Boot 2.6.26 (or similar) on mpc5200 using powerpc

Boot 2.6.26 (or similar) on mpc5200 using powerpc

[Angelo]

[-- Attachment #1: Type: text/plain, Size: 2074 bytes --]

Hi.

Does anyone boots on mpc5200 board, the linux kernel 2.6.26 (or .24 .25) using ARCH=powerpc?

Looking on web i known that it's very different from ppc mode and the right procedure to follow is:

1 - make lite5200b_defconfig ARCH=powerpc CROSS_COMPILE=powerpc-linux-gnu-
2 - make cuImage.lite5200b ARCH=powerpc CROSS_COMPILE=powerpc-linux-gnu-

The output of second command is:
...
  CHK     include/linux/version.h
  CHK     include/linux/utsrelease.h
  CALL    scripts/checksyscalls.sh
  CHK     include/linux/compile.h
  CALL    arch/powerpc/kernel/systbl_chk.sh
  CALL    arch/powerpc/kernel/prom_init_check.sh
arch/powerpc/boot/dtc -O dtb -o arch/powerpc/boot/lite5200b.dtb -b 0  /home/asimmini/Scrivania/linux-2.6.26.3/arch/powerpc/boot/dts/lite5200b.dts
DTC: dts->dtb  on file "/home/asimmini/Scrivania/linux-2.6.26.3/arch/powerpc/boot/dts/lite5200b.dts"
  WRAP    arch/powerpc/boot/cuImage.lite5200b
Image Name:   Linux-2.6.26.3
Created:      Wed Sep  3 15:56:16 2008
Image Type:   PowerPC Linux Kernel Image (gzip compressed)
Data Size:    1597696 Bytes = 1560.25 kB = 1.52 MB
Load Address: 0x00400000
Entry Point:  0x00400550
rm arch/powerpc/boot/lite5200b.dtb 

But, at this point, if i crosscompile in ppc mode i had to modify the Load Address wiht the mkimage tool because the SDRAM at the Bootloader time starts from 0xF8000000, while the uImage has a Load Address and Entry Point at 0x00000000. 
What should i do in powerpc mode??


However, on web i saw that to boot the board it's necessary to load the cuImage and rootfs.uimage on the board with:
=> usb start
=> usb scan
=> fatload usb 0 0x03000000 cuImage.lite5200b
=> fatload usb 0 0x04000000 rootfs.uimage
=> bootm 0x03000000 0x04000000

Does anyone solves this problem?

Thank you in advance.


__________________________________________________
Do You Yahoo!?
Poco spazio e tanto spam? Yahoo! Mail ti protegge dallo spam e ti da tanto spazio gratuito per i tuoi file e i messaggi 
http://mail.yahoo.it 

[-- Attachment #2: Type: text/html, Size: 2861 bytes --]

Re: Boot 2.6.26 (or similar) on mpc5200 using powerpc

[Detlev Zundel]

Hi Angelo,

> Does anyone boots on mpc5200 board, the linux kernel 2.6.26 (or .24 .25) using
> ARCH=powerpc?

Sure, many people do this successfully.

> Looking on web i known that it's very different from ppc mode and the right
> procedure to follow is:
>
> 1 - make lite5200b_defconfig ARCH=powerpc CROSS_COMPILE=powerpc-linux-gnu-

Note that this changes to "make 52xx/lite5200b_defconfig ..." in recent kernels.

> 2 - make cuImage.lite5200b ARCH=powerpc CROSS_COMPILE=powerpc-linux-gnu-

Well, this is debatable.  Here you explicitely request a backward
compatability wrapped uImage (for firmwares being FDT ignorant):

> The output of second command is:
> ...
>   CHK     include/linux/version.h
>   CHK     include/linux/utsrelease.h
>   CALL    scripts/checksyscalls.sh
>   CHK     include/linux/compile.h
>   CALL    arch/powerpc/kernel/systbl_chk.sh
>   CALL    arch/powerpc/kernel/prom_init_check.sh
> arch/powerpc/boot/dtc -O dtb -o arch/powerpc/boot/lite5200b.dtb -b 0  /home/
> asimmini/Scrivania/linux-2.6.26.3/arch/powerpc/boot/dts/lite5200b.dts
> DTC: dts->dtb  on file "/home/asimmini/Scrivania/linux-2.6.26.3/arch/powerpc/
> boot/dts/lite5200b.dts"
>   WRAP    arch/powerpc/boot/cuImage.lite5200b

    ^----  Here the wrapper gets added.

