06:16 <naoki> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/soc/rockchip/io-domain.c?id=0536fa6e6fa3e48f4ca11855b586c277be524fbe will this be included in mainline(v.11)? DT part is already in mainline

06:16 <naoki> v6.11

14:31 <qschulz> Kwiboo: if you're still trying to share time off booting on RK3399, I'm wondering if we cannot have apll_l_600_cfg and apll_b_600_cfg use 816MHz instead of 600MHz. The kernel DT seems to indicate it's the same voltage for both.

14:32 <qschulz> I quickly tried, gain of about 160ms

14:33 <qschulz> BTW, since enabling CONFIG_VIDEO (and specifically HDMI support) in U-Boot, double display support in Linux is broken

14:33 <qschulz> double display = HDMI + DSI. HDMI still works, but DSI is broken unless I unplug HDMI

14:34 <qschulz> there's a whole lot of weird things going on, because if I add CLK_SET_RATE_PARENT to dclk_vop*_frac in the kernel, it now works, but only the dclk_vop*_div rate changes, not _div, but it's enough to make it work

14:34 <qschulz> which seems fishy

14:35 <qschulz> so I removed that, and just tried to swap the order in which the DSI and HDMI is probed in rockchip-drm.... and that made it work too

14:35 <qschulz> the side effect being that DSI is now on VOPL while it is on VOPB right now

16:41 <rtp> qschulz: display on rk3399 on uboot mainline ? how is it these days ? iirc, it has a tendancy of not always playing well with linux, breaking the display

19:17 <Kwiboo> qschulz: would be nice to speed up boot time, I have not worked anything more on it, moved on to focus some time on video decoding for now, will probably cricle back to u-boot optimization in a few weeks :-)

19:18 <Kwiboo> I have also avoided trying to use hdmi/video in u-boot, because it typically may cause issues in linux

19:20 <Kwiboo> in https://lore.kernel.org/all/20240615170417.3134517-14-jonas@kwiboo.se/ I also tried to change display component probe order so that hdmi default to use vopl, to help support 4k hdmi use

21:17 <urja> fascinating (my RK3288/veyron HDMI patches just remove the nondesirable VOP-connector links in the DTS)

21:21 <urja> https://github.com/urjaman/linux-for-my-autobuilder/commit/b96f6dbbdb61b112faf0088117212867970b07ef#diff-6c579597fdbc307a1c05a786158ca11c645d8e69596d7f2d1f0ba04f30d5e31aR96

21:21 <urja> (I gave up on trying to mainline this approach of fixing the RK3288 HDMI *sigh*)

22:26 <Kwiboo> and of course I meant change the probe order to help use vopb for hdmi, as stated in the patch, and not vopl

22:28 <Kwiboo> urja: yeah, in libreelec we also use the npll with a custom pll table dedicated for hdmi to best support hdmi modes on rk3288, and a custom pll table for vpll on rk3399, something that is not upstremable

22:32 <Kwiboo> it will probably not be okay to have something like rockchip,hdmi-rates-hz in DT, as that may be seen as configuration for software and not a description of the hardware