Unless otherwise noted, these patches are intended for the 2.6 kernel.  2.4 is
usually quite different, so it is unlikely that a patch targetting another
kernel version will work without modification.  If you are unsure whether a
patch will apply cleanly, the --dry-run argument to patch(1) is your friend.

---

enable-hidden-HPETs.patch

	* This patch enables the HPET on NForce and VIA s939/am2 motherboard
	  chipsets.  It unifies the quirk handling for both chipsets into a
	  common framework.  This patch is influenced by lots of rather less
	  clean hacks that are floating around, but they should be cleaner and
	  less buggy, as I try to handle all cases of HPET BIOS ignorance.

	  I got the idea of tackling this problem from a patch that enables the
	  HPET for particular NForce machines by Mikko Tiihonen
	  (http://lkml.org/lkml/2007/4/15/119).  I had originally thought that
	  the HPETs on these machines were broken rather than merely badly
	  documented.

	  This patch has been tested to work on Asus A8V and Asus K8N
	  motherboards.  Other motherboards based on the same chipsets (VIA8237
	  and NF4) should work, but haven't been tested, since I don't have
	  access to other hardware.

---

queued-spinlocks*.patch

	* Changes spinlocks so that rather than starting at lock == 1 and
	  decrementing on each acquisition (checking for contention with
	  a sign bit check), locks function by incrementing one counter
	  on lock aquisition, and another paired counter on lock free.
	  This approach allows CPUs to move past the lock in a FIFO
	  manner, and should reduce cacheline thrashing.

	  These patches (and the entire concept) are based off of Nick
	  Piggin's work: http://lkml.org/lkml/2007/3/23/71

	  I've merely provided asm optimized versions for i386 and x86-64
	  instead of the current mostly-C code.  x86-64 is tested and works.
	  i386 is basically the same code and should also work, but is
	  untested (I have no SMP i386 machines).

---

Other patches used to be here, but they are now obsolete.