chdir on linux does not work

chdir on linux does not work

[Noprianto]

Dear All,

Anyone can tell me why this chdir code doesnot work on
linux ?

-> from what i read, chdir is system call number 12


section .text
        global _start

_start:
        mov eax, 12
        mov ebx, dir
        int 80h


        mov eax, 1
        mov ebx, 0
        int 80h



section .data
dir:    db      '/bin'



Thanks,
Pri

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: chdir on linux does not work

[Timo Lindfors]

[-- Attachment #1: Type: text/plain, Size: 623 bytes --]

Hi,

On Tue, May 23, 2006 at 07:49:29PM -0700, Noprianto wrote:
> -> from what i read, chdir is system call number 12

Yep, but better use something like

%define __NR_chdir               12

to make code more readable. Also, your example works just fine here:

$ nasm a.asm -f elf
$ gcc a.o -o a -nostdlib
$ strace ./a
execve("./a", ["./a"], [/* 30 vars */]) = 0
chdir("/bin")                           = 0
_exit(0)                                = ?

Remember that chdir affects only the process that calls chdir. You
can't e.g. write a replacement for shell's builtin "cd" command that
way.

best regards,
Timo Lindfors

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

Re: chdir on linux does not work

[Brian Raiter]

> Anyone can tell me why this chdir code doesnot work on
> linux ?

> section .data
> dir:    db      '/bin'

You forgot to NUL-terminate your string. A dangerous mistake, as it
can become a Heisenbug. You were lucky it failed to work for you. Use
this instead:

dir:	db	'/bin', 0

b

Re: chdir on linux does not work

[Frank Kotler]

Timo Lindfors wrote:

...
> Remember that chdir affects only the process that calls chdir.

I thought Noprianto's problem was the lack of a terminating zero after 
"/bin"... didn't help, but I still think it's a bug waiting to byte. 
Adding a call to sys_getcwd (__NR_ 183) (why is this man 3, not man 2 
???) proves that, sure enough, it *does* work... but we change back on 
exit... as you say...

> You
> can't e.g. write a replacement for shell's builtin "cd" command that
> way.

Okay, how *would* we do that? I know! I'll check "asmutils"! Hmmm... no 
chdir in asmutils... This is apparently harder than it looks!

Good Question!

Best,
Frank

Re: chdir on linux does not work

[Brian Raiter]

> Adding a call to sys_getcwd (__NR_ 183) (why is this man 3, not man
> 2 ???)

Good question, actually. I don't know.

>> You can't e.g. write a replacement for shell's builtin "cd"
>> command that way.
> 
> Okay, how *would* we do that?

It is simply not possible. Like setenv, cd absolutely must be a shell
builtin.

b

Re: chdir on linux does not work

[Frank Kotler]

Brian Raiter wrote:

>>>You can't e.g. write a replacement for shell's builtin "cd"
>>>command that way.
>>
>>Okay, how *would* we do that?
> 
> 
> It is simply not possible. Like setenv, cd absolutely must be a shell
> builtin.

Okay... Thanks for the quick response, Brian - probably saved me a lot 
of frustration!

Best,
Frank

Re: chdir on linux does not work

[Timo Lindfors]

Hi,

On Wed, May 24, 2006 at 12:15:19AM -0700, Brian Raiter wrote:
> It is simply not possible. Like setenv, cd absolutely must be a shell
> builtin.

Indeed, unless you do some dirty tricks by using ptrace(2) to attach
to the parent and force it to do the chdir syscall :)

best regards,
Timo Lindfors

Re: chdir on linux does not work

[jeff]

On Wednesday 24 May 2006 02:49 am, Noprianto wrote:
> Dear All,
>
> Anyone can tell me why this chdir code does not work on
> linux ?

If you  run this program in a terminal  it will do the chdir,
then restore the original directory when the program
exits.  Possibly this is what you are seeing?

>
> -> from what i read, chdir is system call number 12
>
>
> section .text
>         global _start
>
> _start:
>         mov eax, 12
>         mov ebx, dir
>         int 80h
>
>
>         mov eax, 1
>         mov ebx, 0
>         int 80h
>
>
>
> section .data
> dir:    db      '/bin'
>
>
>
> Thanks,
> Pri
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
> -
> To unsubscribe from this list: send the line "unsubscribe linux-assembly"
> in the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: chdir on linux does not work

[Noprianto]

Hi Jeff,

> If you  run this program in a terminal  it will do
> the chdir,
> then restore the original directory when the program
> exits.  Possibly this is what you are seeing?
> 
Yes, thats what i am seeing. 



> >
> > -> from what i read, chdir is system call number
> 12
> >
> >
> > section .text
> >         global _start
> >
> > _start:
> >         mov eax, 12
> >         mov ebx, dir
> >         int 80h
> >
> >
> >         mov eax, 1
> >         mov ebx, 0
> >         int 80h
> >
> >
> >
> > section .data
> > dir:    db      '/bin'
> >
> >
> >


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: chdir on linux does not work

[Adrian Ho]

On Wed, May 24, 2006 at 02:49:24AM -0400, Frank Kotler wrote:
> Adding a call to sys_getcwd (__NR_ 183) (why is this man 3, not man 2 
> ???)

Are you talking about sys_getcwd(3) (which I've never seen or known to
exist) or getcwd(3)?  If the latter, there's a Linux-specific extension
that uses malloc() -- perhaps that's the reason why:

 As an extension to the POSIX.1 standard, Linux (libc4, libc5, glibc)
 getcwd() allocates the buffer dynamically using malloc() if buf is NULL
 on call. In this case, the allocated buffer has the length size unless
 size is zero, when buf is allocated as big as necessary. It is possi-
 ble (and, indeed, advisable) to free() the buffers if they have been
 obtained this way.

- Adrian

Re: chdir on linux does not work

[Frank Kotler]

Adrian Ho wrote:
> On Wed, May 24, 2006 at 02:49:24AM -0400, Frank Kotler wrote:
> 
>>Adding a call to sys_getcwd (__NR_ 183) (why is this man 3, not man 2 
>>???)
> 
> 
> Are you talking about sys_getcwd(3) (which I've never seen or known to
> exist) or getcwd(3)?

getcwd(3). I refer to it as "sys_???" when I mean the int 80h interface, 
sorry.

> If the latter, there's a Linux-specific extension
> that uses malloc() -- perhaps that's the reason why:

I'll bet you're right. If that isn't the reason, it's as good a reason 
as any. Good point!

Best,
Frank