diff options
| author | Felicia Lim <flim@google.com> | 2021-02-22 22:29:14 -0800 |
|---|---|---|
| committer | Felicia Lim <flim@google.com> | 2021-03-02 21:35:16 -0800 |
| commit | 7b05f44f4baadf34d8d1073f4ff69f1806d5cdb4 (patch) | |
| tree | a39ed67321c97dd8a60dcefb347a48cbbf7caa7e /silk/LPC_fit.c | |
| parent | 16286a25fdd865c66a837a73b65fbaa7b25bf484 (diff) | |
| download | opus-7b05f44f4baadf34d8d1073f4ff69f1806d5cdb4.tar.gz | |
celt_lpc: avoid overflows when computing lpcs in fixed point
The LPCs are computed in 32-bit, so increase the allowed range from +/-8
to +/-64 to avoid overflows caught during fuzzing. Before downshifting
back down to the +/-8 range in the final 16-bit output, perform bandwidth
extension to avoid any additional overflow issues.
Diffstat (limited to 'silk/LPC_fit.c')
| -rw-r--r-- | silk/LPC_fit.c | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/silk/LPC_fit.c b/silk/LPC_fit.c index cdea4f3a..c0690a1f 100644 --- a/silk/LPC_fit.c +++ b/silk/LPC_fit.c @@ -31,7 +31,8 @@ POSSIBILITY OF SUCH DAMAGE. #include "SigProc_FIX.h" -/* Convert int32 coefficients to int16 coefs and make sure there's no wrap-around */ +/* Convert int32 coefficients to int16 coefs and make sure there's no wrap-around. + This logic is reused in _celt_lpc(). Any bug fixes should also be applied there. */ void silk_LPC_fit( opus_int16 *a_QOUT, /* O Output signal */ opus_int32 *a_QIN, /* I/O Input signal */ |