Description of problem: kpowersave shows two batteries while my laptop has one only. The cause seems to be that HAL reports two batteries present in my laptop even if there's only one. Version-Release number of selected component (if applicable): hal-0.5.10-1.fc8 How reproducible: always Steps to Reproduce: 1. start kpowersave Actual results: It reports two batteries Expected results: It should report only one battery Additional info: The output of lshal shows that there are two UDI's with info.capabilities = { 'battery' }: udi = '/org/freedesktop/Hal/devices/acpi_BAT0' udi = '/org/freedesktop/Hal/devices/computer_power_supply_0' the first one is the correct one. A temporary fix for the problem is to clear info.capabilities for the 2nd entry: hal-set-property --udi '/org/freedesktop/Hal/devices/computer_power_supply_0' --key info.capabilities --strlist-rem 'battery'
Created attachment 252951 [details] Output of lshal
Any chance we can get this fixed? This is basically happening on all laptops. Looks like a fix has been committed upstream.
Confirmed on Dell Latitude D600 (i386), showing 4 batteries with 2 installed. Severe impact is that the bogus entries (which do not show model or serial) appear to report premature discharge, causing gnome-power-manager to fire shutdown while actual batteries have plenty of charge left. Forces use of non-Linux OS for untethered operation. The freedesktop.org external bug referenced was closed as fixed in early March (and is for kernel-2.6.25.9-40.fc8 hal-0.5.10-3.fc8 just hit repos, does not fix. hal-set-properties temporary fix noted in bug report does not fix.
Same here with a Dell 630. The problem appeared with kernel 2.6.25.4-10.fc8 which is the first with CONFIG_ACPI_SYSFS_POWER=y. The previous kernel, 2.6.24.7-92.fc8, has this option unset and works as expected. I tried to recompile a hal with the patch found in http://lists.freedesktop.org/archives/hal-commit/2008-January/004133.html and indeed there was only one battery visible in hal and kpowermanager. However, kpowermanager was still confused and it would only update the charge level of the battery when a plug or an unplug event occured, and it was unable to compute or display the time remaining. To me, it looks like the simplest fix is to unset the option CONFIG_ACPI_SYSFS_POWER in the kernel.
I suspect the core issue is the same as with Bug 454954, which also began occurring with the first 2.6.25 kernel.
Please try the rpms I scratch built here with CONFIG_ACPI_SYSFS_POWER turned off and let me know if it fixes the issues you've seen: http://koji.fedoraproject.org/koji/taskinfo?taskID=710780 cheers, Kyle
Yes, the kernel in kernel-2.6.25.10-91.fc9.i686.rpm does fix for me the issue with kpowermanager seeing two batteries; thanks for the rpm. (Side note: I have fedora 8, your kernel is for fedora 9. While installing it, rpm complained about mkinitrd not recent enough. Installing with --nodeps, there was an error in the %posttrans scriplet. But it booted anyway and seems to be working very nicely.)
Ok, a F8 kernel is available here if that will help, but I guess it sorts out your issue, and that the other bug might be unrelated. http://koji.fedoraproject.org/koji/taskinfo?taskID=714370 cheers, Kyle
The build in task 714370 cures the two battery bug, and also breaks resume on the Dell D6x0 due to inclusion of a 20-video-quirk-pm-dell.fdi with <merge key="power_management.quirk.radeon_off" type="bool">true</merge> missing under <match key="system.hardware.product" contains_outof="D600;D610;D630"> This broken fdi began shipping with kernels at the same time that CONFIG_ACPI_SYSFS_POWER was turned on. Fortunately I'd already tracked it down previously and knew how to fix it. It's been reported to hal.org. This kernel unfortunately and surprisingly does not fix bug 454954, which also began at the time CONFIG_ACPI_SYSFS_POWER was turned on.
Hi, I am happy with the FC9 kernel, and I won't try your FC8 kernel unless you think it is necessary. Do you ? Thanks for fixing this bug.
I'm pretty sure that this has been fixed now. After updating to kernel-2.6.25.11-60.fc8 I found that the phantom battery has disappeared.
The problem is back in 2.6.26.3-14.fc8. This just bites. Also see bug 452356.
Sorry, looks like the F9 kernel config got copied over the F8 one during an update, and this got turned back on. Fixed again. Really, should just fix Hal, but by the time that happens, F10 will be out and F8 will be gone. regards, Kyle
And with any luck F10 won't have a broken network installer (preupgrade) and will have some of the missing stuff put back in (e.g. gdmsetup). F9 has not been viable for us for those reasons. By the time hal is fixed across the board, we'll have peace on earth. I just had to patch my local 20-video quirk again.
And hooray, the fix finally made it to the F8 repo. Fixed in 2.6.26.5-28.fc8.