Most recent kernel where this bug did not occur: Unknown. Also there under 2.6.13. Distribution: (IMHO irrelevant) Knoppix, self compiled vanilla kernel Hardware Environment: (Probably reproducible with all supporting ix86 processors) seen with a Celeron 1GHz and "Mobile Genuine Intel(R) processor 1100MHz" Software Environment: bash> :-) Problem Description: Kernel Oops on "rmmod processor" Steps to reproduce: boot with "nolapic noapic acpi=off" the "processor" module will still be successfully loaded by some of Knoppix' startup scripts, "rmmod processor" produces: Unable to handle kernel NULL pointer dereference at virtual address 00000004 printing eip: c020d2c9 *pde = 00000000 Oops: 0002 [#1] Modules linked in: shpchp pci_hotplug hw_random ehci_hcd autofs4 af_packet vfat fat e100 mii i2c_i801 i2c_core parport_pc parport ohci1394 ieee1394 usbhid uhci_hcd usbcore processor CPU: 0 EIP: 0060:[<c020d2c9>] Not tainted VLI EFLAGS: 00010246 (2.6.15) EIP is at acpi_bus_unregister_driver+0x21/0x38 eax: 00000000 ebx: d0102640 ecx: ceaa3f68 edx: 00000000 esi: 00000000 edi: 00000880 ebp: ceaa3f58 esp: ceaa3f50 ds: 007b es: 007b ss: 0068 Process rmmod (pid: 2669, threadinfo=ceaa2000 task=ce745070) Stack: d0102a00 00000000 ceaa3f60 d01000fc ceaa3fb4 c012f484 00000000 636f7270 6f737365 b7ef0072 b7efd000 ceaa3fa0 c0147ad6 b7efd000 b7efe000 ce63cc80 ceda3b1c ce63cc80 ce63ccb0 b7efd000 ceaa3fb4 00147b4a b7efd000 bfc124a8 Call Trace: [<c0103a8a>] show_stack+0xa9/0xb1 [<c0103bf6>] show_registers+0x14b/0x1bf [<c0103db0>] die+0xc6/0x135 [<c01161d6>] do_page_fault+0x3f6/0x54a [<c0103763>] error_code+0x4f/0x54 [<d01000fc>] acpi_processor_exit+0x1c/0x2e [processor] [<c012f484>] sys_delete_module+0x13c/0x14f [<c0102ca9>] syscall_call+0x7/0xb Code: 04 89 d0 e8 fd fe ff ff 5d c3 55 89 e5 56 31 f6 85 c0 53 89 c3 74 20 e8 56 ff ff ff 8b 83 a8 00 00 00 85 c0 75 16 8b 53 04 8b 03 <89> 50 04 89 02 89 5b 04 89 1b eb 05 be ea ff ff ff 5b 89 f0 5e I guess, it should be something trivial, just don't know what's the correct fix myself - perhaps, the module should not let itself load at all? Not including any additional information, complete dmesg, lspci, etc. available on request, just don't think they are too important:-)
>nolapic noapic acpi=off It's very wield processor module can be loaded with acpi=off. Can you double check it?
Please reopen this bug if: - it is still present in kernel 2.6.16 and - you can provide the requested information.
Ok, sorry for not answering to the first comment by shaohua.li@intel.com - somehow I didn't recognise it as worthy reading - I do not work with bugzilla too often and it came directly after my submit confirmation and "from" and "subject" looked exactly identical, so, I didn't even read it - just checked it still was unread in my mailbox. As for the request - yes, it was INDEED loaded with acpi=off, I was surprised myself and I did double check. Trying to reproduce it for me would be quite diffecult - I reproduced it twice on 2 different systems with different CPUs and different kernels - second time upon request from akpm, and I don't have access to any of those systems. Really, it would be better if somebody who has a chance to FIX this bug tried to reproduce it. As I said, I had it on 2 quite different systems, so, it is not CPU-dependent, and should be easy to reproduce. I'll leave it as "rejected" for now, but, please, have a look at it. Thanks Guenandi
yakui to check and it should not be possible to insmod processor when acpi=off, then not be able to rmmod...
Hi, Guennadi It is impossible to insmod processor and then unable to rmmod processor. Please reopen it if you meet with the exception again. Thanks.