2.4.6-ac5 not booting on highmem machine

2.4.6-ac5 not booting on highmem machine

[Jeff Lessem]

I have an 8 processor PIII Xeon machine with 8GB of ram.  The recent
-ac5 patch does not boot, panicking immediately after
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
is printed on the console.  The following ksymoops is printed.  If any
more information would be useful in tracking down this error, please
let me know.

ksymoops 2.4.1 on i686 2.4.6.  Options used
     -v /home/lessem/linux/linux-ac/vmlinux (specified)
     -K (specified)
     -L (specified)
     -o /lib/modules/2.4.6-ac5/ (specified)
     -m /boot/System.map-2.4.6-ac5 (specified)

No modules in ksyms, skipping objects
Unable to handle kernel paging request at virtual address f8801000
c024fab6
*pde = 00000000
Oops: 0000
CPU:    0
EIP:    0010:[<c024fab6>]
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010203
eax: f8800000   ebx: f88000e9   ecx: 03ff810c   edx: 000006f0
esi: 00000f17   edi: c0203816   ebp: ecc3ff80   esp: c9cbbfbc
ds: 0018   es: 0018   ss: 0018
Process swapper (pid: 1, stackpage=c9cbb000)
Stack: c027f70c c024bfc0 00000000 0008e000 44000000 03ff80e9 00000024 c024c884
       c9cba000 c024c8c2 c0105075 00010f00 c024bfc0 c01054fc 00000000 00000078
       00098700
Call Trace: [<c0105075>] [<c01054fc>]
Code: 8a 04 1e 00 44 24 13 46 39 ee 72 f4 80 7c 24 13 00 74 08 53

>>EIP; c024fab6 <sbf_init+ee/198>   <=====
Trace; c0105075 <init+29/154>
Trace; c01054fc <kernel_thread+28/38>
Code;  c024fab6 <sbf_init+ee/198>
00000000 <_EIP>:
Code;  c024fab6 <sbf_init+ee/198>   <=====
   0:   8a 04 1e                  mov    (%esi,%ebx,1),%al   <=====
Code;  c024fab9 <sbf_init+f1/198>
   3:   00 44 24 13               add    %al,0x13(%esp,1)
Code;  c024fabd <sbf_init+f5/198>
   7:   46                        inc    %esi
Code;  c024fabe <sbf_init+f6/198>
   8:   39 ee                     cmp    %ebp,%esi
Code;  c024fac0 <sbf_init+f8/198>
   a:   72 f4                     jb     0 <_EIP>
Code;  c024fac2 <sbf_init+fa/198>
   c:   80 7c 24 13 00            cmpb   $0x0,0x13(%esp,1)
Code;  c024fac7 <sbf_init+ff/198>
  11:   74 08                     je     1b <_EIP+0x1b> c024fad1 <sbf_init+109/198>
Code;  c024fac9 <sbf_init+101/198>
  13:   53                        push   %ebx

Re: 2.4.6-ac5 not booting on highmem machine

[Randy.Dunlap]

Hi-

Looks like the bootflag code checksum loop is trying to
access across a page boundary.

Alan et al- is there a good trick to get around this?

Alan- should bootflag.o depend on CONFIG_PNPBIOS, and not
just the one line in it that sets the PNPOS flag?
I.e. why execute sbf_* if not CONFIG_PNPBIOS?
I'll generate a patch for this if you like.

Jeff, can you post (or send me) a good boot log (just the
E820 memory mapping part of it will do).

Do you have the ACPI "pmtools"?  Getting an ACPI tables dump
would be useful.
[http://developer.intel.com/technology/iapc/acpi/downloads.htm]

~Randy


Jeff Lessem wrote:
> 
> I have an 8 processor PIII Xeon machine with 8GB of ram.  The recent
> -ac5 patch does not boot, panicking immediately after
> Linux NET4.0 for Linux 2.4
> Based upon Swansea University Computer Society NET3.039
> is printed on the console.  The following ksymoops is printed.  If any
> more information would be useful in tracking down this error, please
> let me know.
> 
> ksymoops 2.4.1 on i686 2.4.6.  Options used
>      -v /home/lessem/linux/linux-ac/vmlinux (specified)
>      -K (specified)
>      -L (specified)
>      -o /lib/modules/2.4.6-ac5/ (specified)
>      -m /boot/System.map-2.4.6-ac5 (specified)
> 
> No modules in ksyms, skipping objects
> Unable to handle kernel paging request at virtual address f8801000
> c024fab6
> *pde = 00000000
> Oops: 0000
> CPU:    0
> EIP:    0010:[<c024fab6>]
> Using defaults from ksymoops -t elf32-i386 -a i386
> EFLAGS: 00010203
> eax: f8800000   ebx: f88000e9   ecx: 03ff810c   edx: 000006f0
> esi: 00000f17   edi: c0203816   ebp: ecc3ff80   esp: c9cbbfbc
> ds: 0018   es: 0018   ss: 0018
> Process swapper (pid: 1, stackpage=c9cbb000)
> Stack: c027f70c c024bfc0 00000000 0008e000 44000000 03ff80e9 00000024 c024c884
>        c9cba000 c024c8c2 c0105075 00010f00 c024bfc0 c01054fc 00000000 00000078
>        00098700
> Call Trace: [<c0105075>] [<c01054fc>]
> Code: 8a 04 1e 00 44 24 13 46 39 ee 72 f4 80 7c 24 13 00 74 08 53
> 
> >>EIP; c024fab6 <sbf_init+ee/198>   <=====
> Trace; c0105075 <init+29/154>
> Trace; c01054fc <kernel_thread+28/38>
> Code;  c024fab6 <sbf_init+ee/198>
> 00000000 <_EIP>:
> Code;  c024fab6 <sbf_init+ee/198>   <=====
>    0:   8a 04 1e                  mov    (%esi,%ebx,1),%al   <=====
> Code;  c024fab9 <sbf_init+f1/198>
>    3:   00 44 24 13               add    %al,0x13(%esp,1)
> Code;  c024fabd <sbf_init+f5/198>
>    7:   46                        inc    %esi
> Code;  c024fabe <sbf_init+f6/198>
>    8:   39 ee                     cmp    %ebp,%esi
> Code;  c024fac0 <sbf_init+f8/198>
>    a:   72 f4                     jb     0 <_EIP>
> Code;  c024fac2 <sbf_init+fa/198>
>    c:   80 7c 24 13 00            cmpb   $0x0,0x13(%esp,1)
> Code;  c024fac7 <sbf_init+ff/198>
>   11:   74 08                     je     1b <_EIP+0x1b> c024fad1 <sbf_init+109/198>
> Code;  c024fac9 <sbf_init+101/198>
>   13:   53                        push   %ebx

> -