On Sat, 5 Dec 2003, Jonathan S. Shapiro wrote:
On Fri, 2003-12-05 at 12:34, Joshua LeVasseur wrote:
On Dec 5, 2003, at 17:21, Jonathan S. Shapiro wrote:
If a mapping is missing from the kernel's cache...
This statement is very closely related to my earlier question about mechanisms. When you use the term "mapping" above, what exactly do you mean:
- A GPT?
- Some other cache that is somehow constructed from a GPT?
Jon you seem a bit confused ;) As Gernot already mentioned GPTs are a page table implementation. Basically in L4 you have two VM data structures. A page table per address space (GPTs being one such implementation), and the mapping database which maintains the mapping tree for each physical frame of memory in the system, this is only required to implement the unmap/flush operation.
The entire mapping DB is something that I am still failing to understand adequately.
For a summary of the interaction between L4's VM API, the page tables and the mapping database have a look at Chris's u/g thesis. ftp://ftp.cse.unsw.edu.au/pub/users/disy/ug-theses/99-cls.ps.gz
Cheers, Adam
-- Adam "WeirdArms" Wiggins School of Computer Sci. & Eng. PhD Student The University of NSW Phone: +61 2 9385 7359 UNSW SYDNEY NSW 2052, Australia Fax: +61 2 9385 7942 http://www.cse.unsw.edu.au/~awiggins