LATEST STABLE BOOTLDR IS 2.18.54
LATEST EXTREMELY STABLE BOOTLDR IS 2.18.01

2.21.12: jamey 03-nov-03
  -- built bootldr with arm-linux-gcc 2.95.3 for sa1110 ipaqs
2.21.10:
2003-09-24 18:33  tag BOOTLDR_2_21_10
2003-09-24 18:33  jamey
	* Makefile: 2.21.10
2003-09-24 18:30  nelson
	* vfat.c: Dang!  I thought I'd checked this fix in.  This fixes the
	  "file not found" error, or at least it did some months ago.
2003-09-24 12:46  bavery
	* Makefile, params.h, pcmcia.c: fixed the linuxargs to be 2.4.19
	  compliant, reduced some verbosity
2003-09-23 11:37  bavery
	* Makefile, bootldr.c, btflash.c, ide.c, params.h, pcmcia.c,
	  pcmcia.h, serial.c: aded skiff fixes for timer,pcmcia and ide.
2003-09-23 11:31  bavery
	* skiff.h:  skiff specific stuff
2003-09-17 15:46  bavery
	* load_kernel/src/Makefile:  also work with config.local.mk
2003-09-17 15:46  jamey
	* Makefile: define SHORT_ARCH_STR for skiff
2.21.2: 2003-09-17 13:35  bavery
	* load_kernel/src/: Makefile, jffs2.c: skiff updates, allowed arch
	  flags to reach here and use them to ifdef out lcd spec stuff.
	* serial.c:  skiff updates - made structures for skiff
	* sa1100.h:  skiff updates
	* params.h: skiff updates
	* os_linux.mk: removed LINUX_DIR as it is unused and confuses gcc
	  3.2.3 also, added an auto detect to where zlib is so it works off
	  of the existing toolchain. i.e. it looks in the base dir of
	  arm-linux-gcc /arm-linux/{include,lib} for the zlib stuff
	* mmu-pxa.c: added PXA: to mmu message so we know which is which
	* flashif.c: skiff updates
	* config.mk: skiff updates, added some info to help know what
	  setting to use
	* btpci.c: skiff updates
	* btflash.c: skiff updates esp make atmel CB flash sectors right
	* bootlinux.c:	skiff updates includeing the dc21285: for mtdparts
	* bootldr.c, bootldr.h: skiff updates
	* boot.s:  skif updates, added assembeler based serial print for
	  debugging and added boot_flag space to teh stack
	* Makefile, boot_flags.h, bootcmds.c, bootconfig.h: skiff updates
2.20.5:
2003-09-17 13:04  bavery
	* Makefile: made a bogus miro version 5 to provide a tag point
	  before the large skiff checkin

2003-09-06 15:00  joshua
	* serial.c: RXSTAT error stuff removed. If you want it back,
	  compile with -DCONFIG_CONFUSE_USERS.

2003-08-25 14:11  joshua
	* bootldr.c: memtest2 and prelim 1900

2003-08-25 14:09  joshua
	* Makefile: -DCONFIG_USB, not CONFIG_USB.

2003-07-23 07:52  jamey
	* bootldr.c: updated help strings to indicate receive by ymodem or
	  xmodem

2003-07-08 08:50  jamey
	* architecture.h: added XDA_MACH

2003-07-03 10:54  pb
	* boot-pxa.s: change ldrh to ldr to avoid addressing range problems

2003-07-03 10:45  jamey
	* Makefile: updated makefile to include -sa or -pxa in bootldr
	  binary file names

2.20.4: 2003-07-03 10:40  jamey
	* CHANGELOG, Makefile, boot-sa1100.s: roll back to boot-sa1100.s to
	  avoid making bricks

2003-06-28 14:20  joshua
	* boot-pxa.s: boot-pxa update for h1910
	* boot-pxa.s: we can now check action button on 1910

2003-06-18 21:40  joshua
	* boot-pxa.s: Checked in boot-pxa.s which can detect h1900.
	  ***WARNING WARNING!*** THIS CODE MAY NOT WORK AND MAY FAIL TO
	  BOOT! Please test this with ram-bootldr BEFORE flashing to your
	  device!
	* boot_flags.h, bootldr.c, bootldr.h: added defines for h1900 and
	  axim in headers. now to implement..
	* Makefile: fixed USB so that enabling it doesn't break PXA builds.
	* asm-arm/arch-pxa/: h1900-gpio.h, h3900-irqs.h: added gpios for
	  h1900
	* asm-arm/arch-pxa/: axim-gpio.h, h3900-gpio.h: added axim-gpio.h
	* asm-arm/arch-pxa/pxa-regs.h: oops. stupid __REG.
	* asm-arm/arch-pxa/pxa-regs.h: synched pxa-regs.h up to kernel.
	* dma-sa1100.c, usb-sa1110.c: it seems I have changes that I never
	  committed.

