Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id C4105C76196 for ; Fri, 7 Apr 2023 23:04:33 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229456AbjDGXEc (ORCPT ); Fri, 7 Apr 2023 19:04:32 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:57122 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229436AbjDGXEb (ORCPT ); Fri, 7 Apr 2023 19:04:31 -0400 Received: from mail-yb1-xb4a.google.com (mail-yb1-xb4a.google.com [IPv6:2607:f8b0:4864:20::b4a]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 22BF7CA22 for ; Fri, 7 Apr 2023 16:04:30 -0700 (PDT) Received: by mail-yb1-xb4a.google.com with SMTP id 202-20020a2502d3000000b00b8d404ac555so2523309ybc.14 for ; Fri, 07 Apr 2023 16:04:30 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20210112; t=1680908669; h=cc:to:from:subject:mime-version:message-id:date:from:to:cc:subject :date:message-id:reply-to; bh=uTKSSYmniB+LEnhXxM03AxYTLMkdjrfWvT6FhNB+Q7g=; b=nN+dsEwkyiIogTmorI5uSQRsi9cy/FJYwLwU11t1d42HAoqT+L/2mZxubGGkmZS1X1 Av+OCpdIUNl0h5XMEKU5Wie0Wn+H4dqiHsrhlPYEBy9Jpf2G9dwxKsGNaa19QIo3QW0s WNUPb+VEX/PFnz75sD/E1MJgpFiUMGAj7cfmaNzITKVwTlv7KocEFMU3vnpYWedtkTjA iSTxh5IaKbAj3Ig1fQoMcxte9DO/9PkN6NYU1FxQ/B0QmeTbQtC1AaE544B5/JAYPTes EjpYd/ooxMLzA9lTG6QljMFqlmmoSEQmHUhQXEq+MbdZvThkRd/OZ1/LgsjiuMiOV5zM OufQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; t=1680908669; h=cc:to:from:subject:mime-version:message-id:date:x-gm-message-state :from:to:cc:subject:date:message-id:reply-to; bh=uTKSSYmniB+LEnhXxM03AxYTLMkdjrfWvT6FhNB+Q7g=; b=qsRweCN0H83cq0e/+scYHj9EyZSUzNOsXSlxXx5GDQM8YS75MfnkUA1EwhtDTY0TTu 4Dqc9SqLZXOd1swRjl9+qsXhBlR0/y8fVmCy89VToAnFAmg8eP86srO/ddY7qgPtaCId Q7bNCS0SCrnF7ddp3xOdX4tKQEmFukHoPuk6qlHWon537XxHqRQXWjdJBJyHIYqlKs9Y HHsSXiP3Q4fJSdaCqb385AN1NfmL/w+THV+F5m55yQ0rEYa2uWfqgy2FHAv8XmnNecMm 2rWVA5BhY40El07BY5vo7+WQN+W9kkxV67CVhrm47GoBkfwMzHVpsVpM0/edWzcoRKOk HB3Q== X-Gm-Message-State: AAQBX9ekmnj1MoygxamcEgMedeLy49hg0ETXr1dYcJEQvjaGg/V+33xV 3UIx3ThD6Ddtuz37bJMwpzneJLefPuQQ X-Google-Smtp-Source: AKy350Z0K/5LChd8MuOvH3Yl5jqduZI2YW0nHSl1ZgLzbixzQua92cYH6Wpgv4HMaak6d8T2CY26ma/oDEqx X-Received: from irogers.svl.corp.google.com ([2620:15c:2d4:203:b240:9cdf:7861:b23e]) (user=irogers job=sendgmr) by 2002:a25:7341:0:b0:acd:7374:f15b with SMTP id o62-20020a257341000000b00acd7374f15bmr2616504ybc.13.1680908669372; Fri, 07 Apr 2023 16:04:29 -0700 (PDT) Date: Fri, 7 Apr 2023 16:04:00 -0700 Message-Id: <20230407230405.2931830-1-irogers@google.com> Mime-Version: 1.0 X-Mailer: git-send-email 2.40.0.577.gac1e443424-goog Subject: [PATCH v7 0/5] Reference count checker and related fixes From: Ian Rogers To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Mark Rutland , Alexander Shishkin , Jiri Olsa , Namhyung Kim , Thomas Gleixner , Darren Hart , Davidlohr Bueso , James Clark , John Garry , Riccardo Mancini , Yury Norov , Andy Shevchenko , Andrew Morton , Adrian Hunter , Leo Yan , Andi Kleen , Thomas Richter , Kan Liang , Madhavan Srinivasan , Shunsuke Nakamura , Song Liu , Masami Hiramatsu , Steven Rostedt , Miaoqian Lin , Stephen Brennan , Kajol Jain , Alexey Bayduraev , German Gomez , linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org, Eric Dumazet , Dmitry Vyukov , Hao Luo Cc: Stephane Eranian , Ian Rogers Content-Type: text/plain; charset="UTF-8" Precedence: bulk List-ID: X-Mailing-List: linux-perf-users@vger.kernel.org The perf tool has a class of memory problems where reference counts are used incorrectly. Memory/address sanitizers and valgrind don't provide useful ways to debug these problems, you see a memory leak where the only pertinent information is the original allocation site. What would be more useful is knowing where a get fails to have a corresponding put, where there are double puts, etc. This work was motivated by the roll-back of: https://lore.kernel.org/linux-perf-users/20211118193714.2293728-1-irogers@google.com/ where fixing a missed put resulted in a use-after-free in a different context. There was a sense in fixing the issue that a game of wac-a-mole had been embarked upon in adding missed gets and puts. The basic approach of the change is to add a level