RE: [Bochs-developers] Disasm supports x86-64

[Stanislav S. <st...@fi...>]

Thanks for this first review, I already found the problem in RX/ERX/RRX =
methods. Instead of AND with 0xF it should AND with 0x7:

// 16-bit general purpose register
void disassembler::RX(const x86_insn *insn)
{=20
  dis_sprintf("%s", general_16bit_regname[(insn->b1 & 0x7) | =
insn->rex_b]);
}

// 32-bit general purpose register
void disassembler::ERX(const x86_insn *insn)
{=20
  dis_sprintf("%s", general_32bit_regname[(insn->b1 & 0x7) | =
insn->rex_b]);
}

// 64-bit general purpose register
void disassembler::RRX(const x86_insn *insn)
{=20
  dis_sprintf("%s", general_64bit_regname[(insn->b1 & 0x7) | =
insn->rex_b]);
}

I upload the fixed patch to #SF now.

For 64-bit code you should not see it correctly for debugger because of =
several reasons. The debugger totally doesn't supports x86-64, it still =
works with 32-bit EIP address, have no information about long mode =
executing (this is I actually could fix now), it has no access to 64-bit =
registers and etc. As I said you could use instrumentation instead of =
debugger.

Also I would like to ask to look on the opcodes.inc file, may be you'll =
see some inconsistent opcode name there ...

About stand alone debugger, I think the debugger definitely should be =
state less and work with Bochs through specific predefined interface. =
This is not the picture now and current debugger is out-of-date. =
Somebody should start to fix it, but I am not sure I have time to do it =
;(

Stanislav

-----Original Message-----
From: boc...@li... =
[mailto:boc...@li...] On Behalf Of =
Brendan Trotter
Sent: Monday, November 21, 2005 11:35 PM
To: Stanislav Shwartsman
Cc: boc...@li...
Subject: Re: [Bochs-developers] Disasm supports x86-64

Hi,

On 11/20/05, Stanislav Shwartsman <st...@fi...> wrote:
> It is =E2=80=A6 say 90-95% clean. But I theoretically can't validate =
everything so I
> need you help !

I did some testing, and something's not quite right...

I started by single-stepping through the Bochs BIOS's code (which is
real mode code) and found that the disassembly decodes some
instructions with the wrong registers. Examples:

(0) [0x000fe093] f000:e093 (unk. ctxt): mov r8w, 0xfffe           ; =
b8feff
(0) [0x000fe098] f000:e098 (unk. ctxt): mov r8w, 0x0000           ; =
b80000
(0) [0x000f05e6] f000:05e6 (unk. ctxt): pop r11w                  ; 5b
(0) [0x000f05e7] f000:05e7 (unk. ctxt): pop r13w                  ; 5d

Then I single stepped through some 64 bit code - it seems to get the
instruction sizes or prefixes wrong. Examples (lines of original
source code followed by disassembly):

	mov esi,[bootVideoInformation]
	mov eax,[esi+BOOT_VIDEO.videoDepth]
(0) [0x00000354] 0008:0354 (unk. ctxt): mov si, word ptr ds:[si]  ; 8b34
(0) [0x0000035b] 0008:035b (unk. ctxt): mov ax, word ptr ds:[esi] ;

	mov r9d,[esi+BOOT_VIDEO.fontAddress8x16]
	mov [videoDepth],eax
(0) [0x0000036e] 0008:036e (unk. ctxt): mov bp, word ptr ds:[esi+0x18]
; 678b6e18
(0) [0x00000372] 0008:0372 (unk. ctxt): inc sp                    ; 6744
(0) [0x00000377] 0008:0377 (unk. ctxt): inc sp                    ; 6744
(0) [0x0000037c] 0008:037c (unk. ctxt): mov word ptr ds:[si], ax  ; 8904

	shr ebx,3
	cmp ecx,800
(0) [0x000003a6] 0008:03a6 (unk. ctxt): inc sp                    ; 44
(0) [0x000003ae] 0008:03ae (unk. ctxt): inc sp                    ; 44
(0) [0x000003b6] 0008:03b6 (unk. ctxt): shr bx, 0x03              ; =
c1eb03
(0) [0x000003b9] 0008:03b9 (unk. ctxt): cmp cx, 0x0320            ;

	mov edx,esi
	mov edi,KSDMEMvideoMemory
	dec rax
(0) [0x00000466] 0008:0466 (unk. ctxt): mov dx, si                ; 89f2
(0) [0x00000468] 0008:0468 (unk. ctxt): mov r15w, 0x0000          ; =
bf0000
(0) [0x0000046d] 0008:046d (unk. ctxt): dec r8w                   ; 48


> Also collecting feature requests for the disassembler,
> both inside Bochs and stand-alone.

If you're doing a stand-alone debugger too, let me know where I can get =
it! :-)


Cheers,

Brendan

--
BCOS Project - http://bcos.hopto.org
N=18HS=C2=B5=C3=A9=C5=A0X=C2=B2=C5=A1=C2=B2=C5=A0u=C2=BC=E2=80=A6=C3=A2=C2=
=A2=C3=8B=C2=9D=C3=81=C2=B4'=C2=B6'y=C3=94u=C2=ACz=08=C2=B5=C3=AA=C2=A2=C2=
=B6=06=E2=80=B9N=C2=B6=C5=BE)
=E2=80=B9=C2=B1=C3=A1y=C3=A0=C2=AE=C3=98=E2=80=B0=C3=86=C5=A0=E2=80=B0=C3=
=85=C2=A9=C2=A2=C2=B0=E2=80=A2:=C5=A0x=E2=82=AC
zwz=C3=84=C2=AE=E2=80=B9=E2=80=9EI=C2=A1=C3=BD=C3=93=E2=80=98=C2=AEjz)=C2=
=A2=C3=B8=C5=A0=C3=98=C2=B6=C5=B8i=C3=9B=C2=B1=C3=99=C2=A2ov=C3=AFj