Comment 16 for bug 1313196

Hi Christopher, I will try to do my best.

First of all, what about my considerations on the acpi_osi kernel boot parameter?

Secondly, I did a "binary search" of the first mainline kernel version which works. The result is not so linear. I'll explain.

Version 3.11.2 (package name mentions 3.13.11-03131102.201405081435) still has the problem

From 3.14-rc1 onwards, the problem is fixed. However... with 3.14-rc1, 3.14-rc2 and 3.14-rc3 if I type:
sudo rfkill list
I see an output which is similar to what I get if I specify the acpi_osi boot parameter with 3.11 kernel, that is I get the hp-wifi and hp-bluetooth entries and the switch applies a hard block rather than a soft block. The only difference is that the hci0 does not disappear when the wifi card is disabled.
From 3.14-rc4 onwards, the problem is still fixed, but if I type rfkill list, I see the output that I get with 3.15-rc2, that is without the hp-* entries. Also, the wifi switch button produces a soft block instead of a hard block.

This makes me think that there are at least two commits (one for 3.14-rc1 and one for 3.14-rc4) that in some way have effects on this.
Any suggestion?