without loading the ndk example, the packets were not able to be routed to the host cores. in this case, i sent two udp packets from a remote server. the statics print outs from the example shows that the packets did in fact get to the PA, but it wasn't properly routed to the C2 classify engines. Just to reiterate, using the same code, i was able to receive packets if i loaded an NDK example before loading the PA simple example without resetting the DSP board. see attachment for my code. so i was wonder what can cause this issue? why is it that loading ndk helps getting PA to work? maybe switch configuration is different? i've been stuck on this issue for quite a while and needs some assistance.

--- PA STATS ---
C1 number of packets: 98
C1 number IPv4 packets: 0
C1 number IPv6 packets: 0
C1 number custom packets: 0
C1 number SRIO packets: 0
C1 number llc/snap fail: 16
C1 number table matched: 2 //this is the packet i sent from remote server
C1 number failed table matched: 78
C1 number Ingress IP frags: 0
C1 number IP depth overflow: 0
C1 number vlan depth overflow: 0
C1 number gre depth overflow: 0
C1 number mpls packets: 0
C1 number of parse fail: 0
C1 number invalid IPv6 opts: 0
C1 number of Egress IP frags: 0
C1 number of silent discard: 94
C1 number of invalid control: 0
C1 number of invalid states: 0
C1 number of system fails: 1 //this seems really suspicious

C2 number of packets: 1
C2 number of UDP packets: 0
C2 number of TCP packets: 0
C2 number of custom packets: 0
C2 number of silent discard: 0
C2 number of invalid control: 0

Modify number of command file: 0

(Please visit the site to view this file)

↧

RTOS/AM5728: CSL MCASP_Audio Example on C66x DSP

January 10, 2019, 12:20 pm

≫ Next: SMOMAPL138B-HIREL: Read Timing Analysis

≪ Previous: RTOS/TMS320C6678: PA simple example not receiving packets from external devices

Part Number:AM5728

Tool/software: TI-RTOS

Hi,

I generated MCASP_Audio Example CCS project for cortex A15 and it works smoothly. But I'm intrested in use of mcasp csl driver to use on DSP to perform signal processing only there. So my question is: is there some simple way to migrate this example to run on C66x DSP core? If not could you guys provide some useful direction how to do this (or if you have some impediments on mind)? I'm spend some time with this but I cant see where to start.

↧

SMOMAPL138B-HIREL: Read Timing Analysis

January 10, 2019, 1:26 pm

≫ Next: AM5728: AM5728 - Have 1 PRU port write out data to a Backnet device, Have the 2nd PRU port read in data from a Modbus TCP device

≪ Previous: RTOS/AM5728: CSL MCASP_Audio Example on C66x DSP

Part Number:SMOMAPL138B-HIREL

Hi there,

We are trying to perform read timing analysis for EMIFA Asynchronous Memory Interface. We are having trouble showing that we have enough data hold time. The NOR flash device we are interfacing to has an output hold time of 0ns (zero) from the time /OE goes high. The hold time requirement for the SMOMAPL138B-HiRel processor is also 0ns. Since the processor samples the data at the same time it drives /OE high (rising edge of EMA_CLK) its difficult to show positive timing margin. Normally there is a parameter in processor datasheets for internal clock (EMA_CLK in this case) to signal output delay. In this case I'm looking for EMA_CLK to /EMA_OE minimum time. This normally helps to achieve the required timing margin... but we can't see it in the datasheet.

We would appreciate some help finding the missing parameter or any other suggestions you may have.

Regards,

Matthew

↧

AM5728: AM5728 - Have 1 PRU port write out data to a Backnet device, Have the 2nd PRU port read in data from a Modbus TCP device

January 10, 2019, 2:01 pm

≫ Next: Linux/PROCESSOR-SDK-OMAPL138: Unable to load DSP Firmware from Linux using remoteproc.0 because DSP contains partial memory in IRAM

≪ Previous: SMOMAPL138B-HIREL: Read Timing Analysis

Part Number:AM5728

Hello:

In all the documentation for the PRU network communication, I haven't found any information regarding compatibility with Modbus TCP, or Bacnet communication. Would I be able to support Modbus TCP input packets on 1 PRU and Bacnet output packets on the other PRU of an AM5728. These protocols are currently in use on existing equipment and can not be changed (a HW gateway device can not be put between the AN5728 and the 2 existing systems). If each PRU can support one of the 2 protocols (Modbus TCP, and Bacnet) which of the existing TI example code should I pattern my code after:

Dual EMAC and Switch

Simple Open Real-Time Ethernet Protocol (SORTE)

EtherCAT, Profinet, EtherNet/IP, Profibus etc.