2.20.3: jamey 03-JUL-03
  -- rollback boot-sa1100.s to the version from 2.19.73 because it was
  bricking some h3600's.  Will try to come up with a better solution
  when I have one of the bricks to debug with.
2.20.2: pb 06-JUN-03
  -- support action button on h5400
2.20.1: jamey 13-MAY-03
  -- update to 2.20.x to declare stability
  -- major new bootldr hacking will be in linux as bootldr
USE AT YOUR OWN RISK!!!!
2.19.76: pb 29-APR-03
  -- use ATAG_INITRD2 for initrd so that addresses work out right on pxa.
2.19.75: pb 25-APR-03
  -- added extra (optional) machine-type column to reflash.ctl.
  -- tolerate reflash.ctl not ending with newline.
  -- correct left/right joypad confusion on h3900.
  -- fix pxa cache-flushing code.  enable both I & D caches as standard on h3900.
  -- mark RAM non-bufferable on pxa in deference to writeback cache errata.
2.19.74: jamey 23-APR-03
  -- changed boot-sa1100.s to use same SDRAM config values (MDCNFG and MDREFR) as Linux
  -- applied patch from disconnect for initial reflash save support' reflash.c
  -- reduced verbosity of flash_iohandle_prewrite
  -- allow tabs as well as spaces in reflash.ctl
2.19.73: joshua 9-APR-03
  -- usb send over DMA works for real. sorry.
2.19.72: joshua 5-APR-03
  -- jamey's cut-n-paste-happyness in CHANGELOG fixed
  -- usb send support :)
2.19.71: jamey 25-MAR_03
 -- dual bank probe_ram works on h3900 with dual sdram banks, but
    DABT's on single bank h3900.
2.19.70: joshua 15-MAR-03
  -- experimental USB support added.
2.19.69: jamey 14-MAR-03
  -- build with -march=armv4 so that flashif.c will work with 16bit
     flash
  -- check for armv4 option
  -- set flashword = UNCACHED_FLASH_BASE in ram bootldr
2.19.68: jamey 14-MAR-03
  -- debugged support for reflashing bootldr, hopefully it is pretty safe.
  -- enable ymodem by default
2.19.67: jamey 13-MAR-03
  -- add support for reflashing bootldr, including validating bootldr
     image for architecture
2.19.66: pb 7-MAR-03
  -- reflash can handle multiple entries in one md5sum file.
2.19.65: jamey 6-MAR-03
  -- LCD working again on h3900 but h5400 buzzing incessantly
  -- verified action button detection working on h3900
  -- Joshua Wise added a label for reflash to the splash screen.
2.19.64: jamey 6-MAR-03
  -- LCD working again on h3900 but h5400 buzzing incessantly
2.19.63: jamey 6-MAR-03
  -- boot-pxa.s now detects either h5400 or h3900, so one bootldr
     works on both
  -- unfortunately, LCD stopped working on h3900.
2.19.62: pb 2/28/03
  -- go back to checking action button rather than power button on
     H3900, since the latter interferes with suspend.
2.19.61: pb 2/27/03
  -- print a more useful diagnostic if division by zero occurs
  -- add some validity checking to vfat code
  -- re-enable SDRAM size probe
  -- try to make reflash.ctl parser more tolerant of line endings
2.19.60: jamey 2/20/03
  -- iohandle.c: suppressed printing of .\n's in copy to flash
2.19.59: jamey 2/20/03
  -- changed isValidBootldr not to check for parrot if it is bootldr
     but wrong arch, addr, or checksum
  -- added ipcom patch to support multiple flash banks, needs work for
     h3900 and h5400 but it's a start
  -- added definition of H3900_ASIC3_ID_BASE to autodetect h5400/h3900
2.19.48: 
2.19.57: jamey 2/6/03
  -- started h5400 porting
  -- added patch to wait for RTC to stabilize on SA1100
  -- some cleanup
2.19.56: pb 12/11/02
  -- remove setting of green LED in boot-pxa.s, unnecessary now screen works
2.19.55: pb 12/08/02
  -- initialise ASIC2 GPIODIR on h3800, h3900
  -- default to 16 bit ide buffer reading
  -- use slower default pcmcia timings for pxa
2.19.54: pb 12/07/02
  -- reinstate disabling of mmu, etc, in boot-pxa.s::bootLinux
  -- can boot pocketpc on h3900 now
2.19.53: pb 12/06/02
  -- fix parsing problem with reflash.ctl
  -- reflash supports .gz images now; md5sum not required for those
  -- enable "boot wince" on pxa
  -- ifdef out redundant crc32 implementation
  -- support ctrl-u on serial console cmdline
2.19.52: pb 11/27/02
  -- write dram_size0 in probe_ram_size.  I think this will fix the problem