> Image Name:   Linux-2.6.26.3
> Created:      Wed Sep  3 15:56:16 2008
> Image Type:   PowerPC Linux Kernel Image (gzip compressed)
> Data Size:    1597696 Bytes = 1560.25 kB = 1.52 MB
> Load Address: 0x00400000
> Entry Point:  0x00400550
> rm arch/powerpc/boot/lite5200b.dtb
>
> But, at this point, if i crosscompile in ppc mode i had to modify the Load
> Address wiht the mkimage tool because the SDRAM at the Bootloader time starts
> from 0xF8000000, while the uImage has a Load Address and Entry Point at
> 0x00000000.

Well still in newer kernels, uImage has load and entry address 0.  Its
only the cuImage.lite5200b (and I presume other wrapped kernels) that
have this load address.

> What should i do in powerpc mode??

I suggest that you use a recent U-Boot on your board which can
explicitely pass a .dtb to the kernel and use a plain uImage + flat
device tree blob.  I gather this is the best tested combination.  I
myself saw problems with the wrapped image that went away when passing
the device tree explicitely.

So in 2.6.26 I'd advise you to do a "make uImage" + "make lite5200b.dtb"
and use the results in a recent U-Boot.

Cheers
  Detlev

--
DENX Software Engineering GmbH,      MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich,  Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-40 Fax: (+49)-8142-66989-80 Email: dzu@denx.de

Re: Boot 2.6.26 (or similar) on mpc5200 using powerpc

[Angelo]

--- Gio 4/9/08, Detlev Zundel <dzu@denx.de> ha scritto:

> Da: Detlev Zundel <dzu@denx.de>
> Oggetto: Re: Boot 2.6.26 (or similar) on mpc5200 using powerpc
> A: linuxppc-embedded@ozlabs.org
> Data: Gioved=EC 4 settembre 2008, 14:05
> Hi Angelo,
>=20
> > Does anyone boots on mpc5200 board, the linux kernel
> 2.6.26 (or .24 .25) using
> > ARCH=3Dpowerpc?
>=20
> Sure, many people do this successfully.
>=20
> > Looking on web i known that it's very different
> from ppc mode and the right
> > procedure to follow is:
> >
> > 1 - make lite5200b_defconfig ARCH=3Dpowerpc
> CROSS_COMPILE=3Dpowerpc-linux-gnu-
>=20
> Note that this changes to "make
> 52xx/lite5200b_defconfig ..." in recent kernels.

Ok. I use 52xx/lite5200b_defconfig.

>=20
> > 2 - make cuImage.lite5200b ARCH=3Dpowerpc
> CROSS_COMPILE=3Dpowerpc-linux-gnu-
>=20
> Well, this is debatable.  Here you explicitely request a
> backward
> compatability wrapped uImage (for firmwares being FDT
> ignorant):
>=20

Can i use also $ make uImage ... ?! Or only cuImage!?

