Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220408032809.3696798-1-npache@redhat.com> <20220408081549.GM2731@worktop.programming.kicks-ass.net>
 <ee07a31c-c514-4a88-599f-14a30e93f32e@redhat.com> <87k0bzk7e5.ffs@tglx>
In-Reply-To: <87k0bzk7e5.ffs@tglx>
From:   Joel Savitz <jsavitz@redhat.com>
Date:   Fri, 8 Apr 2022 12:13:16 -0400
Message-ID: <CAL1p7m4ukBQdQihkeSTjGHTM+HAF-GVf=QSnNPhRmetC213ANg@mail.gmail.com>
Subject: Re: [PATCH v8] oom_kill.c: futex: Don't OOM reap the VMA containing
 the robust_list_head
To:     Thomas Gleixner <tglx@linutronix.de>
Cc:     Nico Pache <npache@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>, linux-mm@kvack.org,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Rafael Aquini <aquini@redhat.com>,
        Waiman Long <longman@redhat.com>, Baoquan He <bhe@redhat.com>,
        Christoph von Recklinghausen <crecklin@redhat.com>,
        Don Dutile <ddutile@redhat.com>,
        "Herton R . Krzesinski" <herton@redhat.com>,
        David Rientjes <rientjes@google.com>,
        Michal Hocko <mhocko@suse.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Davidlohr Bueso <dave@stgolabs.net>,
        Ingo Molnar <mingo@redhat.com>,
        Darren Hart <dvhart@infradead.org>, stable@kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

> ---
> #include <errno.h>
> #include <fcntl.h>
> #include <pthread.h>
> #include <time.h>
> #include <stdio.h>
> #include <string.h>
> #include <unistd.h>
>
> #include <sys/types.h>
> #include <sys/mman.h>
>
> static char n[4096];
>
> int main(void)
> {
>         pthread_mutexattr_t mat_s, mat_p;
>         pthread_mutex_t *mut_s, *mut_p;
>         pthread_barrierattr_t ba;
>         pthread_barrier_t *b;
>         struct timespec to;
>         void *pri, *shr;
>         int r;
>
>         shr = mmap(NULL, sizeof(n), PROT_READ | PROT_WRITE,
>                    MAP_SHARED | MAP_ANONYMOUS, -1, 0);
>
>         pthread_mutexattr_init(&mat_s);
>         pthread_mutexattr_setrobust(&mat_s, PTHREAD_MUTEX_ROBUST);
>         mut_s = shr;
>         pthread_mutex_init(mut_s, &mat_s);
>
>         pthread_barrierattr_init(&ba);
>         pthread_barrierattr_setpshared(&ba, PTHREAD_PROCESS_SHARED);
>         b = shr + 1024;
>         pthread_barrier_init(b, &ba, 2);
>
>         if (!fork()) {
>                 pri = mmap(NULL, 1<<20, PROT_READ | PROT_WRITE,
>                            MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
>                 pthread_mutexattr_init(&mat_p);
>                 pthread_mutexattr_setpshared(&mat_p, PTHREAD_PROCESS_PRIVATE);
>                 pthread_mutexattr_setrobust(&mat_p, PTHREAD_MUTEX_ROBUST);
One thing I don't understand is what kind of sane use case relies on
robust futex for a process-private lock?
Is there a purpose to a lock being on the robust list if there are no
other processes that must be woken in case the holder process is
killed?
If this usage serves no purpose besides causing races during oom, we
should discourage this use, perhaps by adding a note on the manpage.

>                 mut_p = pri;
>                 pthread_mutex_init(mut_p, &mat_p);
>
>                 // With lock order s, p parent gets timeout
>                 // With lock order p, s parent gets owner died
>                 pthread_mutex_lock(mut_s);
>                 pthread_mutex_lock(mut_p);
>                 // Remove unmap and lock order does not matter
>                 munmap(pri, sizeof(n));
>                 pthread_barrier_wait(b);
>                 printf("child gone\n");
>         } else {
>                 pthread_barrier_wait(b);
>                 printf("parent lock\n");
>                 clock_gettime(CLOCK_REALTIME, &to);
>                 to.tv_sec += 1;
>                 r = pthread_mutex_timedlock(mut_s, &to);
>                 printf("parent lock returned: %s\n", strerror(r));
>         }
>         return 0;
> }
>