syslog', then 'echo cut-here >
> > /var/log/syslog' Left-SHIFT+PAGEUP, arrow up two times, to get the cp..,
> > enter. Then I edited /tmp/syslog and copied only what was after
> > "cut-here". So the keystrokes included here should be Left-SHIFT+PAGEUP,
> > ARROW-UP, ARROW-UP, ENTER. If this works like I think it should. As you
> > can see, it did not generate an "Unknown scancode"...
>
> Can you test whether the attached patch helps?
I was starting to think you had forgot me :)
The patch helped a lot. Now it doesn't crash at all. But when I use
RIGHT-ALT+PAGE-UP, I get these errors a couple of times, then it
suddenly works as it should.
atkbd.c: Unknown key (set 2, scancode 0x1b6, on isa0060/serio0) pressed.
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
input: AT Set 2 keyboard on isa0060/serio0
atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
I think the first key is LEFT-SHIFT+PAGE-UP, and the rest the same with
right-shift. Anyway, after those first errors, it works perfect. No
freeze like before. Thanks :) Great work :)
Best regards,
Stian Jordet
On Mon, Oct 07, 2002 at 09:55:10PM +0200, Stian Jordet wrote:
> man, 2002-10-07 kl. 15:00 skrev Vojtech Pavlik:
> > On Fri, Sep 27, 2002 at 01:51:43PM +0200, Stian Jordet wrote:
> > > fre, 2002-09-27 kl. 09:10 skrev Vojtech Pavlik:
> > > > Unfortunately the little bit of information I needed scrolled away
> > > > already. Can you try with the other shift (right?)? That won't
> > > > probably crash your machine, but will most likely generate an "Unknown
> > > > scancode" message. Again, send me the log lines. This time they should
> > > > make it in the syslog well.
> > > Ok, the combination which freezes the computer is right SHIFT, and
> > > pageup/down, etc. Left SHIFT works just like expected. This time I first
> > > wrote 'cp /var/log/syslog /tmp/syslog', then 'echo cut-here >
> > > /var/log/syslog' Left-SHIFT+PAGEUP, arrow up two times, to get the cp..,
> > > enter. Then I edited /tmp/syslog and copied only what was after
> > > "cut-here". So the keystrokes included here should be Left-SHIFT+PAGEUP,
> > > ARROW-UP, ARROW-UP, ENTER. If this works like I think it should. As you
> > > can see, it did not generate an "Unknown scancode"...
> >
> > Can you test whether the attached patch helps?
>
> I was starting to think you had forgot me :)
In that case you should have reminded me of your problem. I tend to
forget when e-mails accumulate beyond a couple hundreds. ;)
> The patch helped a lot. Now it doesn't crash at all. But when I use
> RIGHT-ALT+PAGE-UP, I get these errors a couple of times, then it
> suddenly works as it should.
>
> atkbd.c: Unknown key (set 2, scancode 0x1b6, on isa0060/serio0) pressed.
> input: AT Set 2 keyboard on isa0060/serio0
> atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
> input: AT Set 2 keyboard on isa0060/serio0
> atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
> input: AT Set 2 keyboard on isa0060/serio0
> atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
> input: AT Set 2 keyboard on isa0060/serio0
> atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
> input: AT Set 2 keyboard on isa0060/serio0
> atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
> input: AT Set 2 keyboard on isa0060/serio0
> atkbd.c: Unknown key (set 2, scancode 0x1c9, on isa0060/serio0) pressed.
>
> I think the first key is LEFT-SHIFT+PAGE-UP, and the rest the same with
> right-shift. Anyway, after those first errors, it works perfect. No
> freeze like before. Thanks :) Great work :)
I still don't like this behavior - the keyboard shouldn't reinitialize.
Can you repeat the same with I8042_DEBUG_IO?
I definitely wasn't able to reproduce this here with very violent
bashing at my keyboard. And l/r alt-pageup works here just fine.
--
Vojtech Pavlik
SuSE Labs
man, 2002-10-07 kl. 22:01 skrev Vojtech Pavlik:
> On Mon, Oct 07, 2002 at 09:55:10PM +0200, Stian Jordet wrote:
[snip]
> > I was starting to think you had forgot me :)
>
> In that case you should have reminded me of your problem. I tend to
> forget when e-mails accumulate beyond a couple hundreds. ;)
I kinda did, in a polite manner. Perhaps I was to polite :)
> > The patch helped a lot. Now it doesn't crash at all. But when I use
> > RIGHT-ALT+PAGE-UP, I get these errors a couple of times, then it
> > suddenly works as it should.
