<Turl>
libv: (or is my system just horribly configured? intel graphics)
<hramrach>
Turl: what flickers?
<Turl>
hramrach: stuff. Monitor goes off and then redetects the signal and shows the OSD again
<Turl>
it's overly annoying because the monitor is slow doing it. And the OSD is ugly and slow to disappear
<hramrach>
generally once bios is over you can just keep going.
<hramrach>
if you are willing to do EFI, that is
<hramrach>
because PC bios does text only and you have to change modes after that
<Turl>
I'm using good ole BIOS w/ GRUB
<Turl>
I don't care if it does it around BIOS time.
<Turl>
but it also does it before loading GDM, and after I log in there
<Turl>
anyway, this is overly offtopic. I should find some other channel to try to figure this out :)
<hramrach>
well, I guess your desktop environment frobs the graphics
<hramrach>
nothing else should *after* you log in
<hramrach>
because the X server is already running then
<hramrach>
it might be somewhat challenging to make the boot process itself flicker-free on legacy bios but with vesa and grub and careful mode picking you can go from grub to desktop without changing mode with possible hiccup when the accelerated driver takes over
<hramrach>
but once kernel modesetting starts there is no need to change mode from there
pseudomind has joined #linux-sunxi
hypophthalmus has joined #linux-sunxi
hypophthalmus has left #linux-sunxi [#linux-sunxi]
pseudomind has quit [Quit: pseudomind]
pseudomind has joined #linux-sunxi
Wizzup has quit [Ping timeout: 248 seconds]
rz2k has quit []
<hramrach>
ok. so the problem now is that the sunxi serial driver does something bogus and breaks the uart, and you cannot build sunxi kernel without a serial driver
<Turl>
hramrach: are you passing sth on cmdline?
<Turl>
like console=...?
<hramrach>
default is console=ttyS0,115200 rdinit=/sbin/init panic=10
<Turl>
hramrach: yeah but you're using uart3? so what did you change it to?
<hramrach>
changing it to console=tty0 rids me of the garbage but I still lose the u-boot configured console
<hramrach>
I am using uart0
<Turl>
make sure you keep the speed, otherwise the default is 9600 and you'll see it hang/spit crap
<hramrach>
or whatever is wired to the SD slot
<Turl>
or you can try the other way round, when it hangs, rerun your uart program at 9600 instead of 115200
popolon has joined #linux-sunxi
<hramrach>
I would have to configure u-boot to 9600 as well
<hramrach>
because it locks up really quick
<Turl>
not really
<Turl>
just let it lock, unplug the usb thingy so the program closes and replug/rerun
<hramrach>
it's normally not specified but maybe the driver has odd default ..
<hramrach>
btw what is card_boot0_para
<Turl>
hramrach: 8n1 I think
<hramrach>
there is no example of writing that so the kernel would understand
<hramrach>
hmm, default is n8 and you cannt specify that 1
mkutsevol has quit [Ping timeout: 252 seconds]
<hramrach>
and I don't even have mmc driver
<Turl>
hramrach: or maybe it's 1n8 and I just remember it backwards :p
<hramrach>
it would be 9600n8
mkutsevol has joined #linux-sunxi
<hramrach>
no room for another number
popolon has quit [Quit: Quitte]
<Turl>
hramrach: it's a long shot but you can try adding to cmdline console=ttyS0,115200 console=ttyS1,115200 console=ttyS2,115200 etc, just to make sure you are setting up the right one
<Turl>
if you're using the sd slot I don't think it's uart0
<hramrach>
you cannnot have 3 consoles on sun5
<hramrach>
it complains that port3 has unknown parameters but adding another uart section does not help
<hramrach>
anyway, I don;t even have a mmc driver so it should not really matter
<hramrach>
also it stops writing anything exactly when serial_sw is initialized, nothing to do with mmc
<libv>
oh, ok, this file looks ok now, not sure what i was looking at
<libv>
hramrach: it is worth a try though
<libv>
heh, the script.bin in your mail is from another revision
FDCX has joined #linux-sunxi
<hramrach>
The pins that aren't multiplexed is PG03 which is supposed to go to camera so with a flat cable breakout you should be able to get at that and get output only
<hramrach>
sadly, I don't have that
<hramrach>
the other option is uart3 which is not connected anywhere according to fex
<hramrach>
actually the other position for uart1 also goes to camera so it seems feasible they would connect the uart there in case of problems
<hramrach>
seems pin 12 is not connected and pin 13 goes to a via
<hramrach>
no, just artifact of the photo
<hramrach>
none of those pins are connected
ZetaNeta has joined #linux-sunxi
km4r has joined #linux-sunxi
<ZetaNeta>
hi
<ZetaNeta>
with uboot, i get tons of arm-linux-gnueabihf-gcc: error trying to exec 'cc1': execvp: No such file or directory
<pacopad>
Hi all, where could i find docs about how vdpau works ?
