| |
| |
| |
|
Page: 1 2
Comments:
<0> mwk: but a push is gonna want to decrease esp...and if esp=ss=0 you got problems right?
<1> seb-: then esp underflows to 0xffffffc. any problems?
<1> besides, if ss=0, it's stack fault. ss descriptor cannot be null.
<0> mwk: are you telling stack can happily grown down from 0 w/o any problems?...0xffff fffc, 0xffff fff8, etc>
<0> s/telling/telling me
<1> yes. it's just that this area of virtual space is used by kernel on some systems, so it'll be segfault.
<1> but if you have an OS that makes those areas accessible to userspace, it's alright.
<0> mwk: that's cool..i didn't know memory was circular like that
<2> mwk i couldnt find intel docs, do they have a special devlopers site or?
<1> actually, it reminds me of xbox story
<1> lukus001: they're just hard to find.
<1> lukus001: http://x-os.homeip.net/x-os/hdir/pdf/intel/
<2> geg, thank you very much
<2> heh*
<2> <3
<1> lukus001: here's my copy. note that this is on my home box, and my internet connection ****s m***ively. can be slow and/or connection can break.
<1> lukus001: 25366x18.pdf
<2> Im using a web based client so i dont think i can transfer
<1> oerm?
<1> this is normal http. you go there with your web browser.
<1> http://x-os.homeip.net/x-os/hdir/pdf/intel/25366*18.pdf, substituting * with numbers from 5 to 9.
<2> okays, thanks
<2> mwk, Downloaded them all now - thank you very much =)
<3> hmm, sse has no cmpxchg instruction :(
<1> you mean, like, 128-bit atomic cmpare-and-swap?
<3> nope
<4> anybody can help me find a pdf of the book "programming the 80386" ?
<4> please anybody has that?
<5> theres a 128 bit cpu now?
<6> __mikem: sse is 128 bit
<5> does the amd64 chips have sse?
<7> hello
<7> i just thought of a good geek challenge...
<7> ... a contest to code the most featureful "os" in a bootsector
<8> http://www.rafb.net/paste/results/3MQvne30.html is a program that reads a file, convert uppercase into lowercase and vice versa, however the output of this program wrong. Any ideas why ?
<9> Someone know how to malloc memory using syscalls?
<9> Also: how to get the start of the heap?
<8> http://www.rafb.net/paste/results/hW03za25.html is giving me segmentation fault, anybody know why ?
<10> iUninstall, use sbrk or similar. that increases the address space of the process. memory management is up to you then.
<10> gnarf!
<10> vick, use a debugger
<8> wobster, i am using one
<8> wobster, i know i am reading memory i am not supposed to
<8> wobster, i just dont know how am i doing that
<9> Sorry, I close the client.
<8> wobster, on the line movb (%eax, %edi, 1), %cl
<10> iUninstall, use sbrk or similar. that increases the address space of the process. memory management is up to you then.
<9> wobster, I can't use sbrk
<9> i need to use brk
<8> wobster, %eax holds the memory address of something which has 0x0 at some place, and i loop till i find the 0x0
<10> iUninstall, whatever :)
<9> (sbrk isn't a syscall, brk is a syscall)
<10> I don't care, iUninstall. The message is just: management is up to you.
<9> wobster, Do you know how to get the start of heap?
<9> wobster, without reading /proc/self/
<11> iUninstall: The point is, you can't malloc with syscalls. Malloc is a c library function which allocates memory from the heap. I suggest using C library malloc or implementing something like a really primitive sequential allocator
<9> oggis__, Yes, but if I want to implement a memory allocator I need: brk syscall and to know where te heap start
<9> because brk need an absolute address
<11> it needs the new END absolute address, and you can get the end with brk'ing with 0 increment
<11> and to that just add how much you want to enlarge, and you'll have the absolute address to p*** to brk
<9> OK.
<9> I put 0 on ebx (first parameter) , and 45 on eax (syscall number)?
<11> yeah
<8> oggis__, can you take a look on http://www.rafb.net/paste/results/r8LmUO81.html
<9> And on eax I will find the start of heap, after I called Inx 0x80?
<8> oggis__, my strlen is not working right :(
<11> iUninstall: yep
<11> vick: yeah, let's see
<8> Thanks :D
<8> ok i knew why now
<8> heh.
<8> because i compared, and then made another instruction, and then used jne
<8> Where i should have use jne, je, etc.. after cmp without anything in the middle
<8> isn't that correct ?
<9> Hi redblue
<8> 'ld' links with libc automatically, or do i have to make it do so ?
<9> Hi mwk
<11> vick: sorry, had to help one guy at work first, i'll take a look at it now
<8> oggis__, it worked already :D
<8> oggis__, Thanks anyway :D
<11> yeah, incl messes up the equal comparisonflags ofc
<11> you swapped the order of incl and cmp right?
<11> note that you need one decl afterwards ;)
<11> (strlen reports length without NUL)
<1> hi iUninstall
<9> mwk, Later can you help me to improve asm inline code?
<1> probably
<8> oggis__, Yep
<8> If i write a character in %eax, it resides in %al, and the rest is zero, is that correct ?
<11> depends
<11> remember sign extension vs zero extension
<11> so it depends on the definition of "write"
<8> i mean
<8> movl 'c', %eax
<11> yeah, that'll do it.
<1> yeah, rest is 0 in this case
<11> but note that in that expression 'c' is actually a 32 bit value
<11> heh, he's gone.
<11> for ascii he won't ofc have to worry about extension...
<12> hi all
<1> hi
<12> I have a question, I want to learn asm, But do all x86 processors have the same instruction set?
<12> Or are they all compatible on like a generic level (sorry for the noobness :D)
<13> not completely, no, depends on manufacturer and model
<12> Ok.
<14> doesn't the asm compiler sorta take care of that
<12> The reason i ask is i want to learn x86 ASM, but the only x86 computer ill have in the house is a laptop, Which has a VIA C7-M processor.
<13> especially as your question is x86 eg 8086 and on
<9> mwk, I've tried to improve my code
<13> TisMe, down load the manual for your chip
<9> mwk, http://phpfi.com/138264
<9> mwk, But don't works
<15> most modern x86 CPUs should be "compatible on a generic level"
<12> Kasperle: ok, thanks. And sorry if that wasn't the best term to use :-)
<15> for learning x86 asm the C7 should be just fine
<1> oh blah. people. all x86's are the same except for sse and other weird extensions
<1> iUninstall: wrong.
<1> iUninstall: those registers are INPUT registers. and you thrown them as output. wrong.
<12> Ok thanks for clearing that up for me everyone :-)
<9> How to fix it?
<1> iUninstall: also, don't declare those registers as 'dword'. use the natural type. buffer should be of type const void *
<1> iUninstall: specify all the registers in input section of asm.
<1> asm volatile("int $0x80" : "=r" (result) : "r" (syscall), "r" (fdesc), "r" (buffer), "r" (size));
<1> also, declare extra:
<1> register int result asm ("%eax");
<1> and don't use 'dword'
<9> mwk, dword is unsigned int
<11> argh
<11> that's junk
<1> iUninstall: so use 'unsigned int'.
<1> iUninstall: or better, uint32_t
<11> why uint32_t but not "dword" which is likely synonymous?
<11> which IS synonymous
<1> oggis__: because uint32_t is standard C.
<11> ah, that was what you were trying to accomplish
<12> Ah, one more question, Whats the best starting point to learn asm? A specific book? online tutorial?
<9> Art of asm
<12> Is that a book?
<12> Nevermind, got it
<9> TisMe, http://webster.cs.ucr.edu/AoA/Linux/index.html
<9> http://webster.cs.ucr.edu/AoA/Windows/index.html
<12> Thank you.
<0> I thought no one used segments on intel...or just set all segment registers to zero??....My Minix OS book (OS D&I) is describing 'segmentation with paging' on Pentium like everyone is doing it
<1> not 'segment registers to 0'
<1> segment registers need to be set to segment ID
<1> which is an entry in a Big Fat Table Of Segments
<1> using segmentation isn't as simple as many think. even if it's just to pretend segmentation doesn't exist.
<0> mwk: i want to understand a simple boot loader.....the boot loader was forced to initialize stuff i never heard of....'page directories, page tables, page directory base register, GDT and LDT"
<1> bootloader initialising paging? crazy. it could be done by the kernel as well.
<1> but well.
Return to
#asm
or
Go to some related
logs:
authProg sasl
#ldap
fedora vmnet device udev problem
x226 debugging
authorized_keys command rssh
mysql_fetch_array undefined non-object
xstartup vncserver centos gnome exit
installing cpkg kubuntu
phishing eails
#perl
|
|