<oliv3r>
hramrach: cnx-software linked it, Greg on Linaro 'i don't take your patch'
ntm12 has joined #linux-sunxi
ntm12 has quit [Client Quit]
tinti has joined #linux-sunxi
<tinti>
hi can someone help me with arch/arm/mach-sun4i/pm/standby/standby.c +61 ?
<wingrime>
what extactly
<wingrime>
*exactly?
<wingrime>
hramrach: strange, drm ol
<wingrime>
*ok
<wingrime>
but mesa use "software raster"
<hramrach>
wingrime: that's because mesa does not have a mali module
<wingrime>
[ 31.556] (EE) AIGLX error: dlopen of /usr/lib/arm-linux-gnueabihf/dri/sunxi-mali_dri.so failed (/usr/lib/arm-linux-gnueabihf/dri/sunxi-mali_dri.so: cannot open shared object file: No such file or directory)
ganbold_ has joined #linux-sunxi
<ssvb>
AIGLX error is not a problem
<ssvb>
wingrime: did you run "modprobe ump", "modprobe mali", "modprobe mali_drm"?
<wingrime>
loaded as lsmod answers
<ssvb>
wingrime: and if mesa is still getting in the way in ubuntu, you may need to do "mv /usr/lib/arm-linux-gnueabihf/mesa-egl/ /usr/lib/arm-linux-gnueabihf/.mesa-egl/" as explained at the bottom of http://linux-sunxi.org/Binary_drivers
<ssvb>
wingrime: what kind of GLES application are you trying to run?
<wingrime>
none: I just see xlog
<wingrime>
and see arror
<wingrime>
and "Bottom" not helped
jelly-home has quit [Ping timeout: 255 seconds]
<wingrime>
ssvb: make test fails /usr/lib/gcc/arm-linux-gnueabihf/4.6/../../../../lib/libEGL.so: undefined reference to `ump_switch_hw_usage_secure_id'
hurtigbuffer has joined #linux-sunxi
<libv>
wingrime: what version of the kernel driver are you running?
<libv>
wingrime: what version was autodetected?
tinti has quit [Ping timeout: 248 seconds]
<oliv3r>
anybody gone over the new A20 source drops yet? Seems like they now duplicate dram init, in march-sun7i/dram and march-sun7i/pm/standby/dram :S
<wingrime>
r3p0
<wingrime>
kernel 3,0
<wingrime>
git 5 days ago
<ssvb>
libv: might be yet another case of a creative interpretation of installation instructions, which involved tossing random libraries at random places
<ssvb>
and now asking for support after the system is fubared
<ssvb>
wingrime: is libEGL.so symlinked to libMali.so?
<ssvb>
can you run ldd for it and paste the log somewhere?
<ssvb>
the sunxi-mali repository provides the binary blobs and should do all the necessary job to install them
<libv>
ssvb: we need packages.
<libv>
ssvb: to stop people from pulling those sorts of stupid stunts
<wingrime>
ssvb: agree, my linaro now like SLACKWARE
<libv>
wingrime: i believe i wasted enough time already with making sure that there is a halfway workable and stable solution there
<libv>
i am not going to debianize this as well
<wingrime>
is there better way than : "make install"
<libv>
wingrime: no
<libv>
wingrime: and count yourself really really lucky that you already have that
<libv>
wingrime: you also have the ability to provide prefix= at the command line
<wingrime>
libv: I have to thank you for you work
<libv>
there's a big fat readme, and an equivalent wiki page
<libv>
there's headers
<libv>
there's a ready to use testcase
<wingrime>
and I still can make it fly after all
<libv>
wingrime: it's already amazingly foolproof
<libv>
wingrime: and the issues mostly depend on what failed attempts you did before
<libv>
wingrime: find the other libraries installed in your system, and fully remove them.
<libv>
then, go and use make config
<libv>
and check whether it autodetected everything correctly.
<libv>
ssvb: 5 years and a few months ago, i and egbert and emmes threw out radeonhd
<libv>
it took around 4 weeks, and then hans niedermann turned up
<libv>
and he fixed all the issues with build system, and provided debian/ for us
<libv>
and he kept it up until the project was dead
<wingrime>
I make sumlinks manualy
<libv>
what happened in those 5 years that people with such a mindset, who don't need to play with C but instead like focusing on making the raw stuff actually useful for real people, can no longer be found?
<libv>
wingrime: clean it all up.
