8259A initialization oops after resume

8259A initialization oops after resume

[Jan Rychter]

People are getting a particular kind of oops after resuming from S4
(implemented by swsusp). I'm posting this here, because the problem
seems not to be specific to swsusp.

Hopefully some people on this list will know how and why the 8259A needs
to be initialized upon resuming from S4 and why it fails.

The oops is reproducible -- one may do a number of suspend/resume cycles
without it, but once it appears, it will reappear on every subsequent
resume.

Any ideas?

--J.

Unable to handle kernel NULL pointer dereference at virtual address 00000004
c02c19d0
*pde = 00000000
Oops: 0000
CPU:    0
EIP:    0010:[<c02c19d0>]    Tainted: P 
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010292
eax: c11acb20   ebx: 00000000   ecx: 000045f1   edx: c110a5ec
esi: 00000004   edi: 00000004   ebp: 00000001   esp: c30e3f2c
ds: 0018   es: 0018   ss: 0018
Process sh (pid: 5263, stackpage=c30e3000)
Stack: c01cb10b 00000000 00000000 00000004 c01cb2de 00000004 00000004 c30e3f7c 
       c30e3f70 00000001 c12ceb40 c01cbb86 00000004 00000000 c705a120 ffffffea 
       00000001 02000000 c026cc86 c026cb60 00000034 00000000 00000000 c01533f6 
Call Trace:    [<c01cb10b>] [<c01cb2de>] [<c01cbb86>] [<c01533f6>] [<c0139aa6>]
  [<c01086df>]
Code: a4 00 75 2c 8b 4b 04 83 c1 fe 89 4d a4 8d 04 8d 00 00 00 00 


>>EIP; c02c19d0 <init_8259A+0/8c>   <=====

>>eax; c11acb20 <_end+eaf368/1050c8a8>
>>ecx; 000045f1 Before first symbol
>>edx; c110a5ec <_end+e0ce34/1050c8a8>
>>esp; c30e3f2c <_end+2de6774/1050c8a8>

Trace; c01cb10b <acpi_system_restore_state+17/68>
Trace; c01cb2de <acpi_suspend+aa/c0>
Trace; c01cbb86 <acpi_system_write_sleep+da/118>
Trace; c01533f6 <proc_file_write+2a/3c>
Trace; c0139aa6 <sys_write+96/f0>
Trace; c01086df <system_call+33/38>

Code;  c02c19d0 <init_8259A+0/8c>
00000000 <_EIP>:
Code;  c02c19d0 <init_8259A+0/8c>   <=====
   0:   a4                        movsb  %ds:(%esi),%es:(%edi)   <=====
Code;  c02c19d1 <init_8259A+1/8c>
   1:   00 75 2c                  add    %dh,0x2c(%ebp)
Code;  c02c19d4 <init_8259A+4/8c>
   4:   8b 4b 04                  mov    0x4(%ebx),%ecx
Code;  c02c19d7 <init_8259A+7/8c>
   7:   83 c1 fe                  add    $0xfffffffe,%ecx
Code;  c02c19da <init_8259A+a/8c>
   a:   89 4d a4                  mov    %ecx,0xffffffa4(%ebp)
Code;  c02c19dd <init_8259A+d/8c>
   d:   8d 04 8d 00 00 00 00      lea    0x0(,%ecx,4),%eax



-------------------------------------------------------
This SF.NET email is sponsored by: eBay
Great deals on office technology -- on eBay now! Click here:
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5

Re: 8259A initialization oops after resume

[Ducrot Bruno]

On Sun, Jun 15, 2003 at 02:00:00PM -0700, Jan Rychter wrote:
> People are getting a particular kind of oops after resuming from S4
> (implemented by swsusp). I'm posting this here, because the problem
> seems not to be specific to swsusp.
> 
> Hopefully some people on this list will know how and why the 8259A needs
> to be initialized upon resuming from S4 and why it fails.
> 
> The oops is reproducible -- one may do a number of suspend/resume cycles
> without it, but once it appears, it will reappear on every subsequent
> resume.

This was done but with S4Bios in mind, not with S4 via swsusp, and is
actually a (bad) hack, since that can not handle all cases.
A better way is done under 2.5.

Actually, there are many others troubles with
'echo 4 > /proc/acpi/sleep' and swsusp (fe, the devices are powered
down twice, via ACPI _and_ swsusp), which is bad.  Better is
to use the swsusp own way to suspend, not via 'echo 4 > /proc/acpi/sleep'
which is IMO broken.

Under 2.5, the situation is different.

-- 
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.Net email is sponsored by: INetU
Attention Web Developers & Consultants: Become An INetU Hosting Partner.
Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission!
INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php

Re: 8259A initialization oops after resume

[Nigel Cunningham]

Hi.

I've got a small patch in the latest swsusp version that disables
reinitialising the 8259A after swsusp. This seems to fix the issues
under 2.4.

Regards,

Nigel

On Tue, 2003-06-17 at 21:58, Ducrot Bruno wrote:
> On Sun, Jun 15, 2003 at 02:00:00PM -0700, Jan Rychter wrote:
> > People are getting a particular kind of oops after resuming from S4
> > (implemented by swsusp). I'm posting this here, because the problem
> > seems not to be specific to swsusp.
> > 
> > Hopefully some people on this list will know how and why the 8259A needs
> > to be initialized upon resuming from S4 and why it fails.
> > 
> > The oops is reproducible -- one may do a number of suspend/resume cycles
> > without it, but once it appears, it will reappear on every subsequent
> > resume.
> 
> This was done but with S4Bios in mind, not with S4 via swsusp, and is
> actually a (bad) hack, since that can not handle all cases.
> A better way is done under 2.5.
> 
> Actually, there are many others troubles with
> 'echo 4 > /proc/acpi/sleep' and swsusp (fe, the devices are powered
> down twice, via ACPI _and_ swsusp), which is bad.  Better is
> to use the swsusp own way to suspend, not via 'echo 4 > /proc/acpi/sleep'
> which is IMO broken.
> 
> Under 2.5, the situation is different.
-- 
Nigel Cunningham
495 St Georges Road South, Hastings 4201, New Zealand

You see, at just the right time, when we were still powerless,
Christ died for the ungodly.
	-- Romans 5:6, NIV.



-------------------------------------------------------
This SF.Net email is sponsored by: INetU
Attention Web Developers & Consultants: Become An INetU Hosting Partner.
Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission!
INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php