Messages in this thread | | | Date | Tue, 13 Jul 1999 22:28:00 -0700 | From | merblich <> | Subject | Re: [RFC] - Some notions that I would like comments on |
| |
> I reckon a spare CPU could happily > > - pre-zero some pages, preferably spread over the range of page colours. > - move pages around to make space for multi-page allocations. > - checksum pages to find duplicates and merge them in the background
I'm just a little lost. Iff the hardware/mmu/page table entry supports multiple page sizes, then why don't you just ...
* Identify the amount of physical memory * Read a config file to determine a non-default number of larger than mininum page sizes, * Else allocate sections of physical memory to support large page sizes at initialization time. Each page size should have the same logic behind the request of the page, replacement of the page, etc. * Even very large physical memory's could have multiple freelists for even the same size of page, * Thus the spare CPU could age the pages of a particular freelist, * In addition, a special type of page could belong to a certain type of freelist that once allocated would not be freed because of its usage. This would in effect lock down the page and eliminate it and other pages of its type from even being search/aged.
* If one looks at memory on a normal functioning system after some period of time, the physical memory is normally filled with many small holes that would prevent the allocation of large pages, even if the system can support it.
And yes, I am "playing" with this multiple page size support and multiple freelist idea on one of my systems. The only real problem that I have is that the freelists are currently aging at different rates. I am slowly looking at this... And lastly, no I have not given this code to anyone..
Mitchell Erblich > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.rutgers.edu > Please read the FAQ at http://www.tux.org/lkml/
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |