Most recent kernel where this bug did not occur: no other kernels tested Distribution: Fedora Core 4 Hardware Environment: AMD Sempron 3400+, GigaByte K8NS Pro, 2x512MB DDR memory modules, custom BIOS version F14b (not publicly available) Software Environment: Problem Description: When cpufreq_ondemand "ondemand" mode is turned on, computer sometimes lose its keyboard input, kernel oops'es, it's hardly possible to reboot 'cause smth stange is happening to the kernel. Here's an oopse I receive when the problem is triggered: Unable to handle kernel paging request at virtual address 044002e8 printing eip: c02b02f9 *pde = 00000000 Oops: 0000 [#1] PREEMPT Modules linked in: snd_rtctimer ppp_synctty ppp_async crc_ccitt ppp_generic slhc cpufreq_powersave cpufreq_ondemand powernow_k8 eeprom it87 hwmon_vid i2c_isa vm net parport_pc parport vmmon ip_conntrack_ftp ipt_state ip_conntrack nfnetlink i pt_REJECT ipt_owner ipt_limit ipt_multiport iptable_filter ipt_ULOG ip_tables bi nfmt_misc ntfs thermal processor fan button ohci1394 ieee1394 ohci_hcd ehci_hcd nvidia amd64_agp agpgart i2c_nforce2 emu10k1_gp gameport snd_intel8x0 snd_emu10k 1_synth snd_emux_synth snd_seq_virmidi snd_seq_midi_emul snd_emu10k1 snd_rawmidi snd_ac97_codec snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device snd_pcm_os s snd_mixer_oss snd_pcm snd_timer snd_ac97_bus snd_page_alloc snd_util_mem snd_h wdep snd skge crc32 sk98lin floppy sata_nv libata scsi_mod CPU: 0 EIP: 0060:[<c02b02f9>] Tainted: P VLI EFLAGS: 00010246 (2.6.14.4-k8) EIP is at cpufreq_notify_transition+0x1a9/0x200 eax: 000f4240 ebx: c1963ed8 ecx: 0003b007 edx: 00000000 esi: c037ac60 edi: 00000000 ebp: 00000000 esp: c1963ea8 ds: 007b es: 007b ss: 0068 Process events/0 (pid: 4, threadinfo=c1962000 task=c1907a50) Stack: 00000012 00000000 c1963ed8 c036408c 00000001 00000000 00000002 00000000 f8cca2e8 c1963ed8 00000001 00000012 00000000 001e8480 000f4240 00000000 f76cb940 00000006 f79806c0 0000000c f8cca45e f76cb940 00000001 00000000 Call Trace: [<f8cca2e8>] transition_frequency+0xc8/0x120 [powernow_k8] [<f8cca45e>] powernowk8_target+0x11e/0x1a0 [powernow_k8] [<f8c93475>] dbs_check_cpu+0x155/0x1e0 [cpufreq_ondemand] [<f8c9351b>] do_dbs_timer+0x1b/0x60 [cpufreq_ondemand] [<c012adad>] worker_thread+0x1dd/0x2d0 [<f8c93500>] do_dbs_timer+0x0/0x60 [cpufreq_ondemand] [<c01159d0>] default_wake_function+0x0/0x20 [<c012abd0>] worker_thread+0x0/0x2d0 [<c012f446>] kthread+0xb6/0xc0 [<c012f390>] kthread+0x0/0xc0 [<c0101069>] kernel_thread_helper+0x5/0xc Code: 85 c9 fe ff ff 8b 43 08 89 42 1c e9 be fe ff ff 89 f6 89 d0 31 d2 f7 f1 89 d5 89 c3 e9 25 ff ff ff 89 f8 31 ff 8b 0d ec 02 40 c0 <f7> 25 e8 02 40 c0 85 d2 89 c5 89 54 24 0c 74 0c 89 d0 31 d2 f7 Alt + ScrollLock partial output (the PC was almost stuck - I couldn't save this output in the file) right after kernel oopse: Pid 0, Comm: Swapper EIP: 0060 :[<f8b61b09>] CPU: 0 EIP is at acpi_processor_idle + 0xf1/0x275 [processor] ... [<c0100da4>] cpu_idle + 0x34/0x50 [<c03ac89a>] start_kernel + 0x17a/0x1a0 [<c03ac3c0>] unknown_bootoption + 0x0/0x1f0 Steps to reproduce: Turn on the PC with given hardware setup. modprobe powernow-k8 modprobe cpufreq_ondemand echo "ondemand" > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor After several hours the keyboard in X server stops functioning, kernel gives an oopse then it takes as long as five minutes (usual shutdown time is around 40 seconds) to shut the PC down using a poweroff PC case button.
Can you reproduce this in 2.6.15 without having ever loaded any proprietary modules (vmware)? Thanks.
I did remove all proprietary modules (nvidia and vmware) and after two hours of working the PC died. (X servers suddenly crashed and started restarting with no success - this way I was forced to reboot). Most possibly it's my hardware (motherboard) or BIOS bug as my Windows XP box is affected by it as well. I can't leave this bug open as it cannot be solved in a software.