On Wed, Feb 1, 2017 at 2:20 AM, Andrew Jeffery [off-list ref] wrote:

On Wed, 2017-02-01 at 11:22 +1030, Andrew Jeffery wrote:

quoted

On Tue, 2017-01-31 at 15:50 +0100, Linus Walleij wrote:

quoted

quoted

quoted

quoted

quoted

On Fri, Jan 27, 2017 at 5:24 AM, Andrew Jeffery [off-list ref] wrote:

This is less straight-forward than one would hope, as some banks only
have 4 pins rather than 8, others are output only, yet more (W and
X, already supported) are input-only, and in the case of the g4 SoC bank
AC doesn't exist.

Add some structs to describe the varying properties of different banks
and integrate mechanisms to deny requests for unsupported
configurations.

quoted

quoted

Signed-off-by: Andrew Jeffery <redacted>

---

Since v2:

Patch applied with some patch -p1 < fuzz
please check the result.

*snip*

quoted

Regardless, I'll try to recreate it myself and inspect the fuzz damage.

Ah, I think I see what's happened. I didn't send "gpio: aspeed: Make
bank names strings" again because you claimed you had applied v1 in the
v2 thread[1]. This patch doesn't fuzz when "gpio: aspeed: Make bank
names strings" has been applied, so is it possible that you tried
applying it to a tree missing "gpio: aspeed: Make bank names strings"?

I mistakedly applied *both* patches to the pinctrl tree, devel branch.

I blame stress.

Oh well I guess I have to keep them there and remove this copy
from the gpio tree.

If the version I applied in the pinctrl tree is the wrong one (which
is possible) then please send an incremental patch on top fixing
the difference, so I can apply that to the pinctrl tree too....

Yours,
Linus Walleij