Thanks!!!!

Kenny

↧

Linux/PROCESSOR-SDK-OMAPL138: Unable to load DSP Firmware from Linux using remoteproc.0 because DSP contains partial memory in IRAM

January 10, 2019, 4:34 pm

≫ Next: AM5728: Sitara GPMC usage rules

≪ Previous: AM5728: AM5728 - Have 1 PRU port write out data to a Backnet device, Have the 2nd PRU port read in data from a Modbus TCP device

Part Number:PROCESSOR-SDK-OMAPL138

Tool/software: Linux

Using TI-PROCESSORSDK-RTOS 05.00.00.15 on the DSP and TI-PROCESSOR-SDK-LINUX-05.00.00.15 on the ARM.

Developed a custom firmware for our custom board using the OMAP-L138, it's based on the LCDK board, however it doesn't have several chips for video, etc and I'm using custom codecs.

I am to get my firmware to load straight into the DSP if I just connect and load the default ARM GEL file and then start my code for the DSP.

I am unable to do this using Linux, the resource table file is attached, all the proper IPC modules are loaded.

This happens whenever I specify any memory to be NOT in the DDR either using the #pragma and specifying the sections to be in IRAM (L2 RAM) using the app.cfg file, or using the config.bld file

Build.platformTable["ti.platforms.evmOMAPL138:dsp"] = {
    externalMemoryMap: [
        [ "DDR", {
            name: "DDR", space: "code/data", access: "RWX",
            base: 0xC3100000, len: 0x800000,
            comment: "DSP Program Memory (8 MB)"
        }]
    ],
    codeMemory:  "DDR",
    dataMemory:  "DDR",
    stackMemory: "DDR",
    l1DMode: "32k",
    l1PMode: "32k",
    l2Mode: "0k"
};

if I specify any of the Memory to be not in DDR, the remoteproc from Linux will not be able to load the firmware.

However, my dataMemory has to be in IRAM because of the drivers and their communication with the codec through McASP and EDMA.

So, how do I go about loading memory from the DDR after the initial load into the other respective memories. I tried to search around, but couldn't find anything.

Side note: If I start Linux and get it running I then can use JTAG to load the DSP firmware into the DSP core using my original settings, and it works fine.

↧

AM5728: Sitara GPMC usage rules

January 10, 2019, 4:38 pm

≫ Next: AM5728: Sitara GPMC Idle cycle insertion

≪ Previous: Linux/PROCESSOR-SDK-OMAPL138: Unable to load DSP Firmware from Linux using remoteproc.0 because DSP contains partial memory in IRAM

Part Number:AM5728

Hi,

I know there are some rules need to be followed for proper GPMC operation.
http://processors.wiki.ti.com/index.php/Tips_for_configuring_Sitara_GPMC_registers

In TRM section 15.4.4.9.1, there is another rule.
It says;
“To ensure a correct duty cycle of GPMC_CLK between accesses, RDCYCLETIME and WRCYCLETIME are expressed in GPMC_FCLK cycles and must be multiples of the GPMC_CLK cycle.”

AM5728 TRM:
http://www.ti.com/lit/ug/spruhz6k/spruhz6k.pdf

Does this also need to be followed?

Thanks and regards,
KoT

↧

AM5728: Sitara GPMC Idle cycle insertion

January 10, 2019, 4:54 pm

≫ Next: TMS320C6748: FFT function

≪ Previous: AM5728: Sitara GPMC usage rules

Part Number:AM5728

Hi,

I have a question for idle cycle insertion in Sitara GPMC.
According to TRM table 15-454, there are some configurations which have no idle cycles in between.

AM5728 TRM:
http://www.ti.com/lit/ug/spruhz6k/spruhz6k.pdf

Suppose one of these configurations is used and accesses to the same CS are well pipelined,
the CS is kept activated all the time?
(no CS inactive period observed?)

