From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-3.8 required=3.0 tests=DKIMWL_WL_HIGH,DKIM_SIGNED, DKIM_VALID,HEADER_FROM_DIFFERENT_DOMAINS,MAILING_LIST_MULTI,SIGNED_OFF_BY, SPF_HELO_NONE,SPF_PASS autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id BAD5DC35280 for ; Fri, 8 May 2020 01:36:16 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id 74510208DB for ; Fri, 8 May 2020 01:36:16 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=kernel.org header.i=@kernel.org header.b="DMSZlX59" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 74510208DB Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=linux-foundation.org Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id 084A4900003; Thu, 7 May 2020 21:36:16 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 035A5900002; Thu, 7 May 2020 21:36:15 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id E66C0900003; Thu, 7 May 2020 21:36:15 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0048.hostedemail.com [216.40.44.48]) by kanga.kvack.org (Postfix) with ESMTP id CF192900002 for ; Thu, 7 May 2020 21:36:15 -0400 (EDT) Received: from smtpin20.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay05.hostedemail.com (Postfix) with ESMTP id 89D08181AEF0B for ; Fri, 8 May 2020 01:36:15 +0000 (UTC) X-FDA: 76791836310.20.gold75_70cf558f8f45a X-HE-Tag: gold75_70cf558f8f45a X-Filterd-Recvd-Size: 6836 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by imf40.hostedemail.com (Postfix) with ESMTP for ; Fri, 8 May 2020 01:36:15 +0000 (UTC) Received: from localhost.localdomain (c-73-231-172-41.hsd1.ca.comcast.net [73.231.172.41]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPSA id 09002208E4; Fri, 8 May 2020 01:36:14 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1588901774; bh=r8XwdHJT7KTjVE6ICsgjUujuod2bK+wHnLIaHqps33k=; h=Date:From:To:Subject:In-Reply-To:From; b=DMSZlX598Khy2XymBS1QXSBfIqfGimmLLRRZ3gpHfbuDTlyYkDXEsvR0jc6rGC4VP RlnyDPHhO+7jKPpNTkJqke1PKwpNeJtfrcrVC6eB64ksA9De8xH6EvtEoYdbuttQ9k TjNBLaB8BuCje/1JQ7rmtnbmp3DSUoDqD81RwP5Q= Date: Thu, 07 May 2020 18:36:13 -0700 From: Andrew Morton To: akpm@linux-foundation.org, jbaron@akamai.com, khazhy@google.com, linux-mm@kvack.org, mm-commits@vger.kernel.org, r@hev.cc, rpenyaev@suse.de, torvalds@linux-foundation.org, viro@zeniv.linux.org.uk Subject: [patch 11/15] kselftests: introduce new epoll60 testcase for catching lost wakeups Message-ID: <20200508013613.QGw54D6Rf%akpm@linux-foundation.org> In-Reply-To: <20200507183509.c5ef146c5aaeb118a25a39a8@linux-foundation.org> User-Agent: s-nail v14.8.16 X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: From: Roman Penyaev Subject: kselftests: introduce new epoll60 testcase for catching lost wakeups This test case catches lost wake up introduced by: 339ddb53d373 ("fs/epoll: remove unnecessary wakeups of nested epoll") The test is simple: we have 10 threads and 10 event fds. Each thread can harvest only 1 event. 1 producer fires all 10 events at once and waits that all 10 events will be observed by 10 threads. In case of lost wakeup epoll_wait() will timeout and 0 will be returned. Test case catches two sort of problems: forgotten wakeup on event, which hits the ->ovflist list, this problem was fixed by: 5a2513239750 ("eventpoll: fix missing wakeup for ovflist in ep_poll_callback") the other problem is when several sequential events hit the same waiting thread, thus other waiters get no wakeups. Problem is fixed in the following patch. Link: http://lkml.kernel.org/r/20200430130326.1368509-1-rpenyaev@suse.de Signed-off-by: Roman Penyaev Cc: Khazhismel Kumykov Cc: Alexander Viro Cc: Heiher Cc: Jason Baron Signed-off-by: Andrew Morton --- tools/testing/selftests/filesystems/epoll/epoll_wakeup_test.c | 146 ++++++++++ 