Re: library for ebtales

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

hello bart,

in future i will enlarge my activities for iptables and ebtables.
currently i have few time for that hobby, because i write my degree
dissertation. yes, add this lib to the cvs tree. it is a good first step
for an little cute lib.

yes, the implementation is correct. i have tested this implementation
and you can find this test in folder 'test' under 'ebtc_test2.c'.

output of this test:

Rule list in chain INPUT before delete
~  Entry jump to 'CONTINUE' (bcnt = 34706; pcnt = 64)
~  Entry jump to 'CONTINUE' (bcnt = 46; pcnt = 1)

Delete second entry

Rule list in chain INPUT after delete
~  Entry jump to 'CONTINUE' (bcnt = 34706; pcnt = 64)

Append entry

Rule list in chain INPUT after commit
~  Entry jump to 'CONTINUE' (bcnt = 34706; pcnt = 64)
~  Entry jump to 'CONTINUE' (bcnt = 0; pcnt = 0)

bye
jens

Bart De Schuymer wrote:
| Op ma, 04-04-2005 te 21:07 +0200, schreef Jens Götze:
|
|>yea multiple insertions, deletions and other chain operations (create,
|>delete, flush, rename, ...) before commit is fully supported, it a
|>naturally. ;o) all counterchanges or other changes will transfer into
|>the kernel with ebtc_commit. that include all entries and all counters.
|>yes, that is another reason for new implementation. i have needed so a
|>library without much gimicks. modules are useful for commandline tools,
|>for c programmer is the dymanic loading only annoying. i use first the
|>structures in ebtables.h of the linux kernel. that way leaves to me the
|>biggest freedom.
|
|
| Yes, I think it's good to have a small, simple and clean standalone
| interface to the kernel.
|
| Are you sure the EBT_SO_SET_COUNTERS implementation is correct? The
| reason why this socket call is needed after the EBT_SO_SET_ENTRIES call
| is because the kernel counters might have changed and it is up to the
| userspace program to figure out what needs to be done to deal with this
| (the kernel doesn't know which rules changed).
|
| example:
| Suppose the kernel table looks like this:
| first entry: pcnt=10, bcnt=10
| second entry: pcnt=10, bcnt=10
| Suppose we now delete the second rule and append a new one instead.
| Suppose the kernel table looks like this right before the
| EBT_SO_SET_ENTRIES call:
| first entry: pcnt=12, bcnt=12
| second entry: pcnt=13, bcnt=13
| After the EBT_SO_SET_ENTRIES call, the kernel table will look like this:
| first entry: pcnt=0, bcnt=0
| second entry: pcnt=0, bcnt=0
| After the EBT_SO_SET_COUNTERS call, the kernel must look like this
| (supposing the counters didn't change):
| first entry: pcnt=12, bcnt=12
| second entry: pcnt=0, bcnt=0
|
|
|>how can i upload that library to the sf cvs from ebtables? can i join to
|>ebtables?
|
|
| Are you planning on supporting ebtables in the future? Unless you
| actively take part in development and answering user questions, I don't
| see a reason to give you access to the cvs tree.
| Of course I'll add your useful library.
|
| cheers,
| Bart
|
|

- --
Jens Götze

E-Mail: [email protected]


GPG/PGP-Public-Key: http://pgp.1in1.de/jens/publickey.asc
GPG-KeyFingerprint: 17D2 FDE6 BC58 DE9B DCF9  EC90 EE58 3E1F DA51 9E6B
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCUarq7lg+H9pRnmsRAl/0AJ9U6aWNwMnqjmwV31zMy7I/y4i5wQCeNyyD
DT/KjIe9jroutQ64or+0jmA=
=6snN
-----END PGP SIGNATURE-----