some questions about XUP and the generation of ace file

windstorm likunarmstrong at gmail.com
Sat Aug 11 02:37:50 EST 2007


1 it's strange. the ns16550.c do exist in the arch/ppc/boot/
directory, not arch/ppc/boot/common. I just download the source
yesterday.

2 Can you show me some reference about the "small bootloader"? Or any
web material about it?

2007/8/10, Grant Likely <grant.likely at secretlab.ca>:
> On 8/10/07, windstorm <likunarmstrong at gmail.com> wrote:
> > Hello all:
> >
> > I am trying to transfer the Linux onto XUP board. The kernel source I
> > used is from the git trees from andrei konovalov or grant likely. But
> > I encounter two questions now.
> >
> > 1 Generally, I should correct the kernel source file
> > arch/ppc/boot/common/ns16550.c, In that file, changing SERIAL_BAUD
> > from 9600 to 38400  makes the bootloader talk at 38400. I always do it
> > when I use standard kernel source. But this time I found out that the
> > ns16550.c's path is arch/ppc/boot/, not arch/ppc/boot/common/, and
> > there was no "SERIAL_BAUD" string any more. So, how can I control the
> > baud rate?
>
> ???
>
> I don't know what you're referring to.  ns16550.c is in
> arch/ppc/boot/common on my tree, and it still contains the SERIAL_BAUD
> #define.
>
> >
> > 2 I have compiled the kernel and obtained a Image.elf whose size is
> > 1.7MB, and then it's time for creating a system.ace file which I can
> > put onto a compact flash card.  I Changed the file xupGenace.opt like
> > the follow:
>
> You can do this, but I don't think it's a particularly good idea.
> Binding the kernel image into an ACE file results in a *very* large
> ace file, and it's slow.  Consider the fact that doing it that way
> means that you're transferring the entire 1.7MB image via JTAG.
>
> I've also seem problems where DDR isn't fully initialized before the
> SystemACE tries to transfer the image which results in a non-booting
> system.
>
> What I do, is I've got a small bootloader that lives in BRAM and is
> just smart enough to load an elf file off of the CF card.  If I can
> get permission to do so, I'll publish the source for my utility.
>
> >
> > -jprog
> > -board user
> > -target ppc_hw
> > -hw ./implementation/download.bit
> > -elf zImage.elf
> > -configdevice devicenr 1 idcode 0x1127e093 irlength 14 partname xc2vp30
> > -debugdevice devicenr 1 cpunr 1
> > -ace system.ace
> >
> > and executed the command: xmd -tcl genace.tcl -opt genace.opt
> >
> > and the result I saw is:  Using GenACE option file : genace.opt
> > Error: Executable zImage.elf does not contain start address..
> >
> > I examine the information by google, some said that it's the bug of
> > development tools without sp, but both my ISE and EDK was patched.
> > Others said that it's because of the "mb-objdump", we should chage it
> > into "powerpc-eaci-objdump", and actually I had done this correction.
> > BUT, the error still here.
> >
> > Could anyone please to help me about this question? I will be very grateful.
>
> Very odd.  I haven't seen this problem myself so I can't be much help. sorry.
>
> >
> > BTW, could any kind person send any basic config file, which had been
> > proved that the kernel compiled based on it could work good on XUP
> > board?
>
> I don't have that board, sorry
>
> g.
>
> --
> Grant Likely, B.Sc., P.Eng.
> Secret Lab Technologies Ltd.
> grant.likely at secretlab.ca
> (403) 399-0195
>


-- 
web: http://www.forwind.cn
msn: likunarmstrong at hotmail.com


More information about the Linuxppc-embedded mailing list