From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753441AbdJSRMv (ORCPT ); Thu, 19 Oct 2017 13:12:51 -0400 Received: from albireo.enyo.de ([5.158.152.32]:39780 "EHLO albireo.enyo.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752007AbdJSRMs (ORCPT ); Thu, 19 Oct 2017 13:12:48 -0400 X-Greylist: delayed 464 seconds by postgrey-1.27 at vger.kernel.org; Thu, 19 Oct 2017 13:12:47 EDT From: Florian Weimer To: Mathieu Desnoyers Cc: Ben Maurer , "Carlos O'Donell" , "Paul E. McKenney" , Boqun Feng , Peter Zijlstra , Paul Turner , Andrew Hunter , Andy Lutomirski , Dave Watson , Josh Triplett , Will Deacon , linux-kernel , Thomas Gleixner , Andi Kleen , Chris Lameter , Ingo Molnar , "H. Peter Anvin" , rostedt , Linus Torvalds , Andrew Morton , Russell King , Catalin Marinas , Michael Kerrisk , Alexander Viro , linux-api Subject: Re: [RFC PATCH v9 for 4.15 01/14] Restartable sequences system call References: <20171012230326.19984-1-mathieu.desnoyers@efficios.com> <20171012230326.19984-2-mathieu.desnoyers@efficios.com> <515879378.43966.1508350299712.JavaMail.zimbra@efficios.com> <14399397.44601.1508412942855.JavaMail.zimbra@efficios.com> Date: Thu, 19 Oct 2017 19:01:52 +0200 In-Reply-To: <14399397.44601.1508412942855.JavaMail.zimbra@efficios.com> (Mathieu Desnoyers's message of "Thu, 19 Oct 2017 11:35:42 +0000 (UTC)") Message-ID: <87efpz2hrz.fsf@mid.deneb.enyo.de> MIME-Version: 1.0 Content-Type: text/plain Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Mathieu Desnoyers: > Speaking of optimization, I think the rseq.c helper library > (and eventually glibc) should define the __rseq_abi TLS > variable with __attribute__((tls_model("initial-exec"))). > It provides faster, and signal-safe, accesses to the TLS > variable from libraries. > > The idea you were suggesting where the application could > override the glibc symbol with its own just ends up enforcing > an initial-exec model, but it looks like we can do this > directly from the library. This really depends on how the programming model turns out once multiple libraries are involved. I think it's premature to discuss such details. From mboxrd@z Thu Jan 1 00:00:00 1970 From: Florian Weimer Subject: Re: [RFC PATCH v9 for 4.15 01/14] Restartable sequences system call Date: Thu, 19 Oct 2017 19:01:52 +0200 Message-ID: <87efpz2hrz.fsf@mid.deneb.enyo.de> References: <20171012230326.19984-1-mathieu.desnoyers@efficios.com> <20171012230326.19984-2-mathieu.desnoyers@efficios.com> <515879378.43966.1508350299712.JavaMail.zimbra@efficios.com> <14399397.44601.1508412942855.JavaMail.zimbra@efficios.com> Mime-Version: 1.0 Content-Type: text/plain Return-path: In-Reply-To: <14399397.44601.1508412942855.JavaMail.zimbra@efficios.com> (Mathieu Desnoyers's message of "Thu, 19 Oct 2017 11:35:42 +0000 (UTC)") Sender: linux-kernel-owner@vger.kernel.org To: Mathieu Desnoyers Cc: Ben Maurer , Carlos O'Donell , "Paul E. McKenney" , Boqun Feng , Peter Zijlstra , Paul Turner , Andrew Hunter , Andy Lutomirski , Dave Watson , Josh Triplett , Will Deacon , linux-kernel , Thomas Gleixner , Andi Kleen , Chris Lameter , Ingo Molnar , "H. Peter Anvin" , rostedt , Linus Torvalds , Andrew Morton , Russell King List-Id: linux-api@vger.kernel.org * Mathieu Desnoyers: > Speaking of optimization, I think the rseq.c helper library > (and eventually glibc) should define the __rseq_abi TLS > variable with __attribute__((tls_model("initial-exec"))). > It provides faster, and signal-safe, accesses to the TLS > variable from libraries. > > The idea you were suggesting where the application could > override the glibc symbol with its own just ends up enforcing > an initial-exec model, but it looks like we can do this > directly from the library. This really depends on how the programming model turns out once multiple libraries are involved. I think it's premature to discuss such details.