15:26 <bamse> kholk[m]: i'm not in favor of skipping the context bank initialization...as it results in the software thinking it has some particular configuration and the hardware having a different one

15:27 <bamse> kholk[m]: and with that in mind i started trying to figure out the exact details of the smmu registers in sdm660 a few months back, but i don't think i reached a conclusion

15:42 <kholk[m]> bamse: but that's not skipping the context bank initialization, or anyway not entirely - it's simply skipping to disable it at iommu init time, but then it is not blocking the rest of the reset and configuration

17:07 <kholk[m]> side note: I've just discovered that Sony Yoshino MSM8998 needs reset-nodisable-cbs for cb12, while the FxTec Pro 1 (8998 again) doesn't... it's just going wild with firmwares... :\

17:07 <kholk[m]> sorry, I mean cb12 on lpass smmu

20:30 <vknecht[m]> konradybcio: thankfully, no need to debug every malloc or dram regulator : fixing pcnoc_a AND reserving 0x86680000 to 0x86681000 was sufficient to get it booting again ;-)

20:51 <konradybcio> Heh, not writing stuff to wrong registers of your dram controller does seem like a good idea 😉

20:52 <konradybcio> Oh wait no i misread

20:52 <konradybcio> Still good that you got it going

21:29 <bamse> kholk[m]: i need to refresh my memory on 660, but there was one of these platforms where the context banks seemed semi-read-only

21:32 <kholk[m]> bamse: yup, let me know if there's anything else that I/we can do for that IOMMU, that's the only blocker on there so it's a pity... anyway, just to be clear, whatever I'm doing in this regard is on 630.. and re-validated on 8998.. :))

21:38 <bamse> kholk[m]: yeah, it seems we have pretty good support for 660 with just this blocker

21:40 <kholk[m]> bamse: definitely... basically, passed that IOMMU issue, we're left with Adreno's rendering issues (all A5xx) and the lack of a battery driver, then it's totally golden... upstream has reached such a great state nowadays ;)

21:41 <bamse> what's the status on the battery side? similar to pm8994/8?

21:42 <bamse> kholk[m]: upstream is really nice, i'm writing you this on my lenovo flex 5g running upstream + some driver fixes

21:42 <kholk[m]> yeah in that regard, as far as I remember (and I don't think that my brain is failing)... SDM630/660 are using the same stuff as 8998

21:42 <bamse> okay good

21:42 <kholk[m]> bamse: I know, I can do the exact same

21:42 <kholk[m]> sc8180x is fun

21:42 <kholk[m]> I wanted to contribute, but then I saw that you were going fast on it, so I've simply left it

21:43 <bamse> :)

21:43 <kholk[m]> and btw, great lovely job

21:44 <bamse> i have a few patches yet to be posted, but it's getting there

21:44 <bamse> think the only big thing left is the usb type-c (pmic_glink) driver

21:44 <bamse> and the dts files

21:44 <kholk[m]> yeah I couldn't resist coding in some fast things, but it runs very nice even without

21:45 <kholk[m]> and yep, I had to write some things in the dts to get it a bit better

21:45 <bamse> i switched from my c630 to the flex 5g was that i managed to get displayport working, so i can use a proper monitor

21:45 <kholk[m]> uhm, DP didn't work for me on Linux

21:45 <bamse> well, i ran with two external monitors for the first week, but wifi crashes every other day

21:46 <kholk[m]> the only puzzling thing that I see here is the unability to get both keyboard and trackpad

21:47 <bamse> hmm, that i don't have any problems with

21:47 <kholk[m]> hm.. perhaps it's just me

21:47 <bamse> well, on c630 we had all kinds of problem with getting the touchpad working on half the devices

21:48 <bamse> so the first c630 i bought works fine, the second one works mostly fine...

21:48 <kholk[m]> yeah, I can imagine... this one though is the first qcom laptop that I ever had