Bluewater Systems Support Forum

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Wed, 21 Sep 2011 22:13:44 -0600

Hello again, sorry to have let this thread go cold. In the end I used a bit-bashing SPI driver in our application program, so the question about the kernel SPI driver became moot. Case closed, thanks!

Alastair

Snapper 9G20 SAM-BA

Stephen — Mon, 13 Dec 2010 17:59:32 -0700

Hi,

Is there a way to make SAM-BA or samba-script work with Snapper 9G20?
Other than samba-script, is the Squid micro-debugger is only other way to recover a bricked device?

Re: Migrating from Snapper9260 to Snapper9G20

Stephen — Sun, 12 Dec 2010 17:06:59 -0700

The Snapper9G20 modules that I have are equipped with 256MB of RAM, but Linux and U-boot can only access 64MB of it. Are you able to give me patches to fully utilize the 256MB?

Re: Migrating from Snapper9260 to Snapper9G20

Stephen — Thu, 09 Dec 2010 23:41:36 -0700

To answer my own question, to support Snapper9G20, go to Linux kernel configuration, change ARCH to AT91SAM9G20.

Re: Migrating from Snapper9260 to Snapper9G20

Stephen — Thu, 09 Dec 2010 21:37:07 -0700

My Snapper9260 linux kernel (with 256k flash block size mods on board-snapper9260.c) works fine on my Snapper9260, but when I tried it on my Snapper9G20, u-boot uncompressed the linux kernel and started printing binary data on the serial port. Is there any other changes I need to make to my kernel to support Snapper9G20? Or is that a u-boot problem?

Re: How to enable watchdog

Stephen — Thu, 09 Dec 2010 04:41:19 -0700

This patch enables the Snapper's internal watchdog (use at your own risk):

diff -rupEbwBN u-boot-1.1.6-buildroot/board/snapper9260/snapper9260.c u-boot-1.1.6/board/snapper9260/snapper9260.c
--- u-boot-1.1.6-buildroot/board/snapper9260/snapper9260.c 2010-05-05 14:14:35.000000000 +1000
+++ u-boot-1.1.6/board/snapper9260/snapper9260.c 2010-12-03 12:20:22.025393450 +1100
@@ -113,6 +113,7 @@ int board_init (void)
AT91C_BASE_PIOC->PIO_PER = AT91C_PIO_PC7;
AT91C_BASE_PIOC->PIO_OER = AT91C_PIO_PC7;
AT91C_BASE_PIOC->PIO_CODR = AT91C_PIO_PC7;
+ hw_watchdog_reset();
#else
hw_disable_watchdog();
#endif
diff -rupEbwBN u-boot-1.1.6-buildroot/include/configs/snapper9260.h u-boot-1.1.6/include/configs/snapper9260.h
--- u-boot-1.1.6-buildroot/include/configs/snapper9260.h 2010-12-03 11:42:28.713121000 +1100
+++ u-boot-1.1.6/include/configs/snapper9260.h 2010-12-03 12:27:30.576379249 +1100
@@ -59,7 +59,7 @@
#define CONFIG_SETUP_MEMORY_TAGS 1
#define CONFIG_INITRD_TAG 1

-//#define CONFIG_HW_WATCHDOG
+#define CONFIG_HW_WATCHDOG
//#define CONFIG_BOOTCOUNT_LIMIT /* support for bootcount limit */
//#define CFG_I2C_RTC_ADDR 0x51
//#define CFG_BOOTCOUNT_ADDR 0x300FF0 /* end of sram */

Re: How to enable watchdog

Stephen — Thu, 09 Dec 2010 04:39:08 -0700

To answer my own question, u-boot-1.16 from Bluewater Systems server currently does not strobe the Snapper's internal watchdog. Another problem is u-boot stays in a loop polling for serial port data, and does not strobe the watchdog while in the loop. I managed to fix both these problems but have not fully tested it.

