> Got it working with "modprobe w83781d force_w83781d=9191,0x290". I > guess the "force" alone tried something different. Not sure if this > problem is solvable. If it is, great. If not, the workaround as > explained in the docs could be further elaborated. "force" skips detection (basically checking if there is a chip at 0x290 or not) while force_w83781d additionally skips identification (basically checking if the chip is a Winbond W83781D or not). Both steps are failing for you, probably for the same reason (BIOS messing with the registers). > It's mentioned in the docs ( > http://www2.lm-sensors.nu/~lm78/cvs/lm_sensors2/doc/chips/w83781d ) > that the chip gets misidentified. However my chip was detected, is > that the same as identified? Also, the docs says: > > (modprobe w83781d force_CHIP=BUS,0x2d where CHIP is the chip name > and BUS is your i2c bus number - > cat /proc/bus/i2c to see a list of bus numbers) > > But I have no /proc/bus/i2c (file or directory). This is for 2.4 kernels. > I can't find the > corresponding data in /sys , either because I don't know where to > look, or it's just not there. There is no file in /sys. Instead you can use "i2cdetect -l" (which happens to also work for 2.4 kernels). > I suggest an update to add a note > explaining how to get the right bus number in kernel 2.4 and 2.6. Most (if not all) of the docs in the lm_sensors2 CVS repository are targeted at 2.4 kernels. The reason is both historical (obviously) and technical (2.6 drivers are not part of the package). But I agree that updating the docs as time permits would be great. Another approach would be to add chip documentation to the 2.6 kernel. > Additionally make a clearer distinction between ISA bus/address and > I2C bus/address and list typical numbers for this device: ISA bus 9191 > / address 0x290 and I2C bus 0 or 1 / address 0x2d. Also helpful would > be both the modprobe syntax (as shown) and the modules.conf syntax. > > # Mail client won't let me enter tabs. > > <DOC_begin> > > W8378* PROBLEMS > ----------------- > Known problems: > - Some chips are misidentified even if successfully detected. > This is caused by the chip having the wrong WCHIPID register > value, so there is no fix. The workaround is to use the > force_CHIP parameter. > > modprobe format: > > modprobe w83781d force_CHIP=BUS,ADDR > > modules.conf format: > > options w83781d force_CHIP=BUS,ADDR > > Example values, comma separated, use only one, use only ISA or > I2C: > > CHIP : w83781d > I2C BUS : 0, 1 This can be about any value, depending on the other devices in the system. On my system, the hardware monitoring bus is 4 (0 to 3 are occupied by my Radeon graphics adapter). > I2C ADDR: 0x2d > ISA BUS : 9191 > ISA ADDR: 0x290 > > Find your specific bus: > > kernel 2.4 I2C: cat /proc/bus/i2c > kernel 2.4 ISA: ??? > kernel 2.6 I2C: ??? > kernel 2.6 ISA: ??? Irrelevant for ISA, bus is always 9191. For 2.6 I2C, "i2cdetect -l" (which also works for 2.4). > Find your specific address: > > kernel 2.4 I2C: ??? > kernel 2.4 ISA: ??? > kernel 2.6 I2C: ??? > kernel 2.6 ISA: ??? Always 0x290 for ISA, there is no way to detect it if a different value is. For I2C, "i2cdetect <BUS NUMBER>" should help, although I never saw a W83781D at a non-0x2d address. > > See "Addresses scanned" sections under "Supported chips" above. > > </DOC_end> Care to provide this documentation update as a patch we could apply to the CVS repository? Thanks. (BTW, I cannot possibly connect to the CVS repository for now, so someone else will have to apply the patch.) As for me, I will propose my extra-debug patch (slightly cleaned up) for inclusion into the 2.6 kernel. It helped us once, so it could prove to be valuable in the future as well. Thanks again. -- Jean "Khali" Delvare http://khali.linux-fr.org/
