changes to the Makefile and uncomment -DHERMES25. This will build
driver wlags49_h25_cs.
-Note: You can detemine the type with command "pccardctrl info"
+Note: You can determine the type with command "pccardctrl info"
MANIFID: 0156,0002 = HERMES - not supported by this driver
MANIFID: 0156,0003 = HERMES II (Wireless B)
MANIFID: 0156,0004 = HERMES II.5 (Wireless B/G)
-After succesfull compile type command
+After successful compile type command
sudo make install
changed; the former ioctl functions are now called before "open" and
after "close", which was not expected. One of the problems was enable/
disable of interrupts in the HCF. Interrupt handling starts at "open"
-so if a former "ioctl" routinge is called before "open" or after "close"
+so if a former "ioctl" routine is called before "open" or after "close"
then nothing should be done with interrupt switching in the HCF. Once
this was solved most HCF_ASSERTS went away.
does not work anymore but it provides some information about all the
settings.
-I have not have personal contact with Agere, but others have. Agere
-agreed to make their software available under the BSD licence.
+I have no personal contact with Agere, but others have. Agere
+agreed to make their software available under the BSD license.
This driver is based on the 7.22 version.
The following was mailed by Agere to Andrey Borzenkov about this:
-First of all, the best thing would be that this driver becomes obsolte by
+First of all, the best thing would be that this driver becomes obsolete by
adding support for Hermes II and Hermes II.5 cards to the existing orinoco
driver. The orinoco driver currently only supports Hermes I based cards.
Since this will not happen by magic and has not happened until now this
TODO:
- verify against a Hermes II.5 card
- - verify with WPA encription (both with H2 and H2.5 cards)
+ - verify with WPA encryption (both with H2 and H2.5 cards)
- sometimes the card does not initialize correctly, retry mechanisms
- are build in to catch most cases but not all
+ are built in to catch most cases but not all
- once the driver runs it is very stable, but I have the impression
- some the crittical sections take to long
+ that some of the critical sections take some time.
- the driver is split into a Hermes II and a Hermes II.5 part, it
would be nice to handle both with one module instead of two
- review by the wireless developer community
DONE:
- verified against a Hermes II card (Thomson Speedtouch 110 PCMCIA
card)
- - verified with WEP encription
+ - verified with WEP encryption
Please send any patches or complaints about this driver to Greg
Kroah-Hartman <greg@kroah.com> and Cc: Henk de Groot <pe1dnn@amsat.org>
* - HCF_ACT_INT_FORCE_ON enable interrupt generation by WaveLAN NIC
* - HCF_ACT_INT_OFF disable interrupt generation by WaveLAN NIC
* - HCF_ACT_INT_ON compensate 1 HCF_ACT_INT_OFF, enable interrupt generation if balance reached
- * - HCF_ACT_PRS_SCAN Hermes Probe Respons Scan (F102) command
+ * - HCF_ACT_PRS_SCAN Hermes Probe Response Scan (F102) command
* - HCF_ACT_RX_ACK acknowledge non-DMA receiver to Hermes
* - HCF_ACT_SCAN Hermes Inquire Scan (F101) command (non-WARP only)
* - HCF_ACT_SLEEP DDS Sleep request
* The F/W is wokenup by the HCF when the NIC Interrupts mode are disabled, i.e. at the first HCF_ACT_INT_OFF
* after going into sleep.
*
- * The following Miscellanuous actions are defined:
+ * The following Miscellaneous actions are defined:
*
* o HCF_ACT_RX_ACK: Receiver Acknowledgement (non-DMA, non-USB mode only)
* Acking the receiver, frees the NIC memory used to hold the Rx frame and allows the F/W to
* If the MSF does not need access (any longer) to the current frame, e.g. because it is rejected based on the
* look ahead or copied to another buffer, the receiver may be acked. Acking earlier is assumed to have the
* potential of improving the performance.
- * If the MSF does not explitly ack te receiver, the acking is done implicitly if:
+ * If the MSF does not explicitly ack the receiver, the acking is done implicitly if:
* - the received frame fits in the look ahead buffer, by the hcf_service_nic call that reported the Rx frame
* - if not in the above step, by hcf_rcv_msg (assuming hcf_rcv_msg is called)
* - if neither of the above implicit acks nor an explicit ack by the MSF, by the first hcf_service_nic after
* The Inquire Tallies command requests the F/W to provide its current set of tallies.
