1*** 2*** Summary of changes between 2005-05-18 and 2005-11-01. 3*** 4*** % setenv TZ UTC 5*** % cvs -d freebsdanoncvs@anoncvs.freebsd.org:/home/ncvs co src 6*** % cd src/sys/net80211/ 7*** % cvs2cl --follow trunk --hide-filenames --no-indent --no-wrap \ 8*** -f CHANGES -l '-d 2005-05-18<=2005-11-01' 9*** 10 112005-08-30 14:27 avatar 12 13Properly set ic_curchan before calling back to device driver to do channel 14switching(ifconfig devX channel Y). This fix should make channel changing 15works again in monitor mode. 16 17Submitted by: sam 18X-MFC-With: other ic_curchan changes 19 202005-08-13 17:50 sam 21 22revert 1.64: we cannot use the channel characteristics to decide when to 23do 11g erp sta accounting because b/g channels show up as false positives 24when operating in 11b. 25 26Noticed by: Michal Mertl 27 282005-08-13 17:31 sam 29 30Extend acl support to pass ioctl requests through and use this to 31add support for getting the current policy setting and collecting 32the list of mac addresses in the acl table. 33 34Submitted by: Michal Mertl (original version) 35MFC after: 2 weeks 36 372005-08-10 17:42 sam 38 39Don't use ic_curmode to decide when to do 11g station accounting, 40use the station channel properties. Fixes assert failure/bogus 41operation when an ap is operating in 11a and has associated stations 42then switches to 11g. 43 44Noticed by: Michal Mertl 45Reviewed by: avatar 46MFC after: 2 weeks 47 482005-08-10 16:22 sam 49 50Clarify/fix handling of the current channel: 51o add ic_curchan and use it uniformly for specifying the current 52 channel instead of overloading ic->ic_bss->ni_chan (or in some 53 drivers ic_ibss_chan) 54o add ieee80211_scanparams structure to encapsulate scanning-related 55 state captured for rx frames 56o move rx beacon+probe response frame handling into separate routines 57o change beacon+probe response handling to treat the scan table 58 more like a scan cache--look for an existing entry before adding 59 a new one; this combined with ic_curchan use corrects handling of 60 stations that were previously found at a different channel 61o move adhoc neighbor discovery by beacon+probe response frames to 62 a new ieee80211_add_neighbor routine 63 64Reviewed by: avatar 65Tested by: avatar, Michal Mertl 66MFC after: 2 weeks 67 682005-08-09 10:19 rwatson 69 70Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and 71IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to 72ifnet.if_drv_flags. Device drivers are now responsible for 73synchronizing access to these flags, as they are in if_drv_flags. This 74helps prevent races between the network stack and device driver in 75maintaining the interface flags field. 76 77Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; 78some less so. 79 80Reviewed by: pjd, bz 81MFC after: 7 days 82 832005-08-08 18:46 sam 84 85Split crypto tx+rx key indices and add a key index -> node mapping table: 86 87Crypto changes: 88o change driver/net80211 key_alloc api to return tx+rx key indices; a 89 driver can leave the rx key index set to IEEE80211_KEYIX_NONE or set 90 it to be the same as the tx key index (the former disables use of 91 the key index in building the keyix->node mapping table and is the 92 default setup for naive drivers by null_key_alloc) 93o add cs_max_keyid to crypto state to specify the max h/w key index a 94 driver will return; this is used to allocate the key index mapping 95 table and to bounds check table loookups 96o while here introduce ieee80211_keyix (finally) for the type of a h/w 97 key index 98o change crypto notifiers for rx failures to pass the rx key index up 99 as appropriate (michael failure, replay, etc.) 100 101Node table changes: 102o optionally allocate a h/w key index to node mapping table for the 103 station table using the max key index setting supplied by drivers 104 (note the scan table does not get a map) 105o defer node table allocation to lateattach so the driver has a chance 106 to set the max key id to size the key index map 107o while here also defer the aid bitmap allocation 108o add new ieee80211_find_rxnode_withkey api to find a sta/node entry 109 on frame receive with an optional h/w key index to use in checking 110 mapping table; also updates the map if it does a hash lookup and the 111 found node has a rx key index set in the unicast key; note this work 112 is separated from the old ieee80211_find_rxnode call so drivers do 113 not need to be aware of the new mechanism 114o move some node table manipulation under the node table lock to close 115 a race on node delete 116o add ieee80211_node_delucastkey to do the dirty work of deleting 117 unicast key state for a node (deletes any key and handles key map 118 references) 119 120Ath driver: 121o nuke private sc_keyixmap mechansim in favor of net80211 support 122o update key alloc api 123 124These changes close several race conditions for the ath driver operating 125in ap mode. Other drivers should see no change. Station mode operation 126for ath no longer uses the key index map but performance tests show no 127noticeable change and this will be fixed when the scan table is eliminated 128with the new scanning support. 129 130Tested by: Michal Mertl, avatar, others 131Reviewed by: avatar, others 132MFC after: 2 weeks 133 1342005-08-08 05:49 sam 135 136use ieee80211_iterate_nodes to retrieve station data; the previous 137code walked the list w/o locking 138 139MFC after: 1 week 140 1412005-08-08 03:30 sam 142 143Cleanup beacon/listen interval handling: 144o separate configured beacon interval from listen interval; this 145 avoids potential use of one value for the other (e.g. setting 146 powersavesleep to 0 clobbers the beacon interval used in hostap 147 or ibss mode) 148o bounds check the beacon interval received in probe response and 149 beacon frames and drop frames with bogus settings; not clear 150 if we should instead clamp the value as any alteration would 151 result in mismatched sta+ap configuration and probably be more 152 confusing (don't want to log to the console but perhaps ok with 153 rate limiting) 154o while here up max beacon interval to reflect WiFi standard 155 156Noticed by: Martin <nakal@nurfuerspam.de> 157MFC after: 1 week 158 1592005-08-06 04:57 sam 160 161fix debug msg typo 162 163MFC after: 3 days 164 1652005-08-06 04:56 sam 166 167Fix handling of frames sent prior to a station being authorized 168when operating in ap mode. Previously we allocated a node from the 169station table, sent the frame (using the node), then released the 170reference that "held the frame in the table". But while the frame 171was in flight the node might be reclaimed which could lead to 172problems. The solution is to add an ieee80211_tmp_node routine 173that crafts a node that does exist in a table and so isn't ever 174reclaimed; it exists only so long as the associated frame is in flight. 175 176MFC after: 5 days 177 1782005-07-31 06:12 sam 179 180close a race between reclaiming a node when a station is inactive 181and sending the null data frame used to probe inactive stations 182 183MFC after: 5 days 184 1852005-07-27 04:41 sam 186 187when bridging internally bypass the bss node as traffic to it 188must follow the normal input path 189 190Submitted by: Michal Mertl 191MFC after: 5 days 192 1932005-07-27 02:53 sam 194 195bandaid ni_fails handling so ap's with association failures are 196reconsidered after a bit; a proper fix involves more changes to 197the scanning infrastructure 198 199Reviewed by: avatar, David Young 200MFC after: 5 days 201 2022005-07-23 00:16 sam 203 204the AREF flag is only meaningful in ap mode; adhoc neighbors now 205are timed out of the sta/neighbor table 206 2072005-07-22 23:25 sam 208 209o move inactivity-related debug msgs under IEEE80211_MSG_INACT 210o probe inactive neighbors in adhoc mode (they don't have an 211 association id so previously were being timed out) 212 213MFC after: 3 days 214 2152005-07-22 21:11 sam 216 217split xmit of probe request frame out into a separate routine that 218takes explicit parameters; this will be needed when scanning is 219decoupled from the state machine to do bg scanning 220 221MFC after: 3 days 222 2232005-07-22 20:48 sam 224 225split 802.11 frame xmit setup code into ieee80211_send_setup 226 227MFC after: 3 days 228 2292005-07-22 17:57 sam 230 231simplify ic_newassoc callback 232 233MFC after: 3 days 234 2352005-07-22 17:54 sam 236 237simplify ieee80211_ibss_merge api 238 239MFC after: 3 days 240 2412005-07-22 17:50 sam 242 243add stats we know we'll need soon and some spare fields for future expansion 244 245MFC after: 3 days 246 2472005-07-22 17:45 sam 248 249simplify tim callback api 250 251MFC after: 3 days 252 2532005-07-22 17:42 sam 254 255don't include 802.3 header in min frame length calculation as it may 256not be present for a frag; fixes problem with small (fragmented) frames 257being dropped 258 259Obtained from: Atheros 260MFC after: 3 days 261 2622005-07-22 17:36 sam 263 264simplify ieee80211_node_authorize and ieee80211_node_unauthorize api's 265 266MFC after: 3 days 267 2682005-07-22 17:31 sam 269 270simplifiy ieee80211_send_nulldata api 271 272MFC after: 3 days 273 2742005-07-22 17:29 sam 275 276simplify rate set api's by removing ic parameter (implicit in node reference) 277 278MFC after: 3 days 279 2802005-07-22 17:21 sam 281 282reject association requests with a wpa/rsn ie when wpa/rsn is not 