From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp.codeaurora.org by pdx-caf-mail.web.codeaurora.org (Dovecot) with LMTP id UzGILWXNGFt+fAAAmS7hNA ; Thu, 07 Jun 2018 06:23:57 +0000 Received: by smtp.codeaurora.org (Postfix, from userid 1000) id 69D6D6089E; Thu, 7 Jun 2018 06:23:57 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on pdx-caf-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-2.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI autolearn=ham autolearn_force=no version=3.4.0 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by smtp.codeaurora.org (Postfix) with ESMTP id 6713E6063F; Thu, 7 Jun 2018 06:23:56 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 smtp.codeaurora.org 6713E6063F Authentication-Results: pdx-caf-mail.web.codeaurora.org; dmarc=none (p=none dis=none) header.from=ah.jp.nec.com Authentication-Results: pdx-caf-mail.web.codeaurora.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752873AbeFGGXx convert rfc822-to-8bit (ORCPT + 25 others); Thu, 7 Jun 2018 02:23:53 -0400 Received: from tyo162.gate.nec.co.jp ([114.179.232.162]:55119 "EHLO tyo162.gate.nec.co.jp" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751372AbeFGGXw (ORCPT ); Thu, 7 Jun 2018 02:23:52 -0400 Received: from mailgate02.nec.co.jp ([114.179.233.122]) by tyo162.gate.nec.co.jp (8.15.1/8.15.1) with ESMTPS id w576NCxn022279 (version=TLSv1.2 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=NO); Thu, 7 Jun 2018 15:23:12 +0900 Received: from mailsv01.nec.co.jp (mailgate-v.nec.co.jp [10.204.236.94]) by mailgate02.nec.co.jp (8.15.1/8.15.1) with ESMTP id w576NCCQ011456; Thu, 7 Jun 2018 15:23:12 +0900 Received: from mail01b.kamome.nec.co.jp (mail01b.kamome.nec.co.jp [10.25.43.2]) by mailsv01.nec.co.jp (8.15.1/8.15.1) with ESMTP id w576NCFc005398; Thu, 7 Jun 2018 15:23:12 +0900 Received: from bpxc99gp.gisp.nec.co.jp ([10.38.151.149] [10.38.151.149]) by mail03.kamome.nec.co.jp with ESMTP id BT-MMP-1041300; Thu, 7 Jun 2018 15:22:20 +0900 Received: from BPXM23GP.gisp.nec.co.jp ([10.38.151.215]) by BPXC21GP.gisp.nec.co.jp ([10.38.151.149]) with mapi id 14.03.0319.002; Thu, 7 Jun 2018 15:22:20 +0900 From: Naoya Horiguchi To: Oscar Salvador CC: Matthew Wilcox , "linux-mm@kvack.org" , "linux-kernel@vger.kernel.org" , Andrew Morton , Michal Hocko , "mingo@kernel.org" , "dan.j.williams@intel.com" , Huang Ying , Pavel Tatashin Subject: Re: kernel panic in reading /proc/kpageflags when enabling RAM-simulated PMEM Thread-Topic: kernel panic in reading /proc/kpageflags when enabling RAM-simulated PMEM Thread-Index: AQHT/GeymmpDVR1XWE69lshFJtmzcqRQRpEAgABpKgCAAWucAIAALt0AgAANvoCAAAOvAIAABOqAgAFfi4A= Date: Thu, 7 Jun 2018 06:22:19 +0000 Message-ID: <20180607062218.GB22554@hori1.linux.bs1.fc.nec.co.jp> References: <20180605005402.GA22975@hori1.linux.bs1.fc.nec.co.jp> <20180605011836.GA32444@bombadil.infradead.org> <20180605073500.GA23766@hori1.linux.bs1.fc.nec.co.jp> <20180606051624.GA16021@hori1.linux.bs1.fc.nec.co.jp> <20180606080408.GA31794@techadventures.net> <20180606085319.GA32052@techadventures.net> <20180606090630.GA27065@hori1.linux.bs1.fc.nec.co.jp> <20180606092405.GA6562@hori1.linux.bs1.fc.nec.co.jp> In-Reply-To: <20180606092405.GA6562@hori1.linux.bs1.fc.nec.co.jp> Accept-Language: en-US, ja-JP Content-Language: ja-JP X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.51.8.80] Content-Type: text/plain; charset="iso-2022-jp" Content-ID: