xen/i386/machdep.c compiler error

Hi all,

I'm running latest Xen3 code from -current but I have a
problem
building the domU kernel:

/usr/src/sys/arch/xen/i386/machdep.c: In function
'idt_vec_alloc':
/usr/src/sys/arch/xen/i386/machdep.c:2457: internal compiler
error:
Segmentation fault

The kernel is the same than XEN3_DOMU with some options
disabled (COMPAT_*, DEBUG/DIAGNOSTIC, etc).

Any idea?

On Sun, 15 Oct 2006 19:00:12 +0200
Juan RP <juanxtrarom.org> wrote:

> 
> Hi all,
> 
> I'm running latest Xen3 code from -current but I have a
problem
> building the domU kernel:
> 
> /usr/src/sys/arch/xen/i386/machdep.c: In function
'idt_vec_alloc':
> /usr/src/sys/arch/xen/i386/machdep.c:2457: internal
compiler error:
> Segmentation fault
> 
> The kernel is the same than XEN3_DOMU with some options
> disabled (COMPAT_*, DEBUG/DIAGNOSTIC, etc).
> 
> Any idea?

Hmm building the installation kernel:

/usr/src/sys/compat/common/compat_util.c: In function
'emul_flags_translate':
/usr/src/sys/compat/common/compat_util.c:259:
internal compiler error: Segmentation fault

Looks like there's some random corruption somewhere.

For the record, this is with latest sources available for
i386/x86/xen:

NetBSD nocturno 4.99.3 NetBSD 4.99.3 (Nocturno-Xen0) #6: Sun
Oct 15
17:49:05 CEST 2006
juannocturno:/home/juan/build/obj/sys/arch/i386/compile/N
octurno-Xen0
i386

On Sun, Oct 15, 2006 at 07:11:36PM +0200, Juan RP wrote:
> On Sun, 15 Oct 2006 19:00:12 +0200
> Juan RP <juanxtrarom.org> wrote:
> 
> > 
> > Hi all,
> > 
> > I'm running latest Xen3 code from -current but I
have a problem
> > building the domU kernel:
> > 
> > /usr/src/sys/arch/xen/i386/machdep.c: In function
'idt_vec_alloc':
> > /usr/src/sys/arch/xen/i386/machdep.c:2457:
internal compiler error:
> > Segmentation fault
> > 
> > The kernel is the same than XEN3_DOMU with some
options
> > disabled (COMPAT_*, DEBUG/DIAGNOSTIC, etc).
> > 
> > Any idea?
> 
> Hmm building the installation kernel:
> 
> /usr/src/sys/compat/common/compat_util.c: In function
> 'emul_flags_translate':
/usr/src/sys/compat/common/compat_util.c:259:
> internal compiler error: Segmentation fault
> 
> Looks like there's some random corruption somewhere.

It's memtest time for your computer.

-- 
Quentin Garnier - cubecubidou.net - cubeNetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.

On Sun, 15 Oct 2006 19:18:02 +0200
Quentin Garnier <cubecubidou.net> wrote:

> It's memtest time for your computer.

Why should it be a hw memory problem? I don't have such
problems
with NetBSD/i386 and Windows.

On Sun, Oct 15, 2006 at 07:20:28PM +0200, Juan RP wrote:
> On Sun, 15 Oct 2006 19:18:02 +0200
> Quentin Garnier <cubecubidou.net> wrote:
> 
> > It's memtest time for your computer.
> 
> Why should it be a hw memory problem? I don't have such
problems
> with NetBSD/i386 and Windows.

gcc just doesn't randomly segfaults.  The most likely cause
of such a
behaviour really is a hardware problem.  In my experience
such issues
have always come down to that (and it manifested in much,
much stranger
ways sometimes).

I'm not saying it's the cause, but it's likely enough to
have a look at
it first.  Especially if it appeared as you didn't touch the
Xen
hypervisor in a while.

-- 
Quentin Garnier - cubecubidou.net - cubeNetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.

