ericxdu has quit [Remote host closed the connection]
ericxdu has joined #linux-sunxi
<swiftgeek>
apritzel: started to work when i have inserted now fdt_addr_r between each and every line xD
<swiftgeek>
can i somehow do sth like line1 ; line2 ?
<swiftgeek>
heh this gave me new error
<swiftgeek>
[ 8.926642] OF: /soc@01c00000/i2c@01c2ac00/touchscreen@0: could not get #gpio-cells for /soc@01c00000/pinctrl@01c20800/i2c0@0
<apritzel>
why don't you just change your source .dts, compile that and load it?
<swiftgeek>
because new features are added to kernel
<swiftgeek>
(and this one is unlikely)
<apritzel>
but apparently hacking up the DT in this way is problematic as well
<apritzel>
overlays are the answer, but I have no insight in how far the support has come
<swiftgeek>
yeah overlays would be better
vagrantc has quit [Quit: leaving]
<apritzel>
you could just create an .dts file on top, including the board .dts and adding your nodes
<apritzel>
that would make dtc add the proper references
<swiftgeek>
newest error seems to be silead_ts 0-0040: could not find pctldev for node /soc@01c00000/drc@01e70000/ports/port@0/endpoint@0, deferring probe
ErwinH has joined #linux-sunxi
<apritzel>
swiftgeek: this doesn't have to be an error: "deferring probe"
<swiftgeek>
apritzel: it has to because it didn't try to load firmware yet
<tkaiser>
KotCzarny: Also one Solid-Run engineer talked about that the SFP+ connector can be configured to 2.5G mode and they have various reports from users saturating this between two Clearfogs. These beasts are made for throughput :)
<KotCzarny>
not too shabby
<KotCzarny>
but i'm wondering about low write speeds of that test, was that ssd used bottleneck?
<JohnDoe_71Rus>
then chine's holidays end? this year
andi_sp has quit [Quit: andi_sp]
<nove>
mripard: you should make this more public, and announce somewhere free-electrons.com/internships/2017/stage-vpu.pdf
<nove>
to who doesn't knows French, that is a paid internship offer to work in the video engine vpu driver
<nove>
and that is good, and i am happy to not been the "sucker" that would do this driver at zero cost
<nove>
when there is only a gratis software community that only wants gratis things, then the only right thing is to expect to pay the full price for the software
<nove>
as it should be
<Ke>
nove: mostly there is no way to pay for what you actually want though
<Ke>
unless you can pay a full salary for other people by yourself
<nove>
Ke, you mean that for people to stop violating the licenses of the software that they are using, (i must pay them a salary?)
<Ke>
are you referring to allwinner now or what?
ErwinH has joined #linux-sunxi
<nove>
Ke: not only allwinner, also here in this "sunxi" community there are some people that think that "license issues" are to be ignored
<nove>
what i want, is only favorable conditions, in which i could do this driver at zero cost will a smile in my face, this conditions still don't exist
ErwinH has quit [Ping timeout: 245 seconds]
<Ke>
nove: which driver?
<Ke>
mali 400 based video decoder?
<KotCzarny>
w00t?
<KotCzarny>
mali is 3d chip, not video engine
<nove>
for that only 1 of two conditions would need to be true. 1 - allwinner stopping creating license issues, 2 - the community refusal to use license issues software
<Ke>
KotCzarny: generic enough shaders can do compute, I am not following though completely
<nove>
still i can't say that any one of this condition are today true
<nove>
Ke, the driver to the video engine (aka VPU)
medvid has quit [Ping timeout: 252 seconds]
<Ke>
my french tells me that this project would be about reverse engineering, not working with allwinner though?
<Ke>
hmm or that most of the re has been already done
<nove>
Ke, the reverse engineering is already done
<Ke>
=o)
<Ke>
see, I can do french
engideavr has quit [Quit: Konversation terminated!]
<willmore>
If the RE is done, then there is a clean room document describing the hardware. If that's the case, then there are no licensing issues.
