Commit message (Collapse)AuthorAgeFilesLines
* Version 0.0.4 release.hostap_0_0_4Jouni Malinen2003-07-282-3/+3
* Added yield() compat macro for Linux 2.2.x.Jouni Malinen2003-07-281-0/+4
* Experimental algorithm in the driver did not include any feedback from theJouni Malinen2003-07-282-0/+27
| | | | | | | measured TX power and could allow TX power values that cause interference or signal quality issues; now only txpower 'off' and 'auto' are allowed. TX power controlling will be returned after proper algorithm using feedback loop has been implemented.
* Added PDA dumping functionality (hostap_diag -p wlan0).Jouni Malinen2003-07-281-2/+106
* Moved PDA loading into common utility file.Jouni Malinen2003-07-283-126/+135
* Yield to other processes instead of busy waiting during card initializationJouni Malinen2003-07-271-19/+23
| | | | and buffer allocation.
* Process beacon frames in kernel driver even when hostapd is used (thisJouni Malinen2003-07-273-5/+17
| | | | updates AP tables and allows automatic WDS link generation).
* Added more details for "Could not find STA for this TX error" message.Jouni Malinen2003-07-271-2/+3
* Removed pseudo-adhoc style communication between APs, i.e., drop TXJouni Malinen2003-07-271-6/+3
| | | | | | | | | | | | | packets to addresses that do not have STA entry (if they are not to WDS peer). Receiving pseudo-adhoc style frames was apparently broken before so this mode would not have worked anyway. In addition, it is not really needed anymore since WDS links can be used for this kind of links. Trying to send this kind of frames will in most cases cause multiply retries and eventual RetryErr event. This can happen, e.g., when using Linux bridge code. Even firmware lockup has been reported (using STA f/w 1.5.6) when a frame is send to own MAC address (addr1=addr2).
* Fixed Ethernet src_addr in TX descriptor when sending out WDS framesJouni Malinen2003-07-271-5/+4
| | | | using standard compliant 4-addr frames.
* Optimized hfa384x_wait_offset()Jouni Malinen2003-07-271-4/+6
| | | | | | | - patch from Jiri Fojtasek, but with longer BAP busy timeout (50000) - do not read register twice if BAP is not busy - use 1 usec resolution instead of 10 usec (average wait time 3 usec with PCI card)
* Divided prism2_hw_init() into two functions so that it is clearer whichJouni Malinen2003-07-261-5/+10
| | | | parts of the initialization is performed during firmware downloading.
* Only allow station firmware to be downloaded to non-volatile memory (flash)Jouni Malinen2003-07-261-0/+5
| | | | since Host AP driver does not yet support Primary f/w downloading.
* Allow compilation with Linux 2.6.x.Jouni Malinen2003-07-261-8/+9
* Added support for manual scan and roaming mode (managed/ad-hoc);Jouni Malinen2003-07-128-13/+144
| | | | | prism2_param host_roaming 2; this disabled automatic IEEE 802.11 scans and gives more control to user space programs using Linux wireless extensions.
* Added support for selection which authentication algorithm to use inJouni Malinen2003-07-124-4/+14
| | | | | | station mode (prism2_param 'ap_auth_algs', 1 = open system, 2 = shared key (requires WEP), 3 = both, default) (ap_auth_algs was previously used only in AP/Master mode, but it is now also used for STA mode configuration).
* Fixed enh_sec setting (hide SSID) to reset Port0 so that the changesJouni Malinen2003-07-123-2/+4
| | | | to the beacon are taken into use immediately.
* Do not report dropping of frames without ToDS or FromDS flags if they areJouni Malinen2003-07-112-7/+10
| | | | | directed to foreign BSSID. This kind of frames are normal broadcast frames from IBSS networks and should be dropped silently.
* Updated.Jouni Malinen2003-07-111-14/+9
* Changed monitor mode to use pseudo-IBSS mode instead of Host AP modeJouni Malinen2003-07-112-0/+4
| | | | to avoid sending out Beacon and ProbeResp frames.
* Use length field in struct iw_point to determine SSID length. This allowsJouni Malinen2003-07-111-2/+2
| | | | both NUL terminated and not NUL terminated data.
* Cleaned up SSID use: do not store extra +1 in ssid_len. Only add it whenJouni Malinen2003-07-114-8/+8
| | | | doing ioctl() to match iwconfig behavior.
* Abort configuration line processing if the invalid line does not containJouni Malinen2003-07-111-0/+1
| | | | '=' to avoid segfaulting on NULL pointer dereference.
* Added module parameter 'dev_template' for setting the device nameJouni Malinen2003-07-102-2/+17
| | | | template (default: 'wlan%d'); this is for hostap_{cs,pci,plx}.o.
* Patch from Pavel Roskin:Jouni Malinen2003-07-101-0/+3
| | | | | | | | | | | Using AccessConfigurationRegister is unsafe is the card has been removed. There is a long-standing bug in the PCMCIA drivers that causes a kernel oops. This bug was been patched in pcmcia-cs, but it's still present in the kernel PCMCIA drivers (both 2.4.x and 2.5.x). I have just sent a patch to LKML, but I think it's a good idea to put protection into the HostAP driver itself, especially because it's reasonable and simple.
* Removed backward compatibility code for Linux Wireless Extensions versionsJouni Malinen2003-07-103-33/+6
| | | | | | older than ver 9. All supported Linux version (2.2.x, 2.4.x, 2.5.x) have had ver 9 or newer available for a long time so this should not be a major issue.