> > The output of second command is:
> > ...
> >   CHK     include/linux/version.h
> >   CHK     include/linux/utsrelease.h
> >   CALL    scripts/checksyscalls.sh
> >   CHK     include/linux/compile.h
> >   CALL    arch/powerpc/kernel/systbl_chk.sh
> >   CALL    arch/powerpc/kernel/prom_init_check.sh
> > arch/powerpc/boot/dtc -O dtb -o
> arch/powerpc/boot/lite5200b.dtb -b 0  /home/
> >
> asimmini/Scrivania/linux-2.6.26.3/arch/powerpc/boot/dts/lite5200b.dts
> > DTC: dts->dtb  on file
> "/home/asimmini/Scrivania/linux-2.6.26.3/arch/powerpc/
> > boot/dts/lite5200b.dts"
> >   WRAP    arch/powerpc/boot/cuImage.lite5200b
>=20
>     ^----  Here the wrapper gets added.
>=20
> > Image Name:   Linux-2.6.26.3
> > Created:      Wed Sep  3 15:56:16 2008
> > Image Type:   PowerPC Linux Kernel Image (gzip
> compressed)
> > Data Size:    1597696 Bytes =3D 1560.25 kB =3D 1.52 MB
> > Load Address: 0x00400000
> > Entry Point:  0x00400550
> > rm arch/powerpc/boot/lite5200b.dtb
> >
> > But, at this point, if i crosscompile in ppc mode i
> had to modify the Load
> > Address wiht the mkimage tool because the SDRAM at the
> Bootloader time starts
> > from 0xF8000000, while the uImage has a Load Address
> and Entry Point at
> > 0x00000000.
>=20
> Well still in newer kernels, uImage has load and entry
> address 0.  Its
> only the cuImage.lite5200b (and I presume other wrapped
> kernels) that
> have this load address.
>=20
> > What should i do in powerpc mode??
>=20
> I suggest that you use a recent U-Boot on your board which
> can
> explicitely pass a .dtb to the kernel and use a plain
> uImage + flat
> device tree blob.  I gather this is the best tested
> combination.  I
> myself saw problems with the wrapped image that went away
> when passing
> the device tree explicitely.
>=20
> So in 2.6.26 I'd advise you to do a "make
> uImage" + "make lite5200b.dtb"
> and use the results in a recent U-Boot.

But lite5200b.dtb doesn't exist. So i'm looking for it and i known that i h=
ad to generate it using this command:

$ arch/powerpc/boot/dtc -b0 -Idts -Odtb -V16 arch/powerpc/boot/dts/lite5200=
b.dts > lite5200b.dtb
I'm wrong?!

So at this point i've got dtb, but somwhere i saw that it's necessary to do=
wnload lite5200.dtb to the system...

And at this point I try to boot the board:

1. usb start
2. usb scan
3. fatload usb 0 0xfc000000 cuImage.lite5200b
(3. fatload usb 0 0xfc000000 uImage - if i use make uImage)
4. fatload usb 0 0xfd000000 rootfs.uimage
5. fatload usb 0 0x???????? lite5200b.dtb
6. bootm 0xfc000000 0x????????

However my uboot version is 1.2.0.=20

Many thanks.
=0A=0A__________________________________________________=0ADo You Yahoo!?=
=0APoco spazio e tanto spam? Yahoo! Mail ti protegge dallo spam e ti da tan=
to spazio gratuito per i tuoi file e i messaggi =0Ahttp://mail.yahoo.it 

powerpc405 versus powerpc440

[Wood, Robert (GE EntSol, Intelligent Platforms)]

We are migrating to the Xilinx Virtex 5 devices and that means migrating
our embedded software from the PowerPC405 to PowerPC440.

Are there any gotchas we should be aware of, or will this be simple case
of S/W migration :-)

Robert Wood
Senior Engineer
GE Fanuc Intelligent Platforms
=20
T +613 749 9241 x270
F +613 749 9461
E robert.wood@gefanuc.com
www.ge.com
=20
5430 Canotek Road
Ottawa, Ontario
Canada K1J 9G2
General Electric Company

Re: Boot 2.6.26 (or similar) on mpc5200 using powerpc

[Detlev Zundel]

Hi Angelo,

>> > 2 - make cuImage.lite5200b ARCH=powerpc
>> CROSS_COMPILE=powerpc-linux-gnu-
>> 
>> Well, this is debatable.  Here you explicitely request a
>> backward
>> compatability wrapped uImage (for firmwares being FDT
>> ignorant):
>> 
>
> Can i use also $ make uImage ... ?! Or only cuImage!?