Thanks and regards,
KoT

↧

TMS320C6748: FFT function

January 10, 2019, 5:10 pm

≫ Next: Linux/AM4384: Command to change TPS65218 FSEAL to enalbe DCDC5/DCDC6 in PDK5.02

≪ Previous: AM5728: Sitara GPMC Idle cycle insertion

Part Number:TMS320C6748

Hi,

I use TI DSPF_sp_fftSPxSP function,why maximum value of the parameter N is 131072?

void DSPF_sp_fftSPxSP (int N, float *ptr_x, float *ptr_w, float *ptr_y, unsigned char *brev, int n_min, int offset, int n_max)

--------------------------------------------------------------------------------

Function Documentation
void DSPF_sp_fftSPxSP ( int N,
float * ptr_x,
float * ptr_w,
float * ptr_y,
unsigned char * brev,
int n_min,
int offset,
int n_max
)

↧

Linux/AM4384: Command to change TPS65218 FSEAL to enalbe DCDC5/DCDC6 in PDK5.02

January 10, 2019, 5:34 pm

≫ Next: Linux/BEAGLEBK: Beaglebone Black peripheral boot USB0

≪ Previous: TMS320C6748: FFT function

Part Number:AM4384

Tool/software: Linux

In the EPOS AM438x EVM use TPS65218, DCDC5/6 supply tamper, Now we want to measure the power consumption of Tamper, but DCDC5/6 is not enabled to supply from CC pin(Coin Cell).

Please tell me the command in UBOOT or Linux to enable FASEAL bit to enable this mode.

My post about the PMIC TPS65218 is here:

Hope do not need to rebuild UBoot and kernel.

Although my question is about AM438x, I saw AM437x IDK also use TPS65218, I thought it is general SDK question, not special for Am438x.

↧

Linux/BEAGLEBK: Beaglebone Black peripheral boot USB0

January 10, 2019, 6:02 pm

≫ Next: AM5728: SPI4 read data error

≪ Previous: Linux/AM4384: Command to change TPS65218 FSEAL to enalbe DCDC5/DCDC6 in PDK5.02

Part Number:BEAGLEBK

Tool/software: Linux

Hi,

I tried to build Beaglebone Black images for booting from USB0 but without success.

The following are what I did:

1.Build the images using Yocto build system from:

Tag "rocko-18.0.3" from git://git.yoctoproject.org/poky

Tag "ti2018.03" from git://git.yoctoproject.org/meta-ti

MACHINE = "am335x-evm"

2. Default u-boot configuration is used (am335x_boneblack_defconfig).

3. Following this instruction http://processors.wiki.ti.com/index.php/Sitara_Linux_Program_the_eMMC_on_Beaglebone_Black#Preparing_the_Flasher_Image, the resulting u-boot.img and u-boot-spl.bin are renamed into u-boot-restore.img and u-boot-spl-restore.bin, respectively. Then placed them under the Uniflash TFTP home folder: C:\AM335x_Flashtool\images.

Below are the messages from the serial console:

U-Boot SPL 2018.01-g8b2f1df4b5 (Jan 10 2019 - 18:54:12)

Trying to boot from USB eth

** Bad device mmc 0 **

Using default environment

<ethaddr> not set. Validating first E-fuse MAC

cpsw, usb_ether

Error: usb_ether address not set.

using musb-hdrc, OUT ep1out IN ep1in STATUS ep2in

MAC c8:df:84:da:88:91

HOST MAC de:ad:be:ef:00:00

RNDIS ready

musb-hdrc: peripheral reset irq lost!

cpsw Waiting for PHY auto negotiation to complete......... TIMEOUT !

Problem booting with BOOTP

SPL: failed to boot from all boot devices

### ERROR ### Please RESET the board ###

Below are the questions I have:

1. What did I miss? I suspect instead of using default u-boot configuration, some of the u-boot configuration needs be modified for peripheral boot from USB0 to work. If yes, what need to be modified? If not, can someone point out what may cause these errors?

2. I noticed that Uniflash does not work on a computer that has an additional network card. I need to use a computer that has only one Ethernet port for it to work. I wonder if this is a known problem? Using Wireshark, I saw BOOTP requests sending out from the Beaglebone on the Linux USB Ethernet/RNDIS Gadget interface. However, the replies from the host PC are not seen. Is there a way to make it work?

Please help. Thank you!

