headers
KIYOHARA Takashi | 3 Jun 2010 08:20
Picon

marvell relation support 20100602

Hi! Frank and all,

I make MV64361 support patch again.  It is working on PegasosII.
It support PCI and Gigabit-Ethernet.  Maybe, this is smarter than
pegasospci.

ftp://ftp.netbsd.org/pub/NetBSD/misc/kiyohara/orion_nas/ofppc-gt-20100602.diff

PCI(gtpci) and Gigabit-Ethernet(mvgbe) already tested by Frank and me.

1. Moreover, call config_found() before than other devices.  And skip.

2. A special bootpath check handling for gt was added in device_register().
   By the way,
   Can our ofwboot handle GbE of MV64361?
   I was not able to confirm bootpath when booting with mvgbe.

Please test this patch.

And I try to support gtidmac(IDMA Controller) now.  Looks like work fine.
But interrupt don't happen...

Thanks,
--
kiyohara
Permalink | Reply |
headers
Frank Wille | 3 Jun 2010 21:56
Picon

Re: marvell relation support 20100602

KIYOHARA Takashi wrote:

> I make MV64361 support patch again.  It is working on PegasosII.
> It support PCI and Gigabit-Ethernet.  Maybe, this is smarter than
> pegasospci.
> [...]
> Please test this patch.

This patch seems to work great! Will continue testing.

--

-- 
Frank Wille
Permalink | Reply |
headers
Matt Thomas | 10 Jun 2010 07:55

Re: marvell relation support 20100602


On Jun 3, 2010, at 12:56 PM, Frank Wille wrote:

> KIYOHARA Takashi wrote:
> 
>> I make MV64361 support patch again.  It is working on PegasosII.
>> It support PCI and Gigabit-Ethernet.  Maybe, this is smarter than
>> pegasospci.
>> [...]
>> Please test this patch.
> 
> This patch seems to work great! Will continue testing.

Maybe for you but for me it failed badly (the dmesg of the previous boot
is at 3am-software.com/dmesg/pegasos.txt ):

NetBSD 5.99.30 (GENERIC) #0: Wed Jun  9 22:41:59 PDT 2010
	matt <at> nixon.home:/u2/netbsd-powerpc/ofppc/obj/sys/arch/ofppc/compile/GENERIC
Model: Pegasos2
total memory = 256 MB
avail memory = 232 MB
bootpath: /pci <at> 80000000/ide <at> C,1/disk <at> 0,0/netbsd
mainbus0 (root)
rtas0 at mainbus0: version 1, entry  <at> pa 0x800000
cpu0 at mainbus0: 7457 (Revision 1.1), ID 0 (primary)
cpu0: HID0
0x84d0c3bc<EMCP,TBEN,HIGH_BAT_EN,NAP,DPM,ICE,DCE,SPD,XBSEN,SGE,BTIC,LRSTK,FOLD,BHT>,
powersave: 1
cpu0: 1000.00 MHz, 512KB L2 cache parity enabled
cpu0: L1 I-cache 32 KB 32B/line, D-cache 32 KB 32B/line
(Continue reading)

Permalink | Reply |
headers
KIYOHARA Takashi | 10 Jun 2010 09:44
Picon

Re: marvell relation support 20100602

Hi! Matt,

From: Matt Thomas <matt <at> 3am-software.com>
Date: Wed, 9 Jun 2010 22:55:26 -0700

> 
> On Jun 3, 2010, at 12:56 PM, Frank Wille wrote:
> 
> > KIYOHARA Takashi wrote:
> > 
> >> I make MV64361 support patch again.  It is working on PegasosII.
> >> It support PCI and Gigabit-Ethernet.  Maybe, this is smarter than
> >> pegasospci.
> >> [...]
> >> Please test this patch.
> > 
> > This patch seems to work great! Will continue testing.
> 
> Maybe for you but for me it failed badly (the dmesg of the previous boot
> is at 3am-software.com/dmesg/pegasos.txt ):
> 
> NetBSD 5.99.30 (GENERIC) #0: Wed Jun  9 22:41:59 PDT 2010
> 	matt <at> nixon.home:/u2/netbsd-powerpc/ofppc/obj/sys/arch/ofppc/compile/GENERIC
> Model: Pegasos2

 :

> cpu0: L3 cache 1 MB 128B/line
> biomask 0 netmask 0 ttymask 0
> Kernelized RAIDframe activated
(Continue reading)

Permalink | Reply |
headers
Matt Thomas | 20 Jun 2010 03:24

Re: marvell relation support 20100602


On Jun 10, 2010, at 12:44 AM, KIYOHARA Takashi wrote:

> Can you see pci and ethernet as a result?  gt checks these to exist in
> arch/ofppc/pci/gt_mainbus.c::gt_match().
> (in my memory:  /ethernet, /pci <at> 80000000, /pci <at> c0000000)
> 
> Or, can the reason why gt is not matched by your PegasosII be examined?

