ALSA: rawmidi: Fix kvfree() call in spinlock

Commit Message

Takashi Iwai Nov. 25, 2024, 2:20 p.m. UTC

At the conversion of locking with guard(), I overlooked that kvfree()
must not be called inside the spinlock unlike kfree(), and this was
caught by syzkaller now.

This patch reverts the conversion partially for restoring the kvfree()
call outside the spinlock.  It's not trivial to use guard() in this
context, unfortunately.

Fixes: 84bb065b316e ("ALSA: rawmidi: Use guard() for locking")
Reported-by: [e-mail]
Reported-by: Eric Dumazet <[e-mail]>
Closes: https://lore.kernel.org/[e-mail]
Cc: <[e-mail]>
Signed-off-by: Takashi Iwai <[e-mail]>
---
 sound/core/rawmidi.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Patch

diff --git a/sound/core/rawmidi.c b/sound/core/rawmidi.c
index 03306be5fa02..348ce1b7725e 100644
--- a/sound/core/rawmidi.c
+++ b/sound/core/rawmidi.c
@@ -724,8 +724,9 @@  static int resize_runtime_buffer(struct snd_rawmidi_substream *substream,
 		newbuf = kvzalloc(params->buffer_size, GFP_KERNEL);
 		if (!newbuf)
 			return -ENOMEM;
-		guard(spinlock_irq)(&substream->lock);
+		spin_lock_irq(&substream->lock);
 		if (runtime->buffer_ref) {
+			spin_unlock_irq(&substream->lock);
 			kvfree(newbuf);
 			return -EBUSY;
 		}
@@ -733,6 +734,7 @@  static int resize_runtime_buffer(struct snd_rawmidi_substream *substream,
 		runtime->buffer = newbuf;
 		runtime->buffer_size = params->buffer_size;
 		__reset_runtime_ptrs(runtime, is_input);
+		spin_unlock_irq(&substream->lock);
 		kvfree(oldbuf);
 	}
 	runtime->avail_min = params->avail_min;