headers
Kel Modderman | 2 Jul 2006 08:58
Favicon

revision 1661 committed

Project     : madwifi
Revision    : 1661
Author      : kelmo (Kel Modderman)
Date        : 2006-07-02 08:58:47 +0200 (Sun, 02 Jul 2006)

Log Message :
There is a lack of hard information as to why ad-hoc has been limited
to channels 10 and 11, although there is a suggestion that this is
possibly the remenants of an old standard for ad-hoc usage. Replace
the table of channels used by ad-hoc mode with that used when in
station mode.

Closes ticket #623

Signed-off-by: Ian M Rawley <imr1 <at> waikato.ac.nz>

Affected Files:
* trunk/net80211/ieee80211_scan_sta.c                 updated             

Modified: trunk/net80211/ieee80211_scan_sta.c
===================================================================
--- trunk/net80211/ieee80211_scan_sta.c	2006-06-30 02:23:09 UTC (rev 1660)
+++ trunk/net80211/ieee80211_scan_sta.c	2006-07-02 06:58:47 UTC (rev 1661)
 <at>  <at>  -1169,7 +1169,7  <at>  <at> 
 	 * of channels for scanning.  Any channels in the ordered
 	 * list not in the master list will be discarded.
 	 */
-	for (scan = adhocScanTable; scan->list != NULL; scan++) {
+	for (scan = staScanTable; scan->list != NULL; scan++) {
 		mode = scan->mode;
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michael Renzmann | 2 Jul 2006 09:19
Favicon

revision 1662 committed

Project     : madwifi
Revision    : 1662
Author      : mrenzmann (Michael Renzmann)
Date        : 2006-07-02 09:19:37 +0200 (Sun, 02 Jul 2006)

Log Message :
Since r1661 adhocScanTable is no longer referenced/used.

Affected Files:
* trunk/net80211/ieee80211_scan_sta.c                 updated             

Modified: trunk/net80211/ieee80211_scan_sta.c
===================================================================
--- trunk/net80211/ieee80211_scan_sta.c	2006-07-02 06:58:47 UTC (rev 1661)
+++ trunk/net80211/ieee80211_scan_sta.c	2006-07-02 07:19:37 UTC (rev 1662)
 <at>  <at>  -427,6 +427,8  <at>  <at> 
 	{ .list = NULL }
 };

+#undef X
+
 static int
 checktable(const struct scanlist *scan, const struct ieee80211_channel *c)
 {
 <at>  <at>  -1130,28 +1132,6  <at>  <at> 
 };

 /*
- * Adhoc mode-specific support.
- */
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michael Renzmann | 3 Jul 2006 06:12
Favicon

revision 1663 committed

Project     : madwifi
Revision    : 1663
Author      : mrenzmann (Michael Renzmann)
Date        : 2006-07-03 06:12:12 +0200 (Mon, 03 Jul 2006)

Log Message :
While patches/install.sh determines the location of the kernel source
that should be patched (either automatically or by using the path
that is provided by the user on the command line) it does not pass
this information to make.

As a result Makefile.inc determines KERNELPATH by looking at the
currently running kernel. If /lib/modules/'uname -r'/build does not
exist make will fail, asking the user to define KERNELPATH - what
he probably thought he already did when he called install.sh.

This is fixed by passing install.sh's idea of the kernel source
path to make where necessary. Closes #731.

Affected Files:
* trunk/patches/install.sh                            updated             

Modified: trunk/patches/install.sh
===================================================================
--- trunk/patches/install.sh	2006-07-02 07:19:37 UTC (rev 1662)
+++ trunk/patches/install.sh	2006-07-03 04:12:12 UTC (rev 1663)
 <at>  <at>  -71,7 +71,7  <at>  <at> 
 MADWIFI=${WIRELESS}/madwifi
 rm -rf ${MADWIFI}
 mkdir -p ${MADWIFI}
(Continue reading)

Permalink | Reply | 0 comments |
headers
Kel Modderman | 4 Jul 2006 12:18
Favicon

revision 1664 committed

Project     : madwifi
Revision    : 1664
Author      : kelmo (Kel Modderman)
Date        : 2006-07-04 12:18:40 +0200 (Tue, 04 Jul 2006)

Log Message :
madwifi-ng works sporadically, or not at all, with wpa_supplicant
when one attempts to use a network whose access points are
"stealthed" (not broadcasting beacons with ESSID).

