[llvm-commits] [llvm-gcc-4.2] r108554 - /llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp

[Bob Wilson]

Dale, I don't think this is correct, at least for ARM.  The 'p' constraint is for a memory address, not the actual memory reference.  If I remember correctly, the use of this reported in pr4521 was kind of questionable, and I haven't yet looked in any detail at pr5314 and pr5533.  Maybe we can talk about this in person sometime today.

On Jul 16, 2010, at 2:00 PM, Dale Johannesen wrote:

> Author: johannes
> Date: Fri Jul 16 16:00:31 2010
> New Revision: 108554
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=108554&view=rev
> Log:
> "p" constraint is a form of "m", not "r".  PR 5314.
> 
> 
> Modified:
>    llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp
> 
> Modified: llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp?rev=108554&r1=108553&r2=108554&view=diff
> ==============================================================================
> --- llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp (original)
> +++ llvm-gcc-4.2/trunk/gcc/llvm-convert.cpp Fri Jul 16 16:00:31 2010
> @@ -4429,12 +4429,11 @@
>       continue;
>     }
> 
> -    // Translate 'p' to 'r'.  This is supposed to check for a valid memory
> +    // Translate 'p' to 'm'.  This is supposed to check for a valid memory
>     // address, but for inline assembly there is no way to know the mode of
> -    // the data being addressed.  Assume that a general register is always
> -    // a valid address.
> +    // the data being addressed.
>     if (ConstraintChar == 'p')
> -      ConstraintChar = 'r';
> +      ConstraintChar = 'm';
> 
>     // See if this is a regclass constraint.
>     unsigned RegClass;
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits