2.4.7-ac10 acpi BUG (again)

2.4.7-ac10 acpi BUG (again)

[Eric Buddington]

I earlier reported a BUG on boot with 2.4.3-ac3 in the acpi
code. Andy's patches made the kernel not boot (out of data while
decompressing), so here's another shot with 2.4.7-ac10.

I enabled ACPI_DEBUG in the kernel config - is that the same as your
earlier patch, Andy? In any case, I get the same problem, pretty much.
Here it is. Oh, yeah - my Penguin is an AMD K6-2. Can'd give an ID on
the motherboard offhand (Gigabyte, I think).

-Eric

------------------------------------------------
ksymoops 2.4.1 on i586 2.4.3-ac3.  Options used
     -V (default)
     -K (specified)
     -L (specified)
     -o /packages/linux/2.4.7-ac10/k6/lib/modules/2.4.7-ac10 (specified)
     -m /packages/linux/2.4.7-ac10/any/src/linux/System.map (specified)

No modules in ksyms, skipping objects
kernel BUG at sched.c:711!
invalid operand: 0000
CPU:    0
EIP:    0010:[<c011310d>]
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010082
eax: 0000001b   ebx: c1216000   ecx: 00000001   edx: 00000b31
esi: c1217dc0   edi: c0189e7c   ebp: c1217d84   esp: c1217d58
ds: 0018   es: 0018   ss: 0018
Process swapper (pid: 1, stackpage=c1217000)
Stack: c02444cf 000002c7 c02444b6 c1217dc0 c1216000 00000000 c1216000 00000711 
       c1216000 c1217dc0 c0189e7c c1217e7c c0106d31 00000711 c0184238 c7fa1460 
       c1217dc0 c0189e7c c1217e7c 00000008 00000018 00000018 00000078 c0105471 
Call Trace: [<c0189e7c>] [<c0106d31>] [<c0184238>] [<c0189e7c>] [<c0105471>] 
   [<c0184386>] [<c0184238>] [<c01844b9>] [<c0189ece>] [<c0189e7c>] [<c0189829>] 
   [<c01897a9>] [<c018a723>] [<c0183dfc>] [<c0107e81>] [<c0183df0>] [<c0107fd4>] 
   [<c010a14e>] [<c0183efe>] [<c018c4a6>] [<c018bff5>] [<c018bc86>] [<c018aef1>] 
   [<c018a922>] [<c0189eed>] [<c0189e94>] [<c01896bc>] [<c0186a61>] [<c01999d9>] 
   [<c0105000>] [<c0105000>] [<c010503f>] [<c0105000>] [<c010547a>] [<c0105034>] 
Code: 0f 0b 83 c4 0c 8d 65 f4 5b 5e 5f 5d c3 89 f6 8b 45 e8 c1 e0 

>>EIP; c011310d <schedule+5d/3c0>   <=====
Trace; c0189e7c <acpi_ev_global_lock_thread+0/18>
Trace; c0106d31 <reschedule+5/c>
Trace; c0184238 <acpi_os_queue_exec+0/e0>
Trace; c0189e7c <acpi_ev_global_lock_thread+0/18>
Trace; c0105471 <kernel_thread+1d/30>
Trace; c0184386 <acpi_os_schedule_exec+6e/118>
Trace; c0184238 <acpi_os_queue_exec+0/e0>
Trace; c01844b9 <acpi_os_queue_for_execution+89/14c>
Trace; c0189ece <acpi_ev_global_lock_handler+3a/44>
Trace; c0189e7c <acpi_ev_global_lock_thread+0/18>
Trace; c0189829 <acpi_ev_fixed_event_dispatch+41/a4>
Trace; c01897a9 <acpi_ev_fixed_event_detect+55/94>
Trace; c018a723 <acpi_ev_sci_handler+1b/2c>
Trace; c0183dfc <acpi_irq+c/10>
Trace; c0107e81 <handle_IRQ_event+31/68>
Trace; c0183df0 <acpi_irq+0/10>
Trace; c0107fd4 <do_IRQ+54/a0>
Trace; c010a14e <call_do_IRQ+5/b>
Trace; c0183efe <acpi_os_out16+a/c>
Trace; c018c4a6 <acpi_hw_low_level_write+176/180>
Trace; c018bff5 <acpi_hw_register_write+91/240>
Trace; c018bc86 <acpi_hw_register_bit_access+28a/37c>
Trace; c018aef1 <acpi_enable_event+6d/ac>
Trace; c018a922 <acpi_install_fixed_event_handler+76/94>
Trace; c0189eed <acpi_ev_init_global_lock_handler+15/2c>
Trace; c0189e94 <acpi_ev_global_lock_handler+0/44>
Trace; c01896bc <acpi_ev_initialize+48/5c>
Trace; c0186a61 <acpi_enable_subsystem+55/80>
Trace; c01999d9 <acpi_init+121/15c>
Trace; c0105000 <_stext+0/0>
Trace; c0105000 <_stext+0/0>
Trace; c010503f <init+b/140>
Trace; c0105000 <_stext+0/0>
Trace; c010547a <kernel_thread+26/30>
Trace; c0105034 <init+0/140>
Code;  c011310d <schedule+5d/3c0>
00000000 <_EIP>:
Code;  c011310d <schedule+5d/3c0>   <=====
   0:   0f 0b                     ud2a      <=====
Code;  c011310f <schedule+5f/3c0>
   2:   83 c4 0c                  add    $0xc,%esp
Code;  c0113112 <schedule+62/3c0>
   5:   8d 65 f4                  lea    0xfffffff4(%ebp),%esp
Code;  c0113115 <schedule+65/3c0>
   8:   5b                        pop    %ebx
Code;  c0113116 <schedule+66/3c0>
   9:   5e                        pop    %esi
Code;  c0113117 <schedule+67/3c0>
   a:   5f                        pop    %edi
Code;  c0113118 <schedule+68/3c0>
   b:   5d                        pop    %ebp
Code;  c0113119 <schedule+69/3c0>
   c:   c3                        ret    
Code;  c011311a <schedule+6a/3c0>
   d:   89 f6                     mov    %esi,%esi
Code;  c011311c <schedule+6c/3c0>
   f:   8b 45 e8                  mov    0xffffffe8(%ebp),%eax
Code;  c011311f <schedule+6f/3c0>
  12:   c1 e0 00                  shl    $0x0,%eax

