Dynamic symbol table

[Oswald Buddenhagen]

On Tue, Jul 20, 2004 at 02:50:23PM +0200, Lubos Lunak wrote:
> I think it'd be better if you tried that yourself,
>
bah. i have a different task now. :)

> How is synchronous page-by-page loading supposed to be faster than
> asynchronous madvise() ?
> 
asynchronicity implies concurrency and concurrency implies trashing.
ergo, forced serialization should improve throughput. so far the theory.
i don't know whether touching each page in sequence will make the kernel
recognize a sequential read pattern and thus make it do big reads with a
lot of read-ahead. maybe in combination with madvice it will work
better. or just read() the relevant sections (which would make it
recognize the sequential pattern for sure), but that implies big
memcpy()s, unfortunately.

-- 
Hi! I'm a .signature virus! Copy me into your ~/.signature, please!
--
Chaos, panic, and disorder - my work here is done.