> -----Original Message----- > From: Greg KH [mailto:gre...@linuxfoundation.org] > Sent: Thursday, December 20, 2012 10:09 PM > To: R, Durgadoss > Cc: Zhang, Rui; linux...@vger.kernel.org; linux-kernel@vger.kernel.org; > hongbo.zh...@linaro.org; w...@nvidia.com > Subject: Re: [PATCH 4/8] Thermal: Add Thermal_trip sysfs node > > On Thu, Dec 20, 2012 at 04:25:32PM +0000, R, Durgadoss wrote: > > > > > -----Original Message----- > > > From: Greg KH [mailto:gre...@linuxfoundation.org] > > > Sent: Thursday, December 20, 2012 9:42 PM > > > To: R, Durgadoss > > > Cc: Zhang, Rui; linux...@vger.kernel.org; linux-kernel@vger.kernel.org; > > > hongbo.zh...@linaro.org; w...@nvidia.com > > > Subject: Re: [PATCH 4/8] Thermal: Add Thermal_trip sysfs node > > > > > > On Thu, Dec 20, 2012 at 07:52:03AM +0000, R, Durgadoss wrote: > > > > > On Tue, Dec 18, 2012 at 02:59:33PM +0530, Durgadoss R wrote: > > > > > > This patch adds a thermal_trip directory under > > > > > > /sys/class/thermal/zoneX. This directory contains > > > > > > the trip point values for sensors bound to this > > > > > > zone. > > > > > > > > > > Eeek, you just broke userspace tools that now can no longer see > these > > > > > entries :( > > > > > > > > > > Why do you need to create a subdirectory? As you found out, doing > so > > > > > isn't the easiest, right? That is on purpose. > > > > > > > > Yes, I observed the complexity. > > > > > > > > > > > > > > I really wouldn't recommend doing this at all, please stick within the > > > > > 'struct device' framework here, don't create new kobjects and hang > sysfs > > > > > files off of them. > > > > > > > > But, we cannot put all _trip directly under ZoneX directory. > > > > > > Why not? What is preventing this? > > > > > > > We can remove the thermal_trip directory, and put sensorY_trip under > > > > /sys/class/thermal/zoneX/. But this sensorY_trip needs to be a > > > > directory which has four sysfs nodes named, active, passive, crit, > > > > hot. > > > > > > > > Rui, What do you think about this ? > > > > > > > > The only other way I see, is directly put > > > sensorY_trip_[active/passive/hot/crit] > > > > which will create way too many nodes, under > /sys/class/thermal/zoneX/. > > > > > > What is "too many"? 20000? 50000? How many are we talking about > here? > > > > Not in 1000's though.. > > > > > What is the limiting factor that is preventing this from all going into > > > one directory? > > > > We support a MAX of 12 sensors per zone today, which will lead to > > 12 * 4, 48 nodes under this directory named > > sensorY_trip_[active/passive/hot/crit], besides the other nodes. > > That's fine, we can easily support that many files, have you tried this > already?
Yes, in fact, this is sort of what was the old implementation.. although with different sysfs nodes. > > The main point is, if you use a kobject like you are, userspace tools > can't "see" these directories and files easily, if at all. Try it out > with libudev yourself to verify it, the attributes will not show up as > owned to that device like you need them to be. I haven't used libudev exactly, but I realized this sort of thing, when I was trying to catch UEvents on this device path. I will give libudev a try.. > > So put them all in one directory, we can handle 10's of thousands of > files quite easily, so 48 is trivial :) Okay, Will make it this way :-) Now I can see the implementation getting much simpler !! Thank you Greg, Durga P.S: I should thank you for this file(samples/kobject/kobject-example.c) also, from where I got how to get this implementation done :-) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