FR^2 has joined #linux-sunxi
<hramrach>
ZetaNeta: your gcc installation is probably broken
* ZetaNeta
cries
<hramrach>
pacopad: there are some vdpau docs from nVidia who invented if
<ZetaNeta>
(
<Faisal>
guys this might be a stupid question but i was wondering what is the fex file actually. i've read this page : http://linux-sunxi.org/Fex_Guide already. it says : It configures the GPIO pins and sets up DRAM, Display, etc parameters. --- isnt it the bootloader and then the kernel thats suposed to do that ? when in the boot process is this fex file involved. i know that the soc doesnt use that fex file , i understand that fex file is used to generate the
<Faisal>
script.bin file used in the boot process but can someone tell me when is this script.bin file needed in the boot process, is it uboot that uses it ? if so, then what if you dont use uboot, do you still need that ?
<hramrach>
Faisal: script bin tells the kernel how to st up things. different boards are connected differently. DTB does that for mainline and PCI BIOS for x86
<hramrach>
u-boot does not use script.bin
<hramrach>
the memory parameters it sets are defined at compile time but are the same you see in script.bin
<Faisal>
hramrach: thx, ok so it contains basically the board config...
<hramrach>
ideally u-boot would read that from script.bin but it's technically not feasible
<hramrach>
but you could like convert the part of script.bin u-boot needs into a simpler format that the SPL could read and make some support for that
tomboy65 has joined #linux-sunxi
tomboy64 has quit [Ping timeout: 272 seconds]
mkutsevol has left #linux-sunxi [#linux-sunxi]
geecko has joined #linux-sunxi
mkutsevol has joined #linux-sunxi
jinzo has joined #linux-sunxi
jinzo has quit [Changing host]
jinzo has joined #linux-sunxi
eebrah_ has joined #linux-sunxi
geecko has quit [Ping timeout: 240 seconds]
bsdfox has quit [Ping timeout: 252 seconds]
bsdfox has joined #linux-sunxi
FreezingCold has quit [Ping timeout: 260 seconds]
Faisal__ has joined #linux-sunxi
Faisal has quit [Ping timeout: 245 seconds]
geecko has joined #linux-sunxi
hansg has quit [Quit: Leaving]
<mnemoc>
libv: I ended up ordering the samsung note 10.1 2014 (exynos 5420, 3GB, 32GB, A5-ish 10.1" 2560x1600, wacom digitalizer) hoping to be able to install mer on it soon. does wayland use mesa? if not, how would lima hook there?
rz2k has joined #linux-sunxi
rellla2 has joined #linux-sunxi
rellla has quit [Ping timeout: 260 seconds]
<libv>
it's a 5xxx, and it's a mali based one, but it's the mali t628 6core
<libv>
i only have basic capture and replay atm
<hramrach>
more importantly you need framebuffer driver
<hramrach>
mali is nice to have but not really important for a notepad
<mnemoc>
i thought the chromeos people had mainlined the framebuffer stuff...
<libv>
my experience with the chromebook is not a good one
<mnemoc>
:(
<mnemoc>
hramrach: unfortunatelly, even if for a notepad/pdf-reader 2d and touch/digitalizer is the most critical part, the general UI tends to need 3d to be usable :\
<hramrach>
is it the SM-P6000
<mnemoc>
yes
<hramrach>
the p6050 with phone uses some snapdragon for some reason
<mnemoc>
built-in lte maybe?
<mnemoc>
the wifi and 3g versions use exynos 5420
<libv>
samsung SoCs lack lte
<libv>
so they tend to ship snapdragons in the .us
<hramrach>
you could like use USB 3G
<hramrach>
free to mount different one for different countries
<hramrach>
it can be internal even when it's usb/sdio/..
<hramrach>
but that's their problem
<mnemoc>
or a mini pcie module... but a (guessed) built-in lte support on the snapdragon makes the whole thing cheaper
<mnemoc>
also slimmer
<hramrach>
probably if it's certified for most of the places they ever dreamt of selling to, anyway
<hramrach>
I guess you are in for some serious hacking with the note then :)
<mnemoc>
hramrach: but hacking with more hope that with the other alternatives
<mnemoc>
exynos 5420 should become easier to "free" than snapdragon or tegra alternatives..
<libv>
mnemoc: in what timeframe?
<libv>
mnemoc: and who is going to do it?
