Fwd: Re: zlib-1.1.4 out - security fix
Bill Maltby LFS Related
lfsbill at wlmlx1.wlmcs.com
Tue Mar 12 13:48:24 PST 2002
I'm glad you brought it up - I was going to let it slide figuring
that the instance was a restriceted/special application installation.
But you are 100% right. _And_, the more general purpose a machine,
they greater the accrued benefits. Not only from the obvious causes,
but consider the effect allso on VM activities. TLB ops are faster,
fewer virtual address ranges need be maintained, etc.
And you already mentioned faster load times. And Linux uses text
files as backing store, so an unexpected benefit is even further
reduced swapping because it also will not swap shared text, it
just discards it, remembering where it got it from, when it needs
the memory. Then it reloads from the original text if it needs it
again. Only heavily swapping machines, esp where swap is a separate
device, beau coup (?) gains.
billm at wlmcs.com
On Tue, 12 Mar 2002, Don Smith wrote:
> "Ian Molton" <spyro at armlinux.org> wrote in message
> news:20020312192959.591fe789.spyro at armlinux.org...
> > On a sunny Tue, 12 Mar 2002 14:02:54 -0500 Gerard Beekmans gathered a
> > of electrons and etched in their motions the following immortal words:
> > > The short answer to WHY is performance, shared libs are slower. SSH
> > > gzip is two of the pretty few programs that I really WANT to have as
> > > optimized as possible.
> > Is the performance hit really that big?
> Um, shared libs are slower? I doubt that. What about shared libs makes
> them slower? If you are talking about load times, shared libs are
> actually faster because you only reload the "static" (sorry about the
> bad terminology) part the second through nth time you run it. If you are
> talking about direct vs. indirect references, that all done in local
> processor registers and is very optimized, usually 1 cycle in modern
> pipelined cpu architectures.
> Having all static greatly increases memory usage and, if you have
> limited memory resources, you start paging. Paging is reaaalllllllly
> Unsubscribe: send email to listar at linuxfromscratch.org
> and put 'unsubscribe lfs-security' in the subject header of the message
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-security' in the subject header of the message
More information about the lfs-security