08:11 <z3ntu> Hi, I'm trying to bring up venus on sm6350 SoC. I get the remoteproc (if that's the correct word here) booted as far as I can tell but the "Check for ARM9 and video core to be idle" condition fails consistently in the driver

08:11 <z3ntu> [ 117.323593] qcom-venus aa00000.video-codec: VenusLow : venus hw version 6.40.0

08:11 <z3ntu> [ 117.323868] qcom-venus aa00000.video-codec: VenusLow : F/W version: 14:VIDEO.VPU.1.2-00038-PROD-1

08:12 <z3ntu> So the readx_poll_timeout(venus_cpu_and_video_core_idle, ...) fails, the register values are cpu_status=0 and ctrl_status=1 but I'm really unsure what that means

08:36 <z3ntu> In downstream (function __prepare_pc_iris2) first it's wfi_status=0 ctrl_status=40000001 and then becomes wfi_status=1 ctrl_status=101 (wfi_status=cpu_status)...

23:07 <lumag_> abhinav___, sboyd_ any chance of a quick review of patches 1-3 from https://patchwork.freedesktop.org/series/98585/? The patches were tested by Kuogee. Patches 4,5 are out of quiestion for now.

23:07 <lumag_> asking here as sboyd_ is not logged on #freedreno

23:08 <sboyd_> I think I got booted from there

23:08 <Mis012[m]> don't think displayport has much to do with freedreno?

23:08 <lumag_> sboyd_, wow?

23:09 <lumag_> the same question for https://patchwork.freedesktop.org/patch/474922/?series=100366&rev=1

23:09 <Mis012[m]> this is a sane-ish platfrom in that sense :P

23:17 <lumag_> Mis012[m], why?

23:20 <Mis012[m]> well, with PCI-E graphics cards, you have something that is almost comparable to a full SoC's peripheral package and which would in a sensible world need a device tree and way more than just one driver

23:20 <Mis012[m]> with arm SoCs, the GPU and the display pipeline are not really that loosely connected as far as I'm aware

23:22 <Mis012[m]> you can pretty much ignore the GPU's existence if you don't care about HW acceleration

23:23 <Mis012[m]> on x86, you would have to use something like a usb->VGA adapter to get that

23:23 <Mis012[m]> well, standard x86

23:24 <Mis012[m]> and then you could use PRIME for acceleration

23:25 <Mis012[m]> obviously most ARM SoCs probably have a way to do zero copy

23:30 <lumag_> Mis012[m], yes, the GPU has a bit loose connection to the rest of the display pipeline.

23:30 <lumag_> However the channel is used for both GPU and DPU discussions

23:35 <Mis012[m]> I see