Quentin Garnier | 1 Oct 18:09 2006
Picon

Re: Shuttle SN27P2: success, mostly

On Sat, Sep 30, 2006 at 11:10:05PM +0200, Quentin Garnier wrote:
[...]
> I had no interrupt trouble, not a glitch with XFree86 from xsrc.
> 
> Oh, wait, it's not true, I do have one glitch with XFree86.  An annoying
> PITA, actually.  It only happens when I use the "nv" driver, so I can
> still fail back on the "vesa" one.  With the nv driver, firefox triggers
> a bug very quickly that makes the X server loop on handling SIGALRM.  It
> happens with all X versions I could test:  XFree86 from xsrc, X.org from
> pkgsrc, modular X.org built mostly by hand (and that, my friends, was
> one _major_ pain).  I'll see how I can debug that in the next few days.
> 
> First I thought it was tied to Xv, but mplayer didn't trigger the bug,
> and later I tried with a firefox on a remote host, and it just did the
> same.

Issue analysed and "fixed", at least in my tree.

https://bugs.freedesktop.org/show_bug.cgi?id=5102

--

-- 
Quentin Garnier - cube <at> cubidou.net - cube <at> NetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.
Juan RP | 1 Oct 18:14 2006

Re: Shuttle SN27P2: success, mostly

On Sat, 30 Sep 2006 23:10:05 +0200
Quentin Garnier <cube <at> cubidou.net> wrote:

> Hi all,
> 
> Just a quick note about a success story.  It went much better than I
> initially expected.
> 
> The machine is a Shuttle SN27P2, equipped with an AMD Athlon X2 3800+
> (the 89W version, I couldn't get my hands on an EE) and 2GB of RAM.
> 
> The motherboard has a nForce 570 Ultra chipset, I added an NVIDIA
> GeForce 6200LE on the PCIe port.  For the story to be complete, it has
> both a SATA hard drive and a SATA DVD burner.
> 
> dmesg: http://taryn.cubidou.net/~cube/netbsd/sn27p2.dmesg
> 
> I quickly installed an amd64 snapshot and later updated to latest
> -current.
> 
> nfe, azalia, viaide, ohci all work perfectly in my use (the USB part
> is still under-tested, but at least it had no trouble with the hub in
> my monitor and the devices behind it;  all USB1).  I haven't tried
> all the "fancy" stuff like recording from azalia, playing with ehci
> or thinking of a way to use the eSATA port.

It's good to know that NetBSD works in modern hardware :-)

Anyway we are not friendly in power management.... maybe in the
future? :-)
(Continue reading)

paul | 2 Oct 05:38 2006

Any hope for getting nForce 4 MoBo's working?

I'm still having no luck at all with getting my ASUS A8N5X
motherboard
to boot and run in 64-bit mode.  It works fine in i386
32-bit mode, but
continues to suffer from the viaide1:0:0 lost interrupt
syndrome.

There's no way for me to disable ACPI in the BIOS (there's
an entry in
the menu, but greyed out and unselectable) and I've tried
disabling all
unused devices.

I'm happy to help in any way I can, but anything more than
booting up
a kernel and reporting results is probably beyond my limited
skill-set!
:)

_________________________________________

Check your Email accounts at MyEmail.com

Login from home, work, school.  Anywhere!

rudolf | 2 Oct 19:27 2006
Picon

Re: Any hope for getting nForce 4 MoBo's working?

Hi,

paul <at> whooppee.com wrote:
> I'm still having no luck at all with getting my ASUS A8N5X
> motherboard
> to boot and run in 64-bit mode.  It works fine in i386
> 32-bit mode, but
> continues to suffer from the viaide1:0:0 lost interrupt
> syndrome.

I've had the same problem with MSI K8N Neo4-FI. I was lucky - the BIOS 
alowes to turn ioapic off. Then I realized that with an acpi-allowed 
kernel I can let it on (the GENERIC kernel for amd64 has the "acpi0 at 
mainbus0" line commented out (netbsd-4)) without problems.

Regards,

r.

paul | 2 Oct 21:39 2006

Re: Any hope for getting nForce 4 MoBo's working?

Hmmmph, that seems to work a lot better now!  I just
whipped up a GENERIC_ACPI.MP and it got all the way
through booting, up to asking me what to use for a
shell!  :)

Now, I just gotta figure out how to build a iso-image
with this kernel and I should be ready to go.  Thanks
for the clue-by-four.

> Hi,
> 
> paul <at> whooppee.com wrote:
> > I'm still having no luck at all with getting my ASUS
> > A8N5X motherboard
> > to boot and run in 64-bit mode.  It works fine in i386
> > 32-bit mode, but
> > continues to suffer from the viaide1:0:0 lost interrupt
> > syndrome.
> 
> I've had the same problem with MSI K8N Neo4-FI. I was
> lucky - the BIOS  alowes to turn ioapic off. Then I
> realized that with an acpi-allowed  kernel I can let it on
> (the GENERIC kernel for amd64 has the "acpi0 at  mainbus0"
> line commented out (netbsd-4)) without problems.
> 
> Regards,
> 
> r.