1 file changed, 146 insertions(+) --- a/tools/testing/selftests/filesystems/epoll/epoll_wakeup_test.c~kselftests-introduce-new-epoll60-testcase-for-catching-lost-wakeups +++ a/tools/testing/selftests/filesystems/epoll/epoll_wakeup_test.c @@ -3,6 +3,7 @@ #define _GNU_SOURCE #include #include +#include #include #include #include @@ -3136,4 +3137,149 @@ TEST(epoll59) close(ctx.sfd[0]); } +enum { + EPOLL60_EVENTS_NR = 10, +}; + +struct epoll60_ctx { + volatile int stopped; + int ready; + int waiters; + int epfd; + int evfd[EPOLL60_EVENTS_NR]; +}; + +static void *epoll60_wait_thread(void *ctx_) +{ + struct epoll60_ctx *ctx = ctx_; + struct epoll_event e; + sigset_t sigmask; + uint64_t v; + int ret; + + /* Block SIGUSR1 */ + sigemptyset(&sigmask); + sigaddset(&sigmask, SIGUSR1); + sigprocmask(SIG_SETMASK, &sigmask, NULL); + + /* Prepare empty mask for epoll_pwait() */ + sigemptyset(&sigmask); + + while (!ctx->stopped) { + /* Mark we are ready */ + __atomic_fetch_add(&ctx->ready, 1, __ATOMIC_ACQUIRE); + + /* Start when all are ready */ + while (__atomic_load_n(&ctx->ready, __ATOMIC_ACQUIRE) && + !ctx->stopped); + + /* Account this waiter */ + __atomic_fetch_add(&ctx->waiters, 1, __ATOMIC_ACQUIRE); + + ret = epoll_pwait(ctx->epfd, &e, 1, 2000, &sigmask); + if (ret != 1) { + /* We expect only signal delivery on stop */ + assert(ret < 0 && errno == EINTR && "Lost wakeup!\n"); + assert(ctx->stopped); + break; + } + + ret = read(e.data.fd, &v, sizeof(v)); + /* Since we are on ET mode, thus each thread gets its own fd. */ + assert(ret == sizeof(v)); + + __atomic_fetch_sub(&ctx->waiters, 1, __ATOMIC_RELEASE); + } + + return NULL; +} + +static inline unsigned long long msecs(void) +{ + struct timespec ts; + unsigned long long msecs; + + clock_gettime(CLOCK_REALTIME, &ts); + msecs = ts.tv_sec * 1000ull; + msecs += ts.tv_nsec / 1000000ull; + + return msecs; +} + +static inline int count_waiters(struct epoll60_ctx *ctx) +{ + return __atomic_load_n(&ctx->waiters, __ATOMIC_ACQUIRE); +} + +TEST(epoll60) +{ + struct epoll60_ctx ctx = { 0 }; + pthread_t waiters[ARRAY_SIZE(ctx.evfd)]; + struct epoll_event e; + int i, n, ret; + + signal(SIGUSR1, signal_handler); + + ctx.epfd = epoll_create1(0); + ASSERT_GE(ctx.epfd, 0); + + /* Create event fds */ + for (i = 0; i < ARRAY_SIZE(ctx.evfd); i++) { + ctx.evfd[i] = eventfd(0, EFD_NONBLOCK); + ASSERT_GE(ctx.evfd[i], 0); + + e.events = EPOLLIN | EPOLLET; + e.data.fd = ctx.evfd[i]; + ASSERT_EQ(epoll_ctl(ctx.epfd, EPOLL_CTL_ADD, ctx.evfd[i], &e), 0); + } + + /* Create waiter threads */ + for (i = 0; i < ARRAY_SIZE(waiters); i++) + ASSERT_EQ(pthread_create(&waiters[i], NULL, + epoll60_wait_thread, &ctx), 0); + + for (i = 0; i < 300; i++) { + uint64_t v = 1, ms; + + /* Wait for all to be ready */ + while (__atomic_load_n(&ctx.ready, __ATOMIC_ACQUIRE) != + ARRAY_SIZE(ctx.evfd)) + ; + + /* Steady, go */ + __atomic_fetch_sub(&ctx.ready, ARRAY_SIZE(ctx.evfd), + __ATOMIC_ACQUIRE); + + /* Wait all have gone to kernel */ + while (count_waiters(&ctx) != ARRAY_SIZE(ctx.evfd)) + ; + + /* 1ms should be enough to schedule away */ + usleep(1000); + + /* Quickly signal all handles at once */ + for (n = 0; n < ARRAY_SIZE(ctx.evfd); n++) { + ret = write(ctx.evfd[n], &v, sizeof(v)); + ASSERT_EQ(ret, sizeof(v)); + } + + /* Busy loop for 1s and wait for all waiters to wake up */ + ms = msecs(); + while (count_waiters(&ctx) && msecs() < ms + 1000) + ; + + ASSERT_EQ(count_waiters(&ctx), 0); + } + ctx.stopped = 1; + /* Stop waiters */ + for (i = 0; i < ARRAY_SIZE(waiters); i++) + ret = pthread_kill(waiters[i], SIGUSR1); + for (i = 0; i < ARRAY_SIZE(waiters); i++) + pthread_join(waiters[i], NULL); + + for (i = 0; i < ARRAY_SIZE(waiters); i++) + close(ctx.evfd[i]); + close(ctx.epfd); +} + TEST_HARNESS_MAIN _