     with only 32MB SDRAM being reported on StrongARM machines.  Dual bank
     support still looks like it is broken.
  -- configure MSC0 on pxa to make flash accesses faster
2.19.51: pb 11/25/02
  -- more powerful syntax for reflash.ctl, largely untested
  -- 3900 version compiles again now, whoops
  -- fix buffer overrun in SA-1100 lcd driver
  -- honour params setting for autoboot time on pxa
2.19.50: pb 11/24/02
  -- add support for YMODEM-g ("set ymodem_g 1")
2.19.49: pb 11/22/02
  -- turn on 3600 frontlight too
  -- write MSC1/2 in boot-sa1100.s to get them right during resume.
  -- embedded reflash seems to work now.
2.19.48: pb 11/21/02
  -- turn on 3900 frontlight
  -- start implementing embedded reflash support
  -- support sleeves and PCMCIA on 3900
2.19.47: pb 11/17/02
  -- light up 3900 green LED during boot to stop users thinking they 
     have bricks.
  -- first cut at LCD/front panel button support for 3900
2.19.46: pb 11/15/02
  -- add ability to read/write sector buffer either 8 or 16 bits at a time
     via  'ide 8bit' and 'ide 16bit' commands
2.19.45: pb 11/14/02
  -- add more data sections to linker script so DRAM copy works right.
2.19.44: pb 11/14/02
  -- suspend/resume works on pxa now
2.19.43: pb 11/12/02
 more pxa changes:
  -- set up GPIOs earlier; check action rather than power button.
  -- first cut at resuming from sleep, almost certainly not sufficient.
2.19.42: pb 11/11/02
  -- implement "reset" command on pxa
  -- copy wince partition check code from boot-sa1100.s to boot-pxa.s
  -- allow pxa to autoboot Linux if power button not held down
2.19.41: pb 11/06/02
  -- on pxa, select turbo mode and execute FCS after configuring clocks
2.19.40: cdm 11/03/02
  -- Fixed the boot_qnx command to use bootLinux() to properly manage
     the i+d caches and pipeline flushing.  Removes the UDEF exceptions.
  -- Fixed some other misc. QNX build issues due to recent changes.
  -- Added in a NO_HILITE_BUTTON macro in aux_micro.c to stop buttons
     presses from inverting areas of the screen.
2.19.39: ??  ??/??/02
  -- Empty log message for the changes not yet logged.
2.19.38: jamey 10/17/02
  -- make sure vioh->vfat is initialized in vfat.c
  -- pass $(CROSS_COMPILE)$(CC) to load_kernel/src/Makefile
  -- remove old mtdpart= support and add rootfstype param
2.19.37: jamey 10/17/02
  -- copied in updated vfat code from stable branch
2.19.36: jamey 10/17/02
  -- added code to enable uart transceiver on h3800
2.19.35: jamey 10/16/02
  -- added a print of a couple of characters on resume to debug resume hang
2.19.34: jamey 10/16/02
  -- added a couple more debug messages in vfat
2.19.33: jamey 10/8/02
  -- CONFIG_ARCH_PXA seemed to conflict with gcc internal symbols, so back
     to CONFIG_PXA, etc
2.19.32: jamey 10/8/02
  -- for each architecture SA110, SA1100, PXA, use CONFIG_aaa
  -- for each machine type, use CONFIG_MACH_mmm 
2.19.31: jamey 10/3/02
  -- removed deprecated atag ptable code
2.19.30: jamey 10/2/02
  -- added jffs2 source/sink and ide partition source/sink for the copy
     command, e.g., 
        copy root:/etc/passwd hda1:/etc/passwd
        copy root hda1
        copy root hda1:root.img
2.19.29: rescinded
2.19.26: jamey 9/04/02
   -- removed requirement that MMU be enabled for PXA ARCH 
      (see MaybeCopyToDRAM in boot-pxa.s)
2.19.25: cdm 9/03/02
   -- QNX boot logic moved from boot_kernel to command_boot_qnx().
   -- Fixed up QNX location macros to be based from DRAM and FLASH BASE.
2.19.24:  bavery 8/21/02
	-- pxa version boots linux kernel (doesnt come all the way up but it does boot it) 
	-- have caches turned off again since they seemed a bit unhappy in the flushing/off routine 
2.19.23:  bavery 8/21/02
	-- pxa version can write to flash...
2.19.22:  bavery 8/21/02
	-- fixed pxa cversion to get to prompt.
2.19.21:  bavery 8/21/02
	-- pxa version of bootloader comes up to the prompt :)
	-- alot of !ifdefs added to allow the pxa to build -- some errors WILL result
	-- allowed CONFIG_LCD=n to build
2.19.18: jamey 8/17/02
