Re: Disassembly difference

[Andrew Cagney <ac131313 at cygnus dot com>]

Gerwyn Davies wrote:
> 
> Hello,
> 
> I'm experiencing a difference in opinion between the disassembly produced by
> objdump and that produced by GDB. As these two traces show. Can anyone offer
> an explanation.

GDB would be using a different dissassembler to objdump - most likely it
is selecting the default mips3 (?) dissassembler where as objdump will
have selected the tx39 variant.  Look for how tm_print_insn is being
(incorrectly?) set.

	Andrew


> 
> GDB TRACE:
> This GDB was configured as "--host=i586-pc-cygwin32 --target=mips-tx39-elf".
> (gdb) target remote com1
> Remote debugging using com1
> 0xbfc09fbc in ?? ()
> (gdb) load led.rom
> Loading section .text, size 0xe34 lma 0xa0008000
> Loading section .ctors, size 0x8 lma 0xa0020e34
> Loading section .dtors, size 0x8 lma 0xa0020e3c
> Loading section .data, size 0x720 lma 0xa0008e34
> Loading section .sdata, size 0x1c lma 0xa0009554
> Start address 0xa0020000 , load size 5504
> Transfer rate: 1693 bits/sec.
> (gdb) disas 0xa0020000 0xa0020030
> Dump of assembler code from 0xa0020000 to 0xa0020030:
> 0xa0020000:     nop
> 0xa0020004:     sllv    $zero,$zero,$zero
> 0xa0020008:     0x2010000
> 0xa002000c:     lb      $zero,512($a0)
> 0xa0020010:     nop
> 0xa0020014:     nop
> 0xa0020018:     sll     $zero,$s0,0x0
> 0xa002001c:     0x1
> 0xa0020020:     0x10a001
> 0xa0020024:     nop
> 0xa0020028:     sll     $zero,$zero,0x8
> 0xa002002c:     nop
> End of assembler dump.
> 
> OBJDUMP TRACE:
> 00000000a0020000 <_start>:
>     a0020000: 00000000  nop
> 
> 00000000a0020004 <zerobss>:
>     a0020004: 3c02a004  lui $v0,0xa004
>     a0020008: 2442073c  addiu $v0,$v0,1852
>     a002000c: 3c03a004  lui $v1,0xa004
>     a0020010: 24630850  addiu $v1,$v1,2128
>     a0020014: ac400000  sw $zero,0($v0)
>     a0020018: 24420004  addiu $v0,$v0,4
>     a002001c: 0062082b  sltu $at,$v1,$v0
>     a0020020: 1020fffc  beqz $at,a0020014 <zerobss+0x10>
>  ...