ID: 21
Name: BSP_Updates_742
Version: 3.89
Dependences: 20, 92, 28
Description: Version 1.0@@@Serial Driver - case 30221@@@Removed extra stop bit that was added between characters.@@@RTS was incorrectly being deasserted for 1us before transmitting each character in RS485 mode.@@@The serial driver was incorrectly using the CPU clock instead of the PLL for baud rate calculations.@@@Serial Driver - case 30049@@@serial_channel() now returns minor as the channel number.@@@max_serial_ports() now uses SERIAL_MAX rather than a hard coded value.@@@Serial Driver - case 29973@@@Fixed select() implementation.  We did not notify the Treck stack about event changes.  Added ioctl option TCGETSIGCHANGE@@@USB Driver@@@Updated ISR to call narm_read_reg and narm_write_reg.@@@See latest API Reference for updates to naUsbDevQueue about zero-length packets.@@@Wireless Driver - case 29954@@@Updated Japanese channels for 802.11a@@@Wireless Driver@@@Fixed Cisco LEAP+WEP problem.@@@SPI MMC Driver - case 28537@@@Fixes to SPI SD card driver to address SD cards that do not respond for various reasons. (usually timeouts)@@@Dialog - case 30203@@@Fixed compilation problems when BSP_WANT_ETHERNET and BSP_WANT_WIRELESS are both FALSE.@@@Version 1.1@@@In customizeBootloader.c, we overwrote the preconfigured GPIO line to check forced recovery status. This would cause problems with I2C when run from flash.  We now restore the GPIO line after checking status.@@@Fixed a problem that could cause the WIEM9210 to sleep forever when an application changes duty cycle time on the fly.@@@Fixed a problem with high packet loss on the WiWM9210 using AES on b-g WLAN with protection enabled. Protection mode allows a mix of B and G devices  on the same channel.  It sends a CTS frame (at a B rate) before every G frame.  The CTS frame tells B-only devices to stay off the air until the G frame is complete.  Protection mode is enabled by the access point. The problem our driver had was that the CTS and the DATA frames are not in the same FIFO so the AP thought they were invalid.@@@Version 1.2@@@Error building NET+OS examples caused by missing file.  Added usbreg_def.h@@@Version 1.3@@@Fixed a problem that would cause cout to stop working after any change to the dialog was made. Case 30472@@@Changed error checking code for BSP_GPIO_MUX_IOP_1_CAN_TXD and BSP_GPIO_MUX_IOP_1_CAN_TXD in gpio.h to allow the 2nd alternative path, which is a valid choice.@@@Added  A band  channels (52, 56, 60, 64, 100, 104, 108, 112,116, 120, 124, 128, 132, 136, 140) to Japanese support.@@@Inside the routine SetChannel() there was missing a setting of the global_channel.  Also added the file mac_11j.o to Makefile.files  Case 30569@@@Version 1.4@@@Corrected bug in NAMCWait which declared the timer handle to be static, thereby making it possible for one thread to alter the variable while a second thread is using it if both threads call into the function simultaneously.@@@Corrected dialog behavior when APP_USE_NETWORK_TIME_PROTOCOL was TRUE and the primary NVRAM SNTP address was 0.0.0.0.   In that case, the dialog never displayed the SNTP addresses and the operator would not know they can update it at that time.@@@Fixed incorrect size used in memcpy call in na_pton.@@@Version 1.5@@@Wireless Driver@@@Changed the receive gain table for airoha chips.@@@Version 1.6@@@Implemented max gain control to improve receive sensitivity.@@@Version 1.7@@@Fix to update the status when we set the tx power.  wln_get_status was not returning the correct value for tx power.@@@Version 1.8@@@Fixed alignment issue in MacBuffer that could cause Wi-Fi to crash when using Green Hills Tools.@@@Version 1.9@@@Fix for Ethernet receive stall after a soft reset under heavy network load on ARM9 based chips. Case 29694@@@Masked off interrupt for 9215 and 9210. This is needed since the hardware does not handle heavy nesting of interrupts correctly. Case 30345@@@Version 2.0@@@Implemented automatic adjustment of the MAX AGC setting based on the receive strength of received frames.