Content-Transfer-Encoding: 8BIT MIME-Version: 1.0 X-TM-AS-MML: disable Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Jun 06, 2018 at 09:24:05AM +0000, Horiguchi Naoya(堀口 直也) wrote: > On Wed, Jun 06, 2018 at 09:06:30AM +0000, Horiguchi Naoya(堀口 直也) wrote: > > On Wed, Jun 06, 2018 at 10:53:19AM +0200, Oscar Salvador wrote: > > > On Wed, Jun 06, 2018 at 10:04:08AM +0200, Oscar Salvador wrote: > > > > On Wed, Jun 06, 2018 at 05:16:24AM +0000, Naoya Horiguchi wrote: > > > > > On Tue, Jun 05, 2018 at 07:35:01AM +0000, Horiguchi Naoya(堀口 直也) wrote: > > > > > > On Mon, Jun 04, 2018 at 06:18:36PM -0700, Matthew Wilcox wrote: > > > > > > > On Tue, Jun 05, 2018 at 12:54:03AM +0000, Naoya Horiguchi wrote: > > > > > > > > Reproduction precedure is like this: > > > > > > > > - enable RAM based PMEM (with a kernel boot parameter like memmap=1G!4G) > > > > > > > > - read /proc/kpageflags (or call tools/vm/page-types with no arguments) > > > > > > > > (- my kernel config is attached) > > > > > > > > > > > > > > > > I spent a few days on this, but didn't reach any solutions. > > > > > > > > So let me report this with some details below ... > > > > > > > > > > > > > > > > In the critial page request, stable_page_flags() is called with an argument > > > > > > > > page whose ->compound_head was somehow filled with '0xffffffffffffffff'. > > > > > > > > And compound_head() returns (struct page *)(head - 1), which explains the > > > > > > > > address 0xfffffffffffffffe in the above message. > > > > > > > > > > > > > > Hm. compound_head shares with: > > > > > > > > > > > > > > struct list_head lru; > > > > > > > struct list_head slab_list; /* uses lru */ > > > > > > > struct { /* Partial pages */ > > > > > > > struct page *next; > > > > > > > unsigned long _compound_pad_1; /* compound_head */ > > > > > > > unsigned long _pt_pad_1; /* compound_head */ > > > > > > > struct dev_pagemap *pgmap; > > > > > > > struct rcu_head rcu_head; > > > > > > > > > > > > > > None of them should be -1. > > > > > > > > > > > > > > > It seems that this kernel panic happens when reading kpageflags of pfn range > > > > > > > > [0xbffd7, 0xc0000), which coresponds to a 'reserved' range. > > > > > > > > > > > > > > > > [ 0.000000] user-defined physical RAM map: > > > > > > > > [ 0.000000] user: [mem 0x0000000000000000-0x000000000009fbff] usable > > > > > > > > [ 0.000000] user: [mem 0x000000000009fc00-0x000000000009ffff] reserved > > > > > > > > [ 0.000000] user: [mem 0x00000000000f0000-0x00000000000fffff] reserved > > > > > > > > [ 0.000000] user: [mem 0x0000000000100000-0x00000000bffd6fff] usable > > > > > > > > [ 0.000000] user: [mem 0x00000000bffd7000-0x00000000bfffffff] reserved > > > > > > > > [ 0.000000] user: [mem 0x00000000feffc000-0x00000000feffffff] reserved > > > > > > > > [ 0.000000] user: [mem 0x00000000fffc0000-0x00000000ffffffff] reserved > > > > > > > > [ 0.000000] user: [mem 0x0000000100000000-0x000000013fffffff] persistent (type 12) > > > > > > > > > > > > > > > > So I guess 'memmap=' parameter might badly affect the memory initialization process. > > > > > > > > > > > > > > > > This problem doesn't reproduce on v4.17, so some pre-released patch introduces it. > > > > > > > > I hope this info helps you find the solution/workaround. > > > > > > > > > > > > > > Can you try bisecting this? It could be one of my patches to reorder struct > > > > > > > page, or it could be one of Pavel's deferred page