linux-kernel - [PATCH 1/3] soundwire: intel: prevent pm_runtime resume prior to system suspend

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20220420023241.14335-2-yung-chuan.liao@linux.intel.com>
Date:   Wed, 20 Apr 2022 10:32:39 +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,
        gregkh@...uxfoundation.org, srinivas.kandagatla@...aro.org,
        pierre-louis.bossart@...ux.intel.com, sanyog.r.kale@...el.com,
        bard.liao@...el.com
Subject: [PATCH 1/3] soundwire: intel: prevent pm_runtime resume prior to system suspend

From: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>

commit e38f9ff63e6d ("ACPI: scan: Do not add device IDs from _CID if _HID is not valid")
exposes a race condition on a TGL RVP device leading to a timeout.

The detailed analysis shows the RT711 codec driver scheduling a jack
detection workqueue while attaching during a spurious pm_runtime
resume, and the work function happens to be scheduled after the
manager device is suspended.

The direct link between this ACPI patch and a spurious pm_runtime
resume is not obvious; the most likely explanation is that a change in
the ACPI device linked list management modifies the order in which the
pm_runtime device status is checked and exposes a race condition that
was probably present for a very long time, but was not identified.

We already have a check in the .prepare stage, where we will resume to
full power from specific clock-stop modes. In all other cases, we
don't need to resume to full power by default. Adding the
SMART_SUSPEND flag prevents the spurious resume from happening.

BugLink: https://github.com/thesofproject/linux/issues/3459
Fixes: 029bfd1cd53cd ("soundwire: intel: conditionally exit clock stop mode on system suspend")
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@...ux.intel.com>
Reviewed-by: Rander Wang <rander.wang@...el.com>
Signed-off-by: Bard Liao <yung-chuan.liao@...ux.intel.com>
---
 drivers/soundwire/intel.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/soundwire/intel.c b/drivers/soundwire/intel.c
index 63101f1ba271..32e5fdb823c4 100644
--- a/drivers/soundwire/intel.c
+++ b/drivers/soundwire/intel.c
@@ -1293,6 +1293,9 @@ static int intel_link_probe(struct auxiliary_device *auxdev,
 	/* use generic bandwidth allocation algorithm */
 	sdw->cdns.bus.compute_params = sdw_compute_params;

+	/* avoid resuming from pm_runtime suspend if it's not required */
+	dev_pm_set_driver_flags(dev, DPM_FLAG_SMART_SUSPEND);
+
 	ret = sdw_bus_master_add(bus, dev, dev->fwnode);
 	if (ret) {
 		dev_err(dev, "sdw_bus_master_add fail: %d\n", ret);
-- 
2.17.1