* See also hcf_get_info with CFG_TALLIES as parameter.
*
- * o HCF_ACT_PRS_SCAN: Inquire Probe Respons Scan command
+ * o HCF_ACT_PRS_SCAN: Inquire Probe Response Scan command
* This command is only operational if the F/W is enabled.
- * The Probe Respons Scan command starts a scan sequence.
+ * The Probe Response Scan command starts a scan sequence.
* The HCF puts the result of this action in an MSF defined buffer (see CFG_RID_LOG_STRCT).
*
* o HCF_ACT_SCAN: Inquire Scan command
* - NIC interrupts are not disabled while required by parameter action.
* - an invalid code is specified in parameter action.
* - HCF_ACT_INT_ON commands outnumber the HCF_ACT_INT_OFF commands.
- * - reentrancy, may be caused by calling hcf_functions without adequate protection against NIC interrupts or
+ * - reentrancy, may be caused by calling hcf_functions without adequate protection against NIC interrupts or
* multi-threading
*
* - Since the HCF does not maintain status information relative to the F/W enabled state, it is not asserted
* change in HREG_EV_STAT matching a bit in HREG_INT_EN, i.e. not if invoked as result of another device
* generating an interrupt on the shared interrupt line.
* Note 1: it has been observed that under certain adverse conditions on certain platforms the writing of
- * HREG_INT_EN can apparently fail, therefor it is paramount that HREG_INT_EN is written again with 0 for
+ * HREG_INT_EN can apparently fail, therefore it is paramount that HREG_INT_EN is written again with 0 for
* each and every call to HCF_ACT_INT_OFF.
* Note 2: it has been observed that under certain H/W & S/W architectures this logic is called when there is
* no NIC at all. To cater for this, the value of HREG_INT_EN is validated. If the unused bit 0x0100 is set,
* - A command other than Continue, Enable, Disable, Connect or Disconnect is given.
* - An invalid combination of the subfields is given or a bit outside the subfields is given.
* - any return code besides HCF_SUCCESS.
- * - reentrancy, may be caused by calling a hcf_function without adequate protection against NIC interrupts or
+ * - reentrancy, may be caused by calling a hcf_function without adequate protection against NIC interrupts or
* multi-threading
*
*.DIAGRAM
* hcf_connect passes the MSF-defined location of the IFB to the HCF and grants or revokes access right for the
* HCF to the IFB. Revoking is done by specifying HCF_DISCONNECT rather than an I/O address for the parameter
* io_base. Every call of hcf_connect in "connect" mode, must eventually be followed by a call of hcf_connect
- * in "disconnect" mode. Clalling hcf_connect in "connect"/"disconnect" mode can not be nested.
+ * in "disconnect" mode. Calling hcf_connect in "connect"/"disconnect" mode can not be nested.
* The IFB address must be used as a handle with all subsequent HCF-function calls and the HCF uses the IFB
* address as a handle when it performs a call(back) of an MSF-function (i.e. msf_assert).
*
* specification for S/W reset
* Note 2: it turns out that on some H/W constellations, the clock to access the EEProm is not lowered
* to an appropriate frequency by HREG_IO_SRESET. By giving an HCMD_INI first, this problem is worked around.
- *2b: Experimentally it is determined over a wide range of F/W versions that waiting for the for Cmd bit in
+ *2b: Experimentally it is determined over a wide range of F/W versions that are waiting for the for Cmd bit in
* Ev register gives a workable strategy. The available documentation does not give much clues.
