|
||||
IPF & PF
Hello, I tried to look up IPF vs PF which led me to this discussion
I stopped reading it cause I felt lost, the reason I am asking is this, if PF was so cool why it has not been adopted by Solaris or Linux tell now like what happened with ZFS *every one is trying to port it right?* I am not bashing PF, I am just asking out of curiosity Thanks in advance |
|
||||
nope, I am not trying to revive a ten year old discussion.
what I am trying to say is this, I know that PF has been ported to Free/Net why a big company like SUN have not ported it to Solaris? and instead they are still using IPF! I am just asking for answer if it's known, if it's not then they just do there thing and the BSD guys do there's. That's all |
|
|||
All anyone here can do is guess.
But if reality is any indication, look at the lag in PF versions found in both FreeBSD & NetBSD. Even though the network stacks are similar, porting PF is not a trivial matter, & I suspect that this is the general answer to your question -- very few engineers exist who have the knowledge, time, & desire to port it. Secondly, it is important to focus on the tight binding between PF & the network stack. FreeBSD & NetBSD are at least close in fundamental structure in network stacks to OpenBSD. Solaris has long diverged from its FreeBSD roots. I suspect (but this is conjecture as well...) that it would be very difficult to shoehorn PF onto Solaris' network stack. Again, someone would have to possess the knowledge, time, & desire to port PF. The other side of tight binding is that PF & OpenBSD's network stack are evolving/melding together more & more. Who is the say that PF's structure would be efficient on another network stack? I suspect (& this is conjecture too...) that the optimum packet filtering functionality for a network stack is specific to that stack. Yes, code can be written to be generic, but performance will likely degrade as a result. Lastly, Sun is no longer. Oracle is calling the shots, & I suspect there is still upheaval going on following the merger. What is the value of having PF on Solaris? If porting is not likely to generate measurable sales, I suspect (again, conjecture...) there is little desire to port PF. Last edited by ocicat; 21st January 2011 at 04:15 AM. |
|
|||
PF relies heavily on BSD kernel internals, (..mbufs for example, the BSD representation of network packets/buffers).
It's not an easy task to port something that's intimately part of the kernel, it's far reaching, even the FreeBSD/NetBSD ports of PF are based on an earlier version. A lot of projects have their own packet filters, adapting another would probably be counterproductive.. and in some cases, it would be a full rewrite. |
|
||||
Quote:
Quote:
any way thanks guys for the answer |
|
||||
These were not answers. They were conjecture.
Quote:
I don't believe OpenSolaris has an open kernel. If I'm right, there's no need to ask that project about it. |
|
||||
Because of the tight binding of pf with the OS network stack mentioned by ocicat and BSDfan666, my conjecture is that "porting pf to Linux" (for example) might end up being more like porting Linux to pf. I kinda don't see that happening.
|
|
||||
IPFW was ported to Linux, along with DummyNet. So it wouldn't be impossible to port PF.
Considering how often they change the packet filter in Linux, I wouldn't discount it from happening. Of course, Linux devs have THE worst case of NIH syndrome that I highly doubt they would even consider it. (FreeBSD devs ported IPFW, for example.) |
|
|||
Quote:
Windows Mac OS Linux The BSDs are not important for the market. And it is a true. Just look what happened with Xfce, KDE 4.6 and who know wha GNOME 3 will bring. Where are drivers for BSD? Flash for BSD? FreeBSD has a Linux layer because users want a flash which is not native for BSD, many games..., web cameras... And if you are looking "wars" between Linux and BSD users, plus Theo Radt talkings about Linux...what do you expect? Do you expect that Linux developers port PF? I don't. |
Thread Tools | |
Display Modes | |
|
|