Re: (ITS#4943) tpool.c pause vs. finish

[h.b.furuseth@usit.uio.no]

I wrote:
> I'll just make [the hash table] chained.  (...)
> Solves the multi-pool problem too.

No it doesn't.  pool_pause() would stilll only pause one pool, while
purgekey() needs all pools to be paused.  I guess I'll chain it anyway.


I forgot one bug in my summary:

Slapd does not always check if pool_submit() and pool_setkey() succeed,
even though failure is not a bug.  (They can run into some limits.)
  pool_submit():
    daemon.c:slapd_daemon_task()            with rtask->routine
    connection.c:connections_shutdown()     with c_clientfunc
    connection.c:connection_read()          with c_clientfunc
    back-bdb/tools.c:bdb_tool_entry_open()  with trickle and index tasks
  pool_setkey():
    connection.c:connection_fake_init2()    for slapi object extensions
    backend.c:be_isroot_pw()                for SLAPD_SPASSWD
    passwd.c:slap_passwd_check()            for SLAPD_SPASSWD
    sl_malloc.c:slap_sl_mem_create()
    back-bdb/search.c:search_stack()
    back-meta/conn.c:meta_back_candidates_get()
    slapd-modules/smbk5pwd/smbk5pwd.c:smbk5pwd_op_bind()

-- 
Regards,
Hallvard