Here's the patch to strobe the Snapper's internal watchdog (use at your own risk):

diff -rupEbwBN u-boot-1.1.6-snapper-org/board/snapper9260/snapper9260.c u-boot-1.1.6-snapper/board/snapper9260/snapper9260.c
--- u-boot-1.1.6-snapper-org/board/snapper9260/snapper9260.c 2010-05-05 14:14:35.000000000 +1000
+++ u-boot-1.1.6-snapper/board/snapper9260/snapper9260.c 2010-12-09 22:19:49.469823000 +1100
@@ -43,6 +43,9 @@
/* Tickle the watchdog(s) */
void hw_watchdog_reset(void)
{
+ /* Reset internal watchdog */
+ AT91C_BASE_WDTC->WDTC_WDCR = (0xA5 << 24) | 1;
+
/* Toggle the WDOG PIO's */
if (AT91C_BASE_PIOC->PIO_SODR & AT91C_PIO_PC6) {
/* Clear the PIO */
@@ -58,6 +61,9 @@ void hw_watchdog_reset(void)
#else
void hw_disable_watchdog(void)
{
+ // Disable internal watchdog
+ AT91C_BASE_WDTC->WDTC_WDMR = AT91C_WDTC_WDDIS;
+
// Disable the watchdog
// This code currently generates a 1Hz pulse on PC6 with a 50% duty cycle
// 1. setup TIOB (peripheral controlled)
@@ -113,6 +119,7 @@ int board_init (void)
AT91C_BASE_PIOC->PIO_PER = AT91C_PIO_PC7;
AT91C_BASE_PIOC->PIO_OER = AT91C_PIO_PC7;
AT91C_BASE_PIOC->PIO_CODR = AT91C_PIO_PC7;
+ hw_watchdog_reset();
#else
hw_disable_watchdog();
#endif
diff -rupEbwBN u-boot-1.1.6-snapper-org/common/main.c u-boot-1.1.6-snapper/common/main.c
--- u-boot-1.1.6-snapper-org/common/main.c 2008-11-03 10:38:50.000000000 +1100
+++ u-boot-1.1.6-snapper/common/main.c 2010-12-09 22:12:29.522252000 +1100
@@ -469,6 +469,7 @@ void main_loop (void)
}
#endif
len = readline (CFG_PROMPT);
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */

flag = 0; /* assume no special flags for now */
if (len > 0)
@@ -493,6 +494,7 @@ void main_loop (void)
puts ("\n");
else
rc = run_command (lastcommand, flag);
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */

if (rc <= 0) {
/* invalid command or not repeatable, forget it */
@@ -739,6 +741,7 @@ static int cread_line(char *buf, unsigne
char esc_save[8];

while (1) {
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
rlen = 1;
ichar = getcmd_getch();

@@ -1227,6 +1230,7 @@ int readline (const char *const prompt)
for (;;) {
#ifdef CONFIG_BOOT_RETRY_TIME
while (!tstc()) { /* while no incoming data */
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
if (retry_time >= 0 && get_ticks() > endtime)
return (-2); /* timed out */
}
@@ -1235,6 +1239,7 @@ int readline (const char *const prompt)

#ifdef CONFIG_SHOW_ACTIVITY
while (!tstc()) {
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
extern void show_activity(int arg);
show_activity(0);
}
diff -rupEbwBN u-boot-1.1.6-snapper-org/cpu/arm926ejs/at91sam926x/serial.c u-boot-1.1.6-snapper/cpu/arm926ejs/at91sam926x/serial.c
--- u-boot-1.1.6-snapper-org/cpu/arm926ejs/at91sam926x/serial.c 2008-11-19 14:36:34.000000000 +1100
+++ u-boot-1.1.6-snapper/cpu/arm926ejs/at91sam926x/serial.c 2010-12-09 22:13:19.337138000 +1100
@@ -35,6 +35,7 @@
*/

#include
+#include
#include
#include

@@ -92,7 +93,9 @@ void serial_putc (const char c)
{
if (c == '\n')
serial_putc ('\r');
- while ((us->US_CSR & AT91C_US_TXRDY) == 0);
+ while ((us->US_CSR & AT91C_US_TXRDY) == 0) {
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
+ }
us->US_THR = c;
}

