intfloat_readwrite: fix signed addition overflows

These additions might overflow the signed range for large
input values.  Converting to unsigned before the addition
rather than after avoids such undefined behaviour.  The
result under normal two's complement wraparound remains
unchanged.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 88d1e2b2b0)

Signed-off-by: Anton Khirnov <anton@khirnov.net>

libavutil/intfloat_readwrite.c

@@ -30,13 +30,13 @@
 #include "intfloat_readwrite.h"
 
 double av_int2dbl(int64_t v){
-    if(v+v > 0xFFEULL<<52)
+    if((uint64_t)v+v > 0xFFEULL<<52)
         return NAN;
     return ldexp(((v&((1LL<<52)-1)) + (1LL<<52)) * (v>>63|1), (v>>52&0x7FF)-1075);
 }
 
 float av_int2flt(int32_t v){
-    if(v+v > 0xFF000000U)
+    if((uint32_t)v+v > 0xFF000000U)
         return NAN;
     return ldexp(((v&0x7FFFFF) + (1<<23)) * (v>>31|1), (v>>23&0xFF)-150);
 }