This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] S/390: Disassembler - pick instruction with most specific mask
- From: Martin Schwidefsky <schwidefsky at de dot ibm dot com>
- To: "Andreas Krebbel" <krebbel at linux dot vnet dot ibm dot com>
- Cc: binutils at sourceware dot org, gingold at adacore dot com
- Date: Mon, 27 Sep 2010 13:46:06 +0200
- Subject: Re: [PATCH] S/390: Disassembler - pick instruction with most specific mask
- References: <20100924145217.GA4769@bart>
On Fri, 24 Sep 2010 16:52:17 +0200
"Andreas Krebbel" <krebbel@linux.vnet.ibm.com> wrote:
> Hi,
>
> currently the disassembler for S/390 always choses the most recent
> version of an instruction. That way when a later machine adds an
> instruction with a new operand while using the same opcode objdump
> would show that new instruction even for old code. Instead it should
> chose the instruction with the most specific mask for an opcode.
>
> Implemented with the attached patch.
>
> No testsuite regressions.
>
> Ok to apply? Would be cool if that patch could make it into 2.21.
Fine with me. In particular I like how the "bcr 0,%rx" instructions
now come out as "nopr %rx" in the disassembly:
> Index: src/gas/testsuite/gas/s390/esa-reloc.d
> ===================================================================
> --- src.orig/gas/testsuite/gas/s390/esa-reloc.d
> +++ src/gas/testsuite/gas/s390/esa-reloc.d
> @@ -48,4 +48,4 @@ Disassembly of section .text:
> [ ]*50: R_390_PLT16DBL foo\+0x2
> 52: 00 00 00 00 [ ]*.long 0x00000000
> [ ]*52: R_390_PLT32 foo\+0x4
> - 56: 07 07 [ ]*bcr 0,%r7
> + 56: 07 07 [ ]*nopr %r7
That makes it more readable.
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.