<aka_[m]>
<konradybcio> "you probably need the ipa clk to..." <- even when ipa hardware is not initialized?
<aka_[m]>
CE what is that?
animist has quit [Remote host closed the connection]
animist has joined #linux-msm
<aka_[m]>
converted xo_boards into rpmcc it didn't restart yet however its stuck on black screen
<aka_[m]>
oh pls
<aka_[m]>
cpufreq cannot depend on rpmcc
<aka_[m]>
because it probe too late
<aka_[m]>
and cpufreq fails with -19
<aka_[m]>
ok it die
<aka_[m]>
i wonder if 8953 works fine
svarbanov_ has joined #linux-msm
svarbanov has quit [Ping timeout: 480 seconds]
<konradybcio>
<aka_[m]> "even when ipa hardware is not..." <- well, qos writes need the clock to be on
<konradybcio>
<aka_[m]> "CE what is that?" <- crypto engine
<konradybcio>
<aka_[m]> "because it probe too late" <- sounds like broken devlinks?
<aka_[m]>
konradybcio: black magic to me
<konradybcio>
same
<aka_[m]>
konradybcio: there is no rpmcc clock for that?
<aka_[m]>
there might be one on 8974
<konradybcio>
well, you can try just commenting out smd clocks to see which ones cause you trouble
<aka_[m]>
then why your rework has no ipa keep_alive
<konradybcio>
as in, in the soc data in the driver
<aka_[m]>
there is not much besides ipa/div/vv ones
<konradybcio>
well, it's one of them!
<aka_[m]>
you are developer >_>
<aka_[m]>
if it dies on 8976 then it might on others like 8953
<konradybcio>
or your interconnect driver is broken..
<aka_[m]>
konradybcio: with ipa commented it works
<aka_[m]>
konradybcio: again
<aka_[m]>
YOU are developer i just generated it
<konradybcio>
so am I supposed to fix the broken generated driver or?
<aka_[m]>
so i have something wrong in icc driver if ipa commented clocks make it not die?
<konradybcio>
yes, that means it's trying to set the QoS registers at the (mas_)IPA node and the clocks were shut down prior to that, meaning you have to add the ipa clock as an intf clock to the bus that hosts mas_ipa
<aka_[m]>
uh i should have mas id in that
<aka_[m]>
and it shows -1
<aka_[m]>
wait
<aka_[m]>
its local
<aka_[m]>
ap_owned
<aka_[m]>
does rpm-id matters then much?
<aka_[m]>
wait i have it added in current
<aka_[m]>
konradybcio: how ds handle these then?
<aka_[m]>
ipa is not keep_alive
<aka_[m]>
in gcc-8976.c
<konradybcio>
keep_alive is a concept for bus clocks
<konradybcio>
the ipa clock that comes from rpmcc is a "normal" clock
<konradybcio>
if it was in gcc, it'd probably be called gcc_ipa_core_clk
<aka_[m]>
so i add ipa clock same way like you add stuff to 8994 ICC?
<aka_[m]>
hmm i wonder if being unable to send request for few of these clocks and -6 on them can be also due to disabled clocks for these
<konradybcio>
the _a clock is just there to represent the active-only/"cpu on" icc bucket
<aka_[m]>
after wakeup display still is quite blank >_>
<aka_[m]>
maybe display driver
<aka_[m]>
ok now im stuck
<aka_[m]>
if i add it as these
<aka_[m]>
what about stock ones
<aka_[m]>
i have to rewrite all of them right
<aka_[m]>
bus,bus_a,ipa
<aka_[m]>
ok im lost now
<konradybcio>
bus and bus_a are handled implicitly if you have the v6 interconnectification fixation
<aka_[m]>
i have one you posted to me
<aka_[m]>
RFT one
<aka_[m]>
ok its broken here
<aka_[m]>
../drivers/interconnect/qcom/msm8976.c:1385:10: error: 'const struct qcom_icc_desc' has no member named 'intf_clocks'; did you mean 'num_clocks'?
<aka_[m]>
konradybcio: ah i see you have new tree
<aka_[m]>
so no reason to bother for now
<aka_[m]>
i will have this patched but won't test it
pespin has quit [Remote host closed the connection]
jhovold has quit [Ping timeout: 480 seconds]
<konradybcio>
well if you don't mind i'm sorta updating a lot of things regarded to bus scaling..
dliviu has quit [Remote host closed the connection]