Touchpad problems with latest kernels
Quoted 8 lines · On 2006-08-14 13:58:09 Luke Sharkey wrote:
While on 2054 it generally works fine, On the latest kernels (2154, 2174 etc.) I have only to e.g. open a konqueror window for the onscreen pointer to start going funny, and jerking about (As happens on computers with v. low RAM). I know its not a RAM problem, as a) everything else works fine, there is no slow down of any of the programs I run, only problems with the mouse and b) I have just upgraded from 512 MB of RAM to 1 GB.
Quoted 4 lines
If I plug in a mouse, the pointer works fine. Though I would happily use a mouse, this is often inconvenient on a laptop.
Do you have any ideas what's wrong?
This is a known problem (and fixed in Windows) with the synaptics touch pad. About one year ago I did a web search amounting to something like "synaptics rubber band" and found a fixed windows driver. But since there is no OS of that kind on this machine, I contacted the developer of the synaptics X driver.
We had a discussion (swedish only) in private mail, where I ran the driver in debug mode - he no longer had a machine with that hardware. Unfortunately I've lost the whole communication due to a voltage frying of everything in the mail machine, so can not give any details.
If Peter Österlund still has the e-mails I hereby give full permission to disclose a translated copy to anyone interested.
But I think it all came down to Peter not being able to do anything... In earlier kernels the issue _seemed_ to lessen if booting with i8042.nomux but nowadays that kernel option only gets rid of the 'lost sync' messages from the pad that turn up in /var/log/messages (Btw, excessive printing of that message Dmitry!)
Mvh
Mats Johannesson
Quoted 26 lines · Voluspa
Quoted 8 lines · On 2006-08-14 13:58:09 Luke Sharkey wrote:
While on 2054 it generally works fine, On the latest kernels (2154, 2174 etc.) I have only to e.g. open a konqueror window for the onscreen pointer to start going funny, and jerking about (As happens on computers with v. low RAM). I know its not a RAM problem, as a) everything else works fine, there is no slow down of any of the programs I run, only problems with the mouse and b) I have just upgraded from 512 MB of RAM to 1 GB.
Quoted 4 lines
If I plug in a mouse, the pointer works fine. Though I would happily use a mouse, this is often inconvenient on a laptop.
Do you have any ideas what's wrong?
This is a known problem (and fixed in Windows) with the synaptics touch pad. About one year ago I did a web search amounting to something like "synaptics rubber band" and found a fixed windows driver. But since there is no OS of that kind on this machine, I contacted the developer of the synaptics X driver.
We had a discussion (swedish only) in private mail, where I ran the driver in debug mode - he no longer had a machine with that hardware. Unfortunately I've lost the whole communication due to a voltage frying of everything in the mail machine, so can not give any details.
Quoted 9 lines
If Peter Österlund still has the e-mails I hereby give full permission to disclose a translated copy to anyone interested.
But I think it all came down to Peter not being able to do anything... In earlier kernels the issue _seemed_ to lessen if booting with i8042.nomux but nowadays that kernel option only gets rid of the 'lost sync' messages from the pad that turn up in
/var/log/messages(Btw, excessive printing of that message Dmitry!)
Hmm, have you tried lowering report rate of the synaptics driver (psmouse.rate=40)? 80 ppsm is quite high and I know some Toshibas have trouble handling the full rate...
Signature · 1 line
Dmitry
Quoted 37 lines · On Mon, 14 Aug 2006 15:27:52 -0400 Dmitry Torokhov wrote:
Quoted 26 lines · Voluspa
Quoted 8 lines · On 2006-08-14 13:58:09 Luke Sharkey wrote:
While on 2054 it generally works fine, On the latest kernels (2154, 2174 etc.) I have only to e.g. open a konqueror window for the onscreen pointer to start going funny, and jerking about (As happens on computers with v. low RAM). I know its not a RAM problem, as a) everything else works fine, there is no slow down of any of the programs I run, only problems with the mouse and b) I have just upgraded from 512 MB of RAM to 1 GB.
Quoted 4 lines
If I plug in a mouse, the pointer works fine. Though I would happily use a mouse, this is often inconvenient on a laptop.
Do you have any ideas what's wrong?
This is a known problem (and fixed in Windows) with the synaptics touch pad. About one year ago I did a web search amounting to something like "synaptics rubber band" and found a fixed windows driver. But since there is no OS of that kind on this machine, I contacted the developer of the synaptics X driver.
We had a discussion (swedish only) in private mail, where I ran the driver in debug mode - he no longer had a machine with that hardware. Unfortunately I've lost the whole communication due to a voltage frying of everything in the mail machine, so can not give any details.
