Messages in this thread | | | Date | Tue, 07 Mar 2006 01:00:45 -0800 | From | Badari Pulavarty <> | Subject | Re: [PATCH 0/3] map multiple blocks in get_block() and mpage_readpages() |
| |
Nathan Scott wrote:
>On Mon, Mar 06, 2006 at 03:33:21PM +0530, Suparna Bhattacharya wrote: > >>On Fri, Feb 24, 2006 at 09:19:08AM -0800, Badari Pulavarty wrote: >> >>>I am thinking of having a "fast path" which doesn't deal with any >>>of those and "slow" path to deal with all that non-sense. >>>... >>>slow_path is going to be slow & ugly. How important is to handle >>>1k, 2k filesystems efficiently ? Should I try ? >>> >>With 64K page size that could include 4K, 8K, 16K, 32K block size filesystems >>as well, not sure how likely that would be ? >> > >A number of architectures have a pagesize greater than 4K. Most >(OK, sample size of 2) mkfs programs default to using 4K blocksizes. >So, any/all platforms not having a 4K pagesize will be disadvantaged. >Search on the definition of PAGE_SHIFT in asm-*/page.h and for all >platforms where its not defined to 12, this will hurt. >
I agree, I haven't made up my mind either on if its really worth doing.
I was hoping that it will help simple cases + it will help filesystems which use page->private for something other than buffer heads.
Thanks, Badari
>
- 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/
| |