On 05/21/2015 01:10 PM, Matthew Garrett wrote:
On Wed, May 20, 2015 at 6:24 PM, Rafael J. Wysocki rjw@rjwysocki.net wrote:
Which doesn't explain why we need a config option per quirk. To me, such config options don't add any value, because (a) everyone will set them anyway and (b) removing the quirks from the source is trivial if needed.
We'd disable this quirk in Fedora the moment jack detection works, because we've got the userspace to handle it and using I2S is preferable to using HDA - but in doing so we might break battery detection on the other Dell that's playing _REV tricks. This seems like a suboptimal choice to have to make.
Having dug into this deeper with you on the other thread, battery detection already wasn't working. That machine did the _REV test and fix for Linux only on Windows 2009 _OSI and the _REV value of 5. The kernel currently responds to a later _OSI that the machine supports so that AML does not get run.
I believe battery detection may be fixed on that Inspiron by 75646e758a0ecbed5024454507d5be5b9ea9dcbf. if it's not, that's a tangential problem that should be fixed to match the Windows behavior for battery detection.