I am glad to see you found it !!
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
Also, if you want to use xds560r to debug dsp, you may need to 'wakeup' the
dsp using the arm.
Regards
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Connecting the emulator is a bit triky and needs you to do several things in
a certain order to work properly.
I don't remember that order, but it was something like: first swich on the
board, then connect the emulator, then open CCS, then ...
If you do that things in a different order then
We don't mind if the output is just sd, but our input would be 720p
Do you know if is it possible to provide a 720p input to TMS320DM355's
H.264 decoder (although its output will be sp) ?
Thank you,
Carlos
___
Davinci-linux-open-source mailing list
Digital output on DC5 is NOT implemented. You have to write the proper
settings to Davinci registers by yourself
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Digital output on DC5 is NOT implemented. You have to write the proper
settings to Davinci registers by yourself. Also note that those values
will depend on which screen you are attaching to DC5
Carlos
___
Davinci-linux-open-source mailing list
Maybe you don't have the undocumented DCLKCTL and DCLKPTN0 registers
set properly. They are crucial for generating correct clock to the
display. Try 0x800 for DCLKCTL and 0x1 for DCLKPTN0.
Cheers
___
Davinci-linux-open-source mailing list
Thank you, Brad.
We are wondering if it is possible (using a DM355) to decode a MPEG-4
from a DVBS signal. It does not matter if the output resolution is
about 780p, but it should decode up to 1080i.
Many thanks,
Carlos
___
Davinci-linux-open-source
Thanks, Brad.
There is something I don't understand yet. So, to decode MPEG4 we can choose :
1.- TMS320DM644x + MPEG4 codec (front fee $20,000 + $.67 per unit @ 10 KU)
or
2.- TMS320DM355 + free MPEG4 codec ($0)
How is that?
Thanks,
Carlos
___
Where could I find TI's HD MPEG-4 SP features and price?
Many thanks,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
VICP provides performance boost on Video Encode applications.
TI gives access to the VICP only to their Authorized Software
Providers (ASPs) or via their multimedia codecs.
Carlos
___
Davinci-linux-open-source mailing list
I am getting the same result. I am able to download other resources
(for example 090s_o_win50ce_sdk_dm644x_3_1.zip) but I can't download
dvflasher_1_11.tar
Regards,
Carlos
___
Davinci-linux-open-source mailing list
I can't download the newest version of the UBL source.
Anyone is getting the same error as me while attempting to download? :
The page cannot be found
The page you are looking for might have been removed, had its name changed,
or is temporarily unavailable.
Please try the following:
If
Try to use something like
'video=dm64xxfb:output=pal:format=component'
on your bootargs
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Look for davinci_psc_all_enable function at board/davinci/davinci.c
Cheers
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
The actual pdf (sprue38a.pdf, june 2006) does not contain any
documentation about preview registers, but the old one does
(sprue38.pdf, december 2005)
Regards,
Carlos
___
Davinci-linux-open-source mailing list
The .x64p file is the .out file
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
1. Does anyone have sources for the first bootloader in NAND (the max 14
kB loader)? We only have a .bin file.
I am also interested in that TI's max 14kb first bootloader (UBL)
2. Does anyone know the proper U-boot configuration for NAND flash
booting?
Undefine CFG_ENV_IS_IN_FLASH at
tvp5146 on DVEVM has also 0x5D address
(Although on DaVinci documentation you would find 0x56 as tvp5146
address, 0x5D is actually the right one)
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Um, maybe POST is not done yet for DaVinci u-boot ...
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
Hello:
Is anybody out there developing CODECs for DaVinci processors?
It seems that VICP coproccessor is somewhat restricted and used only
by Texas CODECs (avaliable at a very high price). Did someone try to
develop a CODEC even without using VICP coprocessor?
Is it possible to get information
Did you try to use fbset command?
in /etc/fb.modes you may have several resolutions you can choose by
using the fbset command like :
fbset -fb /dev/fb/3 640x480-60
(the example is for the video window)
Regards,
Carlos
___
Davinci-linux-open-source
What do you mean?
Do you want to cross-compile an application for davinci on your computer?
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
I erase all the data in NOR flash when I update the u-boot.
The Linux kernel in the NOR flash is erased, too.
How can I re-flash the Linux kernel into NOR flash ?
Sorry that was to write the kernel to NAND flash !