 <0>Kernel panic: Aiee, killing interrupt handler!

RE: 2.4.7-ac10 acpi BUG (again)

[Grover, Andrew]

Hi Eric,

I believe I know what's happening. The stack trace was very helpful.

The Global Lock is used to synchronize access to the hardware between the
BIOS and the OS. This is not a spinlock -- if the BIOS has it when the OS
(that's us) needs it, we set a "pending" bit, so that we get an interrupt
when it's available.

We're enabling the interrupt and we're immediately taking one. This is
strange because at this stage of initialization, we've never attempted to
acquire the Global Lock, which means we've never set the pending bit.

I think this means we either need to explicitly clear it before we enable
the interrupt, or treat it as spurious when we get it.

If I'm right, our next patch should fix this. Thanks for the report.

Regards -- Andy

> From: Eric Buddington [mailto:eric@sparrow.bur.adelphia.net]
> I earlier reported a BUG on boot with 2.4.3-ac3 in the acpi
> code. Andy's patches made the kernel not boot (out of data while
> decompressing), so here's another shot with 2.4.7-ac10.
> 
> I enabled ACPI_DEBUG in the kernel config - is that the same as your
> earlier patch, Andy? In any case, I get the same problem, pretty much.
> Here it is. Oh, yeah - my Penguin is an AMD K6-2. Can'd give an ID on
> the motherboard offhand (Gigabyte, I think).
> 
> -Eric

> >>EIP; c011310d <schedule+5d/3c0>   <=====
> Trace; c0189e7c <acpi_ev_global_lock_thread+0/18>
> Trace; c0106d31 <reschedule+5/c>
> Trace; c0184238 <acpi_os_queue_exec+0/e0>
> Trace; c0189e7c <acpi_ev_global_lock_thread+0/18>
> Trace; c0105471 <kernel_thread+1d/30>
> Trace; c0184386 <acpi_os_schedule_exec+6e/118>
> Trace; c0184238 <acpi_os_queue_exec+0/e0>
> Trace; c01844b9 <acpi_os_queue_for_execution+89/14c>
> Trace; c0189ece <acpi_ev_global_lock_handler+3a/44>
> Trace; c0189e7c <acpi_ev_global_lock_thread+0/18>
> Trace; c0189829 <acpi_ev_fixed_event_dispatch+41/a4>
> Trace; c01897a9 <acpi_ev_fixed_event_detect+55/94>
> Trace; c018a723 <acpi_ev_sci_handler+1b/2c>
> Trace; c0183dfc <acpi_irq+c/10>
> Trace; c0107e81 <handle_IRQ_event+31/68>
> Trace; c0183df0 <acpi_irq+0/10>
> Trace; c0107fd4 <do_IRQ+54/a0>
> Trace; c010a14e <call_do_IRQ+5/b>
> Trace; c0183efe <acpi_os_out16+a/c>
> Trace; c018c4a6 <acpi_hw_low_level_write+176/180>
> Trace; c018bff5 <acpi_hw_register_write+91/240>
> Trace; c018bc86 <acpi_hw_register_bit_access+28a/37c>
> Trace; c018aef1 <acpi_enable_event+6d/ac>
> Trace; c018a922 <acpi_install_fixed_event_handler+76/94>
> Trace; c0189eed <acpi_ev_init_global_lock_handler+15/2c>
> Trace; c0189e94 <acpi_ev_global_lock_handler+0/44>
> Trace; c01896bc <acpi_ev_initialize+48/5c>
> Trace; c0186a61 <acpi_enable_subsystem+55/80>
> Trace; c01999d9 <acpi_init+121/15c>
> Trace; c0105000 <_stext+0/0>
> Trace; c0105000 <_stext+0/0>
> Trace; c010503f <init+b/140>
> Trace; c0105000 <_stext+0/0>
> Trace; c010547a <kernel_thread+26/30>
> Trace; c0105034 <init+0/140>