Danct12 is now known as Guest8981
Danct12 has joined #linux-msm
marvin24_ has joined #linux-msm
marvin24 has quit [Ping timeout: 480 seconds]
Guest8981 has quit [Remote host closed the connection]
Daanct12 has joined #linux-msm
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #linux-msm
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #linux-msm
Daanct12 has quit [Remote host closed the connection]
Danct12 has quit [Remote host closed the connection]
Danct12 has joined #linux-msm
Daanct12 has joined #linux-msm
pespin has joined #linux-msm
Danct12 has quit [Quit: WeeChat 3.8]
<
konradybcio>
check the frequency in your downstream gcc driver
<
konradybcio>
aka_: it's off as far as Linux is concerned, you can't read the actual state from rpm(h), you can only see what Linux thinks it's true
<
konradybcio>
RPM does provide sleep stats but since you're interacting with the device, it's obviously not asleep
<
aka_[m]>
konradybcio: thats quite awful
<
aka_[m]>
would we be able to read it via registers maybe/
<
aka_[m]>
im still woried a bit about this modem supply shit
<
aka_[m]>
S1 domain goes into VDD_MODEM_X nowhere else
<
konradybcio>
no, it's not exposed via mmio at all
<
aka_[m]>
no idea how is that even wired
<
aka_[m]>
konradybcio: fug
<
aka_[m]>
wasnt it like spmi protocol or something?
<
aka_[m]>
extended i2c-like
<
aka_[m]>
konradybcio: how valid would be reusing struct in q6v5_mss.c?
<
aka_[m]>
its static const tho
<
aka_[m]>
you cannot really assign it this way i believe?
<
aka_[m]>
static const are like you cannot modify these at runtime
<
aka_[m]>
or i can just assign msm8909 data tho
<
aka_[m]>
prob same core
<
aka_[m]>
or not it seems
<
konradybcio>
considering that would allocate an entire struct worth of ram anyway, just do it explicitly
<
JIaxyga[m]>
<konradybcio> "check the frequency in your..." <- But if I use 202 frequency in dts, I will get an error that 208 cannot be set
Caterpillar has joined #linux-msm
<
JIaxyga[m]>
I also looked at the sdhc frequency inside the android using a script, then I got exactly 202 frequency
<
JIaxyga[m]>
Today I will test what will happen if I set the values to 208 for dts. I think it will work, but I wouldn't count on long term stability
<
JIaxyga[m]>
<konradybcio> "check the frequency in your..." <- 208 in gcc and 202 in dts
<
JIaxyga[m]>
caf kernel
<
JIaxyga[m]>
When I ported gcc I didn't change frequencies
<
aka_[m]>
My limited understanding is that opp core will select closest freq to provided via dts and it will use it's power stage
<
JIaxyga[m]>
konradybcio: I set the frequency to 208 in dts and left 208 in gcc and it doesn't throw any errors
<
JIaxyga[m]>
But the rate is still set to 201999975
<
aka_[m]>
Happens with alpha-pll
<
aka_[m]>
For me when I was dealing with gcc ones I had like 799.999999mhz
<
aka_[m]>
JIaxyga: you can set it to 202mhz in freq table too
<
aka_[m]>
And when sending you can explain that this is closest what can you get with these dividers
<
konradybcio>
what does debugcc say?
<
konradybcio>
note it will only report frequency when you're actively accessing the emmc
<
konradybcio>
or sdcard
<
aka_[m]>
It's not like you set clock to some 0.1mhz value you just set multiply/divide from source freq there is always margin on clocks
<
JIaxyga[m]>
<konradybcio> "what does debugcc say?" <- Is it enabled by default in downstream?
<
konradybcio>
hard to tell
<
konradybcio>
you'll need to port your platform though
pespin has quit [Remote host closed the connection]
<
JIaxyga[m]>
I think we need to write a generator for this
<
aka_[m]>
konradybcio: oh lol
<
aka_[m]>
msm8976 appears to also have "proxy vote" via s1
<
aka_[m]>
i noticed something
<
aka_[m]>
vdd_pll-supply is nowhere mentioned in our mainline fork
<
aka_[m]>
we should do pll-supply = <&pm8953_l7> i believe
<
aka_[m]>
wait, we do it
<
aka_[m]>
so i wonder, should i add s1 as power domain to 8976?
<
aka_[m]>
ok weird, on 8953 there is vdd_mss-uV
<
aka_[m]>
with level defined
<
aka_[m]>
nothing on 8976 so maybe they don't do "level" votes and might set stuff from driver as regulator
<
aka_[m]>
binding name not specify if its pd or regulator
<
aka_[m]>
huh, main difference between s1 on 3.18 vs 4.9 is that min/max volts are numeric values vs votes
<
aka_[m]>
there is no "level", "floor", "corner"
<
aka_[m]>
there is flag "qcom,use-voltage-level"
<
aka_[m]>
i cannot even find this flag in repo xD
<
aka_[m]>
found,sadly it only assign ops