Re: Comments about IPT_ALIGN

[Laszlo Valko <valko@linux.karinthy.hu>]

Hi Thomas!

On Sun, Jan 26, 2003 at 05:50:21PM +0100, Thomas Heinz wrote:
> > Yes, that's true, except for the fact that you will have to deal
> > with endianness as well, if you are interested in actually exchanging
> > data... 
> 
> The only match/target I'm aware of that has a pointer in its info
> struct is the limit match. Generally this is necessary for each

There's one in struct ipt_replace (indeed that's not a match/target),
for the counters.

> match/target that does not treat the info data as read only on a
> smp system (at least with the current infrastructure as each cpu has
> its own match/target info).
> 
> Anyway, I don't see a reason why one wants to exchange a valid pointer
> between user and kernel space.

I have seen at least two guys here who asked how to exchange arbitrary
amount of data. And a pointer is an easy way out...

> > And don't forget the size change that brings in on every 32-bit
> > platform effectively killing compatibility...
> 
> One could use #ifdef's for compatibility purpose but as I read you
> don't like these :-)

I could put up with that one :) But I doubt that we could get that "solution"
accepted into mainline releases...

Regards,
Laszlo

> 
> 
> Thomas
>