[okl4-developer] Error on Building Image

[Geoffrey Lee]

On Mon, Aug 25, 2008 at 03:55:51PM +0800, Tsai, Tung-Chieh wrote:
> Dear Geoffrey,
> 
>   I attached a patch which is compare to okl4_2.1.1-patch.9.
> My build command is :
> 
>   ./tools/build.py machine=android project=l4test
> 
> I've tried to use VERBOSE_STR=True, but got the same error
>  message.

Hi Tung-Chieh

There are a couple of issues regarding the arm926ejs cpu code
in 2.1, for this specific case Frank posted this fix for it:

http://www.nabble.com/Re:-OKL4-2.1:-Original-version-of'arch-arm-pistachio-cpu-arm926ejs-include-cache.h'-is-not-compilable-td18367028.html

With this patch the error is gone.

> 
> Best Regards,
> Tsai, Tung-Chieh

	-gl

> 
> On Mon, Aug 25, 2008 at 2:03 PM, Geoffrey Lee <glee at ok-labs.com> wrote:
> > On Sat, Aug 23, 2008 at 10:49:09PM +0800, Tsai, Tung-Chieh wrote:
> >> Dear Nelson,
> >
> >
> > Hi Tung-Chieh
> >
> >>
> >> Ya, I'm trying to port to the Android platform , and modified from the
> >>  configuration for s3c2410. But, unless the 'head.spp,' I have changed
> >> everything. I make every functions which need to link for kernel, to be
> >>  empty, return 0, or just an infinite loop(Platform::reboot).
> >
> > Can you let us know what exactly you have changed in the source
> > tree (build system changes, source / header file changes, etc), as
> > well as the build command line you used to build your project?  You
> > can also get the build system to be more verbose with error
> > messages by passing VERBOSE_STR=True to the command line which
> > may give clues to why it is not working.
> >
> >
> >        -gl
> >
> >>
> >> And my machines.py is below:
> >>
> >>   class goldfish(arm926ejs):
> >>       platform = "goldfish"
> >>       memory = arm926ejs.memory.copy()
> >>
> >>   class android(goldfish):
> >>       virtual = False
> >>       memory = goldfish.memory.copy()
> >>       memory['physical'] = [Region(0x0, 0x6000000)]
> >>
> >> So, as you say, I think I miss some  configuration variable.
> >>
> >> Beside, I see my weaver.xml and s3c2410's, they're  different
> >> in base and virtual memory address, and my version lack of some
> >> device's address.But I can't see the what it relative to the error
> >> message.
> >>
> >> Thanks.
> >>
> >> Best Regards,
> >> Tsai, Tung-Chieh
> >>
> >> On Fri, Aug 22, 2008 at 2:51 PM, Nelson Tam <nelson at ok-labs.com> wrote:
> >> > Hi,
> >> >
> >> > It sounds like you are trying to port to the Android platform but using a
> >> > configuration for s3c2410.  That explains why elfweaver works when you
> >> > replaces certain variables with arm920t.
> >> >
> >> > It is likely that there is some configuration variable that needs to be
> >> > changed for goldfish.  I would starting looking at
> >> > platform/x/tools/machines.py and build/images/weaver.xml and see if anything
> >> > looks suspicious.
> >> >
> >> > Nelson.
> >> >
> >> > On 22/08/2008, at 4:26 PM, Tsai, Tung-Chieh wrote:
> >> >
> >> >>
> >> >> I've watched the video of porting okl4 to Android QEMU from :
> >> >>  http://benno.id.au/blog/2008/05/21/porting-okl4-to-a-new-soc
> >> >> and then I'm trying to do it. I use okl4_2.1.1-patch.9( from okl4 offical
> >> >> website), it seen quite different to the okl4 version used in the video.
> >> >> I viewed the source of s3c2410, and put some 'empty' function to my
> >> >> implementation. It seems all right in compiling and linking, but
> >> >> finally I got problem in building ELF image. The output message is below :
> >> >>
> >> >> [XML ] build/images/weaver.xml
> >> >> scons: building `build/images/image.elf' because it doesn't exist
> >> >> [ELF ] build/images/image.elf
> >> >> Error: Could not find address f0021044 in Image.
> >> >> scons: *** [build/images/image.elf] Error 1
> >> >> scons: building terminated because of errors.
> >> >>
> >> >> I think there is something I miss to this version of okl4,but I'm not sure
> >> >> what
> >> >> it is. Initially, I think I miss something in 'machine.py', so I change my
> >> >> machine.py to s3c2410's version, and change the name and value in it, but
> >> >> I
> >> >> got the same result. Accidentally, I found that if I change the cpu
> >> >> of goldfish
> >> >> to arm920t, this problem will disappear.
> >> >>
> >> >> I found the scons setting of arm926ejs in
> >> >> `.../arch/arm/tools/machines.py`,
> >> >> try to see if any thing I can do for it. Finally I think the problem
> >> >> is not here,
> >> >> because I found that if I only change the value of variable cpu in
> >> >> 'class arm926ejs(armv5tej):' to 'arm920t'( without modify any other
> >> >> thing),
> >> >> this problem disapear again.
> >> >>
> >> >> So, I have no idea to deal with this problem. Could anyone give me some
> >> >> advice about what may cause this problem ?
> >> >
> >> > --
> >> > (nt)
> >> >
> >> > Nelson Tam
> >> >
> >> >
> >>
> >> _______________________________________________
> >> Developer mailing list
> >> Developer at okl4.org
> >> https://lists.okl4.org/mailman/listinfo/developer
> >>
> >
> > --
> >
> >



--