Re: [PATCH 7/7] KVM: MMU: don't walk every parent pages while mark unsync

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
From: Xiao Guangrong
Date: Friday, June 11, 2010 - 6:43 am

Xiao Guangrong wrote:

Hi Avi, Marcelo,

About two months ago, i have sent this optimization, it just a little performance
improved (~0.5%) at that time, now, we allow more page become unsync, this path is
called higher frequency, during my test, kernelbench show the performance improved
~2%, so, i think we may apply it now :-)
--
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
[PATCH 2/7] KVM: MMU: cleanup for dirty page judgment, Xiao Guangrong, (Fri Jun 11, 6:29 am)
[PATCH 5/7] KVM: MMU: cleanup for __mmu_unsync_walk(), Xiao Guangrong, (Fri Jun 11, 6:32 am)
[PATCH 6/7] KVM: MMU: clear unsync_child_bitmap completely, Xiao Guangrong, (Fri Jun 11, 6:34 am)
Re: [PATCH 7/7] KVM: MMU: don't walk every parent pages wh ..., Xiao Guangrong, (Fri Jun 11, 6:43 am)