_________________________________________
(Continue reading)

paul | 2 Oct 22:52 2006

Changing which kernel goes into an iso-image

OK, I'm probably dense, but I can't figure out
where in the process the specific kernel to use
gets selected.  There are multiple INSTALL-*
kernels for amd64, but no obvious way to tell
"make iso-image" which one to use.

Anyone got a clue-by-four?

_________________________________________

Check your Email accounts at MyEmail.com

Login from home, work, school.  Anywhere!

Hubert Feyrer | 3 Oct 05:00 2006
Picon

Re: Changing which kernel goes into an iso-image

On Mon, 2 Oct 2006, paul <at> whooppee.com wrote:
> OK, I'm probably dense, but I can't figure out
> where in the process the specific kernel to use
> gets selected.  There are multiple INSTALL-*
> kernels for amd64, but no obvious way to tell
> "make iso-image" which one to use.

Easy: you can't.
Check out src/etc/etc.amd64/Makefile.inc, you'll find MKISOFS_FLAGS there, 
which will specify a floppy image to take. If you want to change the 
kernel, you have to change the floppy (either by using a different image, 
or by putting a different kernel on the boot-big.fs floppy).

Funny enough, I had replacements for the 'iso-image' target for 
src/distrib/{amd64,i386}, but I seem to have lost them. It shouldn't be 
too hard to redo, though...

  - Hubert

paul | 3 Oct 05:49 2006

Re: Changing which kernel goes into an iso-image


> Easy: you can't.
> Check out src/etc/etc.amd64/Makefile.inc, you'll find
> MKISOFS_FLAGS there,  which will specify a floppy image to
> take. If you want to change the  kernel, you have to
> change the floppy (either by using a different image,  or
> by putting a different kernel on the boot-big.fs floppy).

Hmmm, sounds like maybe this could use some work?  :)

> Funny enough, I had replacements for the 'iso-image'
> target for  src/distrib/{amd64,i386}, but I seem to have
> lost them. It shouldn't be  too hard to redo, though...

Well, for now I'll just do it the dirty way - build a kernel
with the right name but with ACPI and MP turned on.

_________________________________________

Check your Email accounts at MyEmail.com

Login from home, work, school.  Anywhere!

Frank van der Linden | 3 Oct 10:20 2006
Picon

Re: Any hope for getting nForce 4 MoBo's working?

On Sun, Oct 01, 2006 at 08:38:01PM -0700, paul <at> whooppee.com wrote:
> I'm still having no luck at all with getting my ASUS A8N5X
> motherboard
> to boot and run in 64-bit mode.  It works fine in i386
> 32-bit mode, but
> continues to suffer from the viaide1:0:0 lost interrupt
> syndrome.

With ACPI support in the kernel, I believe that the interrupt problems
should be mostly gone. However, there are still some reports of people
for whom this doesn't solve the problem completely. I have an Asus
A8V-CSM board with a dual core CPU, on which kernels boot up, but
then things get really weird when init(8) is executed: an illegel
instruction trap is produced seemingly exactly when it jumps to
userspace to run init. The %eip already points at the first userspace
instruction, but the code page hasn't been faulted in yet. I checked
the segmentation setup, etc, but everything seems to be ok.

According to all docs, the iret instruction isn't even supposed to
be able to generate a #UD trap, so I'm stumped so far. My last
theory that I need to verify is that it gets an interrupt or trap
for which the gate isn't properly set up.

Anyway, there is a good chance that your interrupt issues will be fixed
using ACPI support in the kernel.

- Frank

Juraj Hladky | 3 Oct 10:27 2006
Picon

Re: Any hope for getting nForce 4 MoBo's working?

Hi Paul,

I have the same board. You can boot and work in amd64 by disabling APIC 
in BIOS.
You may need update your BIOS, my version is 902.

Juraj

paul <at> whooppee.com wrote:
> I'm still having no luck at all with getting my ASUS A8N5X
> motherboard
> to boot and run in 64-bit mode.  It works fine in i386
> 32-bit mode, but
> continues to suffer from the viaide1:0:0 lost interrupt
> syndrome.
>
> There's no way for me to disable ACPI in the BIOS (there's
> an entry in
> the menu, but greyed out and unselectable) and I've tried
> disabling all
> unused devices.
>
> I'm happy to help in any way I can, but anything more than
> booting up
> a kernel and reporting results is probably beyond my limited
> skill-set!
> :)
>
> _________________________________________
>
(Continue reading)


Gmane