|
Bugzilla – Full Text Bug Listing |
| Summary: | touchpad movement hangs keyboard sometimes | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE Linux 10.1 | Reporter: | k. t. <cooling.crystals> |
| Component: | Kernel | Assignee: | Vojtech Pavlik <vojtech> |
| Status: | RESOLVED FIXED | QA Contact: | E-mail List <qa-bugs> |
| Severity: | Normal | ||
| Priority: | P5 - None | CC: | suse-beta |
| Version: | Beta 3 | ||
| Target Milestone: | --- | ||
| Hardware: | i686 | ||
| OS: | Other | ||
| Whiteboard: | |||
| Found By: | Other | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
| Attachments: |
requested info
output of dmidecode Patch to set rate to 40 based on DMI data for Mitac laptops Patch to set rate to 40 based on DMI data for Mitac laptops dmesg |
||
|
Description
k. t.
2006-02-03 19:43:19 UTC
Is there something interesting in /var/log/messages or another logfile? Created attachment 66413 [details]
requested info
(In reply to comment #2) > Created an attachment (id=66413) [edit] > requested info > I found something interesting. I just can't imagine how can touchpad hang keyboard, itself and make kpowersave think that my battery is empty... Vojtech, please take a look The i8042 embedded controller (EC) in your notebook, which controlls the battery, powersaving, backlight, keyboard and touchpad has crashed. How often does it happen? I assume it never did happen with previous versions of SUSE Linux. Does it happen if you don't run (kill) kpowersave, thus easing the load on the EC? If it weren't a new problem, I'd suggest an EC BIOS upgrade. I know it's available for IBM notebooks, maybe Mitac provides it, too. I have done BIOS update (EC 1.02 ; SYSTEM 1.02 ; both are latest). I have had crashes before with suse 10.0: http://www.linuxquestions.org/questions/t403738.html I have had crashes with latest and older bios (then I installed 10.1 beta3 bios was already updated). I think very important is that I had hang once then I installed 10.0 (can we say that this is not linux problem if crash happens during installation ?). What is strange that all crashes I had so far happened during the touchpad movement. 'not having kpowersave running' - crashes have occured even then I'd suggest lessening the load on the EC in a different way: Set the touchpad to a lower update frequency, by setting the rate to only 40pps. This workaround is already necessary for certain Toshiba notebooks, and it's quite possible they are similar to the Mitac (who sells designs to many other manufacturers). You can do this by the kernel command line parameter psmouse.rate=40, or by using sysfs. If you use the kernel parameter, please verify that it worked using sysfs. (In reply to comment #7) > I'd suggest lessening the load on the EC in a different way: Set the touchpad > to a lower update frequency, by setting the rate to only 40pps. This workaround > is already necessary for certain Toshiba notebooks, and it's quite possible > they are similar to the Mitac (who sells designs to many other manufacturers). > > You can do this by the kernel command line parameter psmouse.rate=40, or by > using > sysfs. If you use the kernel parameter, please verify that it worked using > sysfs. > I'm using kernel parameter in boot options. How do you verify it with sysfs ? Running
cat /sys/bus/serio/devices/serio0/rate
should show "40" if everything works as expected. The default is 100.
kbeta@kohvi:~> cat /sys/bus/serio/devices/serio0/rate cat: /sys/bus/serio/devices/serio0/rate: No such file or directory kbeta@kohvi:~> ls /sys/bus/serio/devices/ serio0 serio1 serio2 serio3 serio4 kbeta@kohvi:~> cat /sys/bus/serio/devices/serio1/rate cat: /sys/bus/serio/devices/serio1/rate: No such file or directory kbeta@kohvi:~> cat /sys/bus/serio/devices/serio2/rate 40 kbeta@kohvi:~> cat /sys/bus/serio/devices/serio3/rate cat: /sys/bus/serio/devices/serio3/rate: No such file or directory kbeta@kohvi:~> cat /sys/bus/serio/devices/serio4/rate cat: /sys/bus/serio/devices/serio4/rate: No such file or directory So far have been no crashes in both : suse10.0 and suse10.1beta3 Ok. With a multiplexing controller, the touchpad will indeed be on serio2, and not serio0. The option worked correctly. Now if you don't experience any more crashes using the touchpad, which is quite likely, please attach the output of 'dmidecode' on your system, so that I can add a DMI blacklist entry, setting the rate of 40 automatically, without the need of a kernel command line option. Created attachment 66493 [details]
output of dmidecode
So is everything okay with the rate of 40? Yes. So far no crashes. Created attachment 66935 [details]
Patch to set rate to 40 based on DMI data for Mitac laptops
Can you please try the attached patch? It should identify your system based on your DMI data and set the rate of 40 automatically.
How to apply patch ? I tried to run it in konsole as root: kohvi:/home/kbeta # /home/kbeta/attachment.cgi diff: unrecognized option `--git' diff: Try `diff --help' for more information. /home/kbeta/attachment.cgi: line 2: ---: command not found /home/kbeta/attachment.cgi: line 3: +++: command not found /home/kbeta/attachment.cgi: line 4: @@: command not found /home/kbeta/attachment.cgi: line 5: syntax error near unexpected token `DMI_PRODUCT_NAME' /home/kbeta/attachment.cgi: line 5: ` DMI_MATCH(DMI_PRODUCT_NAME , "dynabook"),' BTW default value of cat /sys/bus/serio/devices/serio2/rate is 80 Created attachment 67029 [details]
Patch to set rate to 40 based on DMI data for Mitac laptops
cd kernel_sources
patch -p1 < attachment.cgi
(this is an updated version of the patch)
kohvi:/home/kbeta # cd /usr/src/linux-2.6.16-rc1-git3-7/kernel kohvi:/usr/src/linux-2.6.16-rc1-git3-7/kernel # patch -pl < /home/kbeta/attachment.cgi patch: **** strip count l is not a number I'm doing something wrong did I cd into right directory ? bash: cd: kernel_sources: No such file or directory It seems you are not familar with compiling a kernel ;-) Let me try to provide a step-by-step guide (I hope it's complete): - install the "kernel-source" package if not already done - cd /usr/src/linux/ (should be a symlink to the kernel source) - patch -p1 < ... - note that it's the number 1 after -p, not the letter l - make cloneconfig - make I'm not sure if "make install" would be a good idea because it could overwrite your running kernel. Try just copying the module (Vojtech, could you please help with the module name and path? ;-) I'm sorry, but psmouse/synaptics, the patched bits, are built-in on SuSE kernels, not modular. So it's necessary to replace the kernel image (a good idea is to save the original one first and make a grub entry for it). I have done all including "make". I could risk "make install"...since I use 10.0 mostly not beta3. After "make install" just reboot ? In my opinion patching kernel is unnecessary because using boot option "psmouse.rate=40" works great (does patch only apply to touchpad or will it affect any USB mouse connected in the future?). Yes, 'make install' and then reboot should work. Patching the kernel is indeed not absolutely necessary, but it'll save other people having a Mitac laptop like you to go through the effort finding out why it crashes. And the patch will only affect the Synaptics pad, where the option also affects PS/2 mice. USB mice aren't affected by either. Done. kbeta@kohvi:~> cat /sys/bus/serio/devices/serio2/rate 40 Great. Can you please also supply the 'dmesg' output, so that I can check that even the messages are correct? I'll then send the patch upstream, so that 10.1 will have it and close the bug. Created attachment 67179 [details] dmesg boot option "irqpoll" is used for https://bugzilla.novell.com/show_bug.cgi?id=116397 Ok, I'm closing this bug as resolved. Patch has been sent upstream. |