netdev - [QUESTION] ehea memory notifier

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>] [day] [month] [year] [list]

Message-ID: <939f982a-a720-f885-63e3-3ed2a5a2f93f@redhat.com>
Date:   Thu, 17 May 2018 15:38:30 +0200
From:   David Hildenbrand <david@...hat.com>
To:     Douglas Miller <dougmill@...ux.vnet.ibm.com>
Cc:     netdev@...r.kernel.org, "David S. Miller" <davem@...emloft.net>
Subject: [QUESTION] ehea memory notifier

Hi,

looking at the ehea_mem_notifier() and called functions, I wonder if
it can tolerate addresses and sizes that are not aligned to EHEA_SECTSIZE.

Looks like for MEM_ONLINE/MEM_GOING_OFFLINE ehea_update_busmap() will do
nothing in case we don't span at least one EHEA_SECTSIZE.

This implies, that for onlined/offlined memory with unaligned
address/size, we won't mark the usmap entry valid.

start_section = (pfn * PAGE_SIZE) / EHEA_SECTSIZE;
end_section = start_section + ((nr_pages * PAGE_SIZE) / EHEA_SECTSIZE)
...
for (i = start_section; i < end_section; i++) {
...
}

The other way around, if we onlined e.g. 16GB and marked the entry
valid, we won't mark it invalid if e.g. offlining 8GB of that.

Is this the right thing to do? Especially
- is "valid of partially online sections" bad?
- is "invalid of partially online sections" bad?

(working on paravirtualized memory devices that will be able to
online/offline things that would not be possible on real HW and checking
all memory notifiers)

Thanks!

-- 

Thanks,

David / dhildenb