1999-10-12  Andrea Arcangeli  <andrea@suse.de>

	* lil.c: Merged a patch from Rogier Wolff <R.E.Wolff@BitWizard.nl>
	to add a module parameter to allow lil to remains insmodded into the
	kernel. Thanks Roger ;).

1998-09-10  Andrea Arcangeli  <andrea@e-mind.com>

	* lil.c: trimmer_time() need to be run clied() on the local CPU.
	* lil.c: irq_sample() now know that the irq latency has to be longer
	than the trimmer_time(). This will trap hw errors or strange cache
	behavior.

1998-09-08  Andrea Arcangeli  <andrea@e-mind.com>

	* lil.c: updated email.
	* README: updated email.

1998-09-05  Andrea Arcangeli  <arcangeli@mbox.queen.it>

	* lil.c (irq_train): trimmer_time() will not put the code in L1
	cache before continue since the real code don' t neecessary run
	from L1 cache too. trimmer_time() is calc for every irq_sample().

1998-09-04  Andrea Arcangeli  <arcangeli@mbox.queen.it>

	* lil.c: irq_mean need to be volatile to be sure that the asm will
	not more use FPU to pass parameters to printk after the RESTORE_FPU().

1998-09-03  Andrea Arcangeli  <arcangeli@mbox.queen.it>

	* lil.c: calc mean using FPU and output 1000000 * mean since
	printk() doesn' t handle floating point format.
	* lil.c: fixed the signal handling to allow insmod to be interrupted.
	* Version 5 released.
