[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20200817152923.3259-4-yung-chuan.liao@linux.intel.com>
Date: Mon, 17 Aug 2020 23:29:14 +0800
From: Bard Liao <yung-chuan.liao@...ux.intel.com>
To: alsa-devel@...a-project.org, vkoul@...nel.org
Cc: vinod.koul@...aro.org, linux-kernel@...r.kernel.org, tiwai@...e.de,
broonie@...nel.org, gregkh@...uxfoundation.org, jank@...ence.com,
srinivas.kandagatla@...aro.org, rander.wang@...ux.intel.com,
ranjani.sridharan@...ux.intel.com, hui.wang@...onical.com,
pierre-louis.bossart@...ux.intel.com, sanyog.r.kale@...el.com,
mengdong.lin@...el.com, bard.liao@...el.com
Subject: [PATCH v2 03/12] soundwire: intel: fix race condition on system resume
From: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
Previous patches took care of the case where the master device is
pm_runtime 'suspended' when a system suspend occurs.
In the case where the master device was not suspended, e.g. if suspend
occurred while streaming audio, Intel validation noticed a race
condition: the pm_runtime suspend may conflict with the enumeration
started by the system resume.
This can be simply fixed by updating the status before exiting system
resume.
GitHub issue: https://github.com/thesofproject/linux/issues/1482
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@...ux.intel.com>
---
drivers/soundwire/intel.c | 12 ++++++++++++
1 file changed, 12 insertions(+)
diff --git a/drivers/soundwire/intel.c b/drivers/soundwire/intel.c
index 0e21bae3cd19..00c5de1250ec 100644
--- a/drivers/soundwire/intel.c
+++ b/drivers/soundwire/intel.c
@@ -1528,6 +1528,18 @@ static int intel_resume(struct device *dev)
return ret;
}
+ /*
+ * after system resume, the pm_runtime suspend() may kick in
+ * during the enumeration, before any children device force the
+ * master device to remain active. Using pm_runtime_get()
+ * routines is not really possible, since it'd prevent the
+ * master from suspending.
+ * A reasonable compromise is to update the pm_runtime
+ * counters and delay the pm_runtime suspend by several
+ * seconds, by when all enumeration should be complete.
+ */
+ pm_runtime_mark_last_busy(dev);
+
return ret;
}
--
2.17.1
Powered by blists - more mailing lists