Re: [PATCH] Scalable page cache - take two

From: William Lee Irwin III (wli@holomorphy.com)
Date: Sun Dec 23 2001 - 19:28:39 EST


On Mon, Dec 24, 2001 at 12:15:33AM +0200, Momchil Velikov wrote:
> This is the second mutation of the scalable page cache patch. It:
>
> - removes the global page cache hash table and the associated lock
>
> - implements the page cache as a per mapping radix tree. This
> increases the size of ``struct inode'' by 8 bytes (in 32bit
> ports). The branch factor is a compile time constant
> ``RAD_MAP_SHIFT'' (currently 7, 128 pointers)
>
> - decreases the size of ``struct page'' by removing
> ``next_hash'' and ``pprev_hash' fields as they are no longer
> needed for the page cache. NOTE that this curently breaks
> sparc64 and arm ports. The sparc64 port is trivialy fixable.
>
> The patch is stable on UP (survives make -j6) and does not have
> noticeable performance impact (at least for the kernel compile
> benchmark) in either direction.

Beautiful programming!

Cheers,
Bill
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sun Dec 23 2001 - 21:00:31 EST