| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9
Comments:
<0> bush budged plan: 61 billion cuts from the poor people medicaid and elder people medicare, 77 billion to iraq war
<0> bush explains this to be "more fair, more cheap, more accessible and more flexible" sure it's very flexible if there is no thing at all
<1> mur: a bastard - isnt he?
<1> hmm - i tihnk i got myself adicted to 8bit music
<0> ronny, just regular right-wing politician
<1> anyway - the usa are a major problem for the world (as well as china)
<1> casue these bastards dont care for the climate, since its still keeping the being hazardous low
<1> mur: btw - could you take a look at my webforum and give some feedback - http://r0nnyger.mine.nu:3300/forum/
<0> it h*** very interesting aesthetics
<0> but i like it somehow
<0> the ->> arrows dont work but look squares
<0> i like the anonymouss usser colour
<0> i think it is better than registered user
<0> the blue line shoudd cover entire message i think
<0> maybe except the source code
<0> but everythign else
<0> bb code selection shoudl be last and sumbit first
<0> and the forum menu shoudl be accessible without going to forum page and selecting new menu
<0> dropdown menu or if you dont want to make tit compilicated then list below
<1> hmm - a forum quickselect
<0> forum mainmenu
<0> but i like how you have nested the forums
<1> mur: they should be collapsable
<0> they arent
<0> at least not for me
<0> with opera
<1> they are in firefox
<0> but that works better if they are not
<0> becuase it required huge work to collapse them
<0> it's much easier lik ein safari
<1> huge work ?
<1> collapse/decollapse is just a visibility toggle
<0> 4 clicks compared to 0 is big difference :)
<1> later i want it to remember how they where collapsed, so users can blend out forums they dont want
<0> also please remember to keep it web 1.0 compliant
<0> web 2.0 is not required for the page
<1> but the most important part is - keep it working nice in lynx
<0> yeah
<1> web-20 is completely added by the javascripts
<0> if you are interested, you may want to make it also work well in speech synthesis software for blind
<0> like challenge
<0> and if the css file is 404 that it will look useful
<0> mlab.uiah.fi
<1> mur: nice idea - but i have no knowdegde about that
<0> check that webpage for accesibility
<0> medialab is concerned about accessibility very much
<1> neat
<0> it is useless most likely, but you may want to add descriptions on top of forum page
<0> very few text
<0> maybe that green box or a bit lighter
<1> i allready did much work in the style-less look - how do you like it without styles enabled
<0> it is not very logical somehow
<0> site news and links shoudl be in the end of page
<2> what do I need to do to enter userland? just create a pageing table that is setup as a userland? or do I need to do something with the segments?
<0> ronny, now i have to sleep
<0> gn everyone
<1> n8 mur
<2> good night mur
<3> when using grub's module command, is the module loaded at where it is supposed to be(the address specified in the elf header) or could it be anywhere in memory. And if it is not anywhere in memory, how do i make it go anywhere in memory
<3> what does the error unk. ctxt mean in bochs
<2> unknown context?
<3> what does that mean
<3> i'm trying to do paging
<3> for some reason the instruction pointer it is trying to read is in the middle of a mov %ebp, %esp; pop %ebp which is the stack handling code at the end of my ***embly enable paging routine
<2> have you asked in the bochs chatroom?
<3> how should i switch stacks?
<4> load a new one into esp
<3> but how can i then do a ret and won't my old local variables be gone
<4> that's the point
<3> in the bare-bones kernel does it reserve space for the stack in the kernel itself
<3> i mean in the executable image
<5> the multi-boot spec defines the spec as being undefined
<5> meaning, put together a stack as soon as possible
<5> so, if you're using grub
<5> and I'm pretty sure the elf format doesn't do anything like that.
<3> i did
<3> i got paging working yes!
<6> hi
<6> can someone point me to some information where i can start learning about OS development
<5> osdever.net
<4> try some of the links in the topic
<6> didnt bother looking lol, sorry about that
<5> geist: How difficult was it to write your apple ][ emulator?
<4> nah, those are floating all over the place
<4> it wasn't that hard really
<4> i guess the hardest part was getting all the soft siwtches right
<4> and trying to totally emulate the high res screen accurately
<7> Did you emulate the floppy drive?
<7> Because that would be hardcore.
