mei: exclude device from suspend direct complete optimization
authorAlexander Usyskin <alexander.usyskin@intel.com>
Thu, 3 Aug 2017 14:30:19 +0000 (17:30 +0300)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 10 Aug 2017 21:13:18 +0000 (14:13 -0700)
MEI device performs link reset during system suspend sequence.
The link reset cannot be performed while device is in
runtime suspend state. The resume sequence is bypassed with
suspend direct complete optimization,so the optimization should be
disabled for mei devices.

Fixes:
 [  192.940537] Restarting tasks ...
 [  192.940610] PGI is not set
 [  192.940619] ------------[ cut here ]------------ [  192.940623]
 WARNING: CPU: 0
 me.c:653 mei_me_pg_exit_sync+0x351/0x360 [  192.940624] Modules
 linked
 in:
 [  192.940627] CPU: 0 PID: 1661 Comm: kworker/0:3 Not tainted
 4.13.0-rc2+
 #2 [  192.940628] Hardware name: Dell Inc. XPS 13 9343/0TM99H, BIOS
 A11
 12/08/2016 [  192.940630] Workqueue: pm pm_runtime_work <snip> [
 192.940642] Call Trace:
 [  192.940646]  ? pci_pme_active+0x1de/0x1f0 [  192.940649]  ?
 pci_restore_standard_config+0x50/0x50
 [  192.940651]  ? kfree+0x172/0x190
 [  192.940653]  ? kfree+0x172/0x190
 [  192.940655]  ? pci_restore_standard_config+0x50/0x50
 [  192.940663]  mei_me_pm_runtime_resume+0x3f/0xc0
 [  192.940665]  pci_pm_runtime_resume+0x7a/0xa0 [  192.940667]
 __rpm_callback+0xb9/0x1e0 [  192.940668]  ?
 preempt_count_add+0x6d/0xc0 [  192.940670]  rpm_callback+0x24/0x90 [
 192.940672]  ? pci_restore_standard_config+0x50/0x50
 [  192.940674]  rpm_resume+0x4e8/0x800 [  192.940676]
 pm_runtime_work+0x55/0xb0 [  192.940678]
 process_one_work+0x184/0x3e0 [  192.940680]
 worker_thread+0x4d/0x3a0 [ 192.940681]  ?
 preempt_count_sub+0x9b/0x100 [  192.940683]
 kthread+0x122/0x140 [  192.940684]  ? process_one_work+0x3e0/0x3e0 [
 192.940685]  ? __kthread_create_on_node+0x1a0/0x1a0
 [  192.940688]  ret_from_fork+0x27/0x40 [  192.940690] Code: 96 3a
 9e ff 48 8b 7d 98 e8 cd 21 58 00 83 bb bc 01 00 00
 04 0f 85 40 fe ff ff e9 41 fe ff ff 48 c7 c7 5f 04 99 96 e8 93 6b 9f
 ff <0f> ff e9 5d fd ff ff e8 33 fe 99 ff 0f 1f 00 0f 1f 44 00 00 55
 [  192.940719] ---[ end trace
 a86955597774ead8 ]--- [  192.942540] done.

Suggested-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reported-by: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/misc/mei/pci-me.c
drivers/misc/mei/pci-txe.c

index 8621a19..bac3331 100644 (file)
@@ -216,6 +216,12 @@ static int mei_me_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
        pci_set_drvdata(pdev, dev);
 
        /*
+        * MEI requires to resume from runtime suspend mode
+        * in order to perform link reset flow upon system suspend.
+        */
+       pdev->dev_flags |= PCI_DEV_FLAGS_NEEDS_RESUME;
+
+       /*
        * For not wake-able HW runtime pm framework
        * can't be used on pci device level.
        * Use domain runtime pm callbacks instead.
index f811cd5..e38a5f1 100644 (file)
@@ -138,6 +138,12 @@ static int mei_txe_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
        pci_set_drvdata(pdev, dev);
 
        /*
+        * MEI requires to resume from runtime suspend mode
+        * in order to perform link reset flow upon system suspend.
+        */
+       pdev->dev_flags |= PCI_DEV_FLAGS_NEEDS_RESUME;
+
+       /*
        * For not wake-able HW runtime pm framework
        * can't be used on pci device level.
        * Use domain runtime pm callbacks instead.