<KotCzarny>
ke: i wouldnt' expect any good shaders performance, especially that probably there is only some basic and old version of them in mali4x0 (which itself is old design)
<nove>
the work is only about writing the driver in the mainline acceptable way, and is not so trivial as our hardware is stateless, it requires new still not finalized api support in the kernel
<Wizzup>
MoeIcenowy: on the q8 tablets, any clue if the device boots from the sd card first instead of the emmc?
<Wizzup>
I'm also trying to get the emmc of my tablet in the device tree.
<nove>
willmore: to have antiquated video playback in mainline, what is need is not only the vpu driver, but also everything else, you need sound output and display output with preference for hdmi
<TheLinuxBug>
nove: I see you said something about money related to the work, I am sure if there was someone serious enough about it that they showed real progress there would be some people willing to donate some money for the casue, however, the issue is either the donation doesn't appeal or the person isn't serious about it... so its hard to say people won't pay when ther is no one actually serious about doing it
<TheLinuxBug>
I don't think eveyone expects everything for free, but there has to be a good reason to put money into it also
<jelle>
nice, I'm going to check it out in brussels :)
<nove>
at least has bonus points for, is not hiding or behaving that license issues don't exist
<Wizzup>
jelle: me too!
vagrantc has joined #linux-sunxi
<jelle>
1366x768 though
yann has joined #linux-sunxi
<Wizzup>
jelle: *shrug* :) fine by me
<jernej>
MoeIcenowy: I'm going trough your modifications for A64 U-Boot HDMI driver and I think I found something
<jernej>
MoeIcenowy: Can you try changing DE_MUX0_BASE for DE_MUX1_BASE throughout the code?
reinforce has quit [Quit: Leaving.]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 252 seconds]
<TheLinuxBug>
nove: I guess my point is need to show progress and then ask for some contibution to help fund work. I know I am not alergic to throwing down a few $ to help out, but of course I don't have hundreds or thousands to give, lol (I wish). Though I know I would have great interest in seeing the drivers make it to mainline for Mali and I think it would be a big step for the ARM/Allwinner/SBC business that uses it.
<TheLinuxBug>
I would have also though groups like Lemaker, Cubieboard, Olimex, Xunlong, etc would have interest in investing it it, but I am assuming they don't want the headaches of fighting with Allwinner?
<TheLinuxBug>
thought*
<TheLinuxBug>
nove: but I have been here no where as long as you, so you probably know much better than me
<nove>
TheLinuxBug: I understand what you mean.
<nove>
in the question of the video engine(VPU), the reverse engineering work for the most wanted video codecs was completed years ago
<nove>
again, i presume that this is "showing progress"
<nove>
but, i know what we all want is mainline support
<nove>
but, as our hardware (the video engine) is so simples, it belongs to a category of stateless hardware
<nove>
which, complexes the creation of the driver, as the linux kernel still don't have the need driver infrastructure to write drivers for this type of stateless hardware
f0xx has quit [Ping timeout: 240 seconds]
<nove>
but i know. we could always have a out-of-tree driver, by ignoring the requirements of mainlining
<nove>
as says the above VE_Planning, the out-of-tree driver could be done by ignoring the problem of parsing bitstream in kernel space
cptG_ has joined #linux-sunxi
<nove>
this could be done at the time of writing this VE_Planning (was my plan), also this out-of-tree driver could be backported to the multiples vendor kernels
<nove>
even down to kernel 3.0
<nove>
but i know, we want mainline
<nove>
and also i know, that if there is a out-of-tree driver, the "need" for a mainlineable driver will be reduced
cptG has quit [Ping timeout: 240 seconds]
<nove>
together with the "license issues circus" which demonstrated that there is not a community that justifies the work need for this out-of-tree driver
<nove>
because of this it wasn't done,
<nove>
and in here there is reasons, to say there wasn't progress to show
ErwinH has joined #linux-sunxi
<nove>
but, look what i wrote above. "if there is a out-of-tree driver, the need for a mainlineable driver is reduced"
<nove>
from all the time that i have been here, i could observe that the "sunxi" community is a community that prefers to have cheap hardware than to have software without license issues
ErwinH has quit [Ping timeout: 252 seconds]
<nove>
and because of that, the community is "unwilling" to support (even just verbal support) the work for the software
<swiftgeek>
anyone with Q8 and calipers?