283configured on the ap; previously we either ignored the ie or (possibly) 284failed an assertion 285 286Obtained from: Atheros 287MFC after: 3 days 288 2892005-07-22 17:16 sam 290 291missed one in last commit; add device name to discard msgs 292 2932005-07-22 17:13 sam 294 295include device name in discard msgs 296 2972005-07-22 17:12 sam 298 299add diag msgs for frames discarded because the direction field is wrong 300 3012005-07-22 17:08 sam 302 303split data frame delivery out to a new function ieee80211_deliver_data 304 3052005-07-22 17:00 sam 306 307o add IEEE80211_IOC_FRAGTHRESHOLD for getting+setting the 308 tx fragmentation threshold 309o fix bounds checking on IEEE80211_IOC_RTSTHRESHOLD 310 311MFC after: 3 days 312 3132005-07-22 16:55 sam 314 315o add IEEE80211_FRAG_DEFAULT 316o move default settings for RTS and frag thresholds to ieee80211_var.h 317 3182005-07-22 16:50 sam 319 320diff reduction against p4: define IEEE80211_FIXED_RATE_NONE and use 321it instead of -1 322 3232005-07-22 16:37 sam 324 325add flags missed in last merge 326 3272005-07-22 16:36 sam 328 329Diff reduction against p4: 330o add ic_flags_ext for eventual extention of ic_flags 331o define/reserve flag+capabilities bits for superg, 332 bg scan, and roaming support 333o refactor debug msg macros 334 335MFC after: 3 days 336 3372005-07-22 05:17 sam 338 339send a response when an auth request is denied due to an acl; 340might be better to silently ignore the frame but this way we 341give stations a chance of figuring out what's wrong 342 3432005-07-22 05:15 sam 344 345remove excess whitespace 346 3472005-07-22 04:55 sam 348 349use IF_HANDOFF when bridging frames internally so if_start gets 350called; fixes communication between associated sta's 351 352MFC after: 3 days 353 3542005-07-11 03:06 sam 355 356Handle encrypt of arbitarily fragmented mbuf chains: previously 357we bailed if we couldn't collect the 16-bytes of data required 358for an aes block cipher in 2 mbufs; now we deal with it. While 359here make space accounting signed so a sanity check does the 360right thing for malformed mbuf chains. 361 362Approved by: re (scottl) 363 3642005-07-11 03:00 sam 365 366nuke assert that duplicates real check 367 368Reviewed by: avatar 369Approved by: re (scottl) 370 3712005-07-09 23:15 sam 372 373Change default key allocation method to do the right thing for 374legacy parts (i.e. those that have 4 global key slots). We 375blindly assign unicast keys to key slot 0. Devices that need 376alternate allocation logic must override this method. 377 378Reviewed by: avatar 379Approved by: re (scottl) 380 3812005-07-08 22:49 sam 382 383correct check for high priority wme traffic 384 385Noticed by: Ralf Assmann 386Reviewed by: apatti 387Approved by: re (scottl) 388 3892005-07-08 16:36 sam 390 391fix another instance of the MORE_DATA bit handling for frames on the 392power save queue (missed in previous commit) 393 394Submitted by: Bruno Randolf 395Approved by: re (scottl) 396 3972005-07-06 15:38 sam 398 399add "pureg" mode for ap operation: reject association requests from 40011b-only stations when operating in 11g 401 402Reviewed by: avatar 403Approved by: re (scottl) 404 4052005-07-06 01:55 sam 406 407Fix handling of data frames queued for a station in power save mode: 408don't mark the MORE_DATA bit when taking it off the ps queue, there's 409no 802.11 header then; we must wait to do this at encap time so 410mark the mbuf instead. 411 412Reviewed by: avatar 413Approved by: re (scottl) 414Obtained from: Atheros 415 4162005-07-06 01:51 sam 417 418Fix race condition in handling node reference counts for authenticating 419stations in ap mode. Track when a node's first auth frame is 420received and use this to decide whether or not to bump the refcnt. 421This insures we only ever bump the refcnt once. 422 423Reviewed by: avatar 424Approved by: re (scottl) 425 4262005-07-06 01:31 avatar 427 428Only update the scan entry state based on newly received frames. 429This fixes duplicative BSS entries(memory leaks as well) listed in 430"ifconfig dev list scan" when a station fails to associate with an AP. 431 432Reviewed by: sam 433Approved by: re (scottl) 434 4352005-07-05 18:05 sam 436 437remove auto-add of IEEE80211_KEY_GROUP; all the apps that need to 438set it have been fixed 439 440Reviewed by: avatar 441Approved by: re (scottl) 442 4432005-07-05 17:35 sam 444 445o when setting a wpa key, hold a ref on the bss node; 446 fixes a ref cnt leak 447o make unicast key handling on delete identical to set 448o change legacy wep key api to reset the 802.11 state 449 machine for backwards compatibility 450 451Reviewed by: avatar 452Approved by: re (scottl) 453 4542005-07-04 01:29 sam 