@@ -105,7 +108,9 @@ void serial_puts (const char *s)

int serial_getc (void)
{
- while ((us->US_CSR & AT91C_US_RXRDY) == 0);
+ while ((us->US_CSR & AT91C_US_RXRDY) == 0) {
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
+ }
return us->US_RHR;
}

diff -rupEbwBN u-boot-1.1.6-snapper-org/drivers/atmel_usart.c u-boot-1.1.6-snapper/drivers/atmel_usart.c
--- u-boot-1.1.6-snapper-org/drivers/atmel_usart.c 2006-11-13 15:30:51.000000000 +1100
+++ u-boot-1.1.6-snapper/drivers/atmel_usart.c 2010-12-09 22:14:22.021248000 +1100
@@ -16,6 +16,7 @@
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include
+#include

#ifdef CONFIG_ATMEL_USART
#include
@@ -64,7 +65,9 @@ void serial_putc(char c)
if (c == '\n')
serial_putc('\r');

- while (!(usart3_readl(gd->console_uart, CSR) & USART3_BIT(TXRDY))) ;
+ while (!(usart3_readl(gd->console_uart, CSR) & USART3_BIT(TXRDY))) {
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
+ }
usart3_writel(gd->console_uart, THR, c);
}

@@ -76,7 +79,9 @@ void serial_puts(const char *s)

int serial_getc(void)
{
- while (!(usart3_readl(gd->console_uart, CSR) & USART3_BIT(RXRDY))) ;
+ while (!(usart3_readl(gd->console_uart, CSR) & USART3_BIT(RXRDY))) {
+ WATCHDOG_RESET(); /* Trigger watchdog, if needed */
+ }
return usart3_readl(gd->console_uart, RHR);
}

Re: How to enable watchdog

Stephen — Wed, 08 Dec 2010 23:27:35 -0700

I have enabled watchdog on the bootstrap and u-boot (by uncommenting CONFIG_HW_WATCHDOG in snapper9260.h), but looks like u-boot is not resetting the watchdog, and the board keeps rebooting. Is the u-boot-1.16 on bluewater systems server resetting the watchdog correctly? Is there a fix for this?

Re: Migrating from Snapper9260 to Snapper9G20

andre — Mon, 06 Dec 2010 16:03:31 -0700

Hi Stephan,
You are correct, it does require a different binary. This can also be built from the same sources provided above.
However our nightly build is available from [www.bluewatersys.com].

Regards,
Andre

Re: Migrating from Snapper9260 to Snapper9G20

Stephen — Mon, 06 Dec 2010 15:49:21 -0700

Snapper9G20 requires a different bootstrap binary from Snapper9260 right?
It would be great if you can also upload sn9g20_bootstrap.bin so I can use it as a baseline in case I need to roll back to a known working version.
Thanks.

Re: Migrating from Snapper9260 to Snapper9G20

Ryan — Wed, 01 Dec 2010 20:19:04 -0700

It can be downloaded here.

~Ryan

Re: Migrating from Snapper9260 to Snapper9G20

Stephen — Wed, 01 Dec 2010 19:25:01 -0700

Are you able to upload the Snapper 9260 bootstrap source code as well? I need to have the internal watchdog timer enabled.

Re: Migrating from Snapper9260 to Snapper9G20

Ryan — Wed, 01 Dec 2010 18:06:39 -0700

This is correct, the Snapper 9G20 has a block size of 256k. To allow compatibility between the modules we have changed the partition layout for both the Snapper 9260 and 9G20 to be 256k aligned.

I have fixed up a couple of outstanding commits we have. Can you please download tonight's nightly builds for Snapper 9260 bootstrap and U-Boot. The U-Boot image is common and can be used on both Snapper 9260 and 9G20.

