[linux-dvb] Re: OOPS with budget card + cam (in dvb_ca_en50221_io_write)

["Carlo E. Prelz" <fluido@fluido.as>]

	Subject: Re: [linux-dvb] OOPS with budget card + cam (in dvb_ca_en50221_io_write)
	Date: sab, set 11, 2004 at 10:33:46 +0200

Quoting Carlo E. Prelz (fluido@fluido.as):

> > Hi - yeah, I fixed a couple of issues in CVS within the last month which might 
> > fix what you're seeing - the main problem was a locking issue during IRQs.
> 
> Good to read this. Will put the latest CVS to the test in half an
> hour. Good news will be no news ;-) 

Sadly I have some news = bad news. I had another oops:

--8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<--

Unable to handle kernel NULL pointer dereference at virtual address 00000010
 printing eip:
e1a11fd4
*pde = 00000000
Oops: 0000 [#1]
PREEMPT 
Modules linked in: stv0299 budget_ci budget_core dvb_core saa7146 ttpci_eeprom b44 8139too
CPU:    0
EIP:    0060:[<e1a11fd4>]    Not tainted
EFLAGS: 00010206   (2.6.8.1) 
EIP is at dvb_ca_en50221_io_write+0xa4/0x1c0 [dvb_core]
eax: 00000000   ebx: 00000000   ecx: fffffff2   edx: 41d4d102
esi: 00000000   edi: 00000003   ebp: 00000000   esp: d1dbbd40
ds: 007b   es: 007b   ss: 0068
Process harvest_periods (pid: 8743, threadinfo=d1dba000 task=c924ec30)
Stack: d1dbbd56 41d4d101 00000001 00000005 00000003 00010001 ffffffea 01a00001 
       00039601 0b880f00 00010001 e6821891 00000000 00010c00 dd0dfee0 00000000 
       00000003 6a921891 fd921891 00000000 00000000 00000086 d1dbbdc0 00000086 
Call Trace:
 [<c0117d4a>] wake_up_state+0x1a/0x20
 [<c0124dfe>] signal_wake_up+0x1e/0x20
 [<c012579e>] __group_send_sig_info+0x9e/0xf0
 [<c0118375>] scheduler_tick+0x185/0x450
 [<c0352080>] i8042_timer_func+0x0/0x30
 [<c035209f>] i8042_timer_func+0x1f/0x30
 [<c0118375>] scheduler_tick+0x185/0x450
 [<c0123ef5>] do_timer+0x65/0xe0
 [<c0135420>] wake_up_page+0x10/0x40
 [<c013555f>] unlock_page+0x1f/0x30
 [<c0144408>] do_wp_page+0x1b8/0x2a0
 [<c0145293>] handle_mm_fault+0x1b3/0x1c0
 [<c0431fc0>] schedule+0x1a0/0x4c0
 [<c0432767>] schedule_timeout+0x67/0xb0
 [<e1a11f30>] dvb_ca_en50221_io_write+0x0/0x1c0 [dvb_core]
 [<c0153e4c>] vfs_write+0xdc/0x160
 [<c0153fa1>] sys_write+0x51/0x80
 [<c010417b>] syscall_call+0x7/0xb
Code: 8b 55 10 b1 ea 0f b6 c3 c1 e0 06 83 3c 10 02 0f 85 bc 00 00 

--8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<--

Whenever I fail in a recording, I do an unload/reload of the
modules. In this case, the program (the parent thread) generated
another oops right after the first one:

--8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<--

Unable to handle kernel NULL pointer dereference at virtual address 0000000c
 printing eip:
e1a12113
*pde = 00000000
Oops: 0000 [#2]
PREEMPT 
Modules linked in: stv0299 budget_ci budget_core dvb_core saa7146 ttpci_eeprom b44 8139too
CPU:    0
EIP:    0060:[<e1a12113>]    Not tainted
EFLAGS: 00010292   (2.6.8.1) 
EIP is at dvb_ca_en50221_io_read_condition+0x23/0x170 [dvb_core]
eax: daafdf18   ebx: 00000000   ecx: 00000000   edx: 08062669
esi: 00000000   edi: 00000000   ebp: daafdf14   esp: daafdea4
ds: 007b   es: 007b   ss: 0068
Process harvest_periods (pid: 8742, threadinfo=daafc000 task=da5126e0)
Stack: da512c04 00000020 00000000 00000000 daafdf28 da512c04 00000000 ffffffff 
       00000000 da512c04 daafdf28 00000000 00000000 d4b35100 daafdf14 e1a122cd 
       00000000 daafdf14 daafdf18 daafc000 daafc000 daafc000 daafc000 00000000 
Call Trace:
 [<e1a122cd>] dvb_ca_en50221_io_read+0x6d/0x3d0 [dvb_core]
 [<c01198d0>] autoremove_wake_function+0x0/0x60
 [<c0123acb>] update_wall_time+0xb/0x40
 [<c0153c2c>] vfs_read+0xdc/0x160
 [<c011fc13>] __do_softirq+0x43/0x90
 [<c0153f21>] sys_read+0x51/0x80
 [<c010417b>] syscall_call+0x7/0xb
Code: 8b 4f 0c 39 4c 24 18 8b 6f 30 72 2b e9 e1 00 00 00 8d 45 01 

--8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----8<--

Then, the program hung. BUT: killing it by hand now had no ill
consequence, and I could cleanly reload the modules and restart the
program without reboot.

Carlo

--
  *         Se la Strada e la sua Virtu' non fossero state messe da parte,
* K * Carlo E. Prelz - fluido@fluido.as             che bisogno ci sarebbe
  *               di parlare tanto di amore e di rettitudine? (Chuang-Tzu)