C's macro documentation
From: Frank Kotler (fbkotler_at_comcast.net)
Date: 06/29/04
- Next message: Jonathan Bartlett: "Re: Jeff Duntemann Reviews AoA"
- Previous message: Alex McDonald: "Re: Random number question?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 29 Jun 2004 14:49:36 GMT
I thought of sending you guys a message Thursday, saying that I was off
for the weekend partying with old friends, and I'd look at the new code
when I got back. But that sounded ominously like "Debs' Last Message"
(since which she hasn't been heard from - Where are you Debs? We love
you! Hope you're okay.), so I didn't... and it worked, I'm back! :) But
horribly behind on correspondence!
Beth wrote:
>
> Hi,
>
> Subject line says it all, I suppose; Frank said that C has
> provided documentation for the macros but I just can't find
> it...is it a separate file or did C add comments to the source
> code or what?
Just in a message to the SF list, I guess. I'll stick it in CVS
somewhere RSN, if C doesn't beat me to it.
> Unfortunately, with "MailMan" out of commission
Very nice. Not sending anything to you, and bouncing messages back to
me. I'll have a look at "settings", as you suggest, and see if I can see
anything that might fix it. It *does* appear to be some kind of horned
beast... just a head-and-shoulders picture, so we can't see if the tail
is sys_forked or not :)
------
Brief notes on macros (luxasm/_nasm/macros):
(macro syntax is loosly based on Ada language)
[limited overview of code generated]
module / end_module
group procedures into a module (for name
mangling)
[no code generated]
interface / end_interface
specify interface for another module, define
external labels (with name mangling)
[generate _start code if 'start' procedure
found, else no code generated]
procedure / end_procedure
define the call frame and setup data for access
(via esp) for parameters & local variables
[no code generated, end_procedure calls 'return']
locals
define local variables (assumes size is 4bytes,
unless number is specified following label)
[ generates sub esp, <n> ]
return
cleanup stack frame & return from procedure
(return is auto called from end_procedure)
[ generates add esp, <n> \\ ret <n> ]
if / end_if / do / loop / ...
standard condition constructs -- you should be
able to guess the meaning from the name
stack _must_ be balanced else error
[ generates cmp <p>, <p> \\ jcc <l> ]
goto
conditional goto (issue cmp / jcc)
[ generates cmp <p>, <p> \\ jcc <l> or jmp <l> ]
select
generate a jump table & jump (similar to the
C-language 'switch'), parameters are register
for index, default option, <value / jump to
label><pairs> -- takes ages for nasm to
process this macro, but generates really
fast code :-)
[ generates cmp lt <low> \\ jcc <default>
cmp gt <high> \\ jcc <default>
jmp dword [ table + <reg>*4 ]
<jump table> ]
call
mangle name & push params depending on call mode
note: _will_ leave junk on stack when 'ret'
does not cleanup stack -- ie. using the C call
convention, stack is cleaned up on end of
block grouping (so as to clean up multiple
calls with a single 'add esp,<n>') -- user can
force cleanup of junk with 'drop junk', though
as I am using the Pascal calling convention this
is not really an issue.)
[ generates push \\ ... \\ push \\ call <l> ]
push / pop / pushad / ...
modify internal varaibles so 'local' access
points to correct place on stack (via esp).
else same as normal push/pop
[ generates push / pop / ... ]
drop
remove junk / data from stack (pseudo op)
[ generates add esp, <n>*4 ]
clr
set register to zero (pseudo op)
[ generates xor <reg>, <reg> ]
enumerate / end_enumerate
define group, start & stepping for enumerations
defined with 'enum'
[ generates nothing ]
type / end_type
define a struct / union -- can specify alignment
of data types
[ generates nothing ]
variable
define data in the ".data" or ".bss" section
[ generates data specified, or moves bss ptr ]
constant
define data in the ".rodata" section
[ generates data specified ]
assert
check a condition is true (code only generated
when DEBUG flag is defined, else generates
nothing) -- if condition is false 'assertion
failed' message is printed (used a a debugging
aid because gdb is rubbish :-) ).
[ generates assert call, if DEBUG flag]
review
print a review message (contents of g/p regs)
if REVIEW flag is set
[ generates review call, if REVIEW flag]
That should provide a good look at what is
happening, of course this is just an overview
and several features have not been covered
(including the infamous 'stack imbalance' error,
and several bugs which the library can produce
if you are not aware of them -- primarily due
to implicit stack management) but most of the
common macros are sketched here.
-------
Again, thanks for taking the time to write that up, C.
Best,
Frank
- Next message: Jonathan Bartlett: "Re: Jeff Duntemann Reviews AoA"
- Previous message: Alex McDonald: "Re: Random number question?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