There appear to be two closely-related reasons for this:

[1] The ieee80211_ioctl_siwscan handler doesn't process the ioctl
parameters, and will thus ignore the IW_SCAN_THIS_ESSID flag and the
specified ESSID. The VAP's default ESSID is the one actually scanned
for. This problem affects the wpa_supplicant "wext" driver, which
passes the scan ESSID via the SIWSCAN ioctl data. It doesn't affect
the wpa_supplicant "madwifi" driver, which changes the VAP's default
ESSID before scanning.

[2] It's often the case in practice that ieee80211_ioctl_siwscan will
be called, to scan for a desired ESSID, when there's already an
active scan in progress. When this happens, ieee80211_ioctl_siwscan
has no effect... it calls ieee80211_start_scan which simply logs an
"active scan already in progress" and returns. No scan for the
newly-desired ESSID is performed. This problem tends to prevent
wpa_supplicant from working with either the wext or madwifi drivers -
there always seems to be an active scan running (usually with no
ESSID specified) when wpa-supplicant calls for a new active scan, and
the new scan(s) lose out.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Kel Modderman | 4 Jul 2006 12:20
Favicon

revision 1665 committed

Project     : madwifi
Revision    : 1665
Author      : kelmo (Kel Modderman)
Date        : 2006-07-04 12:20:36 +0200 (Tue, 04 Jul 2006)

Log Message :
There is an apparent race condition between scanning and changing
mode. When active scanning changing the mode (iwpriv athX mode Y)
will either work immediately, or will have no effect until a scan
is manually prompted (iwlist athX scan) at which point it changes
to the new mode prior to commencing the scan. If the scanning
debug output is turned on (80211debug +scan) then the change in
mode always occurs immediately.

This patch spins on the flag IEEE80211_F_SCAN which indicates the
scan has been cancelled inside ieee80211_ioctl_setmode() and
delays a short time if the flag indicates the scan is still
running.

At present, the solution is this: Make the max timeout value 100ms,
which should be fine for 99.9% of the time (cancel_scan should, in
theory, only take a maximum of 10ms to fire). However, if cancel_scan
does time out (or another scan is started between cancel_scan firing
and us noticing the cancellation of the scan), then there is a
problem.

Closes #228

Signed-off-by: Ian M. Rawley <imr1 <at> waikato.ac.nz>
Signed-off-by: Scott Raynel <scottraynel <at> gmail.com>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Kel Modderman | 4 Jul 2006 12:22
Favicon

revision 1666 committed

Project     : madwifi
Revision    : 1666
Author      : kelmo (Kel Modderman)
Date        : 2006-07-04 12:22:11 +0200 (Tue, 04 Jul 2006)

Log Message :
This patch improves the ability of VAPs to handle proc entries.
Specifically it allows parts of the driver to register a proc entry
and appropriate handlers with the vap. The vap will then take care of
ensuring that this proc entry exists for the lifetime of the vap and
that it is correctly relocated if the vap is renamed.

Previously these tasks had to be handled by whatever registered the
proc entry, which promoted code duplication.

The list of proc entries for a specific vap is stored as a linked
list. Two new methods are added.

The first is used to add a proc entry to the vap. It is called
ieee80211_proc_vcreate.

 * vap - pointer to the vap to create the entry for
 * fileops - pointer to methods to handle read/writes from the entry.
   Any NULL entries in this structure will be substituted with
   default entries. These default entries will use a
   proc_ieee80211_priv structure to provide a list of associated
   stations. Typically a caller will only need to provide an open
   method to read the desired information into the read buffer of the
   proc_ieee80211_priv structure.
 * name - the name for the proc entry
(Continue reading)

Permalink | Reply | 0 comments |
headers
Kel Modderman | 4 Jul 2006 12:23
Favicon

revision 1667 committed

Project     : madwifi
Revision    : 1667
Author      : kelmo (Kel Modderman)
Date        : 2006-07-04 12:23:35 +0200 (Tue, 04 Jul 2006)

Log Message :
This patch adds support for reporting statistics on the operation
of the rate control modules via /proc. Currently only the sample rate
module actually provides any statistics.

The statistics reported for the sample module were present in the
madwifi-old module but did not seem to make it into madwifi-ng. I
have made some minor changes to how these statistics are reported:

 1. Statistics are now reported via /proc entries instead of using
    sysctls. Sysctls are intended as a method for the user to
    configure parameters of the module and should not be misued for
    reporting statistics.
 2. A new symbol ath_rate_dynamic_proc_register is exported from each
    rate control module. This function is called by the driver just
    after a new VAP has been created to allow the rate control module
    to register the appropriate proc entries. As described in ticket
    #728, all further management of these entries will be handled by
    the VAP, which will call the functions supplied by the rate
    control module to retrieve data when necessary.

