stanu_ has quit [Read error: Connection reset by peer]
pere has quit [Read error: Connection reset by peer]
inky has joined #maemo-leste
inky_ has joined #maemo-leste
inky has quit [Ping timeout: 265 seconds]
uvos has joined #maemo-leste
jkepler has joined #maemo-leste
stanr has joined #maemo-leste
<buZz>
stanr: at what fps?
<stanr>
bout 0.5 in opening scene
<buZz>
lol
<stanr>
that's not at all the appropriate port
<stanr>
there's an iphone one, one for 3DO etc
<buZz>
doom3 for 3DO?
<stanr>
it was just neet to see it render without errors
<stanr>
yes
<uvos>
no
<buZz>
O_o
<uvos>
no way
<jkepler>
Hello, I'm trying to install maemo-leste on my N900. I've got a working freemantle install, so I'm following the instruction on the wiki (https://leste.maemo.org/Nokia_N900). However, after step 1. Install "U-Boot with kernel 2.6.28-omap1", the application installer says, "!!!U-Boot with kernel 2.6.28-omap1 breaks the 3rd party package policy."
<buZz>
3DO is what, 1991?
<uvos>
the 3do can bearly do 3d
<uvos>
there was a (bad doom port for the 3do)
<buZz>
jkepler: yes thats correct, we break nokia's 3rd party policy :)
<Wizzup>
uvos: install u-boot on fremantle so that he can boot leste on sd card I would guess
<jkepler>
uvos, I'm trying to follow https://leste.maemo.org/Nokia_N900 to install leste alongside my existing freemantle, but got an error that step 1. violates 3rd party policies.
<uvos>
so extract the kernel and append the on pc?
<uvos>
instead of installing via fremantle? (im no expert on this subject)
<uvos>
*append the kernel on pc
<sixwheeledbeast>
What version of fremantle do you have at the moment?
<Wizzup>
uvos: I think we want our instructions to encompass instructions for users to install it on fremantle
<Wizzup>
also while keeping fremantle intact
<Wizzup>
so we might need to expand the instructions to install power kernel first, if that is what is required
<uvos>
Wizzup: sure but if it dosent work atm the route that goes by just taking the freemantle zimage and appening that to uboot on pc works fine
<uvos>
i did that recently
<uvos>
if appended kernel_2.6.28-20103103 is fine with you i can give you the image
<uvos>
@jkepler
<sixwheeledbeast>
installing from a base system of stable/testing/thumb and then kp/non-kp I expect to have different results.
<sixwheeledbeast>
I suppose wiki expects cssu testing?
<Wizzup>
uvos: well if one doesn't care to keep fremantle working it becomes a different story sure
<uvos>
Wizzup: ?? i did keep freemantle working
<uvos>
you can just append the kernel from the kernel .deb for fremantle, if you choose the right .deb everything will work fine
<uvos>
once you flash the resulting uboot image via 0xffff
<jkepler>
How do I find out which version of cssu I'm running?
<uvos>
uname -a would be sufficant for your purposes
<uvos>
(to find out what kernel to append)
<Wizzup>
uvos: ok, it just sounds more involved and risky than fixing our wiki instructions is what I mean
<jkepler>
So Wizzup, if I understand correctly, 0. install the power-kernel from the freemantle application manager, then follow steps 1 - 3 to install maemo leste alongside my cssu, yes?
<Wizzup>
I think so, there is a dedicated u-boot thread on the forum somewhere, I've also had this trouble in the past but I do not remember what it was exactly
<Wizzup>
I think you need to install the power kernel and then pick the right u-boot entry from HAM (if you see multiple)
jkepler has quit [Quit: Client closed]
jkepler has joined #maemo-leste
<sicelo>
uvos: the 3d underperfomance is a Leste issue or powervr+mainline?
<sicelo>
jkepler: CSSU version is stated in Settings > About Community SSU
<sicelo>
as for your issue, install the modules it is asking for - should be safe to do it from `apt-get install` in XTerminal
<jkepler>
ah, I didn't realize that I could overrule the app manager with the terminal - I use apt all the time on my Debian stable installation.