initialisation patches. > > > > > > > Or something else ;-) > > > > > > > > > > > > Thank you for the comment. I'm trying bisecting now, let you know the result later. > > > > > > > > > > > > And I found that my statement "not reproduce on v4.17" was wrong (I used > > > > > > different kvm guests, which made some different test condition and misguided me), > > > > > > this seems an older (at least < 4.15) bug. > > > > > > > > > > (Cc: Pavel) > > > > > > > > > > Bisection showed that the following commit introduced this issue: > > > > > > > > > > commit f7f99100d8d95dbcf09e0216a143211e79418b9f > > > > > Author: Pavel Tatashin > > > > > Date: Wed Nov 15 17:36:44 2017 -0800 > > > > > > > > > > mm: stop zeroing memory during allocation in vmemmap > > > > > > > > > > This patch postpones struct page zeroing to later stage of memory initialization. > > > > > My kernel config disabled CONFIG_DEFERRED_STRUCT_PAGE_INIT so two callsites of > > > > > __init_single_page() were never reached. So in such case, struct pages populated > > > > > by vmemmap_pte_populate() could be left uninitialized? > > > > > And I'm not sure yet how this issue becomes visible with memmap= setting. > > > > > > > > I think that this becomes visible because memmap=x!y creates a persistent memory region: > > > > > > > > parse_memmap_one > > > > { > > > > ... > > > > } else if (*p == '!') { > > > > start_at = memparse(p+1, &p); > > > > e820__range_add(start_at, mem_size, E820_TYPE_PRAM); > > > > ... > > > > } > > > > > > > > and this region it is not added neither in memblock.memory nor in memblock.reserved. > > > > Ranges in memblock.memory get zeroed in memmap_init_zone(), while memblock.reserved get zeroed > > > > in free_low_memory_core_early(): > > > > > > > > static unsigned long __init free_low_memory_core_early(void) > > > > { > > > > ... > > > > for_each_reserved_mem_region(i, &start, &end) > > > > reserve_bootmem_region(start, end); > > > > ... > > > > } > > > > > > > > > > > > Maybe I am mistaken, but I think that persistent memory regions should be marked as reserved. > > > > A comment in do_mark_busy() suggests this: > > > > > > > > static bool __init do_mark_busy(enum e820_type type, struct resource *res) > > > > { > > > > > > > > ... > > > > /* > > > > * Treat persistent memory like device memory, i.e. reserve it > > > > * for exclusive use of a driver > > > > */ > > > > ... > > > > } > > > > > > > > > > > > I wonder if something like this could work and if so, if it is right (i haven't tested it yet): > > > > > > > > diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c > > > > index 71c11ad5643e..3c9686ef74e5 100644 > > > > --- a/arch/x86/kernel/e820.c > > > > +++ b/arch/x86/kernel/e820.c > > > > @@ -1247,6 +1247,11 @@ void __init e820__memblock_setup(void) > > > > if (end != (resource_size_t)end) > > > > continue; > > > > > > > > + if (entry->type == E820_TYPE_PRAM || entry->type == E820_TYPE_PMEM) { > > > > + memblock_reserve(entry->addr, entry->size); > > > > + continue; > > > > + } > > > > + > > > > if (entry->type != E820_TYPE_RAM && entry->type != E820_TYPE_RESERVED_KERN) > > > > continue; > > > > > > It does not seem to work, so the reasoning might be incorrect. > > > > Thank you for the comment. > > > > One note is that the memory region with "broken struct page" is a typical > > reserved region, not a pmem region. Strangely reading offset 0xbffd7 of > > /proc/kpageflags is