(Message week:9450)
X-url: http://www.isi.edu/~johnh/
To: faith@cs.unc.edu, linux-laptop@vger.rutgers.edu
Subject: APM on Gateway Solo 9100?
Date: Wed, 30 Jul 1997 18:12:32 -0700
From: John Heidemann <johnh@dash.isi.edu>


I recently got a Gateway Solo 9100.
Although the box claims to support APM 1.1,
it seems to crash Linux kernels which include APM support.
(I've attached details below.)

Has anyone had any experience with Linux/APM on a Gateway Solo 9100?
They worked wonderfully on a Gateway Solo 2100 we had recently.

Alternatively, are there any hints about debugging this sort of thing?
I've tried visually tracing the execution into the BIOS code
but my x86 isn't that great.

Strangely enough, APM seems to run correctly under FreeBSD
on an identical box.

Thanks for any help,
   -John Heidemann


The crash:

boot works until APM initialization starts, then I get:
APM BIOS version 1.2 flags 0x03 (Driver version 1.2)
Entry f000:c6af  cseg16 f000 dseg 40
  cseg len ffff, dseg len ffff
  Connection version 1.1

general protection: e1c8
CPU: 0
EIP: 2048:[<00007586>]
EFLAGS: 00010883
eax: 00000001  ebx: 00000001  ecx: 001e54e0  edx: 00000000
esi: 002d9fdc  edi: 00000014  ebp: 002d9f08  esp: 002d9efe
ds: 2050  es: 0000 fs: 0000 gs: 0000  ss: 0018
Process swapper (pid: 1, process nr: 1, stackapage=002d9000)
Stack: c5179b91 9f08c682 20500000 00010000 20489fdc 00019f1c 54e00000 00065e0a
       c6d20000 00002040 90000000 52bc0000 00100019 02460000 00180000 002b0000
       00180000 00180000 00000000 8e140000 a414001d 9fdc002d 0001002d 90000000

Call trace: (goes into infinite loop)
	[<00180000>] [<00180000>] [<00180000>] [<0018ffff>] [<00180000>] [<00180000>] [<00180000>] 
	[<03e4ffaa>] [<00140000>] [<0018002d>]

Code: 9a ff 1e cb e1 c3 9a 8c 1f cb e1 c3 03 57 10 01 04 09 d3 a2

Code which disassembles to:
	0x804943c <str>:        lcall  0x9ac3,0xe1cb1eff
	0x8049443 <str+7>:      movw   %ds,(%edi)
	0x8049445 <str+9>:      lret   
	0x8049446 <str+10>:     loope  0x804940b <_fini+4059>
	0x8049448 <str+12>:     addl   0x10(%edi),%edx
	0x804944b <str+15>:     addl   %eax,(%ecx,%ecx,1)
	0x804944e <str+18>:     shll   %cl,0x0(%edx)


(I actually had to limit the call trace code in die_if_kernel, it seems
to go into an infinite loop.)


Kernel build options:
CONFIG_APM=y
# CONFIG_APM_IGNORE_USER_SUSPEND is not set
CONFIG_APM_DO_ENABLE=y
CONFIG_APM_CPU_IDLE=y
# CONFIG_APM_DISPLAY_BLANK is not set
CONFIG_APM_POWER_OFF=y

(built under RedHat 4.2, linux-2.0.30)