[BUG] usb: xhci: Two possible sleep-in-atomic-context bugs

[BUG] usb: xhci: Two possible sleep-in-atomic-context bugs

[Jia-Ju Bai]

The driver may sleep with holding a spinlock.
The function call paths (from bottom to top) in Linux-4.16.7 are:

[FUNC] schedule
drivers/base/power/runtime.c, 741: schedule in rpm_resume
drivers/base/power/runtime.c, 1322: rpm_resume in pm_runtime_forbid
drivers/usb/host/xhci-hub.c, 647: pm_runtime_forbid in xhci_enter_test_mode
drivers/usb/host/xhci-hub.c, 632: _raw_spin_lock_irqsave in 
xhci_enter_test_mode

[FUNC] schedule
drivers/base/power/runtime.c, 570: schedule in rpm_suspend
drivers/base/power/runtime.c, 434: rpm_suspend in rpm_idle
drivers/base/power/runtime.c, 1343: rpm_idle in pm_runtime_allow
drivers/usb/host/xhci-hub.c, 670: pm_runtime_allow in xhci_exit_test_mode
drivers/usb/host/xhci-hub.c, 1387: xhci_exit_test_mode in xhci_hub_control
drivers/usb/host/xhci-hub.c, 1038: _raw_spin_lock_irqsave in 
xhci_hub_control

These bugs are found by my static analysis tool (DSAC-2) and checked by my
code review.

I do not know how to correctly fix these bugs, so I just report them.


Best wishes,
Jia-Ju Bai

RE: [BUG] usb: xhci: Two possible sleep-in-atomic-context bugs

[Peter Chen]

 
> 
> The driver may sleep with holding a spinlock.
> The function call paths (from bottom to top) in Linux-4.16.7 are:
> 
> [FUNC] schedule
> drivers/base/power/runtime.c, 741: schedule in rpm_resume
> drivers/base/power/runtime.c, 1322: rpm_resume in pm_runtime_forbid
> drivers/usb/host/xhci-hub.c, 647: pm_runtime_forbid in xhci_enter_test_mode
> drivers/usb/host/xhci-hub.c, 632: _raw_spin_lock_irqsave in xhci_enter_test_mode
> 
> [FUNC] schedule
> drivers/base/power/runtime.c, 570: schedule in rpm_suspend
> drivers/base/power/runtime.c, 434: rpm_suspend in rpm_idle
> drivers/base/power/runtime.c, 1343: rpm_idle in pm_runtime_allow
> drivers/usb/host/xhci-hub.c, 670: pm_runtime_allow in xhci_exit_test_mode
> drivers/usb/host/xhci-hub.c, 1387: xhci_exit_test_mode in xhci_hub_control
> drivers/usb/host/xhci-hub.c, 1038: _raw_spin_lock_irqsave in xhci_hub_control
> 
> These bugs are found by my static analysis tool (DSAC-2) and checked by my code
> review.
> 

I don't think runtime PM API is needed at xhci_hub_control since the controller should
be always active, Mathias, do you agree I delete pm_runtime_forbid and pm_runtime_allow 
to fix this issue?

Peter

Re: [BUG] usb: xhci: Two possible sleep-in-atomic-context bugs

[Mathias Nyman]

On 20.06.2018 06:28, Peter Chen wrote:
>   
>>
>> The driver may sleep with holding a spinlock.
>> The function call paths (from bottom to top) in Linux-4.16.7 are:
>>
>> [FUNC] schedule
>> drivers/base/power/runtime.c, 741: schedule in rpm_resume
>> drivers/base/power/runtime.c, 1322: rpm_resume in pm_runtime_forbid
>> drivers/usb/host/xhci-hub.c, 647: pm_runtime_forbid in xhci_enter_test_mode
>> drivers/usb/host/xhci-hub.c, 632: _raw_spin_lock_irqsave in xhci_enter_test_mode
>>
>> [FUNC] schedule
>> drivers/base/power/runtime.c, 570: schedule in rpm_suspend
>> drivers/base/power/runtime.c, 434: rpm_suspend in rpm_idle
>> drivers/base/power/runtime.c, 1343: rpm_idle in pm_runtime_allow
>> drivers/usb/host/xhci-hub.c, 670: pm_runtime_allow in xhci_exit_test_mode
>> drivers/usb/host/xhci-hub.c, 1387: xhci_exit_test_mode in xhci_hub_control
>> drivers/usb/host/xhci-hub.c, 1038: _raw_spin_lock_irqsave in xhci_hub_control
>>
>> These bugs are found by my static analysis tool (DSAC-2) and checked by my code
>> review.
>>
> 
> I don't think runtime PM API is needed at xhci_hub_control since the controller should
> be always active, Mathias, do you agree I delete pm_runtime_forbid and pm_runtime_allow
> to fix this issue?
> 

The USB2 test mode is a bit special, it will disable all connected devices, and halt the host.
Port should stay in test mode until feature is cleared or host reset.
There is a risk runtime suspend will try to suspend the host as no devices seem preset.

But I think pm_runtime_get_noresume() and pm_runtime_put_noidle() could be used instad,
those should be safe.

-Mathias