[FFmpeg-devel] [PATCH 4/4] avcodec/ra144enc: Fix invalid left shift of negative number

[Andreas Rheinhardt]

by replacing it with a multiplication. Said multiplication can't
overflow an int32_t because lpc_coefs is limited to 16 bit precision.

Fixes the FACE-test acodec-ra144 as well as part of #8217.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt at gmail.com>
---
I am unable to reproduce the division by zero mentioned in ticket #8217;
probably just an error on the reporter's part.

 libavcodec/ra144enc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libavcodec/ra144enc.c b/libavcodec/ra144enc.c
index cc4f381606..059f582334 100644
--- a/libavcodec/ra144enc.c
+++ b/libavcodec/ra144enc.c
@@ -477,8 +477,8 @@ static int ra144_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
                       LPC_ORDER, 16, lpc_coefs, shift, FF_LPC_TYPE_LEVINSON,
                       0, ORDER_METHOD_EST, 0, 12, 0);
     for (i = 0; i < LPC_ORDER; i++)
-        block_coefs[NBLOCKS - 1][i] = -(lpc_coefs[LPC_ORDER - 1][i] <<
-                                        (12 - shift[LPC_ORDER - 1]));
+        block_coefs[NBLOCKS - 1][i] = -lpc_coefs[LPC_ORDER - 1][i]
+                                       * (1 << (12 - shift[LPC_ORDER - 1]));
 
     /**
      * TODO: apply perceptual weighting of the input speech through bandwidth
-- 
2.20.1