        -- added a probe_ram command for dual dram bank ipaqs 
        -- fixed code in bootlinux that was disabling bank0 instead of bank1
2.19.17: bavery 8/17/02
        -- backed out dual bank sdram stuff from bootldr.c so the splash
           screen comes back  
2.19.15: jamey
        -- added code to pass in second dram bank params to linux 
        -- aux micro cleanup
2.19.zz: bavery 8/12/02
        -- pxa250 full function uart can now send characters 
2.19.08: jamey for vaibhav
        -- cleaned up SA1100 entries to unify with PXA250 
2.19.07: jca 8/5/02
	-- rest of dynamic jornada56x machine detection merge
2.19.06: jamey 8/2/02
        -- incorporated jca's code for dynamic jornada56x machine detection
        -- CONFIG_MACH_IPAQ now includes jornada56x
2.19.04: jamey 8/2/02
        -- added code to unmount and eject the sleeve before jumping to kernel in boot_vfat 
2.19.03: bavery 8/1/02
        -- some h3900 work.  do not use the h3900 build unless you have jtag set up 
2.19.02: jamey 7/30/02
        -- added param boot_vfat_partition, which says which ide partition
        to mount vfat on when booting vfat.  Numbered from zero.
        -- renamed param use_mtdparts to use_mtd_cmdline
        -- changed param use_mtd_cmdline to default to 1.  Requires
        drivers/mtd/cmdline.c in linux kernel.   
2.19.01: jamey 7/8/02
	-- applied jornada720 patch from jca
2.19.00: jamey 7/8/02
        -- started adding pxa250 xscale support
============================================================
2.18.55: does not recognize flash on h3100
2.18.54: jamey 10/17/02
        -- make sure vioh->vfat gets initialized. 
2.18.53: jamey 10/17/02
        -- fixed bug comparing attr when should have looked at name[0]
        -- introduced struct vfat_iohandle to clean up ioh->pdata for vfat iohandles 
2.18.52: jamey 10/17/02
        -- reduce vfat verbosity with debug
2.18.51: jamey 10/17/02
        -- more verbose debug message
2.18.50: jamey 10/16/02
	-- added some debug messages to vfat_find_file_in_dir 
2.18.49: jamey 6/12/02
	-- added a command to load backpaq fpga
2.18.49: jamey 6/6/02
        -- removed the crc checking when scanning jffs2, got a 6 fold
	   speedup in one test case
2.18.48: jamey 6/6/02
        -- applied patch to ide that swaps bytes in ide strings
2.18.47: jamey 6/6/02
	-- added jffs2 read command
2.18.46: cdm 6/3/02
	-- updated for qnx 
2.18.45: jamey 5/17/02
	-- one more tweak to the fix to the patch to fix both vfat read and write 
2.18.44: sachac 5/8/02
        -- fixed vfat copy and write again.
2.18.43: jamey 5/7/02
        -- applied sacha's long filename fix 
2.18.42: sachac 4/23/02
        -- copy command seems to work both ways. Have fun. 
2.18.41: jamey 4/23/02
        -- copy command seems to be working from vfat to flash but not vice versa
2.18.40: jamey 4/23/02
        -- stopped modifying PWER because we think PocketPC2002 uses it to
           detect resume 
        -- some serial.c cleanups
2.18.39: jamey 4/18/02
        -- added copy command, which does not quite work yet
        -- modified vfat commands so that they do the sleeve insert, pcmcia
           insert, vfat mount automatically
        -- added some more error checking to the vfat commands   
2.18.38: jamey 4/9/02
        -- have vfat read/write print how many bytes were read/written 
        -- 'load <partname>' prints an error if partition is undefined.
2.18.37: jamey 4/5/02
        -- initialize MSC1 and MSC2 so that we can be sure to control the gpio's 
        -- I think this fixes the bricking problem that arose in 2.18.22
2.18.36: jamey 4/3/02
	-- gpio tweaks to make sure RS232 transceiver is working
        -- sleeve on command for debug 
2.18.35: jamey 3/29/02
	-- silenced flash write for h3100 
	-- sachac and disconnect made a new splash screen
2.18.34: jamey 3/28/02
        -- make sure egpio initially turns on RS232
        -- do not initialize sleeve or pcmcia if we might be in safe mode  
2.18.33: jamey 3/28/02
        -- included George G. Davis's patch to force use of formatted sector size (Jamey)
