I have 2 6509's running L3 and hsrp is configured between them. I read that its a good idea to have the root bridge associated to one of the 6509's running hsrp, too. But why? the root bridge and spanning-tree are layer 2 and hsrp is layer 3. One prevents hardware loops and one prevents software loops. What kinds of things could go wrong?
It is a good idea but it isn't necessary. Also STP prevents L2 loops rather than hardware loops as such and HSRP is nothing to do with preventing loops, it is to do with providing redundancy for end hosts/servers etc.
Lets say your'e 6500 switches are interconnected with a L2 trunk. The idea of matching up STP root and HSRP active gateway is so you can do a rudimentary form of load-balancing from the access-layer
ie. an access-layer switch will have 2 uplinks - one to SW1 (your first 6500) and one to SW2 (your second 6500). If you set SW1 to be STP root for odd vlans and SW2 to be STP root for even vlans then you can utilise both uplinks.
In addition if you ensure that the HSRP active gateway for an odd vlan is SW1 then you don't have to keep going across the L2 trunk that interconnects your 2 6500 switches.
So it's not necessary and nothing disastrous will happen if you don't do it, it is all about optimal paths in your topology.