linux-kernel - Re: [2.6 patch] always enable FW

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: <20080611082320.GM11685@cs181133002.pp.htv.fi>
Date:	Wed, 11 Jun 2008 11:23:20 +0300
From:	Adrian Bunk <bunk@...nel.org>
To:	Greg KH <greg@...ah.com>
Cc:	linux-kernel@...r.kernel.org,
	David Woodhouse <dwmw2@...radead.org>,
	James Bottomley <James.Bottomley@...senPartnership.com>,
	Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [2.6 patch] always enable FW_LOADER unless EMBEDDED=y

On Tue, Jun 10, 2008 at 02:28:59PM -0700, Greg KH wrote:
> On Tue, Jun 10, 2008 at 09:12:06PM +0300, Adrian Bunk wrote:
> > On Tue, Jun 10, 2008 at 09:24:50AM -0700, Greg KH wrote:
> > > On Tue, Jun 10, 2008 at 07:04:08PM +0300, Adrian Bunk wrote:
> > > > James Bottomley recently discovered that we have 
> > > > {request,release}_firmware() dummies for the case of the actual 
> > > > functions not being available and has a fix for the bug that was 
> > > > actually causing build errors for built-in users with 
> > > > CONFIG_FW_LOADER=m.
> > > > 
> > > > But now missing selects on FW_LOADER are no longer visible at 
> > > > compile-time at all and can become runtime problems.
> > > > 
> > > > FW_LOADER is infrastructure with relatively small codesize we can 
> > > > safely enable for everyone, and only for people who really need small 
> > > > kernels (and can be expected to know what they are doing) it matters 
> > > > being able to disable it.
> > > > 
> > > > This patch therefore always sets FW_LOADER=y and allows users only to 
> > > > disable it with EMBEDDED=y.
> > > > 
> > > > As a bonus, we can then get rid of all "select FW_LOADER" plus the due 
> > > > to it required "depends on HOTPLUG" which removes some complexity from 
> > > > our Kconfig files.
> > > 
> > > Well, we can't get rid of that if EMBEDDED is set, right?
> > 
> > No, if EMBEDDED is set and HOTPLUG is not set you will not be able to
> > enable FW_LOADER (the "depends on HOTPLUG" for FW_LOADER has to stay, 
> > but all the options that currently select FW_LOADER no longer need the 
> > dependency).
> 
> Ok, but what about the point that the options that are wanting FW_LOADER
> in that situation?  I know EMBEDDED is tough to get right as you can
> shoot yourself in the foot very easily, but this seems like we are going
> to make it even harder to use properly.

Currently missing selects can result in a build error.

After James' patch to fix the dummy functions (that should have worked 
from day 1) it will instead become a runtime problem.

And when the select gets added the dependency on HOTPLUG often gets 
forgotten.

If you worry about "EMBEDDED is tough to get right" a bigger problem 
would be the linux-tiny patches to add new config options - each might 
save a few kB when disabled but also offers a new way to shoot yourself 
into the foot.

I do care more about:
- that users for the common CONFIG_EMBEDDED=n case don't run into 
  unexpected problems (due to a missing select on FW_LOADER) and
- that we remove complexity from our kconfig dependencies so that
  they break less often.

I'm counting 6 commits that fix bugs in this area introduced since
2.6.25 (sic) under drivers/media/ alone.

And Davids firmware patches offer even more opportunities for people to 
forget the dependency on HOTPLUG (which will break only in the 
CONFIG_EMBEDDED=y case...). [1]

> > > You sent this as an RFC before, I thought people said to just fix up all
> > > of the dependancies with drivers that needed FW_LOADER to be enabled,
> > > that would be easier, right?
> > 
> > As far as I know there were zero answers to my RFC.
> > 
> > What are you referring to?
> 
> Sorry, I thought that was sent to the RFC, I think it was to a thread
> even before that.
>...

If it was in the "build issue #503 ..." thread it was even me who
said it - before I had the idea of hiding FW_LOADER behind EMBEDDED.

> thanks,
> 
> greg k-h

cu
Adrian

[1] and before anyone restarts the "kconfig is broken - select should 
    follow dependencies" FUD: I've seen much talk and zero code for
    this, and in my opinion the problem is much more difficult than
    it appears at first sight

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/