@@@Added support for a NULL terminated array of trusted root certificates passed to the API wpa_set_trusted_certificates().@@@Added firmware to enable the second DC block HP bit in the transceiver.@@@Fixed passing an exception code to customizeExceptionHandler as an agrument, and added documentation about exception codes.@@@Version 2.1@@@Wireless change that will wait for an ack after sending a probe_response before setting the tx state to idle.@@@Moved min and max from fwtypes.h to minmax.h.   Define was causing build errors in C++ when using STL containers vector, map and string. Case 31471.@@@Version 2.2@@@Fixed problem where data overflow error was occurring when the camry serial driver is flow controlled. When the driver receives a XOFF it disables the transmitter, at this time it is unable to send XOFF to the peer if need be, resulting in data overflow. Case 31720@@@Updated readme files to match FIM Application Kit HW Reference.@@@Fixed Pll clk calls and updated FIM code in onewire driver.@@@Updated FIM1 1WIRE and FIM0 UART settings in bsp_iop.h for 9210 based platforms.@@@Fixed tx fifo underrun problem on Connectcorewi9c_a.@@@FPGA file modified to utilize the 4096 byte depth of the recv fifo on Connectcorewi9c_a.@@@Fixed tx fifo reset problem on piper with new macassist code.@@@Changed RF register 0xC setting from 0xA3 to 0x2B for al7230.@@@Limited Connectcorewi9p9215 operational max PSK CCK power to 16dbm.@@@Version 2.3@@@Moved naValidateIpv6Information to nainet.c, renamed validateIpAddress to naValidateIpAddress, Updated API Reference to add documentation to naValidateIpv6Information. Case 31920@@@Fixed a problem caused when there was a low heap situation. Case 31856@@@Changed the MAX_CODE_SIZE for the WiEM, WiSP, WiEM9210, and CCWi9P9215 platforms to 3.5M.  Note that all these platforms already have the APP_MAX_SIZE_IN_FLASH set to 1856K.  It is only natural that the MAX_CODE_SIZE (in RAM) should be about double the SIZE_IN_FLASH due to the compression mechanism. Case 31850@@@Now checking the NVRAM backup copy once, every time the unit comes up, to address the possibility of a corrupted or stale backup sector.  Case 31747@@@Version 2.4@@@Updated file system interface files to use an event flag instead of polling for the result to speed up file transfer. (case 32017)@@@Version 2.5@@@Corrected a problem that could cause the I2C bus to hang if a reset is issued during active I2C communications. (case 32488)@@@Corrected a problem with size check when downloading backup image, and increased the backup image size in the ConnectME and ConnectME9210 to 512K.  (case 32530)@@@Version 2.6@@@Set RTC device count to 0 in all ARM7, ns9750 and ns9210 based platforms. (case 32478)@@@Fixed a problem in the Ethernet driver.  The stack can sometimes send the driver Ethernet packets that are spread out across more than one buffer.  Apparently this happens a lot when using IPSEC, and often 3 buffers are sent for a single packet. The driver was not handling that well, and was getting confused.@@@Fixed customer reported rounding error in NATimerPeriod2Raw.  Error was about (5 100ths) of one percent. (case 32430)@@@Fixed a memory leak in firmware download process. (case 32636)@@@Fixed problem with lock mechanism in spimaster routine.@@@Version 2.7@@@Wireless Updates@@@Now allow duty cycling while scanning.@@@Added LINKUP and LINKDOWN events. Updated mwifimac for LINKUP and LINKDOWN event and added naWlnSetEvent for wpa to setup linkup and linkdown events and startMultiScan for multiple channel mask scan.@@@Added naEkahauSetBatteryCallback prototype and naEkahauSetStatusCallback API.