<M0xCAFEBABE[m]>
btw I've sold that device to barni and he's likely gonna bring it up sooner or later
<M0xCAFEBABE[m]>
it's quite similar to a currently supported device - Redmi 5A
<Wei[m]>
<barni2000[m]> "Which regions?" <- Physical memory regions and where they are mapped to.
<barni2000[m]>
Decompile your downstream dt
<barni2000[m]>
You will find the regions in it
<barni2000[m]>
Mostly for mainline you only need to reserv qseecom and fb region
<Wei[m]>
barni2000[m]: Dt = device tree?
<barni2000[m]>
Y
<barni2000[m]>
adb pull /sys/firmware/fdt
<Wei[m]>
Thanks
<fish4terrisa[m]>
On my xiaomi ugg things is a bit odd. The gpu is actually running(by running glmark2-drm, showed that freedreno is used and it has much better performance than llvmpipe, so gpu is working), but the screen is completely black, with no backlight. From boot after bootloader the screen seems just, well, completely dead. Does anyone has any ideas? Also, cannot find some drm related errors in dmesg except
<fish4terrisa[m]>
`[drm:mdp5_irq_error_handler [msm]] *ERROR* errors: 04000000` but seems that this error is very common and didnt show the cause
<fish4terrisa[m]>
(Also the screen is working fine with twrp so it is physically alive, and I included the gpu firmware in even the initramfs)
<fish4terrisa[m]>
<barni2000[m]> "backlight is flaky on ugg a..." <- I think it's not only the backlight is not working? seems that the screen is just not working. Or the screen should be really dark but it should at least show something?
<barni2000[m]>
[drm:mdp5_irq_error_handler [msm]] *ERROR* errors: 04000000 this is come from mdp5 and not a real issue
<barni2000[m]>
every device have it
<barni2000[m]>
i a mean every mdo5 qcom soc device
<barni2000[m]>
s/a//, s/mdo5/mdp5/
<barni2000[m]>
which display do you have?
<fish4terrisa[m]>
stock one
<barni2000[m]>
i need the type
<barni2000[m]>
lk2nd and downstream cmdline tells you
<fish4terrisa[m]>
barni2000[m]: yeah in lsmod i think it is there
<barni2000[m]>
wrong panel driver is loaded
<fish4terrisa[m]>
oh should I block and remove it
<barni2000[m]>
no i am wrong
<barni2000[m]>
all are loaded
<barni2000[m]>
i have sc variant
<fish4terrisa[m]>
so... is the driver missing? Or is it not functioning
<barni2000[m]>
idk
<barni2000[m]>
yet
<barni2000[m]>
i have tm variant also and it is working
<fish4terrisa[m]>
does this driver need any firmware or other config?
<barni2000[m]>
for fb no, for gpu yes
<fish4terrisa[m]>
the strangest thing is that gpu is working...
<barni2000[m]>
comment out the backlight gpio part from the driver
<fish4terrisa[m]>
let me try it
<fish4terrisa[m]>
also the touchscreen seems also dead. libinput list-devices didnt show any devices other than keyboards, so probably touchscreen is also not detected
<fish4terrisa[m]>
s/also//
<barni2000[m]>
Touch screen could be goodix
<barni2000[m]>
Only edt is enabled atm
<fish4terrisa[m]>
maybe lk2nd is not correctly configed? like the dtb isnt currectly appended? in wiki it wrote that lk2nd is needed or it'll be black screen and the kernel cannot select display panel, maybe in my case it's the cause.