Quoted 9 lines
If Peter Österlund still has the e-mails I hereby give full permission to disclose a translated copy to anyone interested.
But I think it all came down to Peter not being able to do anything... In earlier kernels the issue _seemed_ to lessen if booting with i8042.nomux but nowadays that kernel option only gets rid of the 'lost sync' messages from the pad that turn up in
/var/log/messages(Btw, excessive printing of that message Dmitry!)
Quoted 3 lines
Hmm, have you tried lowering report rate of the synaptics driver (psmouse.rate=40)? 80 ppsm is quite high and I know some Toshibas have trouble handling the full rate...
I've tried it now (without i8042.nomux) though not much progress on this Acer notebook. But first, let me expand on the original issue here.
Doing the "synaptics rubber band" web-search it is easy to locate the _very_ informative windows-info-driver-patch page. Due to copyright I can't quote enough of it to make sense (the page is good all the way to the bottom!) so here's the link about Synaptics 'supersensitive' mode:
http://www.rm.com/Support/TechnicalArticle.asp?cref=TEC360005
The debug runs I performed for Peter confirmed all the symptoms described, but the problem then becomes how to discern this crazy state as not being an intentional 'two-finger' 'hard-pressure' 'palm-touching' one (upon which some kind of reset could be issued).
Unless someone establishes a contact with people at Synaptics or disassembles the win driver, linux will stay with the loony tunes...
Ok, back to psmouse.rate=40 :
Aug 15 04:07:21 sleipner kernel: psmouse.c: TouchPad at isa0060/serio4/input0 lost sync at byte 1 Aug 15 04:07:21 sleipner last message repeated 4 times Aug 15 04:07:21 sleipner kernel: psmouse.c: issuing reconnect request
This came from 1 use of "cat /proc/acpi/thermal_zone/*/temp*" while moving an xterm at the same time by way of the touch pad. Battery readings are fine on this notebook - don't interfere with the input system (anymore? Think ACPI has improved in this area).
Here's the next "cat /proc/acpi/thermal_zone/*/temp*":
Aug 15 04:07:45 sleipner kernel: psmouse.c: TouchPad at isa0060/serio4/input0 lost sync at byte 4 Aug 15 04:07:45 sleipner kernel: psmouse.c: TouchPad at isa0060/serio4/input0 lost sync at byte 1 Aug 15 04:07:45 sleipner kernel: psmouse.c: TouchPad at
isa0060/serio4/input0 - driver resynched.
Mvh
Mats Johannesson
To all concerned,
thank you for your assistance. I sent the previous message concerning my touchpad in order to provide some feedback for the developers of the drivers.
Quoted 1 line
What kind of touchpad is this?
A Synaptics PS/2 Port Touchpad. The driver is "synaptics", according to hwbrowser.
Quoted 1 line
Can you please try vanilla kernels from kernel.org?
Regrettably, no.
Though I am the most technologically minded person I know, I am still many years from being a linux guru and should be considered a linux newbie. Compiling my own vanilla kernel is still beyond my scope as a linux user. I use the precompiled .rpm kernels provided on the fedora download website.
Secondly, as I like to upgrade my kernel as often as possible in order to recieve the latest drivers for all my hardware, frequently compiling my own kernels compared to simply installing a ready made one would be too much of a burden upon my time.
Quoted 1 line
Can you please try vanilla kernels from kernel.org?
...Have I emailed the wrong address though? Should I have contacted someone at redhat / working on the fedora project???
Quoted 1 line
In earlier kernels the issue _seemed_ to lessen if booting with i8042.nomux
Are you sure about this? I mean, I wasn't having any problem at all with the 2054 kernel (sorry to use the fedora-kernel indexing system again). It worked great.
Quoted 2 lines
Unless someone establishes a contact with people at Synaptics or disassembles the win driver, linux will stay with the loony tunes...
But the driver was ok for the 2054 kernel. Can't the driver for my touchpad be rolled back to the 2054 driver until this is fixed? Right now the onscreen pointer is practically unusuable when I so much as open a konqueror window. It frequently freezes on screen for many seconds. (When I said it was jerky, I don't mean that it pings around all over the screen).
As I mentioned above, I only wished to alert somebody to the fact that support for my touchpad was actually getting *worse*; this shouldn't really happen.
(I'd stick with the older kernel except that the newer ones have the support for my bcm43xx wireless card, which though still quite temperamental, I need.)
