Sandpoint 3 + MPC7450 + HHL2

Sandpoint 3 + MPC7450 + HHL2

[Neil Wilson]

Hi,

I managed to get the JourneyMan 2 82XX booting up on my Sandpoint3 MPC7450 -
after adding a enable for the TimeBase - however when a ramdisk is mounted
up I get very slow serial response, like several seconds to print a string,
does anyone have some ideas of what I could look at to resolve this, I have
an Abatron bdi etc but a lot of this is new to me so any help would be very
welcome. Thanks.

Neil
--
-------------------------------------------------------
Neil Wilson                 Airspan Communications Ltd.
nwilson@airspan.com         Cambridge House
Tel: 01895-467265           Oxford Road

                            Uxbridge
                            Middlesex
www.airspan.com             UB8 1UN
-------------------------------------------------------


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Sandpoint 3 + MPC7450 + HHL2

[Mark A. Greer]

Neil Wilson wrote:

> Hi,
>
> I managed to get the JourneyMan 2 82XX booting up on my Sandpoint3 MPC7450 -
> after adding a enable for the TimeBase - however when a ramdisk is mounted
> up I get very slow serial response, like several seconds to print a string,
> does anyone have some ideas of what I could look at to resolve this, I have
> an Abatron bdi etc but a lot of this is new to me so any help would be very
> welcome. Thanks.

I'll virtually guarantee you that its an interrupt-related problem.  Either the
routing is wrong or the IRQ isn't enabled or...

Mark


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Sandpoint 3 + MPC7450 + HHL2

[Dan Malek]

Neil Wilson wrote:

> I managed to get the JourneyMan 2 82XX booting up on my Sandpoint3 MPC7450 -

I'm sitting on a bunch of changes for both Sandpoint X3 and 7450.  With
the new cputable stuff in 2.4_devel, I'll stuff the 7450 changes in
there.

The X3 changes are a big hack to OpenPIC, because the X3 uses the
serial interrupt interface of the EPIC.  I'm still working with Motorola
to understand some discrepancies in the way the board is supposed to
work,
and the way it does.  That is, I have some additional hacks that make
the
board work, but it isn't right.

I'll work on all of this over the weekend.

> ...after adding a enable for the TimeBase

So, you have an old version of DINK.  Get it updated as you are probably
missing more than that.

I don't know what Motorola is shipping to their customers.  I have had
many revisions of hardware and silicon, and just recently have the L2
cache working.  I never received a board with a working L3 cache.  I
guess
there are some magic parts required, and they are ending up in the
hands of people more important than me :-).


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: Sandpoint 3 + MPC7450 + HHL2

[Neil Wilson]

Got it working now....

Not sure if these fixes are just Sandpoint3 or MPC7450 or the combination,
but...

1) The openpic_init call in sandpoint_init_IRQ needs the offset argument to
get the SIO int setup.

2) I also needed to add a enable of the TBEN in HID0 (bit 0x04000000) just
to get the kernel to think about booting at all - I put this in the
decompress_kernel function.

Do these sort of changes need pushing into the source tree ?, and if so
would anyone care to explain how a newcommer does this, thanks.

Neil

> -----Original Message-----
> From:	Mark A. Greer [SMTP:mgreer@mvista.com]
> Sent:	Wednesday, June 27, 2001 5:48 PM
> To:	Neil Wilson
> Cc:	'linuxppc-embedded@lists.linuxppc.org'
> Subject:	Re: Sandpoint 3 + MPC7450 + HHL2
>
> Neil Wilson wrote:
>
> > Hi,
> >
> > I managed to get the JourneyMan 2 82XX booting up on my Sandpoint3
> MPC7450 -
> > after adding a enable for the TimeBase - however when a ramdisk is
> mounted
> > up I get very slow serial response, like several seconds to print a
> string,
> > does anyone have some ideas of what I could look at to resolve this, I
> have
> > an Abatron bdi etc but a lot of this is new to me so any help would be
> very
> > welcome. Thanks.
>
> I'll virtually guarantee you that its an interrupt-related problem.
> Either the
> routing is wrong or the IRQ isn't enabled or...
>
> Mark

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Sandpoint 3 + MPC7450 + HHL2

[Mark A. Greer]

Neil Wilson wrote:

> Got it working now....
>
> Not sure if these fixes are just Sandpoint3 or MPC7450 or the combination,
> but...
>
> 1) The openpic_init call in sandpoint_init_IRQ needs the offset argument to
> get the SIO int setup.

Are you sure you want to do this???  The intsenses stuff in sandpoint_setup.c
is the way it is for a very good reason.  Can you send me your patch so I can
take a look?  Also, can you send the switch settings that you are using.

> 2) I also needed to add a enable of the TBEN in HID0 (bit 0x04000000) just
> to get the kernel to think about booting at all - I put this in the
> decompress_kernel function.

I think that your sandpoint-specific boot head.S or somewhere else like that
would be a much better choice than the ppc-generic decompress_kernel routine.

