mirrors
/
madwifi
mirror of https://github.com/proski/madwifi
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 9c00c5ccf3 Merge -dfs to trunk - r3361 1/2 mentor 2008-02-25 03:17:52 +0000
  • b99d972b64 revert an accidental change to the handling of errors at ath_tx_start from r3346, thx. bruno nbd 2008-02-23 14:13:09 +0000
  • 39f4007bdd fix a merge bug that broke ap channel scanning for different modes nbd 2008-02-21 01:47:52 +0000
  • 7b7aa53f4f Adjust contact information for donations. mrenzmann 2008-02-20 05:48:08 +0000
  • 7d5824e65c In some situations have the hardware PHY calibrations synchronised with beacon sending. There is a greater chance that the mediuim will be quiet during this time, which appears to imporive calibration results. mentor 2008-02-14 06:34:28 +0000
  • c7a001a014 fix errors in ath_info.8 manpage (as detected by debians lintian tool). kelmo 2008-02-11 11:12:17 +0000
  • 6b76be7f11 Merge -dfs to trunk - r3345: This probably should be in trunk mentor 2008-02-10 20:41:19 +0000
  • dd740cbeef Fix for #1769. Now that maximum supported VAPs is up to 64, 32 bits is not large enough for a bit set and rolls over, resulting in VAPs that are not configured to be accessed. mtaylor 2008-02-08 20:59:47 +0000
  • 625d3f967d fix some memory leaks triggered by memory allocation errors, which were most visible in ath_mgtstart nbd 2008-02-08 05:38:41 +0000
  • c3365b93bf Revert r3342 :( mentor 2008-02-06 23:23:48 +0000
  • d3ab3181b0 Move DTIM and CFP count processing out of a loop into closed form mentor 2008-02-06 21:37:05 +0000
  • dc1e5a74cd Also call ath_hal_rxmonitor in IBSS. Maybe this will make Ge0rg all happy. mentor 2008-02-06 21:30:48 +0000
  • 3f4e3ddb40 Only have the Station PS handling routine run when we are actually in Station mode. This probably only affects IBSS. mentor 2008-02-06 21:29:06 +0000
  • 56981e0363 * Adjust for repository structure adjustments made in r2721 * Add support for micro releases * Suppress possible but irrelevant error message mrenzmann 2008-02-05 14:42:15 +0000
  • 8c0469176d Add radar stuff to Makefile.kernel mentor 2008-02-05 06:27:17 +0000
  • c9cc0e7dee Correct pointer arithmetic in the SKB CB. With thanks (but not copyright ;) to Alessandro Erta mentor 2008-02-05 06:22:32 +0000
  • ebc79fce38 Cleanup up de-MIC failure (in hardware) handling. Apparently, HW sometimes gets this wrong and spuriusly fails valid packets. In all cases, the packets are dropped, but we do extra processing to record statistics correctly. This changeset adds a node statistic. If this is not liked, I encourage it to be fixed; I might learn something. Thanks nbd. mentor 2008-02-04 00:16:24 +0000
  • e906bf12da CCMP/AES Crypto module So, ccmp_attach is called for each new key. Hence, only try to allocate a software context if IEEE80211_KEY_SWCRYPT is specified. Also, if we need a software context and fail, fail immediately; thus, the check in ccmp_setkey can be eleminated, as it is not actually a separate condition. mentor 2008-02-02 20:07:01 +0000
  • c7fba7a6ce This is a HACK. With 0.9.30.13 ANI control appears to be broken. ANI is designed only for client (STA/AHDEMO) only mode. This function updates the data used for ANI, so we will only call it for client only mode. This may will not affect ANI problems in client only mode. mentor 2008-02-02 17:47:59 +0000
  • c0c3125da1 Demote these messages to debug only from informational as they are noisy and unneeded mentor 2008-02-02 17:36:29 +0000
  • 89f9f3ed2a Allow ARCH to be set to "x86". That's what it is in Linux 2.6.24 for both i386 and x86_64. When calculating TARGET for x86 platform, use CONFIG_X86 and CONFIG_X86_64 rather than ARCH. proski 2008-01-31 22:48:42 +0000
  • a9788959cd Move some static declarations around for no particular reason mentor 2008-01-31 03:54:19 +0000
  • 877b7727d1 I think howmany(roundup()) should just be howmany() as roundup(x, y) is howmany(x, y) * y mentor 2008-01-31 03:42:51 +0000
