[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v1] vhost-user: fix not send all hugepage files
|
From: |
Paolo Bonzini |
|
Subject: |
Re: [Qemu-devel] [PATCH v1] vhost-user: fix not send all hugepage files to vhost-user |
|
Date: |
Wed, 04 Feb 2015 15:02:19 +0100 |
|
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.4.0 |
On 04/02/2015 14:52, Michael S. Tsirkin wrote:
> I'm not sure: does memory core ever give us two adjacent
> RAM segments that we *can* merge?
I don't think so. Memory core merges ranges already if the following holds:
- same memory region
- same dirty logging mode
- same readonly behavior (ignore reads, trap reads as MMIO, read/write)
- r2 end address is the same as r1 start address
- r2's offset in the memory region is equal to r1's plus the size of r1
Paolo
> If yes it would trigger, and extra memory slots slow down lookups
> linearly so they aren't free.