From patchwork Wed Sep 7 12:11:57 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Naoya Horiguchi X-Patchwork-Id: 12968935 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 189F9C38145 for ; Wed, 7 Sep 2022 12:12:09 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 188AE6B0072; Wed, 7 Sep 2022 08:12:08 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 139456B0073; Wed, 7 Sep 2022 08:12:08 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id F41828D0001; Wed, 7 Sep 2022 08:12:07 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0016.hostedemail.com [216.40.44.16]) by kanga.kvack.org (Postfix) with ESMTP id E24616B0072 for ; Wed, 7 Sep 2022 08:12:07 -0400 (EDT) Received: from smtpin16.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay09.hostedemail.com (Postfix) with ESMTP id B6FE3803E9 for ; Wed, 7 Sep 2022 12:12:07 +0000 (UTC) X-FDA: 79885176294.16.1FD0601 Received: from out2.migadu.com (out2.migadu.com [188.165.223.204]) by imf15.hostedemail.com (Postfix) with ESMTP id E03FBA0072 for ; Wed, 7 Sep 2022 12:12:06 +0000 (UTC) Date: Wed, 7 Sep 2022 21:11:57 +0900 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1662552724; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: in-reply-to:in-reply-to:references:references; bh=ELbwOWzF6t8jyyZfnQLzTZV8DJg2x0pLqTXQhjGKVEA=; b=xRWt0HrgX7XRzsBAlhtyNCGfdG5dscbX29DT70Z6GGIvHtcPB5MpkcIVsJ4q5/lmB1itFv f0iXlGTm2ZeTFA795Z8sUnbrrNVv3nahg2xi8MDF66ithXCLvTFwzQMeJNzhxjpsdbdJWf AiHSzReX9N5yHQiiUQubuFDFDkfDu+4= X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. From: Naoya Horiguchi To: David Hildenbrand Cc: linux-mm@kvack.org, Andrew Morton , Muchun Song , Miaohe Lin , Matthew Wilcox , Michal Hocko , Yang Shi , Naoya Horiguchi Subject: [PATCH] mm/huge_memory: use pfn_to_online_page() in split_huge_pages_all() Message-ID: <20220907121157.GA2954283@ik1-406-35019.vs.sakura.ne.jp> References: <20220907100855.GA2894785@ik1-406-35019.vs.sakura.ne.jp> <1481ce19-0a64-d1dc-4d89-4db71d6f7eb3@redhat.com> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: X-Migadu-Flow: FLOW_OUT X-Migadu-Auth-User: linux.dev ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1662552727; a=rsa-sha256; cv=none; b=2L4crt+lTsLaC5HT0jOU4/cHV05BfPirCdnSNM2HSVKSry079D5D7c+50BFdhhGSCSe6vL vSD09GoBsCKmUiV0PHaKyJhtsHpPIPtw5cS+9zQCPxFfHJy5bqVQFo3hCvB9jWudR9ygO8 478ai+qzmTQMj7pbnfk783zluJJkL2k= ARC-Authentication-Results: i=1; imf15.hostedemail.com; dkim=pass header.d=linux.dev header.s=key1 header.b=xRWt0Hrg; spf=pass (imf15.hostedemail.com: domain of naoya.horiguchi@linux.dev designates 188.165.223.204 as permitted sender) smtp.mailfrom=naoya.horiguchi@linux.dev; dmarc=pass (policy=none) header.from=linux.dev ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1662552727; 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: in-reply-to:in-reply-to:references:references:dkim-signature; bh=ELbwOWzF6t8jyyZfnQLzTZV8DJg2x0pLqTXQhjGKVEA=; b=njN+Y+dhsS9X9cSTddiQ4RwHN8PbCNc6ocDD1aBUsws6bQ4GTa0iNiS1I2qb4JXzhEFcUr sXcv3MPNlmHGI2ZpFTM3ItFTFXEpLTJym8cvXfel8Gm9zd5hg0d/XTPu+p9yLj0oHF0z4W 0aCxx/WH4yLkJ67atishTiAsZazqzh8= X-Rspam-User: X-Rspamd-Queue-Id: E03FBA0072 Authentication-Results: imf15.hostedemail.com; dkim=pass header.d=linux.dev header.s=key1 header.b=xRWt0Hrg; spf=pass (imf15.hostedemail.com: domain of naoya.horiguchi@linux.dev designates 188.165.223.204 as permitted sender) smtp.mailfrom=naoya.horiguchi@linux.dev; dmarc=pass (policy=none) header.from=linux.dev X-Stat-Signature: pnjwx3398kzj3brj7ubzq5yks6zdcany X-Rspamd-Server: rspam03 X-HE-Tag: 1662552726-164126 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 Wed, Sep 07, 2022 at 12:26:18PM +0200, David Hildenbrand wrote: > On 07.09.22 12:23, David Hildenbrand wrote: > > On 07.09.22 12:08, Naoya Horiguchi wrote: > > > Hi MM folks, > > > > Hi, > > > > > > > > When