gabes2 has quit [Quit: Ping timeout (120 seconds)]
gabes2 has joined #linux-amlogic
chewitt has joined #linux-amlogic
cottsay has quit [Ping timeout: 240 seconds]
cottsay has joined #linux-amlogic
Tartarus has quit [Server closed connection]
Tartarus has joined #linux-amlogic
<lvrp16>
f_: I am sure certain bits of it work.
xdarklight has quit [Remote host closed the connection]
xdarklight has joined #linux-amlogic
vkoskiv has joined #linux-amlogic
<narmstrong>
the second cluster init would be missing I think
cyrozap has quit [Server closed connection]
cyrozap has joined #linux-amlogic
psydroid2 has joined #linux-amlogic
<rtp>
hi. I'm not sure I'll have to look at it soon but does someone already seen "meson_clk_pll_set_rate: pll did not lock, trying to restore old rate 5616000000" and loss of network on solitude board (kernel stock debian 6.1.0-13-arm64)
<rtp>
[ and the uboot version is the one coming with K-R boards if it matters ]
<lvrp16>
rtp: this is a long standing issue. I have to spend some time to replicate with and pass it to Amlogic to check.
<lvrp16>
If you have a reliable way of reproducing, let me know.
psydroid2 has joined #linux-amlogic
<minute>
narmstrong: cool!
<jbrunet>
Same happens on all g12/sm1 based devices (odroid, khadas, etc) ... since the beginning of the support.
<jbrunet>
I spent a lot of time on this a few years ago. Could not get to the bottom of it.
<jbrunet>
I tried to tweak the pll driver quite but nothing really helped. With an issue hard to reproduce like this, it is even harder to assert if the problem is solved or not.
<jbrunet>
It is annoying, happens often on boot but fairly hard to reproduce on purpose.
<jbrunet>
Usually it is the SYS_PLL failing to relock on rate change. Seems some HW samples exhibit the issue more than others. Some never do.
<jbrunet>
One thing SBC vendor did is disable dvfs or set the governor to "performance" ... which has the same effect, CPU clock is locked once and for all
<jbrunet>
I remember seeing similar problems reported on odroid forums, using the vendor kernel and going for the performance "governor" as a solution ... so the issue may not be related the PLL driver we are using but maybe we can do something to mitigate it. Could not find a solution so far, sorry
GNUtoo has quit [Remote host closed the connection]
GNUtoo has joined #linux-amlogic
<rtp>
ohhh... so it's going to be hard to solve :(
<jbrunet>
On SoC samples which showed the problem before, I tried to stress DVFS, constantly and randomly changing the cpu freq. It went on for days and could never reproduce the problem that way ... really strange. Might be environment related too, like the temperature maybe
Stricted has quit [Server closed connection]
Stricted has joined #linux-amlogic
<jbrunet>
The best way I found was to do constant reboot, but it may take lot to reproduce the problem.
<rtp>
Im getting it here randomly but happens often. I fear the setup will be hard to reproduce and the board is rebooting as soon as the net is down so it'll be things harder to debug
<lvrp16>
I will have a fleet of these at the end of the month and try to find a way to reproduce.
GNUtoo has quit [Remote host closed the connection]
GNUtoo has joined #linux-amlogic
khilman has quit [Server closed connection]
khilman has joined #linux-amlogic
<djrscally>
narmstrong: One bit of your pmdomain patch for the A311D's ISP is confusing me. In the datasheet I have the Power Sequence for the ISP gives me "Domain Power Sleep bit" as 0xff8000e8[18] and "Input signal isolation bit" as the same
repk_ has quit [Server closed connection]
repk_ has joined #linux-amlogic
<djrscally>
But that doesn't really tally with what you've got the g12a_pwrc_isp's sleep_reg/mask and .iso_reg/mask set to
<djrscally>
I think you've got it set as Domain Power Sleep bit = offset 0xE8[19:18] and Input Signal Isolation bit = offset 0xEC[19:18
<djrscally>
And I can't see where that comes from...but clearly you're right as if I switch it to what I think the datasheet says it ought to be the kernel panics :D
kilobyte_ch has quit [Server closed connection]
kilobyte_ch has joined #linux-amlogic
<narmstrong>
Yeah indeed the struct naming are from old socs and they shuffled the bits in between but the luck is both bits must be set in the same sequence so it works
<djrscally>
You think the ISP power sequence table in the datasheet is likely to reflect the G12A rather than B then or something?
dlan has quit [Server closed connection]
dlan has joined #linux-amlogic
Kwiboo has quit [Server closed connection]
Kwiboo has joined #linux-amlogic
<narmstrong>
AFAIK it’s only in B, same as NNA
buzzmarshall has joined #linux-amlogic
Daanct12 has quit [Quit: WeeChat 4.1.1]
CounterPillow has quit [Server closed connection]
CounterPillow has joined #linux-amlogic
<djrscally>
Hm ok. I'm still confused then I'm afraid. How did you know that you needed to set [19:18] of 0xE8 and 0xEC?
