On Sun, 8 Feb 1998 12:56:02 -0500 (EST), chip@atlantic.net said:
>> As I noticed some people are trying to make the VFS layer 64-bit
>> clean, I announce the filesystem here, so they have something to
>> experiment with.
> Well, some people may have 64-bit-clean as their goal, but I'm
> limiting my horizon to 41- or 42-bit-clean (32-bit block numbers, 512
> byte or 1K blocks).
Make it 64-bit. Things like NFS-V3 require it, and there's no point
in reworking a lot of the VFS code if we're still going to impose
artificial limits in it. Eventually we'll want true 64-bit files for
several reasons --- for example, it's a usefully large address space
for sparse data structures. That would work wonderfully well with
btree mapping pointers in ext2, for example.
> It seems to me that 2048G or 4096G is enough for now, and if I try
> to go bigger, then the "struct page" has to get bigger, and that's
> a performance hit.
Not much of a hit, really, and anyway, we just shrank the struct page
in 2.1.79 (the swap cache now reuses the page cache mechanism), so
there's space available to compensate. :)
> Or am I guilty of chasing a false economy?
Probably. We may as well do it right and stop people moaning!
Cheers,
Stephen.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu