linux-kernel - Re: [PATCH RFC v1] mm: is_mem_section

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

Message-ID: <a5f0bd8d-de5e-9f27-5c94-7746a3d20a95@redhat.com>
Date:   Mon, 20 Jan 2020 10:14:44 +0100
From:   David Hildenbrand <david@...hat.com>
To:     Michal Hocko <mhocko@...nel.org>,
        Dan Williams <dan.j.williams@...el.com>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Linux MM <linux-mm@...ck.org>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Paul Mackerras <paulus@...ba.org>,
        Michael Ellerman <mpe@...erman.id.au>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "Rafael J. Wysocki" <rafael@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Leonardo Bras <leonardo@...ux.ibm.com>,
        Nathan Lynch <nathanl@...ux.ibm.com>,
        Allison Randal <allison@...utok.net>,
        Nathan Fontenot <nfont@...ux.vnet.ibm.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Stephen Rothwell <sfr@...b.auug.org.au>,
        Anshuman Khandual <anshuman.khandual@....com>,
        lantianyu1986@...il.com,
        linuxppc-dev <linuxppc-dev@...ts.ozlabs.org>
Subject: Re: [PATCH RFC v1] mm: is_mem_section_removable() overhaul

On 20.01.20 08:48, Michal Hocko wrote:
> On Fri 17-01-20 08:57:51, Dan Williams wrote:
> [...]
>> Unless the user is willing to hold the device_hotplug_lock over the
>> evaluation then the result is unreliable.
> 
> Do we want to hold the device_hotplug_lock from this user readable file
> in the first place? My book says that this just waits to become a
> problem.

It was the "big hammer" solution for this RFC.

I think we could do with a try_lock() on the device_lock() paired with a
device->removed flag. The latter is helpful for properly catching zombie
devices on the onlining/offlining path either way (and on my todo list).

> 
> Really, the interface is flawed and should have never been merged in the
> first place. We cannot simply remove it altogether I am afraid so let's
> at least remove the bogus code and pretend that the world is a better
> place where everything is removable except the reality sucks...

As I expressed already, the interface works as designed/documented and
has been used like that for years. I tend to agree that it never should
have been merged like that.

We have (at least) two places that are racy (with concurrent memory
hotplug):

1. /sys/.../memoryX/removable
- a) make it always return yes and make the interface useless
- b) add proper locking and keep it running as is (e.g., so David can
     identify offlineable memory blocks :) ).

2. /sys/.../memoryX/valid_zones
- a) always return "none" if the memory is online
- b) add proper locking and keep it running as is
- c) cache the result ("zone") when a block is onlined (e.g., in
mem->zone. If it is NULL, either mixed zones or unknown)

At least 2. already scream for a proper device_lock() locking as the
mem->state is not stable across the function call.

1a and 2a are the easiest solutions but remove all ways to identify if a
memory block could theoretically be offlined - without trying
(especially, also to identify the MOVABLE zone).

I tend to prefer 1b) and 2c), paired with proper device_lock() locking.
We don't affect existing use cases but are able to simplify the code +
fix the races.

What's your opinion? Any alternatives?

-- 
Thanks,

David / dhildenb