[snip}
> I still don't like this behavior - the keyboard shouldn't reinitialize.
> Can you repeat the same with I8042_DEBUG_IO?
>
> I definitely wasn't able to reproduce this here with very violent
> bashing at my keyboard. And l/r alt-pageup works here just fine.
Ok, this is very weird. I tried many times that first boot, and it
wouldn't crash what so ever. Ok, it came up with some errors, but I
couldn't get it to crash. After a recompile with debug enabled,
everything went totally nuts. It didn't crash, but if I pressed S (or
what ever key) I would get a screen full of that. Everything was crazy.
But still no crash. I rebootet, still same behavior. Then I turned of
debugging, and compiled again. And this time it started crashing again
(!). I have no idea why, but now it's like it have always been. I have
tried several times, used a fresh kerneltree, and patched again, no
help. Freezes just like before. I will try some more tomorrow (I'm going
to university in five hours), but I really can't understand what have
happened.
Thanks anyway for you work :)
Best regards,
Stian Jordet
On Tue, Oct 08, 2002 at 02:20:49AM +0200, Stian Jordet wrote:
> man, 2002-10-07 kl. 22:01 skrev Vojtech Pavlik:
> > On Mon, Oct 07, 2002 at 09:55:10PM +0200, Stian Jordet wrote:
> [snip]
> > > I was starting to think you had forgot me :)
> >
> > In that case you should have reminded me of your problem. I tend to
> > forget when e-mails accumulate beyond a couple hundreds. ;)
>
> I kinda did, in a polite manner. Perhaps I was to polite :)
>
> > > The patch helped a lot. Now it doesn't crash at all. But when I use
> > > RIGHT-ALT+PAGE-UP, I get these errors a couple of times, then it
> > > suddenly works as it should.
> [snip}
> > I still don't like this behavior - the keyboard shouldn't reinitialize.
> > Can you repeat the same with I8042_DEBUG_IO?
> >
> > I definitely wasn't able to reproduce this here with very violent
> > bashing at my keyboard. And l/r alt-pageup works here just fine.
>
> Ok, this is very weird. I tried many times that first boot, and it
> wouldn't crash what so ever. Ok, it came up with some errors, but I
> couldn't get it to crash. After a recompile with debug enabled,
> everything went totally nuts. It didn't crash, but if I pressed S (or
> what ever key) I would get a screen full of that. Everything was crazy.
> But still no crash. I rebootet, still same behavior. Then I turned of
> debugging, and compiled again. And this time it started crashing again
> (!). I have no idea why, but now it's like it have always been. I have
> tried several times, used a fresh kerneltree, and patched again, no
> help. Freezes just like before. I will try some more tomorrow (I'm going
> to university in five hours), but I really can't understand what have
> happened.
>
> Thanks anyway for you work :)
I really wonder what your keyboard sends for those keys. To avoid the
freeze (and rescanning of the keyboard), you should be able to
comment-out these lines in atkbd.c:
case ATKBD_KEY_BAT:
serio_rescan(atkbd->serio);
return;
--
Vojtech Pavlik
SuSE Labs
tir, 2002-10-08 kl. 08:58 skrev Vojtech Pavlik:
>
> > But now I decided I should try again. I got 2.5.38 booted after some
> > initial trouble. But, I have a couple of weird problems. First, the
> > mouse. I have a Logitech Cordless Optical mouse. With kernel 2.4.x I use
> > MouseManPlusPS/2 as the XFree mouse-driver. Then I can use the wheel and
> > the fourth button just as expected. But with kernel 2.5.38 neither the
> > wheel or the fourth button works. I change protocol to IMPS/2 in XFree,
> > and everything works like expected, but the fourth button works just
> > like pussing the wheel (third button). This is excactly the same
> > behavior as with 2.4.20-pre7 (that's why I use MouseManPlusPS/2). Anyone
> > have a clue why this doesn't work with kernel 2.5.38?
>
> Does it work when you use "ExplorerPS/2" instead?
Uhm, you answered that 26th september, and I replied the same day.Yes,
it works perfect now :)
> > Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
> > out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
> > it's dead. I have a Logitech cordless keyboard.
>
> It should be fixed in 2.5.39 or 2.5.40. Does it work fine on 2.5.41?