To upgrade the bootstrap and U-Boot on your existing 9260 units (Note, if this process fails the unit may need to be reflashed manually using JTAG or SAMBA):

nand erase 0x0 0x7ffff
tftp $bootaddr /tftpboot/path/to/sn9260_bootstrap.bin
nand write $bootaddr 0x0 0x40000
tftp $bootaddr /tftpboot/path/to/u-boot.bin
nand write $bootaddr 0x40000 0x40000

Hope this helps,
~Ryan

Migrating from Snapper9260 to Snapper9G20

Stephen — Wed, 01 Dec 2010 17:19:45 -0700

Hi,

I'm migrating from Snapper9260 to Snapper9G20. From the 2.6.29 kernel board-snapper9260.c file, it appears that Snapper9260 has 128k flash erase block size while Snapper9G20 has 256k flash erase block size. What are the implications of this change?

In order to support both the Snapper9G20 and the older Snapper9260 with 128k erase block size, I need to program a new bootstrap (preboot) that looks for u-boot at address 0x40000 instead of 0x20000?

Do I also need a new u-boot that looks for u-boot environment variables at address 0x80000 instead of 0x60000?

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Sun, 31 Oct 2010 17:29:11 -0600

FYI, I've checked spidev's operation on the Snapper 9260 QSK with an oscilloscope, and have found that the Snapper SPI drivers don't like nonzero values in the speed_hz and bits_per_word fields of the spi_ioc_transfer argument to the SPI_IOC_MESSAGE ioctl call. The spidev_test example supplied with the 2.6.33.3 kernel runs successfully but produces no activity on the SPI bus unless spidev_test.c is modified to leave these fields at zero. The spidev_fdx example works OK as supplied because it doesn't use these fields.

What value should spi_board_info.max_speed_hz be set to in board-snapper9g20.c? Here's the board info I'm using at present, copied from forum topic [www.designarm.com], to make spidev use the Snapper's SPI1 controller.

static struct spi_board_info spi0_board_info[] __initdata = {
        {
                .modalias       = "spidev",
                .max_speed_hz   = 15 * 1000 * 1000,
                .chip_select    = 0,
                .bus_num        = 1,
        }
};

Thanks,
Alastair

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Thu, 28 Oct 2010 00:00:06 -0600

Have found my I/O expander bug, thanks - had initialised PA24:23 as the 2.6.20 kernel initialised them, as inputs, which clobbered the I2C bus with the later kernels. It's all happy now.

Alastair

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Wed, 27 Oct 2010 23:05:09 -0600

Hi Andre,

The 2.6.33.3 build completes OK with snapper9g20_defconfig. Silly me for not trying that first, d'oh!!

Spidev works OK with the 2.6.33.3 kernel, after adding initialisation code to board-snapper9g20.c, at least to the extent that spidev-test produces plausible-looking output. I won't know more until our target hardware with SPI DACs is built but it's looking good now.

The trouble with the I/O expander appears to be caused by my software, sorry! It initialises the CPU's PIO controllers in a way which works OK with the 2.6.20 kernel, but makes the I/O expander inaccessible under the 2.6.29 or 2.6.33.3 kernels until the Snapper is powered off and on again. I will investigate.

The ADC support appears to be incomplete in the 2.6.33.3 kernel. Presumably that's work in progress.

For some reason I've not been able to boot either the 2.6.29 or the 2.6.33.3 kernel from a root filesystem in flash. Boot fails with a message as follows (this example is from the 2.6.33.3 kernel).

...
yaffs: dev is 32505860 name is "mtdblock4"
yaffs: passed flags ""
VFS: Mounted root (yaffs filesystem) on device 31:4.
Freeing init memory: 120K
Warning: unable to open an initial console.
Kernel panic - not syncing: No init found.  Try passing init= option to kernel.
[] (unwind_backtrace+0x0/0xf0) from [] (dump_stack+0x18/0x1c)
[] (dump_stack+0x18/0x1c) from [] (panic+0x4c/0x140)
[] (panic+0x4c/0x140) from [] (init_post+0xc8/0xfc)
[] (init_post+0xc8/0xfc) from [] (kernel_init+0xf0/0x120)
[] (kernel_init+0xf0/0x120) from [] (kernel_thread_exit+0x0/0x8)

I created the YAFFS2 image of the root filesystem with samba-script and wrote it to flash using mtdtool 0.6.2 running under the 2.6.20 kernel. The 2.6.20 kernel boots it OK from flash. All three kernels can mount and read the rootfs in flash after they have been booted from NFS, although the 2.6.33.3 kernel gives warnings on the first mount, as follows.

