linux-kernel - Re: [PATCH v1] module.h: define __symbol_get_gpl() as a regular __symbol

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: <ZfD9ZdfCYVj9Qh-D@bombadil.infradead.org>
Date: Tue, 12 Mar 2024 18:12:05 -0700
From: Luis Chamberlain <mcgrof@...nel.org>
To: Andrew Kanner <andrew.kanner@...il.com>
Cc: Christoph Hellwig <hch@....de>, linux-modules@...r.kernel.org,
	linux-kernel@...r.kernel.org, mchehab@...nel.org
Subject: Re: [PATCH v1] module.h: define __symbol_get_gpl() as a regular
 __symbol_get()

On Tue, Mar 12, 2024 at 03:25:27PM -0700, Luis Chamberlain wrote:
> On Tue, Feb 13, 2024 at 02:10:45PM +0300, Andrew Kanner wrote:
> > On Thu, Feb 01, 2024 at 10:13:54AM -0800, Luis Chamberlain wrote:
> > > 
> > > While you're at it, if you want to try it, you could see if you can
> > > improve the situation more by looking at symbol_get() users that remain
> > > and seeing if you can instead fix it with proper Kconfig dependency and
> > > at build time. Then we can just remove it as well.
> > > 
> > >   Luis
> > 
> > Sorry for the late reply.
> > 
> > Luis, can you give more details of your idea? I re-read it once, then
> > came back and still don't understand.
> > 
> > I see that there are ~10 users for symbol_get() currently. Do you want
> > to stringify symbol names at build time to completely remove
> > symbol_get() from module.h? Correct me if I'm wrong since using of a
> > fuction which is not declared anywhere sounds confusing.
> 
> As an example look at the code and see if there's a sensible way to make
> some calls built-in instead of part of the module, then the module can
> have a kconfig builtin option, that adds to the built-in code which
> means you don't need the symbol_get().
> 
> For some other pieces of code it may require other strategies.

An example is FW_LOADER_USER_HELPER which is bool only, and is selected
by users. It didn't use symbol_get() before, however its an example of
how through Kconfig you can align requirements and define built-in
components, even if they do come from a module.

  Luis