| |
| |
| |
|
Page: 1 2 3
Comments:
<0> hmm hmmmm
<0> ive looked at the ifconfig code
<0> function pointer always manages to shake my lead of
<0> oki ill use strace
<0> its so damn frustrating
<1> hmm, it looks rather odd.
<1> 18653 socket(PF_INET, SOCK_DGRAM, IPPROTO_IP) = 4
<1> 18653 ioctl(4, SIOCSIFADDR, 0xbfffe760) = 0
<1> 18653 ioctl(4, SIOCGIFFLAGS, 0xbfffe690) = 0
<0> but i guess its because im impatient
<1> those appear to be the appropriate ones.
<0> ahhh oki
<0> yea that actually helps me a little
<1> that was from setting eth0:1 (an alias), however.
<1> I didn't know it'd have to create a socket, though...
<0> yea it does
<0> only i thought since i had to set an ip i should make it af_inet
<0> but it makes sence
<0> it the same when setting a hw addr
<1> hmm, guess I just never looked into it.
<0> mind telling me what you set it too?
<1> 1.2.3.4
<0> k
<1> there's a:
<1> 18653 socket(PF_FILE, SOCK_DGRAM, 0) = 3
<1> as the syscall right before the other mentioned socket, but it never does anything with it.
<1> it's actually relatively short, check it out.
<0> hmm
<0> in installing strace now
<1> strace -fo ifconfig.out ifconfig eth0:0 1.2.3.4
<0> its gentoo so it might take a while
<0> if i ever figure it out ill write a small howto
<1> heh
<0> its almost impossible to find good howtos on the net
<1> oh, one more
<1> 18653 ioctl(4, SIOCSIFFLAGS, 0xbfffe690) = 0
<1> that is the last call right before exit, following the other two
<0> theyre all either copies of eachother or made for old unixes or stolen fron stevens book
<0> thats not it
<1> I didn't notice, but that request is one character different
<0> the juicy one is SIOCSIFADDR
<1> SIOCSIFFLAGS vs. SIOCGIFFLAGS
<1> yeah.
<0> S= set G = get
<1> so, System I/O Control Set/Get Interface Flags I ***ume
<0> exactly
<0> i have 3 fat stevens books at home ,, maybe i should just move myself to my hive
<1> all my books are 6,000 miles away :|
<0> :-(
<1> but I feel justified in having downloaded copies of those 3, as I actually own them.
<0> hmm
<0> wanna share :-D
<1> they're around ;)
<0> i own unix network programming
<0> borrowed unix programming and tcp illu
<1> I've got APUE, UNP, and TCP 1 & 2
<1> no digital copy of APUE, though
<0> its tcp illu 2
<1> 2 is of dubious utility unless you're planning on implementing a full network stack, but it works nicely as a reference into some bizarre details of an actual working implementation
<0> people need some god damn documentation
<1> s/bizarre/esoteric/
<0> infi: wait im not sure its no 2
<0> its 1
<1> 1 is mostly userspace network dev
<0> actually ,, it could be fun implementing a net stack
<0> but one thing at a time
<1> yeah, I just don't have the cycles atm.
<0> but all this C is starting to become fun to me
<0> www programming is so tiresome next to it
<1> yeah
<1> C is great for systems-level work
<0> youre a kernel hacker/koder?
<1> but, it's not the end-all be-all :)
<1> I've been working through the exercises in SICP lately, and have expanded my horizons to a bit of Lisp (well, Scheme)
<1> very nice stuff.
<0> ahh i coded a little scheme during my BSc
<0> i liked it
<0> dunno about its practical applications though
<1> um, not so much, I've pored through the code quite a bit, and have built a couple of modules, but nothing that useful
<1> I've been digging the Scheme work, and it's introduced me to a lot of higher concepts that I'd never even considered before
<0> id like to build userspace tools sometime ,, i mean gtk like apps
<1> had to remember all of my calculus from uni to work through SICP, though ;)
<0> but it seems to cumbersome right now
<0> SICP ?
<1> http://mitpress.mit.edu/sicp/full-text/book/book.html
<1> never before have I had to work 3 hours on exercises in the very first chapter of a book :>
<0> hmm sounds challanging
<1> rewarding anyway
<0> ill bet
<0> infi: have you read some of tannenbaums books?
<0> hes has one about instruction architectures
<0> -s
<1> only the Minix one, and then only briefly
<1> it's in my queue to study more thoroughly
<0> ;-)
<1> I wrote a ghetto boot loader and rudimentary kernel about 10 years ago (in TASM! :), but haven't looked back, recently my interest has waxed again in that direction
<0> nice
<1> Minix is pretty small and simple, and with full annotation in the form of a book, it does make a good reference for OS theory
<0> does it run on anything remotely new hardware ?
<0> annotation is good
<1> yeah, Minix 3 (I just grabbed it a few weeks ago and played with it under qemu) actually has some pretty modern features
<2> minix runs on anything :)
<2> even a 286 :)
<0> sometimes it seems at though the hardcore kernel hacker forget that their code might have to be read by other people at some time
<1> yeah.. which is interesting :)
<0> gb: i mean things lige nic,s etc
<1> cohonen: the linux source is actually a fscking mess :)
<0> infi: yea im beginning to understand that
<0> ive been roaming around in the netstack for some weeks
<1> yeah, minix has network support, etc. Minix 3 is a fairly full featured OS now
<2> infi: only the bits I wrote, everything else is neat by comparison :)
<1> gb, tee hee
<1> crazy aussies with your drop bears and your vegemite
<0> gb: hehe ,, what part did you write? or was it a joke
<2> cohonen: if I told you, I'd have too sh00t you
<0> ;-)
<0> gb: there are worse things than death,,, AND,, i can do all of them
<2> I have enough problems with lemmings now with ppl thinking im a bot
<2> imagine if they thought they could h***le me for help
<0> mnn
<0> makes sence
<0> infi: its self inforcing,, if the kernel was better documented + more structured more people would be able to help improve it futher
<1> well, the bazaar methodology and loose confederation contribute to it. a balance of both that, but with a centralized steering group is more ideal I think.
<2> cohonen: not always, if sumone spends the time writing docs instead of making certain there code functioned then nothing would get done code wise & we'd end up with a heap of documentation about a ****ty kernel
<1> FreeBSD has a nicer development model. I still like Linux better though, from a usability/app support standpoint.
<2> there's a very fine line
<1> indeed
<2> anyone can document after reading code, not everyone can cut the code
<0> gb: yea i KNOW thats the argument
<0> ofcause working fast code is the number one
<0> but thats no excuse for not documenting code
<2> the other prob with linux is the code changes that fast that whatever dics that are written, are soon made obsolete by faster cleaner code
<2> docs rather
<0> well its doesnt have to be a whole fairytale
<2> c is self documenting :)
<0> and besides things like api should bedocumented so fx driver developers actually use whats available
<2> I find in terms of debugging & readability that something as simple as carriage returns & indentation can make the difference between a complete mess & readable code
<0> true
<0> but the basic admin tools in linux simply need to work
<0> i think bsd has an advantage there
<2> thats without comments, comments as well is just the cherry on top
<0> mii-tool and ethtool is only semi-working
<2> caffiene time
<2> gday seabreeze
<0> ahahahahha
<0> 'it worked
<0> weeee
Return to
#kernel
or
Go to some related
logs:
lexmark x7170 ubuntu
#oe
Makefile Makefile.shared jfduke3d
proftpd reallocating sreaddir
#python
gimp uc-logic
#math
#math
uf8 latin1
Error in JAR file format. zip-style comment + linux
|
|