Avoid too large shift exponent in apply_inter_tes() by using fIsLessThan.

Bug: 146936613
Test: atest DecoderTestXheAac ; atest DecoderTestAacDrc
Change-Id: I8414072929ec8f786e0c003f18ab453dbacb9e5e
This commit is contained in:
Fraunhofer IIS FDK 2019-11-13 16:08:57 +01:00 committed by Jean-Michel Trivi
parent 1020e48d6e
commit 5d09563196
1 changed files with 2 additions and 11 deletions

View File

@ -702,20 +702,11 @@ static void apply_inter_tes(FIXP_DBL **qmfReal, FIXP_DBL **qmfImag,
gain_sf[i] += gamma_sf + 1; /* +1 because of fMultDiv2() */ gain_sf[i] += gamma_sf + 1; /* +1 because of fMultDiv2() */
/* set gain to at least 0.2f */ /* set gain to at least 0.2f */
FIXP_DBL point_two = FL2FXCONST_DBL(0.8f); /* scaled up by 2 */
int point_two_sf = -2;
FIXP_DBL tmp = gain[i];
if (point_two_sf < gain_sf[i]) {
point_two >>= gain_sf[i] - point_two_sf;
} else {
tmp >>= point_two_sf - gain_sf[i];
}
/* limit and calculate gain[i]^2 too */ /* limit and calculate gain[i]^2 too */
FIXP_DBL gain_pow2; FIXP_DBL gain_pow2;
int gain_pow2_sf; int gain_pow2_sf;
if (tmp < point_two) {
if (fIsLessThan(gain[i], gain_sf[i], FL2FXCONST_DBL(0.2f), 0)) {
gain[i] = FL2FXCONST_DBL(0.8f); gain[i] = FL2FXCONST_DBL(0.8f);
gain_sf[i] = -2; gain_sf[i] = -2;
gain_pow2 = FL2FXCONST_DBL(0.64f); gain_pow2 = FL2FXCONST_DBL(0.64f);