My mouse will ocassionally freeze and stop responding. I can still operate normally with the keyboard. All applications work fine and nothing besides the mouse seems to be a problem. When I try reseting the xserver, switching to a VT, or just logging out of my wm the whole system will freeze. Reproducible: Sometimes Steps to Reproduce: This problem seems to be random. Here are a few possible causes in order of likelyness (purely arbitrary): 1. I switched from mm-sources, to gdev-sources, to developement-sources in order to eliminate the problem. Each kernel behaved the same way. I really don't want to go back to 2.4 but it may be related to 2.6 2. I am using ati-drivers (radeon 8500) which may be the source of the problem. 3. I use tvtime with my Hauppauge bt878 card. tvtime crashes fairly often and requires me to kill it. This has been happening before I switched to 2.4 though.
I get a bunch of kernel errors saying irq #: nobody cared! It's always 18 or 10. 10: 64862 1 IO-APIC-level uhci_hcd, uhci_hcd 18: 85133 1 IO-APIC-level EMU10K1 This is at the bottom of dmesg in my current session: [fglrx] AGP detected, AgpState = 0x1f000207 (hardware caps of chipset) agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0. agpgart: Putting AGP V2 device at 0000:00:00.0 into 4x mode agpgart: Putting AGP V2 device at 0000:01:00.0 into 4x mode [fglrx] AGP enabled, AgpCommand = 0x1f000304 (selected caps) [fglrx] free AGP = 54800384 [fglrx] max AGP = 54800384 [fglrx] free LFB = 47169536 [fglrx] max LFB = 47169536 [fglrx] free Inv = 0 [fglrx] max Inv = 0 [fglrx] total Inv = 0 [fglrx] total TIM = 0 [fglrx] total FB = 0 [fglrx] total AGP = 16384 atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0). atkbd.c: This is an XFree86 bug. It shouldn't access hardware directly. atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0). atkbd.c: This is an XFree86 bug. It shouldn't access hardware directly. Debug: sleeping function called from invalid context at include/asm/semaphore.h:119 in_atomic():1, irqs_disabled():0 Call Trace: [<c0120fcb>] __might_sleep+0xab/0xd0 [<f09a76f3>] __ke_down_struct_sem+0x33/0x50 [fglrx] [<f09b7764>] drm_getmagic+0x124/0x1c0 [fglrx] [<f09b7640>] drm_getmagic+0x0/0x1c0 [fglrx] [<f09aa49d>] firegl_ioctl+0x15d/0x1e0 [fglrx] [<c0175b37>] sys_ioctl+0x117/0x2d0 [<c010966b>] syscall_call+0x7/0xb This kind of error will appear every time I start tvtime.
I'm currently using development-sources-2.6.4 Today I've left the computer on and when I come back, the monitor is black and the computer won't respond to anything (including Ctrl+Alt+Delete) When I reboot the BIOS cannot find the boot disk and gives a DISK ERROR message. If I power off for a second and then on again things work fine. This kind of scares me and may be related to reiserfs which I seem to be having problems with as of late.