  • 4d5de4b5a2 This patch adds iwpriv values for setting the time in ms, or the number of beacons for the beacon miss alarm. If we haven't had a beacon in this period, the alarm is sounded and action (i.e. roaming) is taken. mtaylor 2008-01-30 23:50:16 +0000
  • 99b36fbd0f Fix section mismatch warning in forthcoming Linux 2.6.25 proski 2008-01-30 21:01:05 +0000
  • 0ad2c8981a Linux 2.4 series compatibility fixes * Based on patches from nbd * av_beacon_alloc code has been reimplemented using atomic bit operations; I think this is safe for SMP mentor 2008-01-30 20:23:49 +0000
  • 2b8281ba31 Update comment for why we set sc_beacons to one when resetting for calibration after txcont, cac, or calibration. mtaylor 2008-01-30 19:34:49 +0000
  • 6fbbcf00a0 This patch makes some debug messages about dropped frames a little more consistent and fixes an issue in hardstart where I was being too aggressive about requeing. mtaylor 2008-01-30 19:33:38 +0000
  • 874abd03ab Reduce the maximum message length for node messages from 1000 to 192 and drop the length needed to display the [atomic] node count. mtaylor 2008-01-30 19:26:13 +0000
  • 5d4e93939c Add CONFIG_CRYPTO as a Kconfig dependency when patching into the kernel mentor 2008-01-30 15:26:06 +0000
  • b8e6b99d2c De-FAIL previous revisions mentor 2008-01-29 17:46:15 +0000
  • b0f92899f4 Formatting - calming... mentor 2008-01-29 11:18:15 +0000
  • 3202c2a0de Change reset_bss semantics so that the node reference is passed to iv_bss, and then use the iv_bss reference. I believe that this is semantically correct, and so have removed the comment. mentor 2008-01-29 11:17:57 +0000
  • 2411a87efc During reset_bss we were incrementing the ref count of the new bss node. The allocation function gives us a reference. With this extra reference, every invocation of reset_bss was causing a leak of one node ~4kb every time the interface was brought back up (or the BSS was reset for other reasons). mtaylor 2008-01-29 07:08:29 +0000
  • 602be360fc Move the node reference up that pins the WDS peer node in the node table. The node reference was being taken in any path where the wds node was found and not just when the wds node was first added. mtaylor 2008-01-29 07:00:09 +0000
  • dead14502d Fix a few more leaked node references when a copy of the bss node is used to send a message we need to make sure all code paths release it when done. This bug combined with repeated auth failures would cause node references to grow with every failure. Then, if you were taking down the interface and bringing it back up you would leak one page (4kb) every time the the node table was flushed on transition to UP. mtaylor 2008-01-29 06:58:22 +0000
  • 8e8ba8882b Some extra SKB operation return code checks. This deviates from nbd's original patch, because I don't think that if the aligning operation fails that we should return the unaligned one. mentor 2008-01-28 23:49:26 +0000
  • 521e00a7c6 Turn on long pre-amble in ERP IE if we have non-ERP stations mentor 2008-01-28 23:40:44 +0000
  • 32b36e8f10 Fix compilation when there is no debugging mentor 2008-01-28 23:35:29 +0000
  • 27c02f3829 add missing #include statement nbd 2008-01-28 21:04:23 +0000
  • 3f689304d6 Fix regression in reverse engineering iwpriv functions now requiring access to sc. mtaylor 2008-01-28 21:02:07 +0000
  • 314dd3b3d8 Merge of madwifi-dfs r3189: Replace ath_buf_counter by sc->sc_txbuf_counter. Counter is specific to a madwifi instance. benoit 2008-01-28 20:44:43 +0000
  • d25a86ffe3 Restore some of the BSS synchronization code, but only add the goto that skips the beacon timer updates for IBSS. mtaylor 2008-01-28 20:32:04 +0000
  • e2ac8356c2 Restore some of the BSS synchronization code, but only add the goto that skips the beacon timer updates for IBSS. mtaylor 2008-01-28 20:26:25 +0000
  • a04a25ead0 This looks ungood, lets not leave it lying around mentor 2008-01-28 20:15:23 +0000
  • 456aaa276a Use standard kernel interface for AHB bus access. mentor 2008-01-28 20:14:20 +0000
  • f0a1f3d537 Null protection for skb1 in multicast node fix mtaylor 2008-01-28 20:05:31 +0000
