$Date$ First created
	10/20/04
	Added sql/ directory for MySQL database of stp results.

	10/15/03
	Merging changes from SuSE.
	
	05/21/03
	setup.sh fix from akpm
	04/14/03
	Fixed installation routine - files now installed in /usr/local/reaim, 
	binary in /usr/local/bin. 
	Tested. 

	04/09/03

	Added workfile arg - 
	Altered shared_memory - increased list of sizes[] - now goes to
		65536 by powers of 2. 
	03/30/03
	Very slow at the change log
	All disk functions now added 
	Speed of disk functions totally depends on FILESIZE param
	10k is good 

	03/17/03
	Code is now compiling
	Replaced static cmdargs[] with single linked list, 
	using a global_list pointer for register_test 
	Removed the dummy NOCMD test, i think this is the 
	end of the static data. 

	Fixed a few issues with the list removal. 

	added command line args. default with no args is 1->5 users. 

	Need to add adaptive timers, crossover test/switch
This will be a long list.

Problem 1. Too many static global data structures, number
	of tests restricted by mystery #defines

	- Replaced the cmdargs array with a linked list
	- linked list now holds both the cmdargs and hit
	count information. 

	- Increased the number of parameters passed to 
	functions in many cases. 

	added aimlist.h for the linked list

Problem 2. Signal handling reported to produce errors.

	Partial solution: use libltp sig_tst function
	to catch unexpected signals
	
	Replaced various signal handlers with the tst_sig()
	function from libltp calling one sighandler.

	I don't think floating point errors are a big
	concern today, so that signal does nothing

	TODO: Look at tst_exit()


	RANTS

	child signal handlers, sets to SIG_IGN, then just
	returns..la de da..shouldn't this exit()??

	line 699 origonal - if args != NOCMD proceed, if not,
	go ahead anyway with invalid pointers. 

	Replaced old-stye testing of wait() with modern macros
	On SIGHUP, there is a call to wait, this needs to be traced with
	a debugger - does the wait() ever return?

-----------------------




