[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: general question about lazy loading of shared libraries
- From: John Ellson <ellson research att com>
- To: arjanv redhat com, Development discussions related to Fedora Core <fedora-devel-list redhat com>
- Cc:
- Subject: Re: general question about lazy loading of shared libraries
- Date: Fri, 19 Nov 2004 07:07:12 -0500
Arjan van de Ven wrote:
On Thu, 2004-11-18 at 20:06 -0500, John Ellson wrote:
I don't know an easy way to tell exactly what pages of the libraries
you are using are touched and thus read off disk at start up... and it's
not even a well defined question - you can't tell if a page was
read because of your app or because of another app.
but you can tell which pages are in memory at least. If people want I
can code a small app for this
Don't go out of your way, but if you have some snippets of code I can
look at it would be appreciated.
I'm working on Graphviz graph layout tools that support multiple
renderers.
that kind of sounds like a plugin situation, for which dlopen is quite
suitable.
Well, yes, thats the direction I was headed in, but I'm wondering if the
performance payoff is going
to be worth the effort of hand coding the dlopen plugin support?
If lazy loading, or deferred mmapping into virtual memory, is
implemented well then
why bother with plugins? (Separate compilation is marginally
attractive, but performance is king.)
Does the deferred paging depend on prelinking having been performed?
no that happens always (even with the main binary btw); the difference
prelink makes is that without prelink you need to touch (and thus load)
more pages even if you don't use the library, for relocations.
So if I understand, deferred paging doesn't depend on prelinking, but if
I am going to try to measure
avoided-page-loads then I should make sure that prelinking has been done?
John
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]