@@@Added naWlnIsWirelessUp function to see whether wireless is ready to transmit. We now make a call to naWlnIsWirelessUp if no LINKUP event is received from the wifi driver to verify wifi is up before sending ELP.@@@wlnNullFramePtrPS and wln_make_null_frame_ps are added for background scans while staying associated.  The unit sends a NULL frame with the power-save bit set, so that the AP knows to buffer data for us. The unit sends a NULL frame with the power-save bit clear when the scan is done.@@@Multipath problem fixed by resetting the transceiver before changing channels.@@@Fixed problem when in A band and in ad-hoc mode that caused the unit to send out tons of RTS.@@@Fixed (Authentication type  WEP (802.1x))@@@Fixed a bug that was causing units to occasionally skip association to the nearest AP because the MacTimer() was not allowing enough time for a beacon to come in before it tried the next AP.@@@Version 2.8@@@Wireless Updates@@@FPGA file updated for Connectcorewi9c_a platform.@@@For Al2236 transceiver@@@Config 7   RSSI  default value@@@Original Setting 0x05B688@@@New Recommended Setting 0x0A8688@@@Note that the Airoha-recommended setting is dependent on the supply voltage - 0x05B688 for 2.8v, 0x0E4688 for 3.0v. The Piper WiEm voltage is actually around 2.88v, so the new value is consistent with the Airoha-recommended values (between the two settings). (case 32800)@@@Serial Updates@@@The multiplexor that controls the output enable for gpio29 is being controlled by the control register for gpio28. The work around and limitation is that gpio28 and gpio29 must always be configured to the same functional mode. (case 1284328)@@@Version 2.9@@@Wireless Updates@@@Fixed problem that casued WiEM 9210 to crash when naPdSleep() was called. (case 33148)@@@Fixed problem where after waking up, the device did not detect it was disassociated from its AP (case 33149)@@@Changed wireless driver so that when we fail EAP FAST authentication two times in a row, we will delay 1 minute before trying again.  We do this because the EAP FAST negotiation is CPU intensive and interferes with other processes. (case 33239)@@@Version 3.0@@@Increased read and write delay on CS1 to support 100ns flash parts on Connectcore9p9215 and Connectcorewi9p9215 modules.@@@Changed the CLI startup mechanism so each service can only be loaded once.  Updated to allow CLI open for serial CLI immediately after dialog, so that a wireless only driver can make use of this to configure its parameters. Telnet and SSH CLI opens take place after the stack is loaded. (case 33290) @@@Fixed an optimization problem on the fixes for serial port D. (case 1284328, 32984, 32985)@@@Now clear serial interrupt before entering sleep mode. (case 33143)@@@Now use the WAKE UP button for forced recovery for the cc9p9215, ccwi9p9215 and the ns9210 modules on the cc9p9215 carrier board. (case 33388)@@@Corrected problem that cause the Image not to execute properly when WriteToFlash is set to No in bootldr.dat (case 33386)@@@Calling wln_set_duty_cycle() on the fly could stall the wireless driver. (case 33142)@@@Version 3.1@@@Updated cli to describe channel index and real channel number. (case 33515)@@@Limited the max rates of Wi9P wireless driver on some A channels@@@Version 3.2@@@IamCallback now returns the gateway address based on interface.  This corrects problems displaying the wireless gateway in systems with more than 1 interface. (case 34058)@@@Fixed compiler warnings  (case 33872)@@@Updated I2C return codes (case 33952)@@@Fixed problem that would cause FIM UART to stop sending data (case 34179)@@@Version 3.3@@@Added NA_ETH_SET_PACKET_TRAILER_SIZE.  This constant can be used to enable a packet trailer in Ethernet packets. It is normally set to 0 and a trailer is not used.@@@Updated a condition check in tx index not to cross boundary. This was causing failures when using FIM1 UART. (case 35008)@@@Updated the NS9215 SPI Master driver to implement additional SPI ports using the FIM processors.  