From: Willem v. E. <wv...@us...> - 2006-08-24 06:45:09
|
Hi, I'm continuing this conversation on apanel-dev, it may contain information useful to read for others. About word interrupts. I wrote a driver for FreeBSD that uses the interrupts (following the OZ990 datasheet). I had to modify the i2c piix driver also, but finally it worked (*) In theory though. I could read /dev/apanel and see what buttons were pressed, but sometimes it would show buttons pressed twice, though they were pressed only once. I don't know the status of smbalert# in Linux, and indeed - you need it hardwired. Either try to find on your motherboard (see if line is connected to i2c controller), or write software to check it. I didn't look at SMI. Regards, - Willem (*) It probably suffers from bitrot, but you can find it at http://apanel.sourceforge.net/stuff/apanel-freebsd-2.0-prealpha.tar.gz On Wed, 2006-08-23 at 17:17 +0200, Hendrik Sattler wrote: > *snip* > OZ990.pdf seems to also (at least partly) apply for the OZ992S: > registers 0x13 and 0x14 match the description: > yavin:/sys/bus/i2c/devices/0-0018# echo 1F > regs_max > yavin:/sys/bus/i2c/devices/0-0018# cat regs > 0x00: 0f04 0000 aa8a aa55 0000 00ff 0000 0000 > 0x08: 0000 0000 0002 0000 00aa 0001 0000 0000 > 0x10: 0000 0000 0000 3004 9200 0000 0000 0000 > 0x18: 0000 0f01 0000 0000 0000 0010 0000 0000 >=20 > As you can see, register 0x13 shows the left-shifted I=C2=B2C address (= 0x30 ->=20 > 0x18) with SMBus-ID 4 (whatever that means) and > register 0x14 shows the chip ID (0x90 for the OZ990, so 0x92 for the OZ= 992=20 > seems logical) and chip revision (0). > Looks good for proper detection verification. > There are GPIOs set to input but only 4 are for the buttons. >=20 > Do you know, how to get an interrupt for them? Polling is not the best=20 > approach. However, is it known if SMBALERT# (probably not handled by I=C2= =B2C host=20 > controller) or SMI (System Management Interrupt) lines are actually=20 > connected? >=20 > HS |