qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] Compile fixes for newer gcc


From: Paul Brook
Subject: Re: [Qemu-devel] Compile fixes for newer gcc
Date: Tue, 10 May 2005 22:10:59 +0100
User-agent: KMail/1.7.2

On Tuesday 10 May 2005 21:47, J. Mayer wrote:
> On Mon, 2005-05-09 at 02:49 +0100, Paul Brook wrote:
> > The attached patch fixes some trivial build problems with newer gcc on
> > amd64. It adds FORCE_RET on load ops,
>
> Index: target-ppc/exec.h
> ===================================================================
> RCS file: /cvsroot/qemu/qemu/target-ppc/exec.h,v
> retrieving revision 1.10
> diff -u -p -r1.10 exec.h
> --- target-ppc/exec.h   13 Mar 2005 17:01:22 -0000      1.10
> +++ target-ppc/exec.h   9 May 2005 01:33:04 -0000
> @@ -33,11 +33,7 @@ register uint32_t T2 asm(AREG3);
>  #define FT1 (env->ft1)
>  #define FT2 (env->ft2)
>
> -#if defined (DEBUG_OP)
> -#define RETURN() __asm__ __volatile__("nop");
> -#else
> -#define RETURN() __asm__ __volatile__("");
> -#endif
> +#define RETURN() FORCE_RET()
>
> Please don't change this. This is usefull and your so-called "fix" only
> makes debug harder.
> Same goes for all other patches: use RETURN macro everywhere in PowerPC
> emulation code.

Ok. This slipped in from a following patch, so we can revisit it when I submit 
those changes. A much better way to do this is to add the NOPs in dyngen. 
Then you get nops after all ops, not just the ones that use RETURN()

> >  and introduces helper functions for floating
> > point negation (these require a literal constant load).
>
> /* fneg */
> +void do_fneg (void);
>  PPC_OP(fneg)
>  {
> -    FT0 = -FT0;
> +    do_fneg();
>      RETURN();
>  }
>
> Where's the problem ? It compiles and run perfectly on my amd64. Calling
> a function for this is a nonsense. Don't apply this.

As I said it's required to avoid a pc-relative constant literal load.
gcc4.0 compiles fneg to:

movlpd 0x1288(%rbp),%xmm0
xorpd  0(%rip),%xmm0        # 3e50 <op_fneg+0x10>
     R_X86_64_PC32     .LC0+0xfffffffffffffffc
movsd  %xmm0,0x1288(%rbp)

Which dyngen can't grok.

> I will do an update to use softfloat functions everywhere in PowerPC
> emulation one of those days...

That would be great.
 
Paul




reply via email to

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