... (boot 2.6.33.3 kernel from an NFS rootfs)
root@snapper:~$ mount -t yaffs2 /dev/mtdblock4 /mnt
yaffs: dev is 32505860 name is "mtdblock4"
yaffs: passed flags ""
uncorrectable error :
uncorrectable error :
uncorrectable error :
uncorrectable error :
uncorrectable error :
uncorrectable error :
root@snapper:~$ ls /mnt
boot        home        linuxrc     proc        sys
etc         lib         lost+found  sbin
root@snapper:~$ umount /mnt
root@snapper:~$ mount -t yaffs2 /dev/mtdblock4 /mnt
yaffs: dev is 32505860 name is "mtdblock4"
yaffs: passed flags ""
root@snapper:~$ ls -l /mnt
drwxrwsr-x    1 513      1505         2048 Oct 10  2010 boot
drwxrwsr-x    1 root     root         2048 Jun 11  2009 etc
drwxrwsr-x    1 513      1505         2048 Oct 10  2010 home
drwxrwsr-x    1 513      1505         2048 Oct 10  2010 lib
lrwxrwxrwx    1 513      1505           11 Oct 13  2010 linuxrc -> bin/busybox
drwx------    1 root     root         2048 Jan  1 00:00 lost+found
drwxrwsr-x    1 513      1505         2048 Oct 10  2010 proc
drwxrwsr-x    1 513      1505         2048 Oct 10  2010 sbin
drwxrwsr-x    1 513      1505         2048 Oct 10  2010 sys
root@snapper:~$ umount /mnt

Is mtdtool under 2.6.20 an inappropriate way to write the rootfs to flash for the newer kernels? The rootfs write encountered bad erase blocks but presumably that doesn't matter.

root@snapper:~$ mtdtool --write_raw Filesystem root_101027b.yaffs2
034%    4M (884.48K/s)            Skipped block at 0x420000 - bad
055%    7M (884.58K/s)            Skipped block at 0x6a0000 - bad
089%   11M (891.95K/s)            Skipped block at 0xac0000 - bad
100%   12M (894.45K/s)
Failed to read from file: Success
FAILURE

Alastair

Re: Cannot debug from Eclipse with Snapper 2.6.29 kernel

alastair — Wed, 27 Oct 2010 22:11:55 -0600

Debugging from Eclipse works OK with the 2.6.33.3 kernel, thanks!

Running from Eclipse (which uses an SSH session instead of gdbserver) doesn't work properly with the 2.6.33.3 kernel unless the previous run has been explicitly terminated with Eclipse's "Remove Launch" or "Remove All Terminated Launches" button - there appears to be a loss of synchronisation between Eclipse and the Snapper that doesn't happen with the 2.6.20 kernel. But pressing the "Remove Launch" button before each run is a work-around I can easily live with.

Alastair

Re: Enabling Snapper peripheral support in 2.6.29 kernel

andre — Tue, 26 Oct 2010 15:48:04 -0600

Hi Alastair,
Sorry, try using the snapper9g20_defconfig.

Regards,
Andre

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Tue, 26 Oct 2010 15:32:53 -0600

Hello Andre,

The 2.6.33.3 kernel build, configured with snapper9g45_defconfig, failed with the following error. I haven't tried to fix it yet - any hints?

