linux-kernel - Re: [GIT PULL] nfsd changes for v6.7 (early)

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: <96800661-0F30-4F9E-89E4-C0B032EFDEB9@oracle.com>
Date:   Mon, 30 Oct 2023 23:21:06 +0000
From:   Chuck Lever III <chuck.lever@...cle.com>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
CC:     Neil Brown <neilb@...e.de>,
        Linux NFS Mailing List <linux-nfs@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Jeff Layton <jlayton@...nel.org>
Subject: Re: [GIT PULL] nfsd changes for v6.7 (early)



> On Oct 30, 2023, at 2:31 PM, Linus Torvalds <torvalds@...ux-foundation.org> wrote:
> 
> On Wed, 25 Oct 2023 at 04:24, Chuck Lever III <chuck.lever@...cle.com> wrote:
>> 
>> This release completes the SunRPC thread scheduler work that was
>> begun in v6.6. The scheduler can now find an svc thread to wake in
>> constant time and without a list walk. Thanks again to Neil Brown
>> for this overhaul.
> 
> Btw, the "help" text for the new Kconfig option that this introduces
> is just ridiculously bad.
> 
> I react to these things, because I keep telling people that our
> Kconfig is one of the nastier parts to people just building and
> testing their own kernels. Yes, you can start with whatever distro
> default config, and build your own, and install it, but when people
> then introduce new options and ask insane and unhelpful questions,
> that scares off any sane person.
> 
> So Kconfig questions really need to make sense, and they need to have
> help messages that are useful..
> 
> Honestly, that LWQ_TEST option probably fails both cases.  The
> "testing" is a toy, and the Kconfig option is horrific. I literally
> think that we would be better off removing that code. Any bug found by
> that testv would be so fundamental as to not be worth testing for.

I have to admit I didn't look too closely at that part
of the series, except to note that there's no maintainer
of record for those files. That's probably why there was
little initial pushback on the scant help text.

Do you need a refreshed PR with the testing bit removed,
or can you live with Neil or me sending a subsequent
fix-up later in the merge window?


--
Chuck Lever