2.18.32: sachac 3/28/02
        -- fixed boot vfat [filename] [args...] bug 
2.18.31: sachac 3/28/02
        -- added IDE and VFAT write support (no reclustering)
2.18.30: sachac 3/27/02
        -- added VFAT long filename support
2.18.29: jamey 3/26/02
        -- added sector_size to struct iohandle
        -- debugging other cf flash cards
        -- cleaned up cis read code  
2.18.28: jamey 3/26/02
        -- trying new wince cookie value
        -- use egpio code backported from h3600 linux module
2.18.27: sachac 3/26/02
        -- added lcdzfile command in jffs2_commands.c, suggested by prh (no bounds checking)
        -- added params splash_filename and splash_partition that control custom splash
        -- removed lcdimg command (tends to crash bootloader)
2.18.26: jamey 3/21/02
        -- stopped writing to PSPR because it interfered with suspend/resume on H3800 
        -- fixed declaration of serial in serial.h and usage in bootldr.c (now a pointer)
2.18.25: jamey 3/20/02
        -- removed dependence on linux source tree
        -- moved linux booting support to bootlinux.c
2.18.24: jamey 3/19/02
        -- added fat12 support
        -- have SUBCOMMAND prepend cmd and subcmd onto help string
2.18.23: sachac 3/18/02
        -- moved to new command execution machinery
        -- moved serial support to serial.c 
        -- started ir uart implementation
        jamey 3/18/02
        -- added struct serial_device
        -- added h3800_get_option_detect to hal.c  
2.18.22: jamey 3/15/02
        -- have vfat mount fail immediately for FAT12
        -- added param use_new_commands to select new command execution
           machinery written by sachac
        -- COMMAND and SUBCOMMAND macros now declare prototypes   
2.18.21: jamey 3/11/02
        -- do not modify MSC1 or MSC2, these should be handled by kernel drivers
2.18.20: jca via jamey 3/8/02
        -- added jornada 56x support.  Not ready for testing by others. 
2.18.19: sachac via jamey 3/8/02
	-- added more command and subcommand declarations 
2.18.17: sachac via jamey 3/7/02
        -- added command definition similar to the new param structure,
        -- enables definition of commands in files other than bootldr.c
2.18.16: jamey 3/6/02
        -- fixed passing of initrd at 0xc02080000
2.18.15: jamey 3/5/02
        -- boot vfat takes precedence over implicit boot wince 
2.18.14: jamey 3/5/02
        -- bound Q button to 'boot vfat'
2.18.13: jamey 3/5/02
        -- debugged vfat directory and file reading
        -- added 'boot vfat' command
2.18.12: jamey 3/4/02
        -- was able to boot ide using 2.4.3 kernel and initrd supplied by Tangent (but no rootfs)
2.18.11: jamey 3/4/02
	-- applied cleanups from jca@alum.mit.edu (split some btflash functions into flashif.c)
2.18.10: jamey 3/3/02
        -- applied correction from Jim Pick for reading partition tables
	   from flash
2.18.9: jamey 3/1/02
	-- change 'boot ide' not to read initrd partition if noinitrd is in linuxargs
2.18.8: jamey 2/14/02
        -- added pspr_wince_cookie to keep track of whether to resume wince
           or else just to jump to pspr on resume (i.e., linux)
        -- made 'boot ide' take preference over booting wince from flash
        -- started work on vfat   
2.18.07: jamey 2/12/02
        -- fixed partition truncation problem in ide_read_partition
        -- added 'program partition srcaddr' command to program flash from dram
2.18.06: jamey 2/12/02
        -- dual pcmcia sleeve detected and controlled
        -- pcmcia_insert() implemented
        -- pcmcia_insert() automatically calls ide_attach()
        -- 'boot ide' command does not require 'sleeve insert' and 'pcmcia insert' commands.  
2.18.05: jamey 2/7/02
        -- now passes initrd from second partition on disk
2.18.04: jamey 2/7/02
        -- boot ide will now read dos-partition-table from ide, and boot zImage from first partition on drive  
2.18.03: jamey 2/6/02
        -- pcmcia cis read implemented, detects ide funcid
        -- first cut at ide implementation, partition table reading 
2.18.02: jamey 1/30/02
