linux-apus-devel

compiler version.

[Ken T. <ke...@we...>]

What's the recommended compiler for APUS ?  I use gcc version 2.95.2.

Since I installed helix (Ximian) gnome about 12 months ago, X has been
unstable under 2.2.10 and 2.4.?.

It runs for a while, if I can get it to start without locking up my
computer that is, and will eventually crash, usually with some interrupt
related panic :

 Kernel panic: Exception in kernel pc c00a4fec signal 4 
 Kernel panic: kernel access of bad area pc c0037330 lr c00373cc address  F80B48 tsk X/555

Another two seen a few times are :

 Page fault in interrupt handler
 Floating point in kernel

The back trace usually shows the crash occured during interrupt handling.

KDE used to be OK IIRC, running without X is rock solid.

I suppose all this suggests the problem is with gnome and not with
kernel/compiler combination.

Anyone running gnome without any problem or what are you using ?

Ken.

Re: compiler version.

[Michel <mic...@ii...>]

Ken Tyler wrote:
>=20
> What's the recommended compiler for APUS ?  I use gcc version 2.95.2.

Upgrading to a later 2.95 version (Debian has 2.95.4-pre) might be a good
idea.


> Since I installed helix (Ximian) gnome about 12 months ago, X has been
> unstable under 2.2.10 and 2.4.?.
>=20
> It runs for a while, if I can get it to start without locking up my
> computer that is, and will eventually crash, usually with some interrup=
t
> related panic :
>=20
>  Kernel panic: Exception in kernel pc c00a4fec signal 4
>  Kernel panic: kernel access of bad area pc c0037330 lr c00373cc addres=
s
>  F80B48 tsk X/555
>=20
> Another two seen a few times are :
>=20
>  Page fault in interrupt handler
>  Floating point in kernel

In particular this last one sounds like a potential compiler problem. The=
re
should be no floating point code in the kernel.


> The back trace usually shows the crash occured during interrupt handlin=
g.
>=20
> KDE used to be OK IIRC, running without X is rock solid.
>=20
> I suppose all this suggests the problem is with gnome and not with
> kernel/compiler combination.

But how do you imagine GNOME would cause these crashes? If anything from =
user
space, I'd rather suspect the X server.


--=20
Earthling Michel D=E4nzer (MrCooper)    \   Debian GNU/Linux (powerpc) de=
veloper
CS student, Free Software enthusiast   \        XFree86 and DRI project m=
ember

Re: compiler version.

[Ken T. <ke...@we...>]

On Wed, 16 May 2001, Michel D=E4nzer wrote:

Hello,

> Upgrading to a later 2.95 version (Debian has 2.95.4-pre) might be a good
> idea.

OK, will try that.

> > Another two seen a few times are :
> >=20
> >  Page fault in interrupt handler
> >  Floating point in kernel
>=20
> In particular this last one sounds like a potential compiler problem. The=
re
> should be no floating point code in the kernel.

I've never seen anything similar not running X, don't rmemember my box
stopping in the last 12 months when not running X, I would have thought
something in that time would have triggered similar panics. =20

> But how do you imagine GNOME would cause these crashes? If anything
> from user space, I'd rather suspect the X server.

I've tried several X servers, off CD, debian, I just compiled X 4.0.2
which works (with the slow logout screen and no SHAPE), they all run.
But I still get the same crashes with all of them.

When it does crash my computer is locked up solid, no heartbeat and no
sysreq which shouldn't happen from user space either.

The task is always X related.

TASK =3D c0b24000[544] 'X' mm->pgd c0d17000 Last syscall: 146
TASK =3Dc1db2000[547] 'gnome-smproxy' mm->pgd c1e61000 Last syscall: 54
TASK =3D c17c4000[576] 'gmc' mm->pgd c1a1c000 Last syscall: 13

Thanks,

Ken.

Re: compiler version.

[Michel <mic...@ii...>]

Ken Tyler wrote:
> 
> On Wed, 16 May 2001, Michel Dänzer wrote:
> 
> > But how do you imagine GNOME would cause these crashes? If anything
> > from user space, I'd rather suspect the X server.
> 
> I've tried several X servers, off CD, debian, I just compiled X 4.0.2
> which works (with the slow logout screen and no SHAPE), they all run.
> But I still get the same crashes with all of them.

The SHAPE extension is probably missing because Load "extmod" isn't in the
Module Section, and the slow GNOME logout should be fixed as of 4.0.99.2 .