*4: clear and initialize the IFB
* The HCF house keeping info is designed such that zero is the appropriate initial value for as much as
#define CFG_CNF_WDS_ADDR6 0xFC16 //[AP] Port 6 MAC Adrs of corresponding WDS Link node
#define CFG_CNF_PM_MCAST_BUF 0xFC17 //[AP] Switch for PM buffereing of Multicast Messages
#define CFG_CNF_MCAST_PM_BUF CFG_CNF_PM_MCAST_BUF //name does not match H-II spec
-#define CFG_CNF_REJECT_ANY 0xFC18 //[AP] Switch for PM buffereing of Multicast Messages
+#define CFG_CNF_REJECT_ANY 0xFC18 //[AP] Switch for PM buffering of Multicast Messages
#define CFG_CNF_ENCRYPTION 0xFC20 //select en/de-cryption of Tx/Rx messages
#define CFG_CNF_AUTHENTICATION 0xFC21 //[STA] selects Authentication algorithm
#define CFG_CURRENT_LINK_STATUS 0x090B //Latest link status got through 0xF200 LinkEvent
/*============================================================ INFORMATION FRAMES =========================*/
-#define CFG_INFO_FRAME_MIN 0xF000 //lowest value representing an Informatio Frame
+#define CFG_INFO_FRAME_MIN 0xF000 //lowest value representing an Information Frame
#define CFG_TALLIES 0xF100 //Communications Tallies
#define CFG_SCAN 0xF101 //Scan results
#include "hcfcfg.h" // to get hcf_16 etc defined as well as
- // possible settings which inluence mdd.h or dhf.h
+ // possible settings which influence mdd.h or dhf.h
#include "mdd.h" //to get COMP_ID_STA etc defined
#include "dhf.h" //used to be "fhfmem.h", to get memblock,plugrecord,
#include "hcfcfg.h" // to get hcf_16 etc defined as well as
- // possible settings which inluence mdd.h or dhf.h
+ // possible settings which influence mdd.h or dhf.h
#include "mdd.h" //to get COMP_ID_STA etc defined
#include "dhf.h" //used to be "fhfmem.h", to get memblock,plugrecord,
/*******************************************************************************
- * function prrottypes
+ * function prototypes
******************************************************************************/
int wl_wep_code( char *szCrypt, char *szDest, void *Data, int nLen );
DESC_STRCT *rx_packet[NUM_RX_DESC];
DESC_STRCT *rx_reclaim_desc, *tx_reclaim_desc; // Descriptors for host-reclaim purposes (see HCF)
int tx_rsc_ind; // DMA Tx resource indicator is maintained in the MSF, not in the HCF
- int rx_rsc_ind; // Also added rx rsource indicator so that cleanup can be performed if alloc fails
+ int rx_rsc_ind; // Also added rx resource indicator so that cleanup can be performed if alloc fails
int status;
} DMA_STRCT;
* constant definitions
******************************************************************************/
-/* Allow support for calling system fcns to access F/W iamge file */
+/* Allow support for calling system fcns to access F/W image file */
#define __KERNEL_SYSCALLS__
/*******************************************************************************
CNV_INT_TO_LITTLE( lp->hcfCtx.IFB_FWIdentity.version_major ),
CNV_INT_TO_LITTLE( lp->hcfCtx.IFB_FWIdentity.version_minor ));
- /* now we wil get the MAC address of the card */
+ /* now we will get the MAC address of the card */
lp->ltvRecord.len = 4;
if ( CNV_INT_TO_LITTLE( lp->hcfCtx.IFB_FWIdentity.comp_id ) == COMP_ID_FW_AP ) {
lp->ltvRecord.typ = CFG_NIC_MAC_ADDR;
lp->ltvRecord.len = 2;
lp->ltvRecord.typ = CFG_CNTL_OPT;
- /* The Card Services build must ALWAYS configure for 16-bit I/O. PCI or
+ /* The Card Services build must ALWAYS be configured for 16-bit I/O. PCI or
CardBus can be set to either 16/32 bit I/O, or Bus Master DMA, but only
for Hermes-2.5 */
#ifdef BUS_PCMCIA
*
* DESCRIPTION:
*
- * This function perfroms a check on the device and calls wl_remove() if
+ * This function performs a check on the device and calls wl_remove() if
* necessary. This function can be used for all bus types, but exists mostly
* for the benefit of the Card Services driver, as there are times when
* wl_remove() does not get called.
lp->portState = WVLAN_PORT_STATE_ENABLED; //;?bad mnemonic, NIC iso PORT
#ifdef ENABLE_DMA
if ( lp->use_dma ) {
- wl_pci_dma_hcf_supply( lp ); //;?always succes?
+ wl_pci_dma_hcf_supply( lp ); //;?always successful?
}
#endif
}
* DESCRIPTION:
*
* This function will perform the tedious task of endian translating all
- * fields withtin a mailbox message which need translating.
+ * fields within a mailbox message which need translating.
*
* PARAMETERS:
*
*
* DESCRIPTION:
*
- * This function will process the mailbox data.
+ * This function processes the mailbox data.
