audio: copier: Fix memory corruption with TRANS_GAIN with gcc

The dst pointer must not used directly while applying TRANS_GAIN to avoid
corrupting memory in heap.

The fade_in_sg_count must be also reset to zero at each startup to correct
the fade effect.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>

Author: ujfalusi

src/audio/copier/copier_gain.c

@@ -49,6 +49,7 @@ __cold int copier_gain_set_params(struct comp_dev *dev, struct copier_gain_param
 			gain_params->silence_sg_length =
 				frames * dmic_glb_cfg->ext_global_cfg.silence_period;
 			gain_params->fade_sg_length = frames * fade_period;
+			gain_params->fade_in_sg_count = 0;
 		}
 		break;
 	default:

src/audio/copier/copier_generic.c

@@ -169,7 +169,7 @@ int copier_gain_input16(struct comp_buffer *buff, enum copier_gain_state state,
 
 			/* Apply fade */
 			for (j = 0; j < nch; j++) {
-				dst += j;
+				dst_tmp = dst + j;
 				/* Quadratic fade part in Q15 format*/
 				gain_env_sq = q_multsr_16x16(gain_env[j], gain_env[j], 15);
 
@@ -180,8 +180,8 @@ int copier_gain_input16(struct comp_buffer *buff, enum copier_gain_state state,
 						      gain_env_sq, 15);
 
 				for (i = 0; i < nmax; i += nch)
-					dst[i] = q_multsr_sat_16x16(dst[i], gain,
-								    GAIN_Q10_INT_SHIFT);
+					dst_tmp[i] = q_multsr_sat_16x16(dst_tmp[i], gain,
+									GAIN_Q10_INT_SHIFT);
 			}
 			samples -= nmax;
 			dst = audio_stream_wrap(&buff->stream, dst + nmax);
@@ -260,7 +260,7 @@ int copier_gain_input32(struct comp_buffer *buff, enum copier_gain_state state,
 
 			/* Apply fade */
 			for (j = 0; j < nch; j++) {
-				dst += j;
+				dst_tmp = dst + j;
 				/* Quadratic fade part in Q15 format*/
 				gain_env_sq = q_multsr_16x16(gain_env[j], gain_env[j], 15);
 
@@ -271,8 +271,8 @@ int copier_gain_input32(struct comp_buffer *buff, enum copier_gain_state state,
 						      gain_env_sq, 15);
 
 				for (i = 0; i < nmax; i += nch)
-					dst[i] =  q_multsr_sat_32x32(dst[i], gain,
-								     GAIN_Q10_INT_SHIFT);
+					dst_tmp[i] =  q_multsr_sat_32x32(dst_tmp[i], gain,
+									 GAIN_Q10_INT_SHIFT);
 			}
 			samples -= nmax;
 			dst = audio_stream_wrap(&buff->stream, dst + nmax);