Hi, I back ported the CS5536 MFGPT based watchdog timer from Kernel 2.6 to Kernel 2.4 (I have to). I am using the Alix2c0 board with TinyBios v0.99. I have been following the discussion on this thread Re: 2.6.24-rc8 hangs at mfgpt-timer". I disabled the MFGPT workaround in the BIOS and it seems to load the module just fine (no more hangs). I get this when I load geode-mfgpt.o:
################ Found vendor: 0x1022h, device: 0x2090h reading from: 6206 reading from: 620e reading from: 6216 reading from: 621e reading from: 6226 reading from: 622e reading from: 6236 reading from: 623e geode-mfgpt: 8 timers available. And when I load the watchdog module geodewdt.o: geode-mfgpt: Registered timer 0 written to: 6206 -- just something I am doing to verify the write operation written to: 6202 Everything seems to be fine except when I try to setup a bit in MFGPT_NR_MSR that allows MFGPT0_CMP2 to perform a hard reset when the event fires, it resets it right away. /* Set up comparator 2 to reset when the event fires */ // geode_mfgpt_set_event(wdt_timer, MFGPT_CMP2, MFGPT_EVENT_RESET); I am not sure if the Clock Switch section 5.16.3 in the CS5536 data book, pointed by Willy Tarreau in one of his posts, is related to it. When the API module is loaded, it only executes read instructions on the MFGPT_REG_SETUP registers. Can anyone speculate as to why the board resets as soon as the watchdog timer module executes the geode_mfgpt_set_event function? Basically when this function is called, (0x100000)h is written to the MFGPT_NR_MSR register. TIA. Regards, Hasan Rashid Embedded Systems Software Engineer -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