It does not. I'll answer your other mail instead :)
Best regards,
Stian Jordet
tir, 2002-10-08 kl. 10:17 skrev Vojtech Pavlik:
> On Tue, Oct 08, 2002 at 02:20:49AM +0200, Stian Jordet wrote:
> > man, 2002-10-07 kl. 22:01 skrev Vojtech Pavlik:
> > > On Mon, Oct 07, 2002 at 09:55:10PM +0200, Stian Jordet wrote:
> > [snip]
> > > > I was starting to think you had forgot me :)
> > >
> > > In that case you should have reminded me of your problem. I tend to
> > > forget when e-mails accumulate beyond a couple hundreds. ;)
> >
> > I kinda did, in a polite manner. Perhaps I was to polite :)
> >
> > > > The patch helped a lot. Now it doesn't crash at all. But when I use
> > > > RIGHT-ALT+PAGE-UP, I get these errors a couple of times, then it
> > > > suddenly works as it should.
> > [snip}
> > > I still don't like this behavior - the keyboard shouldn't reinitialize.
> > > Can you repeat the same with I8042_DEBUG_IO?
> > >
> > > I definitely wasn't able to reproduce this here with very violent
> > > bashing at my keyboard. And l/r alt-pageup works here just fine.
> >
> > Ok, this is very weird. I tried many times that first boot, and it
> > wouldn't crash what so ever. Ok, it came up with some errors, but I
> > couldn't get it to crash. After a recompile with debug enabled,
> > everything went totally nuts. It didn't crash, but if I pressed S (or
> > what ever key) I would get a screen full of that. Everything was crazy.
> > But still no crash. I rebootet, still same behavior. Then I turned of
> > debugging, and compiled again. And this time it started crashing again
> > (!). I have no idea why, but now it's like it have always been. I have
> > tried several times, used a fresh kerneltree, and patched again, no
> > help. Freezes just like before. I will try some more tomorrow (I'm going
> > to university in five hours), but I really can't understand what have
> > happened.
> >
> > Thanks anyway for you work :)
>
> I really wonder what your keyboard sends for those keys. To avoid the
> freeze (and rescanning of the keyboard), you should be able to
> comment-out these lines in atkbd.c:
>
> case ATKBD_KEY_BAT:
> serio_rescan(atkbd->serio);
> return;
>
It didn't help at all. I just tried 2.5.41 instead of .40, and now I get
an Oops. Earlier it just froze. Should I write it down?
I'm starting to feel a little embarassed since I'm the only one with
this problem. It's not reallt a big deal, I can boot with i8042_direct,
and it works fine.
Thanks :)
Best regards,
Stian Jordet
On Tue, Oct 08, 2002 at 07:18:06PM +0200, Stian Jordet wrote:
> tir, 2002-10-08 kl. 10:17 skrev Vojtech Pavlik:
> > On Tue, Oct 08, 2002 at 02:20:49AM +0200, Stian Jordet wrote:
> > > man, 2002-10-07 kl. 22:01 skrev Vojtech Pavlik:
> > > > On Mon, Oct 07, 2002 at 09:55:10PM +0200, Stian Jordet wrote:
> > > [snip]
> > > > > I was starting to think you had forgot me :)
> > > >
> > > > In that case you should have reminded me of your problem. I tend to
> > > > forget when e-mails accumulate beyond a couple hundreds. ;)
> > >
> > > I kinda did, in a polite manner. Perhaps I was to polite :)
> > >
> > > > > The patch helped a lot. Now it doesn't crash at all. But when I use
> > > > > RIGHT-ALT+PAGE-UP, I get these errors a couple of times, then it
> > > > > suddenly works as it should.
> > > [snip}
> > > > I still don't like this behavior - the keyboard shouldn't reinitialize.
> > > > Can you repeat the same with I8042_DEBUG_IO?
> > > >
> > > > I definitely wasn't able to reproduce this here with very violent
> > > > bashing at my keyboard. And l/r alt-pageup works here just fine.
> > >
> > > Ok, this is very weird. I tried many times that first boot, and it
> > > wouldn't crash what so ever. Ok, it came up with some errors, but I
> > > couldn't get it to crash. After a recompile with debug enabled,
> > > everything went totally nuts. It didn't crash, but if I pressed S (or
> > > what ever key) I would get a screen full of that. Everything was crazy.
> > > But still no crash. I rebootet, still same behavior. Then I turned of
> > > debugging, and compiled again. And this time it started crashing again
> > > (!). I have no idea why, but now it's like it have always been. I have
> > > tried several times, used a fresh kerneltree, and patched again, no
> > > help. Freezes just like before. I will try some more tomorrow (I'm going
> > > to university in five hours), but I really can't understand what have
> > > happened.
