Skip to contentSkip to conversation

2.6.9-rc2-mm2 oops on bootup on Acer Aspire 1501LMi (Athlon64) in 32 bit mode

DISCUSSION Bjorn Helgaas 1 participant 1 message Sep 24, 2004 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw
Quoted 20 lines

inserting floppy driver for 2.6.9-0.rc2.1ark

Unable to handle kernel paging request at virtual address e082ed20
...
EIP:    0060:[<c0218c13>]    Not tainted VLI
EFLAGS: 00010202   (2.6.9-0.rc2.1ark)

EIP is at acpi_bus_register_driver+0x34/0x54
eax: ded1b000 ebx: e0d13ba0 ecx: c0334b44 edx: e082ed20
esi: e0d13e00 edi: ffffffed ebp: c03361e0 esp: ded1bf5c
ds: 007b es: 007b ss: 0068
Process modprobe (pid: 1183, threadinfo=ded1b000 task=df22e1f0)

Stack: c03361f8 e0d0d278 e0d13ba0 c03361e0 e0c8f871 0000000a 00000400 e0d0f670
       ded1bfa0 c03361f8 e0d13e00 ded1b000 c01216d7 c03361f8 e0d13e00 ded1b000
       c03361e0 c013ad12 c03f7f88 00000001 e0d13e00 b7fec000 0804f8e8 00000001
Call Trace:
 [<e0d0d278>] acpi_floppy_init+0x18/0x50 [floppy]
 [<e0c8f871>] floppy_init+0x11/0x650 [floppy]
 [<c01216d7>] printk+0x17/0x20
 [<c013ad12>] sys_init_module+0x172/0x250
 [<c010624b>] syscall_call+0x7/0xb
Code: ed ff ff ff 8b 5c 24 08 75 3f 85 db b0 ea 74 39 66 b8 00 f0 21 e0 ff 40 
Quoted 2 lines

14 8b 15 2c 56 37 c0 c7 03 28 56 37 c0 89 1d 2c 56 37 c0 <89> 1a 89 53 04 ff 48 14 8b 40 08 a8 08 74 05 e8 79 6b 0d 00 89

Disassembling your code, I get this:

 mov    0xc037562c,%edx
 movl   $0xc0375628,(%ebx)
 mov    %ebx,0xc037562c
 mov    %ebx,(%edx)  <== oops occurred here
 mov    %edx,0x4(%ebx)

This is the "list_add_tail(&driver->node, &acpi_bus_drivers);" in acpi_bus_register_driver(). 0xc0375628 should be acpi_bus_drivers, and 0xe082ed20 looks like &acpi_floppy_driver.node.

Your acpi_floppy_driver address is a long ways from the floppy code addresses:
0xe0c8f871 (floppy_init).
0xe082ed20 (&acpi_floppy_driver)
I.e., they're about 4.5M apart. I suppose module text and data sections are allocated separately so maybe this isn't a problem.

FWIW, on my box, /proc/kallsyms says:
f881c7d0 t floppy_init [floppy]
f8819d00 d acpi_floppy_driver [floppy]
so they're about 10K apart (the whole floppy module is about 75K).

I am curious about why load_module appears in the backtrace:

Quoted 7 lines
 [<e0d06000>] floppy_hardint+0x0/0x130 [floppy]
 [<c0156e22>] __vunmap+0xc2/0x100
 [<c0156e87>] vfree+0x27/0x40
 [<c013aa96>] load_module+0xa46/0xb50
 [<c0118990>] do_page_fault+0x0/0x5de
 [<c0106ce5>] error_code+0x2d/0x38
 [<c0218c13>] acpi_bus_register_driver+0x34/0x54

Is the oops reproducible, or could this be some sort of race in module loading or something?