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 72063C021AA for ; Tue, 18 Feb 2025 13:02:50 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 0E93728012A; Tue, 18 Feb 2025 08:02:50 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id 0994C280124; Tue, 18 Feb 2025 08:02:50 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id EA37828012A; Tue, 18 Feb 2025 08:02:49 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0011.hostedemail.com [216.40.44.11]) by kanga.kvack.org (Postfix) with ESMTP id C5257280124 for ; Tue, 18 Feb 2025 08:02:49 -0500 (EST) Received: from smtpin22.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay09.hostedemail.com (Postfix) with ESMTP id 0852C81264 for ; Tue, 18 Feb 2025 13:02:32 +0000 (UTC) X-FDA: 83133079386.22.F86DD89 Received: from szxga07-in.huawei.com (szxga07-in.huawei.com [45.249.212.35]) by imf23.hostedemail.com (Postfix) with ESMTP id 0AADA140005 for ; Tue, 18 Feb 2025 13:02:29 +0000 (UTC) Authentication-Results: imf23.hostedemail.com; dkim=none; spf=pass (imf23.hostedemail.com: domain of tongtiangen@huawei.com designates 45.249.212.35 as permitted sender) smtp.mailfrom=tongtiangen@huawei.com; dmarc=pass (policy=quarantine) header.from=huawei.com ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1739883751; 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-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=2dyw0wAFGx2xLRiCAUVf3PCxAhf8EuxQIYAfd8MPrvs=; b=IDYtyL3RyOtecMRIYpp6XoW1olsr4aUb2uuToaV3hR1VT+njiav5Nb3B+fCUGi/jIaP3IF jGVakglLNER00F1I2OAf4FigFQHruba5Iw0IzIb+7pMf18bqv6lfrY6HeFgywypNhaGY0i 07qM7XSbdpfwwB1imkKWRWVH+XaIDEI= ARC-Authentication-Results: i=1; imf23.hostedemail.com; dkim=none; spf=pass (imf23.hostedemail.com: domain of tongtiangen@huawei.com designates 45.249.212.35 as permitted sender) smtp.mailfrom=tongtiangen@huawei.com; dmarc=pass (policy=quarantine) header.from=huawei.com ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1739883751; a=rsa-sha256; cv=none; b=5c4r4WJdPbYBETx6F8J+6fjtu4zVGp4RnIYW4KWEiQVbgpLDSuYMnSDt4PybWptT2hyatj RZ/dyAz617LFEzUICIORVV2Gl+aP9Kls4CoSBpOljPu438fr+ECvfoDp5ezdYqWyE6TTd/ iRSflVF/C5b1DzGVIfAEl/XoN3xf1PM= Received: from mail.maildlp.com (unknown [172.19.88.163]) by szxga07-in.huawei.com (SkyGuard) with ESMTP id 4Yy02G6ckbz1wn8K; Tue, 18 Feb 2025 20:58:30 +0800 (CST) Received: from kwepemk500005.china.huawei.com (unknown [7.202.194.90]) by mail.maildlp.com (Postfix) with ESMTPS id 2688E18005F; Tue, 18 Feb 2025 21:02:25 +0800 (CST) Received: from [10.174.179.234] (10.174.179.234) by kwepemk500005.china.huawei.com (7.202.194.90) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.2.1544.11; Tue, 18 Feb 2025 21:02:23 +0800 Message-ID: <24a61833-f389-b074-0d9c-d5ad9efc2046@huawei.com> Date: Tue, 18 Feb 2025 21:02:22 +0800 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101 Thunderbird/91.8.0 Subject: Re: Add Morton,Peter and David for discussion//Re: [PATCH -next] uprobes: fix two zero old_folio bugs in __replace_page() To: David Hildenbrand , Masami Hiramatsu , Oleg Nesterov , Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim , Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , "Liang, Kan" , Andrew Morton , Peter Xu CC: , , , , , linux-mm References: <20250217123826.88503-1-tongtiangen@huawei.com> <3b893634-5453-42d0-b8dc-e9d07988e9e9@redhat.com> From: