This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: [Fwd: Re: [PATCH] MIPS32 DSP instructions again]
- From: Dominic Sweetman <dom at mips dot com>
- To: ian at airs dot com, Paul Koning <pkoning at equallogic dot com>
- Cc: nigel at mips dot com, echristo at redhat dot com, fu at mips dot com, rsandifo at nildram dot co dot uk, radhika at mips dot com, binutils at sourceware dot org
- Cc: Dominic Sweetman <dom at mips dot com>
- Date: Thu, 9 Jun 2005 18:54:44 +0100
- Subject: Re: [Fwd: Re: [PATCH] MIPS32 DSP instructions again]
- References: <42A86D4B.9010304@mips.com>
Nigel forwarded me your comments on why an assembler should be passed
a flag and on the basis of that flag reject mnemonics from the DSP ASE:
Ian> I disagree. I think it is appropriate and useful for the
Ian> assembler to be able to correctly accept or reject instructions
Ian> based on the command line options and .set options.
And Paul...
> Agreed.
It's clear to me, and I'm sure to all of you, that the assembler must
be handed a flag or a .set when the input should be assembled
differently according to instruction set or CPU variant. So the "l.d"
instruction on MIPS I CPUs is a two-instruction internal macro,
whereas on MIPS III CPUs it's a real machine instruction. Without the
flag, the code is ambiguous: a flag is necessary.
But that's not the case here. The DSP ASE instructions are new names,
never before seen in MIPS code. There's only one way to assemble them.
Which comes down to this: under what circumstances might it be helpful
for the assembler to reject a piece of code which it could have
assembled?
--
Dominic Sweetman
MIPS Technologies