I'm testing memory hotremove with various settings, I found the following > > > NULL-pointer dereference. It reproduces easily with the folloing steps: > > > > > > $ echo offline > /sys/devices/system/memory/memoryN/state > > > $ echo 1 > /sys/kernel/debug/split_huge_pages > > > > > > > That's weird, I don't immediately see how both features are related here, especially because it seems to fail quite early in split_huge_pages_write(). > > > > I was able to trigger it here as well, though, and I get on my kernel with debug options enabled: > > > > [526045.808737] page:00000000467f4890 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x121c000 > > [526045.818306] flags: 0x17fffc00000000(node=0|zone=2|lastcpupid=0x1ffff) > > [526045.824842] raw: 0017fffc00000000 0000000000000000 dead000000000122 0000000000000000 > > [526045.832676] raw: 0000000000000000 0000000000000000 00000001ffffffff 0000000000000000 > > [526045.840498] page dumped because: unmovable page > > [526056.362715] page:000000007d7ab72e is uninitialized and poisoned > > [526056.362720] page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) > > [526056.374837] ------------[ cut here ]------------ > > [526056.379544] kernel BUG at include/linux/mm.h:1248! > > [526056.384429] invalid opcode: 0000 [#1] PREEMPT SMP PTI > > [526056.389570] CPU: 16 PID: 20964 Comm: bash Tainted: G I 6.0.0-rc3-foll-numa+ #41 > > [526056.398347] Hardware name: Dell Inc. PowerEdge R740/0923K0, BIOS 2.11.2 004/21/2021 > > [526056.406087] RIP: 0010:split_huge_pages_write+0xcf4/0xe30 > > [526056.411495] Code: f9 ff ff 48 c7 c6 88 8e 4e a7 48 c7 c7 f8 0b a8 a7 e8 20 1c 3d 00 e9 54 f6 ff ff 48 c7 c6 38 06 4c a7 4c 89 e7 e8 5c 4a f7 ff <0f> 0b 48 8b 7c 24 08 be 02 00 00 00 e8 bb 6c 36 00 e9 11 f5 ff ff > > [526056.430325] RSP: 0018:ffffae50e757fb40 EFLAGS: 00010292 > > [526056.435639] RAX: 0000000000000034 RBX: 0000000002d00000 RCX: 0000000000000000 > > [526056.442858] RDX: 0000000000000001 RSI: ffffffffa751a839 RDI: 00000000ffffffff > > [526056.450077] RBP: 0000000000013937 R08: 0000000000000000 R09: ffffae50e757fa08 > > [526056.457296] R10: 0000000000000003 R11: ffffffffa793c768 R12: fffff00eb4000000 > > [526056.464514] R13: ffffae50e757fb7a R14: 000fffffffffffff R15: fffff00eb4000000 > > [526056.471733] FS: 00007fe35addf740(0000) GS:ffff8b71dc000000(0000) knlGS:0000000000000000 > > [526056.479906] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > > [526056.485738] CR2: 000055e87e606358 CR3: 00000004f5b46004 CR4: 00000000007706e0 > > [526056.492957] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 > > [526056.500176] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 > > [526056.507396] PKRU: 55555554 > > [526056.510196] Call Trace: > > [526056.512734] > > [526056.514929] ? simple_setattr+0x40/0x60 > > [526056.518859] ? vfs_mkobj+0x1b0/0x1c0 > > [526056.522529] ? path_openat+0xb2e/0x1360 > > [526056.526456] ? do_filp_open+0xa1/0x130 > > [526056.530296] full_proxy_write+0x50/0x80 > > [526056.534229] vfs_write+0xd7/0x3e0 > > [526056.537635] ? fpregs_assert_state_consistent+0x22/0x50 > > [526056.542955] ? exit_to_user_mode_prepare+0x3c/0x1d0 > > [526056.547929] ? syscall_exit_to_user_mode+0x17/0x40 > > [526056.552808] ksys_write+0x53/0xd0 > > [526056.556215] do_syscall_64+0x58/0x80 > > [526056.559879] ? exit_to_user_mode_prepare+0x3c/0x1d0 > > [526056.564846] ? syscall_exit_to_user_mode+0x17/0x40 > > [526056.569726] ? do_syscall_64+0x67/0x80 > > [526056.573563] ? do_syscall_64+0x67/0x80 > > [526056.577404] ? do_syscall_64+0x67/0x80 > > [526056.581244] ? syscall_exit_to_user_mode+0x17/0x40 > > [526056.586120] ? do_syscall_64+0x67/0x80 > > [526056.589961] ? do_syscall_64+0x67/0x80 > > [526056.593801] ? do_syscall_64+0x67/0x80 > > [526056.597641] entry_SYSCALL_64_after_hwframe+0x63/0xcd > > [526056.602779] RIP: 0033:0x7fe35ab01c17 > > [526056.606446] Code: 0f 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 48 89 54 24 18 48 89 74 24 > > [526056.625277] RSP: 002b:00007ffc1863c8d8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 > > [526056.632929] RAX: ffffffffffffffda RBX: 0000000000000002 RCX: 00007fe35ab01c17 > > [526056.640147] RDX: 0000000000000002 RSI: 000055e87e7009a0 RDI: 0000000000000001 > > [526056.647366] RBP: 000055e87e7009a0 R08: 0000000000000000 R09: 0000000000000073 > > [526056.654586] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000002 > > [526056.661805] R13: 00007fe35abf8780 R14: 0000000000000002 R15: 00007fe35abf39e0 > > [526056.669028] > > > > > > Looks like there is a page_to_nid() done in an offline memmap, which is wrong. > > > > Usually, this indicates that a pfn_to_online_page() is missing in a PFN walker. > > > > And indeed, something like the following might do the trick: > > diff --git a/mm/huge_memory.c b/mm/huge_memory.c > index e9414ee57c5b..f42bb51e023a 100644 > --- a/mm/huge_memory.c > +++ b/mm/huge_memory.c > @@ -2894,11 +2894,9 @@ static void split_huge_pages_all(void) > max_zone_pfn = zone_end_pfn(zone); > for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) { > int nr_pages; > - if (!pfn_valid(pfn)) > - continue; > - page = pfn_to_page(pfn); > - if (!get_page_unless_zero(page)) > + page = pfn_to_online_page(pfn); > + if (!page || !get_page_unless_zero(page)) > continue; > if (zone != page_zone(page)) > Thank you very much, I confirmed the fix. So I try to write a patch like below. I borrowed your debug message because it has more helpful info than mine. If you have any comment or suggestion in the description, please let me know. Thanks, Naoya Horiguchi Signed-off-by: David Hildenbrand Acked-by: Michal Hocko Reviewed-by: Miaohe Lin Reviewed-by: Oscar Salvador --- From 8a5c284df732943065d23838090d15c94cd10395 Mon Sep 17 00:00:00 2001 From: Naoya Horiguchi Date: Wed, 7 Sep 2022 20:58:44 +0900 Subject: [PATCH] mm/huge_memory: use pfn_to_online_page() in split_huge_pages_all() NULL pointer dereference is triggered when calling thp split via debugfs on the system with offlined memory blocks. With debug option enabled, the following kernel messages are printed out: page:00000000467f4890 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x121c000 flags: 0x17fffc00000000(node=0|zone=2|lastcpupid=0x1ffff) raw: 0017fffc00000000 0000000000000000 dead000000000122 0000000000000000 raw: 0000000000000000 0000000000000000 00000001ffffffff 0000000000000000 page dumped because: unmovable page page:000000007d7ab72e is uninitialized and poisoned page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p)) ------------[ cut here ]------------ kernel BUG at include/linux/mm.h:1248! invalid opcode: 0000 [#1] PREEMPT SMP PTI CPU: 16 PID: 20964 Comm: bash Tainted: G I 6.0.0-rc3-foll-numa+ #41 ... RIP: 0010:split_huge_pages_write+0xcf4/0xe30 This shows that page_to_nid() in page_zone() is unexpectedly called for an offlined memmap. Use pfn_to_online_page() to get struct page in PFN walker. Fixes: 49071d436b51 ("thp: add debugfs handle to split all huge pages") Suggested-by: David Hildenbrand Signed-off-by: Naoya Horiguchi Cc: --- mm/huge_memory.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/mm/huge_memory.c b/mm/huge_memory.c index 5fa2ba86dae4..03149a8f46f9 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -2894,11 +2894,8 @@ static void split_huge_pages_all(void) max_zone_pfn = zone_end_pfn(zone); for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) { int nr_pages; - if (!pfn_valid(pfn)) - continue; - - page = pfn_to_page(pfn); - if (!get_page_unless_zero(page)) + page = pfn_to_online_page(pfn); + if (!page || !get_page_unless_zero(page)) continue; if (zone != page_zone(page))