Re: [PATCH 12/13] ARM: dts: stm32: fix DSI port node on STM32MP15

From: Alexandre TORGUE
Date: Mon Apr 19 2021 - 10:04:34 EST

On 4/19/21 3:57 PM, Arnd Bergmann wrote:
On Thu, Apr 15, 2021 at 2:23 PM Alexandre TORGUE
<alexandre.torgue@xxxxxxxxxxx> wrote:
On 4/15/21 12:43 PM, Ahmad Fatoum wrote:
On 15.04.21 12:10, Alexandre Torgue wrote:
Running "make dtbs_check W=1", some warnings are reported concerning
DSI. This patch reorder DSI nodes to avoid:

soc/dsi@5a000000: unnecessary #address-cells/#size-cells without
"ranges" or child "reg" property

This reverts parts of commit 9c32f980d9 ("ARM: dts: stm32: preset
stm32mp15x video #address- and #size-cells"):

The cell count for address and size is defined by the binding and not
something a board would change. Avoid each board adding this
boilerplate by having the cell size specification in the SoC DTSI.

The DSI can have child nodes with a unit address (e.g. a panel) and ones
without (ports { } container). ports is described in the dtsi, panels are
described in the dts if available.

Apparently, the checker is fine with
ports {
#address-cells = <1>;
#size-cells = <0>;

I think my rationale for the patch above was sound, so I think the checker
taking offense at the DSI cells here should be considered a false positive.

If it's a "false positive" warning then we need to find a way to not
print it out. Else, it'll be difficult to distinguish which warnings are
"normal" and which are not. This question could also be applied to patch[3].

Arnd, Rob what is your feeling about this case ?

I don't have a strong opinion on this either way, but I would just
not apply this one for 5.13 in this case. Rob, Alexandre, please
let me know if I should apply the other patches before the
merge window, I usually don't mind taking bugfixes late before the
merge window, but I still want some level of confidence that they
are actually correct.

For me, we can keep this series for the v5.14 cycle.


Ahmad, if you feel strongly about this particular issue, would you like
to suggest a patch for the checker?