This was done to provide a work-around to problems in the internal NS9215 SPI port.@@@Added new roaming API to the Wi-Fi driver.  See API Reference for details.@@@Updated Wireless Powersave method.@@@Updated MAX_CODE_SIZE from 2M to 3M on ConnectME platform.@@@Added Support for WPA2 Mixed Mode. (cases 33127 and 34393).@@@Fixed IPv6 dialog configuration issue. (case 34923)@@@Version 3.4@@@Added missing file to fix build errors on 9210 related platforms@@@Version 3.5@@@Created new error code NASTATUS_SPI_HARDWARE_ERRATA which will be returned if the user attempts to configure a mode that the NS9215s internal SPI port cannot support.@@@Added new APIs to CAN Driver.  See API Reference for details. (Case 26317, 32022, 34921)@@@Version 3.6@@@Corrected behavior during simultaneous file access. (Case 36061, 34521)@@@Fixed uninitialized variable in spi_read_write. (Case 36106) @@@Added presence signal start and stop timing. Updated FIM to wait and detect presence signal within the start and stop times. Fixed close function to put the GPIO pin back default. (Case 35765)@@@Moved h/shared/nastatus.h to Common Updates Version 1.0@@@This update is dependent on Common Updates Version 1.0@@@Version 3.7@@@Possible serial transmit problem fixed by clearing transmit flow state on initialization during the device open() call. (Case 36318)@@@Implemented work-around to a bug in power save which prevented the  wake on Ethernet feature from working reliably on 100 Megabit  networks.  The work-around is to make sure the CPU clock is no less than 25 MHz.  The software will now set the CPU clock divisor before  going to sleep so that it is at least 25 MHz. (Case 35867)@@@Several changes to the FIM SPI support@@@1. The MISO and CS pins.  This was done to make it possible to support both SPI and serial port A on the ConnectME-9210.  In this case, SPI will be run without CS, but that is okay for most devices if only one device is supported.@@@2. SPI clock rates are now calculated in a different way.  The new  method allows the clock rates to be selected with greater granularity at higher speeds at the cost of not supporting the lower speeds.@@@3. New functions, NASpiWriteImmediately and NASpiReadImmediately, have been implemented that allow users to break up a SPI transfer into separate write and read operations.  The reason for doing this is that combining them as NASpiLowLatencyReadWrite does causes the calling thread to stall while the entire transfer takes place. By performing the operations seperately, the calling thread can do some processing while the transfer is taking place.@@@4. The SPI driver will now power down the FIM port while the SPI bus is idle.  This will only happen on the FIM SPI ports. @@@5. Added a new API, NASpiEmptyReadFifo, into the FIM SPI driver that  removes any data pending in the read FIFO.@@@6. Updated the API to include a write only mode.@@@7. Corrected calculation of FIM SPI clock rate.@@@Version 3.8@@@Fixed type in NASpiMasterFimAddDevice that could prevent the second SPI port from working. (Case 36807)@@@Corrected platform support for CANBUS@@@Fixed a bug where wireless devices may not respond to RTS packets if the data rate drops to 1 Mbps and the short preamble option is enabled.  This can cause the device to disassociate or fail authentication. (Case 36484)@@@Removed the serial select thread, this thread is not used and is left over from the Fusion stack.  This code was also opening two UDP ports unnecessarily. (Case 36437)@@@Fixed a buffer overflow in the wireless dialog, the overflow was overwriting stack variables causing the encryption parameters to be set back to default when modifying the dialog. (Case 36485)@@@Added sendBreak process. Updated serl_monitor to send interrupt command to FIM to set TX low for sendBreak and restore TX after given duration has expired. The FIM code was also updated for sendBreak command. Updated actual bytes written TX instead the number of bytes sent to FIM which includes stop bits and parity. (Case 31247)@@@Version 3.81@@@Fixed the InitTimer() to ensure delays between radio accesses. (Case 36555)@@@Fixed a problem in ParseTimElement() which could cause us to send spurious null-data frames when in low power mode.