I have no ethernet at the top level.  Changing match to allow for 
that allows it to come up.  3am-software.com/dmesg/pegasos.txt has
an updated dmesg.
Permalink | Reply |
headers
KIYOHARA Takashi | 20 Jun 2010 08:04
Picon

Re: marvell relation support 20100602

Hi! Matt,

From: Matt Thomas <matt <at> 3am-software.com>
Date: Sat, 19 Jun 2010 18:24:59 -0700

> 
> On Jun 10, 2010, at 12:44 AM, KIYOHARA Takashi wrote:
> 
> > Can you see pci and ethernet as a result?  gt checks these to exist in
> > arch/ofppc/pci/gt_mainbus.c::gt_match().
> > (in my memory:  /ethernet, /pci <at> 80000000, /pci <at> c0000000)
> > 
> > Or, can the reason why gt is not matched by your PegasosII be examined?
> 
> I have no ethernet at the top level.  Changing match to allow for 
> that allows it to come up.  3am-software.com/dmesg/pegasos.txt has
> an updated dmesg.  

I want to know whether to be able to use Gigabit Ethernet with your
PegasosII.  Can you use your GbE on MorphOS?  Moreover, is the MAC
address obtained?  When the MAC address is not obtained, mvgbe is not
matched.

3am-software.com/dmesg/pegasos.txt:
  mvgbec0 at gt0 unit 0 offset 0x0000-0x3fff: Marvell Gigabit Ethernet Controller
  mvgbe at mvgbec0 port 0 not configured
  mvgbe at mvgbec0 port 1 not configured
  mvgbe at mvgbec0 port 2 not configured

Or, can you netboot on SmartFirmware via geth(GbE)?
(Continue reading)

Permalink | Reply |
headers
Matt Thomas | 20 Jun 2010 23:48

Re: marvell relation support 20100602


On Jun 19, 2010, at 11:04 PM, KIYOHARA Takashi wrote:
> 
> I want to know whether to be able to use Gigabit Ethernet with your
> PegasosII.  Can you use your GbE on MorphOS?  Moreover, is the MAC
> address obtained?  When the MAC address is not obtained, mvgbe is not
> matched.

Never had MorphOS.  I think I have an e

> 3am-software.com/dmesg/pegasos.txt:
>  mvgbec0 at gt0 unit 0 offset 0x0000-0x3fff: Marvell Gigabit Ethernet Controller
>  mvgbe at mvgbec0 port 0 not configured
>  mvgbe at mvgbec0 port 1 not configured
>  mvgbe at mvgbec0 port 2 not configured
> 
> Or, can you netboot on SmartFirmware via geth(GbE)?

Since there's no OFW node for it, I don't think so.
Permalink | Reply |
headers
KIYOHARA Takashi | 4 Jun 2010 08:39
Picon

Re: marvell relation support 20100602

Hi!

From: KIYOHARA Takashi <kiyohara <at> kk.iij4u.or.jp>
Date: Thu, 03 Jun 2010 15:20:31 +0900 (JST)

> And I try to support gtidmac(IDMA Controller) now.  Looks like work fine.
> But interrupt don't happen...

hmm.. I gave up.
IDMA Controller Registers is working.  But interrupt don't happen.
I do not know a clear difference between 64xx0 and 64xx1.  64xx1 doesn't
have Interrupt Controller.  Therefore, the interrupt of IDMA Controller
might not be able to be handled.

Thanks,
--
kiyohara
Permalink | Reply |
headers
Frank Wille | 4 Jun 2010 13:28
Picon

Re: marvell relation support 20100602

KIYOHARA Takashi wrote:

>> And I try to support gtidmac(IDMA Controller) now.  Looks like work
>> fine. But interrupt don't happen...
>
> hmm.. I gave up.
> IDMA Controller Registers is working.  But interrupt don't happen.

Ok. Too bad.

I don't know if garbled <at>  is still reading this list, but from my side you
can commit the ofppc changes now.

