[PATCH] ARM: sun8i: Add Parrot Board DTS

On Tue, Jun 21, 2016 at 2:35 AM, Maxime Ripard
[off-list ref] wrote:

On Tue, Jun 21, 2016 at 12:30:25AM +0800, Chen-Yu Tsai wrote:

quoted

quoted

quoted

quoted

quoted

quoted

+&reg_aldo1 {
+       regulator-always-on;
+       regulator-min-microvolt = <3000000>;
+       regulator-max-microvolt = <3000000>;
+       regulator-name = "aldo1";

What is this for exactly? Would turning it off render the system inoperable?
How was it referenced in the fex file?

If this is for WiFi I/O VCC, then you should specify it in mmc1 with
vqmmc-supply.

In the fex, aldo1 is one of the three power inputs for the WiFi (the
others being dldo1 and dldo2) and in the schematics it is linked to
both VCC-USB and VCC-IO-WIFI.

I tried to turn it off and, indeed, the system becomes inoperable.

I'll add vqmmc-supply in mmc1 with aldo1 regulator. However, I am
wondering what to put in vmmc-supply for mmc1 since the WiFi module has
three power inputs: dldo1, dldo2 and aldo1. In the fex, they are
referenced as, respectively, module_power1, module_power2 and
module_power3 and in the schematics dldo1 and dldo2 are named VCC-WIFI
while aldo1 is used for VCC-IO-WIFI (if it can help in any way).

VCC-WIFI is connected to pin VBAT of the Broadcom AP6212 WiFi chip.
VCC-IO-WIFI is connected to pin VDDIO of the chip.

VCC-IO-WIFI is vqmmc, and VCC-WIFI is vmmc.

Isn't vmmc supposed to be always powered at 3.3v, and vqmmc the one
used to drop to 1.8V with the UHS cards?

For eMMC, vmmc and vqmmc correspond to the pins on the eMMC chip.
vmmc provides power to the internals, while vqmmc provides power to the
I/O buffers on both sides. With SD this is a bit less clear. IIUC the SD
card itself supplies I/O voltage, dropping it to 1.8V if necessary, from
VDD (vmmc). vqmmc only affects the host side, and whatever external pull-ups
that might exist.

Ok.

quoted

quoted

In which case, VCC-IO-WIFI is vmmc, and VCC-WIFI is just one of the
wifi chip power rail, to be handled by a power sequence.

For the WiFi chip, there is a separate IO-VCC rail, which is VCC-IO-WIFI.
This, from what I understand of the datasheet, is for I/O signaling, and
thus vqmmc. VCC-WIFI is what powers the chip, which is what vmmc is.

Ok. but there's still the issue of the two regulators that needs to be
kept in sync. Before, that, I'd rather stice to not tying them to the
MMC bus, and putting a comment on top.

OK. Let's do that for now.

ChenYu