[Varnish] #1630: Varnish crashes with long regex condition

[Varnish]

#1630: Varnish crashes with long regex condition
-------------------------+--------------------
 Reporter:  huguesalary  |       Owner:
     Type:  defect       |      Status:  new
 Priority:  normal       |   Milestone:
Component:  varnishd     |     Version:  4.0.2
 Severity:  normal       |  Resolution:
 Keywords:               |
-------------------------+--------------------

Comment (by paluna):

 Using varnishd (varnish-4.0.2 revision bfe7cd1)

 I get this annoying crash as well (the exact same segfault message).

 It occurs +- 1x/week on a website with 50K hits per day (20% are
 smartphones) but I cannot reproduce it at all. I restart the varnish cache
 daemon every night, it might occur more frequently if I do not do the
 restart every night.

 I see some similarities with this ticket. I also use User-Agent regexp's
 intensively. But your repro command <curl http://yourhost/ --header "User-
 Agent: Mozilla/5.0 (Linux; Android 4.2.1; en-us; Nexus 4 Build/JOP40D)
 AppleWebKit/535.19 (KHTML, like Gecko) Chrome/18.0.1025.166 Mobile
 Safari/535.19"> works fine on my installation.

 There are no traces in the web server logs around that timestamp so it
 occurs before the request is passed to the backend.

 I already changed these settings a month ago but it does not fix it
 unfortunately.
   VARNISH_PCRE_MATCH_LIMIT=2500
   VARNISH_PCRE_MATCH_LIMIT_RECURSION=2500
   VARNISH_THREAD_POOL_STACK=72k

 The only extra artifacts I have are the Ubuntu .crash file and the
 apport.log
 https://dl.dropboxusercontent.com/u/5629939/temp/_usr_sbin_varnishd.65534.crash
 https://dl.dropboxusercontent.com/u/5629939/temp/apport.log

-- 
Ticket URL: <https://www.varnish-cache.org/trac/ticket/1630#comment:4>
Varnish <https://varnish-cache.org/>
The Varnish HTTP Accelerator