Linux on PlayStation 2

Linux on PlayStation 2

[Jun Sun]

I got forwarded by the following link, although in Japanese, which says Sony
is selling a Linux kit for PS2.  Does anybody know more details about this
port?  Specifically, does anyone have the source code?

http://www.jp.playstation.com/linux/

Jun

Re: Linux on PlayStation 2

[Mark Lehrer]

Well, the kernel is GPL'd, so Sony should have some kind of source
code package available if you buy their Linux kit.

Mark


   I got forwarded by the following link, although in Japanese, which says Sony
   is selling a Linux kit for PS2.  Does anybody know more details about this
   port?  Specifically, does anyone have the source code?

   http://www.jp.playstation.com/linux/

Re: Linux on PlayStation 2

[Wayne Gowcher]

Going to link

http://www.jp.playstation.com/linux/dmesg.html

I see that bogo mips are reported as :

Calibrating delay loop... 392.40 BogoMIPS
Estimated CPU clock:  294.240 MHz

I thought the PS2 was a dual pipelined processor and
so maybe would see a higher BogoMIPS figure ? Possibly
double ?

Anybody know the ins and outs of this ?

Wayne

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

Native compile on the target using RedHat 6.1 rpms

[Wayne Gowcher]

Dear All,

I am trying to native compile for mipsel-linux using
the declinuxroot tar file as my NFS and egcs-1.0.3a-1
as my compiler, glibc-2.0.6-5lm as my libraries as
suggested by the README file in
oss.sgi.com/pub/linux/mips/mipsel-linux/README.

But when I try to compile I get the following error :

/usr/bin/ld: /tmp/cca003091.o: uses different e_flags
(0x102) fields than previous modules (0x2)
Bad value: failed to merge target specific data of
file /tmp/cca003091.o
collect2: ld returned 1 exit status
make: *** [pointer] Error 1

When I compile the same program using the RedHat 5.1
rpms / nfs the program compiles to completion OK.

Anybody seen this before ?
Any ideas what I am doing wrong ? missed out ? 

Any help appreciated.

Wayne


__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

Re: Native compile on the target using RedHat 6.1 rpms - Update

[Wayne Gowcher]

I found out what was breaking the compile :

I was setting -mcpu=r4600 -mips2. If I leave this off
everything compiles OK, but now the code isn't
optimized for the processor and does run slower ( of
course).

I also noted that if I use :
-mcpu=r4000 -mips2

I get the error :

/usr/bin/ld: /tmp/cca007591.o: ISA mismatch (-mips3)
with previous modules (-mips1)
/usr/bin/ld: /tmp/cca007591.o: uses different e_flags
(0x102) fields than previous modules (0x2)
Bad value: failed to merge target specific data of
file /tmp/cca007591.o
collect2: ld returned 1 exit status

Which I totally don't understand because I never set
mips3 I set mips2.

I am coming to the conclusion that :

the egcs-1.03a compiler as found on the sgi web site
only supports mips 1.

Can someone confirm or deny this ?

If that is so, how does anyone compile native mips2
code ? You have to build your own compiler / libraries
?

If anyone has / knows of an egcs-1.03a or higher
compiler capapble of compiling mips2 I'd really like
to hear from them.

TIA

Wayne


__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

Re: Native compile on the target using RedHat 6.1 rpms - Update

[Ralf Baechle]

On Mon, Jun 04, 2001 at 12:20:01PM -0700, Wayne Gowcher wrote:

> /usr/bin/ld: /tmp/cca007591.o: ISA mismatch (-mips3)
> with previous modules (-mips1)
> /usr/bin/ld: /tmp/cca007591.o: uses different e_flags
> (0x102) fields than previous modules (0x2)
> Bad value: failed to merge target specific data of
> file /tmp/cca007591.o
> collect2: ld returned 1 exit status
> 
> Which I totally don't understand because I never set
> mips3 I set mips2.

The source itself sets the assembler to mips3 mode.

> I am coming to the conclusion that :
> 
> the egcs-1.03a compiler as found on the sgi web site
> only supports mips 1.

Wrong.

  Ralf

Re: Native compile on the target using RedHat 6.1 rpms - Update

[Wayne Gowcher]

Ralf,

Thanks for the reply.

I was using binutils found in
ftp://oss.sgi.com/pub/linux/mips/mipsel-linux/RedHat-6.1/RPMS/mipsel/

version is 2.10.91-1lm.mipsel.rpm

Do you have a rpm neweer than this or know where I may
get one ?