Tong Tiangen In-Reply-To: <3b893634-5453-42d0-b8dc-e9d07988e9e9@redhat.com> Content-Type: text/plain; charset="UTF-8"; format=flowed Content-Transfer-Encoding: 8bit X-Originating-IP: [10.174.179.234] X-ClientProxiedBy: dggems706-chm.china.huawei.com (10.3.19.183) To kwepemk500005.china.huawei.com (7.202.194.90) X-Rspamd-Server: rspam04 X-Rspamd-Queue-Id: 0AADA140005 X-Stat-Signature: 4izti9u7s88itudspjma463imjoyse1h X-Rspam-User: X-HE-Tag: 1739883749-915788 X-HE-Meta: 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 4JGBBL7S qsVS78quYGABghCDMRsNW4HI1NEkRUXqADEPajSRrB8l1U5tpQ6yoNFQ/wu7CK5pgTWydEOxIxKoTdIEuc21ecLgsTHZ0taXWoTD+x5oPtZk3y2/PrDM9Hy/evVOHvM+tJedhTFeQsTQq1kwhBUr/PbYcA5FLzv4y4aAPyHnTmaX0rLCmbn+2WpQAPK+bMS2BuCQN7TeYOJkF99R8XIMn10lR4gzp+0PtgQFinxnXsi/G99LHNnHjEsAop0J0b8ryiMYTtMgJGB9+FKP4bzAVHV/+Sw== 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: List-Subscribe: List-Unsubscribe: 在 2025/2/18 16:23, David Hildenbrand 写道: > On 18.02.25 03:47, Tong Tiangen wrote: >> >> >> 在 2025/2/17 20:38, Tong Tiangen 写道: >>> We triggered the following error logs in syzkaller test: >>> >>>     BUG: Bad page state in process syz.7.38  pfn:1eff3 >>>     page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 >>> pfn:0x1eff3 >>>     flags: >>> 0x3fffff00004004(referenced|reserved|node=0|zone=1|lastcpupid=0x1fffff) >>>     raw: 003fffff00004004 ffffe6c6c07bfcc8 ffffe6c6c07bfcc8 >>> 0000000000000000 >>>     raw: 0000000000000000 0000000000000000 00000000fffffffe >>> 0000000000000000 >>>     page dumped because: PAGE_FLAGS_CHECK_AT_FREE flag(s) set >>>     Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS >>> 1.13.0-1ubuntu1.1 04/01/2014 >>>     Call Trace: >>>      >>>      dump_stack_lvl+0x32/0x50 >>>      bad_page+0x69/0xf0 >>>      free_unref_page_prepare+0x401/0x500 >>>      free_unref_page+0x6d/0x1b0 >>>      uprobe_write_opcode+0x460/0x8e0 >>>      install_breakpoint.part.0+0x51/0x80 >>>      register_for_each_vma+0x1d9/0x2b0 >>>      __uprobe_register+0x245/0x300 >>>      bpf_uprobe_multi_link_attach+0x29b/0x4f0 >>>      link_create+0x1e2/0x280 >>>      __sys_bpf+0x75f/0xac0 >>>      __x64_sys_bpf+0x1a/0x30 >>>      do_syscall_64+0x56/0x100 >>>      entry_SYSCALL_64_after_hwframe+0x78/0xe2 >>> >>>      BUG: Bad rss-counter state mm:00000000452453e0 type:MM_FILEPAGES >>> val:-1 >>> >>> The following syzkaller test case can be used to reproduce: >>> >>>     r2 = creat(&(0x7f0000000000)='./file0\x00', 0x8) >>>     write$nbd(r2, &(0x7f0000000580)=ANY=[], 0x10) >>>     r4 = openat(0xffffffffffffff9c, &(0x7f0000000040)='./file0\x00', >>> 0x42, 0x0) >>>     mmap$IORING_OFF_SQ_RING(&(0x7f0000ffd000/0x3000)=nil, 0x3000, >>> 0x0, 0x12, r4, 0x0) >>>     r5 = userfaultfd(0x80801) >>>     ioctl$UFFDIO_API(r5, 0xc018aa3f, &(0x7f0000000040)={0xaa, 0x20}) >>>     r6 = userfaultfd(0x80801) >>>     ioctl$UFFDIO_API(r6, 0xc018aa3f, &(0x7f0000000140)) >>>     ioctl$UFFDIO_REGISTER(r6, 0xc020aa00, >>> &(0x7f0000000100)={{&(0x7f0000ffc000/0x4000)=nil, 0x4000}, 0x2}) >>>     ioctl$UFFDIO_ZEROPAGE(r5, 0xc020aa04, >>> &(0x7f0000000000)={{&(0x7f0000ffd000/0x1000)=nil, 0x1000}}) >>>     r7 = bpf$PROG_LOAD(0x5, &(0x7f0000000140)={0x2, 0x3, >>> &(0x7f0000000200)=ANY=[@ANYBLOB="1800000000120000000000000000000095"], &(0x7f0000000000)='GPL\x00', >>> 0x7, 0x0, 