<libv>
wingrime: and then use make config
<libv>
and check its output
<oliv3r>
gah, we have linux-sunxi and some u-boot wiki pages, but they point to github's wiki pages, which are different for hno and sunxi. How confusing :)
<libv>
oliv3r: yeah, the wiki needs a lot of bashing into shape
<libv>
Mali400 should be removed, it contains little relevant information
<hramrach>
it still contains the list of libraries at the very least
<oliv3r>
libv: yeah, i think it should live on linux-sunxi :p; to be centralized and only 1 copy. Gonna build u-boot now and see how up to date the manual is
<libv>
oliv3r: smashing, thanks :)
<libv>
hramrach: this list of libraries is what kept me from removing the page altogether
anunnaki has quit [Ping timeout: 248 seconds]
<libv>
hramrach: but i rather doubt that people use anything but r3p0 today
<wingrime>
we have to remove installation instructions from mali400
<hramrach>
libv: the Binary_drivers page is very incomplete compared to the Mali400 instructions
<libv>
hramrach: r3p0 is also the only one where the egl desctructor is properly neutered
<hramrach>
it does not list the required libraries to build libump nor the lication of libdri2
<hramrach>
*location
<hramrach>
oh, the required libraries are listed in the text
<libv>
hramrach: all of that quickly becomes very distribution specific, but then, most of what i have written is pretty linaro specific as well
<libv>
hramrach: packaging...
<libv>
someone needs to spend some time adding debian/ to sunxi-mali and xf86-video-mali
<hramrach>
so people not running linaro still need to refer to the generic Mali400 instrucions to build the libraries
<libv>
or someone should do the work and provide .spec files
<hramrach>
adding debian/ to xf86-video-mali should be trivial
<hramrach>
but the libraries are problematic
<libv>
why?
<hramrach>
how do you package fb+x11 libraries in a sane way?
<libv>
separate packages
<hramrach>
then nothing installs
<libv>
debian/ allows for that
<libv>
?
<libv>
libmali-fb is an end target
<libv>
libmali-x11 would be a dependency of xf86-video-mali
<wingrime>
libv: same again
<hramrach>
ok, you have a console application and a x11 application
<wingrime>
after full cycle
<hramrach>
to install console application do you remove X?
<libv>
hramrach: this is how the sunxi world works.
<libv>
hramrach: this is how the libs work.
<libv>
hramrach: we do not have the source
<libv>
we are stuck with this.
<libv>
and no other driver does this shit properly either (or yet)
<libv>
because the embedded world didn't need it so far
<hramrach>
mesa does
<libv>
hramrach: oh fuck off.
<hramrach>
:>
<libv>
hramrach: how dare you even state that to the guy who spent the last two years REing the shit out of this
<libv>
hramrach: deal with the binary situation, or stfu.
<hramrach>
mesa does have loadable modules for output to either X11 or or fb ;-)
<libv>
hramrach: and now stop living in a dreamworld
<libv>
and face the facts.
<libv>
and stay on topic and on course.
<libv>
don't go off on this stupid tangent
<libv>
we have these libs
<libv>
and we better deal with it.
<libv>
and we really should count ourselves unbelievably lucky that we have these libs
<hramrach>
they suck so no sane way to package them
<libv>
hramrach: you're dead weight
<hramrach>
libv: you write cool code
<hramrach>
but are very annoying to deal with
<libv>
hramrach: you fail to deal with facts and fail to work towards any solution
<hramrach>
I tell you that mesa has GLES libraries that work for both framebuffer and X11 and you start insulting me
<hramrach>
now who fails to deal with facts
<libv>
hramrach: what mesa does or does not do is absolutely irrelevant
<libv>
and has absolutely no relationship with whether these libraries can be packaged or not
<libv>
they can be packaged
<libv>
but you get to stick to either fb EGL or x11 EGL afterwards
<hramrach>
but however you package them it will be ugly
<libv>
which suits 99.9% of all users just fine.
<libv>
it's fucking binary drivers.
<wingrime>
libv: Lima as 2d may works good
<hramrach>
then I ma not the parson to package the junk because it would not suit me
<wingrime>
you can use textures as windows
<libv>
hramrach: yeah, that would be doing something useful and practical
<libv>
hramrach: did you see the x11 port of test btw?
<hramrach>
no
<libv>
what an impossible and tough task that was...
* libv
goes off to test cwabbotts other vertex shader, in the hope of running q3a on open source compiled vertex shaders later today
torindel has quit [*.net *.split]
MadSpark has quit [*.net *.split]
jochensp has quit [*.net *.split]
rm has quit [*.net *.split]
rm has joined #linux-sunxi
jochensp has joined #linux-sunxi
torindel has joined #linux-sunxi
<oliv3r>
can we detect FEL mode via dmesg/lsusb (on the host?)
<RaYmAn>
probably, but why would you? (as opposed to 'proper' detection?)