...
  CC      sound/soc/atmel/atmel_ssc_dai.o
  CC      drivers/i2c/i2c-dev.o
sound/soc/atmel/atmel_ssc_dai.c: In function 'atmel_ssc_shutdown':
sound/soc/atmel/atmel_ssc_dai.c:275: error: 'struct atmel_ssc_info' has no member named 'substreams_running'
sound/soc/atmel/atmel_ssc_dai.c: In function 'atmel_ssc_hw_params':
sound/soc/atmel/atmel_ssc_dai.c:671: error: 'struct atmel_ssc_info' has no member named 'substreams_running'
sound/soc/atmel/atmel_ssc_dai.c: In function 'atmel_ssc_prepare':
sound/soc/atmel/atmel_ssc_dai.c:710: error: 'struct atmel_ssc_info' has no member named 'substreams_running'
make[3]: *** [sound/soc/atmel/atmel_ssc_dai.o] Error 1
make[2]: *** [sound/soc/atmel] Error 2
make[1]: *** [sound/soc] Error 2
make: *** [sound] Error 2
make: *** Waiting for unfinished jobs....
...

We're using the Snapper 9260 QSK.

Thanks,
Alastair

Re: Cannot debug from Eclipse with Snapper 2.6.29 kernel

andre — Mon, 25 Oct 2010 20:40:44 -0600

Hi Alastair,
Can you please retry this with the new Linux 2.6.33.3 kernel:
[www.bluewatersys.com]

Regards,
Andre

Re: Enabling Snapper peripheral support in 2.6.29 kernel

andre — Mon, 25 Oct 2010 20:40:00 -0600

Hi Alastair,
Is it possible for you to test the Linux 2.6.33.3 kernel listed below?
[www.bluewatersys.com]

There is also an updated rootfs:
[www.bluewatersys.com]

This should resolve the spidev bug.

For the gpio issue - are you using the QSK carrier board, or a custom one?

Regards,
Andre

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Mon, 25 Oct 2010 14:57:39 -0600

Hi Ryan,

Can you post or email me a .config for the 2.6.29 kernel that you know gpiotool works worth? That might be an easy way forward.

Thanks,
Alastair

Cannot debug from Eclipse with Snapper 2.6.29 kernel

alastair — Thu, 21 Oct 2010 21:02:37 -0600

Hi! I'm developing on a Snapper 9260 QSK using the Sourcery G++ Lite 2007q3 tools within Eclipse Helios (build ID 20100617-1415) on Windows XP, connected to the Snapper by Ethernet. When the Snapper is running its Linux kernel 2.6.20 the development environment works well: I can run and debug programs on the Snapper from Eclipse. When the Snapper is running the 2.6.29 kernel, running programs from Eclipse works OK but any attempt to debug a program from Eclipse fails immediately. N.B. running programs uses SSH, debugging programs uses gdbserver.

a) A dialog box pops up "Problem Occurred. An error has occurred. See error log for more details. java.lang.NullPointerException".

b) The debug pane in Eclipse includes the following interesting things.

  gdbserver debugger  (Suspended)
    Thread (1) (Suspended: Signal 'SIGILL' received. Description: Illegal instruction.)
      1 main() \spidev_test.c 150 0x0008864

Have you seen this trouble before? Do you have any tips for getting the Eclipse source debugger to work with the 2.6.29 kernel?

Thanks,
Alastair

Re: Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Thu, 21 Oct 2010 17:28:18 -0600

I/O expander...

gpiotool, which I presume behaves equivalently to libgpio, fails in the following way.

$ gpiotool io_exp_b6
Polarity 1 error: No such device or address [6]
$ gpiotool a1
GPIO A1: input high: GPIO (pullup)
$ gpiotool b1
GPIO B1: input high: periph A

I think that the MAX7311 is not responding on the I2C bus. In my own program, which is derived from gpiotool, the /dev/i2c-0 device node opens OK and the ioctl call to tell the I2C bus driver the slave address of the MAX7311 succeeds, but calls to i2c_smbus_read_byte_data() after that fail.

Does board-snapper9260.c have to be edited as well? Its table sn9260_i2c_devices[] includes an entry for the MAX7311 at address 0x28.

Is enabling the kernel driver support in menuconfig (device drivers > gpio support > PCA953x ...) mutually exclusive with using /dev/i2c-0 as gpiotool does?

spidev...