<sicelo>
you can, and most of the time it's safe. However, of course you want to read all warnings and act accordingly
<sicelo>
do check your CSSU version for sure, especially to know if you have thumb, for which you'll need to be just a bit more careful with setting up your u-boot scripts if you intend to use them
<uvos>
sicelo: ddk1.14 + wayland performs better maybe not quite android par
<bencoh>
sicelo: iirc thumb is unusable with srtock kernel
<uvos>
sicelo: its mostly ddk1.9 + xorg that underperforms vastly
<uvos>
sicelo: for some really wierd reason ddk1.14 plain drm also underperforms vs wayland
<uvos>
no idea why
<uvos>
*ddk1.17
<bencoh>
has anyone tried suspend on droid4 with powervr?
<uvos>
dident work last time i tryed but not sure if its pvrs fault
<bencoh>
(I know it worked decently without, at least back 1~2y ago)
<sicelo>
bencoh: yeah, hence the need to know before one sets up automatic boot. the boot script must specify the thumb kernel in that case
<bencoh>
uvos: oh :(
<Wizzup>
I think I did suspend and it went ok, but X segfaulted on wakeup
<Wizzup>
iirc
<bencoh>
shit
<bencoh>
back then I used to do it all the time with X open, and it didn't crash
<sicelo>
wayland on ddk1.17 works quite alright indeed - phosh runs well enough ... not perfectly
<bencoh>
Wizzup: I think someone already tried disabling it but I'm unsure ... I wonder if disabling the hdmi regulator (currently always-on) would help with power
<Wizzup>
bencoh: no 3d in use and no powervr specific ddx perhaps
<Wizzup>
bencoh: good point, not sure about the hdmi regulator
<mighty17[m]>
<sicelo "wayland on ddk1.17 works quite a"> Does it crash a lot on d4?
<mighty17[m]>
Plus patch wlroots
<bencoh>
Wizzup: for my working suspend test? yeah, that was an old custom ddx, with no 3d/powervr
<Wizzup>
bencoh: that's probably it tbh
<bencoh>
(suspend just happened to work iirc, I didn't do anything for that)
<bencoh>
I'm starting to think it might be the only way to achieve better battery life on droid4 :(
<uvos>
bencoh: hmm not sure why you would want to suspend a mapphone
<uvos>
bencoh: since battery life is pretty good idle allreadt
<uvos>
y
<bencoh>
(we currently get at most two days idling with everything disabled)
<bencoh>
uvos: err .... n900/fremantle idles at ~10mW with modem on
<uvos>
sure so dose android
<uvos>
it dosent suspend
<uvos>
and uses like 15mW
<bencoh>
android on droid4 idles at 15mW?!
<uvos>
just improve kernel support :)
<Wizzup>
bencoh: I think OFF mode would be key here
<uvos>
yeah
<bencoh>
oh and, android uses suspend
<Wizzup>
uvos: is that with modem on?
<uvos>
(with no modem to be fair)
<uvos>
bencoh: no
<uvos>
bencoh: not on d4
<uvos>
it dose not suspend
<uvos>
ever
<uvos>
it uses off mode
<bencoh>
hmm
<uvos>
Wizzup: with modem its more
<bencoh>
(I didn't get great battery life on lineage/droid4 either btw, so I'm a bit puzzled anyway)
<uvos>
linageos has many wakeups
<bencoh>
Wizzup: I thought droid4 already hits OFF mode?
<uvos>
no
<uvos>
it hits RET
<bencoh>
ah
<bencoh>
I see
<uvos>
omap4 lacks support for off in mainline kernel
<bencoh>
alright then
<sicelo>
mmm
<bencoh>
hmm, was that the part where kernel complains about a missing something in firmware to enter OFF mode?
<sicelo>
omap3 has support for what? OFF? RET? or both?
<uvos>
both
<uvos>
bencoh: all the drivers need support to reload the registers after off
<uvos>
bencoh: there might be other stuff missing to
<uvos>
bencoh: tmlind likely knows more details
<stanr>
d4 is tolerant of any extended battery you wire up, even in android
<buZz>
bencoh: is twodays idle 'not great battery life' ?
