[FFmpeg-devel] [PATCH] mips: ac3 downmix updated to the new data layout.

Babic, Nedeljko nbabic at mips.com
Thu Dec 13 11:51:38 CET 2012


I will resubmit this in two patches as you suggested:
one with lines that are not re-indented and one for fixing indention.

From: ffmpeg-devel-bounces at ffmpeg.org [ffmpeg-devel-bounces at ffmpeg.org] on behalf of Carl Eugen Hoyos [cehoyos at ag.or.at]
Sent: Thursday, December 13, 2012 11:13
To: ffmpeg-devel at ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH] mips: ac3 downmix updated to the new        data layout.

Nedeljko Babic <nbabic <at> mips.com> writes:

> -        "lwc1   %[matrix_j],    0(%[matrix_p])                      \n\t"
> -        "lwc1   %[matrix_j2],   4(%[matrix_p])                      \n\t"
> -        "lwc1   %[samples0],    0(%[samples_p])                     \n\t"
> -        "lwc1   %[samples1],    4(%[samples_p])                     \n\t"
> -        "lwc1   %[samples2],    8(%[samples_p])                     \n\t"
> -        "lwc1   %[samples3],    12(%[samples_p])                    \n\t"
> -        "addiu  %[matrix_p],    8                                   \n\t"
> -        "madd.s %[v0],          %[v0],  %[samples0],    %[matrix_j] \n\t"
> -        "madd.s %[v1],          %[v1],  %[samples1],    %[matrix_j] \n\t"
> -        "madd.s %[v2],          %[v2],  %[samples2],    %[matrix_j] \n\t"
> -        "madd.s %[v3],          %[v3],  %[samples3],    %[matrix_j] \n\t"
> -        "madd.s %[v4],          %[v4],  %[samples0],    %[matrix_j2]\n\t"
> -        "madd.s %[v5],          %[v5],  %[samples1],    %[matrix_j2]\n\t"
> -        "madd.s %[v6],          %[v6],  %[samples2],    %[matrix_j2]\n\t"

> +    "lwc1   %[matrix_j],    0(%[matrix_p])                      \n\t"
> +    "lwc1   %[matrix_j2],   4(%[matrix_p])                      \n\t"
> +    "lwxc1  %[samples0],    %[i](%[samples_p])                  \n\t"
> +    "lwxc1  %[samples1],    %[i1](%[samples_p])                 \n\t"
> +    "lwxc1  %[samples2],    %[i2](%[samples_p])                 \n\t"
> +    "lwxc1  %[samples3],    %[i3](%[samples_p])                 \n\t"
> +    "addiu  %[matrix_p],    8                                   \n\t"
> +    "addiu  %[samples_x],   4                                   \n\t"
> +    "madd.s %[v0],          %[v0],  %[samples0],    %[matrix_j] \n\t"
> +    "madd.s %[v1],          %[v1],  %[samples1],    %[matrix_j] \n\t"
> +    "madd.s %[v2],          %[v2],  %[samples2],    %[matrix_j] \n\t"
> +    "madd.s %[v3],          %[v3],  %[samples3],    %[matrix_j] \n\t"
> +    "madd.s %[v4],          %[v4],  %[samples0],    %[matrix_j2]\n\t"
> +    "madd.s %[v5],          %[v5],  %[samples1],    %[matrix_j2]\n\t"
> +    "madd.s %[v6],          %[v6],  %[samples2],    %[matrix_j2]\n\t"

It is your code and I certainly cannot comment on it
but it was often argued here (and I believe with some
good reason) that if you do not re-indent these lines,
it is both easier to review this patch and easier to
do possible future regression tests to find possible
As a side-effect, the interested reader would immediately
see what changed in above block.

You could then send a second patch to fix the indentation
(or leave it slightly broken).

Feel free to ignore, thank you for fixing this!

Carl Eugen

ffmpeg-devel mailing list
ffmpeg-devel at ffmpeg.org

More information about the ffmpeg-devel mailing list