Mess over open(

Mess over open(

[Tommy McCabe]

Although the open( call seems to be used everywhere
(in around 70 source code files), it isn't defined
anywhere. How can ELKS get past booting if a call
isn't defined?


		
__________________________________
Do you Yahoo!?
New and Improved Yahoo! Mail - Send 10MB messages!
http://promotions.yahoo.com/new_mail 

Re: Mess over open(

[David Given]

On Tuesday 20 July 2004 03:19, you wrote:
> Although the open( call seems to be used everywhere
> (in around 70 source code files), it isn't defined
> anywhere. How can ELKS get past booting if a call
> isn't defined?

It might be a syscall entry point, or something else equally as weird; these 
tend to be defined in horrific bits of autogenerated assembly or other 
bizarre things. I'd suggest running nm on the object files and making sure 
that it *really* isn't defined... apart from anything else, the linker should 
bail.

(Not that I'm saying you're wrong, but I've spent so much time trying to find 
weird symbols in the Linux kernel source that I know not to jump to 
conclusions!)

-- 
+- David Given --McQ-+ "I told you to make one longer than another, and
|  dg@cowlark.com    | instead you have made one shorter than the other --
| (dg@tao-group.com) | the opposite." --- Sir Boyle Roche
+- www.cowlark.com --+ 

Re: Mess over open(

[Eduardo Pereira Habkost]

[-- Attachment #1: Type: text/plain, Size: 803 bytes --]

You have made exactly the same question about the "execve" sytem call,
some weeks ago.  8)

System calls are not linked magically using their names, but we should
have a system call interface that makes libc notify the kernel that some
syscall should be called. Take a look at arch/i86/kernel/syscall.dat,
mkentry.sh and the entry.c (it is generated by mkentry.sh), if you want
to know how it works.

If you are looking for the system calls' implementations, look at
sys_execve, sys_open, sys_fork, and others sys_* functions.

On Mon, Jul 19, 2004 at 07:19:51PM -0700, Tommy McCabe wrote:
> Although the open( call seems to be used everywhere
> (in around 70 source code files), it isn't defined
> anywhere. How can ELKS get past booting if a call
> isn't defined?
> 

-- 
Eduardo

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]