dovecot2-pigeonhole / managesieve crash

[Toby Karyadi <toby.karyadi%gmail.com@localhost>]

Hi,

I'm on netbsd 6.1 stable Xen domU using dovecot 2.2.5 (latest) and 
dovecot2-pigeonhole-0.4.1 (also latest) from pkgsrc and am trying to get 
managesieve working. Everytime I do 'telnet <managesieve/dovecot host> 
4190' (that's managesieve's port), the connection is immediately 
disconnected ('Connection closed by foreign host'). Additionally I'm 
trying to get a core dump, but I just couldn't get one.

/var/log/maillog typically shows:
Sep 27 11:12:30 <the_host> dovecot: managesieve-login: Panic: file 
ioloop.c: line 37 (io_add): assertion failed: (fd >= 0)
Sep 27 11:12:30 <the_host> dovecot: managesieve-login: Fatal: master: 
service(managesieve-login): child 3044 killed with signal 6 (core not 
dumped - set service managesieve-login { drop_priv_before_exec=yes })

So, then I did all sort of things to get a coredump, but failed. Setting 
'drop_priv_before_exec = yes', disallows setting chroot = login for the 
managesieve-login, which is the default. At any rate,
I added the following in my dovecot/local.conf just to not get other 
error messages related to permission when accessing config and anvil.

service anvil {
  unix_listener anvil {
    mode = 660
    group = dovenull
  }
}

service config {
  unix_listener config {
    mode = 660
    group = dovenull
  }
}
service managesieve-login {
  chroot =
  drop_priv_before_exec = yes
  user = dovenull
}

In the end I still can't get a coredump, even after I set 
kern.coredump.setid.dump = 1 through sysctl.

Any pointers?

Toby