Incorrect power supply type detection in upower
Hello, on Google Nexus 5, there is a Linux kernel driver used to change charging current according to battery's temperature (too hot -> lower the current)[1]. This driver exports as type "BMS" (battery management system, not present in vanilla kernel) in sysfs, and because upower considers any unknown device which doesn't have an "online" file in sysfs a battery, it thinks the device has two batteries. This causes the GUI to report wrong battery percentage, because it adds values from both "batteries" (one of which is always 0) and divides them by two. The point is I don't think defaulting to battery is a good idea. But I don't know if it is okay not to - will it break stuff like UPS if upower won't default to battery? What if we check for presence of "capacity" file in sysfs (which that temperature driver doesn't have), is that okay? Basically, I'd like to get this properly fixed, but I don't know enough about power supply devices to post a patch, so I'd be glad for any input on this issue. Thank you, Vojtěch Boček PS: This is a re-post of the same e-mail, because I wasn't registered when I sent the first one and it doesn't seem to have made it to the mailing list. [1]: https://android.googlesource.com/kernel/msm/+/android-msm-hammerhead-3.4-kitkat-mr1/drivers/power/battery_temp_ctrl.c ___ devkit-devel mailing list devkit-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/devkit-devel
Re: Incorrect power supply type detection in upower
On Wed, 2014-06-04 at 22:34 +0200, Vojtech Bocek wrote: > Hello, > on Google Nexus 5, there is a Linux kernel driver used to change > charging current according to battery's temperature (too hot -> lower > the current)[1]. This driver exports as type "BMS" (battery management > system, not present in vanilla kernel) in sysfs, and because upower > considers any unknown device which doesn't have an "online" file in > sysfs a battery, it thinks the device has two batteries. This causes > the GUI to report wrong battery percentage, because it adds values > from both "batteries" (one of which is always 0) and divides them by > two. > > The point is I don't think defaulting to battery is a good idea. But I > don't know if it is okay not to - will it break stuff like UPS if > upower won't default to battery? What if we check for presence of > "capacity" file in sysfs (which that temperature driver doesn't have), > is that okay? > > Basically, I'd like to get this properly fixed, but I don't know > enough about power supply devices to post a patch, so I'd be glad for > any input on this issue. Simply fix the driver to export a "type". That's the way you'd fix it for any of the in-kernel drivers. ___ devkit-devel mailing list devkit-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/devkit-devel
Re: Incorrect power supply type detection in upower
The driver exports type. It reads "BMS" (Battery Management System). But upower doesn't know that type (it isn't in vanilla kernel) and while detecting the power supply type[1], it defaults to battery for any unknown power supply device type[2]. [1]: http://cgit.freedesktop.org/upower/tree/src/linux/up-device-supply.c#n955 [2]: http://cgit.freedesktop.org/upower/tree/src/linux/up-device-supply.c#n1019 2014-06-05 0:42 GMT+02:00 Bastien Nocera : > On Wed, 2014-06-04 at 22:34 +0200, Vojtech Bocek wrote: >> Hello, >> on Google Nexus 5, there is a Linux kernel driver used to change >> charging current according to battery's temperature (too hot -> lower >> the current)[1]. This driver exports as type "BMS" (battery management >> system, not present in vanilla kernel) in sysfs, and because upower >> considers any unknown device which doesn't have an "online" file in >> sysfs a battery, it thinks the device has two batteries. This causes >> the GUI to report wrong battery percentage, because it adds values >> from both "batteries" (one of which is always 0) and divides them by >> two. >> >> The point is I don't think defaulting to battery is a good idea. But I >> don't know if it is okay not to - will it break stuff like UPS if >> upower won't default to battery? What if we check for presence of >> "capacity" file in sysfs (which that temperature driver doesn't have), >> is that okay? >> >> Basically, I'd like to get this properly fixed, but I don't know >> enough about power supply devices to post a patch, so I'd be glad for >> any input on this issue. > > Simply fix the driver to export a "type". That's the way you'd fix it > for any of the in-kernel drivers. > ___ devkit-devel mailing list devkit-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/devkit-devel