↧

AM5728: SPI4 read data error

January 10, 2019, 6:11 pm

≫ Next: AM5728: Generating GPMC burst access from ARM core

≪ Previous: Linux/BEAGLEBK: Beaglebone Black peripheral boot USB0

Part Number:AM5728

Customer has problem on using AM5726 SPI4. the returned read data is always wrong, the wave shows data is right from external sensor, but wrong in SPI module.

Send out read ID command, sensor returned the right ID value 0x12, but in SPI_rx register it is 0x0186.

Customer has verified I2C, CAN, UART, UART, SPI1, SPI2, QSPI on the AM5726 board made by themselves. but SPI4 has the issue.

As either GPEVM nor IDK did not connect device on SPI4, I don't know how to verify. Please help.

↧

AM5728: Generating GPMC burst access from ARM core

January 10, 2019, 6:19 pm

≫ Next: TMS320C6670: RSA, RAC and TAC User's Guide and example code on TMS320C6670

≪ Previous: AM5728: SPI4 read data error

Part Number:AM5728

Hi,

Is there any way to generate GPMC burst access from ARM core?
Suppose GPMC is properly configured to handle burst accesses.

I knew;
- using DMA burst accesses
- ARM cache line fill
- LDM/STM instructions
are handled as burst at GPMC.

Are there any other ARM instruction(s) which can generate burst on GPMC?

Thanks and regards,
KoT

↧

TMS320C6670: RSA, RAC and TAC User's Guide and example code on TMS320C6670

January 10, 2019, 6:38 pm

≫ Next: RTOS/TMS320C6678: How to build Linear Algebra Library DSP example under Windows

≪ Previous: AM5728: Generating GPMC burst access from ARM core

Part Number:TMS320C6670

Hi,

I am using C6670 for implement reed-muller codes and I could not find document and example code about RSA, RAC and TAC - its big truble for me.

plz help me

Please send me to lgm0925@dsplab.hanyang.ac.kr

↧

RTOS/TMS320C6678: How to build Linear Algebra Library DSP example under Windows

January 10, 2019, 7:19 pm

≫ Next: TMDSDSK6713: Will Floating point DSK6713programs run in 6544 Fixed point DSP ?

≪ Previous: TMS320C6670: RSA, RAC and TAC User's Guide and example code on TMS320C6670

Part Number:TMS320C6678

Tool/software: TI-RTOS

Hi Champs,

According to the steps in http://processors.wiki.ti.com/index.php/Processor_SDK_Linear_Algebra_Library#Build_and_Run_Examples,

DSP-only example in dsp folder: running on DSP through CCS and JTAG. Examples include:

Matrix multiplication (dgemm)
To run these examples, follow the steps listed below:
- Set environment variable LINALG_DIR to <LINALG_installation_folder>, in addition to the environment variables listed in Rebuilding LINALG.
- Go to folder <LINALG_installation_folder>/examples/dsponly
- Type "make TARGET=<device_name>" to build the examples
- Load the executable (.out file) to CCS and run

The step Type "make TARGET=<device_name>" to build the examples is excuted under linux, how to build the DSP-only example under windows?

Thanks.
Rgds
Shine

↧

TMDSDSK6713: Will Floating point DSK6713programs run in 6544 Fixed point DSP ?

January 10, 2019, 7:59 pm

≫ Next: AM5728: PRU ethernet

≪ Previous: RTOS/TMS320C6678: How to build Linear Algebra Library DSP example under Windows

Part Number:TMDSDSK6713

Hi. I am Ammar from NUST, Pakistan. I am currently working in DSK6713 floating point DSP. But now i need to shift to a high speed kit DSK6455 which is fixed point. Will my programs run in that kit? Because i need high speed DSK. If not, kindly refer me to other Floating point high speed DSP kit, not a processor

Regards

↧

AM5728: PRU ethernet

January 10, 2019, 9:38 pm

≫ Next: RTOS/TDA2PXEVM: h264 ivahd encoder crashes on tda2px evm

≪ Previous: TMDSDSK6713: Will Floating point DSK6713programs run in 6544 Fixed point DSP ?

