[Bug ld/15787] New: Better choice for B(S) value in ARM relocations

[address@hidden]

http://sourceware.org/bugzilla/show_bug.cgi?id=15787

            Bug ID: 15787
           Summary: Better choice for B(S) value in ARM relocations
           Product: binutils
           Version: 2.24 (HEAD)
            Status: NEW
          Severity: normal
          Priority: P2
         Component: ld
          Assignee: unassigned at sourceware dot org
          Reporter: address@hidden

Created attachment 7122
  --> http://sourceware.org/bugzilla/attachment.cgi?id=7122&action=edit
Proposed patch for a better B(S) choice for
R_ARM_{ALU|LDR\LDC|LDRS}_SB_G{0|1|2} relocations

ARM's IHI0044E document specifies the B(S) for the
R_ARM_{ALU|LDR|LDC|LDRS}_SB_G{0|1|2} relocations as:

B(S) is the addressing origin of the output segment defining the symbol S.
The origin is not required to be the base address of the segment. This
value must always be word-aligned.

However currently in bfd/elf32-arm.c the sb value gets calculated as:

    /* sb should be the origin of the *segment* containing the symbol.
       It is not clear how to obtain this OS-dependent value, so we
       make an arbitrary choice of zero.  */
    bfd_vma sb = 0;

IMHO this is wrong and this is not about the origin of the output segment
at run time but about the one which gets determined at linking time.  As
sb group relocations are static ones they are most probably the same so we
better use sym_sec->output_section->vma instead of a fixed value 0.

Please find patch attached.

Concerning tests : we originally had 4 linker groups tests :
group-relocs-alu-bad, group-relocs-ldc-bad, group-relocs-ldr-bad and
group-relocs-ldrs-bad.  The first two were pc-based, the last two were
sb-based.  I've changed the last two in pc-based as well and added 4 new
sb-based tests.

Tested binutils, ld and gas with target arm-unknown-eabi, no regressions.

I've a copyright assignment but no write acccess.

bfd

John Tytgat  <address@hidden>

    * elf32-arm.c (elf32_arm_final_link_relocate): Use origin of output
    segment containing the relocating symbol instead of assuming 0 for
    sb group relocations.

ld/testsuite

John Tytgat  <address@hidden>

    * ld-arm/group-relocs.s: Add comments.  Move symbols used for sb
    group relocations into .data section.  Drop section zero.  Use pc/r0
    as base register when pc/sb group relocations are used.
    * ld-arm/group-relocs.d: Adjust expected result.
    * ld-arm/group-relocs-ldr-bad.s: Use pc_g0/pc_g1 relocs instead of
    sb_g0/sb_g1.
    * ld-arm/group-relocs-ldrs-bad.s: Likewise.
    * ld-arm/group-relocs-ldr-bad.d: Adjust expected result.
    * ld-arm/group-relocs-ldrs-bad.d: Likewise.
    * ld-arm/group-relocs-alu-bad-2.d: New test for sb group relocation.
    * ld-arm/group-relocs-ldr-bad-2.d: Likewise.
    * ld-arm/group-relocs-ldrs-bad-2.d: Likewise.
    * ld-arm/group-relocs-ldc-bad-2.d: Likewise.
    * ld-arm/group-relocs-alu-bad-2.s: New test source.
    * ld-arm/group-relocs-ldr-bad-2.s: Likewise.
    * ld-arm/group-relocs-ldrs-bad-2.s: Likewise.
    * ld-arm/group-relocs-ldc-bad-2.s: Likewise.
    * ld-arm/arm-elf.exp: For group-relocs, drop section zero start
    definition.  Run the new tests.

-- 
You are receiving this mail because:
You are on the CC list for the bug.