Can't capture Vps_printf messages to file

Hello,

Are you using the latest version of the EZSDK 5.05.01.04? Could you please provide your boot arguments and more specific the value of notifyk.vpssm3_sva ?

Regards,

Pavel

Hi Pavel, yes I'm using EZSDK 5.05.01.04.

notifyk.vpssm3=0xbf900000

My relevant other boot arguments are:

mem=364M mem=324M@0x9F900000 vmalloc=500M earlyprintk

I am loading the VPSS module using:

modprobe vpss sbufaddr=0xBFB00000 mode=hdmi:1080p-60,dvo2:1080p-60,hdcomp:1080p-60 i2c_mode=0

Thanks,
Ralph

Can you try with:

mem=364M@0x80000000 mem=320M@0x9FC00000 vmalloc=500M  notifyk.vpssm3_sva=0xBF900000

Regards,

Pavel

Hi Pavel, you were correct as I wasn't using the correct second boot argument. I have corrected this to use 320M however it still produces the same results as before.

Ralph

Try with:

modprobe vpss sbufaddr=0xBFB00000 mode=hdmi:1080p-60,dvo2:1080p-60,hdcomp:1080p-60 i2c_mode=1

The entire flow should be:

echo "Loading HDVICP2 Firmware"
prcm_config_app s
modprobe syslink
until [[ -e /dev/syslinkipc_ProcMgr && -e /dev/syslinkipc_ClientNotifyMgr ]]
do                                                
      sleep 0.5
done
firmware_loader $HDVICP2_ID /usr/share/ti/ti-media-controller-utils/dm816x_hdvicp.xem3 start
echo "Loading HDVPSS Firmware"
firmware_loader $HDVPSS_ID /usr/share/ti/ti-media-controller-utils/dm816x_hdvpss.xem3 start
modprobe vpss sbufaddr=0xBFB00000 mode=hdmi:1080p-60,dvo2:1080p-60,hdcomp:1080p-60 i2c_mode=1
modprobe ti81xxfb vram=0:24M,1:16M,2:6M
fbset -depth 32 -rgba 8/16,8/8,8/0,0/0
modprobe ti81xxhdmi

Regards,

Pavel

Hi Pavel,

I cannot use the argument change you suggested as my ARM must master the I2C bus, not the media controller, due to the way we have our I2C busses set up.

Regarding the flow, mine is identical to yours except for these two lines:

firmware_loader $HDVPSS_ID /tmp/ti-media-controller-utils/dm816x_hdvpss_v4l2.xem3 start -i2c 0
modprobe vpss sbufaddr=0xBFB00000 mode=hdmi:1080p-60,dvo2:1080p-60,hdcomp:1080p-60 i2c_mode=0

I don't see myself how changing the I2C arguments will affect the debug statements anyway...

Ralph

Hi Ralph,

Here we have some threads related to Vps_printf, and how to enable it:

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/717/t/185162.aspx

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/717/t/178312.aspx

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/717/p/232736/818376.aspx#818376

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/717/t/228678.aspx

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/717/p/222543/785599.aspx#785599

Regards,

Pavel

Hi Pavel,

Thanks for the links however I've already read most of those and they don't really give anything new. I've tried attaching a JTAG debugger to the ARM to read the memory using CCS. If I connect to the ARM while capture_encode is running and try to read 0xBF900000 using the memory browser it just comes up with "????????????" and says "Target failed to read memory".

Any more ideas?

Thanks,

Ralph

Here something similar is discussed:

http://e2e.ti.com/support/development_tools/code_composer_studio/f/81/t/195083.aspx

For the last question, you can ask in our CCStudio forum:

http://e2e.ti.com/support/development_tools/code_composer_studio/default.aspx

BR,

Pavel

Hi Pavel,

I tried the recommendation in the first thread but it doesn't change anything and in my experience I've not had to do anything like that before to read physical memory.

I'll post on the CCS forums and see if that gets me anywhere.

Thanks,
Ralph

Hi Ralph,

Meanwhile, you can try if this user-space executable will work for you:http://e2e.ti.com/cfs-file.ashx/__key/communityserver-discussions-components-files/717/4137.readmem

