Problem with k7 optimizations in 2.4.x?

Problem with k7 optimizations in 2.4.x?

[Moses Mcknight]

Hi,

    I don't know if this is a problem with my hardware setup or config 
settings or what, but whenever I try to run a 2.4.x kernel on my machine 
with k7 optimizations the computer will never fully boot and seems to 
give random errors about being "unable to handle kernel NULL pointer 
dereference at virtual address xxxxxxxx" and other errors.
    This has happened with the debian 2.4.2 package and my own 
compilings of 2.4.3, 2.4.3-ac4, and 2.4.3-ac5.  If I compile it with 586 
optimizations it works fine.
    My hardware is as follows: Athlon/Thunderbird 850 Mhz CPU,
Iwill KK266 Mobo. (uses VIA kt133a chipset and 686b southbridge) with 
the latest BIOS update from fullon3d.com which is supposed to fix the 
data corruption problem, 256 MB SDRAM and IBM DTLA-307060 HD.

Thanks for any info/help.
Moses

Re: Problem with k7 optimizations in 2.4.x?

[Disconnect]

Several of us on the list had the same/similar problems.  AFAIK the only
fix at this point is to run w/ no more than K6 optimizations.

Check the archives for
-> Only 10 MB/sec with via 82c686b (towards the end of the thread)
-> thunderbird 1.2G + kk266 + 2.4.x oops

It seems to be something wierd with the way memory is handled on the iwill
board.

On Fri, 13 Apr 2001, Moses Mcknight did have cause to say:

> Hi,
> 
>     I don't know if this is a problem with my hardware setup or config 
> settings or what, but whenever I try to run a 2.4.x kernel on my machine 
> with k7 optimizations the computer will never fully boot and seems to 
> give random errors about being "unable to handle kernel NULL pointer 
> dereference at virtual address xxxxxxxx" and other errors.
>     This has happened with the debian 2.4.2 package and my own 
> compilings of 2.4.3, 2.4.3-ac4, and 2.4.3-ac5.  If I compile it with 586 
> optimizations it works fine.
>     My hardware is as follows: Athlon/Thunderbird 850 Mhz CPU,
> Iwill KK266 Mobo. (uses VIA kt133a chipset and 686b southbridge) with 
> the latest BIOS update from fullon3d.com which is supposed to fix the 
> data corruption problem, 256 MB SDRAM and IBM DTLA-307060 HD.
> 
> Thanks for any info/help.
> Moses
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
---
   _.-=<Disconnect>=-._
|     dis@sigkill.net    | And Remember...
\  shawn@healthcite.com  / He who controls Purple controls the Universe..
 PGP Key given on Request  Or at least the Purple parts!

-----BEGIN GEEK CODE BLOCK-----
Version: 3.1 [www.ebb.org/ungeek]
GIT/CC/CM/AT d--(-)@ s+:-- a-->? C++++$ ULBS*++++$ P+>+++ L++++>+++++ 
E--- W+++ N+@ o+>$ K? w--->+++++ O- M V-- PS+() PE Y+@ PGP++() t 5--- 
X-- R tv+@ b++++>$ DI++++ D++(+++) G++ e* h(-)* r++ y++
------END GEEK CODE BLOCK------

Re: Problem with k7 optimizations in 2.4.x?

[Alan Cox]

> settings or what, but whenever I try to run a 2.4.x kernel on my machine 
> with k7 optimizations the computer will never fully boot and seems to 
> give random errors about being "unable to handle kernel NULL pointer 
> dereference at virtual address xxxxxxxx" and other errors.

I run the K7 optimisations in 2.4.x-ac built with gcc 2.96-69 or later with no
problems. I've not tested them with egcs-1.1.2. The earlier 2.4.* had problems
on later Athlons which have both fxsave and movntq/sfence. That was fixed a 
while back tho

Re: Problem with k7 optimizations in 2.4.x?

[Disconnect]

On the iwill mobo? (I haven't heard of this problem on other
motherboards.)

I'll grab the latest -ac today and give it a try this evening. 

(Just for reference, the system is the Iwill kk266, 1.2G tbird, 512M
pc133, ata66 drive - everything has AMD-approved stickers and nothing is
overclocked.)

On Fri, 13 Apr 2001, Alan Cox did have cause to say:

> > settings or what, but whenever I try to run a 2.4.x kernel on my machine 
> > with k7 optimizations the computer will never fully boot and seems to 
> > give random errors about being "unable to handle kernel NULL pointer 
> > dereference at virtual address xxxxxxxx" and other errors.
> 
> I run the K7 optimisations in 2.4.x-ac built with gcc 2.96-69 or later with no
> problems. I've not tested them with egcs-1.1.2. The earlier 2.4.* had problems
> on later Athlons which have both fxsave and movntq/sfence. That was fixed a 
> while back tho
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
---
   _.-=<Disconnect>=-._
|     dis@sigkill.net    | And Remember...
\  shawn@healthcite.com  / He who controls Purple controls the Universe..
 PGP Key given on Request  Or at least the Purple parts!