0x0, 0x0, 0x0, '\x00', 0x0, @fallback=0x30, >>> 0xffffffffffffffff, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, >>> 0x0, 0x10, 0x0, @void, @value}, 0x94) >>>     bpf$BPF_LINK_CREATE_XDP(0x1c, &(0x7f0000000040)={r7, 0x0, 0x30, >>> 0x1e, @val=@uprobe_multi={&(0x7f0000000080)='./file0\x00', >>> &(0x7f0000000100)=[0x2], 0x0, 0x0, 0x1}}, 0x40) >>> >>> The cause is that zero pfn is set to the pte without increasing the rss >>> count in mfill_atomic_pte_zeropage() and the refcount of zero folio does >>> not increase accordingly. Then, the operation on the same pfn is >>> performed >>> in uprobe_write_opcode()->__replace_page() to unconditional decrease the >>> rss count and old_folio's refcount. >>> >>> Therefore, two bugs are introduced: >>> 1. The rss count is incorrect, when process exit, the check_mm() report >>>      error "Bad rss-count". >>> 2. The reserved folio (zero folio) is freed when folio->refcount is >>> zero, >>>      then free_pages_prepare->free_page_is_bad() report error "Bad >>> page state". >>> >>> To fix it, add zero folio check before rss counter and refcount >>> decrease. >>> >>> Fixes: 7396fa818d62 ("uprobes/core: Make background page replacement >>> logic account for rss_stat counters") >>> Fixes: 2b1444983508 ("uprobes, mm, x86: Add the ability to install >>> and remove uprobes breakpoints") >>> Signed-off-by: Tong Tiangen >>> --- >>>    kernel/events/uprobes.c | 6 ++++-- >>>    1 file changed, 4 insertions(+), 2 deletions(-) >>> >>> diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c >>> index 46ddf3a2334d..ff5694acfa68 100644 >>> --- a/kernel/events/uprobes.c >>> +++ b/kernel/events/uprobes.c >>> @@ -213,7 +213,8 @@ static int __replace_page(struct vm_area_struct >>> *vma, unsigned long addr, >>>            dec_mm_counter(mm, MM_ANONPAGES); >>>        if (!folio_test_anon(old_folio)) { >>> -        dec_mm_counter(mm, mm_counter_file(old_folio)); >>> +        if (!is_zero_folio(old_folio)) >>> +            dec_mm_counter(mm, mm_counter_file(old_folio)); >>>            inc_mm_counter(mm, MM_ANONPAGES); >>>        } >>> @@ -227,7 +228,8 @@ static int __replace_page(struct vm_area_struct >>> *vma, unsigned long addr, >>>        if (!folio_mapped(old_folio)) >>>            folio_free_swap(old_folio); >>>        page_vma_mapped_walk_done(&pvmw); >>> -    folio_put(old_folio); >>> +    if (!is_zero_folio(old_folio)) >>> +        folio_put(old_folio); > >>    >>        err = 0; >>>     unlock: >> > > The whole "manually replace pages" logic is fragile. I tried to rewrite > it a while back: > > https://lkml.kernel.org/r/20240604122548.359952-1-david@redhat.com > > But didn't get to follow-up yet. > > I'm not sure if the page_vma_mapped_walk() really does what we would > expect here. > > The folio_remove_rmap_pte(old_folio, old_page, vma); is certainly wrong > as well for ero folios. > > > I don't think there is a sane use case right now where we would hit the > shared zeropage. > > So for the time being, I think we should just reject them immediately > after get_user_page_vma_remote(). > > At some point I'll follow up with my rewrite that will clean this > nastiness here up a bit. OK, Before your rewrite last merged, How about i change the solution to just reject them immediately after get_user_page_vma_remote()？ Thanks, Tong. >