> When it does crash my computer is locked up solid, no heartbeat and no
> sysreq which shouldn't happen from user space either.
> 
> The task is always X related.
> 
> TASK = c0b24000[544] 'X' mm->pgd c0d17000 Last syscall: 146
> TASK =c1db2000[547] 'gnome-smproxy' mm->pgd c1e61000 Last syscall: 54
> TASK = c17c4000[576] 'gmc' mm->pgd c1a1c000 Last syscall: 13

I doubt the task is very informative other than the fact that you see both the
X server and clients, which IMHO points to a kernel problem, be it compiler or
code related. It's just the process in the time slice of which the panic
happens.

Does ksymoops give any useful information?


-- 
Earthling Michel Dänzer (MrCooper)    \   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \        XFree86 and DRI project member

Re: compiler version.

[Ken T. <ke...@we...>]

On Thu, 17 May 2001, Michel D=E4nzer wrote:

> Ken Tyler wrote:

> The SHAPE extension is probably missing because Load "extmod" isn't in th=
e
> Module Section, and the slow GNOME logout should be fixed as of 4.0.99.2 =
=2E

I'll look into it, I don't think myv XF86Config has a module section...
I don't understand that version number, is it < 4.0.2 ?  I just got the
source from ftp.xfree86.org and did make World.

> Does ksymoops give any useful information?

Just that the crash is usually during interrupt processing, varies a bit,
here are three from my Amiga side dmesg stack trace program.

0xC00895B0 (0xC0089578 + 0x00000038) ide_end_request
0xC0093308 (0xC00931F4 + 0x00000114) read_intr
0xC008B5D4 (0xC008B458 + 0x0000017C) ide_intr
0xC0044798 (0xC0044704 + 0x00000094) amiga_do_irq_list
0xC004522C (0xC0045170 + 0x000000BC) cia_handler
0xC0045B90 (0xC0045B2C + 0x00000064) process_int
0xC0142090 (0xC0142080 + 0x00000010) apus_process_int
0xC000E2F4 (0xC000E2AC + 0x00000048) apus_do_IRQ
0xC0006700 (0xC00066C0 + 0x00000040) do_IRQ
0xC0003C98 (0xC0003B04 + 0x00000194) _switch
0x00000000 (0xC0000000 + 0x40000000) _start

0xC00373CC (0xC00372EC + 0x000000E0) send_sigio
0xC0037440 (0xC00373F0 + 0x00000050) kill_fasync
0xC00ACA6C (0xC00AC8BC + 0x000001B0) mouse_interrupt
0xC0044798 (0xC0044704 + 0x00000094) amiga_do_irq_list
0xC00E142C (0xC00E1284 + 0x000001A8) amifb_interrupt
0xC0045B90 (0xC0045B2C + 0x00000064) process_int
0xC0142090 (0xC0142080 + 0x00000010) apus_process_int
0xC000E2F4 (0xC000E2AC + 0x00000048) apus_do_IRQ
0xC0006700 (0xC00066C0 + 0x00000040) do_IRQ
0xC0003C98 (0xC0003B04 + 0x00000194) _switch

0xC0012F48 (0xC0012E9C + 0x000000AC) update_process_times
0xC00132D4 (0xC0012F60 + 0x00000374) timer_bh
0xC001BB24 (0xC001BAA4 + 0x00000080) do_bottom_half
0xC0003D00 (0xC0003CE0 + 0x00000020) lost_irq_ret

Ken.

Re: compiler version.

[Michel <mic...@ii...>]

Ken Tyler wrote:
> 
> On Thu, 17 May 2001, Michel Dänzer wrote:
> 
> > Ken Tyler wrote:
> 
> > The SHAPE extension is probably missing because Load "extmod" isn't in the
> > Module Section, and the slow GNOME logout should be fixed as of 4.0.99.2 .
> 
> I'll look into it, I don't think myv XF86Config has a module section...

Then create one containing at least this:

Section "Module"
    Load        "extmod"
EndSection

If you don't use a font server, you'll also have to load font renderers here,
so I doubt you don't have it yet. Look at the examples in the Contrib section.


> I don't understand that version number, is it < 4.0.2 ?  I just got the
> source from ftp.xfree86.org and did make World.

Nope, it's a snapshot which will become 4.1.0, the next major release. You'd
have to get it from CVS. Or there are RPMs somewhere.


-- 
Earthling Michel Dänzer (MrCooper)    \   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \        XFree86 and DRI project member