On 12:38 Mon 15 Jun , Giulio Benetti wrote:
Jean-Christophe PLAGNIOL-VILLARD wrote:
On 12:52 Sat 13 Jun , Giulio Benetti wrote:
In this way the linux driver won't work(the one by Renaud CERRATO). Because it expects MR register not to be set, or to be set enabled.
it's work we have test it you need to enable it in the at91bootstrap or in your lowlevel board config
After I don't agree refreshing watchdog inside atmel_usart.c
you MUST do it otherwise you board will be not usable at all simple exammple you stop the boot and get a prompt if you do not type any command you reboot it's not what we expect
If u-boot hangs there instead of execute bootcmd, then the machine lies there forever. In my opinion we should put WATCHDOG_RESET() inside main_loop and nand write and read. What do you think?
NACK for the main_loop
nand read write ok
but if we excecute the bootcmd we have finished and jump to the os so it's over for u-boot
the OS will have to care about the watchdog
Best Regards, J.
Agree with you. But you have to delete hw_watchdog_init(). If you set there MR then linux won't be able to do it. The patch needs only the WATCHDOG_RESET() and nothing more.
you need to update linux, the watchdog can be enable in u-boot but as it can only be disable once if u-boot or any other bootloader init we can not disable it so in this case linux will just have to buy time to the watchdog
Best Regards, J.