| |
| |
| |
|
Page: 1 2
<wildestbo> How do I go about turning my COM port on my XP machine into a port that I can I/O freely as you would with a microcontroller?
<Robert> That's pretty hard.
<Robert> You could do that with the parallel port, though.
<archivist_> write a driver or learn to use what MS gives you
<Robert> http://www.ee.washington.edu/circuit_archive/text/para-port.html
<wildestbo> thanks, I'll try that.
<Worm_Drink> hello
<Robert> Hi, Worm_Drink.
<Robert> Cute nick you got there.
<Worm_Drink> yeah, thats what i thought
<Worm_Drink> listen, can anyone maybe help me out here
<Worm_Drink> I wanto change the EIP in olly debugger, but i cant seem to figure out how
<Robert> olly debugger?
<Worm_Drink> yes
<Worm_Drink> ollydbg
<Robert> Never used that one.
<Worm_Drink> damn
<Worm_Drink> well, its nice, untill you cant change the EIP
<Worm_Drink> you know of any debugging related channels that might be able to help me out ?
<Robert> Sorry, no idea.
<Robert> Tried reading the documentation?
<Robert> There must be SOME way to do jumps.
<Worm_Drink> hmm
<Worm_Drink> yeah, i didnt try jumps as keyword
<Worm_Drink> thanks
<Worm_Drink> ag ****, who whould have thought they will call "jump" "new origin here*"
<Robert> Heh.
<bockman> reading out of Programming from the Ground Up: it asks "what would %eax be if you did 'movl _start, %eax'?". is it the address of _start?
<Robert> Yes.
<Robert> You can treat _start as a constant, being the address.
<bockman> but what about 'movl $_start, %eax'?
<Robert> Er.. actually, I shouldn't say too much now. I'm not using as myself.
<Robert> I'm thinking about what would be written as "mov eax, _start" in NASM/FASM/YASM.
<bockman> ok.
<bockman> Robert: would 'movl $_start, %eax' be the same as 'movl (_start), %eax'?
<Robert> I don't know.
<force> in this code "pusha; push cs; mov bx,rr" what does the rr register do?
<Robert> What do you mean?
<Robert> bx will be set to "rr" (a constant?)
<force> well is it general purpose
<force> :/
<force> hmm
<force> rr:
<force> mov ax, 1 ; return here
<force> so perhaps just an address?
<force> cheers ;)
<Robert> Heh...
<chille_> o_0
<Robert> Hej, du.
<chille_> chille_: Hej, jag.
<chille_> haha
<Robert> Nej.
<Robert> Ah... <3 C.
<Robert> fin[j] = (d->punct[j] & (1<<d->k))? in[(*i_in)++] : 4.0f;
<chille_> <3
<aegray> TST R2, #2; beq addr -> this jumps when (R2&2==0) right?
<d1me0> right
<aegray> thanks
<d1me0> aegray: on which platform do you develop? Xscale or do you have a old ARM at home? Just to know...
<aegray> ipod
<aegray> :)
<d1me0> wow... which processor?
<aegray> arm7tdmi
<d1me0> they have all a nice asm. Thank you, and good work
<aegray> :)
<Charlls> heyo
<Charlls> i'm considering two approaches for a virtual machine specification: 1 would be a single range of untyped raw memory and registers, 2 would be a range of typed memory for each type of data that must be handled natively. I dont know if these alternatives have names but i would like to know if anyone has comments about it, or can suggest me some readings that evaluate pro and cons of each alternative
<wobster> most obvious one will be the performance prolly
<wobster> typechecking is always expensive
<Charlls> the idea more or less is that in case 2, the opcodes would be specific for each set of meaningful operations on the types, hence the opcode itself would tell which type of register is operated on
<Charlls> not that i'm sure that would do any good in the long run :)
<wobster> Charlls, I'd consider looking for apropriate groups in the usenet. you'll prolly get very qualified answers there
<Charlls> i think thats a good idea. thanks
<intero> hello
<intero> i am using gcc with inline at&t asm. i have an array of 3 elements called x_star, and i want to load it's address in ecx. x_star is global
<intero> asm( "movl $_x_star, %ecx" ); gives error: undefined reference to `_x_star'
<intero> where am i wrong?
