changelog

Linux下操作Bios的驱动

ChangeLog for /dev/bios 

** 02/18/02 * 0.3.2 **************************************************

 * change cs5530 driver to map high rom range instead of low one
   and don't use positive decode.
 * remove ruffian flag. Alpha (164LX/UX) almost works with pc code.
 * don't rely on register defaults in intel 8xx driver.
 * updated pci device list. more entries, join amd and via entry.
 * fix error handling in chipset detection.
 * add support for Reliance/ServerWorks chipsets
 * enable 1M 512k on intel 4x0 chips where it's possible
 * cleanup proc file handling

** 02/17/02 **********************************************************

 * rewrote chipset initialisation skeleton.
 * fix pci bios (un)mapping.
 * experimental support for AlphaPC 164UX (Ruffian)
   (probes at 0xfffffffffffc0000 instead of 0xfffffc87C0000000
 * initial code for FWH mode chips
 * Fix Toggle-Until-Ready code.

** 02/16/02 **********************************************************

 * iounmapping fixed. no more address space wasted.
 * /proc/bios shows physical address now. dmesg shows 
   physical address and virtual memory area and offset.

** 02/13/02 **********************************************************

 * added i820/i830 chipset support
 * added AMD 751/760MP(X) support
 * added support for Itanium and 84460GX chipset
 * added experimental support for some flash chips (ST, Intel, 
   Winbond)
 * use spinlocks instead of hard cli()
 
** 02/11/02 **********************************************************

 * added GPL licence tag 
 * remove low bios area access tweaking for intel drivers
 * speed up SST 39SF020 write
 * fix compilation for 2.5 kernels

** 02/05/02 **********************************************************

 * added support for cs5530 (nsc/cyrix mediagx) chipset
 * reorganized shadow/wp handling
 * probe for 2mb high memory area instead of 256k only 
 
** 08/01/01 * 0.3.1 **************************************************

 * compiles and works with Linux kernel 2.4
 * rewrote flash chip probing
 * always use ioremap now
 * flash chips above 128k should work transparent
 * Support for newer VIA chipsets

** 10/15/00 * 0.3.0pre1 **********************************************

 * added patch from Eric Estabrook
 * support for 256k flash chips on intel 430/440 chipsets and via vp3
 * split up source into several files
 * Changes for Ruffian AXP machines. Does not work (yet).

** 07/29/99 * 0.2.101 ************************************************

 * Oh well.. 11 months? Impossible. I am a lazy guy. Implemented
   some support for VIA Apollo VP3. Don't know whether it works, since
   I don't have one.

** 09/06/98 **********************************************************

patches by [e-mail]:
 * The pointer to bios_release in bios.c was on the flush pointer's
   position. This caused Oopses.
 * When bios_read was called with a file position after the actual end
   of bios, it tried to read non-existant memory positions due to size
   being unsigned (it isn't anymore) , causing spontaneous reboots on
   my system

** 08/22/98 **********************************************************

 * Well,.. The diskless spectacle (0.2.100) was caused by a little bug
   in in handling Intel PCI chipsets. Works now.
 * Threw out the chipset_backout stuff. the PCI chipset handling should
   always leave the machine in the same state it was before. ALWAYS.

** 08/18/98 * 0.2.100 ************************************************

 * Threw out the mem_type stuff. There are more important things than
   this.
 * Argh! After flashing fine on an Intel 28F001BT, the computer kept
   hanging in an endless loop and refused writing the emergency boot
   block to the end :-( There's some work until 0.3 is ready.
   Implemented a timeout so that the system will not hang forever if
   the flashchip behaves unexpected.
 * Removed x86 probing in a loop. I think it never found anything else
   but the system bios and *maybe* the graphics adapter bios. On the
   other hand, it reconfigures some networking cards to silence.
   Bad thing on diskless Linux boxes :)

** 08/15/98 **********************************************************

 * added some changes for intel to compile without warnings..

** 08/02/98 **********************************************************

 * What a boring job! Checked some dozen of flash chip entries today
   and added a lot of new ones. I bet it gets hard to find anything
   this driver does not know.

** 07/28/98 **********************************************************

  * Yeah! Atmel Chips finally work.. These Atmel guys are really weird.
  * Testing last instead of first written byte now, when polling for the
    end of a write access.