*
* PARAMETERS:
*
* DESCRIPTION:
*
* Used to start the netif queues of all the "virtual" network devices
- * which repesent the WDS ports.
+ * which represent the WDS ports.
*
* PARAMETERS:
*
* DESCRIPTION:
*
* Used to stop the netif queues of all the "virtual" network devices
- * which repesent the WDS ports.
+ * which represent the WDS ports.
*
* PARAMETERS:
*
* DESCRIPTION:
*
* Used to wake the netif queues of all the "virtual" network devices
- * which repesent the WDS ports.
+ * which represent the WDS ports.
*
* PARAMETERS:
*
* DESCRIPTION:
*
* Used to signal the network layer that carrier is present on all of the
- * "virtual" network devices which repesent the WDS ports.
+ * "virtual" network devices which represent the WDS ports.
*
* PARAMETERS:
*
* DESCRIPTION:
*
* Used to signal the network layer that carrier is NOT present on all of
- * the "virtual" network devices which repesent the WDS ports.
+ * the "virtual" network devices which represent the WDS ports.
*
* PARAMETERS:
*
******************************************************************************/
void wl_adapter_cleanup_module( void )
{
- //;?how comes wl_adapter_cleanup_module is located in a seemingly pci specific module
+ //;?how come wl_adapter_cleanup_module is located in a seemingly pci specific module
DBG_FUNC( "wl_adapter_cleanup_module" );
DBG_ENTER( DbgInfo );
* DESCRIPTION:
*
* Allocates a single Rx packet, consisting of two descriptors and one
- * contiguous buffer. THe buffer starts with the hermes-specific header.
+ * contiguous buffer. The buffer starts with the hermes-specific header.
* One descriptor points at the start, the other at offset 0x3a of the
* buffer.
*
if (retries < 10) {
retries++;
- /* Holding the lock too long, make a gap to allow other processes */
+ /* Holding the lock too long, makes a gap to allow other processes */
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
goto out_unlock;
}
- /* Holding the lock too long, make a gap to allow other processes */
+ /* Holding the lock too long, makes a gap to allow other processes */
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
}
}
- /* Holding the lock too long, make a gap to allow other processes */
+ /* Holding the lock too long, makes a gap to allow other processes */
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
/* Encryption */
- /* Holding the lock too long, make a gap to allow other processes */
+ /* Holding the lock too long, makes a gap to allow other processes */
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
// Retry Limits and Lifetime - NOT SUPPORTED
- /* Holding the lock too long, make a gap to allow other processes */
+ /* Holding the lock too long, makes a gap to allow other processes */
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
DBG_TRACE( DbgInfo, "CFG_SCAN_CHANNEL result : 0x%x\n", status );
- // Holding the lock too long, make a gap to allow other processes
+ // Holding the lock too long, makes a gap to allow other processes
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
DBG_TRACE( DbgInfo, "Reset card to recover, attempt: %d\n", retries );
wl_reset( dev );
- // Holding the lock too long, make a gap to allow other processes
+ // Holding the lock too long, makes a gap to allow other processes
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
status = hcf_put_info( &( lp->hcfCtx ), (LTVP)&( lp->ltvRecord ));
- // Holding the lock too long, make a gap to allow other processes
+ // Holding the lock too long, makes a gap to allow other processes
wl_unlock(lp, &flags);
wl_lock( lp, &flags );
/* Initiate the scan */
/* NOTE: Using HCF_ACT_SCAN has been removed, as using HCF_ACT_ACS_SCAN to
- retrieve probe responses must always be used to support WPA */
+ retrieve probe response must always be used to support WPA */
status = hcf_action( &( lp->hcfCtx ), HCF_ACT_ACS_SCAN );
if( status == HCF_SUCCESS ) {
* Make sure that there is no data queued up in the firmware
* before setting the TKIP keys. If this check is not
* performed, some data may be sent out with incorrect MIC
- * and cause synchronizarion errors with the AP
+ * and cause synchronization errors with the AP
*/
/* Check every 1ms for 100ms */
for (count = 0; count < 100; count++) {
this event BEFORE sending the association event, as there are timing
issues with the hostap supplicant. The supplicant will attempt to process
an EAPOL-Key frame from an AP before receiving this information, which
- is required properly process the said frame. */
+ is required for a proper processed frame. */
wl_wext_event_assoc_ie( dev );
/* Get the BSSID */
projects / linux-beck.git / commitdiff