VC4 GPIO support possibly broken

Michael Zoran mzoran at crowfest.net
Sat Mar 18 22:17:42 PDT 2017


Hi Eric and Stefan,

I did some more testing of the driver I submitted to staging, and even
though the mailbox API is functional, I'm finding that the VC4 driver
doesn't appear to be doing the right thing with GPIO handling.

The DT I submitted does make VC4 functional on the RPI 3, but I added
some debugging prints to the driver I submitted and VC4 isn't even
doing any GPIO queries at all.  

With the DT I submited, I'm seeing that when VC4 looks up the GPIO
number from the DT it's getting back 0 which doesn't make sense.  And
if the GPIO is 0, VC4 is skipping calls to the GPIO.

I also noticed if the GPIO is marked as active-low, VC4 also appears to
be skipping GPIO calls.

Please not that the mailbox API is functional.  I did some manual calls
into the mailbox API, and 132 is really working as a HDMI hotplug
detection.  I plug in HDMI, I get 1 back.  I disconnect HDMI and I get
0 back.  So the actual polarity should be GPIO_ACTIVE_HIGH.





More information about the linux-rpi-kernel mailing list