Subversion Repositories or1k

TLan Device Driver change log.

1.0     - Stopped ignoring devices that didn't has the bus master
          flag set.

0.43    - Changed the id strings of the Compaq devices to their real
          names.
        - Added 2 other Olicom devices with a patch provided by
          Henrik Storner of Olicom.
        - Changed AN so the driver won't try to autonegotiate with
          a partner that doesn't autonegotiate.
        - Added a section to verify whether in full or half duplex.
        - Added value checking for speed and duplex inputs.
        - Added B012 and B030 Compaq devices.
          
0.42    - Coverted tranceiver from misused per-tranceiver functions
          to a timer-oriented path that covers four possible classes
          of tranceivers:
                1. Unmanaged
                2. Manual configuration
                3. Autonegotiation w/ manual configuration
                4. Autonegotiation w/ auto configuration
        - Added ability to force speed and duplex settings.
        - Made speed, duplex, sa_int, etc, to be set per adapter with
          ether= command.
        - Added support for Olicom OC-2326

0.41    - Added non-bounce buffer paths.  Added TLan_FreeLists to
          dispose of unused sk_buff's at device close time.
        - Discovered inlined functions aren't being inlined, or at
          least take up more space than macros would.

0.40    - Refined polarity checking to handle case when polarity
          changes to normal from abnormal.
        - Cleaned up TLan_Probe routine.
        - Added an option for the SA_INTERRUPT flag to be set.
        - Created FAQ.
        - Removed all C++ style comments.
        - Added error message if devices busmastering is inactive.
          Also will now skip device.
        - Put cli and sti back into TLan_HandleInterrupt.  It makes
          me feel better.
        - Moved the code that checks for boot parameter options to 
          tlan_probe.

0.39    - Minor cosmetic cleanups (especially variable declarations).
        - Changes low level TLAN functions to use dev structures instead
          individual data elements.
        - Changed low level TLAN functions not to play with sti and cli
          if in an interrupt routine.
        - Removed cli and sti from TLan_HandleInterrupt.

0.38    - Added code to isolate the external PHY if the internal PHY is
          being used for AUI/BNC connectivity.  Also set the aui and
          debug variables from mem_start and mem_end if the driver is
          built into the kernel.

0.37    - Added TLAN_PHY_ACTIVITY flag for Unmanaged PHY.
        - If aui is selected and the card is not a 0xF130 (unmanaged phy),
          select the builtin PHY and put it in AUI mode.  I don't know if
          this will work, but it is my best guess as to how AUI/BNC
          functionality is being provided for the external managed PHYs,
          which don't support AUI/BNC.
        - If aui is set and the card is a 0xF130, set the MTXD1 bit in
          the ACOMMIT register, to select AUI mode on the unmanaged PHY.
          I don't know if this is necessary.

0.36    - Changed AN_ADV register to not advertise full duplex modes.
          100Mbs should work now on 0xAE32.
        - Fixed a small bug where heartbeat and PHY interrupts were
          always being enabled.
        - Force the driver into Unmanaged PHY mode for 0xF130 devices,
          even if a managed (ie, the built-in one) PHY is detected.
        - Moved the PHY initialization to after the onboard PHY is enabled,
          if selected.

0.35    - Added entry for Level One LXT970 PHY.
        - Commented out instruction to set phyOnline to 1 in DP Check.

0.34    - Revised Reset routine.
        - Added support for 0xF130 device.
        - Added entry for NS DP83840A Rev 0 PHY.

0.33    - Major Reformatting of comments to fit in 80 columns.
        - Changed tabs to 8 spaces, reformatted accordingly.
        - Added code to check and change polarity.
        - Added caveats to README.
        - Small fix to Makefile to make unversioned modules correctly.
        - Redid PhySelect to list all phy ids then choose one.

0.32    - Put in another change for BNC stuff that I mistakenly
          omitted in 0.31.

0.31    - Completed BNC changes (at least one person had it working).
        - Added another device id.
        - Changed debugging messages to hopefully be more useful.

0.30    - Added PCI device IDs provided by Don Dupuis of Compaq.
        - Turned of MINTEN when using 10/100 PHY.
        - Added ability to select AUI at insmod time.  What will this do
          for BNC, if anything?
        - Tweaked with names.  

0.29    Fixed problem with older TLAN chips (pre-PG3.0 didn't have
        EOC bit in lists nor INTDIS register).  Added support for NetFlex
        card.

0.28    Changed a lot of commented out printfs to debugging statements.
        Added detail to product names.  Added 2nd reset to open command.
        Don't know why, just have to, at least for Netelligent 10 card.

0.27    Created a tlan_probe function for compiled-in-kernel use.
        Also added debugging macro, which could be reviewed.  Is there
        a kernel variable which stores desired verbosity/debug level?

0.26    Change Invalid interrupt handler to return 0, not print out debug
        message.  This was causing lots of messages to be printed out with
        shared interrupts.

0.25
        Added IDs for integrated NetFlex-3 controller.  Not sure if it is even
        TLAN yet, but we'll see.

0.24
        Increased second delay in internal PHY check routine, as driver wasn't
        starting on the first insmod.

0.23
        Added documentation to function preambles.
        Removed some commented out printk's.
        Tested Rx and Tx busy frequency with current queue sizes (no busies).

0.22
        Cleared tbusy flag at adapter check.
        Removed some commented out printk's.
        Added TLan_SetMulticastList routine.

0.21
        Got all virt_to_bus calls in place for 2.1 kernels.

0.20
        Rewrote buffer/list subsystem to use bounce buffers as DMA
          doesn't work over 16 Meg.  Why?  This is PCI not ISA.
        Fixed a problem with buffers stalling due to a race condition
          in the kernel.