Date: 13 Nov 1995 06:58:12 +0800 From: peter@haywire.dialix.com (Peter Wemm) To: freebsd-current@freebsd.org Subject: Re: A couple of things Message-ID: <485u64$2he$1@haywire.DIALix.COM> References: <23839.816205152@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
phk@freefall.freebsd.org (Poul-Henning Kamp) writes: >1. The slower sysinstall: we don not use the same scenario >now as in 2.0.5 that's why the exec gzip stuff is so much slower. >in 2.0.5 the MFS disk contained uncompressed binaries, relying >on kzip to compress the entire thing, not the MFS disk contains >gziped bins too. Err.. :-) There's a couple of typos in there that make it a little bit ambiguous.. So, let me get this straight... We're _now_ using a gzipped, crunched executable? If so, isn't that bad? the exec gzip routine makes a copy of the gzipped data in memory, and runs that doesn't it? ie: swap space is allocated because it can't page to the file anymore? I've not looked at the gzip exec code, but what are the odds that each executable launched from the crunched image is using up it's own amount of ram and swap space rather than sharing pages? Or does it try and be smart by associating the the ungzipped pages with the file so that when the crunched image is run, the ungzipped pages are reclaimed? Cheers, -Peter >Poul-Henning
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?485u64$2he$1>