#include <stdio.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <stdint.h>

#define MEM_SIZE 4096
#define MEM_MASK (MEM_SIZE - 1)
#define MEMORY "/dev/mem"

void memaccess(unsigned int reg_addr, unsigned int *reg_val, int access_width)
{
	int fd;
	void *mem_map, *mem_addr;

	fd = open(MEMORY, O_RDWR|O_SYNC);
	if(fd < 0) {
		perror(MEMORY);
		exit(1);
	}

	mem_map = mmap(0, MEM_SIZE, PROT_READ|PROT_WRITE, MAP_SHARED, fd, reg_addr & ~MEM_MASK);

	if(mem_map == (void*) - 1) {
		perror("mmap()");
		exit(1);
	}

	mem_addr = mem_map + (reg_addr & MEM_MASK);

	if (reg_val)
		switch(access_width) {
		case 8:
			*(uint8_t*)mem_addr = *reg_val;
			break;
		case 16:
			*(uint16_t*)mem_addr = *reg_val;
			break;
		case 32:
			*(uint32_t*)mem_addr = *reg_val;
			break;
		default:
			fprintf(stderr, "Invalid access width!\n");
		}
	else
		switch(access_width) {
		case 8:
			printf("0x%02X\n", *(uint8_t*)mem_addr);
			break;
		case 16:
			printf("0x%04X\n", *(uint16_t*)mem_addr);
			break;
		case 32:
			printf("0x%08X\n", *(uint32_t*)mem_addr);
			break;
		default:
			fprintf(stderr, "Invalid access width!\n");
		}
}

static void usage(const char *prog)
{
	fprintf(stderr, "\nUsage:\t%s [-8|-16|-32] <reg_addr> [value]\n\n\n", prog);
}

int main(int argc, char *argv[])
{

	unsigned int reg_addr, reg_val, *reg_val_ptr;
	char **p;
	int n, access_width = 32;

	if (argc  < 2) {
		usage(argv[0]);
		exit(1);
	}

	p = argv+1;
	n = argc-1;

	if ((*p)[0] == '-') {
		if (!strcmp(*p, "-8"))
			access_width = 8;
		else if (!strcmp(*p, "-16"))
			access_width = 16;
		else if (!strcmp(*p, "-32"))
			access_width = 32;
		else {
			usage(argv[0]);
			exit(1);
		}
		n--;
		p++;
	}

	if (n < 1) {
		usage(argv[0]);
		exit(1);
	}

	reg_addr = strtoul(*p, NULL, 0);
	p++;
	n--;

	if (n) {
		reg_val = strtoul(*p, NULL, 0);
		reg_val_ptr = &reg_val;
	}
	else
		reg_val_ptr = NULL;

	memaccess(reg_addr, reg_val_ptr, access_width);
	return 0;
}

I can read/write physical memory without CCS with readmem:

root@dm816x-evm:~# ./readmem 0xBF900000
0xA9C8B7D6
root@dm816x-evm:~#

Regards,

Pavel

Hi Ralph,

      I am having the same problem, which address did you use to solve it??

Thanks in advance,

-David

Search for the definition of Vps_printf in your source files and switch over the "other" one by commenting out the current definition.

Hi Ralph, 

       Thanks for your help, I found where Vps_printf is defined and the other similar function that I found was Vps_rprintf, the only difference between them is a System_printf(buf); call at the end of the Vps_printf which is not present in Vps_rprintf.
       Anyway, I tried commenting that part of the code and creating a new firmware for both scenarios with the line and without the line but I am still getting the same error with that address. My goal is to be able to see the debug messages of the hdvpss driver in order to try to find a bug. I am using EZSDK 5.05.02.00 and I tried with addresses:

1)
0xBF900000 -> Produce syslink error that you got

2) 
 0xBFB00000 
 0xBFD00000 
 0xBFF00000

Produce 

[c6xdsp ] Remote Debug Shared Memory @ 0x9e400000
[m3video] Remote Debug Shared Memory @ 0x9e410820
[m3vpss ] Remote Debug Shared Memory @ 0x9e421040

but no messages are shown from the VPSS.

3) I also tried reading the .map file of the vpss firwmare (http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/717/p/247984/872309.aspx) which says:

                  bfef0000    0000f060                         : remote_debug_server.oem3 (.bss:extMemNonCache:remoteDebugCoreShm)

using the  bfef0000 address I got the same output like 2) no messages coming from the VPSS.

4) Furthermore I have tried running  remote_debug_client.out before to mount the syslink module but the same result is obtained, no messages. 

I am not using CCS, were you able to get messages without using CCS?

Any suggestion is really appreciated,

-David 

Hi,

      Any idea about why it doesn't work in the EZSDK?

Thanks in advance,

-David

Probably because they had the wrong definition uncommented when they built the media controller binaries.

Hi Ralph,

     I have access to the code to build the media controller binaries but I don't see that function that you uncommented, just those two that I mentioned before, can you paste here the definition that you are using?

Thanks

Sorry, I can't easily find out where the issue is as I've modified so many files and it's hard to find the exact definition. All I can suggest is search all the header files for "Vps_printf" and try commenting/uncommenting different definitions. Also, look out for any #defines that may need to be set at the top of source files.

Ralph