        -- applied jornada720 patch from jca@alum.mit.edu
2.18.00: bavery 1/22/02
	-- transposed the 3600 display so its orientation
		matches everyone elses.
2.18.00: bavery 1/17/02
	--Feature version change to indicate gzip support
	--help system fixed
	--debug style menu commands hidden behind a DEBUG_BOOTLDR_CMDS flag
	--new splash command including the action button hint
******************************************************************
2.17.23: bavery 1/10/02
	-- added the ability to download gzipped images to flash.
		This only works with load <partition> currently.
2.17.22: Jamey Hicks 1/10/02
	 -- added pcmcia memory and io mapping routines
	 -- added power control for single CF and PCMCIA sleeves
	 -- implemented struct list_head according to usage in sleeve driver 
2.17.21: Jamey Hicks 1/10/02
	 -- added code to configure sdram bank1 (for 256MB ipaqs) (off by default)
	 -- added constraints to bootldr.ld.in per George Davis -- to
	 reduce size of bootldr when built with other (montavista?) toolchain
2.17.20: Brian Avery 1/8/02
	-- latch 1 button nicely during the jffs2 scan
2.17.19: Jamey Hicks 1/7/02
        - split array of bootldr params into separate declarations
        - bootldr ld script collects them (they can be in separate files)
        - PARAM() macro can be used to define them
        - get_param("name") -> &param_name
        - get_param_value("name", &var) -> var = param_name.value
2.17.18: Brian Avery 1/7/02
	-- buttons now checked during jffs2 scan
	-- graphical feedback added for buttons and scan progress
2.17.17: Jamey Hicks 1/3/02
         -- added pcmcia driver framework
2.17.16: Jamey Hicks 1/3/02
         -- added msleep() uses OSCR
         -- added sleeve driver framework
         -- added sleeve init, sleeve insert, sleeve eject commands
2.17.15: Brian Avery 1/3/02
	-- fixed the 3100 (mono) lcd display code so that the linux
		framebuffer appears correctly
2.17.14: Jamey Hicks 1/3/02
	 -- added code to pass mtd partition info on kernel cmdline
	 -- will remove the ATAG_PTABLE code once kernel supports mtdpart= on cmdline
2.17.13: Jamey Hicks 12/28/01
	 -- added beginning of hardware abstraction layer for H3100/H3600/H3800
2.17.12: Brian Avery 12/18/01
	-- ok, we are back to using the action button to force us into the
		bootldr.  The action button works for all models.
2.17.11: Brian Avery 12/18/01
	-- lcd screen now works for the 3800's.
	-- can turn off lcd for 3100's now too
2.17.10: Brian Avery 12/18/01
	-- lcd screen now works for the 3100's.
	-- made splash screen true 16 bit colour in prep for the 3800
2.17.9: Jamey Hicks 12/18/01
        -- changed xmodem, ymodem, and autoboot timeouts to be measured in seconds
        -- ymodem works with minicom but not hyperterminal
2.17.8: Jamey Hicks 12/17/01
        -- integrated most of John Dorsey's patches, including Assabet,
        md5sum, ymodem, and his jffs/jffs2 support
2.17.7: Jamey Hicks 12/14/01
        -- applied Sean Godsell's patch to reduce RTC runahead during reset
        or resume.
2.17.6: Brian Avery 12/13/01
	-- fixed bug in loading/recognizing German and French
		parrots/wince partitions
2.17.5: Brian Avery 12/12/01
	-- partition table now ordered by base address
	-- 3800 gets an asset partition tacked on the last
		flash sector
2.17.4: Brian Avery 12/12/01
	-- Bootloader now recognizes german wince partitions and has a 
		more liberal is this a parrot bl check.		
2.17.3: Brian Avery 12/12/01
	-- tweaked  RTTR initialization in boot-sa1100.s
	-- added a capabilities bit for 3800 support so BootBlaster and 
		and the Bootloader checks it.
	-- ALSO, NOW YOU NEED TO HOLD THE POWER KEY DOWN TO ACCESS THE BL
		 **NOT** THE ACTION KEY
2.17.2: Jamey Hicks 12/11/01
	-- moved RTTR initialization very early in boot-sa1100.s
	-- probably needs to be tweaked for wince
2.17.1: Brian Avery 12/5/01
	-- added support for the Ipaq 3800
2.17.0: Jamey Hicks 12/5/01
	-- changed version number because 2.16.25 changes boot behavior in
	an incompatible way -- 2.4.7-rmk3-np1-devfs up to hh13 will no
	longer boot on H3100

