Matt Thomas | 1 Feb 2012 06:15

Pegasos II and its gigabit port


It would be nice if it work instead of relying on the vr0.

The openfirmare doesn't even have nodes for them and thus no
properties in which a mac addr could be stored.

Has anyone gotten to work?  So should I just put in a wm?

Frank Wille | 1 Feb 2012 10:34
Picon

Re: Pegasos II and its gigabit port

On Tue, 31 Jan 2012 21:15:25 -0800
Matt Thomas <matt <at> 3am-software.com> wrote:

> It would be nice if it work instead of relying on the vr0.

It doesn't?

Kiyohara implemented mvgbe(4) support together with gt_mainbus.c in
June 2010. The driver is enabled in the GENERIC config file.

> The openfirmare doesn't even have nodes for them and thus no
> properties in which a mac addr could be stored.

This is not needed as mvgbe(4) reads the MAC address from the chip's
registers.

> Has anyone gotten to work?

It works for me, although I have to admit that the performance is very
bad. It is even much slower than the vr(4) port, so I never use it.

> So should I just put in a wm?

If you need a really fast NIC, yes. ;)

--

-- 
Frank Wille

Frank Wille | 1 Feb 2012 11:14
Picon

Current port status

Unfortunately the status of the ofppc port had become constantly worse
over the last year and it is time to address the most important issues
for 6.0.

Maybe others can write something about the status of the IBM 7044/7046
ports, as I can only speak for the Pegasos 2 here.

New issues for components that worked before:

- (since 5.99.61) Problems after some major changes in the uvm/kmem/malloc
  code. The worst problem had been resolved (matt) but it still cannot mount
  the root file system anymore (/dev/console not found) and crashes.

- (since 5.99.60) genfb(4) has wrong colors. Now it display pink kernel
  messages on a black background, while it was green on white before.
  The normal output is probably black on black now, and unreadable.
  The config files had not been changed.

- (since 5.99.56) The disklabel on MBR partitioned disks is no longer
  correctly recognized. This is probably my fault. Will be fixed as soon
  as possible.

- (since ~5.99.55) Firefox has become unstable. It crashes frequently.
  Tabs and the scrolling wheel stop working. Problem can also be observed
  on macppc. Caused by switching gcc to 4.5.3 last summer?

- (since ~5.99.52) Problems with XOrg's mouse pointer acceleration make
  the pointer jump to the left or right screen border, and the only way
  to get it away from there is by typing "xset m 0". Also on macppc.

(Continue reading)

Frank Wille | 3 Feb 2012 10:39
Picon

Re: Current port status

Here is an update:

On Wed, 1 Feb 2012 11:14:52 +0100
Frank Wille <frank <at> phoenix.owl.de> wrote:

> - (since 5.99.61) Problems after some major changes in the
> uvm/kmem/malloc code. The worst problem had been resolved (matt) but
> it still cannot mount the root file system anymore (/dev/console not
> found) and crashes.

Probably solved with http://www.netbsd.org/~matt/uvm_km.txt.
Needs testing.

> - (since 5.99.60) genfb(4) has wrong colors. Now it display pink
> kernel messages on a black background, while it was green on white
> before. The normal output is probably black on black now, and
> unreadable. The config files had not been changed.
> 
> - (since 5.99.56) The disklabel on MBR partitioned disks is no longer
>   correctly recognized. This is probably my fault. Will be fixed as
> soon as possible.

This point was fixed yesterday, just in time!

> - (since ~5.99.55) Firefox has become unstable. It crashes frequently.
>   Tabs and the scrolling wheel stop working. Problem can also be
> observed on macppc. Caused by switching gcc to 4.5.3 last summer?
> 
> - (since ~5.99.52) Problems with XOrg's mouse pointer acceleration
> make the pointer jump to the left or right screen border, and the
(Continue reading)

Frank Wille | 3 Feb 2012 10:51
Picon

Testers wanted: sysinst and disklabel

Hi,

yesterday I fixed problems with writing a disk label on MBR partitioned
disks.

