[Varnish] #629: Fix the (Open)Solaris event port acceptor

[Varnish]

#629: Fix the (Open)Solaris event port acceptor
----------------------+-----------------------------------------------------
 Reporter:  slink     |       Owner:  phk
     Type:  defect    |      Status:  new
 Priority:  normal    |   Milestone:     
Component:  varnishd  |     Version:  2.0
 Severity:  normal    |    Keywords:     
----------------------+-----------------------------------------------------

Comment(by victori):

 The errno issue is still there even with the thread-safety arguments
 passed to SunCC(-mt) or GCC. What causes the panics is the assert lines on
 setting linger in tcp.c and cache_acceptor.c. When I comment out the
 asserts calls around setting linger, varnish works reliably.
 cache_acceptor.c triggers panics a lot more frequently than the one in
 tcp.c. Cache_acceptor.c would trigger a panic every 20 minutes or so while
 tcp.c would do it in a day or two.

 As to the poll vs eventports debate, I have stuck to poll. Poll has been a
 lot more smooth in production than eventports. Last time I checked
 eventports had "stalls" on cache misses at semi-frequent random occasions.

-- 
Ticket URL: <http://varnish-cache.org/ticket/629#comment:10>
Varnish <http://varnish.projects.linpro.no/>
The Varnish HTTP Accelerator