TIA

Wayne


__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/

Re: Native compile on the target using RedHat 6.1 rpms - Update

[Ralf Baechle]

On Mon, Jun 04, 2001 at 01:53:33PM -0700, Wayne Gowcher wrote:

> I was using binutils found in
> ftp://oss.sgi.com/pub/linux/mips/mipsel-linux/RedHat-6.1/RPMS/mipsel/
> 
> version is 2.10.91-1lm.mipsel.rpm
> 
> Do you have a rpm neweer than this or know where I may
> get one ?

There is one in the Redhat 7.0 directory.  I think we only have big endian
binaries, so you may have to build little endian binaries yourself.

   Ralf

Re: Native compile on the target using RedHat 6.1 rpms - Update

[Houten K.H.C. van (Karel)]

> On Mon, Jun 04, 2001 at 01:53:33PM -0700, Wayne Gowcher wrote:
> 
> > I was using binutils found in
> > ftp://oss.sgi.com/pub/linux/mips/mipsel-linux/RedHat-6.1/RPMS/mipsel/
> > 
> > version is 2.10.91-1lm.mipsel.rpm
> > 
> > Do you have a rpm neweer than this or know where I may
> > get one ?
> 
> There is one in the Redhat 7.0 directory.  I think we only have big endian
> binaries, so you may have to build little endian binaries yourself.
> 

I have my RH 7.0 mipsel (glibc 2.2 based) set nearly ready. Should I
upload what I have for now?

Regards,
Karel.


-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

USB OHCI on 2.4.2 kernel using r4k processor

[Wayne Gowcher]

Dear All,

Has anyone got USB OHCI running on a 2.4.2 kernel with
a mips r4k processor ?

I have successfully used the above with a r3k mips
processor but now after porting the same base code to
a r4k processor the kernel throws an OOPS.

I believe the problem may lay with the writeback cache
because when I disable the d cache on the r4k 2.4.2
kernel USB works fine !

I did some digging around and noted that the working
aic7xxx driver I have on the 2.4.2 r4k makes use of
pci_map_single to handle cache flushing ( ??? please
correct me if I am wrong ) and that this type of code
doesn't make it into the usb code until 2.4.4 of the
sgi linux kernel.

Does anyone have a "back patch" to apply to the 2.4.2
usb code ?

If not I am thinking of attempting this myself and
would welcome any comments / advice about doing this

TIA

Wayne



__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail
http://personal.mail.yahoo.com/

Re: USB OHCI on 2.4.2 kernel using r4k processor

[Pete Popov]

Wayne Gowcher wrote:

> Dear All,
> 
> Has anyone got USB OHCI running on a 2.4.2 kernel with
> a mips r4k processor ?


Yes, on a couple of different boards.  
 
> I have successfully used the above with a r3k mips
> processor but now after porting the same base code to
> a r4k processor the kernel throws an OOPS.
> 
> I believe the problem may lay with the writeback cache
> because when I disable the d cache on the r4k 2.4.2
> kernel USB works fine !
 
> I did some digging around and noted that the working
> aic7xxx driver I have on the 2.4.2 r4k makes use of
> pci_map_single to handle cache flushing ( ??? please
> correct me if I am wrong ) and that this type of code
> doesn't make it into the usb code until 2.4.4 of the
> sgi linux kernel.
 
> Does anyone have a "back patch" to apply to the 2.4.2
> usb code ?
 
> If not I am thinking of attempting this myself and
> would welcome any comments / advice about doing this


Steve at MontaVista did the usb work for mips about six months ago on an 
ITE 8172 board. Since then we've also ported it to the Alchemy Au1000 
board, both running on 2.4.2.  He forwarded all patches to the usb 
maintainer but it looks like it took a bit of time for all the patches 
to be applied.  I don't know if we still have patches against the 2.4.2 
tree.  One thing you can do is next Monday check our ftp site (HHL2.0 
for mips should be released by early next week), grab the generic HHL2.0 
mips kernel, and diff all the usb code against the stock 2.4.2. I don't 
think it would be too bad of a job.

Pete

Re: Native compile on the target using RedHat 6.1 rpms

[Ralf Baechle]

On Mon, Jun 04, 2001 at 10:48:18AM -0700, Wayne Gowcher wrote:

> /usr/bin/ld: /tmp/cca003091.o: uses different e_flags
> (0x102) fields than previous modules (0x2)
> Bad value: failed to merge target specific data of
> file /tmp/cca003091.o
> collect2: ld returned 1 exit status
> make: *** [pointer] Error 1
> 
> When I compile the same program using the RedHat 5.1
> rpms / nfs the program compiles to completion OK.
> 
> Anybody seen this before ?
> Any ideas what I am doing wrong ? missed out ? 

This is a bug in certain binutils versions which gets triggered by
certain kernel configurations.  Upgrade to newer binutils.

   Ralf