Re: [PATCH] libffi on XScale (BE)

[Quaritsch Markus <quam at qwws dot net>]

On Tue, Oct 26, 2004 at 05:04:16PM +0100, Richard Earnshaw wrote:

> I think I've figured out most of what you are trying to do now, and as a
> result I think your approach is wrong.

but it works ;-)

> --- sysv.S.orig 2004-10-21 09:35:22.000000000 +0200
> +++ sysv.S      2004-10-21 10:59:17.204423684 +0200
> @@ -88,10 +88,17 @@
>         mov   sp, fp
>  
>         # Load a3 with the pointer to storage for the return value
> -       ldr   a3, [sp, #24]
> +# @@@ quam
> +       ldr  a3, [sp, #36]
>  
> 
> This bit doesn't make sense at all.  There are only 32 bytes in the
> stack frame for this call, so you can't peek outside it.

But when you look at the code-snipped below, you can see, thet the
pointer for the ecif-struct is at sp + 32, and therefore the address of
the return-value must be at sp + 36 (its the parameter before the
ecif-struct). I have not reconstructed all modifications of the
stackpointer within this function - just guessing... 
I've also compared the address of the returnvalue in the preceding
c-function with the one in the assembler-routine and they are (now) the
same.

>         # Load a4 with the return type code 
> -       ldr   a4, [sp, #12]
> +# @@@ quam
> +       # load address of ecif.cif  
> +       ldr a4, [sp, #32] 
> +       # load address of ecif.cif->rtype 
> +       ldr a4, [a4, #12]
> +       # load value of ecif.cif->rtype->type 
> +       ldrh a4, [a4, #6]
>  
> No, you don't want to be looking up the fully-qualified type here, you
> want the simplified type, as calculated in arm/ffi.c
> (ffi_prep_cif_machdep).  That function needs extending to support a
> single 64-bit integer type (the low-level code doesn't care whether it's
> signed or not, just the size and the registers holding the value).

Well I just looked at the parameter passed to the assembler-routine and
tried to grab the type of the return-value. The only way I found was the
ecif.cif-struct...
The original code was (from my point of view) wrong, because a4 always held 
the value 1 (FFI_TYPE_INT) - even if the return-type was 64bit value. 

I'm sure, you have a better solution because this are my first
experiences with the XScale (especially asm on this platform) and
libffi (got to know it last week).

bye
quam
-- 
|  .''`. | Quaritsch Markus   quam@qwws.net   Student of Telematics |
| : :' : |                                         TU-Graz, Austria |
| `. `'  |                                                          |
|   `-   | Enter any 11-digit prime number to continue...           |

Attachment: signature.asc
Description: Digital signature