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 Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id AD8FBC46467 for ; Sat, 7 Jan 2023 08:07:22 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id E99B38E0002; Sat, 7 Jan 2023 03:07:21 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id E49088E0001; Sat, 7 Jan 2023 03:07:21 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id D103F8E0002; Sat, 7 Jan 2023 03:07:21 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0017.hostedemail.com [216.40.44.17]) by kanga.kvack.org (Postfix) with ESMTP id C1C8A8E0001 for ; Sat, 7 Jan 2023 03:07:21 -0500 (EST) Received: from smtpin01.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay04.hostedemail.com (Postfix) with ESMTP id 9C0151A021F for ; Sat, 7 Jan 2023 08:07:21 +0000 (UTC) X-FDA: 80327273082.01.746C071 Received: from mail3-166.sinamail.sina.com.cn (mail3-166.sinamail.sina.com.cn [202.108.3.166]) by imf11.hostedemail.com (Postfix) with ESMTP id 26CCC4000B for ; Sat, 7 Jan 2023 08:07:17 +0000 (UTC) Authentication-Results: imf11.hostedemail.com; dkim=none; spf=pass (imf11.hostedemail.com: domain of hdanton@sina.com designates 202.108.3.166 as permitted sender) smtp.mailfrom=hdanton@sina.com; dmarc=none ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1673078840; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=VhIdgPSQxwimZfnFDlAq9nsOAZmdlKWRK+BS1wrFsvA=; b=rrr7ctwkLHoYhZamX9J0Ow0ZWt8upVXwscwPD4rwL99yOiprS97ccsRT4f8Rb9R7Ovs2pb K9d6qpF8cmjW2Z9Fsr5zAM9DZYXAgNdLIM7ghZEEFL7VM1qCQ7AQQL80Sb+mk/3tNOxExe 7cV7HNQvp5UrAlaVP8bMqihmighNHeo= ARC-Authentication-Results: i=1; imf11.hostedemail.com; dkim=none; spf=pass (imf11.hostedemail.com: domain of hdanton@sina.com designates 202.108.3.166 as permitted sender) smtp.mailfrom=hdanton@sina.com; dmarc=none ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1673078840; a=rsa-sha256; cv=none; b=akiIx8K5b6tGQZjoX4wFXyqW7NNnUdzVwZ0bIR+liH9qpDH/T1i7nrLkypGNNFX4ia6pjO j96HwEgm66zORrLFzBgFF3j7Mh8+a2Pgp+73ww7N1E49rcyK+KVBsJgrK88OyT6OC23VGf dJhmIGl8W2aTYqRkPk5Z6vOSnNdu+oI= Received: from unknown (HELO localhost.localdomain)([114.249.57.238]) by sina.com (172.16.97.23) with ESMTP id 63B9277D0002729E; Sat, 7 Jan 2023 16:04:15 +0800 (CST) X-Sender: hdanton@sina.com X-Auth-ID: hdanton@sina.com X-SMAIL-MID: 35420054920248 From: Hillf Danton To: Munehisa Kamata Cc: hannes@cmpxchg.org, ebiggers@kernel.org, mengcc@amazon.com, Suren Baghdasaryan , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: Re: another use-after-free in ep_remove_wait_queue() Date: Sat, 7 Jan 2023 16:07:02 +0800 Message-Id: <20230107080702.4907-1-hdanton@sina.com> In-Reply-To: <20230106224859.4123476-1-kamatam@amazon.com> References: <20230106224859.4123476-1-kamatam@amazon.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Stat-Signature: 4z5ptigwcofnqfpzmtx9beozyqinepr5 X-Rspam-User: X-Rspamd-Queue-Id: 26CCC4000B X-Rspamd-Server: rspam06 X-HE-Tag: 1673078837-228240 X-HE-Meta: U2FsdGVkX19yI/z0nuDnl+BnLdS6lThPLkJUeltwn+cXiZzZv5Gc/CoczJ8r3qYi/8qXRUowlgRyM5k1rVl+es6mtJFyiCVthLaVioL/oTKc6I+M5d/NRn8ovrQ2CdTUPIYUxe1hzX+rE3eJNzWgo0yW/ncnVkEqqpwDzw30R2UfhFIv2I4o7/bI6P/kgkmDZ9gSIoO7+7df3yXYYyp8opnuJKel1/6ZXeWKC0svWE0zKoiZwsooD/ghvC1V5ntgAfRkGs+A6k9v6X16dEmrf+YPXT/acGIWABSsXmWyu3FgSxN9LdO/I8lCPfLmdV+cgHsgHlI6YxrJgi9lIU83puLRA5PCidwinoneVP/N263/vzwCqEHmRVRAyVZbZn2gunf365M0qqFVgfU3MTazO7RqUvQcwwPuUEt9Lw/lJ6W5lI795SV0v4tBJDXBLN0qVc/Tqjc2vtEImOi0rT56ZjGnmebLXIVu8EUCWnNvwbb8fgYDFxCpbLQSlPngkA5KAyMMDfSKvNorsi0F9S2n3g2PUiZW8n0cP9qFJbcKF0MTDs4D7K8k5dnUg9aZHwIGKc4XFYlTiEA50MLsGQPcMiQ7tSSGjF9nSF6jNNqOcoZpei6QofwXT+fltMMtk/RF4dgGaGSEhA4SJtz/E7t2qsfrysNSkNdmM9bCQpmtjAAMG0kjq+eKawGRKOA/oPQQjN72dkYg9LAZllgDlyVA5URXa6SH0h3zkVqZMCZOGYPCrIr/nqv49+ooTDMZy21ElQwBeRGvGANSe0OspBufgkZeIFsjOA9Is9ZOagxPNLaJ0e3BCD2cMkQKWOrxzKPY4HczOLQ+Ii1Up6JT280CVdEu9lz8iFOi4vpeuh3XIT3wZe6MKeq4oxeYjggzCD+tvKge1Gh6woJkwMqQNPvzz8IHoMzQVjaGRFD+bY5/pXLUB7LlMRk0Zg== 