Home | History | Annotate | Line # | Download | only in ld.elf_so
TODO revision 1.4 
      1 rtld:
      2 * resolve MIPS binding lossage
      3 
      4 binutils/gcc:
      5 * alpha: why are there GLOB_DAT relocs in ld.elf_so?
      6 * alpha: bogus textrels in rtti info
      7 * mips: why are there global GOT relocs in ld.elf_so?
      8 * mips: bogus textrels in .rel.dyn
      9 * powerpc: make sure the .got section is always executable
     10 * powerpc: fix ld scripts so that .rodata is still used for PIC
     11 
     12 machine-specific:
     13 * #ifdefs:
     14 headers.c:#ifndef __mips__
     15 headers.c:#ifdef __mips__
     16 reloc.c:#if defined(__hppa__)
     17 rtld.c:#if !defined(__hppa__)
     18 rtld.c:#if !defined(__mips__) && !defined(__hppa__)
     19 rtld.c:#if !defined(__arm__) && !defined(__mips__) && !defined(__sh__)
     20 rtld.c:#ifdef __mips__
     21 rtld.h:#ifdef __mips__
     22 symbol.c:#ifndef __mips__
     23 
     24 issues to research:
     25 * sh, x86_64 have different absolute and GLOB_DAT relocs; nothing else does
     26 * sh, vax, x86_64 use addend on PLT; nothing else does
     27 
     28 
     29 rela: alpha hppa m68k powerpc sh sparc sparc64 vax x86_64
     30 rel: arm i386 mips
     31 
     32 - mycroft
     33   2003/09/24
     34