To write your kernel to NOR I think you could load it via tftp
Does .bss belong to BIOS CODE section?
None of BIOS CODE sections can be stored in DDR2
(just a thought)
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Sorry I think I was wrong, BIOS CODE sections can be in DDR2.
The sections that can't be in DDR2 are BIOS DATA ones, that includes .stack
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
I don't know if it is related to your problem, but I think you have 2
bytes per pixel (try to adjust OSDWIN0XL, etc to (num_pixels*2))
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Do you compile without errors when you uncheck your USB options?
(just to dismiss things like environment variables, links, and so on)
Thanks,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
When i'm trying to compile the kernel with the new configuration, it's showing
so many errors.
I didn't have errors. Please, post the errors you have.
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Where could we find that TI BIOS LINK ?
Thanks,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
I would ask for the procedure to enable UART1
I add port-line number at serial.c
Do you mean to change this line at ./arch/mips/vr41xx/common/serial.c?:
int vr41xx_serial_ports = 0;
In my kernel tree I have these 'serial.c':
$ find -name serial.c
./arch/mips/vr4181/common/serial.c
Though we don't have a driver available, VLYNQ may be an option worth
looking at.
Thank you Thom.
Do you know where could I find more information about VLYNQ protocol
in order to implement it on fpga side? (sprue36 is not very thorough)
Many thanks,
Carlos
Any ideas from on how to debug this and to confirm UART on board is
good/bad are appreciated.
To see if UART on board is bad, you could configure S3 to boot from
UART. When you power on your DVEVM you must see the messages BOOTME
BOOTME
Regards,
Carlos
Hello. Many thanks for your replies!
We want to transmit about 4kbytes periodically from fpga to davinci
(unidireccional data operation) using 16 data lines if possible, at a
maximum data rate of 20 Mbytes/sec (peak)
Data is buffered on fpga side and we also need a 'control path'
(through the
Then tried l (load into memory). Then there was no response.
As if the console was hung.
I think I now need to send u-boot.bin. how should I send this ? As a ASCII
TEXT ?
No, you should send 'l' + Load address + data length + start address + u-boot
(u-boot in binary form)
Note that you should
Hello
We want to get data from a fpga and I we are looking for the best
'port' to attach our fpga. EMIF, ATA interface... what do you think is
the best way ?
Thanks!
Carlos
___
Davinci-linux-open-source mailing list
Using the below I can Read/Write ARM/DSP DDR memory using
Physical addresses, load and execute programs in the DSP
etc. (maps entire 256MB)...
int fd;
void *map;
if ((fd=open(/dev/mem,O_RDWR|O_SYNC))==-1) exit(1);
map=mmap((void
The best way I think is to use vpfe by using CCD raw data mode:)
I am sorry, I forgot to say that we need vpfe to capture video...
Thanks,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Accorging to DDR2 Memory Controller User's Guide (SPRUE22B), the
peripheral bus burst priority register (PBBPR) helps prevent command
starvation within the DDR2 memory controller. To avoid command
starvation, the DDR2 memory controller momentarily raises the priority
of the oldest command in the
The OSD appears shifted for just a frame time and then goes back. It seems
to occur most often when I am moving data with memcpy.
That is exactly what I see if I do a memcpy on the dsp (if I don't do
the change to PBBPR register)
I am at a loss as to what to do next!
Try to change the
Hello !
Thank you very much for your reply.
May be some bandwidth issue or bus sharing
Is that a big single memcpy() ?
If yes, can you try looping around smaller shots ?
I changed my dsp's program to be only the following lines:
while (1) {
for (m=0; mlongFFT; m++) {
(I am using rgb888 mode)
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
Could this problem be becouse of Advisory 1.1.6 on sprz241d.pdf
(silicon errata) that stands: DDR2: Multiple Master Access to the DDR2
at the Same Time may Cause Master to Stop. Revision(s) Affected: 1.1
How could I know my silicon revision?
How could I know DSP, ARM, VCLK and MCLK clock rates?
The effect appears even if I don't execute any program on the arm.
(it happens only on the LCD attached to DC5 and specially when the DSP
do some memcpy)
The 'shift' on the OSD is not always the whole screen...
Many thanks,
Carlos
___
Btw, do you know why there is no SPI driver in MV by default? As I know
there is some complex SPI infrastructure which is not ready yet... Do you
know where can find more information on this (assuming I'd like to share my
driver with the rest of community [but I need to talk to my boss first])?
