Re: GNU linker and Function stubs on mips.

[Koundinya K <knk at dde dot dk>]

Hi Ian & Mark,

Ian Lance Taylor wrote:

>    Date: Thu, 02 Mar 2000 12:29:20 -0700
>    From: Koundinya K <knk@dde.dk>
>
>    *************** _bfd_mips_elf_adjust_dynamic_symbol (inf
>    *** 7612,7618 ****
>
>        /* For a function, create a stub, if needed. */
>        if (h->type == STT_FUNC
>    !       || (h->elf_link_hash_flags & ELF_LINK_HASH_NEEDS_PLT) != 0)
>          {
>            if (! elf_hash_table (info)->dynamic_sections_created)
>            return true;
>    --- 7708,7714 ----
>
>        /* For a function, create a stub, if needed. */
>        if (h->type == STT_FUNC
>    !       && (h->elf_link_hash_flags & ELF_LINK_HASH_NEEDS_PLT) != 0)
>          {
>            if (! elf_hash_table (info)->dynamic_sections_created)
>            return true;
>    *************** _bfd_mips_elf_adjust_dynamic_symbol (inf
>    *** 7642,7647 ****
>    --- 7738,7751 ----
>              return true;
>            }
>          }
>    +   else if (h->type == STT_FUNC
>    +            && (h->elf_link_hash_flags & ELF_LINK_HASH_NEEDS_PLT) == 0)
>    +     {
>    +       /* This will set the entry for this symbol in the GOT to 0, and
>    +          the dynamic linker will take care of this */
>    +       h->root.u.def.value = 0;
>    +       return true;
>    +     }
>
>    Please provide your views on this change.
>
> I think the first test can be for ELF_LINK_HASH_NEEDS_PLT only.  I
> don't think we need to test for STT_FUNC at all.
>

Yes you are right. In fact even in the second test we don't need to test for
STT_FUNC at all.

*************** _bfd_mips_elf_adjust_dynamic_symbol (inf
*** 7611,7618 ****
                                           hmips->possibly_dynamic_relocs);

    /* For a function, create a stub, if needed. */
!   if (h->type == STT_FUNC
!       || (h->elf_link_hash_flags & ELF_LINK_HASH_NEEDS_PLT) != 0)
      {
        if (! elf_hash_table (info)->dynamic_sections_created)
        return true;
--- 7707,7713 ----
                                           hmips->possibly_dynamic_relocs);

    /* For a function, create a stub, if needed. */
!   if ( (h->elf_link_hash_flags & ELF_LINK_HASH_NEEDS_PLT) != 0)
      {
        if (! elf_hash_table (info)->dynamic_sections_created)
        return true;
*************** _bfd_mips_elf_adjust_dynamic_symbol (inf
*** 7642,7647 ****
--- 7737,7749 ----
          return true;
        }
      }
+   else
+     {
+       /* This will set the entry for this symbol in the GOT to 0, and
+          the dynamic linker will take care of this */
+       h->root.u.def.value = 0;
+       return true;
+     }

because in libstdc++.so generated bt GNU linker, cout is defined to be a
object. I don't understand why it has to put cout as an object type.

[187324]        |0x60070e54|0x00000028|OBJT |GLOB |0    |65282  |cout

So the second change was failing. cout happens to be GOT16 reloc in hello.o
and again it does not need a stub and also the fact that it is defined in the
shared library.

So I feel it is better to check against whether a symbol requires a PLT or
not.


>
> However, I'm not convinced that it will always work.  If a symbol has
> both call relocations and ordinary relocations, then I think that
> NEEDS_PLT will be set, but we want to set the dynamic symbol table
> entry to 0.  Try something like this with the native tools and with
> your linker, to see what happens:
>     extern int foo ();
>     int (*pfn) = foo;
>     int bar () { return foo (); }
>

I  did not quite understand this. Here is the objdump -r of the above you
asked me to check. I just wrote a main and tried to link main.o and ruf.o. I
got undefined references to foo. I am not sure what I am supposed to do.
Please clarify ? What results from the Linker am I supposed to expect ??

halfway:2708 [knk/work/hello] objdump -r ruf.o

ruf.o:     file format elf32-tradbigmips

RELOCATION RECORDS FOR [.text]:
OFFSET   TYPE              VALUE
00000000 R_MIPS_HI16       _gp_disp
00000004 R_MIPS_LO16       _gp_disp
00000024 R_MIPS_GOT16      foo


RELOCATION RECORDS FOR [.data]:
OFFSET   TYPE              VALUE
00000000 R_MIPS_32         foo

Thanks a lot.

koundinya