Hi Folks,

The 0.1.13-pre3 release is up at:
ftp://ftp.linux-wlan.org/pub/linux-wlan-ng/

The PF_PACKET patch still hasn't been rolled into this release. We're
deferring it to 0.1.14.

If you've forgotten the procedure for "-pre" releases, please see
doc/releasepolicy.txt.

Showstopper bug reports for 0.1.13 have slowed down and I'm considering
this one to be the release candidate for 0.1.13. I know USB is still
troublesome, but I'm not going to worry too much about it until the new
firmware is released. On another USB note: I'm seeing radically different
reports from UHCI and OHCI users. We need to figure out what's going on
here. If someone who knows more about USB cares to chime in, I'm happy to
listen to anyone's suggestions. Frankly, I'm not using USB that much at
the moment so you folks are probably more experienced with prism2_usb than
I am.

The one major change that I (still) need to point out to everyone is a new
wlanctl command. The command is called "lnxreq_ifstate" and it controls
whether the Driver+MAC is enabled or not. The driver+MAC _must_ be
enabled before issuing any other commands. So the basic initialization
sequence now looks like:

wlanctl-ng wlan0 lnxreq_ifstate ifstate=enable
wlanctl-ng wlan0 lnxreq_autojoin ssid=XYZZY authtype=opensystem
ifconfig wlan0 X.Y.Z.A
  ...etc...

Here's the change overview from the CHANGES file:

0.1.13
-pre3
  - crankin's patch for a bug in usb rrid, DBFENTER/EXIT, and Actiontec
    USB IDs.
  - hmuurimaa's ID patch for the Omnibook500 integrated USB adapter
  - proskin's patch for MODULE_LICENSE conditional
  - dstates (JediElite) submission (others submitted it too) for the
    NDC PLX card.
  - crankin's patch for src/prism2/driver/Makefile. PCI and USB were
    linking with PLX's prism2wext.o file. DOH!
-pre2
  - Synch'd the etc/pcmcia/wlan-ng scripts with the etc/wlan* scripts.
    This should help PCI/PLX/USB Adhoc users.
  - Added cmulliner's patch for set_mac_address(). Also changed it to
    use dot11StationID instead of the prism2 specific mib item...Oh
    Yeah, I also changed to comments to /*...*/ ;-)
  - Fixed PDA reads and PDA validation.
  - KNOWNBUG: prism2_usb + linux-2.4.17 + uhci.o : Very slow performance.
    We need to look into having multiple posted tx and rx urbs (this is
    also a TODO).
  - KNOWNBUG: prism2_usb + linux-2.2.20 + uhci.o : Physically
    disconnecting the device causes a Panic. The logs look like the
    usbdev is still holding some buffers (urbs?) associated with the
    device despite the fact that I've explicitly unlinked them. This
    problem does not occur on 2.4.17 and, unfortunately, I don't have
    time to debug 2.2.20 this time around.
  - Made all corereset() calls dependent on the prism2_doreset module
    argument.
  - Added corereset() support for all device types.
  - Added the lnxreq_ifstate request message. This request is used
    to bring the device up to an operable state _after_ the
    [hostbus]_probe() function is finished. It can also be used to
    shutdown and/or restart the driver+device. This takes the place
    of the "activate hardware on ifconfig up" code that was introduced
    in 0.1.11. This is the change that should fix most of our dhcp
    client problems.
  - Massive changes/cleanups to the driver init/shutdown code for all
    device types (pcmcia|pci|plx|usb).
  - Removed the add-ons/mibedit application from the linux-wlan-ng package.
    We'll be re-releasing it as a separate tarball.
  - Modified hfa384x_usb.c per dbrownell's suggestion to mark the request
    urb's with the USB_ASYNC_UNLINK flag. This because we're calling
    unlink_urb() from the timer callback (intterupt context).
  - Added arichter's patch for the RRID framelength. It's a good catch
    to make things truly correct. Unfortunately, I don't think it will
    have any effect. Because of the 64-byte minimum, the actual RID value
    is being transferred regardless. Also, I don't think the RRID handler
    in the MAC even looks at the framelen field of RRID frames.
  - Added mwelwarsky's patch for the tx path that may fix some of the
    "queue empty" errors. Had to update it for 0.1.13.
  - Added new PLX device IDs from cyokoyama's patch, but deferred the rest
    until we can get a clean patch against the latest release.
  - Added pizza's v2.1 NETLINK enable/disable patch. Note that this whole
    NETLINK enable/disable thing is only barely tested by me. I _always_
    have netlink available so maintaining a separate kernel setup just to
    test this feature is a pain.
  - Added jdiedrich's pcmcia card ids for the Compaq ipaq PCMCIA card.
  - Added crankin's patch fixing some more wext induced problems and adding
    the Actiontec USB ids.
  - Added abridgett's patch for PRIV_GENSTRING handling.
  - Added emckee's malloc.h -> slab.h patch and made the mods to the usb
    and wext files he missed. Some kind list soul told me
    that the change to slab.h goes way back in the 2.2.x history, so we
    should be OK. If this change proves to be a problem for someone with
    an old kernel, please submit a patch with ifdefs for the version where
    things changed (not just the version you're using....thanks).
  - Added jtourrhiles patch for wext reporting level and noise in dbm.
    To enable the reporting in dbm, you must add "#define WEXT_IN_DB"
    to the top of src/prism2/driver/prism2wext.c.
  - Added support for the dbm-normalized RID for CommsQuality in prep
    for working with Jean's patch. The mibs are called p2dbmCommsQuality*.
    Note that this will only work for secondary firmware versions 1.2.0
    and higher. Earlier versions will return with a
    'resultcode=implementation_failure'. Note also: The values returned
    are signed 16-bit integers that have been stuffed into unsigned 32-bit
    integers for the trip up to userland. The following shell expression
    will convert them back to the signed quantity you want:
       if [ $val -lt 32767 ]; then
           echo $1
       else
           echo $[ $1 | 0xffff0000 ]
       fi
  - Added mkershaw's suggestion to lengthen the wait interval for
    commands in hfa384x.c.
  - Added a note to the README about dleffler's observation regarding
    netlink troubles and using 'make mrproper'.
  - Added gback's patch to initialize wireless extensions for non-pcmcia
    targets.
-pre1
  - Fixed a patch induced (not sure which patch) Makefile syntax error.
  - Added rfloeter's cool Wireless Extensions patch which was kindly
    updated to 0.1.12 by pizza. Note that my own testing of this patch has
    been minimal but I've had reports of folks who are using it successfully.
    NOTE: The addition of this patch adds the requirement that Wireless
    Extensions (version >= 10) be compiled into one's kernel.
  - DOH! some somewhat kind soul on the lists pointed out that the keygen
    program wasn't in the distributed tarball. Basically, I forgot to
    check it into CVS, therefore the distribution build script couldn't
    find it. Ahem...it's now in CVS and should be included in the tarballs.
  - Added pizza's patch fixing the "netlink disable" patch. The
    p80211ind_* functions need to be handled individually instead of two
    giant #if's, but that's a task for another day.
  - Added theBS's patch for making the /etc stuff installation relative and
    installing the FIXINST scripts for USB.
  - Added crankin's SMP deadlock fix patch.

Have Fun!
-Mark

-- 
Mark S. Mathews
AbsoluteValue Systems      Web:    http://www.linux-wlan.com
715-D North Drive          e-mail: marklinux-wlan.com
Melbourne, FL 32934        Phone:  321.259.0737
USA                        Fax:    321.259.0286
_______________________________________________
Linux-wlan-announce mailing list
Linux-wlan-announcelists.linux-wlan.com
http://lists.linux-wlan.com/mailman/listinfo/linux-wlan-announce

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]