On Sun, 15 Oct 2006 19:00:12 +0200
Juan RP <juanxtrarom.org> wrote:

> 
> Hi all,
> 
> I'm running latest Xen3 code from -current but I have a
problem
> building the domU kernel:
> 
> /usr/src/sys/arch/xen/i386/machdep.c: In function
'idt_vec_alloc':
> /usr/src/sys/arch/xen/i386/machdep.c:2457: internal
compiler error:
> Segmentation fault
> 
> The kernel is the same than XEN3_DOMU with some options
> disabled (COMPAT_*, DEBUG/DIAGNOSTIC, etc).
> 
> Any idea?

Hmm building the installation kernel:

/usr/src/sys/compat/common/compat_util.c: In function
'emul_flags_translate':
/usr/src/sys/compat/common/compat_util.c:259:
internal compiler error: Segmentation fault

Looks like there's some random corruption somewhere.

For the record, this is with latest sources available for
i386/x86/xen:

NetBSD nocturno 4.99.3 NetBSD 4.99.3 (Nocturno-Xen0) #6: Sun
Oct 15
17:49:05 CEST 2006
juannocturno:/home/juan/build/obj/sys/arch/i386/compile/N
octurno-Xen0
i386

On Sun, Oct 15, 2006 at 07:11:36PM +0200, Juan RP wrote:
> On Sun, 15 Oct 2006 19:00:12 +0200
> Juan RP <juanxtrarom.org> wrote:
> 
> > 
> > Hi all,
> > 
> > I'm running latest Xen3 code from -current but I
have a problem
> > building the domU kernel:
> > 
> > /usr/src/sys/arch/xen/i386/machdep.c: In function
'idt_vec_alloc':
> > /usr/src/sys/arch/xen/i386/machdep.c:2457:
internal compiler error:
> > Segmentation fault
> > 
> > The kernel is the same than XEN3_DOMU with some
options
> > disabled (COMPAT_*, DEBUG/DIAGNOSTIC, etc).
> > 
> > Any idea?
> 
> Hmm building the installation kernel:
> 
> /usr/src/sys/compat/common/compat_util.c: In function
> 'emul_flags_translate':
/usr/src/sys/compat/common/compat_util.c:259:
> internal compiler error: Segmentation fault
> 
> Looks like there's some random corruption somewhere.

It's memtest time for your computer.

-- 
Quentin Garnier - cubecubidou.net - cubeNetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.

On Sun, 15 Oct 2006 19:18:02 +0200
Quentin Garnier <cubecubidou.net> wrote:

> It's memtest time for your computer.

Why should it be a hw memory problem? I don't have such
problems
with NetBSD/i386 and Windows.

On Sun, Oct 15, 2006 at 07:20:28PM +0200, Juan RP wrote:
> On Sun, 15 Oct 2006 19:18:02 +0200
> Quentin Garnier <cubecubidou.net> wrote:
> 
> > It's memtest time for your computer.
> 
> Why should it be a hw memory problem? I don't have such
problems
> with NetBSD/i386 and Windows.

gcc just doesn't randomly segfaults.  The most likely cause
of such a
behaviour really is a hardware problem.  In my experience
such issues
have always come down to that (and it manifested in much,
much stranger
ways sometimes).

I'm not saying it's the cause, but it's likely enough to
have a look at
it first.  Especially if it appeared as you didn't touch the
Xen
hypervisor in a while.

-- 
Quentin Garnier - cubecubidou.net - cubeNetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.

On Sun, 15 Oct 2006 19:26:45 +0200
Quentin Garnier <cubecubidou.net> wrote:

> gcc just doesn't randomly segfaults.  The most likely
cause of such a
> behaviour really is a hardware problem.  In my
experience such issues
> have always come down to that (and it manifested in
much, much
> stranger ways sometimes).
> 
> I'm not saying it's the cause, but it's likely enough
to have a look
> at it first.  Especially if it appeared as you didn't
touch the Xen
> hypervisor in a while.

Maybe I'll try, anyway I'm not convinced because I've built
the kernels
on NetBSD/i386 without any problem.

I remember once I had a similar problem with Xen2 and Manuel
fixed it.