[FFmpeg-cvslog] swresample/rematrix: Use clipping s16 rematrixing if overflows are possible

[Michael Niedermayer]

ffmpeg | branch: master | Michael Niedermayer <michael at niedermayer.cc> | Sun May 15 21:34:37 2016 +0200| [2f76157eb05bf63725f96167feda6b2e07501c7e] | committer: Michael Niedermayer

swresample/rematrix: Use clipping s16 rematrixing if overflows are possible

Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=2f76157eb05bf63725f96167feda6b2e07501c7e
---

 libswresample/rematrix.c          |   19 ++++++++++++++++---
 libswresample/rematrix_template.c |    7 ++++++-
 2 files changed, 22 insertions(+), 4 deletions(-)

diff --git a/libswresample/rematrix.c b/libswresample/rematrix.c
index f18c9f6..ddba043 100644
--- a/libswresample/rematrix.c
+++ b/libswresample/rematrix.c
@@ -32,6 +32,9 @@
 
 #define TEMPLATE_REMATRIX_S16
 #include "rematrix_template.c"
+#define TEMPLATE_CLIP
+#include "rematrix_template.c"
+#undef TEMPLATE_CLIP
 #undef TEMPLATE_REMATRIX_S16
 
 #define TEMPLATE_REMATRIX_S32
@@ -367,23 +370,33 @@ av_cold int swri_rematrix_init(SwrContext *s){
             return r;
     }
     if (s->midbuf.fmt == AV_SAMPLE_FMT_S16P){
+        int maxsum = 0;
         s->native_matrix = av_calloc(nb_in * nb_out, sizeof(int));
         s->native_one    = av_mallocz(sizeof(int));
         if (!s->native_matrix || !s->native_one)
             return AVERROR(ENOMEM);
         for (i = 0; i < nb_out; i++) {
             double rem = 0;
+            int sum = 0;
 
             for (j = 0; j < nb_in; j++) {
                 double target = s->matrix[i][j] * 32768 + rem;
                 ((int*)s->native_matrix)[i * nb_in + j] = lrintf(target);
                 rem += target - ((int*)s->native_matrix)[i * nb_in + j];
+                sum += FFABS(((int*)s->native_matrix)[i * nb_in + j]);
             }
+            maxsum = FFMAX(maxsum, sum);
         }
         *((int*)s->native_one) = 32768;
-        s->mix_1_1_f = (mix_1_1_func_type*)copy_s16;
-        s->mix_2_1_f = (mix_2_1_func_type*)sum2_s16;
-        s->mix_any_f = (mix_any_func_type*)get_mix_any_func_s16(s);
+        if (maxsum <= 32768) {
+            s->mix_1_1_f = (mix_1_1_func_type*)copy_s16;
+            s->mix_2_1_f = (mix_2_1_func_type*)sum2_s16;
+            s->mix_any_f = (mix_any_func_type*)get_mix_any_func_s16(s);
+        } else {
+            s->mix_1_1_f = (mix_1_1_func_type*)copy_clip_s16;
+            s->mix_2_1_f = (mix_2_1_func_type*)sum2_clip_s16;
+            s->mix_any_f = (mix_any_func_type*)get_mix_any_func_clip_s16(s);
+        }
     }else if(s->midbuf.fmt == AV_SAMPLE_FMT_FLTP){
         s->native_matrix = av_calloc(nb_in * nb_out, sizeof(float));
         s->native_one    = av_mallocz(sizeof(float));
diff --git a/libswresample/rematrix_template.c b/libswresample/rematrix_template.c
index 95a3b9a..add65e3 100644
--- a/libswresample/rematrix_template.c
+++ b/libswresample/rematrix_template.c
@@ -31,11 +31,16 @@
 #    define INTER double
 #    define RENAME(x) x ## _double
 #elif defined(TEMPLATE_REMATRIX_S16)
-#    define R(x) (((x) + 16384)>>15)
 #    define SAMPLE int16_t
 #    define COEFF int
 #    define INTER int
+#  ifdef TEMPLATE_CLIP
+#    define R(x) av_clip_int16(((x) + 16384)>>15)
+#    define RENAME(x) x ## _clip_s16
+#  else
+#    define R(x) (((x) + 16384)>>15)
 #    define RENAME(x) x ## _s16
+#  endif
 #elif defined(TEMPLATE_REMATRIX_S32)
 #    define R(x) (((x) + 16384)>>15)
 #    define SAMPLE int32_t