From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mathieu Desnoyers Subject: Re: [RFC PATCH v4 1/5] glibc: Perform rseq(2) registration at nptl init and thread creation Date: Thu, 22 Nov 2018 11:47:42 -0500 (EST) Message-ID: <1045257294.10291.1542905262086.JavaMail.zimbra@efficios.com> References: <20181121183936.8176-1-mathieu.desnoyers@efficios.com> <20181122143603.GD23599@brightrain.aerifal.cx> <782067422.9852.1542899056778.JavaMail.zimbra@efficios.com> <20181122151444.GE23599@brightrain.aerifal.cx> <686626451.10113.1542901620250.JavaMail.zimbra@efficios.com> <87wop5xeit.fsf@oldenburg.str.redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <87wop5xeit.fsf@oldenburg.str.redhat.com> Sender: linux-kernel-owner@vger.kernel.org To: Florian Weimer Cc: Rich Felker , carlos , Joseph Myers , Szabolcs Nagy , libc-alpha , Thomas Gleixner , Ben Maurer , Peter Zijlstra , "Paul E. McKenney" , Boqun Feng , Will Deacon , Dave Watson , Paul Turner , linux-kernel , linux-api List-Id: linux-api@vger.kernel.org ----- On Nov 22, 2018, at 11:28 AM, Florian Weimer fweimer@redhat.com wrote: > * Mathieu Desnoyers: > >> Here is one scenario: we have 2 early adopter libraries using rseq which >> are deployed in an environment with an older glibc (which does not >> support rseq). >> >> Of course, none of those libraries can be dlclose'd unless they somehow >> track all registered threads. > > Well, you can always make them NODELETE so that dlclose is not an issue. > If the library is small enough, that shouldn't be a problem. That's indeed what I do with lttng-ust, mainly due to use of pthread_key. > >> But let's focus on how exactly those libraries can handle lazily >> registering rseq. They can use pthread_key, and pthread_setspecific on >> first use by the thread to setup a destructor function to be invoked >> at thread exit. But each early adopter library is unaware of the >> other, so if we just use a "is_initialized" flag, the first destructor >> to run will unregister rseq while the second library may still be >> using it. > > I don't think you need unregistering if the memory is initial-exec TLS > memory. Initial-exec TLS memory is tied directly to the TCB and cannot > be freed while the thread is running, so it should be safe to put the > rseq area there even if glibc knows nothing about it. Is it true for user-supplied stacks as well ? > Then you'll only > need a mechanism to find the address of the actually active rseq area > (which you probably have to store in a TLS variable for performance > reasons). And that part you need whether you have reference counter or > not. I'm not sure I follow your thoughts here. Currently, the __rseq_abi TLS symbol identifies a structure registered to the kernel. The "currently active" rseq critical section is identified by the field "rseq_cs" within the __rseq_abi structure. So here when you say "actually active rseq area", do you mean the currently registered struct rseq (__rseq_abi) or the currently running rseq critical section ? (pointed to by __rseq_abi.rseq_cs) One issue here is that early adopter libraries cannot always use the IE model. I tried using it for other TLS variables in lttng-ust, and it ended up hanging our CI tests when tracing a sample application with lttng-ust under a Java virtual machine: being dlopen'd in a process that possibly already exhausts the number of available backup TLS IE entries seems to have odd effects. This is why I'm worried about using the IE model within lttng-ust. So using the IE model for glibc makes sense, because nobody dlopen glibc AFAIK. But it's not so simple for early adopter libraries which can be dlopen'd. > >> The same problem arises if we have an application early adopter which >> explicitly deal with rseq, with a library early adopter. The issue is >> similar, except that the application will explicitly want to unregister >> rseq before exiting the thread, which leaves a race window where rseq >> is unregistered, but the library may still need to use it. >> >> The reference counter solves this: only the last rseq user for a thread >> performs unregistration. > > If you do explicit unregistration, you will run into issues related to > destructor ordering. You should really find a way to avoid that. The per-thread reference counter is a way to avoid issues that arise from lack of destructor ordering. Is it an acceptable approach for you, or you have something else in mind ? Thanks, Mathieu -- Mathieu Desnoyers EfficiOS Inc. http://www.efficios.com