setjmp/longjmp

I've implemented setjmp and longjmp for PowerPC. Annoyingly, because PPC still has no module support, files are compiled with -DGRUB_UTIL=1, and that causes include/grub/setjmp.h to redefine both functions. I guess that means we used to link in the libc setjmp/longjmp (which then failed miserably when one tried to run them).

Anyways, I have done some standalone tests and it seems to work fine. So pretending the GRUB_UTIL stuff is fixed, this wouldn't break the build...

-Hollis

2004-11-15 Hollis Blanchard <[email protected]>

* kern/powerpc/ieee1275/init.c (grub_setjmp): Remove function. (grub_longjmp): Likewise.
* include/grub/powerpc/setjmp.h (grub_jmp_buf): Set array size to 20.
* normal/powerpc/setjmp.S (grub_setjmp): New function. (grub_longjmp): Likewise.

Index: kern/powerpc/ieee1275/init.c
===================================================================
RCS file: /cvsroot/grub/grub2/kern/powerpc/ieee1275/init.c,v retrieving revision 1.7

diff -u -p -r1.7 init.c
@@ -111,15 +111,3 @@ grub_get_rtc (void)
 {
   return 0;
 }
-
-int
-grub_setjmp (grub_jmp_buf env __attribute ((unused)))
-{
-  return 0;
-}
-
-void
-grub_longjmp (grub_jmp_buf env __attribute ((unused)),
-	      int val __attribute ((unused)))
-{
-}

Index: include/grub/powerpc/setjmp.h
===================================================================
RCS file: /cvsroot/grub/grub2/include/grub/powerpc/setjmp.h,v retrieving revision 1.2

diff -u -p -r1.2 setjmp.h
@@ -20,6 +20,6 @@
 #ifndef GRUB_SETJMP_CPU_HEADER
 #define GRUB_SETJMP_CPU_HEADER	1
 
-typedef unsigned long grub_jmp_buf[6];
+typedef unsigned long grub_jmp_buf[20];
 
 #endif /* ! GRUB_SETJMP_CPU_HEADER */

Index: normal/powerpc/setjmp.S
===================================================================
RCS file: normal/powerpc/setjmp.S

Index: conf/powerpc-ieee1275.rmk
===================================================================
RCS file: /cvsroot/grub/grub2/conf/powerpc-ieee1275.rmk,v retrieving revision 1.13

diff -u -p -r1.13 powerpc-ieee1275.rmk
@@ -41,6 +41,7 @@ grubof_SOURCES = boot/powerpc/ieee1275/c
 	kern/powerpc/ieee1275/init.c term/powerpc/ieee1275/ofconsole.c \
 	kern/powerpc/ieee1275/openfw.c fs/ext2.c fs/ufs.c fs/minix.c fs/hfs.c \
 	fs/jfs.c normal/cmdline.c normal/command.c normal/main.c normal/menu.c \
+	normal/powerpc/setjmp.S \
 	disk/powerpc/ieee1275/ofdisk.c disk/powerpc/ieee1275/partition.c \
 	kern/env.c normal/arg.c loader/powerpc/ieee1275/linux.c \
 	loader/powerpc/ieee1275/linux_normal.c commands/boot.c

Nice. This evening I will try to fix that -DGRUB_UTIL thing and test this code of yours.

You can better say:

* normal/powerpc/setjmp.S: New file.

When you make a new file and you did not move old code there saying it is a new file is sufficient.

...

Ok.

Yeah, I wasn't even going to include that part of the patch originally since it breaks the build. Then I did, and forgot the changelog entry. Then I didn't bother to reply to myself, since I knew you'd do it for me. ;)

-Hollis

LOL!

Don't worry. I will have a look at it and check in your patch myself including the required changes.

Feel free to bother me about this when I forget about it.

Thanks,
Marco

Thanks for your patch.

For now I have added a macro called "GRUBOF" which is defined for grubof. In include/grub/setjmp.h I made a change to make an exception for grubof. I did this because a lot of GRUB_UTILS things are still required for grubof.

This can all change when module loading is supported.

I just committed the patch and it works great. I've also committed another patch which fixes some problem I had on the pegasos.

Thanks,
Marco

Ugghaa, very dirty. Please remove such a macro as soon as possible.

Okuji

Yes, I agree. I will remove it as soon as module loading is possible on the PPC. It is one of my highest priorities.

Thanks,
Marco