<libv>
and when will the next SoC be there?
<libv>
ratrace.
<hramrach>
you do not need a new soc when the current one works
<hramrach>
but for that the device should have some decent specs
<hramrach>
3G ram is nice but you still run out easily
<mnemoc>
libv: my biggest hope with the 5420 is chromeos people and those wanting to turn chromebooks into linux laptops
<libv>
hramrach: your statements are mutually exclusive.
<libv>
hramrach: over any time frame over 3 months
<libv>
mnemoc: how long has the arm chromebook been around?
<hramrach>
chromeos people tell you to install chromeos or go fuck yourself
<libv>
14 months.
<mnemoc>
libv: :(
<libv>
there is no community. period.
<mnemoc>
the odroid u3 can fix that...
<libv>
at least, there wasn't one 5 months ago
<libv>
no
<hramrach>
they do not support anythin than the chrome ROM monolith
<libv>
hardkernel doesn't get community
<libv>
they will not help change anything
<mnemoc>
:'(
<libv>
they actually work against it, be it knowingly, or be it by being stupid
<libv>
anyway, 5 months ago, there was no community, and the people who originally hacked the arm chromebook had moved on
<libv>
this means that there never will be a community for them
<hramrach>
libv: problem is that with current specs you *will* want the new device that comes out in 3 months because the current specs are best what there is but not best you can make use of
<libv>
the people who jumped on the chromebook are rats
<libv>
and they love running around their little mazes
<libv>
hramrach: how long has linux-sunxi existed?
<libv>
hramrach: how long is the average lifespan of any mobile device?
<hramrach>
I don't know. Haven't been htere at the start
<mnemoc>
the exynos chromebook has a crap display... but commits are teasing about one with 5420 and good display
<libv>
about a year and a half, for sunxi
<libv>
about 2 years, for mobile devices
<mnemoc>
we got a very nice community :)
<libv>
now... is mnemoc going to start a new community?
<mnemoc>
:(
<libv>
is he going to use his exynos tablet or is he going to play with it?
<mnemoc>
work tool
<libv>
and where will he be by the time he stops using it and is ready to play with it?
<libv>
hramrach here will then state "no chance, get the new whatever and hack that! it still has life in it!"
<mnemoc>
:(
<libv>
mnemoc: it's all true, isn't it?
<libv>
it's just reality
<libv>
it's how this works
<hramrach>
libv: a tablet that you can no longer buy but has support is kind of not useful
<libv>
hramrach: how long have you been using linux?
<hramrach>
if the new one is based on the same hardware as the old one then hacking the old one makes supporting new one easier
<libv>
hramrach: what graphics cards that have full support can you still buy new in the shops?
<hramrach>
but if not making it work is only good for people who already have the old tablet
<mnemoc>
libv: but but but... there is more people out there who can start communities too...
<voltagex>
I see the same problem with all ARM SoCs though
<libv>
voltagex: all except 1
<hramrach>
libv: there are no graphics cards with full support. However, many cards you can buy have useful support as in you can run a desktop on htem
<libv>
voltagex: because we have stuck with these slow bangers
<hramrach>
you cannot run a d\esktop on exynos
<voltagex>
libv: let me guess, sunxi? ;)
<voltagex>
I've been around for a while now
<hramrach>
also there are very few graphics cards with full support under Windows. There are always bugs
<libv>
hramrach: i was around when even basic support wasn't there for basic cards
<voltagex>
I've got an ARMv5te system here, was in use until the SD reader died - how much dev effort do you think is going into it now?
<libv>
hramrach: heck, thanks to me you have some hope of getting something out of your ATI graphics card within a few months
<libv>
now luckily, the graphics hardware market is no longer on a 6 month release cycle
<hramrach>
I have a few ATI cards to play with so if they get some improved support that might be nice :)
<libv>
with a major changes every 6 months
<libv>
that was 7 years ago
<libv>
it's cooled down
<voltagex>
aha! knew I recognised the name libv. Thanks for your work.
<libv>
SoCs are on that cycle now
<libv>
voltagex: that work was mostly in vain anyway
<voltagex>
libv: what's up?
<hramrach>
because they are not good enough. they will cool down when the new ones do not bring any useful real world features
<libv>
hramrach: when?
<voltagex>
hramrach: good enough for what?
<libv>
hramrach: and by that time, there will be something new, and you will have moved on with it
<voltagex>
libv: don't get too discouraged.
<libv>
voltagex: it's called realism, sorry if it is not positive.
<mnemoc>
what's after an octa a15/a7 in the chain? armv8... what's after 10.1" 2560x1600? nothing in the next half decade at least... how long have we been stuck with 1366x768?
