linux-kernel - Re: [PATCH v2] make buffer_locked provide an acquire semantics

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: <20220801032045.GZ2860372@paulmck-ThinkPad-P17-Gen-1>
Date:   Sun, 31 Jul 2022 20:20:45 -0700
From:   "Paul E. McKenney" <paulmck@...nel.org>
To:     Matthew Wilcox <willy@...radead.org>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Mikulas Patocka <mpatocka@...hat.com>,
        Will Deacon <will@...nel.org>,
        Ard Biesheuvel <ardb@...nel.org>,
        Alexander Viro <viro@...iv.linux.org.uk>,
        Alan Stern <stern@...land.harvard.edu>,
        Andrea Parri <parri.andrea@...il.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Boqun Feng <boqun.feng@...il.com>,
        Nicholas Piggin <npiggin@...il.com>,
        David Howells <dhowells@...hat.com>,
        Jade Alglave <j.alglave@....ac.uk>,
        Luc Maranget <luc.maranget@...ia.fr>,
        Akira Yokosawa <akiyks@...il.com>,
        Daniel Lustig <dlustig@...dia.com>,
        Joel Fernandes <joel@...lfernandes.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-arch <linux-arch@...r.kernel.org>,
        linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH v2] make buffer_locked provide an acquire semantics

On Sun, Jul 31, 2022 at 11:48:32PM +0100, Matthew Wilcox wrote:
> On Sun, Jul 31, 2022 at 10:30:11AM -0700, Paul E. McKenney wrote:
> > That said, I confess that I am having a hard time finding the
> > buffer_locked() definition.  So if buffer_locked() uses normal C-language
> > accesses to sample the BH_Lock bit of the ->b_state field, then yes,
> > there could be a problem.  The compiler would then be free to reorder
> > calls to buffer_locked() because it could then assume that no other
> > thread was touching that ->b_state field.
> 
> You're having a hard time finding it because it's constructed with the C
> preprocessor.  I really wish we generated header files using CPP once
> and then included the generated/ file.  That would make them greppable.
> 
> You're looking for include/linux/buffer_head.h and it's done like this:
> 
> enum bh_state_bits {
> ...
>         BH_Lock,        /* Is locked */
> ...
> 
> #define BUFFER_FNS(bit, name)                                           \
> ...
> static __always_inline int buffer_##name(const struct buffer_head *bh)  \
> {                                                                       \
>         return test_bit(BH_##bit, &(bh)->b_state);                      \
> }
> 
> BUFFER_FNS(Lock, locked)
> 
> (fwiw, the page/folio versions of these functions don't autogenerate
> the lock or uptodate ones because they need extra functions called)

Thank you!

Another thing that would have helped me find it would have been to leave
the "BH_" prefix on the bit name in the BUFFER_FNS() invocation, as in
ditch the "BH_##" in the definition and then just say "BUFFER_FNS(BH_Lock,
locked)".

But what is life without a challenge?  ;-/

Anyway, on x86 this will use the constant_test_bit() function, which
uses a volatile declaration for its parameter.  So it should avoid
vulnerability to the vicissitudes of the compiler.

So I feel much better now.

							Thanx, Paul