After setting the menuconfig items as you said, editing board-snapper9260.c as in another forum topic (http://www.designarm.com/forum/read.php?3,159), and building and running the resulting 2.6.29 kernel, mdev creates a device node /dev/spidev1.0 but attempting to use it causes a crash in the kernel.

boot@snapper:/home/ams$ ls /dev/spi*
/dev/spidev1.0
root@snapper:/home/ams$ ./spidev_test -D /dev/spidev1.0
spi modkernel BUG at include/linux/spi/spi.h:609!
Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = c3a3c000
[00000000] *pgd=2383a031, *pte=00000000, *ppte=00000000
Internal error: Oops: 817 [#1]
Modules linked in:
CPU: 0    Not tainted  (2.6.29 #9)
PC is at __bug+0x20/0x2c
LR is at release_console_sem+0x198/0x1b0
pc : []    lr : []    psr: 60000093
sp : c3a65e50  ip : c3a65d88  fp : c3a65e5c
r10: 00000000  r9 : 00000001  r8 : c3a64000
r7 : c3a5b026  r6 : c38eae00  r5 : c3967244  r4 : c3a65ea4
r3 : 00000000  r2 : c02b6270  r1 : 00001a0c  r0 : 0000002e
Flags: nZCv  IRQs off  FIQs on  Mode SVC_32  ISA ARM  Segment user
Control: 0005317f  Table: 23a3c000  DAC: 00000015
Process spidev_test (pid: 947, stack limit = 0xc3a64268)
Stack: (0xc3a65e50 to 0xc3a66000)
5e40:                                     c3a65e84 c3a65e60 c0166a68 c002a830
5e60: c3967220 00000000 c3a65e68 c3a65e68 00000001 c39b8240 c3a65efc c3a65e88
5e80: c0167198 c0166a08 c013d774 c3931430 c39b8220 c3931420 00000026 c3967220
5ea0: 00000000 c396723c c396723c 00000000 00000000 c0166ca8 c3a65e64 00000000
5ec0: 00000000 00000000 00000000 00000000 c3a64000 be8cbcd0 40206b00 00000003
5ee0: c38cb660 be8cbcd0 c3a64000 00000000 c3a65f14 c3a65f00 c0090c44 c0166ccc
5f00: c38cb660 be8cbcd0 c3a65f7c c3a65f18 c0091188 c0090c20 00000000 c3a64000
5f20: c3a65f3c c3a65f30 c013e3d0 c01293a0 c3a65f54 c3a65f40 c0042500 c013e3cc
5f40: 00000100 c3a64000 c3a65f84 c3a65f58 c0042354 c00421d8 00000003 be8cbcd0
5f60: 40206b00 c38cb660 c0026f28 00000000 c3a65fa4 c3a65f80 c009121c c0090cc0
5f80: c00425a4 00000000 be8cbcd0 00000008 be8cbc84 00000036 00000000 c3a65fa8
5fa0: c0026d80 c00911ec be8cbcd0 00000008 00000003 40206b00 be8cbcd0 be8cbc74
5fc0: be8cbcd0 00000008 be8cbc84 00000036 00000000 00000000 40024000 00000000
5fe0: 00000003 be8cbc70 000087a9 400deb5c 80000010 00000003 00000000 00000000
Backtrace:
[] (__bug+0x0/0x2c) from [] (spidev_sync+0x70/0xc0)
[] (spidev_sync+0x0/0xc0) from [] (spidev_ioctl+0x4dc/0x590)
 r4:c39b8240
[] (spidev_ioctl+0x0/0x590) from [] (vfs_ioctl+0x34/0x78)
[] (vfs_ioctl+0x0/0x78) from [] (do_vfs_ioctl+0x4d8/0x52c)
 r5:be8cbcd0 r4:c38cb660
[] (do_vfs_ioctl+0x0/0x52c) from [] (sys_ioctl+0x40/0x64)
[] (sys_ioctl+0x0/0x64) from [] (ret_fast_syscall+0x0/0x2c)
 r7:00000036 r6:be8cbc84 r5:00000008 r4:be8cbcd0
Code: e1a01000 e59f000c eb004ed0 e3a03000 (e5833000)
e:---[ end trace 2f518758c11dd51e ]---
 0
bits per word: 8
max speed: 500000 Hz (500 KHz)
Segmentation fault
root@snapper:/home/ams$ ./spidev_test -D /dev/spidev1.0
(the system hangs)

This looks like the same crash as [www.designarm.com] describes. Do you know a fix for this?