455 456when operating in ap mode, explicitly drop associated/authenticated 457stations when transitioning to INIT state (e.g. as a result of 458changing state at the 802.11 level) 459 460Approved by: re (scottl) 461 4622005-06-13 21:01 sam 463 464revert 1.53; it breaks ibss merge 465 466Noticed by: Bruno Randolf 467Approved by: re (dwhite) 468 4692005-06-10 21:30 sam 470 471don't look at the wme ie in a beacon unless we negotiated use 472 4732005-06-10 16:49 brooks 474 475Stop embedding struct ifnet at the top of driver softcs. Instead the 476struct ifnet or the layer 2 common structure it was embedded in have 477been replaced with a struct ifnet pointer to be filled by a call to the 478new function, if_alloc(). The layer 2 common structure is also allocated 479via if_alloc() based on the interface type. It is hung off the new 480struct ifnet member, if_l2com. 481 482This change removes the size of these structures from the kernel ABI and 483will allow us to better manage them as interfaces come and go. 484 485Other changes of note: 486 - Struct arpcom is no longer referenced in normal interface code. 487 Instead the Ethernet address is accessed via the IFP2ENADDR() macro. 488 To enforce this ac_enaddr has been renamed to _ac_enaddr. 489 - The second argument to ether_ifattach is now always the mac address 490 from driver private storage rather than sometimes being ac_enaddr. 491 492Reviewed by: sobomax, sam 493 4942005-06-10 16:14 sam 495 496validate the bssid for non-data frames too when operating in 497adhoc/ahdemo/hostap modes 498 4992005-06-10 16:11 sam 500 501o fix wpa w/ wme: don't strip the QoS header on recv as tkip requires 502 it; instead pass the space occupied by the header down into the 503 crypto modules (except in the demic case which needs it only when 504 doing int in s/w) 505o while here fix defrag to strip the header from 2nd and later frames 506o teach decap code how to handle 4-address frames 507 5082005-06-10 05:04 sam 509 510mark stations authorized during recv processing instead of doing it 511as a side effect of sending an auth success frame; sending mgmt 512frames should not have side effects 513 5142005-06-10 04:42 sam 515 516move AID implementation defines from the protocol definitions to 517where they are used 518 5192005-06-10 04:37 sam 520 521accept diassoc frame in ASSOC state 522 5232005-06-10 01:48 sam 524 525pull some debug msgs up so they're seen more often 526 5272005-06-10 01:47 sam 528 529kick the state machine when we receive failure notice from an ap (when 530operating in sta mode); this speeds up the state machine, previously 531we were acting on a timeout 532 5332005-06-10 01:43 sam 534 535discard open auth requests in adhoc mode 536 5372005-06-10 01:41 sam 538 539reject open auth requests when shared key auth is configured 540 541Obtained from: Atheros 542 5432005-06-10 01:40 sam 544 545add ieee80211_send_error to encapsulate an idiom 546 5472005-06-10 01:38 sam 548 549o always check if ic_set_tim is !NULL before using it 550o add missing call to clear tim after flushing ps q 551 5522005-06-10 01:35 sam 553 554mark state for protection only when operating in 11g 555 5562005-06-10 01:33 sam 557 558don't reject station based on the PRIVACY bit in the capabilities; 559the 802.11 spec says not to 560 561Obtained from: Atheros 562 5632005-06-10 01:31 sam 564 565correct checks for rate set compatibility 566 5672005-06-10 01:29 sam 568 569record tstamp from beacons received in station mode when associated; 570this is needed by drivers that want to resync their timers based on 571the tsf of the last recv'd beacon frame 572 5732005-06-09 04:05 sam 574 575o collect dtim period+count from beacons in station mode so drivers 576 can better program beacon timers 577o leave placeholder in com structure for future ap/adhoc mode tim support 578 579Reviewed by: avatar 580 5812005-06-07 23:37 sam 582 583Change the MLME ASSOCIATE ioctl to accept either a ssid, a bssid, 584or a bssid+ssid. This is needed for later versions of wpa_supplicant 585and for forthcoming addons to wpa_supplicant. 586 587Note this is an api change and applications must be rebuilt. 588 5892005-06-07 23:31 sam 590 591Don't clock the state machine in various cases when roaming is set 592to manual; this helps keep wpa_supplicant in sync. 593 5942005-06-07 00:08 sam 595 596WPA/802.11i interoperability fixes: 597o only include capabilities word in the WPA ie when non-zero and 598 not preauth 599o always include the capabilities in the RSN ie 600 601Obtained from: Atheros 602 6032005-06-06 04:04 sam 604 605add force flag to enmic/demic crypto api for use in xmit fragmentation 606and h/w mic verification 607 608Reviewed by: avatar 609 610