netdev - Re: [PATCH 2/2 v2] sierra_net: fix issues with SYNC/RESTART messages and interrupt pipe setup

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]

Message-ID: <87a9r8cub8.fsf@nemi.mork.no>
Date:	Wed, 13 Feb 2013 12:44:11 +0100
From:	Bjørn Mork <bjorn@...k.no>
To:	Dan Williams <dcbw@...hat.com>
Cc:	Oliver Neukum <oliver@...kum.org>,
	Elina Pasheva <epasheva@...rrawireless.com>,
	netdev@...r.kernel.org, linux-usb@...r.kernel.org,
	Rory Filer <rfiler@...rrawireless.com>,
	Phil Sutter <phil@....cc>
Subject: Re: [PATCH 2/2 v2] sierra_net: fix issues with SYNC/RESTART messages and interrupt pipe setup

Dan Williams <dcbw@...hat.com> writes:

> It doesn't need to run exactly at probe, but it appears to need to be
> the first thing the driver does when communicating with the firmware to
> ensure clear state and whatnot.  Possibly like the QMI SYNC message that
> clears all the client IDs and resets the internal stack.  (the driver
> also sends a "shutdown" message to the firmware when unbinding).
>
> So I do think that somewhere around probe() is the best time to do this,
> because it's best to initialize the device when the driver binds to it
> and react to errors as soon as possible, rather than trying to set
> everything up on open/IFF_UP and then fail right before you want to
> actually use the device.  Late-fail is quite unhelpful for applications.
>
> I don't really care if it happens in probe() or somewhere else right
> after the driver is bound to the device, but it should be part of the
> initialization process.

I was looking for something else in the rndis_host driver right now and
noticed that it has the same sort of problem.  The generic_rndis_bind()
function will call rndis_command() which does

         usb_control_msg(.., USB_CDC_SEND_ENCAPSULATED_COMMAND, ..);
         usb_interrupt_msg(.., dev->status->desc.bEndpointAddress, .. );
         for (count = 0; count < 10; count++) {
             usb_control_msg(.., USB_CDC_GET_ENCAPSULATED_RESPONSE, ..);
             if (ok)
                return 0;
             msleep(20);
         }

Somewhat ugly, but it is a way to send and receive commands while
probing.  This driver also sends a command in unbind().

Looks like your patch would allow this to be solved a lot cleaner,
replacing the loop over USB_CDC_GET_ENCAPSULATED_RESPONSE with proper
interrupt endpoint handling.


Bjørn
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html