<barni2000[m]>
you can verify panel selection with # dtc /sys/firmware/fdt | grep panel
<barni2000[m]>
at mainline
<barni2000[m]>
panel selection is working on ugglite and same selection was copied to ugg so it should work
<barni2000[m]>
<barni2000[m]> "you can verify panel selection..." <- you should see something like this `compatible = "xiaomi,ugglite-ili9881c-sc\0xiaomi,ugg-panel";`
<barni2000[m]>
it seems phosh finally fixed on systemd
<fish4terrisa[m]>
<barni2000[m]> "comment out the backlight gpio..." <- Is there a patch about it? I'm not sure about which part to comment
<fish4terrisa[m]>
However, dmesg seems cleaner. I'll give you my dmesg output:
<barni2000[m]>
[ 1.254684] msm_mdp 1a01000.display-controller: [drm:adreno_request_fw [msm]] loaded qcom/a530_pm4.fw from new location
<barni2000[m]>
[ 1.254974] msm_mdp 1a01000.display-controller: [drm:adreno_request_fw [msm]] loaded qcom/a530_pfp.fw from new location
<barni2000[m]>
[ 1.255647] msm_mdp 1a01000.display-controller: Zap shader not enabled - using SECVID_TRUST_CNTL instead
<barni2000[m]>
its up and running
<fish4terrisa[m]>
but no output
<fish4terrisa[m]>
no backlight
<fish4terrisa[m]>
do you have a similiar device?
<barni2000[m]>
yes
<barni2000[m]>
btw 1 of my tm panel is not working
<fish4terrisa[m]>
is there a way to power off and on the screen?
<barni2000[m]>
but all my ugglite have replacement tm panel
<barni2000[m]>
fish4terrisa[m]: yes with fbdebug
<ValPackett[m]>
<barni2000[m]> "8917 can wake up from sleep with..." <- pretty sure it would completely freeze on boot for me if i added that state, let me try again
<fish4terrisa[m]>
edt seems cannot probe the touchscreen
<fish4terrisa[m]>
get `edt_ft5x06: 1-0038: touchscreen probe failed`
<fish4terrisa[m]>
barni2000: which kernel version are you using? Maybe something's broken in 6.13-r0?
<ValPackett[m]>
<ValPackett[m]> "pretty sure it would completely..." <- yep, on nora it just doesn't even get to showing anything on the display, screen turns black and it hard freezes
<barni2000[m]>
strang it is working fine on xiaomis
<barni2000[m]>
s/strang/strange/
<barni2000[m]>
<fish4terrisa[m]> "edt seems cannot probe the..." <- you have goodix
<ValPackett[m]>
hm, looks like the mic ultra quiet issue is just the ucm stuff not actually applying the 'ADC1 Volume' from the BootSequence
<barni2000[m]>
hm now i have blue screen
<barni2000[m]>
it is started to boot and crashed i have seen the splash
<barni2000[m]>
let me try with my other 8917 devices
<ValPackett[m]>
well, the moment at which it first decides to go into that state might vary, so it might hang at different moments
<ValPackett[m]>
for me it's been right after loading the panel driver and before even showing anything on the panel
<ValPackett[m]>
but either way looks like no devices in mainline ever have that full system pc state, all the socs that have cluster pc are multi cluster socs
<ValPackett[m]>
<barni2000[m]> "ig some driver could cause we..." <- i'm pretty sure it's mainline missing the notify-rpm thing that's actually really important
<barni2000[m]>
it wakes up from sleep so it seems it is not missing
<ValPackett[m]>
if somehow it works on some device.. that's a weird case. maybe that device isn't actually 8917? xD
<barni2000[m]>
it should be checked what downstream rpm driver is do
<ValPackett[m]>
the comment on msm_rpm_enter_sleep is "During power collapse, the rpm driver disables the SMD interrupts to make sure that the interrupt doesn't wakes us from sleep."
<ValPackett[m]>
so that's seemingly actually to not get woken up when we don't want to be
<ValPackett[m]>
oh but then it also calls msm_mpm_enter_sleep
<barni2000[m]>
ig pinctrl driver should be extended
<barni2000[m]>
mpm map is not defined yet
<ValPackett[m]>
right we don't seem to have a qcom,mpm in our dtsi
<barni2000[m]>
i have started to define it is on my git stash
<ValPackett[m]>
yea yea irq-qcom-mpm talks about "When AP is about to get power collapsed" (tho it doesn't seem to actually set timers, huh)
<barni2000[m]>
what is strange it can wake up when it is booted