linux-kernel - Re: [PATCH 2/2] usb: dwc3: gadget: Preserve UDC max speed setting

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-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 28 Oct 2020 19:00:59 -0700
From:   Wesley Cheng <wcheng@...eaurora.org>
To:     Thinh Nguyen <Thinh.Nguyen@...opsys.com>,
        "balbi@...nel.org" <balbi@...nel.org>,
        "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>
Cc:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>,
        "jackp@...eaurora.org" <jackp@...eaurora.org>
Subject: Re: [PATCH 2/2] usb: dwc3: gadget: Preserve UDC max speed setting

On 10/28/2020 5:43 PM, Thinh Nguyen wrote:
> Hi,
> 
> Wesley Cheng wrote:
>> The USB gadget/UDC driver can restrict the DWC3 controller speed using
>> dwc3_gadget_set_speed().  Store this setting into a variable, in order for
>> this setting to persist across controller resets due to runtime PM.
> 
> Why do we need to do this? DCFG should persist unless we overwrite it.
> The current PM shouldn't update the current speed setting.
> 
> BR,
> Thinh
> 

Hi Thinh,

During runtime PM suspend, the dwc3_suspend_common() will call
dwc3_core_exit().  On some platforms they register the DWC3 reset
control to the DWC3 core driver (otherwise could be handled in the DWC3
glue drivers), which will be asserted here:

static void dwc3_core_exit(struct dwc3 *dwc)
{
...
	reset_control_assert(dwc->reset);

>From the SNPS databook (Table 2-2 Resets for Registers) it mentions that
assertion of the reset signal will reset the DCFG register.

In addition to the above, with the change to allow runtime PM suspend
during UDC unbind, we need a way to avoid writing to the DCFG during the
UDC bind path. (if we entered suspend before re-binding the UDC)  If we
add an early exit based on the PM state (in
dwc3_gadget_set_udc_speed()), then we basically ignore the max speed
request from the UDC/gadget layer.

Since it looks like the DWC3 gadget driver doesn't like using
synchronous PM runtime resumes, by going this route, we can allow the
async runtime resume handler to do everything, such as writing the speed
config and re-enabling the controller.

Thanks

Regards,
Wesley Cheng
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project