OrgACRN
/
sof
mirror of https://github.com/thesofproject/sof.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Math: IIR: Fix overflow in DF2T generic C version biquad output 

With coefficients scaling that creates above 1.0 absolute value
internal biquad output before scaling gain and shift there's
overflow in variable tmp calculation that is used for output
and for recursive a1, a2 coefficients multiplication.

Fixes: #3678

Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>
This commit is contained in:
Seppo Ingalsuo 2020-12-10 16:42:55 +02:00 committed by Liam Girdwood
parent 918f22c206
commit 9c33333cf5
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/math/iir_df2t_generic.c
View File

@ -65,10 +65,10 @@ int32_t iir_df2t(struct iir_state_df2t *iir, int32_t x)
for (i = 0; i < iir->biquads_in_series; i++) {
/* Compute output: Delay is Q3.61
* Q2.30 x Q1.31 -> Q3.61
* Shift Q3.61 to Q3.31 with rounding
* Shift Q3.61 to Q3.31 with rounding, saturate to Q1.31
*/
acc = ((int64_t)iir->coef[c + 4]) * in + iir->delay[d];
tmp = (int32_t)Q_SHIFT_RND(acc, 61, 31);
acc = ((int64_t)iir->coef[c + 4]) * in + iir->delay[d]; /* Coef b0 */
tmp = (int32_t)sat_int32(Q_SHIFT_RND(acc, 61, 31));
/* Compute first delay */
acc = iir->delay[d + 1];