Search |
 |
|
 |
| Author |
Message |
|
|
 Posted: Feb 08, 2009 - 09:44 AM |
|
Joined: Feb 03, 2008
Posts: 56
|
|
| About a year ago I received a lot of wonderful help getting all the serial ports going on an NGW board. Here is that thread:
Code:
http://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&t=60005
I tried to set up a build to do this again, and found that the code has changed enough so that what I did last year can't be done the same way.
Using the latest (or very recent) buildroot-avr32-v2.1.0 completely from source, I modified the project_build_avr32/atngw100/linux-2.6.27.6/arch/avr32/boards/atngw100/setup.c file. (See the comments):
Code:
static int __init atngw100_init(void)
{
unsigned i;
/*
* ATNGW100 uses 16-bit SDRAM interface, so we don't need to
* reserve any pins for it.
*/
at32_add_device_usart(0);
at32_add_device_usart(1); /* DocuGeek added USART1 device */
at32_add_device_usart(2); /* DocuGeek added USART2 device */
at32_add_device_usart(3); /* DocuGeek added USART3 device */
I was concerned because these three ports might conflict with other items:
Quote:
USART0 shares with PSIF Clock0/Data0
USART2 shares with EXTINT 1 and 2
USART3 shares with SSC2 Frame/Clock
but I could not find anything but the PSIF in the code.
Code:
/* DocuGeek removed this:
at32_add_device_spi(0, spi0_board_info, ARRAY_SIZE(spi0_board_info));
*/
I looked through the project_build_avr32/atngw100/linux-2.6.27.6/arch/avr32/mach-at32ap/at32ap700x.c file, and it appears that all the configurations have been moved elsewhere.
So, I was hoping that, since many things had been changed, that was all I needed to do. However, after
Code:
make rootclean
make
and extracting the rootfs.avr32.tar file from binaries/atngw100 to an SD Flash, and booting it, the dmesg gave me the same old single SERIAL line it had before.
Code:
...
io scheduler cfq registered (default)
atmel_usart.0: ttyS0 at MMIO 0xffe01000 (irq = 7) is a ATMEL_SERIAL
MACB_mii_bus: probed
...
To me, this is a clear sign that the serial ports have not been configured.
What am I missing? |
|
|
| |
|
|
|
|
|
|
Posted: Feb 16, 2009 - 10:58 AM |
|
Joined: Sep 14, 2003
Posts: 4046
Location: Queanbeyan, Australia
|
|
If there are pin conflicts the boot logs will be full of backtraces and scary things like that.
You look like you`re simply missing the calls to at32_map_usart() in the board setup function. I`m not in the office this at the moment so can`t be of more specific help, but look through setup.c for lines which are commented with something like
Code:
USARTA -> ttyS1
-S. |
_________________
Blag: http://www.niasdigital.com/blag
|
| |
|
|
|
|
|
|
Posted: Feb 16, 2009 - 11:39 AM |
|
Joined: Jan 07, 2003
Posts: 3437
Location: Trondheim, Norway
|
|
| Remember to do a
Code:
make linux26-force
after changing the kernel code. |
|
|
| |
|
|
|
|
|
|
Posted: Feb 17, 2009 - 06:09 AM |
|
Joined: Apr 02, 2007
Posts: 24
|
|
Find at32_map_usart call in your setup.c and
add the lines like this
Code:
at32_map_usart(1, 0); /*usart1 as ttyS0*/
at32_map_usart(0, 1); /*usart0 as ttyS1*/
at32_map_usart(2, 2); /*usart2 as ttyS2*/
at32_map_usart(3, 3); /*usart3 as ttyS3*/
And here is the chunk of my startup log
Code:
io scheduler noop registered
io scheduler cfq registered (default)
atmel_usart.0: ttyS0 at MMIO 0xffe01000 (irq = 7) is a ATMEL_SERIAL
atmel_usart.1: ttyS1 at MMIO 0xffe00c00 (irq = 6) is a ATMEL_SERIAL
atmel_usart.2: ttyS2 at MMIO 0xffe01400 (irq = 8) is a ATMEL_SERIAL
atmel_usart.3: ttyS3 at MMIO 0xffe01800 (irq = 9) is a ATMEL_SERIAL
MACB_mii_bus: probed
eth0: Atmel MACB at 0xfff01800 irq 25 (00:04:25:1c:8b:cc)
eth0: attached PHY driver [Generic PHY] (mii_bus:phy_addr=0:01, irq=-1)
MACB_mii_bus: probed
|
|
|
| |
|
|
|
|
|
|
Posted: Feb 19, 2009 - 08:07 AM |
|
Joined: Feb 03, 2008
Posts: 56
|
|
Thanks for the continued help!
Just to document exactly what I changed:
Code:
void __init setup_board(void)
{
at32_map_usart(1, 0); /* USART 1: /dev/ttyS0, DB9 */
at32_setup_serial_console(0);
at32_map_usart(0, 1); /* DocuGeek maps USART0 to /dev/ttyS1 */
at32_map_usart(2, 2); /* DocuGeek maps USART2 to /dev/ttyS2 */
at32_map_usart(3, 3); /* DocuGeek maps USART3 to /dev/ttyS3 */
}
Code:
static int __init atngw100_init(void)
{
unsigned i;
/*
* ATNGW100 uses 16-bit SDRAM interface, so we don't need to
* reserve any pins for it.
*/
at32_add_device_usart(0);
at32_add_device_usart(1); /* DocuGeek added USART1 device */
at32_add_device_usart(2); /* DocuGeek added USART2 device */
at32_add_device_usart(3); /* DocuGeek added USART3 device */
set_hw_addr(at32_add_device_eth(0, ð_data[0]));
set_hw_addr(at32_add_device_eth(1, ð_data[1]));
/* DocuGeek removed this:
at32_add_device_spi(0, spi0_board_info, ARRAY_SIZE(spi0_board_info));
*/
at32_add_device_mci(0, &mci0_data);
Both of those are in this file:
/home/l1g8/AtmelNGW/buildroot-avr32-v2.3.0/project_build_avr32/atngw100/linux-2.6.27.6/arch/avr32/boards/atngw100/setup.c
I have my doubts about removing that at32_add_device_spi - the /dev/ list no longer has an mtdblock3, and so I cannot map the Serial Flash. I was trying to turn off the PSIF-Clock0 and -Data0 and became confused.
Anyhow, all the above was done that way from the beginning of this post - I had just forgotten to mention the setup_board(void) call in my previous post.
What actually helped was this:
Code:
make linux26-force
So, hce, from what part of the entire Earth did you come up with that?
GONE is my small feeling of competence at discovering the end of /home/l1g8/AtmelNGW/buildroot-avr32-v2.3.0/Makefile had a list of "make" commands and meanings.
I checked again, and linux26-force is nowhere in that file. Anyhow, thanks for the magical incantation!
And thanks to you all for getting me by this snag and on to the next step. |
|
|
| |
|
|
|
|
|
|
Posted: Feb 19, 2009 - 08:56 AM |
|
Joined: Jan 07, 2003
Posts: 3437
Location: Trondheim, Norway
|
|
| Take a peek in target/linux/Makefile.in or target/linux/Makefile.in.advanced, whatever you have enabled. Most likely the latter. |
|
|
| |
|
|
|
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
