mirror of
https://github.com/xen-project/xen.git
synced 2026-10-29 18:50:05 +01:00
Malcolm Crossley
7430a86de0
The page scrubbing is done in 128MB chunks in lockstep across all the non-SMT CPU's. This allows for the boot CPU to hold the heap_lock whilst each chunk is being scrubbed and then release the heap_lock when the CPU's are finished scrubing their individual chunk. This allows for the heap_lock to not be held continously and for pending softirqs are to be serviced periodically across the CPU's. The page scrub memory chunks are allocated to the CPU's in a NUMA aware fashion to reduce socket interconnect overhead and improve performance. Specifically in the first phase we scrub at the same time on all the NUMA nodes that have CPUs - we also weed out the SMT threads so that we only use cores (that gives a 50% boost). The second phase is for NUMA nodes that have no CPUs - for that we use the closest NUMA node's CPUs (non-SMT again) to do the job. This patch reduces the boot page scrub time on a 128GB 64 core AMD Opteron 6386 machine from 49 seconds to 3 seconds. On a IvyBridge-EX 8 socket box with 1.5TB it cuts it down from 15 minutes to 63 seconds. Signed-off-by: Malcolm Crossley <malcolm.crossley@citrix.com> Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Reviewed-by: Tim Deegan <tim@xen.org> Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>