On Wed, 2012-02-01 at 10:19 -0500, Jean Delvare wrote: > On Wed, 1 Feb 2012 06:48:54 -0800, Guenter Roeck wrote: > > Another key question: Can I remove the vrm attribute entirely from all drivers, > > and/or can I even make it read-only, without going through the feature removal > > process ? It is documented in the ABI, and libsensors supports it. > > It is documented in the ABI as an attribute which can exist, not one > which is mandatory. Thus removing it shouldn't break anything. > > As far as libsensors is concerned, version from lm-sensors 3.x does > _not_ know of VRM. At all. I already did not like the implementation > back when I converted libsensors to make use of the standard sysfs > interface, so I decided to not include it :) > > Version from lm-sensors 2.x does support it, but at least in the latest > version it is printed through a common helper function (print_vid_info) > which will silently skip the VRM value if not readable. So we're safe > here too. > > > Maybe I should only apply the patch to hwmon-vid.c now, and schedule vrm removal > > for something like 2013 ? > > I'm fine both ways, thought 2013, is probably too far away. I really > have no problem with you killing this whole legacy thing right now if > you feel like it, but if you prefer to announce it now and do it in 6 > months, that's up to you really. > I am all with you. So I'll go ahead and just remove the code. Another question: Several drivers read vid only once. That doesn't seem to be correct; I thought the whole point of having vid is to have dynamic voltage to the CPU. Should this be changed as well ? Guenter _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