of indirection at the get and put calls. Get allocates a level of indirection that, if no corresponding put is called, becomes a memory leak (and associated stack trace) that leak sanitizer can report. Similarly if two puts are called for the same get, then a double free can be detected by address sanitizer. This can also detect the use after put, which should also yield a segv without a sanitizer. Adding reference count checking to cpu map was done as a proof of concept, it yielded little other than a location where the use of get could be cleaner by using its result. Reference count checking on nsinfo identified a double free of the indirection layer and the related threads, thereby identifying a data race as discussed here: https://lore.kernel.org/linux-perf-users/CAP-5=fWZH20L4kv-BwVtGLwR=Em3AOOT+Q4QGivvQuYn5AsPRg@mail.gmail.com/ Accordingly the dso->lock was extended and use to cover the race. The v3 version addresses problems in v2, in particular using macros to avoid #ifdefs. The v3 version applies the reference count checking approach to two more data structures, maps and map. While maps was straightforward, struct map showed a problem where reference counted thing can be on lists and rb-trees that are oblivious to the reference count. To sanitize this, struct map is changed so that it is referenced by either a list or rb-tree node and not part of it. This simplifies the reference count and the patches have caught and fixed a number of missed or mismatched reference counts relating to struct map. A wider discussion of the approach is on the mailing list: https://lore.kernel.org/linux-perf-users/YffqnynWcc5oFkI5@kernel.org/T/#mf25ccd7a2e03de92cec29d36e2999a8ab5ec7f88 Comparing it to a past approach: https://lore.kernel.org/all/20151209021047.10245.8918.stgit@localhost.localdomain/ and to ref_tracker: https://lwn.net/Articles/877603/ v7. rebase on top of merged and Arnaldo fixed changes. The remaining 5 patches no longer refactor APIs but just add the necessary reference count checking macros and usage. v6. rebase removing 5 merged changes. Fix missed issues with libunwind. v5. rebase removing 5 merged changes. Add map_list_node__new to the 1st patch (perf map: Move map list node into symbol) as suggested by Arnaldo. Remove unnecessary map__puts from patch 12 (perf map: Changes to reference counting) as suggested by Adrian. v4. rebases on to acme's perf-tools-next, fixes more issues with map/maps and breaks apart the accessor functions to reduce individual patch sizes. The accessor functions are mechanical changes where the single biggest one is refactoring use of map->dso to be map__dso(map). Ian Rogers (5): libperf: Add reference count checking macros. perf cpumap: Add reference count checking perf namespaces: Add reference count checking perf maps: Add reference count checking. perf map: Add reference count checking tools/lib/perf/Makefile | 2 +- tools/lib/perf/cpumap.c | 94 ++++++++------- tools/lib/perf/include/internal/cpumap.h | 4 +- tools/lib/perf/include/internal/rc_check.h | 94 +++++++++++++++ tools/perf/builtin-inject.c | 2 +- tools/perf/builtin-top.c | 4 +- tools/perf/tests/cpumap.c | 4 +- tools/perf/tests/hists_link.c | 2 +- tools/perf/tests/maps.c | 20 ++-- tools/perf/tests/thread-maps-share.c | 29 ++--- tools/perf/tests/vmlinux-kallsyms.c | 4 +- tools/perf/util/annotate.c | 2 +- tools/perf/util/cpumap.c | 40 +++---- tools/perf/util/dso.c | 2 +- tools/perf/util/dsos.c | 2 +- tools/perf/util/machine.c | 27 +++-- tools/perf/util/map.c | 69 ++++++----- tools/perf/util/map.h | 32 ++--- tools/perf/util/maps.c | 64 +++++----- tools/perf/util/maps.h | 17 +-- tools/perf/util/namespaces.c | 132 ++++++++++++--------- tools/perf/util/namespaces.h | 3 +- tools/perf/util/pmu.c | 8 +- tools/perf/util/symbol-elf.c | 26 ++-- tools/perf/util/symbol.c | 55 +++++---- tools/perf/util/unwind-libdw.c | 2 +- tools/perf/util/unwind-libunwind-local.c | 2 +- tools/perf/util/unwind-libunwind.c | 2 +- 28 files changed, 448 insertions(+), 296 deletions(-) create mode 100644 tools/lib/perf/include/internal/rc_check.h -- 2.40.0.577.gac1e443424-goog