This is one of two things: HC is broken, or SMM BIOS messes with the
hardware behind our backs.

Please try usb-handoff option in grub.conf command line (where root=X is).
Also (separately), try to disable everything that resembles "USB Legacy
Support" or "Compatibility" or whatever

I tried both usb-handoff in grub.conf and disabling/altering things in the BIOS.
 No luck yet.

Here is the entry I used in grub.conf:

title Fedora Core (2.6.14-1.1637_FC4smp)
        root (hd0,0)
        kernel /vmlinuz-2.6.14-1.1637_FC4smp ro root=/dev/VolGroup00/root
usb-handoff rhgb quiet
        initrd /initrd-2.6.14-1.1637_FC4smp.img

As for the BIOS, I actually didn't find "USB Legacy Support" or "USB
Compatibility".  I did turn off "USB Keyboard Support", but it had no affect on
the problem.  I also disabled ACPI and tried MP=1.1 to see if that would have an
effect---no luck.

Any additional suggestions?

Thanks,

Paul

No ideas at present time. We may be sure that HC is all right though, I did
not read the report closely enough. All signs point to a rude SMM BIOS.
I don't know what made it act up in recent kernels...

I'm seeing something similar (jerky/stuttery mouse movement, terrible gamepad
control) on a MSI Master2-FAR (VIA K8T800 chipset) with the AMD64
kernel-smp-2.6.14-1.1644_FC4. It doesn't happen immediately (or always), but
once it starts, it'll occur if I plug the mouse/gamepad into any of the USB
controllers and it survives a warm reboot.

USB Legacy is disabled (iirc, I'm going to reboot, check that, and then try
usb-handoff after I finish this comment).

I'm still seeing this with 1653, but I've found a workaround that doesn't
require a reboot -- forcing the UHCI driver to reinit the host controller (by
unbinding and then rebinding it to the PCI devices) seems to clear it up.

For the unitiated, unbinding/rebinding can be accomplished by something similar
to the following. Obviously, you really want to script this if you have a USB
keyboard. I also do the EHCI controller for good measure.

# cd /sys/bus/pci/drivers/uhci_hcd
# ls
0000:00:10.0  0000:00:10.1  0000:00:10.2  bind  new_id  unbind
# echo -n "0000:00:10.0" > unbind
# echo -n "0000:00:10.1" > unbind
# echo -n "0000:00:10.2" > unbind
# echo -n "0000:00:10.0" > bind
# echo -n "0000:00:10.1" > bind
# echo -n "0000:00:10.2" > bind
# cd /sys/bus/pci/drivers/ehci_hcd
# ls
0000:00:10.4  bind  new_id  unbind
# echo "0000:00:10.4" > unbind
# echo "0000:00:10.4" > bind

2.6.14-1.1656_FC4smp is still doing this, in case you were wondering. :)

My handy script automating the workaround. --

#!/bin/bash
PCI_DRIVERS=/sys/bus/pci/drivers
UHCI_HCD=$PCI_DRIVERS/uhci_hcd
OHCI_HCD=$PCI_DRIVERS/ohci_hcd
EHCI_HCD=$PCI_DRIVERS/ehci_hcd

if [ -d "$UHCI_HCD" ]; then
        UHCI_DRIVERS=$(find $UHCI_HCD -type l -printf "%f\n");
fi

if [ -d "$OHCI_HCD" ]; then
        OHCI_DRIVERS=$(find $OHCI_HCD -type l -printf "%f\n");
fi

if [ -d "$EHCI_HCD" ]; then
        EHCI_DRIVERS=$(find $EHCI_HCD -type l -printf "%f\n");
fi

# unbind them all

for f in $UHCI_DRIVERS; do
        echo -n "$f" > $UHCI_HCD/unbind;
done

for f in $OHCI_DRIVERS; do
        echo -n "$f" > $OHCI_HCD/unbind;
done

for f in $EHCI_DRIVERS; do
        echo -n "$f" > $EHCI_HCD/unbind;
done

# rebind them
for f in $UHCI_DRIVERS; do
        echo -n "$f" > $UHCI_HCD/bind;
        sleep 1
done

for f in $OHCI_DRIVERS; do
        echo -n "$f" > $OHCI_HCD/bind;
        sleep 1
done

for f in $EHCI_DRIVERS; do
        echo -n "$f" > $EHCI_HCD/bind;
        sleep 1
done

This is a mass-update to all currently open kernel bugs.

A new kernel update has been released (Version: 2.6.15-1.1830_FC4)
based upon a new upstream kernel release.

Please retest against this new kernel, as a large number of patches
go into each upstream release, possibly including changes that
may address this problem.

This bug has been placed in NEEDINFO_REPORTER state.
Due to the large volume of inactive bugs in bugzilla, if this bug is
still in this state in two weeks time, it will be closed.

Should this bug still be relevant after this period, the reporter
can reopen the bug at any time. Any other users on the Cc: list
of this bug can request that the bug be reopened by adding a
comment to the bug.

If this bug is a problem preventing you from installing the
release this version is filed against, please see bug 169613.

Thank you.

Actually, while the update above was scripted, I am indeed interested if
current kernels fix anything. Don't forget to mention exact version tested,
do not use word "current". Thanks...

Closing per previous comment.