<mnemoc>
oliv3r: on lsusb you'll see a line without description
torqu3e has quit [Read error: Connection reset by peer]
<mnemoc>
don't remember the exact id
<oliv3r>
ok, i'll double check that (my tablet is in a reboot loop, won't even loot the kernel as far as I can tell)
<oliv3r>
I doubt it's trying to go to FEL mode and back though, so I think it boots fine, but the kernel/nanda may be corrupt
<wingrime>
mnemoc: in new source drop in your repo I find that AW make difference between MEM and STANDBY pm modes
<wingrime>
new mode calls "super standby"
<mnemoc>
I remember the "super standby" thing in a10 as well
<mnemoc>
wingrime: but I haven't got time to look at the new code yet. need to finish sanitizing a{10,13,10s}-dev first
torqu3e has joined #linux-sunxi
<mnemoc>
wingrime: one interesting thing is that the A31 has a companion core (openrisc, ar100) caring about the "little things"
<oliv3r>
sounds like an Embedded Controller as found in laptops?
<mnemoc>
haven't reviewed it in deep, only what I saw while rebasic/cleaning the history
<wingrime>
mnemoc: realy intersting .... but why thay add MIPS , I not even imagine how it builds crossplatform
<mnemoc>
they build is separated steps. pm/standby was not intended to be compiled in-tree for a10/a13 either
<mnemoc>
look at their build.sh thing
<oliv3r>
mnemoc: in sunxi-boards, we keep original 'dumps' of fex files, right? even if they are downright stupid (for the inital commit)
<mnemoc>
also, ar100 has .c/.h and a blob, but it seems all the code to compile the blob (.code) is already there
<wingrime>
thats good
<wingrime>
but they didn't add my TS
<mnemoc>
oliv3r: the original is the reference starting point, then we fix dram_para, and then disable stupid things
ganbold_ has quit [Remote host closed the connection]
<oliv3r>
I see there's very few tablets, so I guess my fex file can be used as the 'series default', on what shall I base it's name?
<oliv3r>
It's a Yarvik Tab264, Android uses the momo9 as it's "mother"
<mnemoc>
oliv3r: there is no naming rule. make a wiki page for it, with specs, pictures and location of the pads, and tell the chosen board name there
<oliv3r>
the tablet IS listed on this page as a 'clone' allready, so making a new page seems redundant
<mnemoc>
#REDIRECT ?
<mnemoc>
anyone, if we already have it as "bora", improve that one :)
<oliv3r>
fair nuff, but have each clone use an indicidual fex file then?
<mnemoc>
anyway*
<mnemoc>
diff both and see if the clones are .fex compatible too
<oliv3r>
I'll add some pics, still don't know what to name the fex file :p
<oliv3r>
I don't have fex files for those (in git)
<mnemoc>
many times the "clones" aren't clones, but rebrands
<mnemoc>
rebrandings*
<oliv3r>
well, we don't have that many fex files to begin with, it could be 'compatible' with one of the coby's, but how would I know
<oliv3r>
yep, seems none of those have a fex file
<oliv3r>
so i'll just name it yarvik_tab264.fex and mention that on the wiki
<mnemoc>
if we don't have u-boot/fex for this "bora", let's take yours and rename the Bora page
<oliv3r>
u-boot/fex? you mean sunxi-boards/fex!
<mnemoc>
u-boot-sunxi needs to know about the board too :)
<mnemoc>
and we try to keep the same name on both
<oliv3r>
u-boot-sunxi/board/*?
