03:53 <lvrp16> xdarklight: spectek is a micron subsidiary. they use micron's low bin tiers, remanufactured chips stripped from dead ddr sticks, and ett ram from the open market. i would not think of it as a micron product because those are new chips from good bin tiers and usually go to crucial.

10:04 <xdarklight> lvrp16: that does sound pretty questionable indeed

16:01 <narmstrong> Quite common, very used in the cheap USB keys and sdcards ;-)

22:08 <xdarklight> mjourdan: have you considered using genpd for controlling AO_RTI_GEN_PWR_SLEEP0 and AO_RTI_GEN_PWR_ISO0 instead of doing it in the video decoder driver by controlling the AO syscon?

22:08 <xdarklight> I'm wondering because I was curious what it takes to get the video decoder running on the 32-bit SoCs and I discovered that there's no AO syscon on the 32-bit SoCs

22:09 <xdarklight> (and I don't want to add it because there's no overlapping memory region with an AO clock controller)

22:10 <xdarklight> a separate AO_RTI_GEN_PWR_* power controller driver would work on all SoC generations - apart from the initialization sequence: GX SoCs take the parent syscon, 32-bit SoCs have to create a regmap after ioremap'ing the registers

22:12 <xdarklight> (meson-gx-pwrc-vpu would also be converted to take the AO_RTI power domains from that new driver, which would also allow me to re-use the VPU power domain driver on the 32-bit SoCs)