I committed a simplified version of this 
(2743fdee849be4a00a9ea72d601004c95b22dd56) which passes ulogentries as a 
separate argument to print_update().

I omitted the change to the "if (entry ..." test since it didn't appear 
related to the purpose of this change.

I omitted the change to the start_sno assignment because I couldn't figure 
out the purpose of the change.  ulog->kdb_first_sno - 1 seems like a fine 
starting point even if the ulog is underfull, as long as kdb_first_sno is 
maintained properly.