[K42-discussion] Complete build/mambo regress with GIT Makefile.new

[Patrick G. Bridges]

On Apr 11, 2007, at 11:48 AM, Livio Soares wrote:

>> # Build linux object files that K42 needs
>>   cd .. &&
>>   git-clone http://www.cs.unm.edu/~k42/git/linux-030.git &&
>>   cd linux-030 &&
>>   git-checkout k42-port &&
>
>   I tried the checkout and got the following:
>
> error: pathspec 'k42-port' did not match any file(s) known to git.
> Did you forget to 'git add'?
>
>   Now, I admit to knowing anything of how GIT works. Since I didn't  
> want to
> spend the time  learning it, or understading what could  be wrong,  
> I simply
> hacked the git meta-data, in the following way, and things _seemed_  
> to work
> fine. Please yell if I have committed some atrocity with the  
> following:
>
>     cd linux-030
>     cp .git/refs/remotes/origin/k42-port .git/refs/heads/
>     git-checkout k42-port

I really have no idea. The git checkout worked for me, so that's odd.  
Amos knows more about this sort of thing with GIT, so maybe he can  
chime in here.

>>   make ARCH=ppc64 CROSS_COMPILE=powerpc64-linux- \
>>    CROSS32_COMPILE=powerpc-linux- k42_defconfig &&
>>   make -f Makefile.k42 MKANCHOR=~/k42 \
>>    CROSS_COMPILE=powerpc64-linux- CROSS32_COMPILE=powerpc-linux-  
>> k42libs
>
>   This is probably obvious, but I had to do:
>
> ln -s $MY_K42_DEVEL_DIRECTORY ~/k42

I'll look into this so that it works better. Probably an issue in the  
location spec file or something.

>> # Run regression test
>> k42console -m mambo -f ~/k42/powerpc/fullDeb/os/k42-vmlinux --
>> MAMBO_BOOT_VMLINUX=1 MAMBO_MAPLE=1 MAMBO_TCL_STAMP=-120305
>> MAMBO_TYPE=gpul K42_INITSCR=/home/regress/sysinit MAMBO_EXIT=yes
>
>   This is mostly  running. I forgot to setup some of  my NFS  
> variables, and
> therefore K42 crashes when trying to  access the file-system. But I  
> get all
> the  way to  the  beginning of  login.  I  bet  that after  fixing  
> the  NFS
> problem, regress here will run fine.

There may be a few problems with the permissions for some of the test  
scripts and such, too. I'm sorting out these details and I'll check  
in more fixes as I come across them. Right now when regress finishes,  
I get a message that says "/home/regress/sysinit: line 88: /ksys/ 
console: No such file or directory" that I haven't finished out yet,  
for example

>   Thanks again for this,

No problem. I'm going to attack gcc 4.0 in a few weeks, but I have  
another research project I'd like to do some prototyping on first  
before I get to that.

-Patrick