xen 2.0.7 on VIA C3 (EPIA)

xen 2.0.7 on VIA C3 (EPIA)

[Chris Brookes]

Hello,

Has anyone tried to get Xen 2.0.7 booting on a VIA EPIA board with a
VIA C3 CPU? I know some of the EPIA mainboards have a CPU with
"features" (4kb pages and no cmov instruction) that require the patch
Adam Sulmicki put together for 2.0.0 and 2.0.1
(http://lists.xensource.com/archives/html/xen-devel/2004-12/txt8D0L7RWimG.txt),
and I have managed to patch, compile and boot a NetBSD dom0
sucessfully with this version.

However, I would really like to try and get 2.0.7 going, so I've
attempted to apply the original patch where it still seemed
appropriate. The patch does manage to get me some output from Xen as
it boots (which is a great step forward from the immediate reboot
without it), however the whole thing halts with "(XEN) Unknown
interrupt" just after "
XEN) System RAM: 509MB (521852kB)".

With a few well placed printk's I was able to determine that the halt
was happening inside init_frametable() in arch/x86/memory.c, at the
point where it calls memset(frame_table, 0, frame_table_size).

Unfortunately my C & ASM skills being what they are (i.e not great)
I'm making  slow progress. Anyone more accomplished been there and
done this, or can lend a hand?

Regards

Chris

cpu0: VIA C3 Samuel 2/Ezra (686-class), 533.36 MHz, id 0x673
cpu0: features 80803035<FPU,DE,TSC,MSR,MTRR>
cpu0: features 80803035<PGE,MMX>
cpu0: features 80803035<3DNOW>
cpu0: I-cache 64 KB 32B/line 4-way, D-cache 64 KB 32B/line 4-way
cpu0: L2 cache 64 KB 32B/line 4-way
cpu0: ITLB 128 4 KB entries 8-way
cpu0: DTLB 128 4 KB entries 8-way
cpu0: 4 page colors

Re: xen 2.0.7 on VIA C3 (EPIA)

[Rafael Ferreira]

Please let me know if you figure how to get Xen 2.07/3 going on your C3.
I have the same problem with a Via C3 Samuel 2. 

- raf

On Sat, 2005-09-03 at 13:05 -0500, Chris Brookes wrote:
> Hello,
> 
> Has anyone tried to get Xen 2.0.7 booting on a VIA EPIA board with a
> VIA C3 CPU? I know some of the EPIA mainboards have a CPU with
> "features" (4kb pages and no cmov instruction) that require the patch
> Adam Sulmicki put together for 2.0.0 and 2.0.1
> (http://lists.xensource.com/archives/html/xen-devel/2004-12/txt8D0L7RWimG.txt),
> and I have managed to patch, compile and boot a NetBSD dom0
> sucessfully with this version.
> 
> However, I would really like to try and get 2.0.7 going, so I've
> attempted to apply the original patch where it still seemed
> appropriate. The patch does manage to get me some output from Xen as
> it boots (which is a great step forward from the immediate reboot
> without it), however the whole thing halts with "(XEN) Unknown
> interrupt" just after "
> XEN) System RAM: 509MB (521852kB)".
> 
> With a few well placed printk's I was able to determine that the halt
> was happening inside init_frametable() in arch/x86/memory.c, at the
> point where it calls memset(frame_table, 0, frame_table_size).
> 
> Unfortunately my C & ASM skills being what they are (i.e not great)
> I'm making  slow progress. Anyone more accomplished been there and
> done this, or can lend a hand?
> 
> Regards
> 
> Chris
> 
> cpu0: VIA C3 Samuel 2/Ezra (686-class), 533.36 MHz, id 0x673
> cpu0: features 80803035<FPU,DE,TSC,MSR,MTRR>
> cpu0: features 80803035<PGE,MMX>
> cpu0: features 80803035<3DNOW>
> cpu0: I-cache 64 KB 32B/line 4-way, D-cache 64 KB 32B/line 4-way
> cpu0: L2 cache 64 KB 32B/line 4-way
> cpu0: ITLB 128 4 KB entries 8-way
> cpu0: DTLB 128 4 KB entries 8-way
> cpu0: 4 page colors
> 
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
> 

Re: xen 2.0.7 on VIA C3 (EPIA)

[Helmut Wollmersdorfer]

Chris Brookes wrote:

> Has anyone tried to get Xen 2.0.7 booting on a VIA EPIA board with a
> VIA C3 CPU? 

