00:03 <bryanodonoghue> aka_[m] i wonder if the TPG has received that kind of extensive testing on vfe 480

00:03 <bryanodonoghue> this might not be a problem specific to you at all

00:03 <bryanodonoghue> I'll test it on sm8250 tomorrow and see if I get the same result

11:13 <marc|gonzalez> bryanodonoghue: did you see anything useful in the venus debug output dump?

14:40 <marc|gonzalez> bryanodonoghue: I didn't understand your latest email.

14:41 <bryanodonoghue> the video ahb clock should probably be always-on

14:42 <bryanodonoghue> venus ahb clock

14:42 <marc|gonzalez> That means setting a specific flag somewhere?

14:43 <bryanodonoghue> .flags = CLK_SET_RATE_PARENT | CLK_IS_CRITICAL,

14:43 <bryanodonoghue> but for preference we'd remove the AHB clock description

14:43 <bryanodonoghue> and write the always-on bit per the email

14:44 <marc|gonzalez> Are you going to send a patch for the sm8250?

14:44 <bryanodonoghue> no that's already upstream

14:45 <marc|gonzalez> oooh, you meant "you can look at sm8250 to do the same for 8998" now I get it.

14:45 <marc|gonzalez> I will be brutally honest: my experience sending short patches has been awful, and I hesitate even to send one-liners.

15:15 <bryanodonoghue> meh fuck the begrudgers marc|gonzalez

15:37 <arnd> bamse: you sent multiple patches that mark DT nodes as dma-coherent for 6.10. Are you sure those shouldn't be in v6.9 as well? I would expect that a missing dma-coherent annotation causes data corruption when data gets put into the cache but gets flushed by the CPU

15:41 <arnd> mani_s: I'm confused by all your "arm64: dts: qcom: xxx: Add PCIe bridge node" patches. What issue does this address?

15:42 <arnd> I would expect that any PCIe controller has a bridge behind it and that we probe that automatically

15:43 <mani_s> arnd, We need to model PCI bridges for multiple reasons:

15:43 <mani_s> arnd, 1. Bridge specific properties are going to be added in the future. Like PERST#, WAKE# etc.. currently these are added in controller node, but that's not correct

15:44 <mani_s> arnd, 2. We need bridge to add child nodes. Like when we use the ongoing 'power sequencing' series

15:45 <arnd> right, that makes more sense then. I was mainly confused because you don't add any properties or children that make any difference at the moment

15:45 <arnd> thanks

15:45 <mani_s> arnd, np. I mentioned it in the cover letter, but that got missed

16:03 <marc|gonzalez> bryanodonoghue: did you have a chance to look at the venus debug output? Did you see anything useful?

20:44 <bryanodonoghue> marcgonzalez honestly you should ask dikshita - maybe ping her direclty

20:44 <bryanodonoghue> she can probably tell you straight out or even read the fw sources if not