Also sysinst was reworked to allow both, installation on MBR-partitioned
and RDB-partitioned disks. In the last case you are not allowed to change
of rewrite the disklabel. The NetBSD label is defined by the RDB partitions.

Both types seem to work on my Pegasos2, but I would like to hear whether
sysinst still does its job for IBM 7044/7046 as well.

--

-- 
Frank Wille

matthew green | 5 Feb 2012 00:03
Picon
Favicon

re: Current port status


my pegasosII just crashed with a very recent -current kernel:

mvgbe0: watchdog timeout
panic: kernel diagnostic assertion "m->m_type != MT_FREE" failed: file
"/usr/src/sys/kern/uipc_mbuf.c", line 571
Stopped in pid 375.1 (cvs) at   netbsd:cpu_Debugger+0x10:       lwz     r0, 0x14(r1)
db> bt
0xaa353d60: at vpanic+0x21c
0xaa353d90: at kern_assert+0x68
0xaa353dd0: at m_freem+0x124
0xaa353e00: at ether_input+0x338
0xaa353e40: at mvgbe_intr+0x320
0xaa353e90: at intr_deliver+0x7c
0xaa353ed0: at pic_handle_intr+0x1ac
0xaa353f20: at trapstart+0x684
saved LR(0xfffffffb) is invalid.

i'm not convinced that kmem changes are fully working on ppc.

.mrg.

Matt Thomas | 5 Feb 2012 00:25

Re: Current port status


On Feb 4, 2012, at 3:03 PM, matthew green wrote:

> 
> my pegasosII just crashed with a very recent -current kernel:
> 
> mvgbe0: watchdog timeout
> panic: kernel diagnostic assertion "m->m_type != MT_FREE" failed: file
"/usr/src/sys/kern/uipc_mbuf.c", line 571
> Stopped in pid 375.1 (cvs) at   netbsd:cpu_Debugger+0x10:       lwz     r0, 0x14(r1)
> db> bt
> 0xaa353d60: at vpanic+0x21c
> 0xaa353d90: at kern_assert+0x68
> 0xaa353dd0: at m_freem+0x124
> 0xaa353e00: at ether_input+0x338
> 0xaa353e40: at mvgbe_intr+0x320
> 0xaa353e90: at intr_deliver+0x7c
> 0xaa353ed0: at pic_handle_intr+0x1ac
> 0xaa353f20: at trapstart+0x684
> saved LR(0xfffffffb) is invalid.
> 
> i'm not convinced that kmem changes are fully working on ppc.

These panics happen on 5.99.60 which pre-kmem changes.
Frank Wille | 5 Feb 2012 00:38
Picon

Re: Current port status

On Sun, 05 Feb 2012 10:03:47 +1100
matthew green <mrg <at> eterna.com.au> wrote:

> mvgbe0: watchdog timeout
> panic: kernel diagnostic assertion "m->m_type != MT_FREE" failed: file
> "/usr/src/sys/kern/uipc_mbuf.c", line 571
> Stopped in pid 375.1 (cvs) at

I haven't seen this panic yet. What do you have to do to create it?
Just run cvs on mvgbe0?

--

-- 
Frank Wille

Matt Thomas | 5 Feb 2012 02:25

Re: Current port status


On Feb 4, 2012, at 3:38 PM, Frank Wille wrote:

> On Sun, 05 Feb 2012 10:03:47 +1100
> matthew green <mrg <at> eterna.com.au> wrote:
> 
>> mvgbe0: watchdog timeout
>> panic: kernel diagnostic assertion "m->m_type != MT_FREE" failed: file
>> "/usr/src/sys/kern/uipc_mbuf.c", line 571
>> Stopped in pid 375.1 (cvs) at
> 
> I haven't seen this panic yet. What do you have to do to create it?
> Just run cvs on mvgbe0?

I did it on vr0.


Gmane