<5> I'm kinda thinking of doing a snes/nes emulator once I lose interest in my OS
<8> question, how long have you guys been learning about OS development?
<5> os dev is just filling in the holes so you can put together a complete system... if you've been doing any multithreaded stuff and know memory management then the rest is just learning about the hardware and some filesystem stuff.
<3> is setting the last entry of the pdt to the pdt itself a good idea?
<5> why?
<5> unless you want the machine to see the remaining entries to the pdt as page table entries when you access the last bits of virtual memory
<5> last 4 megs that is... and then you can't touch the last 4k... if I understand the page hole thing correctly
<3> so you can access the process's page table at a known address sp1nm0nkey
<3> and then it is the same for all processes
<5> then you put the page directory address in a page table.
<5> and map it like any other page.
<5> unless you're using 4 meg pages.
<5> err... scratch that last one
<3> sp1nm0nkey: won't that just waste a page table?
<5> the page directory is only 4k
<5> you put it in one of your page tables.
<3> yeah and then you'll have to map your page tables in that page table too
<4> ah, the trick is p***ed on :)
<3> what trick, geist?
<5> iammisc: you'll have to map your page tables to modify them
<5> you know where they are since they're in your page dir.
<5> the ones that are per process are tracked in the task struct
<5> when you need to modify a page table, you map it
<5> if it doesn't exist, you grab a free physical page from your physical allocator
<5> and map it, insert what you need to, unmap it
<5> you can dedicate 4 megs of address space and a page table as a circular queue to create an lru sorta thing, just for mapping whatever explicit physical pages you need.
<5> if it's already mapped, you don't need to flush the tlb
<5> http://bcos.zapto.org/bcos/misc/snap/dev22-11-06.jpg how do I get my console to look all cool like that? Is that just a vesa console? How do you get the blue background?
<4> it's a vesa text mode
<4> and you can just display spaces with blue backcolor
<5> ah... I was hoping there was a better way than just iterating all the way through filling console memory with blue-ness
<4> not that i know of
<4> ultimately it ends up with the console memory being full of something
<4> either you do it or there's some automated way, but I dont know of an automated way
<4> also, it's not worth optimizing.
<5> nah, I guess it's only done once at boot.
<4> what do you mean?
<4> putting it into the high graphics mode? I guess
<3> should i index my memory before setting up paging? or should i allocate two pages at the end of my kernel to set up paging and then get memory allocation working?
<5> huh?
<5> link your kernel to the higher half, and get paging set up with instructions that use purely relative addressing, and jump to the higher half on the provisional page directory and page table
<5> index physical memory, allocate a new page dir, and make your virtual allocator aware of your kernel and it'll set up page tables, then switch to the new page dir/page tables and free the provisional ones
<3> sp1nm0nkey: i'm using grub as a boot loader and i'm scared of overwriting its multiboot info structure with my page tables
<3> sp1nmonkey:i'm using it for modules too
<5> the way I do it in my OS is I tell grub to load the kernel in the second meg, then link to the 3rd gig. I tell the linker to define a symbol called "END" at the very end of the kernel and to page align "END"
<3> sp1nmonkey i'm doing the same thing
<5> then you create your page dirs/tables at END
<3> that is also what i;m doing
<3> but what if grub's multiboot info is there
<5> check for yourself
<3> ok
<3> that is what i was going todo
<5> if you're worried, save the multiboot info
<3> sp1nm0nkey: and the modules too
<5> define some space for your multiboot info in the data segment, and copy it first thing, make a stack that should overwrite the multiboot info if grub actually puts it after the kernel and call main.
<5> since grub doesn't give you a stack
<3> why would i make a stack that will overwrite it? can't i just leave it be and write over it whenever?
<5> could do that, but I'd rather trust everything but the kernel and bios space as free space as soon as possible
<5> probably doesn't matter
<5> thing is, if you copy multiboot info into the kernel as soon as possible, you can trust everything after END as free.
Return to
#osdev
or
Go to some related
logs:
archlinux nvidia-glx
installer script aquabou gimp
#perl
Media change ubuntu nfs
gentoo rev-rebuild
ubuntu 6 avi decoder
#linux
SIOCSIFFLAGS: Cannot assign requested address+ubuntu
#gimp
#php
|
|
