Received-SPF: pass (google.com: domain of linux-kernel+bounces-144626-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
From: =?utf-8?Q?Bj=C3=B8rn_Mork?= <bjorn@mork.no>
To: Aleksander Morgado <aleksandermj@chromium.org>
Cc: oneukum@suse.com, linux-usb@vger.kernel.org, gregkh@linuxfoundation.org,
        linux@roeck-us.net, linux-kernel@vger.kernel.org,
        ejcaruso@chromium.org
Subject: Re: [PATCH] usb: cdc-wdm: close race between read and workqueue
Organization: m
References: <385a3519-b45d-48c5-a6fd-a3fdb6bec92f@chromium.org>
Date: Mon, 15 Apr 2024 08:47:19 +0200
In-Reply-To: <385a3519-b45d-48c5-a6fd-a3fdb6bec92f@chromium.org> (Aleksander
	Morgado's message of "Mon, 15 Apr 2024 05:42:06 +0000")
Message-ID: <87mspvi0lk.fsf@miraculix.mork.no>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/28.2 (gnu/linux)
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

Aleksander Morgado <aleksandermj@chromium.org> writes:
> On 3/14/24 11:50, Oliver Neukum wrote:
>> wdm_read() cannot race with itself. However, in
>> service_outstanding_interrupt() it can race with the
>> workqueue, which can be triggered by error handling.
>> Hence we need to make sure that the WDM_RESPONDING
>> flag is not just only set but tested.
>> Fixes: afba937e540c9 ("USB: CDC WDM driver")
>> Signed-off-by: Oliver Neukum <oneukum@suse.com>
>
> We are not aware of all the details involved in this patch, but we had
> to revert it in all the different ChromeOS kernel versions where we
> had it cherry-picked, because it broke the MBIM communication with the
> Intel XMM based Fibocomm L850 modem. Other modems shipped in
> Chromebooks like the QC based Fibocomm FM101 don't seem to be
> affected.
>
> Attached is an example output of mbimcli talking directly to the
> cdc-wdm port (i.e. without ModemManager or the mbim-proxy). In the
> example, we are receiving a bunch of different messages from previous
> mbimcli runs. Looking at the timestamps, it looks as if we only
> receive a message right after we have sent one, e.g. after each "open
> request" we end up receiving responses for requests sent in earlier
> runs; or something along those lines.
>
> Is this bad behavior of this specific modem chipset, and if so, how
> can we workaround it? If you need any additional information or help
> to test new patches, let us know.

I'm not sure I understand what problem that patch is supposed to fix.
Which means that everything I write could be completely wrong...

But to me it looks like the described issue is exactly what you should
expect if that change ever triggers.  I believe we must resubmit the
urb from service_outstanding_interrupt(). That's why it was added. See
the explanation Robert wrote when introducing it:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/d=
rivers/usb/class/cdc-wdm.c?id=3Dc1da59dad0ebd3f9bd238f3fff82b1f7ffda7829

As for the XMM behaviour: it's been a long time since I tried any of
those, but AFAIR one the major differences compared to Qualcomm was the
strict queue handling in the firmware.  This caused a number of problems
where the cdc-wdm driver wanted to skip a message for some reason.  So
I'm not surprised that a bug like this is triggered by one of those
modems. That's probably the only thing they are good for :-)


Bj=C3=B8rn