| |
| |
| |
|
Page: 1 2
Comments:
<0> hmm - typo
<1> :)
<1> I propably end up keeping refcounts of some kind, unless I invent better way.
<0> all: mind to help me a bit by populating the database at http://r0nnyger.mine.nu:3300/database/ it would be great to have a database with 30-40 snippets for the start
<0> just found the first stupid bug ;P
<2> r0nny: What, no ASM? And what's "PASAL"?
<0> nyef: its not yet complete - need some reasonably sane testing data
<0> and PASAL is a typo
<3> you don't say
<0> fixed the typo
<4> how goes it folks
<5> hey geist !
<5> :)
<5> long time no seen
<5> :D
<4> heya. I'm stuck in japan
<4> flight to hong kong got redirected, so I'm at the airport in Nagoya
<2> Pick us up some Pocky while you're there?
<5> hmm ... my father was there in last december
<5> I've developed some interesting stuff though :)
<5> hehehe
<5> I didn't come here for some months
<5> has anyone seen Robet ?
<5> *Robert
<4> nyef: there is tons of pocky in the store downstairs
<6> heeh too long
<5> the guy ignored me without a reason
<5> hehe
<5> I just joined the channel and appeared:
<5> -Mathis- alphakiller!*@* You are being ignored from now on!
<2> alphakiller: Umm... Mathis has been doing that for the past several days to everybody.
<5> hmm ..
<5> he may be on drugs :P
<2> Possibly not even really /ignoring people, maybe just sending the /notify.
<7> alphakiller: he's an ***hat, ignore him =)
<5> hmm ... he used to be cool with me, sometime ago
<7> *shock*
<7> (confusion, disarray!)
<5> heheh
<5> hi :D
<5> silence :D
<8> sweet silence
<8> moimoi
<5> hmm, still joining #uuu mur ?
<0> yo all
<9> hi
<10> BONG
<10> what's a good executable format to start with? something either with a simple, portable library suitable for embedding in near-kernel mode code, or that's simple enough for me to write a loader for... i'd prefer something i can either compile directly to, or with a way of converting from elf or pe...
<11> ELF
<10> any urls that are particularly helpful? i'm googling about the format atm...
<11> the specification
<10> ok thx
<2> Wouldn't statically-linked a.out be easier to work with, at least from the perspective of the kernel?
<2> (I realize that the tool support has been disappearing over the past few years, though.)
<11> statically linked ELF is easy enough too
<2> I suppose that I should be glad that my design doesn't call for using executables.
<10> ive been reading up on L4 and want to do a rather conventional [in terms of execution & low level stuff] framework around it using D
<11> D??
<11> heh this language really seems to spread
<10> why shouldn't it?
<11> if the language only had better tool-support and would already be stable...
<10> as far as conventional [compiled, imperative, OO] languages go it's awesome
<11> Kirk wasn't admiral for a very long time...
<11> erm, does endianness also affect the bit order?
<10> i don't think so
<10> i mean what is bit order? the smallest addressable unit is a byte...
<10> if it did, it would make endian changing a real bitch
<10> in fact im fairly certain it's just byte order
<12> it's byte order...
<10> ive had to write endian conversion stuff before, and it was simple, but it might not of been even remotely right :)
<11> some guy in #SDL just claimed it would change bit order too
<11> and yes, bit order doesn't make really sense
<12> i suppose every document on the internet about endianness is wrong then.
<11> who knows, #SDL guys are as smart as libsdl heh
<11> ouch that was evil
<10> heh
<10> i dont get ELF
<11> llama32: what's the problem?
<10> just dont understand much at all
<10> hmm
<10> im sorta getting it now
<10> but im trying to work out how user security would interoperate with it... i mean you'd want only one copy of each unique library in ram at one time... but in a system where files often [probably more often than not] won't reside on disk, there arent any neat device:inode pairs i can check to see if user A's library is the same as user B's...
<8> btw, if anyone understands slavic languages; ) http://www.dusky.sk/pics/2006-05/2206_asus.gif
<11> srat asus?
<13> ;))
<13> jebat asus.
<11> I don't know czech
<11> but a company which has such rules must be sceweed
<2> llama32: You don't have any unique identifier for a file?
<8> well well
<8> ii think that's pretty self-explanatory if you know the words :D
<10> nyef: i hope to have most programs downloaded from HTTP or other means on demand [and cached of course]... so there's no real way to get a unique ID for a file... also /lib/libtaco.so for Bob might be different to /lib/libtaco.so for Eliza
<2> If the files reside entirely in core, do you think you can get away with an md5sum?
<10> yeah that's what i was thinking, but still there might be some sort of less disk-intensive way...
<11> llama32: in what cases will it really be the "same" file??
<10> i guess md5sums stored along with the file could be trusted in some circumstances, but a lot of the time you'd have to check it for security
<10> well if the user doesn't have their own version of the file
<2> So now you're worried about the security of your on-disk cache?
<11> llama32: you maybe should consider the file equal when you know by your caching mechanism that it _must_ be the same file
<2> Note also that, as soon as you've determined that two files are the "same", you can blow the data for one of them out of the cache and link their metadata.
<10> oh yeah i didn't even think about caching it safely on disk lol... i was just thinking of having a checksum at the start of the actual file [in which case a user could convince the OS that a malicious one is the same one as root should use]
<10> undesktop: but the HTTP code is not in the kernel... in fact where the file comes from is arbitrary... it could be local, could be directly from a CD, could be from internet/HTTP, could be from a nearby file server, could be from a running compiler's memory
<14> hello
<11> llama32: then the source is untrusted... maybe you should compare the files byte per byte (using the md5 sum as an optimization)
<14> what's up?
<10> yeah i guess
<10> goodnight all
<15> Someone know where is the heap on linux?
<16> wtf?
<11> WindowsUninstall: doesn't it normally just extend the data section
<15> undesktop, I need to know how to get heap start address using linux syscalls
<11> also try /proc/self/maps
<16> I'd say he has to ask the libc not the kernel
<2> Well -damn-.
<2> WindowsUninstall: Thank you.
<2> That's such an -obvious- solution to one of my problems.
<15> nyef, Thank you :D You have helped me with FDC
<17> hi
<18> hi
<14> hi
<15> hi marco_g
<11> haha it's funny how often UNIX tools explode in your face
<14> with great power comes great instability
<11> false
<14> uh, that's thermodynamics
<11> I can't see any power in that "tga2ppm filename" dumps the converted file to stdout
<11> although it had a file as input, not stdin
<14> that's a pretty common UNIXism. redirect it to a file if you want, or pipe it into some other program
<14> hey shackan
<19> "with great power comes great instability" - #osdev on thermodynamics
<19> heh'
<16> ugh. what an ugly day. guys, computers are crap!
<11> of course
<11> but they give us jobs
<11> because of that
<19> we have known this for many years now
<19> please do not tell everyone else
<16> maybe I should order some chinese that do the coding for me.
<19> we like our jobs, and prefer that we keep them
<19> now, the bulk of the GNU/Linux kernel itself is mostly drivers, right?
<16> yes.
<19> the core components of any kernel (memory allocation, paging, etc?) were at a usable stage quite a while ago, yes?
<14> more or less
<16> apart from the locking, I'd agree
<19> thread locking?
<16> ak kinds of. it became quite fine grained, lock-wise. and I guess that's continuing.
<16> s/ak/all/
<14> depends on the arch
Return to
#osdev
or
Go to some related
logs:
#math
UPS- BGLAD
#gentoo
#debian
meaning of chupa meu pau
#linux
$ENV in php
#python
#ubuntu
0x3A28213A
|
|