Maybe I was unclear - "make uImage" generates a regular kernel image for
an FDT aware U-Boot, whereas "make cuImage" wraps the FDT blob _into_
the cuImage so that even old non-FDT aware U-Boots can boot the kernel.

But this option of course uses a wrapper for the kernel which sometimes
exhibits behaviour different from what one sees when booting a regular
uImage + dtb.  Of course these are bugs and should be fixed.

>> So in 2.6.26 I'd advise you to do a "make
>> uImage" + "make lite5200b.dtb"
>> and use the results in a recent U-Boot.
>
> But lite5200b.dtb doesn't exist. So i'm looking for it and i known
> that i had to generate it using this command:
>
> $ arch/powerpc/boot/dtc -b0 -Idts -Odtb -V16 arch/powerpc/boot/dts/lite5200b.dts > lite5200b.dtb

> I'm wrong?!

Not really, but you could also simply follow my advise - use a current
kernel, trust me, type "make lite5200b.dtb" and rejoice.

> So at this point i've got dtb, but somwhere i saw that it's necessary
> to download lite5200.dtb to the system...
>
> And at this point I try to boot the board:
>
> 1. usb start
> 2. usb scan
> 3. fatload usb 0 0xfc000000 cuImage.lite5200b
> (3. fatload usb 0 0xfc000000 uImage - if i use make uImage)
> 4. fatload usb 0 0xfd000000 rootfs.uimage
> 5. fatload usb 0 0x???????? lite5200b.dtb
> 6. bootm 0xfc000000 0x????????
>
> However my uboot version is 1.2.0. 

Uh right, U-Boot for lite5200 has fdt support since 1.1.6, but I still
advise to use a more recent U-Boot as a lot changed in this area.  If
you have a recent U-Boot, you'll need to use the three parameter version
of bootm - i.e. <kernel addr> <ramdisk addr> <fdt addr>.  A "-" for the
ramdisk is ok if you do not intend to use a ramdisk.

Cheers
  Detlev

-- 
Every generation laughs at the old fashions, but follows religiously
the new.
                                        -- Henry David Thoreau
--
DENX Software Engineering GmbH,      MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich,  Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-40 Fax: (+49)-8142-66989-80 Email: dzu@denx.de

Re: powerpc405 versus powerpc440

[Josh Boyer]

On Thu, Sep 04, 2008 at 09:28:01AM -0400, Wood, Robert (GE EntSol, Intelligent Platforms) wrote:
>We are migrating to the Xilinx Virtex 5 devices and that means migrating
>our embedded software from the PowerPC405 to PowerPC440.
>
>Are there any gotchas we should be aware of, or will this be simple case
>of S/W migration :-)

Erm.  405 and 440 have entirely different MMUs.  405 has real-mode, and 440
is BookE and therefore has the MMU always on.

That is mostly an impact to the kernel itself and not to userspace.  But it
can be annoying if you aren't used to dealing with BookE style CPUs.

josh

Re: powerpc405 versus powerpc440

[Grant Likely]

On Thu, Sep 04, 2008 at 11:07:25AM -0400, Josh Boyer wrote:
> On Thu, Sep 04, 2008 at 09:28:01AM -0400, Wood, Robert (GE EntSol, Intelligent Platforms) wrote:
> >We are migrating to the Xilinx Virtex 5 devices and that means migrating
> >our embedded software from the PowerPC405 to PowerPC440.
> >
> >Are there any gotchas we should be aware of, or will this be simple case
> >of S/W migration :-)
> 
> Erm.  405 and 440 have entirely different MMUs.  405 has real-mode, and 440
> is BookE and therefore has the MMU always on.
> 
> That is mostly an impact to the kernel itself and not to userspace.  But it
> can be annoying if you aren't used to dealing with BookE style CPUs.

As Josh said, the MMUs are entirely different, but it is mostly isolated
to impacting the kernel.  The mainline kernel already has basic support
for the Virtex5 440 in the latest release.  Also, the xilinx git server
(git.xilinx.com) has a tree with drivers for a bunch of the xilinx
devices that aren't quite ready for mainline yet.

g.