Jivin John Williams lays it down ...
> Attached is a patch that updates elf2flt support for microblaze:
>
> * Handling for new reloc type R_MICROBLAZE_32_PCREL_LO (we can just ignore
> it)
> * remove workarounds for old microblaze linker bugs, fixed in 2.16
> * Tweak configure{.in} to indicate that MicroBlaze needs the
> --disable-emit-relocs switch
>
> signed-off-by: John Williams <john.willi...@petalogix.com>
Applied the elf2flt changes.
Still awaiting a conclusion to the emit-relocs bits discussion :-)
Seems like MB would be better served with a preset default like Mike
suggested ?
Thanks,
Davidm
> Index: configure
> ===================================================================
> RCS file: /var/cvs/elf2flt/configure,v
> retrieving revision 1.17
> diff -u -r1.17 configure
> --- configure 7 May 2009 23:09:35 -0000 1.17
> +++ configure 17 May 2009 03:10:00 -0000
> @@ -1355,7 +1355,7 @@
> --disable-FEATURE do not include FEATURE (same as
> --enable-FEATURE=no)
> --enable-FEATURE[=ARG] include FEATURE [ARG=yes]
> --disable-got-check - disable check for GOT (needed on H8)
> - --disable-emit-relocs - don't use the --emit-relocs (-q) linker option
> + --disable-emit-relocs - don't use the --emit-relocs (-q) linker option
> (needed on MicroBlaze)
> --enable-emit-ctor-dtor manually create ctor/dtor list
>
> Optional Packages:
> Index: configure.in
> ===================================================================
> RCS file: /var/cvs/elf2flt/configure.in,v
> retrieving revision 1.16
> diff -u -r1.16 configure.in
> --- configure.in 7 May 2009 06:14:27 -0000 1.16
> +++ configure.in 17 May 2009 03:10:00 -0000
> @@ -50,7 +50,7 @@
> )
>
> AC_ARG_ENABLE(emit_relocs,
> - [ --disable-emit-relocs - don't use the --emit-relocs (-q) linker
> option ],
> + [ --disable-emit-relocs - don't use the --emit-relocs (-q) linker
> option (needed on MicroBlaze)],
> [ emit_relocs=$enableval ],
> [ emit_relocs=yes ]
> )
> Index: elf2flt.c
> ===================================================================
> RCS file: /var/cvs/elf2flt/elf2flt.c,v
> retrieving revision 1.57
> diff -u -r1.57 elf2flt.c
> --- elf2flt.c 7 May 2009 06:23:02 -0000 1.57
> +++ elf2flt.c 17 May 2009 03:10:01 -0000
> @@ -522,6 +522,7 @@
> {
> case R_MICROBLAZE_NONE:
> case R_MICROBLAZE_64_NONE:
> + case R_MICROBLAZE_32_PCREL_LO:
> continue;
> }
> #endif /* TARGET_microblaze */
> @@ -979,15 +980,11 @@
> the relocation symbol. */
> {
> unsigned char *p = r_mem;
> - uint32_t offset;
> pflags=0x80000000;
>
> /* work out the relocation */
> sym_vma = bfd_section_vma(abs_bfd,
> sym_section);
> - /* grab any offset from the text */
> - offset = (p[2]<<24) + (p[3] << 16) +
> (p[6] << 8) + (p[7]);
> - /* Update the address */
> - sym_addr += offset + sym_vma +
> q->addend;
> + sym_addr += sym_vma + q->addend;
> /* Write relocated pointer back */
> p[2] = (sym_addr >> 24) & 0xff;
> p[3] = (sym_addr >> 16) & 0xff;
> @@ -1019,36 +1016,21 @@
> case R_MICROBLAZE_32:
> {
> unsigned char *p = r_mem;
> - unsigned long offset;
>
> - /* grab any offset from the text */
> - offset = (p[0]<<24) + (p[1] << 16) +
> (p[2] << 8) + (p[3]);
> sym_vma = bfd_section_vma(abs_bfd,
> sym_section);
> - /* This is a horrible kludge. For some
> - reason, *sometimes* the offset is in
> - both addend and the code. Detect
> - it, and cancel the effect. Otherwise
> - the offset gets added twice - ouch.
> - There should be a better test
> - for this condition, based on the
> - BFD data structures */
> - if(offset==q->addend)
> - offset=0;
> -
> - sym_addr += offset + sym_vma +
> q->addend;
> + sym_addr += sym_vma + q->addend;
> relocation_needed = 1;
> break;
> }
> case R_MICROBLAZE_64_PCREL:
> sym_vma = 0;
> - //sym_addr = (*(q->sym_ptr_ptr))->value;
> sym_addr += sym_vma + q->addend;
> sym_addr -= (q->address + 4);
> sym_addr = htonl(sym_addr);
> /* insert 16 MSB */
> - * ((unsigned short *) (r_mem+2)) |=
> (sym_addr) & 0xFFFF;
> + * ((unsigned short *) (r_mem+2)) =
> (sym_addr) & 0xFFFF;
> /* then 16 LSB */
> - * ((unsigned short *) (r_mem+6)) |=
> (sym_addr >> 16) & 0xFFFF;
> + * ((unsigned short *) (r_mem+6)) =
> (sym_addr >> 16) & 0xFFFF;
> /* We've done all the work, so continue
> to next reloc instead of break */
> continue;
> _______________________________________________
> uClinux-dev mailing list
> uClinux-dev@uclinux.org
> http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
> This message was resent by uclinux-dev@uclinux.org
> To unsubscribe see:
> http://mailman.uclinux.org/mailman/options/uclinux-dev
--
David McCullough, david_mccullo...@securecomputing.com, Ph:+61 734352815
McAfee - SnapGear http://www.snapgear.com http://www.uCdot.org
_______________________________________________
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
