headers
Jouni Malinen | 1 Jan 2005 07:45
Picon
Picon

Re: [PATCH] fully clean binaries in hostap-utils

On Mon, Dec 27, 2004 at 07:02:32AM -0500, Andres Salomon wrote:

> In hostap-utils, during `make clean`, the hostap_io_debug binary is not
> deleted.  This patch corrects that.

Thanks, applied to both development and stable branches.

--

-- 
Jouni Malinen                                            PGP id EFC895FA
Permalink | Reply | 0 comments |
headers
Jouni Malinen | 1 Jan 2005 07:51
Picon
Picon

Re: ifrename problems

On Mon, Dec 27, 2004 at 06:06:04AM -0500, Andres Salomon wrote:

> "Card inserts normally as wlan0.  ifrename renames the card to e200
> based
> on MAC address per /etc/iftab.  Removing the card, with the interface up
> or down, then produces:
> 
> hostap_cs: Driver unloaded
> Badness in remove_proc_entry at fs/proc/generic.c:688
>  [<c018c83a>] remove_proc_entry+0x10a/0x150

This is related to a known issue of what to do with a procfs directory
with an interface name when the name of that interface is changed.. The
current version of the driver just ignores the change, but does not save
the old interface name which seems to trigger this case since the procfs
entries are not removed. I have consider of changing the driver to
rename that directory, but haven't yet seen this important enough to
spent much time on..

> This is with hostap-driver-0.2.5 and linux 2.6.9.  Note that they also
> have issues w/ the MAC address after an ifrename, as well.  The original
> bug reports can be seen here:

This does not look like ifrename specific.. I would guess that this
would be a card without firmware image on flash and ifrename being run
before firmware has been downloaded, i.e., when the card MAC address is
not yet available.

--

-- 
Jouni Malinen                                            PGP id EFC895FA
(Continue reading)

Permalink | Reply | 2 comments |
headers
Jouni Malinen | 1 Jan 2005 07:59
Picon
Picon

Re: ifrename problems

On Fri, Dec 31, 2004 at 10:51:29PM -0800, Jouni Malinen wrote:

> > This is with hostap-driver-0.2.5 and linux 2.6.9.  Note that they also
> > have issues w/ the MAC address after an ifrename, as well.  The original
> > bug reports can be seen here:
> 
> This does not look like ifrename specific.. I would guess that this
> would be a card without firmware image on flash and ifrename being run
> before firmware has been downloaded, i.e., when the card MAC address is
> not yet available.

Actually, not.. After a closer look at the kernel log, this does not
seem to be downloading a firmware image.

--

-- 
Jouni Malinen                                            PGP id EFC895FA
Permalink | Reply | 0 comments |
headers
Whoopie | 1 Jan 2005 19:28
Picon

EAPOL-Key timeout

Hi,
can´t nobody help me with my problem with hostapd, madwifi and WinXP 
supplicant.
Here is my post: 
http://lists.shmoo.com/pipermail/hostap/2004-December/008945.html

Would be great if someone gave my an advice.
Thanks a lot,
Whoopie
Permalink | Reply | 0 comments |
headers
Miles Lane | 2 Jan 2005 10:15
Picon
Gravatar

Problem building the Broadcom driver -- no typedefs.h on my Linux x86 system.

Dear developers,

I cannot get the broadcom driver to build.  Both the main driver file
and wlioctl.h are including typedefs.h.  How can I either work around
this or locate a suitable typedefs.h file?  I am running Debian SID
and a 2.6 kernel.  I have the linux-2.6 kernel tree on my machine and
/usr/src/linux is linked to /usr/src/linux-2.6.

Thanks,
      Miles
Permalink | Reply | 0 comments |
headers
Derek Chan | 2 Jan 2005 10:22
Picon

invalid cis with plx adapter

Hi folks,

I've got a US Robotics 2415 (USR 2410 + plx 9052 pci adapter) that I've
been using for over 2 years in my laptop (orinoco_cs, linux-wlan-ng and
hostap) with no trouble at all.  I finally broke down and upgraded to a
54g card for my laptop for mobile use, and put the 2410 into the adapter
into my desktop.  Unfortunately, as a few posters here and on the net
have found, just because it works straight as a pcmcia card, it doesn't
in the plx.  Same as is for me.

When I attempt to load the plx drivers, I get an "Invalid CIS data"
error.

hostap_plx: 0.2.6 - 2004-12-25 (Jouni Malinen <jkmaline <at> cc.hut.fi>)
PCI: Found IRQ 10 for device 00:0d.0
PLX9052 PCI/PCMCIA adapter: mem=0xe0004000, plx_io=0x6800, irq=10, pccard_io=0x6c00
hostap_plx: CIS: 00 af 7a fb 3d f8 ...
hostap_plx: invalid CIS data
Unknown PC Card CIS - not a Prism2/2.5 card?
hostap_plx: No devices found, driver not installed.
hostap_crypt: unregistered algorithm 'NULL' (deinit)

I can post the details from the other failings if that is any help. (The
orinoco seemed to pick a slightly different offset... orinoco_plx: CIS:
FF00:AF:AF7A:7AFB:FB3D:3DF8:F836:36C8:C899:9131:85:81E3:E0BA:B050:509F:99D9:)

