Date:   Thu, 21 May 2020 14:44:44 +0200
From:   Michal Hocko <mhocko@...nel.org>
To:     Hugh Dickins <hughd@...gle.com>
Cc:     Naresh Kamboju <naresh.kamboju@...aro.org>,
        Chris Down <chris@...isdown.name>,
        Yafang Shao <laoar.shao@...il.com>,
        Anders Roxell <anders.roxell@...aro.org>,
        "Linux F2FS DEV, Mailing List" 
        <linux-f2fs-devel@...ts.sourceforge.net>,
        linux-ext4 <linux-ext4@...r.kernel.org>,
        linux-block <linux-block@...r.kernel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        open list <linux-kernel@...r.kernel.org>,
        Linux-Next Mailing List <linux-next@...r.kernel.org>,
        linux-mm <linux-mm@...ck.org>, Arnd Bergmann <arnd@...db.de>,
        Andreas Dilger <adilger.kernel@...ger.ca>,
        Jaegeuk Kim <jaegeuk@...nel.org>,
        Theodore Ts'o <tytso@....edu>, Chao Yu <chao@...nel.org>,
        Andrea Arcangeli <aarcange@...hat.com>,
        Matthew Wilcox <willy@...radead.org>,
        Chao Yu <yuchao0@...wei.com>, lkft-triage@...ts.linaro.org,
        Johannes Weiner <hannes@...xchg.org>,
        Roman Gushchin <guro@...com>, Cgroups <cgroups@...r.kernel.org>
Subject: Re: mm: mkfs.ext4 invoked oom-killer on i386 - pagecache_get_page

On Thu 21-05-20 05:24:27, Hugh Dickins wrote:
> On Thu, 21 May 2020, Michal Hocko wrote:
> > On Thu 21-05-20 16:11:11, Naresh Kamboju wrote:
> > > On Thu, 21 May 2020 at 15:25, Michal Hocko <mhocko@...nel.org> wrote:
> > > >
> > > > On Wed 20-05-20 20:09:06, Chris Down wrote:
> > > > > Hi Naresh,
> > > > >
> > > > > Naresh Kamboju writes:
> > > > > > As a part of investigation on this issue LKFT teammate Anders Roxell
> > > > > > git bisected the problem and found bad commit(s) which caused this problem.
> > > > > >
> > > > > > The following two patches have been reverted on next-20200519 and retested the
> > > > > > reproducible steps and confirmed the test case mkfs -t ext4 got PASS.
> > > > > > ( invoked oom-killer is gone now)
> > > > > >
> > > > > > Revert "mm, memcg: avoid stale protection values when cgroup is above
> > > > > > protection"
> > > > > >    This reverts commit 23a53e1c02006120f89383270d46cbd040a70bc6.
> > > > > >
> > > > > > Revert "mm, memcg: decouple e{low,min} state mutations from protection
> > > > > > checks"
> > > > > >    This reverts commit 7b88906ab7399b58bb088c28befe50bcce076d82.
> > > > >
> > > > > Thanks Anders and Naresh for tracking this down and reverting.
> > > > >
> > > > > I'll take a look tomorrow. I don't see anything immediately obviously wrong
> > > > > in either of those commits from a (very) cursory glance, but they should
> > > > > only be taking effect if protections are set.
> > > >
> > > > Agreed. If memory.{low,min} is not used then the patch should be
> > > > effectively a nop. Btw. do you see the problem when booting with
> > > > cgroup_disable=memory kernel command line parameter?
> > > 
> > > With extra kernel command line parameters, cgroup_disable=memory
> > > I have noticed a differ problem now.
> > > 
> > > + mkfs -t ext4 /dev/disk/by-id/ata-TOSHIBA_MG04ACA100N_Y8NRK0BPF6XF
> > > mke2fs 1.43.8 (1-Jan-2018)
> > > Creating filesystem with 244190646 4k blocks and 61054976 inodes
> > > Filesystem UUID: 3bb1a285-2cb4-44b4-b6e8-62548f3ac620
> > > Superblock backups stored on blocks:
> > > 32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,
> > > 4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968,
> > > 102400000, 214990848
> > > Allocating group tables:    0/7453                           done
> > > Writing inode tables:    0/7453                           done
> > > Creating journal (262144 blocks): [   35.502102] BUG: kernel NULL
> > > pointer dereference, address: 000000c8
> > > [   35.508372] #PF: supervisor read access in kernel mode
> > > [   35.513506] #PF: error_code(0x0000) - not-present page
> > > [   35.518638] *pde = 00000000
> > > [   35.521514] Oops: 0000 [#1] SMP
> > > [   35.524652] CPU: 0 PID: 145 Comm: kswapd0 Not tainted
> > > 5.7.0-rc6-next-20200519+ #1
> > > [   35.532121] Hardware name: Supermicro SYS-5019S-ML/X11SSH-F, BIOS
> > > 2.2 05/23/2018
> > > [   35.539507] EIP: mem_cgroup_get_nr_swap_pages+0x28/0x60
> > 
> > Could you get faddr2line for this offset?
> 
> No need for that, I can help with the "cgroup_disabled=memory" crash:
> I've been happily running with the fixup below, but haven't got to
> send it in yet (and wouldn't normally be reading mail at this time!)
> because of busy chasing a couple of other bugs (not necessarily mm);
> and maybe the fix would be better with explicit mem_cgroup_disabled()
> test, or maybe that should be where cgroup_memory_noswap is decided -
> up to Johannes.

Thanks Hugh. I can see what is the problem now. I was looking at the
Linus' tree and we have a different code there

	long nr_swap_pages = get_nr_swap_pages();

        if (!do_swap_account || !cgroup_subsys_on_dfl(memory_cgrp_subsys))
                return nr_swap_pages;

which would be impossible to crash so I was really wondering what is
going on here. But there are other changes in the mmotm which I haven't
reviewed yet. Looking at the next tree now it is a fallout from "mm:
memcontrol: prepare swap controller setup for integration".

!memcg check slightly more cryptic than an explicit mem_cgroup_disabled
but I would just leave it to Johannes as well.

> 
> ---
> 
>  mm/memcontrol.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> --- 5.7-rc6-mm1/mm/memcontrol.c	2020-05-20 12:21:56.109693740 -0700
> +++ linux/mm/memcontrol.c	2020-05-20 12:26:15.500478753 -0700
> @@ -6954,7 +6954,8 @@ long mem_cgroup_get_nr_swap_pages(struct
>  {
>  	long nr_swap_pages = get_nr_swap_pages();
>  
> -	if (cgroup_memory_noswap || !cgroup_subsys_on_dfl(memory_cgrp_subsys))
> +	if (!memcg || cgroup_memory_noswap ||
> +            !cgroup_subsys_on_dfl(memory_cgrp_subsys))
>  		return nr_swap_pages;
>  	for (; memcg != root_mem_cgroup; memcg = parent_mem_cgroup(memcg))
>  		nr_swap_pages = min_t(long, nr_swap_pages,

-- 
Michal Hocko
SUSE Labs

Powered by blists - more mailing lists