* Moved Linux Wireless Extensions compatibility code into a new header file,Jouni Malinen2003-07-108-92/+92
| | | | hostap_wext.h.
* Updated.Jouni Malinen2003-07-091-10/+13
* Cleaned up RX handling of frames with too long or negativeJouni Malinen2003-07-091-19/+36
| | | | | | data_len. Accept these in monitor mode, but drop in all other modes. Handle negative length properly so that dev_alloc_skb() is used with large enough buffer (system crashes reported in monitor mode).
* Fixed thinko.Jouni Malinen2003-07-091-1/+2
* Do not include linux/wireless.h if CONFIG_NET_RADIO is not set.Jouni Malinen2003-07-041-2/+2
| | | | | Use iw_mode module param even without WIRELESS_EXT. Patch from Pavel Roskin.
* Updated to match current code.Jouni Malinen2003-07-041-0/+17
* Removed extra linefeed from frame header dump.Jouni Malinen2003-07-031-1/+1
* Fixed TX using wlan#ap (i.e., from hostapd) to buffer (other than auth/Jouni Malinen2003-06-251-1/+20
| | | | | (re)assoc) frames to power saving stations. This fixes inactivity polling from hostapd and IEEE 802.1X when STA is using power saving.
* Changed STA activity polling to use TX status callback instead ofJouni Malinen2003-06-256-46/+57
| | | | | | | counting TX Exc events. This fixes expiration of power saving STAs when using kernel driver for 802.11 management. This should also fix hostapd, but since kernel driver does not yet use power save buffering for wlan#ap interface, this does not really work properly.
* Changed IEEE 802.11 authentication to mark STA authenticated when sendingJouni Malinen2003-06-091-0/+8
| | | | | | | | authentication success message and not only after receiving ACK for this frame. IEEE 802.11 standard does not clearly state whether authentication state is changed on ACK or Auth sending. Setting it without waiting for ACK should not cause any problems and is not a security issue. Old behavior can be configured by defining IEEE80211_REQUIRE_AUTH_ACK.
* External pcmcia-cs package renames CONFIG_NET_RADIO toJouni Malinen2003-06-074-8/+8
| | | | | CONFIG_NET_PCMCIA_RADIO, so use both of these to select whether linux/wireless.h should be included.
* Use less aggressive transmit rate decreasing algorithm: decrease rate onlyJouni Malinen2003-06-062-3/+12
| | | | | | after WLAN_RATE_DECREASE_THRESHOLD consecutive dropped packets. Old behavior matches with value 1, new default is 2. Based on patch from Martin Lamb / Joe Parks.
* Added a warning if kernel configuration does not enable wireless extensionsJouni Malinen2003-06-061-0/+6
| | | | | (CONFIG_NET_RADIO) since they are needed for most configuration commands and hostapd does not work without them.
* Fixed compilation when kernel is configured not to support Linux WirelessJouni Malinen2003-06-065-11/+15
| | | | Extensions (CONFIG_NET_RADIO).
* Verify that eapol_sm_alloc() succeeded before derefering the returnedJouni Malinen2003-06-061-1/+2
| | | | pointer.
* Fixed ap_bridge_packets to use correct destination address (local copy,Jouni Malinen2003-06-031-2/+2
| | | | | instead of addr3 from modified header). This fixes bridging frames between two associated STAs.
* Verify that basic rate set is a non-empty subset of supported rate set toJouni Malinen2003-05-291-0/+7
| | | | avoid invalid configuration.
* Update basic rate set automatically, when changing supported rate set toJouni Malinen2003-05-291-1/+16
| | | | not include all previous basic rates.
* Do not decrement sta->users when in the case it was not incremented (whenJouni Malinen2003-05-291-0/+1
| | | | STA entry existed, but it was not marked authenticated).
* Patch from Pavel Roskin (note: the part about changing INCLUDES wasJouni Malinen2003-05-261-1/+25
| | | | | | | | | | | | | | | | | | | | | | | not applied): The attached patch exports PCMCIA_PATH and also introduces some basic checks to avoid common errors. If PCMCIA_PATH is set to an invalid or missing directory, exit with an error. If both CONFIG_PCMCIA and PCMCIA_PATH are set, it's an error too - you don't need to specify PCMCIA_PATH if kernel PCMCIA support is present. If neither CONFIG_PCMCIA nor PCMCIA_PATH are defined, it's a warning. Ideally, hostap_cs shouldn't be built and installed in this case, but let's leave it for another patch. Since depmod is run after every installed modules, make can fail because of unresolved symbols if the already installed modules that are about to be replaced. I see two ways to avoid this problem. One is to create private targets like _install_pcard without depmod and make sure that public targets only call depmod at the end. Another approach is to introduce "uninstall" target that would remove old modules before depmod is run the first time. The patch implements the second approach.
* Revert changes in rev 1.41 since top-level Host AP Makefile is still usedJouni Malinen2003-05-261-0/+31
| | | | | when compiling for Linux 2.2.x. In addition, it can be used as an alternative method for compiling Linux 2.4.x versions.
* Fixed modprobe examples not to include '.o' in module names.Jouni Malinen2003-05-241-2/+2
* Show IntEn register contents in "command was not completed" error messageJouni Malinen2003-05-231-2/+3
| | | | to ease debugging.
* TX callback handler is now called from a tasklet, not hw irq handler.Jouni Malinen2003-05-232-41/+1
| | | | | | Remove tx_callback_list and separate tasklet for processing these frames since there is no need to avoid spending more time in hw irq handler anymore.