I also found out why my dmesg is scrolling faster with gtpci: The second pci
bus at 0xc0000000 is attached first. So the graphics card is initialized
much earlier than with pegasospci. I like it. :)

--

-- 
Frank Wille
Permalink | Reply |
headers
KIYOHARA Takashi | 9 Jun 2010 07:04
Picon

Re: marvell relation support 20100602

Hi!

From: KIYOHARA Takashi <kiyohara <at> kk.iij4u.or.jp>
Date: Thu, 03 Jun 2010 15:20:31 +0900 (JST)

> I make MV64361 support patch again.  It is working on PegasosII.
> It support PCI and Gigabit-Ethernet.  Maybe, this is smarter than
> pegasospci.
> 
> ftp://ftp.netbsd.org/pub/NetBSD/misc/kiyohara/orion_nas/ofppc-gt-20100602.diff

Merged.  ;-)
--
kiyohara
Permalink | Reply |
headers
matthew green | 9 Jun 2010 07:25
Picon
Favicon

re: marvell relation support 20100602


> Hi!
> 
> 
> From: KIYOHARA Takashi <kiyohara <at> kk.iij4u.or.jp>
> Date: Thu, 03 Jun 2010 15:20:31 +0900 (JST)
> 
> > I make MV64361 support patch again.  It is working on PegasosII.
> > It support PCI and Gigabit-Ethernet.  Maybe, this is smarter than
> > pegasospci.
> > 
> > ftp://ftp.netbsd.org/pub/NetBSD/misc/kiyohara/orion_nas/ofppc-gt-20100602.diff
> 
> Merged.  ;-)

great job! i'm gonna have to turn on my pegasosII again soon. :)

.mrg.
Permalink | Reply |
headers
Frank Wille | 9 Jun 2010 09:46
Picon

Re: marvell relation support 20100602

On Wed, 09 Jun 2010 15:25:24 +1000
matthew green <mrg <at> eterna.com.au> wrote:

> > > ftp://ftp.netbsd.org/pub/NetBSD/misc/kiyohara/orion_nas/ofppc-gt-20100602.diff
> > 
> > Merged.  ;-)
> 
> 
> great job! i'm gonna have to turn on my pegasosII again soon. :)

Yes, it works fine! Now we have the best and cleanest Pegasos2 port ever
(even better than Linux ;) !

The only feature missing for me is support for RDB partitions, besides MBR.
So I can use NetBSD together with MorphOS and AmigaOS on the same disk.

I already implemented RDB in ofwboot last year, but I was too weak to
find a good, preferable MI, solution for the kernel. I have to motivate
me to work on that again... :)

--

-- 
Frank Wille
Permalink | Reply |
headers
matthew green | 9 Jun 2010 10:03
Picon
Favicon

re: marvell relation support 20100602


> The only feature missing for me is support for RDB partitions, besides MBR.
> So I can use NetBSD together with MorphOS and AmigaOS on the same disk.
> 
> I already implemented RDB in ofwboot last year, but I was too weak to
> find a good, preferable MI, solution for the kernel. I have to motivate
> me to work on that again... :)

sys/kern/subr_disk_rdb.c ? :)  or something with wedges?
Permalink | Reply |
headers
Frank Wille | 9 Jun 2010 12:02
Picon

Re: rdb for ofppc (was: marvell relation support 20100602)

On Wed, 09 Jun 2010 18:03:47 +1000
matthew green <mrg <at> eterna.com.au> wrote:

> > I already implemented RDB in ofwboot last year, but I was too weak
> > to find a good, preferable MI, solution for the kernel. I have to
> > motivate me to work on that again... :)
> 
> sys/kern/subr_disk_rdb.c ? :)

That would be easy, but as bad as subr_disk_mbr.c. It wouldn't allow to
use MBR and RDB together in one kernel.

Optimal would be a solution that allows us to combine all partition table
styles which are needed for a port, with auto-detection. And in the GENERIC
config for ofppc we would just add:

options		MBR_PARTITIONS		# MsDOS MasterBootRecord
options		RDB_PARTITIONS		# AmigaDOS RigidDiskBlock
options		APM_PARTITIONS		# Apple Partition Map
options		RAW_PARTITIONS		# Raw NetBSD disklabel

This could be useful for several ports which are supporting native
partitions besides MBR and disklabel, e.g. sgimips, sparc, macppc,
atari, etc.

But that's quite a huge project for a newbie like me alone. So I would
probably just add a special disksubr.c for ofppc which handles all the
required partition tables, as already done by other ports.

> or something with wedges?
(Continue reading)

Permalink | Reply |

Gmane