linux-kernel - [PATCH 0/9] VMware balloon: Large page ballooning and VMCI support

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

Message-Id: <1429032576-2656-1-git-send-email-moltmann@vmware.com>
Date:	Tue, 14 Apr 2015 10:29:26 -0700
From:	"Philip P. Moltmann" <moltmann@...are.com>
To:	gregkh@...uxfoundation.org
Cc:	linux-kernel@...r.kernel.org, xdeguillard@...are.com,
	akpm@...ux-foundation.org, pv-drivers@...are.com
Subject: [PATCH 0/9] VMware balloon: Large page ballooning and VMCI support

The VMware balloon driver optimizes the memory reclamation when Linux
runs in a VM of a VMware hypervisor (such as ESXi, VMware Workstation,
or VMware Fusion). The hypervisor signals the balloon driver how much
memory to reclaim. The balloon driver allocates this amount of memory
via regular Linux allocation mechanisms. As the memory appears to be
used by the driver linux does not touch it and the hypervisor can unmap
it and give the memory to a different VM. This is not different from
any other balloon driver.

This series of patches improves the balloon drivers performance,
latency, and reduces the overhead to reclaim memory. All of this is done
while staying backwards compatible with all ever shipping versions of
VMware's hypervisors.

There are three main improvements:
- The balloon driver and hypervisor communicate about the capabilities
  of each other during balloon load.
- Instead of inflating the memory a single 4 KB page at a time, the
  balloon allocates a whole list of pages and hands it off to the
  hypervisor at once.
- We can now allocate 2 MB and 4 KB pages which further reduces the
  overhead and increases the performance
- If there is an demand to reclaim memory the hypervisor signals the
  balloon driver via VMCI that there is something to do. Before the
  balloon driver polled ever second.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/