On Tue, 28 May 2013 13:35:27 +0200 Jean Delvare <khali@xxxxxxxxxxxx> wrote: > Hi Richard, > > On Fri, 24 May 2013 08:28:40 -0700, rh wrote: > > On Fri, 24 May 2013 08:59:27 +0200 > > Jean Delvare <khali@xxxxxxxxxxxx> wrote: > > > On Fri, 24 May 2013 03:21:29 -0700, rh wrote: > > > > # sensors-detect revision 6085 (2012-10-30 18:18:45 +0100) > > > > # System: VIA Technologies Ltd. VX800 [1.0] > > > > (...) > > > > Next adapter: SMBus Via Pro adapter at 0500 (i2c-3) > > > > Do you want to scan it? (YES/no/selectively): > > > > Client at address 0x2f can not be probed - unload all client > > > > drivers first! > > > > > > This is your W83L786G chip. sensors-detect would tell you if the > > > driver was built as a module rather than embedded in the kernel. > > > > > > I have improved sensors-detect so that it can handle monolithic > > > kernels better. Please give it a try and report the output: > > > http://khali.linux-fr.org/devel/lm-sensors/sensors-detect > > > > I tried this and sensors-detect produces the same output. > > I tested it myself and it doesn't produce the exact same output for > me. The new script properly reports that the busy I2C address has a > built-in driver attached and it reports that driver's name (which in > most cases matches the module name.) I re-ran and you're correct, I have several versions of sensors-detect and I had been adding/removing from my PATH. > > This could be improved by remembering this finding and using it when > later presenting the summary, but I don't really have the time for > that right now. > > > It seems I don't need to rely on sensors-detect though and > > need to find out why pwmconfig cannot control the fans. > > That's right, once you have the right modules built-in, sensors-detect > is almost useless, except for distributions which will not start > dependent services (e.g. sensord or fancontrol) if the configuration > file is missing. But anyway, once you go for a monolithic kernel you > are moving away from the distribution and must be ready to experience > some hiccups. That's putting it mildly. I usually run nconfig (or menuconfig) from a console and I can tell you that in regards to building custom kernels things have not improved over time and they have not stayed the same either. I've seen things like Depends on: This: [y]=n etc., etc. I do not have the desire to track those down. Usually I disable anything that looks wrong. Once the kernel boots I use it. > > > But it's another guess. I did see that in the 3.9.2 kernel there's > > a separate flag for PWM that I had never noticed before. I think > > it's new. I checked the box for that. > > This is unrelated. The new PWM "subsystem" is essentially used on > embedded systems, hwmon drivers aren't using it (yet.) > > > I just want to have fan control and it seems all the pieces are > > there but something's missing. It could be PEBKAC but I've looked > > at a lot of stuff....the vx800 spec, the W83l786NG,NR,G,R > > datasheets, linux/Documentation/hwmon, thermal, etc., etc. > > The W83L786* chips aren't very popular, I don't think the driver > received too much testing and fan control is the last part of hwmon > driver to receive proper testing coverage. So it is entirely possible > that there is a bug remaining in this area of the driver. > The driver was written by Winbond and seems to have been tossed into the fray. Also some Via use only DC line direct. There are GPIO pins too plus this MB has more SMBus pins and some others as well. I've fiddled with pwm*_mode too and saw bizarre things. So instead of bricking something I'm going to let it stand as-is. The delays in the hwmon or pwm or ??? are perplexing from a testing standpoint. i.e. changing pwm1,2 there's a delay > If you share the output of pwmconfig, it will be easier for us to help > you. You said it didn't work, but we have no idea how it failed > exactly. I did already, maybe in other thread. But it's okay for now. Thank you again for time/consideration. _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
