On Wed, Sep 19, 2012 at 03:17:39PM +0200, Åsa Uhrenius wrote: > Hi Guenter, > Please don't top-post, and don't drop the mailing list from your replies. > Thanks a lot for your answer! I get your point. > > If I understand the documents I find correctly, it's possible to write a > voltage level to a pmbus-compatible device with pmbus. Correct? > If so, in which version was that included? > All versions. See chapter 8 and 9 of the PMBus specification. Before you play with it, though, make sure you read and understand the data sheet for your specific chip, as the operation of the related commands is highly chip specific. Guenter > Thanks and Regards / Åsa Uhrenius > > > > > > > From: Guenter Roeck <linux@xxxxxxxxxxxx> > To: Åsa Uhrenius <asa.uhrenius@xxxxxxxxxxxxx> > Cc: lm-sensors@xxxxxxxxxxxxxx > Date: 2012/09/14 18:08 > Subject: Re: pmbus-questions > > > > On Fri, Sep 14, 2012 at 11:25:07AM +0200, Åsa Uhrenius wrote: > > Hi, > > > > I have some questions concerning pmbus. If you are not the persons I > > should contact concerning this please let me know and advise me to whom > I > > should send this. > > > > Questions: > > Is it correct that pmbus is used for monitoring only and can't be used > for > > writing data to a pmbus compatible devices? > > If so, are there any plans to include writing capability and when if > > Yes and no. You can write data, but writes are limited to setting limits. > > > that's the case? > > > > The reason to my questions is that I would like to write a powerlevel to > a > I assume you refer to setting power levels. > > There are a number of answers to that. > > First, so far there has to be demand. My need and the need of others using > the > pmbus driver is for monitoring purposes, so there was no need to add > support for > actually regulating power levels. > > Adding support for setting power levels can be done. It would, however, > require > the driver to be reworked as multi-function driver, with hardware > monitoring and > (probably) regulator sub-drivers. That is a substantial effort, for which > I don't > have time for right now. That doesn't mean it won't happen, but unless my > employer > decides to assign such a task to me (which is highly unlikely), someone > else would > have to pick it up and submit patches. > > That leads to the next, and more substantial, set of problems: The ability > to > modify power levels is nice, but if misused can lead to physical damage to > the > board. After all, PMBus chis are typically not used to set, say, battery > power > levels, but to provide power to on-board components. Adding support for > manipulating > such power levels would have to be done very carefully. In addition to > that, > power level manipulation is an area which is not too well defined in the > PMBus > standard. For some chips, it is just a matter of changing the expected > power > level registers. Other chips need to follow an elaborate sequence of > commands, > and it is effectively only possible to change the power level on those > chips with > the tools provided by the chip manufacturer. Everything else tends to > cause such > chips to shut down, or in bad cases to literally blow up. So this is > something I > would only want to get into on a per-chip basis, and only permit it if I > am > really sure that this specific chip does support "simple" means to set the > power > level. > > Another question is _why_ you want to be able to set the power level. If > it is > for diagnostics purposes, you might be better off by writing the required > I2C > commands directly into the chip, eg with the i2cset command from > i2c-tools. > > Hope this helps, > Guenter > _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