@@@Added LED support to Connect Wi-EM9210 platform.@@@Version 3.82@@@Fixed mis-matched index for tx_in and tx_out when IOP is done transmitting and the serial buffer is full of data. Updated TCSETA in ioctl to flush Tx and Rx queues. Updated IOP to flush DMA buffers in ISR, and reset DMA index from the register. When setting CA bit to flush DMA buffers, naIopDmaFlushQueue must wait until either all DMA buffers have been cleared in the ISR or the DMA state is idle (no interrupts pending).  (Case 37914)@@@Swapped CS and MISO so that users can use UARTA and FIM SPI by configuring SPI without CS on ConnectME9210. (Case 37833)@@@Added support for the 3033 variant of the cc9p9360 module.  This variant runs at 155MHz instead of 177MHz. (Case 34834)@@@Version 3.83@@@NET+OS serial driver should come up with no flow control. The serial driver was initialized to use SW flow control while loading the driver. This could halt the transmitter if it received a serial data frame with 0x13 in it.@@@Corrected typos and incorrect paths in IRQ1 and IRQ2    (Case 38288, 38337)@@@Fixed a problem when in ad-hoc mode and using search for channel. (Case 38469)@@@Fix problem in spectrum management (802.11h) and  multi domain (802.11d) (Case 43043)@@@Fixed error in roaming parameters. (Case 43278)@@@Version 3.84@@@Changed read and write access to 32-bit words in scratchpad driver.   Updated nascratchpad example readme file.  (Case 44550)@@@Removed calls to naIopPause() which was causing the SPI FIM clock to get stuck high, and fixed naIopPause() and naIopResume() for FIM 1. (Case 43026)@@@Fixed problem that was causing TFTP recovery to fail due to DHCP offer handling error. (Case 44186)@@@Set ENOENT errno instead of none in file system interface when open non-existing file to read.@@@Version 3.85@@@Improved power save mode, improved beacon handling and additional communication fixes.  (Case 45536, 45402, 45183)@@@Version 3.86@@@Fixed coding error to ensure the promiscuousMode bit is set if in monitor mode. (NET+OS-13)@@@Version 3.87@@@Fixed a problem with setting relative time alarm in RTC driver on 02/29 in a leap year. (NETOS-5)@@@Added code to skip empty calibration entries for the newly manufactured modules, which eliminated 4.9 GHz band. (NETOS-28)@@@Replaced calls to deprecated function NATimerConfigure with NATimerConfig in the natimer sample. (NETOS-7)@@@Fixed SSH security problem that allowed users to login into the module using SSH without user name/password. (NETOS-17)@@@Version 3.88@@@There was a bug such that the utility function CoreDump produced output on ARM7 processors, but produced nothing on ARM9 processors. It was found that the CoreDump code was executed, but no output was produced on the serial line through bsp_printf. The fix was to call function setupSimpleSerial at the beginning of function CoreDump. This ensured that all necessary GPIOs were in a known state prior to running CoreDump. ( NETOS-74)@@@Version 3.89@@@In function na9215A2dConfigure the mask used against the n value (sampling rate) was 0x3f. The problem is that the register section is 10 bits and 0x3f is only 6 bits. Mask should be 0x3ff  (10 bits)   (Case 00194980)
platformversion: 7.4
Revision: 2
platform: netos
Relevance: RECOMMENDED
Filename: BSP_Updates_742_21.dipk
Date: 21/10/2014
Type: fix
Target: environment
neededfiles: none
rootfs: none
minidescription: BSP Updates for NET+OS 7.4.2
Size: 5798kb
Installedsize: 20749kb
Checksum: d97ebe8d9a808384e4d99da081bee452