<swiftgeek>
what is the dimension of power jack?
Andy-D has joined #linux-sunxi
<swiftgeek>
oh found 2.5mm x 0.7
<nove>
as the community is only looking for gratis things, to the point of even not helping the people writing the software to not get involved in "license issues"
<nove>
and of course, in this way don't work
<nove>
as if the community gets their gratis out-of-tree driver, nobody will pay for the mainline driver
<nove>
so this is me saying, that not doing the out-of-tree driver was showing progress
apritzel has joined #linux-sunxi
<nove>
and is the best way to "accelerate" the work for a mainline driver
iamfrankenstein1 has joined #linux-sunxi
<swiftgeek>
yay i guess
iamfrankenstein has quit [Ping timeout: 240 seconds]
iamfrankenstein1 is now known as iamfrankenstein
<swiftgeek>
nove: all i can do is pass that url with slides around
<nove>
swiftgeek: which url?
<TheLinuxBug>
nove: thanks for that info, I didn't know you had done so much work on it already
<TheLinuxBug>
nove: I understand your frustration and your words on license issue, obviously most here want it to be open license and not have any infringement issue
<TheLinuxBug>
but that isn't always possible
iamfrankenstein has quit [Quit: iamfrankenstein]
<swiftgeek>
sorry for being ignorant to those issues but what issues are there?
<TheLinuxBug>
nove: do we describe this as 'being between a rock and a hard place" ? ;p
<swiftgeek>
i can understand issues with implementing silicon IP
ErwinH has joined #linux-sunxi
<swiftgeek>
but software?
<TheLinuxBug>
swiftgeek: Allwinner/ARM refuse to release a version of the driver code that isn't some internal license or with GPL so its hard to write code that later if its used in a product you can be sued for because you don't own rights to the drivers.
<swiftgeek>
but this is about open driver
<swiftgeek>
based on reversing thing
<TheLinuxBug>
their concern is if they make a driver that doesn't have GPL that people will assume its GPL license and use it without permission resulting in issue with Allwinner/ARM
<nove>
swiftgeek: that url, take notice that there is a reason that it is in French, but yes do adverting the people doing the right things
<swiftgeek>
but reversing is pretty much done
<swiftgeek>
nove: what reason?
<swiftgeek>
french only?
<swiftgeek>
maybe sb would claim EU citizenship and score it anyway? xD
<nove>
swiftgeek: is only open for french people that can be in site, but that is a question for mripard
<swiftgeek>
nove: i don't see french requirement
<swiftgeek>
only location
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 252 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 252 seconds]
<TheLinuxBug>
nove: thanks for coming and spending time to explain :)
<TheLinuxBug>
I know thanks /= $ but I know I appriciate your time on it :)
ssvb has joined #linux-sunxi
<nove>
TheLinuxBug: no problem, it is alot more easy to explain when there is willingness to listen
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 252 seconds]
IgorPec has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
ErwinH_ has joined #linux-sunxi
<swiftgeek>
TheLinuxBug: ≠
<swiftgeek>
there you go :P
ErwinH has quit [Ping timeout: 252 seconds]
<swiftgeek>
any idea if there is a touchscreen ic with reversed firmware?
<swiftgeek>
like actual cpu firmware, not just data blob
ErwinH_ has quit [Ping timeout: 252 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
nove has quit [Quit: nove]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
<apritzel>
can someone confirm that SD boot is broken with U-Boot 2017.03-rc1?
<apritzel>
reverting b3d2861eb20a ("spl: Remove overwrite of relocated malloc limit") fixed it for me
<vagrantc>
on all platforms?
<apritzel>
tested on Pine64, Opi PC2 and OPi zero
<apritzel>
I guess it applies to all sunxi boards
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
<apritzel>
yup, same on a BPi-M1 (A20)
<apritzel>
I sent an email to the list already, just as a heads up ...
<vagrantc>
looks like there were significant changes for pine64... does it still require jumping through all the boot0 hoops?