** 07/28/98 **********************************************************

  * Well, I am definitely spending too much time in IRC, but detecting
    PCI cards' bioses works now (at least for me)
  * Thrown out some obsolete stuff.
  * Declared PCI and Flash reading/writing __inline__. Don't know,
    whether this is a good idea. But let's try it for a while.
  * Aaaargh! Some major mistakes in handling whether a flash has to
    be erased before programming. FIXED!
  * Even worse. An endless loop made it into writing in 0.2.99. Sorry!
    I had no chance to test writing on an intel board with that release.
    At least my warning, not to write, made sense.
  * Intel flashchips are supported now!! It's at least tested on my
    Alpha AXP LX164 Board (1MByte i28f008 chip) But all Intel flash chips
    seem to work in the same way. 
  * Atmel 64kByte flash chips supported.

** 07/27/98 **********************************************************

  * Split up flash_probe in 2 parts to be able to expand probing on
    PCI bioses and others correctly.
  * Turned around 1st and 2nd probing codes. This is funny, Atmel 
    Flashroms give some wrong numbers if they are probed with the
    0x80/0x60 way. I only hope that no flashchips react on the
    0x90 method with wrong values.

** 07/19/98 * V0.2.99 ************************************************

  * Reading the flashchip works now on Alpha AXP (at least on my LX164
    Board)
    Writing ought to work, too, but Intel Flashchips are not supported
    yet. This should be done until 0.3.0.
    NOTE: I have no idea whether this driver still works on intel
    boards or not. There have been too many changes. Please try, but
    do not flash with this release of the driver.
  * Minor Changes and fixes. Naming scheme changed a bit. This version
    might work on James Mastros' machine again ?!?

** 07/11/98 **********************************************************

  * Started porting stuff to Alpha AXP architecture to continue testing
    the flashing routines. We have a lot of tests next week, so I
    won't get much stuff done..
    Porting to AXP seems to be much more work than I thought. It may
    take some time until the next version is released.
  * Moved major number again. This time we have an official major
    number for /dev/bios. Thanks to Hans Peter Anvin.
    (Well, we have this one since May 1st, sorry for the delay)

** 06/26/98 * V0.2.95 ************************************************

  * added all Manufacturer IDs from the JEDEC standards publication.
  * sorry for not having released a new version since months, but
    my x86 machine died and I have no chance to do any testing right
    now. I guess I must get a new Intel box, as Alpha AXP are all
    delivered with the same Intel flash chips.

** 04/30/98 * V0.2.9 *************************************************

  * removed ioctls. They have been really unneccesary and did not fit
    into the new driver layout.
  * cleaned up the code. Hey, it should be readable again.
  * Moved device minors from 10+ to 0+
  * Rewrote most of the documentation
  * changed intel shadowing routines. Now original values are saved
    and shadowing is turned off for 0xc0000 to 0xdffff, too (This
    was needed to support 2MBit system bios flash chips. Thanks again
    to Matthew Harrell for intensive testing.
  * Removed dirty hacks from bios_read_proc()
  * Added some fields to struct flashdevice to support all ROM types,
    not only flash roms. Probing for other types still missing.
  * Implemented probing for some strange Winbond chips (0x80/0x20).

** 04/27/98 * V0.2.8 *************************************************

	*** Attention *** This version has a lot of changes since
            0.2.7, so be very careful, when testing. Things may 
                        be broken that used to work.

  * Rewrote big parts of the driver to (theoretically) support
    multiple flash chips and/or ROM areas.
  * Tried to implement support for 2MBit System BIOS chips, but
    I have no idea, whether it works. I don't have one.
  * added some more OPTi, SiS and VIA PCI chipsets to chipset list.
    They have no function yet, though.
  * Some weird computers have an ISA bridge, but don't have it declared
    as one. Now probing for known ISA bridge IDs. (Thanks to Matthew 
    Harrell for reporting this.)
  * Added some new flashchip IDs and made some old ones work.

** 04/24/98 * V0.2.7 *************************************************

  * rewrote shadowing and wp functions to use a pci_functions structure
    This makes it very easy to include new PCI chipsets.
  * function chipset_init() detects PCI chipset.
  * modversions support. Thanks to Matthew Harrell.
  * moved PCI bridge detection to chipset_init()

** 04/23/98 * V0.2.6 *************************************************

  * repaired flashchip_ready_toggle and flashchip_ready_poll.
  * Set WRITE_DELAY to 300 as it should be (works now)
  * NOTE: These two changes make the operation of /dev/bios
    theoretically correct, and by that quite secure.

**********************************************************************

There was no ChangeLog for versions prior to 0.2.6

Stefan Reinauer, <[e-mail]>