[Varnish] #144: Race in backend socket address replacement.

[Varnish]

#144: Race in backend socket address replacement.
---------------------------------------+------------------------------------
 Reporter:  anders                     |        Owner:  phk     
     Type:  defect                     |       Status:  assigned
 Priority:  normal                     |    Milestone:          
Component:  varnishd                   |      Version:  1.1     
 Severity:  normal                     |   Resolution:          
 Keywords:  varnishd core dump sig 10  |  
---------------------------------------+------------------------------------
Comment (by anders):

 I applied commit 1809 (http://anders.fupp.net/patch-varnish-sigbus-r1809).
 PHK said it should help, but instead I get sigbus more often now:

 {{{
 Aug  7 18:03:07 cache10 kernel: pid 27495 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:04:25 cache10 kernel: pid 27714 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:05:44 cache10 kernel: pid 27747 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:07:03 cache10 kernel: pid 27751 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:08:20 cache10 kernel: pid 27758 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:09:38 cache10 kernel: pid 28003 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:10:53 cache10 kernel: pid 28010 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:12:09 cache10 kernel: pid 28050 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:13:28 cache10 kernel: pid 28064 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:14:47 cache10 kernel: pid 28317 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:16:03 cache10 kernel: pid 28318 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:17:20 cache10 kernel: pid 28328 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:18:36 cache10 kernel: pid 28355 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:19:55 cache10 kernel: pid 28580 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 Aug  7 18:21:11 cache10 kernel: pid 28607 (varnishd), uid 0: exited on
 signal 10 (core dumped)
 }}}

 Did a backtrace on one of the new core-dumps, and got:

 {{{
 Program terminated with signal 10, Bus error.
 #0  0x0000000000408840 in vbe_sock_conn (ai=0xd0d0d0d0d0d0d0d0)
     at cache_backend.c:161
 161     cache_backend.c: No such file or directory.
         in cache_backend.c
 (gdb) bt
 #0  0x0000000000408840 in vbe_sock_conn (ai=0xd0d0d0d0d0d0d0d0)
     at cache_backend.c:161
 #1  0x00000000004088ec in vbe_conn_try (bp=0xaf4d00, pai=0x7ffffa7d2868)
     at cache_backend.c:179
 #2  0x0000000000408ac4 in vbe_connect (sp=0xc1a008, bp=0xaf4d00)
     at cache_backend.c:227
 #3  0x0000000000408ee3 in vbe_nextfd (sp=0xc1a008) at cache_backend.c:298
 #4  0x0000000000409122 in VBE_GetFd (sp=0xc1a008) at cache_backend.c:335
 #5  0x000000000040c700 in Fetch (sp=0xc1a008) at cache_fetch.c:278
 #6  0x0000000000409e24 in cnt_fetch (sp=0xc1a008) at cache_center.c:290
 #7  0x000000000040addd in CNT_Session (sp=0xc1a008) at steps.h:40
 #8  0x00000000004124b8 in wrk_do_one (w=0x7ffffa7d2ad0) at
 cache_pool.c:193
 #9  0x0000000000412814 in wrk_thread (priv=0x856300) at cache_pool.c:234
 #10 0x000000080088655b in ?? ()
 #11 0x0000000000000000 in ?? ()
 #12 0x0000000000000000 in ?? ()
 #13 0x0000000000bd2e00 in ?? ()
 #14 0x0000000000000000 in ?? ()
 #15 0x0000000000af7000 in ?? ()
 #16 0x0000000000000000 in ?? ()
 Cannot access memory at address 0x7ffffa7d5000
 (gdb)
 }}}

-- 
Ticket URL: <http://varnish.projects.linpro.no/ticket/144#comment:2>
Varnish <http://varnish.projects.linpro.no/>
The Varnish HTTP Accelerator