linux-kernel - Re: [PATCH v3 18/19] kselftests: cgroup: add kernel memory accounting tests

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20200527204522.GE396660@carbon.dhcp.thefacebook.com>
Date:   Wed, 27 May 2020 13:45:22 -0700
From:   Roman Gushchin <guro@...com>
To:     Vlastimil Babka <vbabka@...e.cz>
CC:     Andrew Morton <akpm@...ux-foundation.org>,
        Johannes Weiner <hannes@...xchg.org>,
        Michal Hocko <mhocko@...nel.org>, <linux-mm@...ck.org>,
        <kernel-team@...com>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v3 18/19] kselftests: cgroup: add kernel memory
 accounting tests

On Wed, May 27, 2020 at 07:00:30PM +0200, Vlastimil Babka wrote:
> On 5/26/20 5:45 PM, Roman Gushchin wrote:
> > On Tue, May 26, 2020 at 05:24:46PM +0200, Vlastimil Babka wrote:
> >> 1 << 20 ?
> >> 
> >> Anyway I was getting this:
> >> not ok 1 test_kmem_basic
> >> ok 2 test_kmem_memcg_deletion
> >> ok 3 test_kmem_proc_kpagecgroup
> >> not ok 4 test_kmem_kernel_stacks
> >> ok 5 test_kmem_dead_cgroups
> >> 
> >> Adding some debugging into kmem_basic I found I get memory.stat == 0 at this
> >> point thus it fails the fixed test (otherwise it was failing the <= 0 test after
> >> writing to memory.high). But it's just a VM spinned by virtme which has a very
> >> simple init, so perhaps things are not as initialized as expected.
> > 
> > Hm, it's strange, do you have any values in memory.stat::slab for any cgroups?
> > Or can you send me your config (and kvm setup), I'll take a look.
> 
> Config is attached, KVM setup is just running virtme [1] on the git gree where I
> compiled the kernel:
> 
> virtme-run --mods=auto --kdir /path/to/linux.git/

Thanks!

So, test_kmem_kernel_stacks fails because there is not enough memory in your setup:
it tries to spawn 1000 threads, and virtme-run sets the total memory in 100M,
which is not enough. Adding --memory 4G option makes the test pass.

> 
> So there's only the cgroup the test creates, and the memry.stat::slab is zero
> after alloc_dcache(). Although I can see by strace that it does all those
> stat()'s. Also /proc/slabinfo also doesn't seem to be increasing number of
> dentries. Could be that because root fs is 9p in virtme?

The basic test seems to fail because of 9p. It also fails on the unpatched
kernel, so it's definitely not a regression.

So, maybe it's better to allocate kernel memory in the test using a different
method? I'm open for any ideas here.

Thanks!