Over the weekend a new mailing list thread began that was entitled "Massive power regression going 3.4->3.5" pertaining to a power problem in this most recent Linux kernel release. This just wasn't a random user complaining of a "massive power regression" but James Bottomley, a Linux kernel developer veteran.
With Bottomley's Lenovo ThinkPad X220i, the 3.5 vanilla kernel is burning through much more power than Linux 3.4. Bottomley tested some of the 3.5-merged commits from power management expert Rafael J. Wysocki to no avail. The Linux 3.4 kernel meanwhile had improved the power management situation compared to Linux 3.3.
Bottomley did some bisecting of the Linux 3.5 kernel and then narrowed down the Linux 3.5 kernel power regression as happening within the DRM sub-system. Unfortunately he can't nail the specific DRM commit causing a power regression due to Git branch issues. The ThinkPad X220i laptop that Bottomley is using sports Intel HD graphics.
The mailing list discussion so far pertaining to this power regression in Linux 3.5 can be found on the mailing list.
Time to fire up some systems and power meters myself to see if the Phoronix Test Suite can nail this power problem on the latest stable Linux kernel.