<mnemoc>
yes, check the history
<oliv3r>
ok, lots more there
<oliv3r>
i take it boards.cfg should contain anything from board
<mnemoc>
fexc -O u-boot will take the (verified) dram_init section in a .fex into a board.c for u-boot
<mnemoc>
boards.cfg is the "index"
<oliv3r>
good, i can grep that then
<mnemoc>
and adds fancy -DFOO at the end of the line
<oliv3r>
so, to summarize (i'll check the wiki and supplement it where needed), get original fex file, add it to sunxi-boards git. Optionally clean it up. Get dram parameters from running the tool (i have that on the ML somewhere luckly), modify the fex file and commit. then generate u-boot board.c for u-boot-sunxi
<mnemoc>
and improve the "Submiting boards" wiki page :)
<oliv3r>
that was step 0! :p 'supplment the wiki where needed'
<oliv3r>
though, yes, the wiki is a bit too general
<oliv3r>
hopefully, I should get all that done without actually powering up the tablet :) (mem paramters are on ML, script.bin is in the extracted livesuit image)
<oliv3r>
oh! submitting board I wrote initially :p
ZaEarl has joined #linux-sunxi
<mnemoc>
*g*
<oliv3r>
lol, yeah i'll fix it up
jquip has joined #linux-sunxi
<jquip>
Hi all :)
jquip has quit [Read error: Connection reset by peer]
<wingrime>
funny in zet6221 disasemply (code that we not have) "license=GPL v2"
<mnemoc>
for them it's just a token required to get the modules working
jquip has joined #linux-sunxi
jquip has quit [Read error: Connection reset by peer]
jquip has joined #linux-sunxi
hansg has quit [Remote host closed the connection]
jquip has quit [Read error: Connection reset by peer]
jquip has joined #linux-sunxi
jquip has quit [Read error: Connection reset by peer]
jquip has joined #linux-sunxi
ZaEarl has quit [Ping timeout: 264 seconds]
jquip has quit [Ping timeout: 245 seconds]
jquip has joined #linux-sunxi
torqu3e has quit [Quit: torqu3e]
hurtigbuffer has quit [Read error: Operation timed out]
<Turl>
mnemoc: so the ar100 is a companion core, I was wondering what was it the other day :|
<wingrime>
so my new driver at least feels toches and recives interupts
hurtigbuffer has joined #linux-sunxi
christopher has joined #linux-sunxi
christopher is now known as Guest46872
jochensp has quit [Quit: WeeChat 0.4.0]
rellla has quit [Remote host closed the connection]
torqu3e has joined #linux-sunxi
mab has joined #linux-sunxi
jochensp has joined #linux-sunxi
rellla has joined #linux-sunxi
jquip has quit [Remote host closed the connection]
<wingrime>
dumping i2c regs so funny
<wingrime>
and see what is changes when you touch keys
mab has quit [Ping timeout: 245 seconds]
<mnemoc>
i2c pr0n
<wingrime>
mnemoc: intresting feature one finger one bit , ts remember release order !
<RaYmAn>
that's a decent ts then!
<RaYmAn>
most cheap ts's lack individual finger tracking
<RaYmAn>
you just get a bundle of points in some order
<wingrime>
it some cheap MCU
<wingrime>
inside
mab has joined #linux-sunxi
MadSpark has joined #linux-sunxi
drachensun has joined #linux-sunxi
<drachensun>
which branch has the latest allwinner code? is the latest wip/lichee3-sunxi/import-a10s-sdk ?
<drachensun>
I need to bring over a new camera driver and there is no news bringing an older leak if there is a newer one to merge in
mab has quit [Ping timeout: 248 seconds]
<mnemoc>
drachensun: lichee/a10s-dev in my github is the same code buy with pristine -dev history
<drachensun>
ok, perfect
<mnemoc>
beware that branch is not santized
<mnemoc>
so you'll find funky encodings, ^Ms and sick whitespace issues
<mnemoc>
will try to push a sanitized import/lichee-3.0/a10s-dev in the next days
<drachensun>
ok
<drachensun>
turns out the A10s branch doesn't have the camera drivers
<drachensun>
which I guess makes sense, the A10s is for sticks
<drachensun>
but the ionik source has them
<mnemoc>
lichee/a10-dev and lichee/a13-dev
<mnemoc>
both complete -dev up to their 1.5 release, same as i-oink but with detailed history
wingrime has quit [Ping timeout: 260 seconds]
rellla has quit [Read error: Connection reset by peer]
zub_ has joined #linux-sunxi
ssvb_ has joined #linux-sunxi
<techn__>
hramrach: which suspand patches should I apply?
<techn__>
or is there coming v2
Liberato1 has joined #linux-sunxi
zub has quit [Ping timeout: 246 seconds]
Liberator has quit [Ping timeout: 246 seconds]
ssvb has quit [Ping timeout: 246 seconds]
rz2k has quit []
<hramrach>
techn__: the wingrime's v2 and the 2/2 in my series I guess
<hramrach>
I would merge it but google mundes patches so amything from the ML is useless for me :s
torqu3e has quit [Quit: torqu3e]
titankiller has joined #linux-sunxi
<titankiller>
Why is it that my IR irq gets called even when my remote is not pressed
<mnemoc>
hramrach: I save-to-file patches on gmail using imap just fine...
<titankiller>
when i log at the entrance of irq handler , the kernal logs show atleast a handler getting called randomly
torqu3e has joined #linux-sunxi
<Turl>
mripard: I just moved the code to "hardware numbers" and documented all the #s, so far everything works still
<Turl>
mripard: tag sunxi-clk-9 is there with all the unsquashed fixups if you want to compare with the previous set
<mripard>
Turl: ok, great
n01 has joined #linux-sunxi
<mripard>
I'm exhausted, I'll take a look at it during the week
Guest46872 is now known as anunnaki
<Turl>
mripard: yeah, no hurry :)
<Turl>
mripard: wemac seems to be working ok, it's pretty fast judging by ping time
<Turl>
mripard: round-trip min/avg/max = 0.000/0.000/0.000 ms