Thanks,
Alastair

Re: Enabling Snapper peripheral support in 2.6.29 kernel

Ryan — Thu, 21 Oct 2010 14:12:34 -0600

For the max7311 IO Expander you can use our libgpio library, which has a user space driver for the IO expander and also understands the symbolic names for the IO expander pins on the Snapper 9260. If you want to use the kernel driver it is in menuconfig under device drivers, gpio support, PCA953x, PCA955x, TCA64xx, and MAX7310 I/O ports. You will probably also want to enable /sys/class/gpio in the same menu (see linux/Documentation/gpio.txt for details).

The SPI bus controller is in menuconfig under device drivers, spi support, Atmel SPI Controller. You can enable spidev by select the User mode SPI device driver support option. To use spidev you will need to run /etc/init.d/mdev.sh to generate the spidev device nodes (see linux/Documentation/spi/spidev for details).

~Ryan

Enabling Snapper peripheral support in 2.6.29 kernel

alastair — Wed, 20 Oct 2010 22:56:46 -0600

What steps are needed to enable support in the 2.6.29 kernel build for Snapper's I/O expander and SPI bus controller? I've been experimenting with the menuconfig and have got the AT91 ADC driver working, but not the I/O expander support or spidev.

Thanks,
Alastair

Re: Kernel binary with ADC support for Snapper 9260

alastair — Wed, 20 Oct 2010 20:55:11 -0600

I had been pursuing recovery using samba-script because the Snapper would not boot the 2.6.20 kernel, mtdtool does not work under 2.6.29, and the filesystem's pseudofiles were absent when booted from 2.6.29.

However I've since learned how to set up an NFS server on Linux, and have been able to boot the Snapper from it and restore the root filesystem with mtdtool as you described. We can close this case now, thanks!

Re: Kernel binary with ADC support for Snapper 9260

Ryan — Mon, 18 Oct 2010 20:15:04 -0600

The easiest way to write a YAFFS partition is to simply mount it and copy files to it

mount /dev/mtdblock4 /mnt
cp -a /root_files/* /mnt
umount /mnt

You can find out which mtdblock device to mount by looking at the contents of /proc/mtd.

If you want to flash a binary image of a YAFFS filesystem, then you can use our mtdtool utility:

mtdtool --eraseall filesystem
mtdtool --write_raw filesystem yaffs.bin

You can use the mkyaffs utility (or our version) to create a binary YAFFS image. Note that you will need to make sure that you set up a mapping for the correct NAND layout.

~Ryan

Re: Kernel binary with ADC support for Snapper 9260

alastair — Mon, 18 Oct 2010 20:04:32 -0600

The Linux start-up messages from the Snapper include the following information about the flash partitions.

Creating 5 MTD partitions on "NAND 256MiB 3,3V 8-bit":
0x00000000-0x00020000 : "Preboot"
0x00020000-0x00060000 : "Bootloader"
0x00060000-0x00080000 : "Environment"
0x00080000-0x00480000 : "Kernel"
0x00480000-0x10000000 : "Filesystem"

Does this mean that the following samba-script commands would correctly rewrite the Snapper's root filesystem from a YAFFS2 image file? I'm reluctant to try this without confirmation for fear of damaging the partition structure of the flash.

AT91> open sn9260
AT91> nand_erase 0x480000 0xFB80000
AT91> nand_write_raw_file 0x480000 root.yaffs2
AT91> quit

Thanks,
Alastair