[FFmpeg-cvslog] avcodec/aacsbr_fixed: Check shift in sbr_hf_assemble()

[Michael Niedermayer]

ffmpeg | branch: release/2.8 | Michael Niedermayer <michael at niedermayer.cc> | Tue Jun 13 16:25:59 2017 +0200| [fd0da0eef0076732a65d173186081c125e28068f] | committer: Michael Niedermayer

avcodec/aacsbr_fixed: Check shift in sbr_hf_assemble()

Fixes: runtime error: shift exponent -10 is negative

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
(cherry picked from commit d1992448d37f7cfa2acda5cc729dc0ff1b019390)
Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=fd0da0eef0076732a65d173186081c125e28068f
---

 libavcodec/aacsbr_fixed.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/libavcodec/aacsbr_fixed.c b/libavcodec/aacsbr_fixed.c
index 2410c25d80..a12ecd19c9 100644
--- a/libavcodec/aacsbr_fixed.c
+++ b/libavcodec/aacsbr_fixed.c
@@ -578,22 +578,30 @@ static void sbr_hf_assemble(int Y1[38][64][2],
 
                 SoftFloat *in  = sbr->s_m[e];
                 for (m = 0; m+1 < m_max; m+=2) {
+                    int shift2;
                     shift = 22 - in[m  ].exp;
+                    shift2= 22 - in[m+1].exp;
+                    if (shift < 1 || shift2 < 1) {
+                        av_log(NULL, AV_LOG_ERROR, "Overflow in sbr_hf_assemble, shift=%d,%d\n", shift, shift2);
+                        return;
+                    }
                     if (shift < 32) {
                         round = 1 << (shift-1);
                         out[2*m  ] += (in[m  ].mant * A + round) >> shift;
                     }
 
-                    shift = 22 - in[m+1].exp;
-                    if (shift < 32) {
-                        round = 1 << (shift-1);
-                        out[2*m+2] += (in[m+1].mant * B + round) >> shift;
+                    if (shift2 < 32) {
+                        round = 1 << (shift2-1);
+                        out[2*m+2] += (in[m+1].mant * B + round) >> shift2;
                     }
                 }
                 if(m_max&1)
                 {
                     shift = 22 - in[m  ].exp;
-                    if (shift < 32) {
+                    if (shift < 1) {
+                        av_log(NULL, AV_LOG_ERROR, "Overflow in sbr_hf_assemble, shift=%d\n", shift);
+                        return;
+                    } else if (shift < 32) {
                         round = 1 << (shift-1);
                         out[2*m  ] += (in[m  ].mant * A + round) >> shift;
                     }