From: "Ruslan Nikolaev via gcc" <gcc@gcc.gnu.org>
To: Torvald Riegel <triegel@redhat.com>
Cc: Alexander Monakov <amonakov@ispras.ru>,
Szabolcs Nagy <szabolcs.nagy@arm.com>,
"gcc@gcc.gnu.org" <gcc@gcc.gnu.org>, "nd@arm.com" <nd@arm.com>
Subject: Re: GCC interpretation of C11 atomics (DR 459)
Date: Mon, 26 Feb 2018 18:59:00 -0000 [thread overview]
Message-ID: <1738288554.4867171.1519671329114@mail.yahoo.com> (raw)
In-Reply-To: <1519670138.15077.663.camel@redhat.com>
Torvald, thank you for your output. See my response below.
On Monday, February 26, 2018 1:35 PM, Torvald Riegel <triegel@redhat.com> wrote:
> ... does not imply this latter statement. The statement you cited is
> about what the standard itself requires, not what makes sense for a
> particular implementation.
True but makes sense to provide true atomics when they are available. Since the standard seem to allow atomic_load implementation using RMW, does not seem to be a problem.
In fact, lock_free flag for this type can return true only if mcx16 is specified; otherwise -- it returns false (since it can only be determined during runtime, assuming worst case scenario)
> So, in such a case, using the wide CAS for
> atomic loads breaks a reasonable assumption. Moreover, it's also a
> special case, in that 32b atomics do work as intended.
But in this case a programmer already makes an assumption that atomic_load does not use RMW which C11 does not seem to guarantee.Of course, for single-width operations, the programmer may in most practical cases assume it (even though there is no guarantee).
Anyway, there is no good solution here for double-width operations, and the programmer should not assume it is possible when writing portable code.In fact, lock-based solution is even more confusing and potentially error-prone (e.g., cannot be safely used inside signal handlers since it is not lock-free, etc)
> The behavior you favor would violate that, and
> there's no portable way to distinguish one from the other.
There is already a similar problem with IFFUNC (when used with Linux and glibc). In fact, I do not see any difference here. Redirection to libatomic when mcx16 is specified just adds extra cost + less predictable behavior. Moreover, it seems counterintuitive -- I specify a flag that mcx16 is supported but gcc still does not use it (at least directly). It is possible to make a change to libatomic to always use cmpxchg16b when available (even on systems without IFFUNC), this way it is totally consistent and binary compatible for code compiled with and without mcx16.
> I see your point in wanting to have a builtin or such for the 64b atomic
> CAS. However, IMO, this doesn't fit into the world of C11/C++11
> atomics, and thus rather should be accessible through a separate
> interface.
Why not? If atomic_load is not really an issue, then it may be good to use standardized interface.
next prev parent reply other threads:[~2018-02-26 18:59 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <1615980330.4453149.1519617655582.ref@mail.yahoo.com>
2018-02-26 4:01 ` Ruslan Nikolaev via gcc
2018-02-26 5:50 ` Alexander Monakov
2018-02-26 7:24 ` Fw: " Ruslan Nikolaev via gcc
2018-02-26 8:20 ` Alexander Monakov
2018-02-26 8:43 ` Ruslan Nikolaev via gcc
2018-02-26 19:07 ` Torvald Riegel
2018-02-26 19:43 ` Ruslan Nikolaev via gcc
2018-02-26 22:49 ` Ruslan Nikolaev via gcc
2018-02-27 3:33 ` Ruslan Nikolaev via gcc
2018-02-27 10:34 ` Ramana Radhakrishnan
2018-02-27 11:14 ` Torvald Riegel
2018-02-27 12:39 ` Torvald Riegel
2018-02-27 13:04 ` Ruslan Nikolaev via gcc
2018-02-27 13:08 ` Szabolcs Nagy
2018-02-27 13:17 ` Ruslan Nikolaev via gcc
2018-02-27 16:40 ` Torvald Riegel
2018-02-27 17:07 ` Ruslan Nikolaev via gcc
2018-02-27 16:21 ` Torvald Riegel
2018-02-27 16:16 ` Torvald Riegel
2018-02-27 16:46 ` Simon Wright
2018-02-27 16:52 ` Florian Weimer
2018-02-27 17:30 ` Torvald Riegel
2018-02-27 17:33 ` Ruslan Nikolaev via gcc
2018-02-27 19:32 ` Torvald Riegel
2018-02-27 17:59 ` Simon Wright
2018-02-27 10:40 ` Fw: " Torvald Riegel
2018-02-26 18:56 ` Torvald Riegel
2018-02-26 12:30 ` Szabolcs Nagy
2018-02-26 13:57 ` Alexander Monakov
2018-02-26 14:51 ` Szabolcs Nagy
2018-02-26 14:53 ` Ruslan Nikolaev via gcc
2018-02-26 18:35 ` Torvald Riegel
2018-02-26 18:59 ` Ruslan Nikolaev via gcc [this message]
2018-02-26 19:20 ` Torvald Riegel
2018-02-26 18:16 ` Florian Weimer
2018-02-26 18:34 ` Ruslan Nikolaev via gcc
2018-02-26 18:36 ` Janne Blomqvist
2018-02-27 10:22 ` Florian Weimer
[not found] <886227277.5611063.1519759959364.ref@mail.yahoo.com>
2018-02-27 20:20 ` Ruslan Nikolaev via gcc
2018-02-27 22:19 ` Torvald Riegel
2018-02-28 1:46 ` Ruslan Nikolaev via gcc
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1738288554.4867171.1519671329114@mail.yahoo.com \
--to=gcc@gcc.gnu.org \
--cc=amonakov@ispras.ru \
--cc=nd@arm.com \
--cc=nruslan_devel@yahoo.com \
--cc=szabolcs.nagy@arm.com \
--cc=triegel@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).