I was able to catch the error this time. The kernel actually disables irq 10 causing the mouse to freeze. Sorry to post so much but I'm not exactly sure what's relevant. [fglrx] AGP detected, AgpState = 0x1f000207 (hardware caps of chipset) agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0. agpgart: Putting AGP V2 device at 0000:00:00.0 into 4x mode agpgart: Putting AGP V2 device at 0000:01:00.0 into 4x mode [fglrx] AGP enabled, AgpCommand = 0x1f000304 (selected caps) [fglrx] free AGP = 54800384 [fglrx] max AGP = 54800384 [fglrx] free LFB = 47169536 [fglrx] max LFB = 47169536 [fglrx] free Inv = 0 [fglrx] max Inv = 0 [fglrx] total Inv = 0 [fglrx] total TIM = 0 [fglrx] total FB = 0 [fglrx] total AGP = 16384 atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0). atkbd.c: This is an XFree86 bug. It shouldn't access hardware directly. atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0). atkbd.c: This is an XFree86 bug. It shouldn't access hardware directly. Debug: sleeping function called from invalid context at include/asm/semaphore.h:119 in_atomic():1, irqs_disabled():0 Call Trace: [<c0120fcb>] __might_sleep+0xab/0xd0 [<f09a76f3>] __ke_down_struct_sem+0x33/0x50 [fglrx] [<f09b7764>] drm_getmagic+0x124/0x1c0 [fglrx] [<f09b7640>] drm_getmagic+0x0/0x1c0 [fglrx] [<f09aa49d>] firegl_ioctl+0x15d/0x1e0 [fglrx] [<c0175b37>] sys_ioctl+0x117/0x2d0 [<c010966b>] syscall_call+0x7/0xb bttv0: PLL can sleep, using XTAL (28636363). tuner: tv freq set to 67.25 tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. tuner: tv freq set to 67.25 tuner: tv 0x07 0x10 0x8e 0xa0 i2c_adapter i2c-0: master_send: writing 4 bytes. irq 18: nobody cared! Call Trace: [<c010baca>] __report_bad_irq+0x2a/0x90 [<c010bbc0>] note_interrupt+0x70/0xb0 [<c010bf00>] do_IRQ+0x160/0x1a0 [<c0109fd8>] common_interrupt+0x18/0x20 [<c0396ba7>] established_get_first+0x97/0x100 [<c0396d6a>] established_get_idx+0x1a/0x50 [<c0396e56>] tcp_get_idx+0xb6/0xe0 [<c0396ed8>] tcp_seq_start+0x58/0x60 [<c01830be>] seq_read+0x11e/0x340 [<c0161248>] vfs_read+0xb8/0x130 [<c01614f2>] sys_read+0x42/0x70 [<c010966b>] syscall_call+0x7/0xb handlers: [<f0a1b6c0>] (snd_emu10k1_interrupt+0x0/0x420 [snd_emu10k1]) Disabling IRQ #18 cdrom: This disc doesn't have any tracks I recognize! cdrom: This disc doesn't have any tracks I recognize! Debug: sleeping function called from invalid context at include/asm/semaphore.h:119 in_atomic():1, irqs_disabled():0 Call Trace: [<c0120fcb>] __might_sleep+0xab/0xd0 [<f09a76f3>] __ke_down_struct_sem+0x33/0x50 [fglrx] [<f09b7764>] drm_getmagic+0x124/0x1c0 [fglrx] [<f09b7640>] drm_getmagic+0x0/0x1c0 [fglrx] [<f09aa49d>] firegl_ioctl+0x15d/0x1e0 [fglrx] [<c0175b37>] sys_ioctl+0x117/0x2d0 [<c010966b>] syscall_call+0x7/0xb cdrom: This disc doesn't have any tracks I recognize! cdrom: This disc doesn't have any tracks I recognize! Debug: sleeping function called from invalid context at include/asm/semaphore.h:119 in_atomic():1, irqs_disabled():0 Call Trace: [<c0120fcb>] __might_sleep+0xab/0xd0 [<f09a76f3>] __ke_down_struct_sem+0x33/0x50 [fglrx] [<f09b7764>] drm_getmagic+0x124/0x1c0 [fglrx] [<f09b7640>] drm_getmagic+0x0/0x1c0 [fglrx] [<f09aa49d>] firegl_ioctl+0x15d/0x1e0 [fglrx] [<c0175b37>] sys_ioctl+0x117/0x2d0 [<c010966b>] syscall_call+0x7/0xb cdrom: This disc doesn't have any tracks I recognize! cdrom: This disc doesn't have any tracks I recognize! Debug: sleeping function called from invalid context at include/asm/semaphore.h:119 in_atomic():1, irqs_disabled():0 Call Trace: [<c0120fcb>] __might_sleep+0xab/0xd0 [<f09a76f3>] __ke_down_struct_sem+0x33/0x50 [fglrx] [<f09b7764>] drm_getmagic+0x124/0x1c0 [fglrx] [<f09b7640>] drm_getmagic+0x0/0x1c0 [fglrx] [<f09aa49d>] firegl_ioctl+0x15d/0x1e0 [fglrx] [<c0175b37>] sys_ioctl+0x117/0x2d0 [<c010966b>] syscall_call+0x7/0xb irq 10: nobody cared! Call Trace: [<c010baca>] __report_bad_irq+0x2a/0x90 [<c010bbc0>] note_interrupt+0x70/0xb0 [<c010bf00>] do_IRQ+0x160/0x1a0 [<c0109fd8>] common_interrupt+0x18/0x20 [<c0107030>] default_idle+0x0/0x40 [<c010705c>] default_idle+0x2c/0x40 [<c01070eb>] cpu_idle+0x3b/0x50 [<c01244b8>] printk+0x178/0x1f0 handlers: [<c02e8540>] (usb_hcd_irq+0x0/0x70) [<c02e8540>] (usb_hcd_irq+0x0/0x70) Disabling IRQ #10
Please post that stuff as attachments in the future.