<hramrach>
for anything. The cool new device has 3G ram. Can you not make use of more on a desktop/media center?
<libv>
mnemoc: i do not know when the mobile device races will end
<libv>
mnemoc: but it doesn't look like it will any time soon
<voltagex>
libv: I understand that, but your work is still very useful. There are lots and lots and lots of older A10 and A20 boards floating around now
<hramrach>
somewhere around 8G~16G you stop seeing any benefit for 90%+ people but the SoCs are not there yet
<libv>
mnemoc: especially not with apple laying down the gauntlet with the A7 like they did
<mnemoc>
uhm
<libv>
not sure if that proverb was correct though
<libv>
a7 is armv8 + rogue
<libv>
and everyone was surprised
<voltagex>
hramrach: I wish someone would decide to use less - I shouldn't need 2GB of RAM in my phone
<hramrach>
same for storage bandwidth, networking, CPU power
<libv>
arm first and foremost
nedko has quit [Ping timeout: 245 seconds]
<libv>
and samsung shouted "we will do that too!!! ... somewhere next year..."
<hramrach>
voltagex: when you can run a web browser, media player, word processor, and play games on your phone you do need the ram
<libv>
now samsung will really come back in full force
<libv>
so for the next 2 years, this thing is going to continue
nedko has joined #linux-sunxi
<libv>
and there's just not enough people with enough spare time to keep up.
<voltagex>
libv: I see it in all the fragmented ARM communities
<libv>
and samsung is really incapable of learning how to work with open source software.
<voltagex>
libv: *plug, Pi, ODROID, UDOO, Ouya. No one can keep up and there's not enough focus anywhere - except maybe the Pi due to the size of the community
<libv>
voltagex: we're right behind the rpi
<libv>
voltagex: and ahead of everyone else
<libv>
which is why it is important to stick with it
<libv>
and hope that allwinner produces halfdecent SoCs
<libv>
(unlike A31 and A80)
Derry has joined #linux-sunxi
<libv>
if we really want to keep up though, we need to start working on A23
<voltagex>
libv: if I could get my hands on an A23 device, I'd probably devote some mental cycles
<libv>
voltagex: they are all over the place now
<voltagex>
hrm
<voltagex>
I like the Cubie* series
<libv>
voltagex: the really cheap tablets come with a23 now
<voltagex>
libv: I've been burnt by one too many Kickstarter so far
<libv>
voltagex: ebay, amazon
<hramrach>
I personally have not much interest in a23. If AW does some more useful soc that would be nice but a23 is not it. Then again a23 might be a testbed for new hardware features just as 13 was bu who knows
<libv>
dx
<libv>
a23 is where it should be at
<mnemoc>
voltagex: the a80 reference board is also made by cubietech
<voltagex>
libv: any recommendations?
<libv>
it has many bits from a31 taken over
<libv>
voltagex: where do you live?
<voltagex>
Australia, away from everywhere
<mnemoc>
eva announced a kitkat sdk for a23 on linkedin two days ago
<hramrach>
dx does Australia
<voltagex>
took two months for me to get a Udoo board, boy was that a mistake
<mnemoc>
and roumors tell about a quad variant of the a23
<mnemoc>
a43
<mnemoc>
but where is the a40?!
<hramrach>
how much ram on a43 and how wide bus to it?
<voltagex>
I'm actually just looking for a replacement for my Dreamplug, pulled the Cubie1 out of the drawer and came here ;)
<mnemoc>
hramrach: same as a31 i guess
<hramrach>
because that's the single bottleneck on all AW hardware so far except maybe a31
Faisal__ has quit [Remote host closed the connection]
Faisal has joined #linux-sunxi
Derry has quit [Ping timeout: 245 seconds]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
notmart has joined #linux-sunxi
<Faisal>
guys what is this : enable sunxi_is_foo() even for other mach-sunNi in the kernconf
<Faisal>
no help is available
rellla has joined #linux-sunxi
<ccaione>
Faisal: context?
<hramrach>
kernel configuration I guess
<Faisal>
ccaione: yep kernel config
<Faisal>
its just not documented what it is
<Faisal>
so im not sure
t3st3r has quit [Ping timeout: 240 seconds]
<ccaione>
I think that is related to sunxi_is_foo() function that is used to identify the hardware the kernel is running on
<mnemoc>
mach_sunNi() tells you only the family based on build time options, while sunxii_is_foo() gives you up to soc revision level and by hardware detection
<Faisal>
doesnt the kernel use devicetree for that
<Faisal>
oh ok
<mnemoc>
Faisal: mainline doesn't use sunxi_is_foo()
<mnemoc>
Faisal: only legacy and the hybrid kernels have sunxi_is_foo()
<hramrach>
hmm, g_ether only connects afetr disconnecting and reconnecting the cable
<libv>
hramrach: nm on your host?