> > >
> > > Thanks anyway for you work :)
> >
> > I really wonder what your keyboard sends for those keys. To avoid the
> > freeze (and rescanning of the keyboard), you should be able to
> > comment-out these lines in atkbd.c:
> >
> > case ATKBD_KEY_BAT:
> > serio_rescan(atkbd->serio);
> > return;
> >
> It didn't help at all. I just tried 2.5.41 instead of .40, and now I get
> an Oops. Earlier it just froze. Should I write it down?
Yes!
> I'm starting to feel a little embarassed since I'm the only one with
> this problem. It's not reallt a big deal, I can boot with i8042_direct,
> and it works fine.
Once 2.5 is more public, more people will be bitten by the problem you
have, so we need to fix it. Thanks a lot for your cooperation, and
please pardon my confusion about who sumbitted which problem, and if I
solved it already. :)
--
Vojtech Pavlik
SuSE Labs
On Tue, Oct 08, 2002 at 07:14:19PM +0200, Stian Jordet wrote:
> tir, 2002-10-08 kl. 08:58 skrev Vojtech Pavlik:
> >
> > > But now I decided I should try again. I got 2.5.38 booted after some
> > > initial trouble. But, I have a couple of weird problems. First, the
> > > mouse. I have a Logitech Cordless Optical mouse. With kernel 2.4.x I use
> > > MouseManPlusPS/2 as the XFree mouse-driver. Then I can use the wheel and
> > > the fourth button just as expected. But with kernel 2.5.38 neither the
> > > wheel or the fourth button works. I change protocol to IMPS/2 in XFree,
> > > and everything works like expected, but the fourth button works just
> > > like pussing the wheel (third button). This is excactly the same
> > > behavior as with 2.4.20-pre7 (that's why I use MouseManPlusPS/2). Anyone
> > > have a clue why this doesn't work with kernel 2.5.38?
> >
> > Does it work when you use "ExplorerPS/2" instead?
>
> Uhm, you answered that 26th september, and I replied the same day.Yes,
> it works perfect now :)
Sorry, found it again on the 2.5 problems kernel page. :)
> > > Second problem, if I press SHIFT+PAGEUP, my computer freezes. It spits
> > > out this message: "input: AT Set 2 keyboard on isa0060/serio0, and then
> > > it's dead. I have a Logitech cordless keyboard.
> >
> > It should be fixed in 2.5.39 or 2.5.40. Does it work fine on 2.5.41?
>
> It does not. I'll answer your other mail instead :)
:)
--
Vojtech Pavlik
SuSE Labs
tir, 2002-10-08 kl. 19:25 skrev Vojtech Pavlik:
> On Tue, Oct 08, 2002 at 07:18:06PM +0200, Stian Jordet wrote:
> > tir, 2002-10-08 kl. 10:17 skrev Vojtech Pavlik:
[snip]
> > > I really wonder what your keyboard sends for those keys. To avoid the
> > > freeze (and rescanning of the keyboard), you should be able to
> > > comment-out these lines in atkbd.c:
> > >
> > > case ATKBD_KEY_BAT:
> > > serio_rescan(atkbd->serio);
> > > return;
> > >
> > It didn't help at all. I just tried 2.5.41 instead of .40, and now I get
> > an Oops. Earlier it just froze. Should I write it down?
>
> Yes!
It won't oops anymore (?) This is really weird, I don't know what I did
last night, I was terribly tired, so that must be the reason. .41 and
your alt-alt patch, does NOT crash my computer now. No oops, nothing.
Attached is a parts of my syslog with debug enabled (and with your
alt-alt patch). I logged in, did a echo cut-here >> /var/log/syslog to
know what to send you (didn't want everyone at lkml to know my root
password, for instance). Then I did this:
Right Shift+Pageup
Right Shift+Pageup
Right Shift+Pagedown
Right Shift+Pagedown
Left Shift+Pageup
Left Shift+Pageup
Left Shift+Pagedown
Left Shift+Pagedown
Right Shift+Pageup
Right Shift+Pageup
Right Shift+Pagedown
Right Shift+Pagedown
Left Shift+Pageup
Left Shift+Pageup
Left Shift+Pagedown
Left Shift+Pagedown
ARROW-UP (to get the echo-line back)
ENTER (to make a new mark in syslog).
What I have attached is what is between those cut-here marks.
When I press the right-shift and Page up, I first get unknown scancode
0x1b6. Then, I'll get 0x1aa on every button I press untill I started to
press the left-shift instead. After that, the right shift works like it
should. I'm totally confused. But I guess you're better reading the logs
than I'm.
> > I'm starting to feel a little embarassed since I'm the only one with
> > this problem. It's not reallt a big deal, I can boot with i8042_direct,
> > and it works fine.
