Since Kernel gentoo-dev-source-2.6.9-r1 (built with a make oldconfig from gentoo-dev-sources-2.6.8-r10) my keyboard does not work anymore after resuming from suspend. Some keys produce double output of another key, some keys don't work at all. In the above URL you can see kernel messages of what happens when I push a key. If you need further informations (about chipsets for example), I can provide them. Reproducible: Always Steps to Reproduce: 1. Close the lid of the laptop 2. Open the lid of the laptop 3. Type something on the keyboard. Actual Results: Some keys produce double output of another key, some keys don't work at all. Expected Results: Echoing the correct key. relevant output of dmesg Linux version 2.6.9-gentoo-r1Y (root@pingu) (gcc version 3.3.4 20040623 (Gentoo Linux 3.3.4-r1, ssp-3.3.2-2, pie-8.7.6)) #5 SMP Sat Oct 23 14:11:04 CEST 2004 [...] No local APIC present or hardware disabled [...] Local APIC not detected. Using dummy APIC emulation. [...] apm: BIOS version 1.2 Flags 0x03 (Driver version 1.16ac) [...] input: AT Translated Set 2 keyboard on isa0060/serio0 Kernel .config will be attached as a file
Created attachment 42567 [details] Kernel Configuration for gentoo-dev-sources-2.6.9-r1
Do any strange messages appear in dmesg after the resume? Can you reproduce this on 2.6.10-rc1?
there are way to many new updates in 2.6.9 to run make oldconfig I suggest building the kernel again this time going threw each step I have a funny feeling you will see your problem disappear.
There are absolutely no strange messages in dmesg after the resume. Everything is normal. 2.6.10-rc1? According to packages.gentoo.org, this version is not yet out for gentoo-dev-sources. I'm gogint o build my kernel without make oldconfig tonight. But IIRC, make oldconfig only prompted like 6-8 changes over 2.6.8-r10 and none of them seems to have something to do with keyboard devices.
I mean development-sources-2.6.10-rc1
I just tried to build develoment-sources-2.6.9 and I have the same phenomenon. But I could nail down the problem a litle bit. It has to do somehow with sound (interrupt sharing of IRQ 9, see below (this is the /proc/interrupts from 2.6.8-r10)). I can send the machine to standby (in the shell) as many times as I want and coming back with the keybaord alive. However, after several beep outputs (tab completion or backspace at the beginning of the command line), the system keyboard does not come back anymore after resume. I tried to disable alsa completely but the system beep keeps outputing and as a consequence, after several beeps, I can not get my keyboard working correctly anymore. I've had no time yet to try out 2.6.10. CPU0 0: 461930 XT-PIC timer 1: 1175 XT-PIC i8042 2: 0 XT-PIC cascade 9: 7781 XT-PIC Intel 440MX, yenta, yenta, eth0 12: 11684 XT-PIC i8042 14: 8525 XT-PIC ide0 15: 0 XT-PIC uhci_hcd NMI: 0 LOC: 0 ERR: 0 MIS: 0 After a resume from standby, I get the following in /var/log/messages Oct 30 17:31:45 pingu PCI: Found IRQ 9 for device 0000:00:00.1 Oct 30 17:31:45 pingu PCI: Sharing IRQ 9 with 0000:00:13.1 Oct 30 17:31:45 pingu PCI: Setting latency timer of device 0000:00:00.1 to 64 So I'm quite sure it has to do something with IRQ sharing of IRQ9 as soon as that IRQ passes a particular value...
I'll attach dmesg of 2.6.9 and dmesg of 2.6.8-r10. Maybe it helps someone.
Created attachment 42912 [details] dmesg with development-sources-2.6.9
Created attachment 42913 [details] dmesg with gentoo-dev-sources-2.6.8-r10 (working)
I just tried with 2.6.10-r1 and the keyboard still fails after resume. But it seems that the sound output has nothing to do but it's with IRQ 1 (i8042) that after several keys being pressed, the resume fails. I can see only one relevant config option in menuconfig Device Drivers ---> Input device support ---> PCI PS/2 keyboard and PS/2 mouse controller (which is compiled in the kernel). How can I further debug this failure? Have there been code changes for that part of the kernel? Thanks Marco
Is there a specific reason why you're using the PCI PS/2 driver. Gnerally it's only needed if you're using a docking station (maybe thats the case with you)? Additionaly can you please post your APM specific Kernel configuration?
I thought I needed the pcips2 option since I'm using a port replicator. But already the maintainer of pcips2 told me that it's unlike I'm using pcips2. Today, I completely removed support for pcips2 and since then, my system seems to work well (at least I can not reproduce the error until now). The APM related configuration can be seen in attachment 1 [details]. So after all: is this a bug or not? Enabling a feature where there is no hardware available should not interfere, right? Thanks Marco
> So after all: is this a bug or not? Enabling a feature where there is no > hardware available should not interfere, right? Yeah right. It _should_ not interfere. But somtimes it does nonetheless (due to bad code or whatever). I think you could close _this_ bug safely since removing pcips2 support from kernel solved the problem, but I would report the problem to upstream if I were you so the Kernel hackers get the chance to fix it.
Yep, closing as this is somewhat user-error. It's upto the pcips2 maintainer if he regards it as a bug or not - perhaps you could ask?
unfortunately, I have to reopen this bug again :-( It seems it was coincidence that after removing the pcips2 support, the keyboard kept on working, because now, the bug is reproducable again whenever I want. My current kernel config will be attached again. Thanks for further investigation. If you need to know more, just let me know. Marco
Created attachment 43237 [details] current 2.6.9-r1 config file
Ok. If it's definately also a problem on 2.6.10-rc1 then its an upstream issue. Could you please file a bug at the kernel bugzilla? http://bugzilla.kernel.org Thanks
upstream solved the problem. as of gentoo-dev-sources-2.6.11-r4, the problem is inexistant. thanks
