Re: porting to hppa-ipl

[Jeff Bailey]

On Sun, Jul 30, 2006 at 04:11:26PM +0200, Yoshinori K. Okuji wrote:
> On Saturday 29 July 2006 15:35, Jeff Bailey wrote:
> > Hi!  I started working on the port to hppa-ipl and have run into three
> > things:
> 
> Amazing. ;)

*lol* It's amazing when I'm succesful, in the meantime, it's just
insane. =)

One of the nice things, though, is that I have the support of the folks
who wrote palo.

> > 1) Dependancy generation seems to skip out on files that don't exist.
> > This means that in the Makefile, hello.c doesn't wind up depending on
> > grub_script.tab.h.  "make grub_script.tab.h" works fine, and then the
> > build can proceed.  Current ports work around this by defining _HEADERS
> > lines, apparently, but this dependancy should probalby be explicitely
> > declared somewhere, or this file should be forcibly generated.
> 
> Why does hello.c depend on grub_script.tab.h? AFAIK, only normal/lexer.c 
> depends on grub_script.tab.h, and the dependency is automatically generated 
> in my environment.

According to GCC:

gcc -Ihello -I./hello -I. -Iinclude -I./include -Wall -W -Wall -W -Wshadow 
-Wpointer-arith -Wmissing-prototypes                  -Wundef 
-Wstrict-prototypes -g -Os -fno-builtin  -c -o hello_mod-hello_hello.o 
hello/hello.c
In file included from include/grub/normal.h:28,
                 from hello/hello.c:27:
include/grub/script.h:27:29: error: grub_script.tab.h: No such file or directory

> > 2) When hacking on the build env, how do I force things to be
> > regenerated?  When I do "make conf/hppa-ipl.mk", it doesn't generate the
> > file for me.
> 
> Isn't it normal in make??? When I play with make, I do "rm -f FILE; make 
> FILE". I don't know anything else more elegant.

Right it is normal in make, which is why I was surprised it didn't work:

address@hidden:~/Programming/cvstree/grub2/conf$ rm hppa-ipl.mk
address@hidden:~/Programming/cvstree/grub2/conf$ cd ..
address@hidden:~/Programming/cvstree/grub2$ make conf/hppa-ipl.mk
Makefile:105: conf/hppa-ipl.mk: No such file or directory
make: *** No rule to make target `conf/hppa-ipl.mk'.  Stop.
address@hidden:~/Programming/cvstree/grub2$ cd conf
address@hidden:~/Programming/cvstree/grub2/conf$ make hppa-ipl.mk
make: *** No rule to make target `hppa-ipl.mk'.  Stop.
address@hidden:~/Programming/cvstree/grub2/conf$ ls hppa-ipl.rmk
hppa-ipl.rmk
address@hidden:~/Programming/cvstree/grub2/conf$ 

> > 3) The files for grub-emu and others seems to be copied into each file
> > in conf.  Should these be refactored out into common.mk?  It seems a
> > waste for me to just copy them into hppa-ipl for yet another time.
> 
> Well, the original idea I had was that it could be possible to implement 
> grub-emu differently among architectures. Currently we only emulate PC. I 
> cannot tell you if this is a good thing. As grub-emu does not use much in 
> i386- or PC-specific code, this might be sufficient. If so, it is a good idea 
> to share the same config in common.rmk.

Okay.  As I'm hacking along, I'll see what I do that looks identical and
supply patches to refactor.

Tks,
Jeff Bailey

-- 
I do not agree with a word you say, but I will defend to the death your
right to say it. 
 - Voltaire