The link above refer to the details of what I did trying to isolate the problem. When I am plugging in my Iomega jaz card (pcmcia to fast scsi II adapter) which is using the aha152x_cs driver (it worked fine with Redhat 7.x and kernel 2.4.x) the system (IBM Thinkpad T20 with 256MB RAM) is slowing down and the keyboard doesn't respond anymore to anything else than CTRL-ALT-DEL. The behavior is exactly the same if I boot the laptop with the adapter already inserted and the drive (Iomega 2GB Jaz) switched on. Seems the /etc/pcmcia/scsi script is invoked with wrong arguments or the udev system is not creating the appropriate device nodes. Seems these are not talking the same language with they come to scsi devices. I don't know which naming scheme is the appropriate one. The details are in the forum discussion about the arguments. So, I modified the scsi script to have it exit before running offending commands (which I was thinking were responsible for the system starting to trash), namely /sbin/scsi_info /dev/scsi0 while this file doesn't exist. But it apprears it is not the problem. So, I guess the driver itself is doing some weird things or something is ran after the scsi script and initiated by the cardmgr. From here I don't know how to investigate further. But it seems clear to me there is a bug somewhere. Reproducible: Always Steps to Reproduce: 1.Switch on the Iomega 2GB Jaz drive 2.Insert the jaz card in the pcmcia socket 3.Wait few seconds and the keyboard is no longer responding and the system cannot be bring down cleanly. Actual Results: System behaves like it is thrashing. Slow to respond, doesn't respond to keyboard except CTRL-ALT-DEL, in graphical mode, the mouse will continue to respond but moves very slowly and the pointer doesn't follow the mouse. Expected Results: The system should continue to respond appropriately, the drive should be mountable and accessible.
I did some further testing. I updated my udev to latest 0.52 without anything resolved. But, now I can also reproduce the problem by just plugging the PC Card adapter into the socket. The reason I didn't notice this behavior before, it is taking more time to evolve into a thrashing like situation. Seems to me a strong indication there is a memory leak in the aha15x_cs driver.
Yeah, not a hotplug issue then :) Closing the bug.