<wobster> you cannot p*** a variable into that inline-asm that way
<wobster> intero, http://www.ibiblio.org/gferg/ldp/GCC-Inline-***embly-HOWTO.html
<intero> wobster: okay, i tried "m" also
<intero> lets say: asm("movl %0, %ecx" :: "m" (x_star));
<intero> err
<intero> errtwo %
<intero> lets say: asm("movl %0, %ecx" :: "m" (x_star));
<intero> mmm
<wobster> we count from 1 not from 0
<intero> ecx has two %, xchat doesnt show them
<wobster> %%
<wobster> it does
<intero> Numbering is done as follows. If there are a total of n operands (both input and output inclusive), then the first output operand is numbered 0
<intero> %
<intero> not here
<intero> i was quoting the url you gave me
<wobster> however .. should work like __asm__("movl %1, %%eax" : : "m" (xxx) );
<wobster> dont forget to add you ecx into the clobber-list
<intero> wobster: thanks. so that puts the address of xxx in eax
<intero> or it's value?
<wobster> always the value
<intero> and to get the address? __asm__("leal %1, %eax" : : "m" (xxx) ); ?
<wobster> see what these obscure m/r/etc flags mean in the howto
<wobster> &xxx would work
<intero> wobster: which compiler are you using?
<wobster> mostly gcc
<wobster> __asm__ is gcc-only, by the way
<intero> that's so weird :-/
<wobster> what is weird?
<xark> wobster: Mostly true, but its supported by ICC too (and a few others).
<wobster> yep. but let's say, I only not it for gcc for sure =)
<wobster> know
<xark> Mostly don't expect inline asm to ever be fully portable. :)
<intero> here parameters start counting at %0 (otherwise: error: invalid 'asm': operand number out of range) and if i do &xxx i get error: memory input 0 is not directly addressable
<xark> intero: Paste the line.
<intero> xark: just a min to make it "readable", thanks
<xark> intero: Also, newer gcc now supports named arguments (in addition to the %n numbered ones).
<wobster> gcc4?
<intero> xark: really? acn you make an example please?
<intero> i am using gcc4
<wobster> that could be the source of your errors :p
<intero> better use 3.4?
<wobster> I still have gcc 3.3.5
<xark> This might be handy -> http://www.ibiblio.org/gferg/ldp/GCC-Inline-***embly-HOWTO.html
<wobster> newer versions don't increase the skills, I noticed some time ago ;)
<intero> xark: there only %0 %1 operands are used
<xark> intero: Yeah, not related to the named params, but handy none the less. Still looking for a good reference on those...
<intero> right, 3.4 allows &xxx
<xark> intero: Here is an example of the named params (from official doc) -> http://gcc.gnu.org/onlinedocs/gcc-4.0.2/gcc/Extended-Asm.html#Extended-Asm
<intero> thanks xark
<xark> Hmm, it says "As of GCC version 3.1", so the named parms aren't as new as I thought...
<xark> Not all that much more readable, but at least you avoid "parameter renumbering" bugs if you change the parameters. :)
<intero> xark: http://rafb.net/paste/results/pbqOBg57.html
<intero> s/x_star/coeff/
<xark> intero: Hmmm, are you doing inline asm mostly for SSE use?
<intero> xark: yes
<intero> i know of intrinsics
<xark> intero: If so, the SSE intrinsics seem to work quite well (and are portable to ICC, VC++ and even AMD64)....OK
<intero> but there are too many load and store
<xark> It seemed to work pretty well for me (its just created with inline one-line asm routines).
<xark> But I have an SSE asm example...one min
<xark> Crap. I rewrote it with intrinsics so it would compile on my AMD64 machine. Sorry. :)
<intero> xark: see this http://rafb.net/paste/results/Q9UcM779.html thats why i dont like intrinsics (for critical parts)
<intero> xark: i have amd64 too, no need to rewrite
<xark> intero: Inline asm isn't the same for x86 and AMD64. I had issues.
<xark> (or #ifdef)
<intero> i agree. i am working in chroot to have 32 bits
<xark> intero: I suspect those loads aren't caused by the intrinsics.
<xark> intero: You are using -O2, right?
<intero> xark: the source: http://rafb.net/paste/results/ZsmO3080.html
<intero> nope
<intero> -g -O0
Return to asm
or
Go to some related
logs:
UseSidePixmap+change+debian
canada
politics
chatzone
WoW honored buying mount
football
|
|