lightning
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Lightning] About _ASM_SAFETY


From: Paolo Bonzini
Subject: Re: [Lightning] About _ASM_SAFETY
Date: Sun, 29 Aug 2010 12:06:07 +0200

2010/8/28 Paulo César Pereira de Andrade
<address@hidden>:
>  Hi,
>
>  I made a somewhat experimental commit yesterday,
> what should be ok for a "playground" repository :-), but
> I would like to comment about it.

I liked it and pulled it already.

As long as there's no confusion between MMX/XMM vs. general purpose
registers, I think the improvement is good enough.

> Possible way to simplify:
> 1. Define registers with only a class modifier, example:
>    #define _RA    0x10
>    ...
>    #define _R15   0x1f
>    ...
>    #define _RAH  0x20  /* %ah and other top 8 bits of a 16 bit register */
>    ...
>    #define _RX0   0x40  /* mmx, sse, sse2 first register */
>
>    after, that, do not use names like _AL, _AH, _AX, _EAX, and _RAX,
> but just the _RA, and let the _ASM_SAFETY check ensure it is a valid
> argument.

I think this is too complicated.

> 2. Add support for inline functions, to have compiler support
>    for strict type checking, and use enums. This is tricky as
>    it requires maintaining two implementations. But should
>    allow the compiler to generate significantly better code for
>    macro/inline function invocations with values not known at
>   compile time.

Changing all of lightning to inline functions may make sense, but it's
a big chunk of work.

> /* more strict patch, possibly only accepting 64 bit registers where
> 32 bit ones are required
> * (explaining better :-) 64 bit registers actually mean an abstract
> type/value check, and
> * is a bitmask removed before generating code, so, effectively a 32
> bit and a 64 bit register
> * specification in lightning macros generate the same jit code) */

_ASM_STRONG_SAFETY or something like that.  I thought about it too
while reviewing your patch, but I think _ASM_SAFETY as you defined it
is good enough for lightning. It was not "safe enough" for the ccg
project on which asm*.h was based, but for lightning the assemblers
are an implementation detail and the register sizes are not visible to
the user.

Paolo



reply via email to

[Prev in Thread] Current Thread [Next in Thread]