logs archiveBotHelp.net / Freenode / #1gam / 2015 / September / 11 / 2
Llamageddon
Sounds like a kinda dumb thing to learn
Hirsch
why should it?
its a logical address right?
Llamageddon
E.g. I don't see how this knowledge would enrich your ability as a programmer
What do you mean by logical address?
Hirsch
logical address is the term intel uses for virtual address
Llamageddon
Ah
Hirsch
if i remember correctly
Llamageddon
Yeah, but you still have a memory space
That is probably randomized
Hirsch
and then thy use virtual address for something else - totally confusing those intel guys
Llamageddon
And even if not, unlikely to have anything specific at some random specific address
Hirsch
Llamageddon: i am currently at chapter 3/5 alignment and caches
:D
Llamageddon
Sounds fun
Hirsch: Will you get to create your own VM?
Hirsch
then i have to learn pipelining (+speed up), multi threading, simd, CUDA, memory management (paging, segemntation), optimization with TLB
and then i have time for my VM
Llamageddon: i hope so
sandbox*
Llamageddon
Fuck, that's a lot
Hirsch
yeah
Llamageddon
I don't know much about pipelining, so tell me about that later
Hirsch
Llamageddon: after that i have to learn maths and electric engineering :(
Llamageddon
:(
Hirsch
and then i have time for sandbox stuff
Llamageddon
Sandbox, as in?
Hirsch
virtualization
Llamageddon
Mhm
x86?
As in, not a VM for some other bytecode
Hirsch
./my_program
sandbox --no-network --root="/home/me/profile_warez" ./my_program
:D
Llamageddon
How are you going to block the network?
Blocking syscalls?
Hirsch
trapping syscalls and modifying them
with ptrace
Llamageddon
Trapping?
Wouldn't it be better to just replace them with something else at load time?
Hirsch
i think its easy to intercept those syscalls and then simply appending stuff to make "/" lay anywhere else
Llamageddon: man ptrace ;)
Llamageddon
I always thought that it was for observing the syscalls only
Hirsch
you can trap them so that they are not passed to the os
Llamageddon
Isn't trapping an OS mechanism as well?
Hirsch
yes
Llamageddon
Sooo
Hirsch
a segfault is a trap for example
Llamageddon
Your VM is going to make multiple round trips from kernel mode?
Hirsch
:D
Llamageddon
E.g.
Program makes syscall, kernel traps it, tells your VM, your VM does stuff, possibly syscalls it again now, blah blah
I dunno
Hirsch
i have not thought of it much yet
Llamageddon
It sounds like a sandbox that it'd be possible to break out of
Say, by specifying a new trap handler
Or whatever
Hirsch
fopen("../../") -> vm -> fopen("/home/haha/" + "../../") // DANGEROUS!
Llamageddon
And that is what chroot is for
Hirsch
Llamageddon: i hate sandy bridge
its internally RISC :(
Llamageddon: http://hirsch.blinkenshell.org/sandy.jpg
Llamageddon
Hirsch: Aren't all x86-64 CPUs internally RISC?
I thought that was one of the main improvements over x86
Hirsch
CISC
also the commands have not the same length
which makes pipelining impossible
Llamageddon
Eh
Hirsch
so they predecode it to RISC first afaik
but i am currently learning about that i might now more tomorrow :D
Llamageddon
I'm pretty sure that I read that all x86-64 decodes x64 to internal RISC representation
http://stackoverflow.com/questions/5806589/why-does-intel-hide-internal-risc-core-in-their-processors
Also http://stackoverflow.com/questions/13071221/is-x86-risc-or-cisc
Hirsch: And I'm pretty sure that AMD does the same thing, since RISC is just plain better
And this kind of decoding is a smart approach to implementing(especially CISC) processors, anyway
Hirsch
ye
Llamageddon
Hirsch: So why do you hate Sandy Bridge specifically? @_@
Anyway, time for me to log off
Bye bye
« prev 1 2