Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [RFC PATCH 2/4] rseq: Allow extending struct rseq
To:     Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
        Florian Weimer <fweimer@redhat.com>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        paulmck <paulmck@linux.ibm.com>,
        Boqun Feng <boqun.feng@gmail.com>,
        "H. Peter Anvin" <hpa@zytor.com>, Paul Turner <pjt@google.com>,
        linux-api <linux-api@vger.kernel.org>,
        Christian Brauner <christian.brauner@ubuntu.com>
References: <20200714030348.6214-1-mathieu.desnoyers@efficios.com>
 <20200714030348.6214-3-mathieu.desnoyers@efficios.com>
 <87mu42bepq.fsf@oldenburg2.str.redhat.com>
 <131549905.11442.1594731035989.JavaMail.zimbra@efficios.com>
 <87a7028d5u.fsf@oldenburg2.str.redhat.com>
 <2452161.11491.1594732791558.JavaMail.zimbra@efficios.com>
From:   Carlos O'Donell <carlos@redhat.com>
Organization: Red Hat
Message-ID: <71f08b3a-56f5-0e0f-53b0-cc680f7e8181@redhat.com>
Date:   Tue, 14 Jul 2020 17:30:49 -0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.7.0
MIME-Version: 1.0
In-Reply-To: <2452161.11491.1594732791558.JavaMail.zimbra@efficios.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 7/14/20 9:19 AM, Mathieu Desnoyers wrote:
> Is there an arch-agnostic way to get the thread pointer from user-space code ? That
> would be needed by all rseq critical section implementations.

Yes, and no. We have void *__builtin_thread_pointer (void), but
few architectures implement the builtin so we'd have to go through
a round of compiler updates and backports. All targets know how to
access the thread pointer because the compiler has to generate
IE-mode accesses to the TLS variables.

I have filed an enhancement request:
Bug 96200 - Implement __builtin_thread_pointer() and 
            __builtin_set_thread_pointer() if TLS is supported
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96200

We have glibc internal macro APIs to access the thread pointer,
but I would rather the compiler handle the access since it can
schedule the resulting sequence better.

On some arches setting the therad pointer needs a syscall or
equivalent operation (hppa), and for some arches there is no
fixed register (arm) hence the need for __builtin_thread_pointer()
to force the compiler to place the pointer into a register for
function return.

-- 
Cheers,
Carlos.