2.16.25: Brian Avery 12/4/01
         -- added a discovery function to dertermine 31/36/38xx models and 
		use that to set mach_type
2.16.24: Jamey Hicks 11/27/01
         -- use ATAG_PTABLE_COMPAT for now
2.16.23: Jamey Hicks 11/27/01
         -- moved bootldr tag_ptable partition table defs to linux kernel
            source include/linux/mtd/bootldr.h 
2.16.22: Jamey Hicks 11/27/01
         -- make mach_type not conditionalized by CONFIG_MACH_SKIFF'
2.16.21: Jamey Hicks 11/27/01
         -- added mach_type parameter: as of 2.4.15-rmk1, H3100, H3600, and
         H3800 distinguished by mach_type
            mach_type  22 for H3600 and H3700
            mach_type 136 for H3100
            mach_type 137 for H3800

STABLE BOOTLDR:
2.16.19: Brian Avery 11/2/01
	-- Behaviour change-> if wince is installed we jump to it 
		directly after a reset. This maintains parrot's 
		pseudo soft-reboot fxnality.  If you hold down the 
		action button you jump to our bootloader.  From here
		if you boot wince, you WILL clear dram.  If wince is not
		detected, the bootloader behaves as before.
	--  if the action button is held down (wince or linux), we will not
		autoboot.
2.16.17: Brian Avery 11/2/01
	-- load bootloader will now upload an OHH bootloader OR a parrot bootloader.
2.16.16: Brian Avery 11/1/01
	-- shortened default timeout to jump to os param
	-- put in smarts to rescan the flash if things have changed.
		--e.g. ls, reflash root, ls will result in another scan now.
2.16.12: Jamey Hicks 10/18/01
	 -- merged 2.16.11 onto cvs head branch
2.16.10: Brian Avery 10/16/01

