Ed Greshko wrote:
John Summerfield wrote:
bloated kernels increase the probability of exploits
If someone gets close enough to use an otherwise-unused kernel module,
the battle's already gone to the intruder.
Funny you should mention that.... I think some folks have no idea what they
are talking about when they use the word "bloat". At least they don't
define what they are calling "bloat".
For fun...and since it was raining here part of the day....
I did a default "all" install of the current stable slackware and an "all"
install of the latest F8-beta on the exact same hardware. I then started
them both in run level 3.
As soon as the systems were up I checked top and lsmod. Now, this wasn't
entirely one-to-one since on my F8 install I did use LVM while I didn't on
the slackware. Also, slackware is at 2.6.21.5 while F8 is 2.6.23.0.
That should advantage Slack.
F8 loaded a total of 60 modules
Slackware loaded 38 modules
The /lib/modules of F8 was 57MB while slackware is 52MB.
Now, the really interesting thing was that top showed F8 to be using 147.9MB
out of 512MB while slackware was consuming 179.4MB.
I find that interesting.....
You should also include the kernel size. I was amazed when i first used
debian to find the bare kernel didn't support my IDE hardware: the IDE
drivers were modules.
If you still have both, find how many modules each has under
/lib/modules. And, if you still haven't got a life, see if you can pick
the differences: one might, for example, have madwifi even though the
HAL is binary-only.
Part of the requirement of any commercial distro is that it must support
all likely hardware. That's constrained, in part, by attitudes to lack
of source: Debian hacks stuff out of the kernel because some is not
DFSG-free.
--
Cheers
John
-- spambait
1aaaaaaa@xxxxxxxxxxxxxxxx Z1aaaaaaa@xxxxxxxxxxxxxxxx
Please do not reply off-list
