16:00 <macromorgan> so the datasheet says the rk3566 supports 2 independent displays, and the rk3568 supports 3. Does that mean the vp0 (or 1, or 2) is not present on the rk3566?

16:00 <macromorgan> (I'm getting a null pointer exception with the VOP2 driver when I try to use vp0 on my rk3566)

16:05 <macromorgan> I mean null pointer dereference

18:19 <jakllsch> macromorgan: quite possibly

18:19 <jakllsch> macromorgan: but it'd help to know which pointer is null

18:20 <macromorgan> not sure yet, looking now

18:20 <jakllsch> if there are seperate TRMs for each chip, you might compare

18:27 <macromorgan> there don't appear to be, the TRM I have says rk3566/rk3568

18:28 <macromorgan> there's like 2 or 3 notes total though that detail the differences, and I don't think that's all

18:34 <CounterPillow> yeah the RK3566/3568 TRM is one and the same unfortunately

20:08 <macromorgan> jakllsch: with some debug printks in place it looks like it's taking a shit at drm_crtc_init_with_planes for the 2nd port (port 1 if 0 and 1 are enabled, 2 if 1 and 2 are enabled, etc). Specifically, it looks like plane may be the NULL culprit

20:11 <macromorgan> still sad my screen looks wonky, but I'm starting to wonder if it's actually the VOP2 and not the dsi or dphy or panel driver...

20:12 <macromorgan> I also get some really weird format errors too

21:37 <macromorgan> okay, think I figured out the NPE in the VOP2 driver when you set one port to one value and another port to a different value. tl;dr, it's because there's only 1 primary plane to be shared among multiple ports.

21:38 <macromorgan> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c#n2284 (right here it iterates through the wins)

21:39 <macromorgan> when it comes to a win that is primary, it assigns it to a port that doesn't have primary win

21:39 <macromorgan> problem is, for an rk3566 there is only 1 primary (since the switch statement excludes 3 wins, one of which is the other primary)

21:41 <macromorgan> so the first port gets a primary win, additional ports get nothing, and when you get to the drm_crtc_init_with_planes it takes a crap because the plane is null for the second port on an rk3566 (or null for the 3rd port on an rk3568)

21:41 <macromorgan> don't think this came up in testing because the device tested was an rk3568 evaluation board