2.16.10: Brian Avery 10/16/01
	-- prevets flash operations other than load bootldr when 
		running from DRAM
2.16.8: Brian Avery 10/15/01
	-- can load splash screen from a file 
		-- supports binary ppm (P6) and binary pgm (P5) file formats
	--re-enabled splash screen during jffs2 partition scan, overhead now 5%
2.16.0-2.16.6: Brian Avery 10/12/01
 - boots wince and linux.  suspends and resumes wince and linux and the bootldr.
 - autoboots wince if it is there.

============================================================
2.15.19 -- Jamey Hicks 10/10/01
- turn on the LCD after scanning the root partition for params
- look for wince at 0x40000, do not attempt to scan jffs2 if found

2.15.18 -- David Neuer 10/1/01
- make root partition not overlap bootldr on monochrom configuration

2.15.17 -- Jamey Hicks 9/28/01
- increase bootldr partition to 0x40000 on monochrome configuration 

2.15.16 -- Brian Avery 9/27/01
- added icache on by default
- made the bootldr,root partition the default one.

2.15.15 -- Brian Avery 9/26/01
- corrected params file reading bugs
- added cat command

2.15.14 -- Brian Avery 9/21/01
- converted boot_linux to pass params in a tags list
- enables bootldr to read params and partition table from root partition
  /boot/params rather than a sector

2.15.13 -- Brian Avery 9/19/01
- removed problematic enabling of cache at init time
- sped up jffs2 scan for empty fs

2.15.12 -- Jamey Hicks 9/18/2001
- initialize RTTR in bootldr.c so RTC will continue to count at 1Hz

============================================================

2.14.15  -- Jamey Hicks 8/19/01
- made minor change to params_save so that saving parameters formatted
PF_DECIMAL would not truncate the parameter list.  There is an probably 
still an issue with strlen, as pointed out by Dave Panariti.

2.14.14 -- Dave Panariti 6/22/2001
- Added kpart=xxx parsing to boot flash to allow specifying a kernel
  partition on the command line.

2.14.13 -- Dave Panariti 6/22/2001
- Added ability to run multiple commands separated by ';'
- Added param which names the partition in which to find the kernel.

2.14.12 -- Dave Panariti 6/14/2001
- Fixed to to say 8b/color rather than to imple 8bpp.
- Removed debug code in lcdzimg which prevented it from converting the
  image to the lcd's color format.
- Added back the CONFIG_GIMP_SPLASH_PNM vars to os xxx.mk which someone had
  nuked.
- Added target to make a raw compressed splash image for use with lcdzimg
- Added some doc to makefile about splash image format, etc.
- Added os_freebsd.mk so I can build at home again.

2.14.11 -- Dave Panariti 6/13/2001
- Made parameter execution overridden if booting verbosely (becoming
  debug boot)
- Added cvs_ver variable and tagver target to allow easy tagging w/
  make
- Made params save smarter about strings: now does strcmp.
- Found bug in wince-bootldr w.r.t params save.  In this case, the default
  params, which should point into flash at the original version of the
  params, ends up pointing somewhere else distinctly non-default-paramish.
  This needs to be fixed.  For the c000 (wince) bootldr, there is no flash
  copy of the params for reference.  Perhaps we can make a copy of the
  initialized params and use that as the reference in wince-bootldr.
  A function to get the "defaults" will need to be conditionalized to
  return the proper thing.  And it should be const.
- Added -n option to params save to have it just show what it would
  do. It also dumps the vars used to find the default params.

2.14.10  -- Brian Avery 6/7/2001
- partition save command now calls command_params_save as it ought to.
- renamed the save [kernel World bootldr...] command to upload[ ...] 
- if you try to load in a kernel larger than your kernel partition size 
	there are now enough copies of the error message to overcome 
	minicoms desire to eat them.  

2.14.8 -- Jamey
- dram_size was being probed but not saved in the dram_param

2.14.7 -- Jamey
- moved set_vppen and clr_vppen to btflash.c

2.14.6  -- Chris McKillop 5/11/2001
- Added new parameter values: domainname, dns1, dns2 and netcfg.  The first three 
  are pretty obvious and the last can be used to flag for manual or dhcp network
  configuration.
- Updated the build process to use OS specific makefiles (os_<name>.mk).  


2.14.5  -- Jamey 5/11/2001
- added "jffs2" as valid flags argument for define partition
- call params_eval() after load_params

2.14.4  -- Jamey 5/11/2001
- eflash <partitionname> now erases the whole named partition
- partition define <partitionname> <base> <size> 16 
  defines a JFFS2 partition.
- load <partitionname> erases the whole partition if it is JFFS2, and marks
  the unwritten sectors with jffs2_sector_marker0, jffs2_sector_marker1, jffs2_sector_marker2,
  saving and erase cycle on the sectors.

2.14.3
- merged in mitsy mods

2.14.2
- fixed btflash_reset_partitions so it initializes default partitions even
if no partition table is loaded into params partition.
- reduced xmodem_timeout and xmodem_initial_timeout to 100000.  We might
need to increase xmodem_initial_timeout to 100000 or 300000 for minicom.
3,000,000 causes hyperterminal to fail to transfer.

2.14.1
- fixed dwordtodecimal so it prints '0' for 0 and not '' for 0.  (Tested
before updating version number.)

2.14.0 Jamey 5/3/2001
- renumber to 2.14.0 because removal of headers and addition of partition
tables is actually a relatively signicant change

2.13.8 Jamey 5/3/2001
- the load and store commands now use the partition table to upload and
download regions by name
- the kernel no longer has a 12 byte header

2.13.7: Jamey 5/2/2001
- implemented flash partition table

2.13.6: Jamey 5/2/2001
- removed the special partitions other than bootldr, params, and kernel
- others will be supported by the currently unimplemented define_partition command