Thank you,
LS
_________________________________________________________________
The new Windows Live Toolbar helps you guard against viruses http://toolbar.live.com/?mkt=en-gb
Quoted 1 line · On Tue, 15 Aug 2006 15:57:05 +0100 Luke Sharkey wrote:
To all concerned,
[...]
Quoted 4 lines
Quoted 1 line
In earlier kernels the issue _seemed_ to lessen if booting with i8042.nomux
Are you sure about this? I mean, I wasn't having any problem at all with the 2054 kernel (sorry to use the fedora-kernel indexing system again). It worked great.
I'm referring back to kernels 2.6.11, .12, .13 etc. You're at 2.6.16 or thereabouts with that distro.
Quoted 7 lines
Quoted 2 lines
Unless someone establishes a contact with people at Synaptics or disassembles the win driver, linux will stay with the loony tunes...
But the driver was ok for the 2054 kernel. Can't the driver for my touchpad be rolled back to the 2054 driver until this is fixed? Right now the onscreen pointer is practically unusuable when I so much as open a konqueror window. It frequently freezes on screen for many seconds. (When I said it was jerky, I don't mean that it pings around all over the screen).
Ok, there seems to be two distinct issues here. 1) Synaptics 'supersensitive' mode, 2) pointer 'freezing'.
Number 1 can not be fixed by the kernel developers (assumingly) and not by the Synaptics X driver author. Number 2 I've seen myself, even when using i8042.nomux, but it is _very_ sporadic on this machine.
Pointer freezing would be Dmitry's domain, but he'd have to work with someone who can trigger it easily, and who can write scripts to capture debug data, since it's 'hard' to move a frozen pointer to a terminal and issue commands...
Mvh
Mats Johannesson
Quoted 6 lines · Voluspa
Pointer freezing would be Dmitry's domain, but he'd have to work with someone who can trigger it easily, and who can write scripts to capture debug data, since it's 'hard' to move a frozen pointer to a terminal and issue commands...
The trick is to have a terminal open and then do alt-tab (presuming that it does not unfreeze the pointer)...
Signature · 1 line
Dmitry
Quoted 10 lines · On Tue, 15 Aug 2006 13:12:58 -0400 Dmitry Torokhov wrote:
Quoted 6 lines · Voluspa
Pointer freezing would be Dmitry's domain, but he'd have to work with someone who can trigger it easily, and who can write scripts to capture debug data, since it's 'hard' to move a frozen pointer to a terminal and issue commands...
The trick is to have a terminal open and then do alt-tab (presuming that it does not unfreeze the pointer)...
Yeah :-) Hitting send a bit too quickly there... But still, from what I saw of the freezes, they are in the region of 5 to 10 seconds, so having a prepared script ready (maybe even tied to a key-combo) would ensure success and ease the stress.
Mvh
Mats Johannesson
Quoted 2 lines · Luke Sharkey
But the driver was ok for the 2054 kernel. Can't the driver for my touchpad be rolled back to the 2054 driver until this is fixed?
The synaptics driver itself was not changed for many month, it is the surrounding code that was changed and somehow it interferes with your box.
Quoted 5 lines
Right now the
onscreen pointer is practically unusuable when I so much as open a konqueror window. It frequently freezes on screen for many seconds. (When I said it was jerky, I don't mean that it pings around all over the screen).
Does it exibit the same behavior when you switch to text console?
Also, is there programs that poll status of your battery or monitor box's temperature? How often do they do the polling? Have you updated any of them recently?
Oh, another one... try booting with "ec_intr=0" on the kernel command line to disable embedded controller interrupt mode.
And finally, can I mples get a dmesg (or /var/log/messages) of boot with "i8042.debug=1 log_buf_len=131072" please?
Signature · 1 line
Dmitry
Hello
Quoted 2 lines
Ok, there seems to be two distinct issues here. 1) Synaptics 'supersensitive' mode, 2) pointer 'freezing'.
Hmmm... perhaps "jerky" was the wrong choice of word. I don't at all mean I've had a supersensitive touchpad: quite the opposite. What I was trying to convey was the fact that when my pointer freezes, sometimes it becomes unfrozen again suddenly (for a brief time) and then moves forward a little, then gets stuck again, and in this way is "jerky". So not supersensitivity.
(Apologies for any confusion).
Quoted 1 line
The trick is to have a terminal open and then do alt-tab (presuming that it does not unfreeze the pointer)
Yes, I've been doing that as a worthwhile precaution. Sometimes the pointer seems to get so stuck even when I plug in a normal mouse, it remains frozen.