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: On Fri, Jan 6, 2023 at 2:49 PM Munehisa Kamata wrote: > Hi, > > We found the following use-after-free with userspace code polling on > a pressure file in a non-root cgroup using epoll. > > [ 57.183661] BUG: KASAN: use-after-free in _raw_spin_lock_irqsave+0x76/0x130 > [ 57.186662] Write of size 4 at addr ffff888114976428 by task a.out/2426 > > [ 57.190551] CPU: 0 PID: 2426 Comm: a.out Not tainted 6.2.0-rc2+ #16 > [ 57.193384] Hardware name: Amazon EC2 c5a.large/, BIOS 1.0 10/16/2017 > [ 57.196272] Call Trace: > [ 57.197565] > [ 57.198714] dump_stack_lvl+0x8f/0xc0 > [ 57.200494] print_report+0x16c/0x4e0 > [ 57.202084] ? _raw_spin_lock_irqsave+0x76/0x130 > [ 57.204077] kasan_report+0xc3/0xf0 > [ 57.205587] ? entry_SYSCALL_64_after_hwframe+0x63/0xcd > [ 57.207760] ? _raw_spin_lock_irqsave+0x76/0x130 > [ 57.209685] kasan_check_range+0x2d2/0x310 > [ 57.211477] _raw_spin_lock_irqsave+0x76/0x130 > [ 57.213355] remove_wait_queue+0x25/0x130 > [ 57.215102] ep_free+0x12d/0x220 > [ 57.216506] ep_eventpoll_release+0x3c/0x40 > [ 57.218254] __fput+0x32b/0x700 > [ 57.221486] task_work_run+0x1db/0x230 > [ 57.224885] exit_to_user_mode_prepare+0xfd/0x100 > [ 57.228662] syscall_exit_to_user_mode+0x20/0x40 > [ 57.232360] do_syscall_64+0x52/0x90 > [ 57.235691] entry_SYSCALL_64_after_hwframe+0x63/0xcd > [ 57.239572] RIP: 0033:0x7fadf96e1c44 > [ 57.242865] Code: 00 00 b8 ff ff ff ff eb 9c b8 ff ff ff ff eb 95 e8 01 e2 01 00 90 8b 05 2a ac 2c 00 48 63 ff 85 c0 75 11 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 3a f3 c3 48 83 ec 18 48 89 7c 24 08 e8 e4 a0 > [ 57.255244] RSP: 002b:00007ffd1d1b7b98 EFLAGS: 00000246 ORIG_RAX: 0000000000000003 > [ 57.261714] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 00007fadf96e1c44 > [ 57.266293] RDX: 0000000000000000 RSI: 00007ffd1d1b7b60 RDI: 0000000000000004 > [ 57.270979] RBP: 00007ffd1d1b7bf0 R08: 00000000004007e0 R09: 00007fadf9a0f240 > [ 57.275856] R10: 00000000000006ba R11: 0000000000000246 R12: 00000000004005e0 > [ 57.280478] R13: 00007ffd1d1b7cd0 R14: 0000000000000000 R15: 0000000000000000 > [ 57.285059] > > [ 57.290402] Allocated by task 2426: > [ 57.293705] kasan_set_track+0x3d/0x60 > [ 57.297102] __kasan_kmalloc+0x85/0x90 > [ 57.300491] psi_trigger_create+0x155/0x850 > [ 57.304040] pressure_write+0x200/0x510 > [ 57.307508] cgroup_file_write+0x1de/0x3e0 > [ 57.310949] kernfs_fop_write_iter+0x27d/0x380 > [ 57.314601] vfs_write+0x7d7/0xaa0 > [ 57.317891] ksys_write+0xd7/0x1a0 > [ 57.321152] do_syscall_64+0x43/0x90 > [ 57.324496] entry_SYSCALL_64_after_hwframe+0x63/0xcd > > [ 57.330887] Freed by task 2429: > [ 57.334053] kasan_set_track+0x3d/0x60 > [ 57.337448] kasan_save_free_info+0x27/0x40 > [ 57.340977] ____kasan_slab_free+0x11d/0x170 > [ 57.344570] slab_free_freelist_hook+0x87/0x150 > [ 57.348236] __kmem_cache_free+0xcb/0x180 > [ 57.351710] psi_trigger_destroy+0x458/0x550 > [ 57.355274] cgroup_file_release+0x96/0x110 > [ 57.358779] kernfs_drain_open_files+0x238/0x420 > [ 57.362519] kernfs_drain+0x191/0x2a0 > [ 57.365901] __kernfs_remove+0x3a6/0x600 > [ 57.369363] kernfs_remove_by_name_ns+0xc2/0x120 > [ 