[FFmpeg-cvslog] r18936 - in trunk/libavcodec: Makefile mlpdsp.c

[Ramiro Polla]

ramiro wrote:
> Author: ramiro
> Date: Mon May 25 03:53:43 2009
> New Revision: 18936
> 
> Log:
> mlp: Re-enable x86-optimized DSP functions.

 From r18903:
> +static void mlp_filter_channel_x86(int32_t *state, const int32_t *coeff,
> +                                   int firorder, int iirorder,
> +                                   unsigned int filter_shift, int32_t mask,
> +                                   int blocksize, int32_t *sample_buffer)
> +{
[...]
> +        : /* 0*/"+r"(state),
> +          /* 1*/"+r"(coeff),
> +          /* 2*/"+r"(sample_buffer),
> +          /* 3*/RDWRVAL(blocksize)
> +        :
> +          /* 4*/READVAL((x86_reg)mask),

llvm gives a lovely error message:

/home/fate/fate64/source/libavcodec/x86/mlpdsp.c:169:30: error: invalid 
use of a cast in a inline asm context requiring an l-value: remove the 
cast or build with -fheinous-gnu-extensions
                 "m"((x86_reg)mask),
                     ~~~~~~~~~^~~~

It only triggers on x86_32. The commit for that error message in llvm is 
http://www.mail-archive.com/cfe-commits at cs.uiuc.edu/msg07397.html .

I thought we used these kind of casts all over the place in swscale and 
such. Any suggestions on what should be done?

Ramiro Polla