[linux-dvb] Re: Announcement: new saa7146 driver core for the av7110 dvb driver

["Pete Martin" <petemartin@ntlworld.com>]

Michael,

As the original author of this driver, I was wondering if you have any ideas
on tracking down the kernel panic that sometimes occurs in this driver when
starting up from cold. Several people have reported similar problems. I have
a ksymoops output which points to the routine fidbirq being called before
the pointer to the grabbing areas has been setup (attached).

Regards

Pete Martin
ksymoops 2.4.5 on i686 2.4.19-16mdk-dvb-t.  Options used
     -V (default)
     -k /proc/ksyms (specified)
     -l /proc/modules (default)
     -o /lib/modules/2.4.19-16mdk-dvb-t/ (default)
     -m /boot/System.map-2.4.19-16mdk-dvb-t (default)

Oops: 0000
CPU:    0
EIP:    0010:[<c8905510>]    Tainted: GF
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010202
eax: c8951c00   ebx: 00000001   ecx: 0095b635   edx: c8951c00
esi: c1c30000   edi: 5aa65fd0   ebp: c0297f2c   esp: c0297f1c
ds: 0018   es: 0018   ss: 0018
Process swapper (pid: 0, stackpage=c0297000)
Stack: c1c30000 00000000 c02bc578 fffffff7 c0297f3c c0120164 c1c30000 00000001
       c0297f58 c012000a c02bc578 00000046 c02baa60 0000000b c0297f78 c0297f70
       c010a416 c32aa7c0 000255b7 c0296000 00000010 c0297fc4 c010c8c8 000255b7
Call Trace:    [<c0120164>] [<c012000a>] [<c010a416>] [<c010c8c8>] [<c01071a4>]
  [<c0114b35>] [<c0114a80>] [<c0107212>] [<c0105000>]
Code: 80 3f 47 74 08 8d 65 f4 5b 5e 5f 5d c3 51 8d 86 d0 0e 00 00


>>EIP; c8905510 <[msdos].data.end+1c01d/23b6d>   <=====

>>eax; c8951c00 <[dvb-ttpci]Root+1f020/3e200>
>>ecx; 0095b635 Before first symbol
>>edx; c8951c00 <[dvb-ttpci]Root+1f020/3e200>
>>esi; c1c30000 <_end+1937428/8509488>
>>edi; 5aa65fd0 Before first symbol
>>ebp; c0297f2c <init_task_union+1f2c/2000>
>>esp; c0297f1c <init_task_union+1f1c/2000>

Trace; c0120164 <tasklet_action+44/70>
Trace; c012000a <do_softirq+aa/b0>
Trace; c010a416 <do_IRQ+b6/c0>
Trace; c010c8c8 <call_do_IRQ+5/d>
Trace; c01071a4 <default_idle+24/30>
Trace; c0114b35 <apm_cpu_idle+b5/150>
Trace; c0114a80 <apm_cpu_idle+0/150>
Trace; c0107212 <cpu_idle+42/60>
Trace; c0105000 <_stext+0/0>

Code;  c8905510 <[msdos].data.end+1c01d/23b6d>
00000000 <_EIP>:
Code;  c8905510 <[msdos].data.end+1c01d/23b6d>   <=====
   0:   80 3f 47                  cmpb   $0x47,(%edi)   <=====
Code;  c8905513 <[msdos].data.end+1c020/23b6d>
   3:   74 08                     je     d <_EIP+0xd> c890551d <[msdos].data.end+1c02a/23b6d>
Code;  c8905515 <[msdos].data.end+1c022/23b6d>
   5:   8d 65 f4                  lea    0xfffffff4(%ebp),%esp
Code;  c8905518 <[msdos].data.end+1c025/23b6d>
   8:   5b                        pop    %ebx
Code;  c8905519 <[msdos].data.end+1c026/23b6d>
   9:   5e                        pop    %esi
Code;  c890551a <[msdos].data.end+1c027/23b6d>
   a:   5f                        pop    %edi
Code;  c890551b <[msdos].data.end+1c028/23b6d>
   b:   5d                        pop    %ebp
Code;  c890551c <[msdos].data.end+1c029/23b6d>
   c:   c3                        ret    
Code;  c890551d <[msdos].data.end+1c02a/23b6d>
   d:   51                        push   %ecx
Code;  c890551e <[msdos].data.end+1c02b/23b6d>
   e:   8d 86 d0 0e 00 00         lea    0xed0(%esi),%eax

Kernel panic: Aiee, killing interrupt handler!

--------------------------

Looks to me like this is at av7110.c->fidbirq near the end where it
is trying to read from mem[0] while comparing th econtents of mem9[0] 
to 0x47.