Seeing as Linux is less easily controlled with the keyboard compared to
MS-Windows, sometimes all I can do is Alt-tab to the terminal window and do a command line reboot.
(Off topic question: why not have the K menu open when the "MS-Windows" button on the keyboard is pressed, as happens with the "Start" button on MS-windows?)
*However*, I have noticed that sometimes when I alt-tab between any windows / programs I have open, sometimes the pointer becomes unstuck again...
Quoted 1 line
Also, is there programs that poll status of your battery or monitor box's temperature?
Battery charge level, yes, but not the box's temperature. I have tried lots of apps to try and do this, e.g. gkrellm, but there is no support for this on my laptop. Besides, these problems often evidence themselves soon after I switch on, so I don't think laptop temperature would still be a problem that early. Also, I rip the occasional DVD, sometimes leaving my computer on for 5+ hours: it doesn't overheat.
Quoted 3 lines
Oh, another one... try booting with "ec_intr=0" on the kernel command line to disable embedded controller interrupt mode.
And finally, can I mples get a dmesg (or
/var/log/messages) of boot with "i8042.debug=1 log_buf_len=131072" please?
Thank you for the suggestions. I will try the kernel option, and post the results of the error log.
Yours,
LS
_________________________________________________________________
Be the first to hear what's new at MSN - sign up to our free newsletters! http://www.msn.co.uk/newsletters
Quoted 2 lines · Luke Sharkey
Seeing as Linux is less easily controlled with the keyboard compared to
MS-Windows,
Careful, you are threading dangerous waters here ;)
Quoted 8 lines
sometimes all I can do is Alt-tab to the terminal window and do a command line reboot.
(Off topic question: why not have the K menu open when the "MS-Windows" button on the keyboard is pressed, as happens with the "Start" button on MS-windows?)
It doesn't? Indeed it does not! Alt-F1 does it though and it is much harder to hid by accident (I can't count how many times I hit that key by accident on windows box and then have to grab mouse to switch focus back to the window that had it before). Anyway, it is indeed offtopic for kernel, try asking on KDE lists...
Quoted 6 lines
*However*, I have noticed that sometimes when I alt-tab between any windows / programs I have open, sometimes the pointer becomes unstuck again...
Quoted 2 lines
Also, is there programs that poll status of your battery or monitor box's temperature?
Quoted 7 lines
Battery charge level, yes, but not the box's temperature. I have tried lots of apps to try and do this, e.g. gkrellm, but there is no support for this on my laptop. Besides, these problems often evidence themselves soon after I switch on, so I don't think laptop temperature would still be a problem that early. Also, I rip the occasional DVD, sometimes leaving my computer on for 5+ hours: it doesn't overheat.
No, it is not overheating, just the act of polling battery status or temperature may cause pointer stalls on some boxes.
Signature · 1 line
Dmitry
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 10 lines · Luke Sharkey
Quoted 6 lines
Quoted 2 lines
Seeing as Linux is less easily controlled with the keyboard compared
to
Quoted 1 line
MS-Windows,
Careful, you are threading dangerous waters here ;)
Well, I wasn't trying to be inflammatory. It's just that Microsoft seems to make such a big deal of how their OS can be controlled solely from keyboard.
I guess there is a difference as to what you call controlling. You probably mean that in Windows is is easier to navigate GUI with keyboard whereas I mean that in Linux or Unix you just start a terminal program and do all necessary setup from within it (with keyboard ;) )
Quoted 4 lines
Quoted 1 line
Alt-F1 does it though
Thanks for that.
Quoted 8 lines
Quoted 2 lines
Oh, another one... try booting with "ec_intr=0" on the kernel command line to disable embedded controller interrupt mode.
I tried this. Was this meant to cause a major improvement in mouse control?
If there *was* a difference, it was only subtle. I'd have to boot in to
the kernel with and without this option a few times to see whether it truly makes a difference or not.
Well, it was just a thing to try. On some boxes interrupt mode of EC was reported to hurt mice, while on others there was no effect or even was an improvement.
Quoted 6 lines
Quoted 2 lines
And finally, can I mples get a dmesg (or
/var/log/messages) of boot with "i8042.debug=1 log_buf_len=131072" please?Yes. Here is the output of dmesg with "i8042.debug=1 log_buf_len=131072" appended to the kernel line:
Hmm, don't see anything bad happening here.. Could you please send me your /var/log/messages (still after booting with i8042.debug=1 log_buf_len=131072)? You should probably spare other people's mailboxes and send it to me directly... Or put it on FTP somewhere.
Thanks!
Signature · 1 line
Dmitry