Did you set J7 to host position?
How many devices are you connecting to your USB?
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
I finally tried to setup SPI according to data sheet (SPRUE32) but I think
data sheet is wrong. It states (see below) that configuration should be done
while in reset mode (SPIGCR0: RESET bit). But it doesn't work.
I agree and I also saw that behaviour. I write a 0x0001 to SPIGCR0
(spi out
Did you write your SPI driver yourself?
Maybe SPIFMT0 or SPIDELAY are not properly configured.
Where are you configuring SPI registers?
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
SPIDELAY |= (0xF 24); // maximum setup time
SPIDELAY |= (0xF 16); // maximum hold time
I am using 0 for these two.
3) Do you know maybe TI supplies examples for SPI controller which shows how
to communicate with SPI slave (SPI EEPROM or smth like this)? I need to hold
CS active during data
Rudy did. Look for his 27-Sep post
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
We got it !! It was necessary to transmit the two fields and their
corresponding blanking periods.
Thank you and regards!
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
What mode have you configured the VPFE in? This behavior that
you are observing sounds like the VPFE is in BT.656 mode where only the
clock and data lines are used and the VSYNC and HSYNC are encoded in the
data stream.
That's right. I tried to change the registers to use raw mode
Some customers expect to 'see something' just after power on. Becouse
of that I am modifying davinci's u-boot-594 to get a splash screen. I
archieved in making u-boot showing the 'color bars' on my screen (LCD)
Now, I want to see an image, so I changed u-boot to configure all VPBE
registers
I saw in sprue37, page 39, figure 22 the signal: venc_sclk_osd
What register should I use to change venc_sclk_osd?
Could it be my problem? ('Setting Up Clocks for DM420 OSD' was not the problem)
Thanks and regards,
Carlos
___
Davinci-linux-open-source
Hello
I want to change the logo that appears when loading montavista linux.
To make that, I should modify drivers/video/logo/logo_linux_clut224.c
and compile the kernel, but I don't know how to generate
logo_linux_clut224.c from an image
Anybody knows?
Thanks,
Carlos
Can anyone advise on how to make the kernel boot log be more verbose. Im
getting the system hanging just before the usual time when the console
login appears.
Just a short in the dark: Did you change /etc permissions?
Carlos
___
So: Should I assume VLYNQ is 'propietary' and there is no
specification avaliable out there?
Thanks and regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
The VLYNQ peripheral has consumes 2 memory regions within the
DM6446 memory map, the control registers at base address 0x01e0 and
the memory window to access the remote devices' memory map at base
address 0x0c00.
That was it! Thank you!
I have started with a minimal bin,dev,etc,lib type structure, but the
Kernel Panics!
I think at least sbin directory is also necessary
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
Does somebody know if davinci's uboot is capable of displaying a splash screen?
I modified include/configs/davinci.h to add 'bmp' command (adding
CFG_CMD_BMP to the list of CONFIG_COMMANDS) but I get an error when
uboot compiles:
make[1]: *** [.depend] Error 1
make[1]: Leaving directory
I thought those boot splash (i.e. the montavista logo) were hardcoded in arrays in one of the header files?
Do you know in wich file ismontavista's logo hardcoded? (I will need to modify it in the future)
For now I want also uboot's splash screen becouse it would show very shortly after power
Hello!
VLYNQ REVID register does not appear on sprue14.pdf. I think it could
be 0x0C00 but according to sprue14.pdf 0x0C00 address is
'Reserved'. At what address VLYNQ REVID is located?
By the way I tried to monitor VLYNQ activity on DVEVM's R17
(VLYNQ_CLK) but I could only see the
In 'USB HID devices' thread (24 August) David Brownell said:
The basic issue is that the 2.6.10 driver version only supports two periodic
endpoints. The hub takes one, the mouse takes the second ... nothing left
for a keyboard.
I don't know if that issue is fixed in the last kernel.
Regards,
Finally I archieved to configure SPI registers.
It was neccesary to 'power up' SPI module with this sequence:
1.- Wait until PTSTAT is 0
2.- Set enable state in MDCTL22 (SPI)
3.- Set 'GO' bit in PTCMD
4.- Wait until PTSTAT is 0
Now I think my registers are properly configured, but I still can't
I am seeing a weird behaviour when I write to SPI registers.