> Do these sort of changes need pushing into the source tree ?, and if so
> would anyone care to explain how a newcommer does this, thanks.

Let's hold off on this until people who have sandpoints can take a look.

Mark


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Sandpoint 3 + MPC7450 + HHL2

[Mark A. Greer]

BTW, what tree are you basing your code off of?
--

"Mark A. Greer" wrote:

> Neil Wilson wrote:
>
> > Got it working now....
> >
> > Not sure if these fixes are just Sandpoint3 or MPC7450 or the combination,
> > but...
> >
> > 1) The openpic_init call in sandpoint_init_IRQ needs the offset argument to
> > get the SIO int setup.
>
> Are you sure you want to do this???  The intsenses stuff in sandpoint_setup.c
> is the way it is for a very good reason.  Can you send me your patch so I can
> take a look?  Also, can you send the switch settings that you are using.
>
> > 2) I also needed to add a enable of the TBEN in HID0 (bit 0x04000000) just
> > to get the kernel to think about booting at all - I put this in the
> > decompress_kernel function.
>
> I think that your sandpoint-specific boot head.S or somewhere else like that
> would be a much better choice than the ppc-generic decompress_kernel routine.
>
> > Do these sort of changes need pushing into the source tree ?, and if so
> > would anyone care to explain how a newcommer does this, thanks.
>
> Let's hold off on this until people who have sandpoints can take a look.
>
> Mark
>


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: Sandpoint 3 + MPC7450 + HHL2

[Neil Wilson]

Yours !
It is the JourneyMan 2 ppc_82xx build.

-----Original Message-----
From: Mark A. Greer
To: Neil Wilson; 'linuxppc-embedded@lists.linuxppc.org'
Sent: 28/06/01 18:13
Subject: Re: Sandpoint 3 + MPC7450 + HHL2

BTW, what tree are you basing your code off of?
--

"Mark A. Greer" wrote:

> Neil Wilson wrote:
>
> > Got it working now....
> >
> > Not sure if these fixes are just Sandpoint3 or MPC7450 or the
combination,
> > but...
> >
> > 1) The openpic_init call in sandpoint_init_IRQ needs the offset
argument to
> > get the SIO int setup.
>
> Are you sure you want to do this???  The intsenses stuff in
sandpoint_setup.c
> is the way it is for a very good reason.  Can you send me your patch
so I can
> take a look?  Also, can you send the switch settings that you are
using.
>
> > 2) I also needed to add a enable of the TBEN in HID0 (bit
0x04000000) just
> > to get the kernel to think about booting at all - I put this in the
> > decompress_kernel function.
>
> I think that your sandpoint-specific boot head.S or somewhere else
like that
> would be a much better choice than the ppc-generic decompress_kernel
routine.
>
> > Do these sort of changes need pushing into the source tree ?, and if
so
> > would anyone care to explain how a newcommer does this, thanks.
>
> Let's hold off on this until people who have sandpoints can take a
look.
>
> Mark
>

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: Sandpoint 3 + MPC7450 + HHL2

[Neil Wilson]

>Are you sure you want to do this???  The intsenses stuff in
>sandpoint_setup.c is the way it is for a very good reason.  Can you send
>me your patch so I can take a look?  Also, can you send the switch
>settings that you are using.

Okay, when I work out how to do patches, I am new at this stuff.  It was
just changing the offset argument value in the function call.

As it is an SP3 the switch settings don't sort of match up to SP2 which I
think is what the build is aimed at in the first place, but the switch
settings are (for the 7450 card) :-

SW2-1:  ON      RCS0 on PCI
SW2-2:  OFF     Map B / CHRP
SW2-3:  ON      MOT MPPMC
SW2-4:  ON      Free agent
SW2-5:  OFF     Local flash is bootable
SW2-6:  OFF     Standard flash selected
SW2-7:  ON      33MHz PCI
SW2-8:  OFF     COP resets MPPMC only

SW1 on the 7450 card and SW1 & 2 on the SP3 motherboard are set to the
Motorola defaults.

>I think that your sandpoint-specific boot head.S or somewhere else like
>that would be a much better choice than the ppc-generic decompress_kernel
>routine.

Yes that would be more sensible.

>Let's hold off on this until people who have sandpoints can take a look.

Okay.

Neil

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: Sandpoint 3 + MPC7450 + HHL2

[Neil Wilson]

>I'm sitting on a bunch of changes for both Sandpoint X3 and 7450.  With
>the new cputable stuff in 2.4_devel, I'll stuff the 7450 changes in
>there.

sounds good to me

>> ...after adding a enable for the TimeBase

>So, you have an old version of DINK.  Get it updated as you are probably
>missing more than that.

I am using dink32 v12.2 of 16Feb2001.  My Abatron sets the TBEN bit, I run &
halt dink32 after getting the dink prompt and TBEN is still set but when I
run the downloaded zImage and halt when it "appears" to hang TBEN has been
cleared.

Neil

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/