Skip to contentSkip to conversation

sound/core/seq: Initialize structure pointer to NULL to prevent undefined behavior

PATCH In review [email protected] 3 participants 3 messages Oct 2, 2025 – Oct 6, 2025 Sound

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sound/core/seq: Initialize structure pointer to NULL to prevent undefined behavior

From: HariKrishna Sagala <[email protected]>

This change ensures the structure pointer is explicitly initialized to NULL,preventing potential access to uninitialized memory. It improves code safety and avoids undefined behavior during pointer dereferencing.

---

Note:
Turned on the settings needed for sequencer MIDI and built a kernel image with those settings. The system booted up fine with no errors. However, couldn’t get the sequencer emulation to start.

 sound/core/seq/seq_midi_emul.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
sound/core/seq/seq_midi_emul.c +2 -2
index 81d2ef5e5811..f24c5a475c41 100644
@@ -647,7 +647,7 @@ static void snd_midi_channel_init(struct snd_midi_channel *p, int n)
  */
 static struct snd_midi_channel *snd_midi_channel_init_set(int n)
 {
-	struct snd_midi_channel *chan;
+	struct snd_midi_channel *chan = NULL;
 	int  i;
 
 	chan = kmalloc_array(n, sizeof(struct snd_midi_channel), GFP_KERNEL);
@@ -686,7 +686,7 @@ reset_all_channels(struct snd_midi_channel_set *chset)
  */
 struct snd_midi_channel_set *snd_midi_channel_alloc_set(int n)
 {
-	struct snd_midi_channel_set *chset;
+	struct snd_midi_channel_set *chset = NULL;
 
 	chset = kmalloc(sizeof(*chset), GFP_KERNEL);
 	if (chset) {

base-commit: 50c19e20ed2ef359cf155a39c8462b0a6351b9fa

Signature · 1 line
2.43.0
#1 lore raw
Quoted 7 lines · On Thu, 02 Oct 2025 19:43:00 +0200, [email protected] wrote:

From: HariKrishna Sagala <[email protected]>

This change ensures the structure pointer is explicitly initialized to NULL,preventing potential access to uninitialized memory. It improves code safety and avoids undefined behavior during pointer dereferencing.

Quoted 7 lines

---

Note:
Turned on the settings needed for sequencer MIDI and built a kernel image with those settings. The system booted up fine with no errors. However, couldn’t get the sequencer emulation to start.

Something really wrong in your test, I'm afraid.
See your patch below more closely:

Quoted 9 lines
 sound/core/seq/seq_midi_emul.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
sound/core/seq/seq_midi_emul.c +0 -0
index 81d2ef5e5811..f24c5a475c41 100644

sound/core/seq/seq_midi_emul.c static void snd_midi_channel_init(struct snd_midi_channel *p, int n) · line 653

Quoted 8 lines
sound/core/seq/seq_midi_emul.c +1 -1
@@ -647,7 +647,7 @@ static void snd_midi_channel_init(struct snd_midi_channel *p, int n)
  */
 static struct snd_midi_channel *snd_midi_channel_init_set(int n)
 {
-	struct snd_midi_channel *chan;
+	struct snd_midi_channel *chan = NULL;
 	int  i;
 
 	chan = kmalloc_array(n, sizeof(struct snd_midi_channel), GFP_KERNEL);

The variable chan is initialized at the very beginning. NULL initialization is utterly nonsense.

sound/core/seq/seq_midi_emul.c reset_all_channels(struct snd_midi_channel_set *chset) · line 691

Quoted 8 lines
sound/core/seq/seq_midi_emul.c +1 -1
@@ -686,7 +686,7 @@ reset_all_channels(struct snd_midi_channel_set *chset)
  */
 struct snd_midi_channel_set *snd_midi_channel_alloc_set(int n)
 {
-	struct snd_midi_channel_set *chset;
+	struct snd_midi_channel_set *chset = NULL;
 
 	chset = kmalloc(sizeof(*chset), GFP_KERNEL);

Here, too.

So all changes make really no sense.

thanks,

Takashi

#2 lore raw
In reply to [email protected]
Quoted 7 lines · On 10/2/25 13:43, [email protected] wrote:

From: HariKrishna Sagala <[email protected]>

This change ensures the structure pointer is explicitly initialized to NULL,preventing potential access to uninitialized memory. It improves code safety and avoids undefined behavior during pointer dereferencing.

Hi Harikrishna,

Thank you for detailing your attempt at testing. Unfortunately, your patch makes no substantive change. The variable you are setting to NULL is assigned before use, so there is no risk of undefined behavior.

I know this can seem discouraging to hear, but I do want you to encourage you to keep looking for ways you can make contributions.

Thanks,
David