  • 6600ba247b Restore changes to if_ath_ahb.c from r3267, and move statics and includes to the top of the file. mtaylor 2008-01-28 19:39:44 +0000
  • 0037b77ceb Revert changes to if_ath_ahb.c from r3267 mtaylor 2008-01-28 19:36:03 +0000
  • 43f0aa1604 Undo regression from madwifi-dfs merge mtaylor 2008-01-28 19:34:03 +0000
  • c61bd72fbc Remove improper handling of non-cloned frames in hardstart. mtaylor 2008-01-28 19:27:26 +0000
  • db307a76c6 Move rate control module parameter text to the correct section in INSTALL mentor 2008-01-28 14:31:55 +0000
  • d136b9b1f9 Edit token names for clarity, consistency and length. mentor 2008-01-28 01:36:25 +0000
  • 70be1ef404 Edit token names for clarity, consistency, and to remove EXTREME length TO THE MAX mentor 2008-01-28 01:21:19 +0000
  • b165b8b28c Typo mentor 2008-01-27 20:00:41 +0000
  • 4a60ebccec 'Context:' is used to denote the kernel context that the process is running in (i.e., hard IRQ, soft IRQ, process, preemption disabled, etc.) mentor 2008-01-27 19:50:09 +0000
  • c4952a9d3d Make CHANNEL_FOREACH slightly more readable mentor 2008-01-27 06:49:24 +0000
  • cf6e609462 Formatting mentor 2008-01-27 05:48:23 +0000
  • 60e9b06ffa Use a logical XOR mentor 2008-01-27 04:40:14 +0000
  • 28cf53aa95 Fix merge failures - pass one mentor 2008-01-27 04:26:38 +0000
  • 8041aa65bc in ath_intr, we will need HW TSF value if we get HAL_INT_RX, HAL_INT_RXPHY, or HAL_INT_SWBA. In such cases, we will collect the earliest hw tsf sample we can get and this is passed into send_beacon and the trigger processing helper functions. mtaylor 2008-01-27 00:46:22 +0000
  • 30e6030d36 Remove one unreferenced (but populated) hw tsf and the expensive call. Move one hw tsf read into a conditional block as it is only used in a debug printout. Replace one hw tsf read with a reference to the existing hw tsf in the buffer (which is more accurate anyway). mtaylor 2008-01-26 23:55:10 +0000
  • e00f351af1 Restore the evil printf macro I accidentally damaged mtaylor 2008-01-26 20:56:20 +0000
  • c2bce77197 Merge madwifi-dfs branch to trunk. mtaylor 2008-01-26 20:48:11 +0000
  • 0b74c2ebe0 More cases to use ieee80211_skb_copy_noderef mentor 2008-01-26 17:46:36 +0000
  • 776a840f28 The netif_queue appears to be causing some performance problems. I'm going to go ahead and guess that we always want to prod the networking system to start sending packets; thus use netif_wake_queue() mentor 2008-01-26 17:31:51 +0000
  • c14fa8bf98 Merge -dfs to trunk - r3246:r3254 mentor 2008-01-26 02:47:27 +0000
  • d05f18eef2 Have internal bridged packets belong to the BSS node mentor 2008-01-26 02:30:35 +0000
  • e479155497 Eliminate ether_sprintf in favor of new MAC_FMT and MAC_ADDR macros (which are faster and thread safe). mtaylor 2008-01-25 22:49:48 +0000
  • 77d9f966c0 - Add a iwpriv function that dumps a mapping of nasty obfuscated symbol names to their corresponding HAL function names (ah.h) - Add a script to generate a SED script from the current HAL (ah.h) that will replace instances of obfuscated names with the correct names. - Add the SED script for cleaning up text containing obfuscated function names for the current HAL mtaylor 2008-01-25 00:29:30 +0000
  • 8f5e980418 From madwifi-dfs. If ieee80211_find_channel is passed 0 for flags, then omit flag comparisons from the search criteria. mtaylor 2008-01-23 18:53:34 +0000
  • 2bfb48c560 frame injection: let IEEE80211_RADIOTAP_DATA_RETRIES specify the number of retries instead of the total number of tries. br1 2008-01-22 07:11:13 +0000
  • c585507798 beacon backoff configuration like the 802.11 standard describes br1 2008-01-22 07:04:39 +0000
  • e04d68cce7 ieee80211_dev_kfree_skb will NULL ni_rxfrag. mtaylor 2008-01-21 23:19:39 +0000
  • 72a19df0ff Fix svnversion.h generation proski 2008-01-21 15:55:13 +0000
  • 3f942d694e Add git svnversion.h generation code Signed-off-by: Russell Harmon <eatnumber1@gmail.com> mentor 2008-01-21 08:59:27 +0000
  • 52319f4d34 Remove humourous typo mentor 2008-01-20 03:34:45 +0000
