linux-kernel - Re: [PATCH] usbip: vhci_hcd: do proper error handling

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] [day] [month] [year] [list]

Message-ID: <0ff0a28a-e369-91c8-81f9-c6e2cbe4bc26@linuxfoundation.org>
Date:   Thu, 8 Apr 2021 09:44:45 -0600
From:   Shuah Khan <skhan@...uxfoundation.org>
To:     Muhammad Usama Anjum <musamaanjum@...il.com>, shuah@...nel.org,
        gregkh@...uxfoundation.org, linux-kernel@...r.kernel.org,
        linux-usb@...r.kernel.org, valentina.manea.m@...il.com,
        stern@...land.harvard.edu, Shuah Khan <skhan@...uxfoundation.org>
Subject: Re: [PATCH] usbip: vhci_hcd: do proper error handling

On 3/31/21 5:23 AM, Muhammad Usama Anjum wrote:
> On Fri, 2021-03-26 at 14:24 -0600, Shuah Khan wrote:
>> On 3/25/21 5:46 AM, Muhammad Usama Anjum wrote:
>>> The driver was assuming that all the parameters would be valid. But it
>>> is possible that parameters are sent from userspace. For those cases,
>>> appropriate error checks have been added.
>>>
>>
>> Are you sure this change is necessary for vhci_hcd? Is there a
>> scenario where vhci_hcd will receive these values from userspace?
> I'm not sure if these changes are necessary for vhci_hcd. I'd sent
> this patch following the motivation of a patch (c318840fb2) from
> dummy_hcd to handle some cases. Yeah, there is scenario where vhci_hcd
> will receive these values from userspace. For example, typReq =
> SetPortFeature and wValue = USB_PORT_FEAT_C_CONNECTION can be received
> from userspace. As USB_PORT_FEAT_C_CONNECTION case isn't being
> handled, default case will is executed for it. So I'm assuming
> USB_PORT_FEAT_C_CONNECTION isn't supported and default case shouldn't
> be executed.
> 

The way dummy_hcd handles USB_PORT_FEAT_C_CONNECTION isn't applicable
to vhci_hcd.

rh_port_connect() and  rh_port_disconnect() set the 
USB_PORT_FEAT_C_CONNECTION this flag to initiate port status polling.
This flag is set by the driver as a result of attach/deatch request
from the user-space. Current handling of this flag is correct.

>> Is there a way to reproduce the problem?
> I'm not able to reproduce any problem. But typReq = SetPortFeature and
> wValue = USB_PORT_FEAT_C_CONNECTION may trigger some behavior which
> isn't intended as USB_PORT_FEAT_C_CONNECTION may not be supported for
> vhci_hcd.
> 

If you reproduce the problem and it impacts attach/detach and device
remote device access, we can to look into this further.

thanks,
-- Shuah