linux-kernel - [PATCH] drivers:usb:disable usb hub&port async 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-next>] [day] [month] [year] [list]

Date: Wed, 28 Feb 2024 00:56:07 -0800
From: buckzhangwh@...il.com
To: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: linux-usb@...r.kernel.org,
	linux-kernel@...r.kernel.org,
	weihui zhang <buckzhangwh@...il.com>
Subject: [PATCH] drivers:usb:disable usb hub&port async suspend

From: weihui zhang <buckzhangwh@...il.com>

we prepare 30 mobile phones only for usb plugg in and out.
testers plug USB in and out very frequently after phone in deep.
Repeat the test again and again,some phones are crashed.
our analysis：
We analyze ten kernel-dumps,we found something common kernel is blocked.
pasre in crash,all the dump are directed to usb(device port/hub).
here is a kdump,task 446 &365&4511 are UN.
446:
... ->|kobj_attr_store
          |state_store
            |pm_suspend
              |enter_state
                |suspend_devices_and_enter
                 |dpm_resume_end
                   |dpm_resume
                    |dpm_resume
                     |async_synchronize_full
                      |async_synchronize_cookie_domain
                       |schedule
446 is waiting for 365&4511,no doubtful usb thread.
here is some warning log:
[83.958310] musb device disconnect detected from VBUS GPIO.
.........
[84.908017] musb device connection detected from VBUS GPIO.
[84.911946] typec port1-partner: parent port1 should not be sleeping
task 365 & 4511:
.. ->worker_thread
       |process_one_work
        |async_run_entry_fn
          |async_resume
            |device_resume
              |dpm_wait_for_superior
                |wait_for_completion
                  |wait_for_common
                     |schedule_timeout
I guess usb async resume/suspend are disordered,So I try to disable.
After that,we tested the case for a month,the bug never happened again.
the fn device_enable_async_suspend set the dev->power.async_suspend= 1.
dev->power.async_suspend=1&pm_async_enabled=1,fork task like 365
   ---> dpm_resume
           |dpm_async_fn
             |async_resume  
dev->power.async_suspend=0,disable async
     --->dpm_resume
            |device_resume
              |call device resume fn.
here is a demo:
Only few devices such as scsi/pci/usb call device_enable_async_suspend.
but scsi call device_disable_async_suspend at drivers/scsi/hosts.c

Signed-off-by: weihui zhang <buckzhangwh@...il.com>
---
 drivers/usb/core/hub.c  | 2 +-
 drivers/usb/core/port.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
index e38a4124f..de74f70e5 100644
--- a/drivers/usb/core/hub.c
+++ b/drivers/usb/core/hub.c
@@ -2602,7 +2602,7 @@ int usb_new_device(struct usb_device *udev)
 		add_device_randomness(udev->manufacturer,
 				      strlen(udev->manufacturer));
 
-	device_enable_async_suspend(&udev->dev);
+	device_disable_async_suspend(&udev->dev);
 
 	/* check whether the hub or firmware marks this port as non-removable */
 	set_usb_port_removable(udev);
diff --git a/drivers/usb/core/port.c b/drivers/usb/core/port.c
index c628c1abc..97696c415 100644
--- a/drivers/usb/core/port.c
+++ b/drivers/usb/core/port.c
@@ -760,7 +760,7 @@ int usb_hub_create_port_device(struct usb_hub *hub, int port1)
 	pm_runtime_set_active(&port_dev->dev);
 	pm_runtime_get_noresume(&port_dev->dev);
 	pm_runtime_enable(&port_dev->dev);
-	device_enable_async_suspend(&port_dev->dev);
+	device_disable_async_suspend(&port_dev->dev);
 
 	/*
 	 * Keep hidden the ability to enable port-poweroff if the hub
-- 
2.17.1