linux-kernel - Re: [PATCH v2 4/7] mm/sparse.c: only use subsection map in VMEMMAP case

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: <20200226035336.GF24216@MiWiFi-R3L-srv>
Date:   Wed, 26 Feb 2020 11:53:36 +0800
From:   Baoquan He <bhe@...hat.com>
To:     Michal Hocko <mhocko@...nel.org>
Cc:     linux-kernel@...r.kernel.org, linux-mm@...ck.org,
        akpm@...ux-foundation.org, richardw.yang@...ux.intel.com,
        david@...hat.com, osalvador@...e.de, dan.j.williams@...el.com,
        rppt@...ux.ibm.com, robin.murphy@....com
Subject: Re: [PATCH v2 4/7] mm/sparse.c: only use subsection map in VMEMMAP
 case

On 02/25/20 at 10:57am, Michal Hocko wrote:
> On Thu 20-02-20 12:33:13, Baoquan He wrote:
> > Currently, subsection map is used when SPARSEMEM is enabled, including
> > VMEMMAP case and !VMEMMAP case. However, subsection hotplug is not
> > supported at all in SPARSEMEM|!VMEMMAP case, subsection map is unnecessary
> > and misleading. Let's adjust code to only allow subsection map being
> > used in VMEMMAP case.
> 
> This really needs more explanation I believe. What exactly happens if
> somebody tries to hotremove a part of the section with !VMEMMAP? I can
> see that clear_subsection_map returns 0 but that is not an error code.
> Besides that section_deactivate doesn't propagate the error upwards.
> /me stares into the code
> 
> OK, I can see it now. It is relying on check_pfn_span to use the proper
> subsection granularity. This really begs for a comment in the code
> somewhere.

Yes, check_pfn_span() guards it. People have no way to hot add/remove
on non-section aligned block with !VMEMMAP.

I have added extra comment to above section_activate() to note this,
please check patch 5/7. Let me see how to add words to reflect the
check_pfn_span() guard thing.