Received-SPF: pass (google.com: domain of linux-bluetooth-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
To:     marcel@holtmann.org, linux-bluetooth@vger.kernel.org
Cc:     len.brown@intel.com, chromeos-bluetooth-upstreaming@chromium.org,
        linux-pm@vger.kernel.org, rafael@kernel.org,
        todd.e.brandt@linux.intel.com, rui.zhang@intel.com,
        Abhishek Pandit-Subedi <abhishekpandit@chromium.org>,
        "David S. Miller" <davem@davemloft.net>,
        Johan Hedberg <johan.hedberg@gmail.com>,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        Jakub Kicinski <kuba@kernel.org>
Subject: [PATCH v3] Bluetooth: Allow suspend even when preparation has failed
Date:   Fri,  5 Jun 2020 13:50:15 -0700
Message-Id: <20200605135009.v3.1.I0ec31d716619532fc007eac081e827a204ba03de@changeid>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Sender: linux-bluetooth-owner@vger.kernel.org
Precedence: bulk

It is preferable to allow suspend even when Bluetooth has problems
preparing for sleep. When Bluetooth fails to finish preparing for
suspend, log the error and allow the suspend notifier to continue
instead.

To also make it clearer why suspend failed, change bt_dev_dbg to
bt_dev_err when handling the suspend timeout.

Fixes: dd522a7429b07e ("Bluetooth: Handle LE devices during suspend")
Reported-by: Len Brown <len.brown@intel.com>
Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
---
To verify this is properly working, I added an additional change to
hci_suspend_wait_event to always return -16. This validates that suspend
continues even when an error has occurred during the suspend
preparation.

Example on Chromebook:
[   55.834524] PM: Syncing filesystems ... done.
[   55.841930] PM: Preparing system for sleep (s2idle)
[   55.940492] Bluetooth: hci_core.c:hci_suspend_notifier() hci0: Suspend notifier action (3) failed: -16
[   55.940497] Freezing user space processes ... (elapsed 0.001 seconds) done.
[   55.941692] OOM killer disabled.
[   55.941693] Freezing remaining freezable tasks ... (elapsed 0.000 seconds) done.
[   55.942632] PM: Suspending system (s2idle)

I ran this through a suspend_stress_test in the following scenarios:
* Peer classic device connected: 50+ suspends
* No devices connected: 100 suspends
* With the above test case returning -EBUSY: 50 suspends

I also ran this through our automated testing for suspend and wake on
BT from suspend continues to work.


Changes in v3:
- Changed printf format for unsigned long

Changes in v2:
- Added fixes and reported-by tags

 net/bluetooth/hci_core.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index dbe2d79f233fba..83ce665d3cbfb0 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -3289,10 +3289,10 @@ static int hci_suspend_wait_event(struct hci_dev *hdev)
 				     WAKE_COND, SUSPEND_NOTIFIER_TIMEOUT);
 
 	if (ret == 0) {
-		bt_dev_dbg(hdev, "Timed out waiting for suspend");
+		bt_dev_err(hdev, "Timed out waiting for suspend events");
 		for (i = 0; i < __SUSPEND_NUM_TASKS; ++i) {
 			if (test_bit(i, hdev->suspend_tasks))
-				bt_dev_dbg(hdev, "Bit %d is set", i);
+				bt_dev_err(hdev, "Suspend timeout bit: %d", i);
 			clear_bit(i, hdev->suspend_tasks);
 		}
 
@@ -3360,12 +3360,15 @@ static int hci_suspend_notifier(struct notifier_block *nb, unsigned long action,
 		ret = hci_change_suspend_state(hdev, BT_RUNNING);
 	}
 
-	/* If suspend failed, restore it to running */
-	if (ret && action == PM_SUSPEND_PREPARE)
-		hci_change_suspend_state(hdev, BT_RUNNING);
-
 done:
-	return ret ? notifier_from_errno(-EBUSY) : NOTIFY_STOP;
+	/* We always allow suspend even if suspend preparation failed and
+	 * attempt to recover in resume.
+	 */
+	if (ret)
+		bt_dev_err(hdev, "Suspend notifier action (%lu) failed: %d",
+			   action, ret);
+
+	return NOTIFY_STOP;
 }
 
 /* Alloc HCI device */
-- 
2.27.0.278.ge193c7cf3a9-goog