From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-it0-f69.google.com (mail-it0-f69.google.com [209.85.214.69]) by kanga.kvack.org (Postfix) with ESMTP id B37E56B0003 for ; Mon, 11 Jun 2018 05:15:36 -0400 (EDT) Received: by mail-it0-f69.google.com with SMTP id m12-v6so8674941ita.6 for ; Mon, 11 Jun 2018 02:15:36 -0700 (PDT) Received: from tyo162.gate.nec.co.jp (tyo162.gate.nec.co.jp. [114.179.232.162]) by mx.google.com with ESMTPS id r143-v6si6580691ita.0.2018.06.11.02.15.34 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Mon, 11 Jun 2018 02:15:35 -0700 (PDT) From: Naoya Horiguchi Subject: Re: kernel panic in reading /proc/kpageflags when enabling RAM-simulated PMEM Date: Mon, 11 Jun 2018 09:05:00 +0000 Message-ID: <20180611090500.GA20480@hori1.linux.bs1.fc.nec.co.jp> References: <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> <20180607062218.GB22554@hori1.linux.bs1.fc.nec.co.jp> <20180607065940.GA7334@techadventures.net> <20180607094921.GA8545@techadventures.net> <20180607100256.GA9129@hori1.linux.bs1.fc.nec.co.jp> In-Reply-To: <20180607100256.GA9129@hori1.linux.bs1.fc.nec.co.jp> Content-Language: ja-JP Content-Type: text/plain; charset="iso-2022-jp" Content-ID: <0C1CA386112E9742BDC2E4797561174B@gisp.nec.co.jp> Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Sender: owner-linux-mm@kvack.org List-ID: 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 On Thu, Jun 07, 2018 at 10:02:56AM +0000, Horiguchi Naoya(=1B$BKY8}=1B(B = =1B$BD>Li=1B(B) wrote: > On Thu, Jun 07, 2018 at 11:49:21AM +0200, Oscar Salvador wrote: > > On Thu, Jun 07, 2018 at 08:59:40AM +0200, Oscar Salvador wrote: > > > On Thu, Jun 07, 2018 at 06:22:19AM +0000, Naoya Horiguchi wrote: > > > > On Wed, Jun 06, 2018 at 09:24:05AM +0000, Horiguchi Naoya(=1B$BKY8}= =1B(B =1B$BD>Li=1B(B) wrote: > > > > > On Wed, Jun 06, 2018 at 09:06:30AM +0000, Horiguchi Naoya(=1B$BKY= 8}=1B(B =1B$BD>Li=1B(B) 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 wrot= e: > > > > > > > > On Wed, Jun 06, 2018 at 05:16:24AM +0000, Naoya Horiguchi w= rote: > > > > > > > > > On Tue, Jun 05, 2018 at 07:35:01AM +0000, Horiguchi Naoya= (=1B$BKY8}=1B(B =1B$BD>Li=1B(B) wrote: > > > > > > > > > > On Mon, Jun 04, 2018 at 06:18:36PM -0700, Matthew Wilco= x wrote: > > > > > > > > > > > On Tue, Jun 05, 2018 at 12:54:03AM +0000, Naoya Horig= uchi wrote: > > > > > > > > > > > > Reproduction precedure is like this: > > > > > > > > > > > > - enable RAM based PMEM (with a kernel boot parame= ter like memmap=3D1G!4G) > > > > > > > > > > > > - read /proc/kpageflags (or call tools/vm/page-typ= es with no arguments) > > > > > > > > > > > > (- my kernel config is attached) > > > > > > > > > > > > > > > > > > > > > > > > I spent a few days on this, but didn't reach any so= lutions. > > > > > > > > > > > > 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 { /* Pa= rtial 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 readin= g kpageflags of pfn range > > > > > > > > > > > > [0xbffd7, 0xc0000), which coresponds to a 'reserved= ' range. > > > > > > > > > > > > > > > > > > > > > > > > [ 0.000000] user-defined physical RAM map: > > > > > > > > > > > > [ 0.000000] user: [mem 0x0000000000000000-0x0000= 00000009fbff] usable > > > > > > > > > > > > [ 0.000000] user: [mem 0x000000000009fc00-0x0000= 00000009ffff] reserved > > > > > > > > > > > > [ 0.000000] user: [mem 0x00000000000f0000-0x0000= 0000000fffff] reserved > > > > > > > > > > > > [ 0.000000] user: [mem 0x0000000000100000-0x0000= 0000bffd6fff] usable > > > > > > > > > > > > [ 0.000000] user: [mem 0x00000000bffd7000-0x0000= 0000bfffffff] reserved > > > > > > > > > > > > [ 0.000000] user: [mem 0x00000000feffc000-0x0000= 0000feffffff] reserved > > > > > > > > > > > > [ 0.000000] user: [mem 0x00000000fffc0000-0x0000= 0000ffffffff] reserved > > > > > > > > > > > > [ 0.000000] user: [mem 0x0000000100000000-0x0000= 00013fffffff] persistent (type 12) > > > > > > > > > > > > > > > > > > > > > > > > So I guess 'memmap=3D' parameter might badly affect= the memory initialization process. > > > > > > > > > > > > > > > > > > > > > > > > This problem doesn't reproduce on v4.17, so some pr= e-released patch introduces it. > > > > > > > > > > > > I hope this info helps you find the solution/workar= ound. > > > > > > > > > > > > > > > > > > > > > > Can you try bisecting this? It could be one of my pa= tches to reorder struct > > > > > > > > > > > page, or it could be one of Pavel's deferred page ini= tialisation patches. > > > > > > > > > > > Or something else ;-) > > > > > > > > > > > > > > > > > > > > Thank you for the comment. I'm trying bisecting now, le= t 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 co= ndition and misguided me), > > > > > > > > > > this seems an older (at least < 4.15) bug. > > > > > > > > > > > > > > > > > > (Cc: Pavel) > > > > > > > > > > > > > > > > > > Bisection showed that the following commit introduced thi= s issue: > > > > > > > > > > > > > > > > > > commit f7f99100d8d95dbcf09e0216a143211e79418b9f > > > > > > > > > Author: Pavel Tatashin > > > > > > > > > Date: Wed Nov 15 17:36:44 2017 -0800 > > > > > > > > > > > > > > > > > > mm: stop zeroing memory during allocation in vmemma= p > > > > > > > > > > > > > > > > > > This patch postpones struct page zeroing to later stage o= f memory initialization. > > > > > > > > > My kernel config disabled CONFIG_DEFERRED_STRUCT_PAGE_INI= T 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=3D setting. > > > > > > > > > > > > > > > > I think that this becomes visible because memmap=3Dx!y crea= tes a persistent memory region: > > > > > > > > > > > > > > > > parse_memmap_one > > > > > > > > { > > > > > > > > ... > > > > > > > > } else if (*p =3D=3D '!') { > > > > > > > > start_at =3D memparse(p+1, &p); > > > > > > > > e820__range_add(start_at, mem_size, E820_TY= PE_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 reg= ions 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(voi= d) > > > > > > > > if (end !=3D (resource_size_t)end) > > > > > > > > continue; > > > > > > > > > > > > > > > > + if (entry->type =3D=3D E820_TYPE_PRAM || en= try->type =3D=3D E820_TYPE_PMEM) { > > > > > > > > + memblock_reserve(entry->addr, entry= ->size); > > > > > > > > + continue; > > > > > > > > + } > > > > > > > > + > > > > > > > > if (entry->type !=3D E820_TYPE_RAM && entry= ->type !=3D E820_TYPE_RESERVED_KERN) > > > > > > > > continue; > > > > > > > > > > > > > > It does not seem to work, so the reasoning might be incorrect= . > > > > > >=20 > > > > > > Thank you for the comment. > > > > > >=20 > > > > > > One note is that the memory region with "broken struct page" is= a typical > > > > > > reserved region, not a pmem region. Strangely reading offset 0x= bffd7 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 caus= e the crash, > > > > > > so pmem region seems properly set up. > > > > > >=20 > > > > > > [ 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 =3D=3D=3D> "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) =3D> pmem region > > > > > > [ 0.000000] user: [mem 0x0000000140000000-0x000000023fffffff= ] usable > > > > > >=20 > > > > >=20 > > > > > I have another note: > > > > >=20 > > > > > > My kernel config disabled CONFIG_DEFERRED_STRUCT_PAGE_INIT so t= wo callsites of > > > > > > __init_single_page() were never reached. So in such case, struc= t pages populated > > > > > > by vmemmap_pte_populate() could be left uninitialized? > > > > >=20 > > > > > I quickly checked whether enabling CONFIG_DEFERRED_STRUCT_PAGE_IN= IT affect > > > > > the issue. And found that the kernel panic happens even with this= config enabled. > > > > > So I'm still confused... > > > >=20 > > > > Let me share some new facts: > > > >=20 > > > > I gave accidentally an inconvenient memmap layout like 'memmap=3D1G= !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=3D" option is provided. > > > >=20 > > > > (messages from free_area_init_nodes() for no "memmap=3D" case > > > > [ 0.000000] Early memory node ranges > > > > [ 0.000000] node 0: [mem 0x0000000000001000-0x000000000009= efff] > > > > [ 0.000000] node 0: [mem 0x0000000000100000-0x00000000bffd= 6fff] > > > > [ 0.000000] node 0: [mem 0x0000000100000000-0x000000013fff= ffff] // <--- > > > > [ 0.000000] node 1: [mem 0x0000000140000000-0x000000023fff= ffff] > > > >=20 > > > > When "memmap=3D1G!4G" is given, the range [0x0000000100000000-0x000= 000013fffffff] > > > > disappears and kernel messages are like below: > > > >=20 > > > > (messages from free_area_init_nodes() for "memmap=3D1G!4G" case > > > > [ 0.000000] Early memory node ranges > > > > [ 0.000000] node 0: [mem 0x0000000000001000-0x000000000009= efff] > > > > [ 0.000000] node 0: [mem 0x0000000000100000-0x00000000bffd= 6fff] > > > > [ 0.000000] node 1: [mem 0x0000000140000000-0x000000023fff= ffff] > > > >=20 > > > > This makes kernel think that the end pfn of node 0 is 0 0xbffd7 > > > > instead of 0x140000, which affects the memory initialization proces= s. > > > > memmap_init_zone() calls __init_single_page() for each page within = a zone, > > > > so if zone->spanned_pages are underestimated, some pages are left u= ninitialized. > > > >=20 > > > > If I provide 'memmap=3D1G!7G', the kernel panic does not reproduce = and > > > > kernel messages are like below. > > > > =20 > > > > (messages from free_area_init_nodes() for "memmap=3D1G!7G" case > > > > [ 0.000000] Early memory node ranges > > > > [ 0.000000] node 0: [mem 0x0000000000001000-0x000000000009= efff] > > > > [ 0.000000] node 0: [mem 0x0000000000100000-0x00000000bffd= 6fff] > > > > [ 0.000000] node 0: [mem 0x0000000100000000-0x000000013fff= ffff] > > > > [ 0.000000] node 1: [mem 0x0000000140000000-0x00000001bfff= ffff] > > > > [ 0.000000] node 1: [mem 0x0000000200000000-0x000000023fff= ffff] > > > >=20 > > > >=20 > > > > I think that in order to fix this, we need some conditions and/or p= rechecks > > > > for memblock layout, does it make sense? Or any other better approa= ches? > >=20 > All this is handled in parse_memmap_one(), so I wonder if the right to do= would be that in > case we detect that an user-specified map falls in an usable map, we just= back off and do not insert it. >=20 > Maybe a subroutine that checks for that kind of overlapping maps before c= alling e820__range_add()? >=20 This problem seems to happen when the end address of the user-defined memma= p is equal to the end address of NUMA node (0x140000000 or 5GB in this case.) So that's the condition to be checked, I think. However we don't initialize numa at parse_memmap_one(), so we need identify right place to do this, so I'll do this next. Thanks, Naoya Horiguchi=