-----BEGIN GEEK CODE BLOCK-----
Version: 3.1 [www.ebb.org/ungeek]
GIT/CC/CM/AT d--(-)@ s+:-- a-->? C++++$ ULBS*++++$ P+>+++ L++++>+++++ 
E--- W+++ N+@ o+>$ K? w--->+++++ O- M V-- PS+() PE Y+@ PGP++() t 5--- 
X-- R tv+@ b++++>$ DI++++ D++(+++) G++ e* h(-)* r++ y++
------END GEEK CODE BLOCK------

Re: Problem with k7 optimizations in 2.4.x?

[Alan Cox]

> On the iwill mobo? (I haven't heard of this problem on other
> motherboards.)

Im not using an Iwill board. I'm using an AMD751/756 based board and an old
'Uncle Fester' reference board.

Re: Problem with k7 optimizations in 2.4.x?

[Disconnect]

[-- Attachment #1: Type: text/plain, Size: 1048 bytes --]

Blew sky high. Booted stock 2.4.3-ac3, K7 optimizations, init=/bin/sh.  
First time it went fine for quite some time (10-15 mins, find / -print,
etc) and then locked up hard on "open".  Second time, it oops'd before
hardlocking (which I copied out by hand.) The oops and decode are
attached.

On Fri, 13 Apr 2001, Alan Cox did have cause to say:

> > On the iwill mobo? (I haven't heard of this problem on other
> > motherboards.)
> 
> Im not using an Iwill board. I'm using an AMD751/756 based board and an old
> 'Uncle Fester' reference board.
> 
---
   _.-=<Disconnect>=-._
|     dis@sigkill.net    | And Remember...
\  shawn@healthcite.com  / He who controls Purple controls the Universe..
 PGP Key given on Request  Or at least the Purple parts!

-----BEGIN GEEK CODE BLOCK-----
Version: 3.1 [www.ebb.org/ungeek]
GIT/CC/CM/AT d--(-)@ s+:-- a-->? C++++$ ULBS*++++$ P+>+++ L++++>+++++ 
E--- W+++ N+@ o+>$ K? w--->+++++ O- M V-- PS+() PE Y+@ PGP++() t 5--- 
X-- R tv+@ b++++>$ DI++++ D++(+++) G++ e* h(-)* r++ y++
------END GEEK CODE BLOCK------

[-- Attachment #2: decode.txt --]
[-- Type: text/plain, Size: 1380 bytes --]

ksymoops 2.3.7 on i686 2.4.3.  Options used
     -v vmlinux (specified)
     -K (specified)
     -L (specified)
     -O (specified)
     -m System.map (specified)
     -1

EIP: c023873a
Call Trace: [<c0238847>] [<c0121295>] [<c0121a3b>] [<c0110d18>] [<c0110bc0>] [<c011be74>] [<c011ce4a>] [<c011d1f5>] [<c010714c>]
Code: 0f 6f 03 0f e7 06 0f 6f 4b 08 0f e7 4e 08 0f 6f 53 10 0f e7
Using defaults from ksymoops -t elf32-i386 -a i386

Trace; c0238847 <mmx_copy_page+37/40>
Trace; c0121295 <do_wp_page+1c5/230>
Trace; c0121a3b <handle_mm_fault+9b/d0>
Trace; c0110d18 <do_page_fault+158/440>
Trace; c0110bc0 <do_page_fault+0/440>
Trace; c011be74 <rm_sig_from_queue+14/20>
Trace; c011ce4a <do_sigaction+ba/120>
Trace; c011d1f5 <sys_rt_sigaction+75/d0>
Trace; c010714c <error_code+34/3c>
Code;  00000000 Before first symbol
00000000 <_EIP>:
Code;  00000000 Before first symbol
   0:   0f 6f 03                  movq   (%ebx),%mm0
Code;  00000003 Before first symbol
   3:   0f e7 06                  movntq %mm0,(%esi)
Code;  00000006 Before first symbol
   6:   0f 6f 4b 08               movq   0x8(%ebx),%mm1
Code;  0000000a Before first symbol
   a:   0f e7 4e 08               movntq %mm1,0x8(%esi)
Code;  0000000e Before first symbol
   e:   0f 6f 53 10               movq   0x10(%ebx),%mm2
Code;  00000012 Before first symbol
  12:   0f e7 00                  movntq %mm0,(%eax)


[-- Attachment #3: oops.txt --]
[-- Type: text/plain, Size: 316 bytes --]

null pointer defref 00000000
EIP: c023873a
pgd entry d9040000 - 00000000
pmd entry d9040000 - 00000000
pmd not present!

Call Trace: [<c0238847>] [<c0121295>] [<c0121a3b>] [<c0110d18>] [<c0110bc0>] [<c011be74>] [<c011ce4a>] [<c011d1f5>] [<c010714c>]
Code: 0f 6f 03 0f e7 06 0f 6f 4b 08 0f e7 4e 08 0f 6f 53 10 0f e7