  • 751589f640 Revert r3220 mentor 2008-01-19 15:45:28 +0000
  • 70339802ca Fix a bug where bf->bf_skb needs to be set to NULL because the skbuff have already been freed by ath_tx_capture, for each frame in the list we have already freed the skb but the original pointer to the list of skbs is still in bf->bf_skb. So when bf is returned with ath_return_txbuf it tries to double-free, resuilting a spew of backtraces/warnings when reference debugging is enabled. mtaylor 2008-01-19 01:02:26 +0000
  • 583f54226e Do not count requeued packets as discarded. mtaylor 2008-01-19 00:56:25 +0000
  • f5245bf3d0 Trivial spelling error mtaylor 2008-01-19 00:52:41 +0000
  • 9132c847c9 Add state debug messages when AP or STA returns to RUN state, indicating whether or not beacons were going to be reconfigured. Add logic so that any time a STA on an AP mode radio comes back to RUN state we will reconfigure beacons. Also reconfigure beacons on AP return to RUN state in a AP+STA configuration. In this AP+STA mode, we need to be careful to update the beacons any time a STA or AP comes online to make sure the APs start beaconing. This fixes issues where beacons are not restarted correctly after the station returns from SCAN/ASSOC/AUTH/INIT to RUN state while APs are pending completation of the STAs activity. mtaylor 2008-01-18 22:14:41 +0000
  • 9e039dcd6a Two more places where running out of txbuf should cause requeue mtaylor 2008-01-18 22:01:03 +0000
  • ac89d5ecc3 For any hardstart failure resulting from insufficient resources, set requeue. On requeue, stop the kernel queue and flag the fact that we stopped it in sc_dev_stopped so that it can be restarted automatically when resources become available. This is already done in the case of taking a txbuf, but can/should be done for every code path leading to requeuing. mtaylor 2008-01-18 21:32:17 +0000
  • 278c57c4e7 Move channel non-occupancy limits out of ieee80211_channel structure to avoid hitting the size limitations of wireless extensions. wext is a real pain. Now the values are stored in ic_chan_non_occupy[] instead. mtaylor 2008-01-18 21:25:05 +0000
  • a27f996ac4 Comment cleanup. No tagging in comments mentor 2008-01-18 17:28:43 +0000
  • 249e5b919b Bad merge from r3161; see r3123 for rationale mentor 2008-01-18 17:26:28 +0000
  • ac2124cce3 Another pass on r3156:r3200: * Lots of coding style and formatting, tedious and annoying * Quite a few bad merges as far as I can tell; at least one of which was serious - reintroduced a leak mentor 2008-01-18 01:28:43 +0000
  • 981463bb22 Minor formatting mentor 2008-01-17 23:31:09 +0000
  • bd7108a853 Do not try and find a device from a non-existent SKB mentor 2008-01-17 23:22:51 +0000
  • beb3ffc4eb Formatting; grammar - r3179 mentor 2008-01-17 02:51:32 +0000
  • 307d970bf5 Formatting r3185 mentor 2008-01-17 02:49:14 +0000
  • 3dd1a89219 Formatting r3187 - getting slightly annoying now mentor 2008-01-17 02:48:09 +0000
  • 86891cd55e Formatting and style for r3189 mentor 2008-01-17 02:45:41 +0000
  • 09ffca4d23 Formatting - r3191 mentor 2008-01-17 02:30:15 +0000
  • f1d15f48ce Invoke netif_stop_queue and requeue skb if we run out of txbuf instances in the pool mtaylor 2008-01-16 22:46:31 +0000
  • 7559e86881 Add support for variable number of VAPs. The hard coded maximum of 4 VAPs is replaced by a module parameter so that you can specify from 2 to 64 VAPs. The only quirk with high numbers of VAPs is that you must not attempt to use staggered beacons because the hardware cannot fire them fast enough. If more than 4 VAP maximum is specified, beacons are always bursted. This code has been tested successfully against 16-VAP APs without problems. mtaylor 2008-01-16 22:43:30 +0000
  • 4bf21e04c5 More meaningful error message mtaylor 2008-01-16 22:29:57 +0000
  • 1da2be285c Debug output when beacons are reconfigured. mtaylor 2008-01-16 22:28:49 +0000
  • d25dbe6d49 Add the VAP pointer and nickname when dumping state transitions mtaylor 2008-01-16 22:26:32 +0000