Re: select()ing on a pcap descriptor in Solaris

[Fernando Gont <fernando () gont com ar>]

Hi, Guy,

On 08/14/2014 06:01 PM, Guy Harris wrote:
> On Aug 14, 2014, at 1:34 PM, Fernando Gont <fernando () gont com ar>
> wrote:
>
> > While trying to select() on a pcap descriptor in solaris
>
> Which version?  Solaris 11 (in which pcap descriptors should refer to
> BPF devices), or earlier (in which pcap descriptors should refer to
> DLPI STREAMS devices)?

fgont@solaris:~/ipv6toolkit$ uname -a
SunOS solaris 5.11 11.1 i86pc i386 i86pc


> > I found the following:
> >
> > 1) It seems that pcap descriptors are never readable or writeable.
> > (-- This is different from *BSD and Linux, where at least you seem
> > to be able to check for readability)
> >
> > 2) It would seem that even trying to select() on such descriptors
> > causes trouble. For example, I get a kind of random "Bad file
> > number" error from select when I try to check for
> > readability/writeability.
>
> I wouldn't expect that prior to Solaris 11 - select() and poll() Just
> Work on STREAMS devices, as I remember.

You wouldn't expect them to fail, or to work properly?



> It could, however, be that Sun^WOracle screwed up with BPF;
> unfortunately, Solaris 11 is ClosedSolaris, so I can't just look at
> the source and see what they did wrong.

Seems like a long time since I last installed it... But I could try with
the CD they released circa 2009 -- that one was open source.



> I'll try this on my Solaris 11 virtual machine.
>
> > Is the above assessment correct? Should I not even try to call
> > select() on pcap descriptors in Solaris?
>
> If they screwed up on Solaris 11, you probably shouldn't try to call
> select(); instead, you should try to call Oracle and report this as a
> bug (and tell them that, if they fix this, they should fix it so that
> if you have a timeout set with BIOCSRTIMEOUT, the timer should start
> as soon as a select() or poll() is done, so that, when the timeout
> expires, it can report the descriptor as readable - the original BPF
> didn't do that, and all the *BSDs and OS X eventually fixed it).

Point noted. I'll wait for your response above, and will proceed as
indicated if necessary.

Thanks so much (yet one more time)!

Best regards,
-- 
Fernando Gont
e-mail: fernando () gont com ar || fgont () si6networks com
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1



_______________________________________________
tcpdump-workers mailing list
tcpdump-workers () lists tcpdump org
https://lists.sandelman.ca/mailman/listinfo/tcpdump-workers