Skip to contentSkip to conversation

Reserving a syscall number

DISCUSSION John Richard Moser 3 participants 4 messages Oct 27, 2004

lore
Share & tools
lore
#0 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

How would one go about having a specific syscall number reserved in entry.S? I'm exploring doing a kill inside the kernel from a detection done in userspace, which would allow the executable header of the binary to indicate whether the task should be killed or not; if it works, the changes will likely not go into mainline, but will still require a non-changing syscall index (assuming I understood the syscall manpage properly).

On a side note, if a syscall doesn't exist, how would that be detected in userspace?
- --
All content of all messages exchanged herein are left in the Public Domain, unless otherwise explicitly stated.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFBf+1thDd4aOud5P8RAkeNAJsFJD2l7Up62+/P+SHbJ3l7MwbM0gCfUE/Y vDgYr0SXlrkrwXZyEZw86QE=
=jmbP
-----END PGP SIGNATURE-----

#1 lore raw
Quoted 11 lines · John Richard Moser

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

How would one go about having a specific syscall number reserved in entry.S? I'm exploring doing a kill inside the kernel from a detection done in userspace, which would allow the executable header of the binary to indicate whether the task should be killed or not; if it works, the changes will likely not go into mainline, but will still require a non-changing syscall index (assuming I understood the syscall manpage properly).

Quoted 3 lines

On a side note, if a syscall doesn't exist, how would that be detected in userspace?
- --

Look at ld.so.preload for potential capabilities to control any executable.

Also what's the problem with sending the task a signal when the detection has been done?

If the usual capabilites are not sufficient, then make a driver (module).

Cheers,
Dick Johnson
Penguin : Linux version 2.6.9 on an i686 machine (5537.79 BogoMips).
Notice : All mail here is now cached and reviewed by John Ashcroft.
98.36% of all statistics are fiction.

#2 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

linux-os wrote:

| On Wed, 27 Oct 2004, John Richard Moser wrote:
|
|> -----BEGIN PGP SIGNED MESSAGE-----
|> Hash: SHA1
|>
|> How would one go about having a specific syscall number reserved in
|> entry.S?  I'm exploring doing a kill inside the kernel from a detection
|> done in userspace, which would allow the executable header of the binary
|> to indicate whether the task should be killed or not; if it works, the
|> changes will likely not go into mainline, but will still require a
|> non-changing syscall index (assuming I understood the syscall manpage
|> properly).
|>
|> On a side note, if a syscall doesn't exist, how would that be detected
|> in userspace?
|> - --
|
|
| Look at ld.so.preload for potential capabilities to control any
| executable.
|
| Also what's the problem with sending the task a signal when
| the detection has been done?
|
| If the usual capabilites are not sufficient, then make
| a driver (module).
|

I'm attempting to figure a way to control the IBM stack smash protector via a flag in the ELF header, without opening the executable image on disk and checking manually. If there is a way to do this from userspace, that would be acceptable.

|
| Cheers,
| Dick Johnson
| Penguin : Linux version 2.6.9 on an i686 machine (5537.79 BogoMips).
|  Notice : All mail here is now cached and reviewed by John Ashcroft.
|                  98.36% of all statistics are fiction.
|

- --
All content of all messages exchanged herein are left in the Public Domain, unless otherwise explicitly stated.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFBf/04hDd4aOud5P8RAm0AAJ9FWZ2d0hJpS5qDhogRPM6mWZJDOwCfe5YC BynHiZzH94hn5XnSLZlNqyc=
=jMqN
-----END PGP SIGNATURE-----

#3 lore raw
In reply to John Richard Moser
Quoted 7 lines · * John Richard Moser ([email protected]) wrote:

How would one go about having a specific syscall number reserved in entry.S? I'm exploring doing a kill inside the kernel from a detection done in userspace, which would allow the executable header of the binary to indicate whether the task should be killed or not; if it works, the changes will likely not go into mainline, but will still require a non-changing syscall index (assuming I understood the syscall manpage properly).

To reserve a syscall there needs to be some users and some eventual hope of merging. The idea, btw, means anyone can specify the value in the binary, so it could just as easily be done via prctl or something similar that makes the out of tree patch easier to maintain. Although, I don't actually see the value with the description above.

Quoted 2 lines

On a side note, if a syscall doesn't exist, how would that be detected in userspace?

ENOSYS error.

thanks,
-chris

Signature · 1 line
Linux Security Modules     http://lsm.immunix.org     http://lsm.bkbits.net