On Thu, Aug 20, 2020 at 1:23 AM Mauro Carvalho Chehab mchehab+huawei@kernel.org wrote:
(added c/c Rob Herring)
Em Wed, 19 Aug 2020 18:53:06 -0700 John Stultz john.stultz@linaro.org escreveu:
On Wed, Aug 19, 2020 at 4:46 AM Mauro Carvalho Chehab mchehab+huawei@kernel.org wrote:
@@ -376,7 +355,7 @@ static int kirin_drm_platform_resume(struct platform_device *pdev) }
static const struct of_device_id kirin_drm_dt_ids[] = {
{ .compatible = "hisilicon,hi3660-dpe",
{ .compatible = "hisilicon,kirin960-dpe",One issue, elsewhere in your patch stack you still refer to the hisilicon,hi3660-dpe compatible string. This should probably be consistent one way or the other.
Agreed with regards to consistency.
It sounds to me that calling those as Kirin 9xx (and the previous one as Kirin 620) is better than using the part number.
Here, googling for "Kirin 970" gave about 6.9 million hits, while "Hi3670" gave only 75,5K hits.
Kirin 620 has similar results: 6.85 million hits, against 61,9 hits for "Hi3620".
Hi6620 is kirin 620 I believe.
With "Kirin 960", the numbers are a lot higher: had 21,4 million hits, against 423K hits for "Hi3660".
So, my preference is to use "Kirin 620, 960 and 970" for future changes.
I think traditionally the DTS is developed with the hardware documentation sometimes before the SoC is announced, so they tend to stick with whatever those documents call it, rather than (later more google-able) marketing names.
That said, I don't have a preference, as long as it's consistent, and we don't change compatible strings that are already upstream.
I would love to make this consistent among the Kernel. However, I'm not sure if changing "compatible" would be acceptable by DT maintainers.
Existing bindings are already ABI. So we can't change those. New bindings can be set to what makes the most sense.
thanks -john