Hi Diego,
On 2024-02-11 at 14:19:49 +0100, Frank Oltmanns <[email protected]> wrote: > Hi Diego, > > On 2024-02-09 at 16:17:13 +0100, Diego Roversi <[email protected]> wrote: >> Thanks for advices. I found that the bug was introduced in the kernel >> pre-6.5 after the rc1 with this patch: >> >> https://lore.kernel.org/all/[email protected]/ >> >> For newer kernel the workaround you suggested, disabling >> CLK_SET_RATE_PARENT, works fine. >> > > What you are saying means that in 6.5 the Teres-I doesn't work with > pll-mipi, but with 6.6 + disabling CLK_SET_RATE_PARENT it does. That is > quite confusing for me. I should try to get one of those on my desk for > testing. > > Thank you for your analysis! > > Best regards, > Frank Ok, as I wrote in IRC this is my current hypothesis: Probably in pre-6.5 the SoC was able to closely fulfill the panels request for a specific rate (because tcon was using pll-video0-2x as it's parent). 6.5 only allowed pll-mipi as the parent and pll-mipi was probably not able to give the rate as closely as requested by the panel. In 6.6 I added support for pll-mipi to set a closer rate by a) overshooting and b) setting the parent rate. It seems that allowing to overshoot fixes the issues on teres-I but the fact that it can set the parent rate, breaks it. Could you please send the content of /sys/kernel/debug/clk/clk_summary for - 6.4 or earlier, - 6.5 (assuming you can log in via ssh/serial console) - 6.6 (same) I noticed that the anx6345 is only used in the pinebook and Teres-I which are both based on the A64. So, maybe there is something in that driver that fixes the broken calculation of the tcon-pixel-clock that was fixed in 6.5 as well. Thank you and best regards, Frank -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web, visit https://groups.google.com/d/msgid/linux-sunxi/87jzn9yebl.fsf%40oltmanns.dev.