<uvos>
buZz: well no
<uvos>
buZz: compeard to motorola stock android that really bad
<buZz>
ah ok, i didnt try that
<bencoh>
uvos: that wouldn't be the most difficult part, considering I had pm_suspend working (without pvr)
<bencoh>
buZz: it drops pretty quickly to less than a day overall once you start using it a bit
<sicelo>
mighty17[m]: it did crash, but i think yours is more crash happy - i could use phosh reasonably ok on the droid4.
<buZz>
bencoh: yeah i know
<buZz>
imho thats still pretty ok
<buZz>
but well, longer would be very welcome obv :)
<bencoh>
that's more than enough for a dev phone and playong around yeah :)
<bencoh>
it's great already, but ... still :)
<uvos>
the real problem is that it uses way to much power while using the device (imo)
<uvos>
this is related to pvr
<bencoh>
(tbf 8Mm amazed at the progress that was done there)
<uvos>
the cpu ends up doing a lot of rendering work and has to copy the frame around
<uvos>
so while scrollin in android the cpu remains idle on leste its hammerd 100%
<uvos>
also android is way more agressvie with gateing sgx while the display is on
<uvos>
it will gate sgx between frames
<mighty17[m]>
<sicelo "mighty17: it did crash, but i th"> i mean many apps dont work :P
<sicelo>
or i'm more patient ;-)
pere has joined #maemo-leste
<mighty17[m]>
oh damn i have been replying sorry irc
<mighty17[m]>
i mean many apps dont work :P > well idk whom to blame pmos or sgx
<stanr>
spexking of sgx, what are you seeing with glmark2-es
<stanr>
./glmark2-es2 --fullscreen
<bencoh>
uvos: is redrawing rate capped, or does it get as high as possible?
<stanr>
though for visible results you need to dsmetool -k hildon and run it without
<uvos>
its capped to 60 but it dosent matter it never acchives this with a gles2 client
<bencoh>
hildon compositing counts as gles client?
<uvos>
no - sorta
<bencoh>
(I have scrolling in mind)
<uvos>
hildon gets a special path
<uvos>
the xclients that are not the compositor get hammerd
<mighty17[m]>
./glmark2-es2 --fullscreen oh rightt i forgot about this
<uvos>
because the dri2 buffer flip path is broken
<uvos>
(it causes high cpu use - very likely the cpu i copying the frame)
<uvos>
*is
<stanr>
well we get strange results with a lot of fullscreen gles2
<stanr>
like it's getting the wrong resolution
<stanr>
but it's faster than windowed/composited
<mighty17[m]>
gimme a sec, i hv diff powervr device
<stanr>
i get glmark2 score 34 on hdmi
<stanr>
most tests top out at 42-43 fps though which is odddly consistent
<jkepler>
Duh! Just realized that this N900 (I have two) is the one I *didn't* upgrade to CSSU. Gotta do that first, then power-kernel, then U-Boot with kernel 2.6.28-omap1, so I can use Freemantle CSSU and Maemo Leste. Thanks for the help!
<uvos>
jkepler: you can just use my image then thats with stock freemantle kernel
<uvos>
or apppend yourself
<uvos>
no need for cssu
<Wizzup>
jkepler: great, maybe cssu should be mentioned as requirement on the wiki if not already
<uvos>
(unless you want it ofc)
<buZz>
i thought CSSU was already mentioned on it?
<jkepler>
@buZz I though so too, but I remembered incorrectly, confusing this n900 with my other one. I recently moved and this one was in storage, so its been a while since I'd picked it up.
<buZz>
jkepler: if you want, you could be making your first addition to maemo leste today!
<buZz>
:D
jkepler has quit [Quit: Client closed]
<buZz>
lol
<buZz>
i scared them off, didnt i
jkepler has joined #maemo-leste
<buZz>
wb
<jkepler>
So if I get maemo leste installed with CSSU that's something others haven't done? Or is it more that its not yet been documented?
<uvos>
its been done
<uvos>
its just poorly documented maybe
<buZz>
i think whoever wrote the documentation didnt consider that n900 might -not- be using CSSU ;)
jkepler has quit [Client Quit]
<buZz>
geebus
<uvos>
buZz: maybe he just hates you
<uvos>
:P
<buZz>
understandable
<buZz>
must have spoken to my gf
<uvos>
:D
jkepler has joined #maemo-leste
<stanr>
modifying a scummvm theme removes it from the themes list. i can't even.
