[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20200210122426.682758235@linuxfoundation.org>
Date: Mon, 10 Feb 2020 04:29:03 -0800
From: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To: linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
stable@...r.kernel.org,
syzbot+30edd0f34bfcdc548ac4@...kaller.appspotmail.com,
Takashi Iwai <tiwai@...e.de>
Subject: [PATCH 5.5 030/367] ALSA: pcm: Fix memory leak at closing a stream without hw_free
From: Takashi Iwai <tiwai@...e.de>
commit 66f2d19f8116e16898f8d82e28573a384ddc430d upstream.
ALSA PCM core recently introduced a new managed PCM buffer allocation
mode that does allocate / free automatically at hw_params and
hw_free. However, it overlooked the code path directly calling
hw_free PCM ops at releasing the PCM substream, and it may result in a
memory leak as spotted by syzkaller when no buffer preallocation is
used (e.g. vmalloc buffer).
This patch papers over it with a slight refactoring. The hw_free ops
call and relevant tasks are unified in a new helper function, and call
it from both places.
Fixes: 0dba808eae26 ("ALSA: pcm: Introduce managed buffer allocation mode")
Reported-by: syzbot+30edd0f34bfcdc548ac4@...kaller.appspotmail.com
Cc: <stable@...r.kernel.org>
Link: https://lore.kernel.org/r/20200129195907.12197-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@...e.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
sound/core/pcm_native.c | 24 +++++++++++++++---------
1 file changed, 15 insertions(+), 9 deletions(-)
--- a/sound/core/pcm_native.c
+++ b/sound/core/pcm_native.c
@@ -785,10 +785,22 @@ end:
return err;
}
+static int do_hw_free(struct snd_pcm_substream *substream)
+{
+ int result = 0;
+
+ snd_pcm_sync_stop(substream);
+ if (substream->ops->hw_free)
+ result = substream->ops->hw_free(substream);
+ if (substream->managed_buffer_alloc)
+ snd_pcm_lib_free_pages(substream);
+ return result;
+}
+
static int snd_pcm_hw_free(struct snd_pcm_substream *substream)
{
struct snd_pcm_runtime *runtime;
- int result = 0;
+ int result;
if (PCM_RUNTIME_CHECK(substream))
return -ENXIO;
@@ -805,11 +817,7 @@ static int snd_pcm_hw_free(struct snd_pc
snd_pcm_stream_unlock_irq(substream);
if (atomic_read(&substream->mmap_count))
return -EBADFD;
- snd_pcm_sync_stop(substream);
- if (substream->ops->hw_free)
- result = substream->ops->hw_free(substream);
- if (substream->managed_buffer_alloc)
- snd_pcm_lib_free_pages(substream);
+ result = do_hw_free(substream);
snd_pcm_set_state(substream, SNDRV_PCM_STATE_OPEN);
pm_qos_remove_request(&substream->latency_pm_qos_req);
return result;
@@ -2466,9 +2474,7 @@ void snd_pcm_release_substream(struct sn
snd_pcm_drop(substream);
if (substream->hw_opened) {
- if (substream->ops->hw_free &&
- substream->runtime->status->state != SNDRV_PCM_STATE_OPEN)
- substream->ops->hw_free(substream);
+ do_hw_free(substream);
substream->ops->close(substream);
substream->hw_opened = 0;
}
Powered by blists - more mailing lists