On Wed, 22 Jan 2014 16:17:42 -0500 (EST)
Alan Stern [off-list ref] wrote:
On Wed, 22 Jan 2014, Jonas Gorski wrote:
quoted
If it's safe to set ehci->big_endian_{desc,mmio} from the _probe()
routine, then maybe the pdata sets in _reset() should be moved into here
instead of adding extra cludges/checks into _reset().
Why? What difference would it make?
Effectivewise none, but to me it seems to be cleaner to set them once in
probe() instead of everytime reset() is called.
I admit I don't know the code flow good enough if reset() is called
more than once in the lifetime of a hcd device.
And as I said, it would allow doing the checks the patch adds for both
DT and !DT, not just DT only.
Regards
Jonas