OK if pmem region does not exist, but NG if pmem region exists. > > Reading the offset like 0x100000 (on pmem region) does not cause the crash, > > so pmem region seems properly set up. > > > > [ 0.000000] user-defined physical RAM map: > > [ 0.000000] user: [mem 0x0000000000000000-0x000000000009fbff] usable > > [ 0.000000] user: [mem 0x000000000009fc00-0x000000000009ffff] reserved > > [ 0.000000] user: [mem 0x00000000000f0000-0x00000000000fffff] reserved > > [ 0.000000] user: [mem 0x0000000000100000-0x00000000bffd6fff] usable > > [ 0.000000] user: [mem 0x00000000bffd7000-0x00000000bfffffff] reserved ===> "broken struct page" region > > [ 0.000000] user: [mem 0x00000000feffc000-0x00000000feffffff] reserved > > [ 0.000000] user: [mem 0x00000000fffc0000-0x00000000ffffffff] reserved > > [ 0.000000] user: [mem 0x0000000100000000-0x000000013fffffff] persistent (type 12) => pmem region > > [ 0.000000] user: [mem 0x0000000140000000-0x000000023fffffff] usable > > > > I have another note: > > > My kernel config disabled CONFIG_DEFERRED_STRUCT_PAGE_INIT so two callsites of > > __init_single_page() were never reached. So in such case, struct pages populated > > by vmemmap_pte_populate() could be left uninitialized? > > I quickly checked whether enabling CONFIG_DEFERRED_STRUCT_PAGE_INIT affect > the issue. And found that the kernel panic happens even with this config enabled. > So I'm still confused... Let me share some new facts: I gave accidentally an inconvenient memmap layout like 'memmap=1G!4G' in 2 NUMA node with 8 GB memory. While I didn't intended this, but 4GB is the address starting some memory block when no "memmap=" option is provided. (messages from free_area_init_nodes() for no "memmap=" case [ 0.000000] Early memory node ranges [ 0.000000] node 0: [mem 0x0000000000001000-0x000000000009efff] [ 0.000000] node 0: [mem 0x0000000000100000-0x00000000bffd6fff] [ 0.000000] node 0: [mem 0x0000000100000000-0x000000013fffffff] // <--- [ 0.000000] node 1: [mem 0x0000000140000000-0x000000023fffffff] When "memmap=1G!4G" is given, the range [0x0000000100000000-0x000000013fffffff] disappears and kernel messages are like below: (messages from free_area_init_nodes() for "memmap=1G!4G" case [ 0.000000] Early memory node ranges [ 0.000000] node 0: [mem 0x0000000000001000-0x000000000009efff] [ 0.000000] node 0: [mem 0x0000000000100000-0x00000000bffd6fff] [ 0.000000] node 1: [mem 0x0000000140000000-0x000000023fffffff] This makes kernel think that the end pfn of node 0 is 0 0xbffd7 instead of 0x140000, which affects the memory initialization process. memmap_init_zone() calls __init_single_page() for each page within a zone, so if zone->spanned_pages are underestimated, some pages are left uninitialized. If I provide 'memmap=1G!7G', the kernel panic does not reproduce and kernel messages are like below. (messages from free_area_init_nodes() for "memmap=1G!7G" case [ 0.000000] Early memory node ranges [ 0.000000] node 0: [mem 0x0000000000001000-0x000000000009efff] [ 0.000000] node 0: [mem 0x0000000000100000-0x00000000bffd6fff] [ 0.000000] node 0: [mem 0x0000000100000000-0x000000013fffffff] [ 0.000000] node 1: [mem 0x0000000140000000-0x00000001bfffffff] [ 0.000000] node 1: [mem 0x0000000200000000-0x000000023fffffff] I think that in order to fix this, we need some conditions and/or prechecks for memblock layout, does it make sense? Or any other better approaches? Thanks, Naoya Horiguchi