This bug is 100% reliable for me on my system (IBM Thinkpad T41p, ATI FireGL Mobility T2). If I exit X or switch to a VT, the whole system hangs, and needs a hard reboot.
I posted a bug at the kernel bug tracker and it turned out to be a duplicate of another known bug. "VIA Apollo PRO133x quirk is already filed. We'll put the patch in the existing bug report, please test it when it is ready. thanks, -Len" Existing Bug Report: http://bugme.osdl.org/show_bug.cgi?id=2243 Gentoo Forums post: http://forums.gentoo.org/viewtopic.php?p=1050753#1050753
OK, thanks. I'm just adding a patch for the other issue in this bug (atkbd.c etc)
I see all these errors and I'm not using a VIA apollo. This is an ati-drivers bug. Is this being pushed upstream to kernel, to ATI, to xfree, or what? Either way, I don't have a working solution here and now.
Shevek can you reproduce with various versions of ati-drivers or is it specific to one?
I was using the latest stable and unstable versions. I think I had to go to the latest unstable to get it to even load. ATI have done a remarkably shoddy job with these drivers, especially compared to nvidia. Unfortunately I only have one ATI machine, which is a core business machine. It's running radeon framebuffer right now and I can't afford to crash or reboot it, so testing is limited. I was using 2.6.5-mm6 kernel and ati-drivers-3.7.6-r1. The computer is an IBM Thinkpad T41p with an ATI FireGL Mobility T2. I tried with Linux agpgart and ATI's AGP driver. All didn't load or hung. Then I gave up and went back to framebuffer. S.
please try the 2.6.6 Can you track down the issue to a specific kernel version? looks like more an agpgart incompatibilty
Well I had to reboot so I had another hack at this while I was at it. I think it's an incompatibility with the radeon framebuffer. It is not an AGP problem. I now have the fglrx drivers running nearly correctly, except I don't get the screen back after an S3 recovery, and I have funny coloured pixels all over the screen (some anti-aliasing bug?). There are really two problems in this thread. One is the incompatibility between radeon framebuffer and fglrx. The other is the ACPI problem. There is also now the anti-aliasing problem. I guess that if it's noted never to use radeon framebuffer with fglrx, then this bug can be closed and others opened in its place referencing this one.
Is this bug still valid?
Probably. I haven't gone near the ATI FireGL drivers since I reported this bug. I've just been using Radeon 2D, which is a shame because I really need Quake3 on the tech support laptops. And ATI did win an award for the worst vendor linux support. I do have an LRMI hack which I fire up from acpid to fix the S3 issue while using Radeon 2D from Xorg though. Look for a package called boot-radeon and get it into portage or ibm-acpi or something. Remember to chvt to a non-X11 vt before running it.
Well, somewhere among all the system updates I believe this bug was fixed, at least for my system. linux-2.6.10 ati-drivers-8.8.25 xorg-6.8.0
I hope I could close it now
