linux-kernel - Re: [PATCH 1/5] seqlock: introduce SEQLOCK_READ

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: <20251005160753.GA6063@redhat.com>
Date: Sun, 5 Oct 2025 18:07:54 +0200
From: Oleg Nesterov <oleg@...hat.com>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Alexander Viro <viro@...iv.linux.org.uk>,
	Boqun Feng <boqun.feng@...il.com>,
	David Howells <dhowells@...hat.com>, Ingo Molnar <mingo@...hat.com>,
	Li RongQing <lirongqing@...du.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Waiman Long <longman@...hat.com>, Will Deacon <will@...nel.org>,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/5] seqlock: introduce SEQLOCK_READ_SECTION()

OK. I'll send v2 but let me ask...

On 10/05, Linus Torvalds wrote:
>
> As far as the users are concerned, the result doesn't end up being
> really any different from our scoped guards, so I'd actually suggest
> you just make this helper look like our scoped_guard() macro does.

I swear, I too thought about scoped_seqlock_xxx ;)

> And instead of making people pass in a NULL 'flags', just do a
> separate version of it, exactly like we do for locking. Even if the
> internal implementation then ends up sharing most of the code, please
> don't make people pass in NULL just because they don't want the
> irqsave version.
>
> So make it two different things:
>
>    scoped_seqlock_read(lock) { .... }
>
>    scoped_seqlock__read_irqsave(lock, flags) { ... }

OK. But if you don't object I'd like to avoid another DEFINE_LOCK_GUARD()
or something like it in this case. To me it won't buy anything.

And I think that the "generic" seqlock_read_section_retry(flags) makes
sense, the "if (flags)" checks should not generate the extra code.

Will you agree?

> (Maybe 'flags' can even be local to that scope?)

The problem is that you can't declare "int lockless/seq" and
"unsigned long flags" inside "for (...)",  but I'll try to think about
it more.

Oleg.