[Patch] Fix another unbound alloca (BZ 13761)

[Jeff Law <law at redhat dot com>]

This is a follow-up to fully resolve 13761.

Using the same procedures mentioned in BZ 13761, it's still possible to 
trigger an unbound alloca in nscd.  Basically Siddhesh's patch fixed the 
first unbound alloca, but left a later one unfixed.

Basically DATASET in cache_addgr can grow without bound in one 
particular case.  When we can not permanently add the result, but can 
provide the result as-is we allocate temporary memory to hold that 
DATASET.  If DATASET is large, we blow out the stack.

Attachment: patch
Description: Text document