HID: intel-ish-hid: ipc: make ish suspend paths clear
authorZhang Lixu <lixu.zhang@intel.com>
Thu, 8 Aug 2019 10:21:12 +0000 (18:21 +0800)
committerJiri Kosina <jkosina@suse.cz>
Mon, 19 Aug 2019 12:01:50 +0000 (14:01 +0200)
For suspend-to-idle, send suspend message and set N0_D3 flag to put
the ISH into D0i3 state.
For suspend-to-mem, disable the DMA bit before ISH entering D3, and
NO_D3 flag is cleared by default, then the ISH would enter D3.

Signed-off-by: Zhang Lixu <lixu.zhang@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
drivers/hid/intel-ish-hid/ipc/pci-ish.c

index 89b14d2..35081f2 100644 (file)
@@ -223,6 +223,8 @@ static void __maybe_unused ish_resume_handler(struct work_struct *work)
         * it means ISH isn't powered off, in this case, send a resume message.
         */
        if (fwsts >= FWSTS_SENSOR_APP_LOADED) {
+               disable_irq_wake(pdev->irq);
+
                ishtp_send_resume(dev);
 
                /* Waiting to get resume response */
@@ -255,27 +257,36 @@ static int __maybe_unused ish_suspend(struct device *device)
        struct pci_dev *pdev = to_pci_dev(device);
        struct ishtp_device *dev = pci_get_drvdata(pdev);
 
-       enable_irq_wake(pdev->irq);
-       /*
-        * If previous suspend hasn't been asnwered then ISH is likely dead,
-        * don't attempt nested notification
-        */
-       if (dev->suspend_flag)
-               return  0;
-
-       dev->resume_flag = 0;
-       dev->suspend_flag = 1;
-       ishtp_send_suspend(dev);
-
-       /* 25 ms should be enough for live ISH to flush all IPC buf */
-       if (dev->suspend_flag)
-               wait_event_interruptible_timeout(dev->suspend_wait,
-                                                !dev->suspend_flag,
-                                                 msecs_to_jiffies(25));
-
        if (ish_should_enter_d0i3(pdev)) {
-               /* Set the NO_D3 flag, the ISH would enter D0i3 */
-               pdev->dev_flags |= PCI_DEV_FLAGS_NO_D3;
+               /*
+                * If previous suspend hasn't been asnwered then ISH is likely
+                * dead, don't attempt nested notification
+                */
+               if (dev->suspend_flag)
+                       return  0;
+
+               dev->resume_flag = 0;
+               dev->suspend_flag = 1;
+               ishtp_send_suspend(dev);
+
+               /* 25 ms should be enough for live ISH to flush all IPC buf */
+               if (dev->suspend_flag)
+                       wait_event_interruptible_timeout(dev->suspend_wait,
+                                       !dev->suspend_flag,
+                                       msecs_to_jiffies(25));
+
+               if (dev->suspend_flag) {
+                       /*
+                        * It looks like FW halt, clear the DMA bit, and put
+                        * ISH into D3, and FW would reset on resume.
+                        */
+                       ish_disable_dma(dev);
+               } else {
+                       /* Set the NO_D3 flag, the ISH would enter D0i3 */
+                       pdev->dev_flags |= PCI_DEV_FLAGS_NO_D3;
+
+                       enable_irq_wake(pdev->irq);
+               }
        } else {
                /*
                 * Clear the DMA bit before putting ISH into D3,
@@ -304,7 +315,6 @@ static int __maybe_unused ish_resume(struct device *device)
        ish_resume_device = device;
        dev->resume_flag = 1;
 
-       disable_irq_wake(pdev->irq);
        schedule_work(&resume_work);
 
        return 0;