I verified that the card is indeed plugged in, and in the right
direction (face up from the pci card's board, face down when in the
tower case).  The PCI slot also works -- I moved the Promise ATA
controller out of the slot the 2415 is in now.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jouni Malinen | 3 Jan 2005 03:25
Picon
Picon

Re: ifrename problems

On Fri, Dec 31, 2004 at 10:51:29PM -0800, Jouni Malinen wrote:
> On Mon, Dec 27, 2004 at 06:06:04AM -0500, Andres Salomon wrote:
> > "Card inserts normally as wlan0.  ifrename renames the card to e200
> > based
> > on MAC address per /etc/iftab.  Removing the card, with the interface up
> > or down, then produces:
> > 
> > hostap_cs: Driver unloaded
> > Badness in remove_proc_entry at fs/proc/generic.c:688
> >  [<c018c83a>] remove_proc_entry+0x10a/0x150
> 
> This is related to a known issue of what to do with a procfs directory
> with an interface name when the name of that interface is changed.. The
> current version of the driver just ignores the change, but does not save
> the old interface name which seems to trigger this case since the procfs
> entries are not removed. I have consider of changing the driver to
> rename that directory, but haven't yet seen this important enough to
> spent much time on..

I was too lazy to start renaming the entry, but CVS versions (devel &
stable) are at least now deleting the correct directory to avoid this
badness report.

--

-- 
Jouni Malinen                                            PGP id EFC895FA
Permalink | Reply | 0 comments |
headers
Jouni Malinen | 3 Jan 2005 03:56
Picon
Picon

Re: ifrename problems

On Mon, Dec 27, 2004 at 06:06:04AM -0500, Andres Salomon wrote:

> This is with hostap-driver-0.2.5 and linux 2.6.9.  Note that they also
> have issues w/ the MAC address after an ifrename, as well.  The original
> bug reports can be seen here:

> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=286726

I cannot reproduce the exact reported behavior with the current CVS
versions from either development or stable branches. However, wifi0
interface does show up with all zeroes address and so would wlan0 if the
card does not come with firmware included. The workaround for this is to
delay ifrename call to happen at later phase when the driver has
completed firmware initialization. See
http://hostap.epitest.fi/bugz/show_bug.cgi?id=6 for details.

--

-- 
Jouni Malinen                                            PGP id EFC895FA
Permalink | Reply | 0 comments |
headers
Jouni Malinen | 3 Jan 2005 04:54
Picon
Picon

hostap/hostapd/wpa_supplicant - new development release v0.3.3

New versions of the Host AP driver, hostapd, and wpa_supplicant were
just released and are now available from http://hostap.epitest.fi/

This release is from the development (0.3.x) branch. Please note that
0.2.x branch continues to be the current source of stable releases.

Most of the new features are for EAP authentication. wpa_supplicant
got support for EAP-FAST and the integrated EAP authenticator in
hostapd for EAP-PEAP, EAP-GTC, EAP-TTLS, and EAP-SIM. In addition,
hostapd can now be used as a RADIUS authentication server that
supports the same EAP methods as the integrated EAP authenticator.

hostap-driver:
* fixed /proc/net/hostap/wlan# removing for cases where the interface
  name has been changed

hostapd:
* added support for EAP-PEAP in the integrated EAP authenticator
* added support for EAP-GTC in the integrated EAP authenticator
* added support for configuring list of EAP methods for Phase 1 so that
  the integrated EAP authenticator can, e.g., use the wildcard entry
  for EAP-TLS and EAP-PEAP
* added support for EAP-TTLS in the integrated EAP authenticator
* added support for EAP-SIM in the integrated EAP authenticator
* added support for using hostapd as a RADIUS authentication server
  with the integrated EAP authenticator taking care of EAP
  authentication (new hostapd.conf options: radius_server_clients and
  radius_server_auth_port); this is not included in default build; use
  CONFIG_RADIUS_SERVER=y in .config to include
(Continue reading)

Permalink | Reply | 6 comments |
headers
Jouni Malinen | 3 Jan 2005 05:07
Picon
Picon

Bug tracking and feature requests

It looks like it could be time to get a bit more organized with bug
reporting and feature requests in order to get all information collected
into one place and to avoid losing useful information. I haven't had
enough time to keep up with all email I'm receiving for long time and
every now and then, I seem to be able to concentrate on adding new
features thoroughly enough to get even longer backlog in Host AP related
email..

I installed Bugzilla for bug tracking and feature requests and hope that
it will help in this area of the Host AP/hostapd/wpa_supplicant
development. I would appreciate it if all bug reports and feature
requests would be filed to that database at
http://hostap.epitest.fi/bugz/. The direct URL for filing a new bug is
http://hostap.epitest.fi/bugz/enter_bug.cgi. Both addresses are also
linked from the other web pages of the Host AP project.

There are number of bug reports and/or patches pending in my email
queues. I may or may not have enough time to go through these at some
point, but in order to speed this up for bugs or features that you
believe are important, please consider adding these reports/requests to
bugzilla even if they have already been sent to the mailing list or my
private email address.

This mailing list can, of course, still be used for reporting bugs and
discussing them, but I would appreciate it if people would also enter
the details to bugzilla. This should also allow attaching longer debug
log files to the bugs without having to send long messages to the
mailing list.

--

--
(Continue reading)

Permalink | Reply | 0 comments |

Gmane