>
> Once 2.5 is more public, more people will be bitten by the problem you
> have, so we need to fix it. Thanks a lot for your cooperation, and
> please pardon my confusion about who sumbitted which problem, and if I
> solved it already. :)
Well, as I'm still the only one with the problem, I'm very, very
greatful you're taking the time to try to solve this :) Sorry for the
delay.
Best regards,
Stian Jordet
tir, 2002-10-08 kl. 19:25 skrev Vojtech Pavlik:
> On Tue, Oct 08, 2002 at 07:18:06PM +0200, Stian Jordet wrote:
> > tir, 2002-10-08 kl. 10:17 skrev Vojtech Pavlik:
[snip]
> > >
> > > I really wonder what your keyboard sends for those keys. To avoid the
> > > freeze (and rescanning of the keyboard), you should be able to
> > > comment-out these lines in atkbd.c:
> > >
> > > case ATKBD_KEY_BAT:
> > > serio_rescan(atkbd->serio);
> > > return;
> > >
> > It didn't help at all. I just tried 2.5.41 instead of .40, and now I get
> > an Oops. Earlier it just froze. Should I write it down?
>
> Yes!
*argh* I hate this! Just to doublecheck after I sent you the last mail
an hour or so ago, I did a recompile, fresh kernel-tree, just unrar'ed,
apply patch (just like I did earlier today), this time it Oopses again.
Every time. Earlier today, with same kernel, never any Oops. This is
incredibly weird. Anyway, here comes the Oops. I had to write it down by
hand, so it _might_ be some errors, but I really don't think so, I did
my best, and I even understood my notes when I wrote this textfile :)
Best regards,
Stian Jordet
ons, 2002-10-09 kl. 08:22 skrev Vojtech Pavlik:
> On Wed, Oct 09, 2002 at 01:48:11AM +0200, Stian Jordet wrote:
> > tir, 2002-10-08 kl. 19:25 skrev Vojtech Pavlik:
> > > On Tue, Oct 08, 2002 at 07:18:06PM +0200, Stian Jordet wrote:
> > > > tir, 2002-10-08 kl. 10:17 skrev Vojtech Pavlik:
> > [snip]
> > > > > I really wonder what your keyboard sends for those keys. To avoid the
> > > > > freeze (and rescanning of the keyboard), you should be able to
> > > > > comment-out these lines in atkbd.c:
> > > > >
> > > > > case ATKBD_KEY_BAT:
> > > > > serio_rescan(atkbd->serio);
> > > > > return;
> > > > >
> > > > It didn't help at all. I just tried 2.5.41 instead of .40, and now I get
> > > > an Oops. Earlier it just froze. Should I write it down?
> > >
> > > Yes!
> > It won't oops anymore (?) This is really weird, I don't know what I did
> > last night, I was terribly tired, so that must be the reason. .41 and
> > your alt-alt patch, does NOT crash my computer now. No oops, nothing.
> > Attached is a parts of my syslog with debug enabled (and with your
> > alt-alt patch). I logged in, did a echo cut-here >> /var/log/syslog to
> > know what to send you (didn't want everyone at lkml to know my root
> > password, for instance). Then I did this:
> >
> > Right Shift+Pageup
> > Right Shift+Pageup
> > Right Shift+Pagedown
> > Right Shift+Pagedown
> > Left Shift+Pageup
> > Left Shift+Pageup
> > Left Shift+Pagedown
> > Left Shift+Pagedown
> > Right Shift+Pageup
> > Right Shift+Pageup
> > Right Shift+Pagedown
> > Right Shift+Pagedown
> > Left Shift+Pageup
> > Left Shift+Pageup
> > Left Shift+Pagedown
> > Left Shift+Pagedown
> >
> > ARROW-UP (to get the echo-line back)
> > ENTER (to make a new mark in syslog).
> >
> > What I have attached is what is between those cut-here marks.
> >
> > When I press the right-shift and Page up, I first get unknown scancode
> > 0x1b6. Then, I'll get 0x1aa on every button I press untill I started to
> > press the left-shift instead. After that, the right shift works like it
> > should. I'm totally confused. But I guess you're better reading the logs
> > than I'm.
>
> Thanks a lot for helping me here ....
>
> > Well, as I'm still the only one with the problem, I'm very, very
> > greatful you're taking the time to try to solve this :) Sorry for the
> > delay.
>
> Actually, Linus was the second who was bitten by it. :) And I managed to
> reproduce it locally yesterday as well, so it's now fixed.
Thank you very, very much. :) It works like a charm now. Maybe I'll get
to sleep the next night :) Great work, and I appreciate that you've been
taking the time sort this out.
Best regards,
Stian Jordet