linux-kernel - Re: [PATCH] dwc3: make PM functions as __maybe

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: <12901342.P3on6xiMtS@wuerfel>
Date:   Wed, 16 Nov 2016 17:08:45 +0100
From:   Arnd Bergmann <arnd@...db.de>
To:     Felipe Balbi <balbi@...nel.org>
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Heikki Krogerus <heikki.krogerus@...ux.intel.com>,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dwc3: make PM functions as __maybe_unused

On Wednesday, November 16, 2016 1:13:48 PM CET Felipe Balbi wrote:
> Arnd Bergmann <arnd@...db.de> writes:
> > A change to the suspend/resume handling in dwc3-pci introduced a
> > harmless warning:
> >
> > drivers/usb/dwc3/dwc3-pci.c:169:12: error: ‘dwc3_pci_dsm’ defined but not used [-Werror=unused-function]
> >
> > Replacing the #ifdef around the PM functions with __maybe_unused
> > annotations is the easiest way to make sure this doesn't happen
> > again. A similar problem happened two months earlier and we
> > ended up updating the #ifdef, but as it has come back now,
> > I'd suggest going back to my earlier approach.
> >
> > Fixes: 9cecca75b5a0 ("usb: dwc3: pci: call _DSM for suspend/resume")
> > Link: https://patchwork.kernel.org/patch/9318887/
> > Signed-off-by: Arnd Bergmann <arnd@...db.de>
> 
> I'll just move the ifdef around. We really need a real fix for this. Why
> couldn't we just always add PM callbacks and assume they won't be used
> if !PM && !PM_SLEEP?

I fully agree. This is now the most common warning that gets added
to linux-next, and there are sometimes several new ones on one day, so
I'm playing whack-a-mole here.

I have a rough plan for a proper solution already, but it's going to
be hard to do since there are so many drivers using the current
method. See https://www.mail-archive.com/netdev@vger.kernel.org/msg136759.html
for what I suggested last week.

> Adding __maybe_unused everywhere is rather unelegant 

I agree, we shouldn't need the #ifdef or the __maybe_unused. I always
use the latter since it's easier to get right. In this particular case,
if you had applied the original patch, it the warning would not have
come back now.

	Arnd