linux-kernel - Re: [GIT pull] locking/urgent for v5.15-rc2

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: <YUhwHsx8n0KJWgQP@hirez.programming.kicks-ass.net>
Date:   Mon, 20 Sep 2021 13:27:26 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        the arch/x86 maintainers <x86@...nel.org>
Subject: Re: [GIT pull] locking/urgent for v5.15-rc2

On Sun, Sep 19, 2021 at 01:10:45PM -0700, Linus Torvalds wrote:
> On Sun, Sep 19, 2021 at 11:28 AM Thomas Gleixner <tglx@...utronix.de> wrote:
> >
> >
> > -       atomic_add(READER_BIAS - bias, &rwb->readers);
> > +       /*
> > +        * _release() is needed in case that reader is in fast path, pairing
> > +        * with atomic_try_cmpxchg() in rwbase_read_trylock(), provides RELEASE
> > +        */
> > +       (void)atomic_add_return_release(READER_BIAS - bias, &rwb->readers);
> 
> Ugh. This really needs fixing.
> 
> atomic_add() is already much more than release-ordered on x86, and
> atomic_add_return_release() is much more expensive on some uarchs.

Is that XADD really more expensive? I don't think I've ever seen that.
Anyway, sure, we can go create atomic_{add,sub}_release() and replace
the few occurences we have of this (a quick grep shows this is the 3rd
such).