[Bug binutils/13964] New: ARM libopcode force-thumb sticks on

[dave at treblig dot org]

http://sourceware.org/bugzilla/show_bug.cgi?id=13964

             Bug #: 13964
           Summary: ARM libopcode force-thumb sticks on
           Product: binutils
           Version: 2.22
            Status: NEW
          Severity: normal
          Priority: P2
         Component: binutils
        AssignedTo: address@hidden
        ReportedBy: address@hidden
    Classification: Unclassified


Created attachment 6332
  --> http://sourceware.org/bugzilla/attachment.cgi?id=6332
Test wrapper for ARM disassembly showing sticky force-thumb

If I call the libopcodes ARM disassembler for arch/subarch arm/arm and don't
pass any options it defaults to disassembling in ARM's normal 4byte opcode
format, if I pass the force-thumb option it disassembles in the more compact
2/4 byte thumb format - so far so good.

However, if I call the disassembler with force-thumb as an option, and then
call it again without the option it'll still use thumb mode - I don't think
this is intentional.

In src/opcodes/arm-dis.c:1650 we have:


static bfd_boolean force_thumb = FALSE;

Then in parse_arm_disassembler_option somewhere around line 4398 we have:

  else if (CONST_STRNEQ (option, "force-thumb"))
    force_thumb = 1;
  else if (CONST_STRNEQ (option, "no-force-thumb"))
    force_thumb = 0;

and that's the only places force_thumb are set; so once it's set by a
force-thumb, if you miss the option off it defaults differently - I believe
that if you call the disassembler twice with the same options it should do the
same thing, hence I say that force_thumb needs initialising in
parse_arm_disassembler_option or similar.

The following test is a demo of the bug; it currently outputs:


address@hidden:~$ gcc disassbug.c -lbfd-2.22-multiarch -lopcodes-2.22-multiarch
address@hidden:~$ ./a.out 
Result is : svc 0x00001234 nprocessed=4
Result is : asrs        r4, r6, #8 nprocessed=2
Result is : asrs        r4, r6, #8 nprocessed=2

but I believe it should output:

Result is : svc 0x00001234 nprocessed=4
Result is : asrs        r4, r6, #8 nprocessed=2
Result is : svc 0x00001234 nprocessed=4

given that the 1st and 3rd calls are done without force-thumb.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.