JohnnyonFlame has joined #linux-amlogic
<narmstrong>
I extrapolated from VPU and NNA configs
<narmstrong>
documentation is crap, nothing correlates
<djrscally>
Joy of joys
<djrscally>
OK; thanks
f_[xmpp] has joined #linux-amlogic
<f_>
narmstrong: I agree!
iprusov has quit [Server closed connection]
iprusov has joined #linux-amlogic
<f_>
Aaaand the CM4 boots u-boot!
<narmstrong>
f_: it's expected, except if you did spl boot ;-p
<f_>
I didn't boot using SPL.
<f_>
and nothing is on the eMMC :P
<f_>
Booted from USB
<f_>
the eMMC appears to be empty.
<narmstrong>
yup
<narmstrong>
better empty than crappy sw
<f_>
Who thought hosting stuff on drive was a good idea?
<f_>
narmstrong: ...which is what I'm about to install :D
<f_>
Am downloading their Android image.
<narmstrong>
awwwwwwwwwww
<f_>
awwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww
<narmstrong>
don't do that
<f_>
why not?
<f_>
Except I doubt the vendor U-Boot will last even just a few minutes :P
<f_>
I usually end up overriding u-boot and al.
<narmstrong>
:-p
<f_>
Oh I think my computer doesn't like it
<f_>
It froze upon extracting lol
<f_>
or I filled /tmp
<f_>
After some magic sysrq it's back up
tsegers has quit [Server closed connection]
tsegers has joined #linux-amlogic
f_[xmpp] has quit [Ping timeout: 240 seconds]
<f_>
I guess f_[xmpp] also doesn't like it
JerryXiao has quit [Server closed connection]
JerryXiao has joined #linux-amlogic
<f_>
haha what a weird Android build
<f_>
Is it Android or Android TV?
<f_>
lol
<f_>
Installing Android on it was a mistake
<phh>
if you're asking I'm guessing you have the ""mbox"" build, which is ""chinese"" android without google for tv boxes
<phh>
which isn't exactly android tv, because of google apps, and because they largely don't include aosp tv apps either, or props
<phh>
it is indeed a pretty weird build
<f_>
That build includes some AOSP tv apps.
<phh>
I would guess it includes TvSettings but not LeanbackIME for instance
<f_>
At least there's an ATV settings app
<phh>
but dpi is 240 instead of 320
<f_>
(along with Amlogic's own settings app)
<f_>
But yeah I must have a modified MBOX build.
<f_>
Which also means shady apps like what was on my S905 box.
<f_>
They did replace that "MBOX" bootanimation though
<narmstrong>
yeah it's a frankenstein android, and it must be like Android 9
<f_>
correct
<f_>
Yeee old Android 9.
* f_
zeroes-out @AML header.
<f_>
Installing Android was a mistake.
<f_>
(I mean, I have nothing against Android..but installing *this* specific build was a mistake)
<f_>
I was reminded how bad amlogic MBOX builds are.
<f_>
so I'll run postmarketOS on it
pinchartl has quit [Server closed connection]
pinchartl has joined #linux-amlogic
vagrantc has joined #linux-amlogic
dliviu has quit [Server closed connection]
dliviu has joined #linux-amlogic
<f_>
Anyway I *won't* work on anything newer than gxl for now.
<f_>
Let's remind everyone that.....gxl is still not booting!
<f_>
And I haven't done much since (lack of time).
<phh>
i probably already asked before, but do you have a nice TV environement you can build in pmos?
<f_>
(hopefully now I should have more time)
<f_>
phh: You didn't ask before, and apart from Plasma BigScreen (not really usable) and Kodi..not really
<f_>
But there's one interface called SXMO, which I'm planning to turn into a nice 10-foot environment some day.
<phh>
k that's what i thought. you think pmos is more worth it than libreelec?
<f_>
('TV mode')
<f_>
phh: I think you should use what works best.
<phh>
cool, i wish you good luck in that endeavor :-)
<phh>
f_: well, my current experience is that I spent 4 hours in jelos failing to build, and 4 hours in pmos failing to build
<f_>
Besides having ported pmOS to some amlogic boards...
<f_>
phh: hmm
<phh>
(jelos because they didn't maintain libreelec's 32 bits. pmos largely because there is no standard way to build a kernel, but even then I was stuck)
<f_>
phh: "no standard way to build a kernel"?
<f_>
kernel builds in pmOS are mostly `cp kernelconfig linux/.config` and `make`.
<phh>
that's definitely not my experience, and that's definitely not what commited kernel configs does
<phh>
I'm not saying it was a lot more, but that little more was painful to find
<phh>
(I don't have a pure mainline tree, only mainline + patches)
<f_>
while this calls the default `prepare` function
<f_>
phh: in the build() function, I mean.
<f_>
anyway, this calls alpine's default `prepare` function and copies the kernel config to the tree
<phh>
honestly building a true mainline kernel with only patches and defconfig should be only contain the list of files, the base version the .config and the hashes and nothing