Crash Fix: Fix HRTF resampler delay misuse

The delay value should be scaled by the resampling ratio, similar to
how it already is when allocating the impulse buffer. This went
undetected, as it scribbled over other memory without causing immediate
crashes, but instead later heap corruption.

Signed-off-by: Christopher Snowhill <kode54@gmail.com>
This commit is contained in:
Christopher Snowhill 2025-02-14 18:45:18 -08:00
parent 75441bc5fa
commit ab62de0fa2

View file

@ -218,8 +218,10 @@ static impulseSetCache *_sharedController = nil;
data->get_direction_data(elevation, azimuth, speaker.distance, hrtfLeft, hrtfRight);
if(resampling) {
soxr_oneshot(sampleRateOfSource, sampleRate, 1, &hrtfLeft.impulse_response[0], sampleCountExact, NULL, &hrtfData[((hrtfLeft.delay + 2) >> 2) + actualSampleCount * i * 2], sampleCountResampled, NULL, &io_spec, &q_spec, &runtime_spec);
soxr_oneshot(sampleRateOfSource, sampleRate, 1, &hrtfRight.impulse_response[0], sampleCountExact, NULL, &hrtfData[((hrtfRight.delay + 2) >> 2) + actualSampleCount * (i * 2 + 1)], sampleCountResampled, NULL, &io_spec, &q_spec, &runtime_spec);
ssize_t leftDelay = (ssize_t)((double)(hrtfLeft.delay) * 0.25 * sampleRate / sampleRateOfSource);
ssize_t rightDelay = (ssize_t)((double)(hrtfRight.delay) * 0.25 * sampleRate / sampleRateOfSource);
soxr_oneshot(sampleRateOfSource, sampleRate, 1, &hrtfLeft.impulse_response[0], sampleCountExact, NULL, &hrtfData[leftDelay + actualSampleCount * i * 2], sampleCountResampled, NULL, &io_spec, &q_spec, &runtime_spec);
soxr_oneshot(sampleRateOfSource, sampleRate, 1, &hrtfRight.impulse_response[0], sampleCountExact, NULL, &hrtfData[rightDelay + actualSampleCount * (i * 2 + 1)], sampleCountResampled, NULL, &io_spec, &q_spec, &runtime_spec);
} else {
cblas_scopy(sampleCountExact, &hrtfLeft.impulse_response[0], 1, &hrtfData[((hrtfLeft.delay + 2) >> 2) + actualSampleCount * i * 2], 1);
cblas_scopy(sampleCountExact, &hrtfRight.impulse_response[0], 1, &hrtfData[((hrtfRight.delay + 2) >> 2) + actualSampleCount * (i * 2 + 1)], 1);