Messages in this thread | | | From | (Alan Cox) | Subject | Re: 2.1.76, nfs client, and memory fragmentation | Date | Fri, 2 Jan 1998 00:37:11 +0000 (GMT) |
| |
> > No, no reverse page tables. The data structures to keep them up-to-date > > worries me (it would essentially involve doubling all the page tables). > > One word: Eh? > > Based on what I did with Nachos it would take only one page table entry * > per page of physical memory; (alternately, the order-0 free bitmap [cw]ould
How do you handle shared pages - you need more than one pointer in this case.
> Right now I see it as a cheap and elegant solution to the fragmentation > problem which involves low-coefficient O(1) work to maintain and one kbyte > of storage per megabyte of physical memory. My 32M machine can certainly > trade 32k it can't use when it needs it, for 16k (and larger) fragments > on demand...
Put it this way - Im willing to drop 32K of memory for the right results and test the patch if you do it. To start with I can get that 32K back simply by putting Andy's select/poll patches into my kernel and non inlining the copy_to/from_user which btw my benchmarks say speeds up the machine
Alan
| |