Odd TimeQuest slack time issue

Hi,

I am having a rather odd issue with timequest. I get many errors (slack time fail) of very small amounts no matter how I adjust the optimization. The weirdest part is that the errors are from a node to itself. How is that even possible? Here is an example, this was worst case fast model, i have similar ones from the slow model, time (ns) and then "from node" and then "to node":

*******************

-0.005 RX_SIG_PROC:\sig_proc_gen:DATA_PROC|dec5fir3:LPF3_I|dec5fir3_ast:dec5fir3_ast_inst|dec5fir3_st_wr:fircore|mr_dnc_wr:u1|cnt_reg[2] RX_SIG_PROC:\sig_proc_gen:DATA_PROC|dec5fir3:LPF3_I|dec5fir3_ast:dec5fir3_ast_inst|dec5fir3_st_wr:fircore|mr_dnc_wr:u1|cnt_reg[2]**********************************

-0.005 RX_SIG_PROC:\sig_proc_gen:DATA_PROC|dec5fir3:LPF3_I|dec5fir3_ast:dec5fir3_ast_inst|dec5fir3_st_wr:fircore|mr_dnc_wr:u1|cnt_reg[1] RX_SIG_PROC:\sig_proc_gen:DATA_PROC|dec5fir3:LPF3_I|dec5fir3_ast:dec5fir3_ast_inst|dec5fir3_st_wr:fircore|mr_dnc_wr:u1|cnt_reg[1]

**********************************

I get a lot more of these as hold violations in slow model than fast model. These aren't signals that feed back it's just part of a normal data source->do_stuff->do_stuff->output . Anyone seen this before? I don't understand how you can have a slack violaiton on the same node. This is a relatively slow 25-30Mhz Cyclone 3 design.