The information that this patch adds back into the driver is very
useful and is referred to in the documentation which currently causes
some confusion when the statistics is not found. The documentation
should be updated to reflect that the statistics will now live at
(Continue reading)

Permalink | Reply | 0 comments |
headers
Kel Modderman | 4 Jul 2006 12:25
Favicon

revision 1668 committed

Project     : madwifi
Revision    : 1668
Author      : kelmo (Kel Modderman)
Date        : 2006-07-04 12:25:41 +0200 (Tue, 04 Jul 2006)

Log Message :
With enhanced iwspy support it is possible to define signal strength
thresholds. If the signal strength of a currently spyed node exceeds
the defined thresholds, the driver raises an iwspy event.

The patch adds adds iwspy event capability for station and ad-hoc
mode.

Signed-off-by: Christian Buennig <masala <at> web.de>

Affected Files:
* trunk/net80211/ieee80211_input.c                    updated             
* trunk/net80211/ieee80211_var.h                      updated             
* trunk/net80211/ieee80211_wireless.c                 updated             

Modified: trunk/net80211/ieee80211_input.c
===================================================================
--- trunk/net80211/ieee80211_input.c	2006-07-04 10:23:35 UTC (rev 1667)
+++ trunk/net80211/ieee80211_input.c	2006-07-04 10:25:41 UTC (rev 1668)
 <at>  <at>  -46,6 +46,9  <at>  <at> 
 #include <linux/etherdevice.h>
 #include <linux/random.h>
 #include <linux/if_vlan.h>
+#include <net/iw_handler.h> /* wireless_send_event(..) */
+#include <linux/wireless.h> /* SIOCGIWTHRSPY */
(Continue reading)

Permalink | Reply | 0 comments |
headers
Pavel Roskin | 5 Jul 2006 04:21
Favicon

revision 1669 committed

Project     : madwifi
Revision    : 1669
Author      : proski (Pavel Roskin)
Date        : 2006-07-05 04:21:30 +0200 (Wed, 05 Jul 2006)

Log Message :
Fix KERNELRELEASE calculation for Linux 2.6.18+

linux/version.h doesn't define UTS_RELEASE in the forthcoming Linux
2.6.18.  UTS_RELEASE is defined in linux/utsrelease.h.

Affected Files:
* trunk/kernelversion.c                               updated             

Modified: trunk/kernelversion.c
===================================================================
--- trunk/kernelversion.c	2006-07-04 10:25:41 UTC (rev 1668)
+++ trunk/kernelversion.c	2006-07-05 02:21:30 UTC (rev 1669)
 <at>  <at>  -7,4 +7,10  <at>  <at> 
  * $Id$
  */
 #include <linux/version.h>
+
+/* Linux 2.6.18+ uses <linux/utsrelease.h> */
+#ifndef UTS_RELEASE
+#include <linux/utsrelease.h>
+#endif
+
 char *uts_release = UTS_RELEASE;
(Continue reading)

Permalink | Reply | 0 comments |
headers
Pavel Roskin | 5 Jul 2006 06:35
Favicon

revision 1670 committed

Project     : madwifi
Revision    : 1670
Author      : proski (Pavel Roskin)
Date        : 2006-07-05 06:35:06 +0200 (Wed, 05 Jul 2006)

Log Message :
Fix extended iwspy support for WE 16

IW_QUAL_*_UPDATED appeared first in WE 17, so we need compatibility
definitions for WE 16.

Affected Files:
* trunk/net80211/ieee80211_input.c                    updated             

Modified: trunk/net80211/ieee80211_input.c
===================================================================
--- trunk/net80211/ieee80211_input.c	2006-07-05 02:21:30 UTC (rev 1669)
+++ trunk/net80211/ieee80211_input.c	2006-07-05 04:35:06 UTC (rev 1670)
 <at>  <at>  -130,6 +130,12  <at>  <at> 
 #ifdef CONFIG_NET_WIRELESS
 #if WIRELESS_EXT >= 16

+#ifndef IW_QUAL_QUAL_UPDATED
+#define IW_QUAL_QUAL_UPDATED	0x01
+#define IW_QUAL_LEVEL_UPDATED	0x02
+#define IW_QUAL_NOISE_UPDATED	0x04
+#endif /* IW_QUAL_QUAL_UPDATED */
+
 /**
  * This function is a clone of set_quality(..) in ieee80211_wireless.c
(Continue reading)

Permalink | Reply | 0 comments |

Gmane