<hramrach>
no nm
<hramrach>
it does not show as USB device
<libv>
hrm, someone probably should look at the usb otg code to fix that one
<libv>
but i guess you can rootwait around it, no?
<hramrach>
no. it does not connect
<hramrach>
I think somebody complained about OTG being flaky but no test devices at that time
<libv>
hramrach: is this the only tablet you own?
<hramrach>
I also own a Tegra tablet ;-)
* libv
shoots hramrach
<hramrach>
does it have to be a tablet?
<libv>
what other allwinner hw with otg do you have?
<hramrach>
cubieboard
<hramrach>
the issue was supposedly sun5i specific
_whitelogger has joined #linux-sunxi
<libv>
with me, with the standard script.bin, basic u-boot, and sun5i_defconfig, it just worked
<libv>
all of it
mkutsevol has left #linux-sunxi [#linux-sunxi]
<libv>
if i hadn't been stupid and put the script.bin in place (instead of just removing the a10 one), i would've never have looked into usbboot at all
<libv>
and your mmc is clearly recognized, so that's not it
<hramrach>
well, the display does not work
<hramrach>
I kind of expected that to work out of the box so I would fix any remaining issues with display and USB keyboard
<hramrach>
but that's not the case
<hramrach>
with sunxi the display is totally dead
<hramrach>
btw the filesystem on the card was clean so there *was* some boot issue with the SD card
<hramrach>
it never got to writing to it
km4r has quit [Ping timeout: 240 seconds]
<hramrach>
but I have no idea what it was because there was no screen output and no serial
<hramrach>
btw powering off does not work. the tablet stays on
cajg has quit [Ping timeout: 248 seconds]
<libv>
hramrach: the 86vs script.bin is quite similar to the 86vz
<hramrach>
it was written by the same guy, probably
<libv>
my 86vz just brings up the display np with disp/lcd and console=tty0
<hramrach>
well, 86vs does not for some reason
<hramrach>
I would really like to know how to turn it on but can't really measure the signal on the wires
paulk-collins has joined #linux-sunxi
TomiK has quit [Quit: KVirc a enfin son Script! [AndarisScript 4.3 sur www.andaris.net]]
cajg has joined #linux-sunxi
<Faisal>
I'm trying to compile a kernel but it stops here : http://pastebin.com/XtD4FTW0 just wondering if any of you had an idea ?
<hramrach>
something is broken. try a different branch ;-)
<Faisal>
hramrach: thx
<hramrach>
if you said what branch you are using somebody might perhaps reproduce and fix the issue
<hramrach>
or tell you it's obsolete anyway :p
<Faisal>
using the 3.4 branch
<Faisal>
:s
<Faisal>
i might try 3.10
<Turl>
Faisal: 3.4 is fine
<hramrach>
I used to use it and was fine
<Turl>
3.10 is gonna be pretty broken :)
<Turl>
Faisal: what config are you using?
<mnemoc>
no 3.10 yet
<Faisal>
well i did the menuconfig myself, trying to compile it on the cubietruck maybe thats why
<Turl>
Faisal: try make sun7i_defconfig first
<Faisal>
Turl: thx
<Faisal>
Turl: will do
<hramrach>
save your config first
<hramrach>
so you can debug which option caused the problem
<hramrach>
if the config was not too insane it should work
<hramrach>
and ideally kconfig should prevent insane configs
<Faisal>
done
<Faisal>
looking much better now, what make sun7i_defconfig does ?
<hramrach>
replaces your config with minimal kernel that should boot on a20
<Faisal>
i already had all sun7i related config done in the config file
<Faisal>
ok so that means my config is gone i presume
<hramrach>
it might be in .config.old or something
<Faisal>
i mean im not compiling with my options now
<hramrach>
no
<hramrach>
you are compiling the defconfig
<mnemoc>
it's good to start from a known to work base ;-) and then slowly trim it/improve it
pseudomind has joined #linux-sunxi
<Faisal>
mnemoc: right
mkutsevol has joined #linux-sunxi
<mnemoc>
or you'll never know what broke it
selsinork has joined #linux-sunxi
* hramrach
building pre-usb-unification kernel
pseudomind has quit [Quit: pseudomind]
mkutsevol has left #linux-sunxi [#linux-sunxi]
mkutsevol has joined #linux-sunxi
rz2k has quit []
<hramrach>
with 3.4.29 the OTG works flawlessly on sun5i \o/