They seem to hold their value only for a short period of time, and
then they go to 0x00.
With this code:
#define SPIGCR0 0x01C66800
*(int *__iomem) IO_ADDRESS(SPIGCR0) |= 0x0001;
printk (SPIGCR0 %X\n, *(int
Hello!
If you are using the ubl_nand.bin that we provide (or maybe got from
spectrum digital) then you will need to set your start address to
0x100(or 0x120 to be more exact).
Your memory dumps look good though.
Keep me advised.
There may be a CRC problem as Hiroshi said :
So the ubl
Hello!
Daniel will let us know the correct behavior of RBL, but I think you need to
modify u-boot in order to write ubl using u-boot's nand command.
RBL (and ubl_nand) requires CRC generated by DaVinci's hardware,
but its algorithm is different from the one implemented in u-boot.
So the ubl
2, Added follows to Makefile:
ARCH = $ARCH
CROSS_COMPILE = $CROSS_COMPILE
export ARCH CROSS_COMPILE
I can compile the kernel and modules without any change to the Makefile...
Why did you add that lines?
Regards,
Carlos
Um, I wrote u-boot-567-nand.bin (and the header) as I did with ubl_nand.bin.
The header starting at 0x18000 and u-boot-567-nand.bin starting at
0x18200, but I get :
Booting PSP Boot Loader
Starting NAND Copy
NAND Boot failed. Starting UART
BOOTPSP BOOTPSP
Could I have an error on the header?
Can the User Boot Loader (UBL) be used to upload UBOOT, thus allowing
the user to restore UBOOT to flash?
Yes, load Rudy's UBL via uart. Then using Rudy's UBL load uboot.
Once you have uboot, you could write the uboot you have in memory to
nor flash using 'cp.b' command.
Regards,
Carlos
I have written the equivalent in Linux and it appears to download
correctly, I get the 'DONE' Suggesting DaVinci has used downloaded the
the UBL but then it just starts again with the 'BOOTME'.
What are you sending as your header?
I send ACK 0BB80100
Are you sending your
Reading the ubl descriptor and the fist bytes of ubl_nand.bin stored
in NAND I think see everthing ok... I am puzzled :
DaVinci EVM # nand read 0x8070 0x4000 0x4000
NAND read: device 0 offset 16384, size 16384 ... 16384 bytes read: OK
DaVinci EVM # md 0x8070
8070: 00edaca1 2000
Emm sorry, I meant:
Since you have a bad file name, try to set the 'bootfile' environment
variable before 'dhcp' command:
DaVinci EVM # setenv bootfile uImage
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
TFTP from server 192.168.100.70; our IP address is 192.168.100.64
Filename '#65533;'.
Load address: 0x8070
Loading: *
TFTP error: 'File not found' (1)
Starting again
As you has a bad file name, try to set the bootfile environment
variable before 'dhcp' command:
DaVinci EVM # setenv
The version of u-boot that I'm using has the commands:
nand erase offset length
nand write address offset length
That was it !
'nand erase' and 'nand write' instead of 'erase' and 'cp.b'
Thank you so much!
Regards,
Carlos
___
Davinci-linux-open-source
Hello,
Thank you all for your kind replies. I wish to ask if I am doing
something wrong, since my DVEVM doesn't boot from nand yet:
1.- To use the NAND to boot, I put S3 positions 1 and 0 to 00. I also
put J4 in NAND.
2.- I made a UBL descriptor in a hex editor, sent it to uboot using
tftp and
Hello! I loaded u-boot-567-nand.bin using Rudy's UBL and executed.
Now I have some doubts about writing ubl_nand.bin and
u-boot-567-nand.bin into nand.
I loaded ubl_nand.bin through tftp to memory:
DaVinci EVM # tftp
TFTP from server 192.168.1.34; our IP address is 192.168.1.33
Filename
Hello list:
When you have S3 in 10 position and J4 in NAND position, it
isn't harmful for NOR to write to addresses - 0001 since
that addresses correspond now to NAND flash and NOT to NOR flash. Is
this correct?
(I don't want to overwrite my NOR uboot)
Thank you,
Carlos
Thank you very much! And is there any more froum talking about te
davinci-linux?
Andy Ye created one QQ Davinci discussing group, Group number is 30456209.
Regards,
Carlos
___
Davinci-linux-open-source mailing list
Thanks!.. Now my uart boot works..
