Comment 522 for bug 88746

I have this exact problem with a particular SATA external hard-drive using kernel 2.6.24 (Hardy) and 2.6.27 (Jaunty). My observation is that plugging in the device with only one USB connection results in the problem (infinite resets) but plugging in the device with 2 connections (1 data+power and 1 power only) results in correct polling and mounting.
I am wondering if the USB 2.0 driver is supposed to sense and allocate power (or more precisely current)? Incorrect powering would explain why the problem is only seen with certain hardware combinations and is especially seen with USB hubs, which split power.