No. But I compiled Kernel 2.6.12.4 yesterday with 'CONFIG_M386=y', got 
some warnings from the compiler related to '486 and higher instruction 
set', and will try this kernel today on my EPIAs ME6000.

> With a few well placed printk's I was able to determine that the halt
> was happening inside init_frametable() in arch/x86/memory.c, at the
> point where it calls memset(frame_table, 0, frame_table_size).

IMHO you should get warnings or errors, if you compile for VIAC3_2 (386, 
486, 586 should also work), which lets you find the locations.

But maybe there are other glitches, as I would assume from my (bad) 3 
years experience with this hardware.

> Unfortunately my C & ASM skills being what they are (i.e not great)

Same here;-)

Helmut Wollmersdorfer

Re: [Xen-devel] Re: xen 2.0.7 on VIA C3 (EPIA)

[Chris Brookes]

> No. But I compiled Kernel 2.6.12.4 yesterday with 'CONFIG_M386=y', got
> some warnings from the compiler related to '486 and higher instruction
> set', and will try this kernel today on my EPIAs ME6000.

Let us know how you get on. I have a ME6000 board in another system,
so if you have any success I'll have to swap them out. I think it has
the exact same CPU (aside from speed) though :-/

> IMHO you should get warnings or errors, if you compile for VIAC3_2 (386,
> 486, 586 should also work), which lets you find the locations.

The freeze is occurring in the actual xen kernel, rather than in the
domain 0 linux/netbsd kernel that gets loaded.

> > Unfortunately my C & ASM skills being what they are (i.e not great)
> 
> Same here;-)

The original patch for Xen 2.0.0 wasn't particularly complicated so
I'm hopeful that this won't be a big problem. Well, you have to be
optimistic eh?
=

Re: [Xen-devel] Re: xen 2.0.7 on VIA C3 (EPIA)

[Helmut Wollmersdorfer]

Chris Brookes wrote:
>>No. But I compiled Kernel 2.6.12.4 yesterday with 'CONFIG_M386=y', got
>>some warnings from the compiler related to '486 and higher instruction
>>set', and will try this kernel today on my EPIAs ME6000.

> Let us know how you get on. 

Seems fine. See the graphs at 
http://212.186.65.206/munin/nack.at/via2.nack.at.html (reboot at 17:00).
Just for avoidance of confusing: This is a stock Debian Kernel 2.6.12.4 
plus patches vserver 2.0 and vt1211.

> I have a ME6000 board in another system,

I assume it in the XEN source, if other kernels are running (2.4.24, 
2.6.8 and now 2.6.12.4 work fine).

> so if you have any success I'll have to swap them out. I think it has
> the exact same CPU (aside from speed) though :-/

Yes:
via2:~# cat /proc/cpuinfo
processor       : 0
vendor_id       : CentaurHauls
cpu family      : 6
model           : 7
model name      : VIA Samuel 2
stepping        : 3
cpu MHz         : 599.993
cache size      : 64 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 1
wp              : yes
flags           : fpu de tsc msr cx8 mtrr pge mmx 3dnow
bogomips        : 1187.84

