Hi... I just lend a small help, hopefully it helps...
On Sun, Sep 19, 2010 at 17:56, 海峰 陈 <gg...@yahoo.com.cn> wrote: > > Dear friends, > > I'm a student and i will do some work about QEMU memory management > mechanism. However, i am a > > freshman, i need friends to give me some suggestion or send some helpfull > documentation to me. Thanks. I suggest, next time, send plain ASCII email. No HTML format, no picture embedded as background....etc, please? In this kind of technical discussion list, they are just wasting spaces. So unless it is useful to explain something (.eg diagram..), don't embed pictures. OK, about qemu memory management. Have you ever read the Qemu source code? If you haven't, please download, extract and start reading it. The very basic fundamental of Qemu memory management is that it's actually a file mmap()-ed into the guest address space. The size of the this simulated RAM is a large as the size you pass at -m parameter (by default, it's 256 MB if I remember correctly). The start of the mmap()-ed address, is considered as offset zero in this virtual RAM. Every access to this virtual RAM is like the same like actual RAM e.g there is page fault, reserved memory address and so on. It's just it's not so complicated. IIRC, there is no actual L1 or L2 cache. Probably it's due to the fact that they are simulated as mmaped file. -- regards, Mulyadi Santosa Freelance Linux trainer and consultant blog: the-hydra.blogspot.com training: mulyaditraining.blogspot.com