Skip to contentSkip to conversation

aplay: Fix bugs in peak calculations

PATCH New Takashi Iwai 1 participant 6 messages Aug 24, 2021

lore
Share & tools
lore
#0 lore raw

cover aplay: Fix bugs in peak calculations

The code in aplay for sample peak calculations and VU-meter representation seems to have a lot of bugs. Let's fix them.

Takashi

===

Takashi Iwai (5):
aplay: Fix conversion of unsigned samples in peak calculation
aplay: Handle 16bit sample negative overflow in peak calculations
aplay: Don't pass most negative integer to abs() in peak calculations
aplay: Handle upper bound in peak calculations
aplay: Fix out-of-bound access in stereo VU meter drawing

 aplay/aplay.c | 28 +++++++++++++++++++---------
 1 file changed, 19 insertions(+), 9 deletions(-)
Signature · 1 line
2.26.2
#1 lore raw

PATCH 1/5 aplay: Fix conversion of unsigned samples in peak calculation

The XOR with the mask has to be applied before calculating abs value.

---
 aplay/aplay.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
aplay/aplay.c +6 -3
index cc51dcb48bba..91af244edb60 100644
@@ -1828,7 +1828,8 @@ static void compute_max_peak(u_char *data, size_t samples)
 				sval = le16toh(*valp);
 			else
 				sval = be16toh(*valp);
-			sval = abs(sval) ^ mask;
+			sval ^= mask;
+			sval = abs(sval);
 			if (max_peak[c] < sval)
 				max_peak[c] = sval;
 			valp++;
@@ -1848,11 +1849,12 @@ static void compute_max_peak(u_char *data, size_t samples)
 			} else {
 				val = (valp[0]<<16) | (valp[1]<<8) | valp[2];
 			}
+			val ^= mask;
 			/* Correct signed bit in 32-bit value */
 			if (val & (1<<(bits_per_sample-1))) {
 				val |= 0xff<<24;	/* Negate upper bits too */
 			}
-			val = abs(val) ^ mask;
+			val = abs(val);
 			if (max_peak[c] < val)
 				max_peak[c] = val;
 			valp += 3;
@@ -1871,7 +1873,8 @@ static void compute_max_peak(u_char *data, size_t samples)
 				val = le32toh(*valp);
 			else
 				val = be32toh(*valp);
-			val = abs(val) ^ mask;
+			val ^= mask;
+			val = abs(val);
 			if (max_peak[c] < val)
 				max_peak[c] = val;
 			valp++;
Signature · 1 line
2.26.2
#2 lore raw

PATCH 2/5 aplay: Handle 16bit sample negative overflow in peak calculations

In reply to Takashi Iwai

The handling of 16bit samples in the peak calculations has a bug when a sample with 0x8000 is passed. As abs() treats 32bit int, it returns 0x8000. And yet the code stores back into 16bit value again.

To fix that overflow, use 32bit value (i.e. val instead of sval) for the further calculations.

---
 aplay/aplay.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
aplay/aplay.c +3 -3
index 91af244edb60..c884346c9f25 100644
@@ -1829,9 +1829,9 @@ static void compute_max_peak(u_char *data, size_t samples)
 			else
 				sval = be16toh(*valp);
 			sval ^= mask;
-			sval = abs(sval);
-			if (max_peak[c] < sval)
-				max_peak[c] = sval;
+			val = abs(sval);
+			if (max_peak[c] < val)
+				max_peak[c] = val;
 			valp++;
 			if (vumeter == VUMETER_STEREO)
 				c = !c;
Signature · 1 line
2.26.2
#3 lore raw

PATCH 3/5 aplay: Don't pass most negative integer to abs() in peak calculations

In reply to Takashi Iwai

The return value from abs() for the most negative integer is undefined. Cap it properly for the 32bit sample handling.

---
 aplay/aplay.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
aplay/aplay.c +4 -1
index c884346c9f25..2543de5b6cd8 100644
@@ -1874,7 +1874,10 @@ static void compute_max_peak(u_char *data, size_t samples)
 			else
 				val = be32toh(*valp);
 			val ^= mask;
-			val = abs(val);
+			if (val == 0x80000000U)
+				val = 0x7fffffff;
+			else
+				val = abs(val);
 			if (max_peak[c] < val)
 				max_peak[c] = val;
 			valp++;
Signature · 1 line
2.26.2
#4 lore raw

PATCH 4/5 aplay: Handle upper bound in peak calculations

In reply to Takashi Iwai

Make sure that the calculated max_peak[] won't go beyond the sample max resolution.

---
 aplay/aplay.c | 2 ++
 1 file changed, 2 insertions(+)
aplay/aplay.c +2 -0
index 2543de5b6cd8..a51a37ba34bd 100644
@@ -1898,6 +1898,8 @@ static void compute_max_peak(u_char *data, size_t samples)
 		max = 0x7fffffff;
 
 	for (c = 0; c < ichans; c++) {
+		if (max_peak[c] > max)
+			max_peak[c] = max;
 		if (bits_per_sample > 16)
 			perc[c] = max_peak[c] / (max / 100);
 		else
Signature · 1 line
2.26.2
#5 lore raw

PATCH 5/5 aplay: Fix out-of-bound access in stereo VU meter drawing

In reply to Takashi Iwai

The left channel drawing of a stereo VU meter has a bug where it may access a negative array index.

---
 aplay/aplay.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
aplay/aplay.c +6 -4
index a51a37ba34bd..63a4e3437fd9 100644
@@ -1758,10 +1758,12 @@ static void print_vu_meter_stereo(int *perc, int *maxperc)
 		if (c)
 			memset(line + bar_length + 6 + 1, '#', p);
 		else
-			memset(line + bar_length - p - 1, '#', p);
-		p = maxperc[c] * bar_length / 100;
-		if (p > bar_length)
-			p = bar_length;
+			memset(line + bar_length - p, '#', p);
+		p = maxperc[c] * bar_length / 100 - 1;
+		if (p < 0)
+			p = 0;
+		else if (p >= bar_length)
+			p = bar_length - 1;
 		if (c)
 			line[bar_length + 6 + 1 + p] = '+';
 		else
Signature · 1 line
2.26.2