Le mardi 21 octobre 2008 à 19:54 +0200, Remy Bohmer a écrit :
Hello Stelian,
Hi Remy,
Are all your USB stick problems solved by now? Including that stick that did not work?
Nope. I still have one (kinda old) stick which does not work, on any AT91 board. But I suspect the issues are no longer in the board/cpu code, but in the core USB layer.
I will try your patches tomorrow, but I have one question:
#define AT91_MAIN_CLOCK 18432000 /* 18.432 MHz crystal */ #define AT91_MASTER_CLOCK 100000000 /* peripheral */ #define AT91_CPU_CLOCK 200000000 /* cpu */
Are such nice rounded values possible with that unrounded crystal?
Looking at sam9261 I see: #define AT91_MAIN_CLOCK 198656000 /* from 18.432 MHz crystal */
This one is clearly wrong. 18.432 MHz is 18432000 Hz.
#define AT91_MASTER_CLOCK 99328000 /* peripheral = main / 2
And the master clock is not main / 2, but PLLA / 2, where PLLA is initialized by the bootstrap to (main * 0x60) / 9.
This gives: PLLA = 18432000 * 0x60 / 9 = 196608000 Hz MASTER = 196608000 / 2 = 98304000 Hz.
The defines in the header files contain rounded values, I don't think it's too important to be precise here.
Stelian.