linux-kernel - Re: [PATCH 1/7] lib/crc: remove unnecessary prompt for CONFIG

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: <20250402164101.GC1235@sol.localdomain>
Date: Wed, 2 Apr 2025 09:41:01 -0700
From: Eric Biggers <ebiggers@...nel.org>
To: Randy Dunlap <rdunlap@...radead.org>
Cc: linux-kernel@...r.kernel.org, Ard Biesheuvel <ardb@...nel.org>,
	linux-crypto@...r.kernel.org, linux-arch@...r.kernel.org
Subject: Re: [PATCH 1/7] lib/crc: remove unnecessary prompt for CONFIG_CRC32
 and drop 'default y'

On Tue, Apr 01, 2025 at 10:56:32PM -0700, Randy Dunlap wrote:
> 
> 
> On 4/1/25 10:02 PM, Eric Biggers wrote:
> > On Tue, Apr 01, 2025 at 09:50:57PM -0700, Randy Dunlap wrote:
> >>
> >>
> >> On 4/1/25 8:51 PM, Eric Biggers wrote:
> >>> On Tue, Apr 01, 2025 at 08:42:41PM -0700, Randy Dunlap wrote:
> >>>> Hi 
> >>>>
> >>>> On 4/1/25 3:15 PM, Eric Biggers wrote:
> >>>>> From: Eric Biggers <ebiggers@...gle.com>
> >>>>>
> >>>>> All modules that need CONFIG_CRC32 already select it, so there is no
> >>>>> need to bother users about the option, nor to default it to y.
> >>>>>
> >>>>
> >>>> My memory from 10-20 years ago could be foggy, but ISTR that someone made at least
> >>>> CRC16 and CRC32 user-selectable in order to support out-of-tree modules...
> >>>> FWIW.
> >>>> But they would not need to be default y.
> >>>
> >>> That's not supported by upstream, though.
> >>
> >> Which part is not supported by upstream?
> > 
> > Having prompts for library kconfig options solely because out-of-tree modules
> > might need them.
> 
> Well, I think that is was supported for many years. I don't see how it would become
> unsupported all of a sudden. IMHO.

Most kernel-internal options aren't user-selectable, though.  It's mainly just
some older ones that were made user-selectable for some reason, and that is a
mistake that has been getting cleaned up over time.

Consider that the upstream community has no visibility into out-of-tree modules
in general, so there is no reasonable policy that could be applied in deciding
which options should be user-selectable purely for the benefit of out-of-tree
modules.  The only reasonable policy is to consider in-tree users only.  Just
like we don't add EXPORT_SYMBOL() just because an out-of-tree module wants it.

And of course downstreams always can, and do, just add a new kconfig option that
selects any non-visible options they want.

- Eric