Maybe your chipset is a little bit different. Here is mine:
via2:~# lspci
0000:00:00.0 Host bridge: VIA Technologies, Inc. VT8623 [Apollo CLE266]
0000:00:01.0 PCI bridge: VIA Technologies, Inc. VT8633 [Apollo Pro266 AGP]
0000:00:0d.0 FireWire (IEEE 1394): VIA Technologies, Inc. IEEE 1394 Host 
Controller (rev 80)
0000:00:10.0 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 
1.1 Controller (rev 80)
0000:00:10.1 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 
1.1 Controller (rev 80)
0000:00:10.2 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 
1.1 Controller (rev 80)
0000:00:10.3 USB Controller: VIA Technologies, Inc. USB 2.0 (rev 82)
0000:00:11.0 ISA bridge: VIA Technologies, Inc. VT8235 ISA Bridge
0000:00:11.1 IDE interface: VIA Technologies, Inc. 
VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)
0000:00:11.5 Multimedia audio controller: VIA Technologies, Inc. 
VT8233/A/8235/8237 AC97 Audio Controller (rev 50)
0000:00:12.0 Ethernet controller: VIA Technologies, Inc. VT6102 
[Rhine-II] (rev74)
0000:00:14.0 Ethernet controller: Advanced Micro Devices [AMD] 79c970 
[PCnet32 LANCE] (rev 36)
0000:01:00.0 VGA compatible controller: VIA Technologies, Inc. VT8623 
[Apollo CLE266] integrated CastleRock graphics (rev 03)

> The freeze 

These low power EPIAs are "cool" - no wonder, that they freeze;-)

> is occurring in the actual xen kernel, rather than in the
> domain 0 linux/netbsd kernel that gets loaded.

To be serious EPIAs are known to freeze often, which can have several 
reasons:
- weak power supply
- round "air flow" IDE/DMA cables
- buggy kernel (all before 2.4.23)
- the DMA bug
- the longhaul driver bug
- weak memory
- to warm
- to cold

I have two with exactly same hardware, firmware and software, one 
running since 2 years (only interrupted at power outage or kernel 
change), the other freezing sometimes. A few weeks ago I changed the 
mainboard (had a blown condensator), but after 2 weeks it again freezed. 
Today I will change memory as a last trial.
As a result I have now parts for a third EPIA, which can be expected to 
run some days - good enough for testing XEN kernels;-)

Maybe you know the lock-up thread at 
http://forums.viaarena.com/messageview.aspx?catid=28&threadid=60131&enterthread=y 
with funny esoteric solutions.

Helmut Wollmersdorfer

Re: [Xen-users] Re: Re: xen 2.0.7 on VIA C3 (EPIA)

[Chris Brookes]

> Seems fine. See the graphs at
> http://212.186.65.206/munin/nack.at/via2.nack.at.html (reboot at 17:00).

Can't get to that URL. :-/

> Just for avoidance of confusing: This is a stock Debian Kernel 2.6.12.4
> plus patches vserver 2.0 and vt1211.

vserver?  Not Xen? If you are using Xen, which version is it? Did you patch it?

Chris

Re: [Xen-users] Re: Re: xen 2.0.7 on VIA C3 (EPIA)

[Helmut Wollmersdorfer]

Chris Brookes wrote:
>>Seems fine. See the graphs at
>>http://212.186.65.206/munin/nack.at/via2.nack.at.html (reboot at 17:00).

> Can't get to that URL. :-/

Ooops, this should work:
http://212.186.222.206/munin/nack.at/via2.nack.at.html

>>Just for avoidance of confusing: This is a stock Debian Kernel 2.6.12.4
>>plus patches vserver 2.0 and vt1211.

> vserver?  Not Xen? If you are using Xen, which version is it? Did you patch it?

No, I am not using Xen on the VIAs. But Xen would be an alternative to 
my configuration (vserver-guests on a DRBD device in a heartbeat cluster 
- see my HOWTO on http://linux-vserver.org/Vserver+DRBD).

In reverse you could use vserver instead of Xen, but this will not work 
with non-Linux guests.

Helmut Wollmersdorfer