<buZz>
:D
<sicelo>
CSSU is not a requirement. Even stock/vanilla can handle u-boot just fine
<jkepler>
@buZz do you have some reasons you prefer running Freemantle without CSSU? I used CSSU on my other N900 for years, so I don't recall if plain freemantle has any advantages, particularly in 2021.
<sicelo>
there's no good reason to
<sicelo>
but, if you flash the standard image, you have non-CSSU, and if you're itching to play with u-boot right away ... there's a reason
<jkepler>
I want to start using maemo leste to the extent that I can, as I really like the idea of the project. My main phone rune LineageOS (without any Google services), but I'd much prefer having a mainline GNU/linux kernel and standard tools on a command line with a pocket-sized physical keyboard.
<stanr>
it very fun to have the same base OS on phone as rest of my devices
sunshavi has quit [Ping timeout: 265 seconds]
inky has joined #maemo-leste
mardy has quit [Quit: WeeChat 2.8]
inky_ has quit [Ping timeout: 252 seconds]
jkepler has quit [Quit: Client closed]
<bencoh>
Wizzup: I doubt CSSU is a hard requirement
crab has quit [Ping timeout: 240 seconds]
crab has joined #maemo-leste
<stanr>
'i'll just quick make a large font SCUMMVM theme', he said...
<bencoh>
haha
<bencoh>
is there a way to disable HAM automatic repository update?
<Wizzup>
bencoh: I disabled wifi with the switcher but I am still able to connect to wifi it seems?
cockroach has joined #maemo-leste
<bencoh>
Wizzup: hmm, what do ifconfig/wpa_cli say?
<bencoh>
oh and, I just saw wpasupplicant and friends got an upgrade today when updating repos ... I hope it doesn't enable the interface now
<Wizzup>
bencoh: it doesn't enable the interface
<bencoh>
then if you were still able to connect, it means the applet didn't disable it (?)
<bencoh>
are you able to reproduce it?
<bencoh>
oh and, I might have an idea
<bencoh>
apparently the wifi.sh script doesn't get a parameter from the applet, but instead toggles the interface based on its current status
<bencoh>
so if the interface was disabled manually (not updating the applet status), then you tried disabling it from the applet ...
<bencoh>
it will actually enable it (I think)
<uvos>
bencoh: maybe try /dev/rfkill (isent in leste kernel you need to compile a kernel with support for this) then nothing can unblock the interface except your widget
<Wizzup>
bencoh: I was connected to a wifi ap when I tried to disable it
<Wizzup>
uvos: I don't think we necessarily want rfkill?
<uvos>
Wizzup: why not?
<uvos>
Wizzup: the widget is to disable wifi no?
<uvos>
i thought
<bencoh>
Wizzup: can you reproduce it?
<Wizzup>
bencoh: I think maybe the iaps were cached
<Wizzup>
that's why they showed up in the scan dialog
<Wizzup>
I suppose I could flush more quickly
<Wizzup>
probably should
<bencoh>
Wizzup: did you really manage to connect though?
<uvos>
the widget is just bringing down the wlan0 interface or?
<bencoh>
uvos: ifconfig wlan0 0.0.0.0 down
<uvos>
dose that even prevent scanning? or monitor mode for that matter?
<bencoh>
I think it should, unless I missed something
<bencoh>
at least I don't get any scanning result after ifconfig down from the connections dialog
<bencoh>
uvos: is rfkill even supported by our driver(s)
<bencoh>
?
<Wizzup>
bencoh: I think so, kinda weird
<Wizzup>
uvos: it does prevent scanning
<uvos>
bencoh: probubly
<uvos>
bencoh: not that i tried
<uvos>
Wizzup: why would that be wierd?
<Wizzup>
uvos: because the interface is down and wpa_supplican cannot use it
<uvos>
Wizzup: that dosent disable scannin per say
<Wizzup>
uvos: try it
<uvos>
wpa just dosent use down interfaces
<Wizzup>
it doesn't work
<uvos>
right but is that wpa
<uvos>
or kernel interface
<uvos>
im not so sure
<Wizzup>
pretty sure it's kernel iface
<uvos>
ok
<uvos>
i dont see why you think rfkill is wierd
<bencoh>
he was referring to the fact that he still connected, not to rfkill :)