I am trying it and I get the last DONE message from Davinci, but thats
all, I don't get the Rboot- (nor anything).
Should I send UBL binary file least significant byte first?
Thanks,
Carlos
___
For send UBL binary file least significant byte first I mean, If a
word is 77073096... should I send 96 30 07 77 (and in ASCII HEX)?
or 77 07 30 96 (ASCII HEX)?
Thanks and regards,
Carlos
___
Davinci-linux-open-source mailing list
Vsync are toggling correctly but I can't get VCLK to toggle.
I think your VCLK problem could happen if you are using RGB666 mode.
RGB666 mode is not working. Use instead a LCD888 mode and don't
connect 2 LSB singals for each color component.
Maybe the DCLK is not setup (No documenation) or
Hello Rudy,
Thank you so much for this invaluable contribution!
Carlos
___
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
Hello:
I tried to download DVEVM 1.10 and 1.00 cd's from ti but... if you try to download v1.00 cd's you will download v1.10 !!
For example, if you try to downloadmvl_setuplinux_1_00_00_32.bin, you will get mvl_setuplinux_1_10_00_30.bin instead !!
Regards,
Carlos
I am using DirectFB-0.9.22 to draw some graphics on davinci's /dev/fb/0.
Everything is fine when I use the analog output to see the screen on
the Color TFT that comes with DVEVM.
Recently I attached a LCD and I see this effect on the LCD:
If I draw my graphics in WHITE (or using a color close to
There may be others that may better satisfy your particular use case -
others on the list, feel free to jump in. Does your script processing
thread need to communicate with a particular codec
Probably not. But the DSP thread should be a non-blocking task, means it's
initially started and
Anybody knows where can I find information about these 'undocummented'
registers? :
0x01C4 0044VPSS_CLKCTL (VPSS clock control)
0x01C4 0D18(PLL2CTL ??)
0x01C7 2614??
0x01C7 2464DCLKCTL
0x01C7 2468DCLKPTN0
0x01C7 246C ??DCLKPTN1
0x01C7 2470 ??
If you need to see the transparency, you need to choose the value of the
color in RGB for which you need the transparency effect (which in turn
takes the blending ratio from the register (OSDWIN0MD.BLND0)( make sure
it is any other value than 0 or 7 to achieve blending. For a value of 0
you
BOOTP broadcast 1
DHCP clieant broundto address 192.168.3.101
TFTP from server 192.168.3.1:our IP address is 192.168.3.101
Filename 'tftpboot/uImage'
Load address:0x8070
Loading...
TFTP error: File not found (1)
Starting again
If you are exporting 'tftpboot' directory, use
setenv bootfile
Is it possible to load an updated kernel like 2.6.18?
are they supported by TI Davinci or rather by Montavista?
You could find the last kernel in http://source.mvista.com/git/
I didn't try that kernel. Comments from people on this list about this
kernel are welcome !
In
while trying to manipulate the serial port - i got some errors
root@(none):/mnt/fileSys/appl# stty -F /dev/ttyS0 115200 cs8 -parenb
-cstopb
stty: /dev/ttyS0: Inappropriate ioctl for device
does anyone have any idea why that may happen ...
Don't know, but: Could it be becouse of ttyS0 is in
I have downloaded the second stage bootloader source
code(uboot_bl2_MontaVista_tar.gz) and made it for both Nor and Nand,and boot
Linux successfuly.I want to know how to get and make the ubl_nand.bin (14 KB
first stage boot-loader) and where is the download-URL for ubl_nand.bin
source code?
Currently the LCD configuration is not supported on the VPBE driver. So
format=rgb will not work.
I saw in .../drivers/video/davincifb.c that format=component is
supported (dm64xxfb_pal_component_config function)
I try with format=component but still have no signal on DC5.
Shouldn't be HSYNC
For Component and for all analog outputs you will see the proper signals
at the 4 DACs (DAC_IOUTA - D). In case of component, you will see that
the three DACs carry the Y, Cr and Cb signals, with Y containing the
sync signals.
Right, I see the signals there.
I will try to write a new function
Currently the LCD configuration is not supported on the VPBE driver. So
format=rgb will not work.
Thank you Hadli!
So: Should I find the proper configuration in VPBE registers to get
video in DC5?
Any clue?
Thanks!
Carlos
___
95 matches
Mail list logo