Re: Bug caused by 53965c79c2db (USB: Fix device driver race)

From: Alan Stern <stern@rowland.harvard.edu>
To: Bastien Nocera <hadess@hadess.net>
Cc: "M. Vefa Bicakci" <m.v.b@runbox.com>,
	Pany <pany@fedoraproject.org>,
	linux-usb@vger.kernel.org
Subject: Re: Bug caused by 53965c79c2db (USB: Fix device driver race)
Date: Wed, 21 Oct 2020 16:11:27 -0400	[thread overview]
Message-ID: <20201021201127.GA1002979@rowland.harvard.edu> (raw)
In-Reply-To: <5d2dc161951d0717d1ccfc88049c241c8ce8d1e6.camel@hadess.net>

On Wed, Oct 21, 2020 at 03:18:06PM +0200, Bastien Nocera wrote:
> On Wed, 2020-10-21 at 09:08 -0400, M. Vefa Bicakci wrote:
> > On 21/10/2020 07.53, Bastien Nocera wrote:
> > [Snipped by Vefa]
> > > 
> > > I have no idea why there isn't a match function. Patch v1 had a
> > > huge
> > > table:
> > > https://marc.info/?l=linux-usb&m=157062863431186&w=2
> > > and v2 already didn't had that comment, but no .match function:
> > > https://marc.info/?l=linux-usb&m=157114990905421&w=2
> > > 
> > > I'll prepare a patch that adds a match function. I'll let you
> > > (Vefa)
> > > look at which of your patches need backporting though, as I'm
> > > really
> > > quite a bit lost in the different patch sets and branches :/
> > 
> > Hello Bastien,
> > 
> > Having found the root cause of the issue by going through Pany's
> > logs and having proposed a solution, I was hoping to get credit
> > for the resolution of the issue by authoring the patch.
> 
> I don't care either way. Attached are the 2 patches I had made and was
> in the process of compiling and testing, feel free to adapt them,
> change the authorship, etc.
> 
> Note that you mentioned you'd need to "replace the ID table with
> a match function", which will mean that the driver isn't automatically
> loaded when a device gets plugged in. So that wouldn't have worked.
> 
> Let me know when you've made your patch, as I'll need to update this
> bug when there's something to test:
> https://bugzilla.redhat.com/show_bug.cgi?id=1878347
> 
> Cheers

> From 6652af5b46f39d9690d72dc11902b36a44c242a1 Mon Sep 17 00:00:00 2001
> From: Bastien Nocera <hadess@hadess.net>
> Date: Wed, 21 Oct 2020 14:31:58 +0200
> Subject: [PATCH 2/2] USB: apple-mfi-fastcharge: don't probe unhandled devices
> 
> Contrary to the comment above the id table, we didn't implement a match
> function. This meant that every single Apple device that was already
> plugged in to the computer would have its device driver reprobed
> when the apple-mfi-fastcharge driver was loaded, eg. the SD card reader
> could be reprobed when the apple-mfi-fastcharge after pivoting root
> during boot up and the module became available.
> 
> Make sure that the driver probe isn't being run for unsupported
> devices by adding a match function that checks the product ID, in
> addition to the id_table checking the vendor ID.
> 
> Fixes: 249fa8217b84 ("USB: Add driver to control USB fast charge for iOS devices")
> Signed-off-by: Bastien Nocera <hadess@hadess.net>
> ---

Another note: The patch description should include a pointer to Pany's
RedHat Bugzilla bug report.

Alan Stern