Part Number:AM5728

Hi,

we are using am578 SDK ti-processor-sdk-linux-am57xx-evm-03.02.00.05-Linux-x86-Install.bin

In our custom board we connected PRUSS2 ethernet to micrel ethernet driver IC. we connected rx,tx data lines clocks and ctl signals

I changed the DTS file as below:

/* Dual-MAC Ethernet application node on PRU-ICSS2 */
   pruss2_eth {
       compatible = "ti,am57-prueth";
       pruss = <&pruss2>;
       sram = <&ocmcram1>;
       interrupt-parent = <&pruss2_intc>;

       pruss2_emac0: ethernet-mii0 {
           phy-handle = <&pruss2_eth0_phy>;
           phy-mode = "mii";
           /*interrupts = <20>, <22>;
           interrupt-names = "rx", "tx";
           /* Filled in by bootloader */
           local-mac-address = [00 00 00 00 00 00];
       };

       pruss2_emac1: ethernet-mii1 {
           phy-handle = <&pruss2_eth1_phy>;
           phy-mode = "mii";
           /*interrupts = <21>, <23>;
           interrupt-names = "rx", "tx";
           /* Filled in by bootloader */
           local-mac-address = [00 00 00 00 00 00];
       };
   };

&pruss2 {
   status = "okay";
   pinctrl-names = "default";
   pinctrl-0 = <&pru2_mii_pins_default>;
   /*dual_emac;*/

   pru2_0: pru0@4b2b4000 {
       mboxes=<&mailbox4 &mbox_pru2_0>;
       status = "okay";
   };

   pru2_1: pru1@4b2b8000 {
       mboxes=<&mailbox4 &mbox_pru2_1>;
       status = "okay";
   };

&pruss2_mdio {
   status = "okay";

   pinctrl-names = "default";
   pinctrl-0 = <&pru2_mii_davinci_mdio_pins_default>;

   pruss2_eth0_phy: ethernet-phy@5 {
       reg = <5>;
   dual_emac_res_vlan = <3>;

   };

   pruss2_eth1_phy: ethernet-phy@6 {
       reg = <6>;

   dual_emac_res_vlan = <4>;
   };
};

we used physical address of 5 and 6

we included ti,am572x-beagle-x15 in pru_rproc.c

the log is as follows regarding pru ethernet

[ 2.318576] prueth pruss2_eth: unable to get pruss handle

[    2.532482] ti-pruss 4b280000.pruss: creating PRU cores and other child platform devices
[    2.586488] davinci_mdio 4b2b2400.mdio: davinci mdio revision 1.6
[    2.586491] libphy: 4b2b2400.mdio: probed
[    2.591095] davinci_mdio 4b2b2400.mdio: phy[5]: device 4b2b2400.mdio:05, driver Micrel KSZ9031 Gigabit PHY
[    2.591101] davinci_mdio 4b2b2400.mdio: phy[6]: device 4b2b2400.mdio:06, driver Micrel KSZ9031 Gigabit PHY
[    2.592507] remoteproc4: 4b2b4000.pru0 is available
[    2.592512] remoteproc4: Note: remoteproc is still under development and considered experimental.
[    2.592515] remoteproc4: THE BINARY FORMAT IS NOT YET FINALIZED, and backward compatibility isn't yet guaranteed.
[    2.592583] pru-rproc 4b2b4000.pru0: PRU rproc node /ocp/pruss@4b280000/pru0@4b2b4000 probed successfully
[    2.592756] remoteproc5: 4b2b8000.pru1 is available
[    2.592759] remoteproc5: Note: remoteproc is still under development and considered experimental.
[    2.592763] remoteproc5: THE BINARY FORMAT IS NOT YET FINALIZED, and backward compatibility isn't yet guaranteed.
[    2.592831] pru-rproc 4b2b8000.pru1: PRU rproc node /ocp/pruss@4b280000/pru1@4b2b8000 probed successfully
[    2.593758] remoteproc4: Direct firmware load for ti-pruss/am57xx-pru0-prueth-fw.elf failed with error -2
[    2.593765] remoteproc4: failed to load ti-pruss/am57xx-pru0-prueth-fw.elf
[    2.593815] remoteproc5: Direct firmware load for ti-pruss/am57xx-pru1-prueth-fw.elf failed with error -2
[    2.593821] remoteproc5: failed to load ti-pruss/am57xx-pru1-prueth-fw.elf

what else to be changed

Regards,

Kumari

↧