Summary: | mounting a partition on a disk connected to a Promise PDC20267 mass storage controler crashes the system if listening to music | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Matthias Langer <m.langer798> |
Component: | [OLD] Core system | Assignee: | Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel> |
Status: | RESOLVED CANTFIX | ||
Severity: | critical | ||
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Matthias Langer
2005-04-14 11:01:13 UTC
Can you do a cat /proc/interrupts, please? This is $ cat /proc/interrupts CPU0 0: 11092509 XT-PIC timer 1: 7573 XT-PIC i8042 2: 0 XT-PIC cascade 9: 0 XT-PIC acpi 10: 535082 XT-PIC ide0, ide1, EMU10K1, uhci_hcd, uhci_hcd, uhci_hcd, eth0 11: 1037563 XT-PIC ide4, nvidia 12: 496280 XT-PIC i8042 14: 109038 XT-PIC ide2 15: 10 XT-PIC ide3 NMI: 0 ERR: 6 Matthias, is ide0/ide1 the Promise controller? If so, then I assume, that this is really an IRQ conflict. Try to move your SB to another PCI slot and check /proc/interrupts again. Ok, i tried some differnt [not all] combinations with the cards in my pci slots: The only combination i found that really works is: -------- Promise [PCI - IDE] -------- Silicon Image [PCI - IDE] -------- -------- Realtek [network] -------- -------- Emu10k1 $ cat /proc/interupts CPU0 0: 803413 XT-PIC timer 1: 1487 XT-PIC i8042 2: 0 XT-PIC cascade 9: 0 XT-PIC acpi 10: 2620 XT-PIC ide0, ide1 11: 79771 XT-PIC ide4, EMU10K1, uhci_hcd, uhci_hcd, uhci_hcd, eth0, nvidia 12: 24549 XT-PIC i8042 14: 40199 XT-PIC ide2 15: 10 XT-PIC ide3 NMI: 0 ERR: 0 The following combination worked partley, as the system didn't crash, but the sound was disturbed by short beeping noises from time to time, espacially when network load was high: -------- -------- Silicon Image [PCI - IDE] -------- -------- Promise [PCI - IDE] -------- Realtek [network] -------- Emu10k1 These combinations didn't work: -------- -------- Silicon Image [PCI - IDE] -------- -------- Realtek [network] -------- Promise [PCI - IDE] -------- Emu10k1 -------- -------- Silicon Image [PCI - IDE] -------- -------- Realtek [network] -------- Emu10k1 -------- Promise [PCI - IDE] So, the question that remains for me, as my problem is more or less solved: Why does windows2k not suffer from these issues ? Is it possible that with a well choosen kernel configuration these problems disapear in linux too ? As far as I know the order in which pci cards are plugged in should not make a differnece for the end user, at least whith an ideal motherboard and kernel ... Try enabling APIC in your kernel config. THis is probably the way that windows does it. Hmm, according to /proc/cpuinfo my processor doesn't support apic. Is it possible that this problems are caused because '[] Plug and Play ACPI support' and/or '[] ISA support' (allthough i have no isa slots, there is an pci to isa bridge on my bord according to lspci) ? It would be very unlikely that your processor doesn't support APIC. Please just try turning the config option on. You are right; APIC was just disabled by my BIOS - thus i passed lapic=yes to the kernel ... Local APIC disabled by BIOS -- reenabling. Found and enabled local APIC! mapped APIC to ffffd000 (fee00000) But this didn't solve my problem ... Maybe i should mention that i had similar problems (strange beeping noises while listening to music and system/network load is high) with 2.4 kernels and without the promise mass storage controler too: At first I circumvented this issue by using my onboard sound chip instead of emu10k1. After some discussions in the gentoo.user mailing list i moved my soundblaster to the last pci slot - and that solved the problem. Thus, i think that my 'Promise PDC20267' has nothing to do with my problem, as most likely i would have experinced similar behaviour with any other additional pci card, so the summary for this bug (which may be a result of a misconfigured kernel) is somehow misleading. In my opinion this bug should be closed/maked as dublicate in favour of another one. I think that this issue is, if it has to do at least something with my hardware, related to my mainbord, which is an eopx-kh8+. I also had major troubles with my VIA Apollo board, especially with SB live. I ignored the problems until I bought newer hardware. However the problem is almost certainly that there are so many drivers on one interrupt: 10: 535082 XT-PIC ide0, ide1, EMU10K1, uhci_hcd, uhci_hcd, uhci_hcd, eth0 I'd be interested to see the contents of /proc/interrupts now that APIC is enabled. You could try manually assigning IRQ's in your BIOS setup so that they are spaced out more evenly. Closing as this seems more like a BIOS issue.. see comment #10 |