ASoC: SOF: ipc: fix a race, leading to IPC timeouts

Currently on all supported platforms the IPC IRQ thread first signals
the sender when an IPC response is received from the DSP, then unmasks
the IPC interrupt. Those actions are performed without holding any
locks, so the thread can be interrupted between them. IPC timeouts
have been observed in such scenarios: if the sender is woken up and it
proceeds with sending the next message without unmasking the IPC
interrupt, it can miss the next response. This patch takes a spin-lock
to prevent the IRQ thread from being preempted at that point. It also
makes sure, that the next IPC transmission by the host cannot take
place before the IRQ thread has finished updating all the required IPC
registers.

Fixes: 53e0c72d98b ("ASoC: SOF: Add support for IPC IO between DSP and Host")
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

1 files changed, 0 insertions, 13 deletions

diff --git a/sound/soc/sof/ipc.c b/sound/soc/sof/ipc.c
index 894e68cbd69d..10304a90cf25 100644
--- a/sound/soc/sof/ipc.c
+++ b/sound/soc/sof/ipc.c
@@ -308,19 +308,8 @@ EXPORT_SYMBOL(sof_ipc_tx_message);
 int snd_sof_ipc_reply(struct snd_sof_dev *sdev, u32 msg_id)
 {
 	struct snd_sof_ipc_msg *msg = &sdev->ipc->msg;
-	unsigned long flags;
-
-	/*
-	 * Protect against a theoretical race with sof_ipc_tx_message(): if the
-	 * DSP is fast enough to receive an IPC message, reply to it, and the
-	 * host interrupt processing calls this function on a different core
-	 * from the one, where the sending is taking place, the message might
-	 * not yet be marked as expecting a reply.
-	 */
-	spin_lock_irqsave(&sdev->ipc_lock, flags);
 
 	if (msg->ipc_complete) {
-		spin_unlock_irqrestore(&sdev->ipc_lock, flags);
 		dev_err(sdev->dev, "error: no reply expected, received 0x%x",
 			msg_id);
 		return -EINVAL;
@@ -330,8 +319,6 @@ int snd_sof_ipc_reply(struct snd_sof_dev *sdev, u32 msg_id)
 	msg->ipc_complete = true;
 	wake_up(&msg->waitq);
 
-	spin_unlock_irqrestore(&sdev->ipc_lock, flags);
-
 	return 0;
 }
 EXPORT_SYMBOL(snd_sof_ipc_reply);