57.373073] cgroup_addrm_files+0x90f/0xcf0 > [ 57.376610] cgroup_destroy_locked+0x48a/0x730 > [ 57.380260] cgroup_rmdir+0x2b/0x130 > [ 57.383650] kernfs_iop_rmdir+0x17a/0x230 > [ 57.387201] vfs_rmdir+0x196/0x410 > [ 57.390442] do_rmdir+0x1c7/0x3f0 > [ 57.393651] __x64_sys_rmdir+0x45/0x50 > [ 57.397000] do_syscall_64+0x43/0x90 > [ 57.400340] entry_SYSCALL_64_after_hwframe+0x63/0xcd > > [ 57.406689] The buggy address belongs to the object at ffff888114976400 > which belongs to the cache kmalloc-128 of size 128 > [ 57.414907] The buggy address is located 40 bytes inside of > 128-byte region [ffff888114976400, ffff888114976480) > > [ 57.425474] The buggy address belongs to the physical page: > [ 57.429541] page:000000008c5ecb31 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x114976 > [ 57.436725] flags: 0x2fffff80000200(slab|node=0|zone=2|lastcpupid=0x1fffff) > [ 57.441235] raw: 002fffff80000200 ffff8881000418c0 dead000000000100 dead000000000122 > [ 57.447793] raw: 0000000000000000 0000000000100010 00000001ffffffff 0000000000000000 > [ 57.454274] page dumped because: kasan: bad access detected > > [ 57.460990] Memory state around the buggy address: > [ 57.464744] ffff888114976300: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb > [ 57.471126] ffff888114976380: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc > [ 57.477447] >ffff888114976400: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb > [ 57.483833] ^ > [ 57.487541] ffff888114976480: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc > [ 57.493976] ffff888114976500: 00 00 00 00 00 00 00 00 00 00 00 00 00 01 fc fc > > Here is the simple repro. > > #include > #include > #include > > int main(void) > { > const char trigger_str[] = "some 100000 1000000"; > int fd, epfd; > struct epoll_event event; > struct epoll_event events[1]; > > fd = open("/cgroup2/test/cpu.pressure", O_RDWR); > write(fd, trigger_str, sizeof(trigger_str)); > epfd = epoll_create(1); > event.events = EPOLLPRI | EPOLLET; > event.data.fd = fd; > epoll_ctl(epfd, EPOLL_CTL_ADD, fd, &event); > epoll_wait(epfd, events, 1, -1); /* returns after rmdir */ > sleep(5); > close(epfd); > close(fd); > > return 0; > } > > # mkdir /cgroup2 > # mount -t cgroup2 none /cgroup2 > # mkdir /cgroup2/test > # ./a.out & > # rmdir /cgroup2/test > > Looks like calling wake_up_pollfree() in psi_trigger_destroy() can properly > clear the queue and then avoid this use-after-free, but POLLFREE wasn't > considered enough there for the past similar issue[1]. While > wake_up_pollfree() could *also* be called in psi_trigger_destroy(), it may > be awkward and there can be more appropriate solution. It would be great if > experts could have a look. Thanks for your report. In the wakeup pattern below, cpu0 cpu2 --- --- wake_up(&t->event_wait); HZ later kfree(t); HZ *3 later the last sleeper gets on CPU remove_wait_queue waker should wait for every sleeper to go off waitqueue. See if the diff below could survive your repro. Hillf --- mainline/kernel/sched/psi.c +++ y/kernel/sched/psi.c @@ -1346,7 +1346,7 @@ void psi_trigger_destroy(struct psi_trig * Wakeup waiters to stop polling. Can happen if cgroup is deleted * from under a polling process. */ - wake_up_interruptible(&t->event_wait); + wake_up_all(&t->event_wait); mutex_lock(&group->trigger_lock); @@ -1394,6 +1394,10 @@ void psi_trigger_destroy(struct psi_trig kthread_stop(task_to_destroy); atomic_set(&group->poll_scheduled, 0); } + + while (wq_has_sleeper(&t->event_wait)) + schedule_timeout_idle(HZ); + kfree(t); } --