<MoeIcenowy>
but A31s/A23 have still nobody working on
<netchip>
Ah
<netchip>
I have no experience with kernel development yet, so I don't think I'm of any help (is that the right expression?)
leviathanch has quit [Ping timeout: 252 seconds]
<montjoie>
netchip: for a begin, try eudyptula challenge
<netchip>
montjoie: seems cool
<netchip>
I have experience with kernel compiling and patching for Android devices, but nothing more advanced than that
<netchip>
KotCzarny: BTW, what's the thing with all the different Cortex A* nowadays? I know the Cortex A8/A9/A53, but what's this one for example? Rockchip RK3288 ARM Cortex-A17
<rellla>
montjoie: nice thing, the challenge. sadly i'm standing at #5 due to lack of time :(
<MoeIcenowy>
montjoie: what's this?
<KotCzarny>
basically a53 is few a7 cores and some more
<wens>
the problem is getting stereo controls in first
<KotCzarny>
um, no. scratch that
<MoeIcenowy>
wens: isn't it sun8i-analog-codec driver by Mylene ?
<wens>
MoeIcenowy: no, i wrote that driver, but didn't test it, and i sent it to them, since they are working on A33
<wens>
and there's a specific structure i want for the codecs
<wens>
basically forcing my will on others :p
<MoeIcenowy>
so the driver is written by you and tested by Mylene ?
mzki_ has quit [Quit: leaving]
popolon has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
<wens>
MoeIcenowy: yes
<netchip>
KotCzarny: A7 is the power efficiënt variant right?
<wens>
netchip: yup
<apritzel>
netchip: Cortex-A8: first ARMv7 implementation from ARM Ltd., single core only, no LPAE or virt extension (read: limited to 4GB phys mem and no KVM/Xen)
<KotCzarny>
netchip: nope. a53 is 64bit
<apritzel>
Cortex-A7: power efficient, in-order, up to four cores, LPAE + virt extension
<netchip>
Ah
<apritzel>
Cortex-A15: performance optimized, out of order, LPAE + virt extension, can be paired with A7 in big.LITTLE systems
<apritzel>
Cortex-A12/A17: more power _efficient_ than A15, more powerful than A7, LPAE + virt
<MoeIcenowy>
apritzel: I have a problem: can Cortex-A53 virtualization simulate a A7?
<apritzel>
MoeIcenowy: yes
<apritzel>
well, at least AArch32
<MoeIcenowy>
for example, on Pine64, use KVM to simulate a RPI2
<apritzel>
kind of
<MoeIcenowy>
does A7 have some special quirks?
<apritzel>
KVM doesn't cover any special register a certain core has
<netchip>
So for performance A15 > A12/A17?
<apritzel>
netchip: good question ;-)
<apritzel>
I think A12 can be clocked higher in the same setting, so they maybe faster
<apritzel>
MoeIcenowy: KVM only really supports the same core in the guest
<MoeIcenowy>
apritzel: ah-oh
<apritzel>
MoeIcenowy: but: it seems to work at least with Linux to pretend to be another core
<MoeIcenowy>
it's worse than the virtualization on x86
<apritzel>
since Linux usually doesn't access IMPDEF registers
<apritzel>
MoeIcenowy: no
<netchip>
apritzel, ah OK
<netchip>
And if I wanted high performance 64-bits, I should go for a A53?
<MoeIcenowy>
but on x86 I can make my IvyBridge simulate a 486...
<KotCzarny>
though i hate those presentation type pages which can be done in one paragraph
<KotCzarny>
;)
<apritzel>
netchip: in terms of U-Boot support AW SoCs are quite good as well (again thanks to the community, not AW)
<apritzel>
KotCzarny: indeed, these type of web pages are getting increasingly annoying, but seemingly more popular
yann has joined #linux-sunxi
<MoeIcenowy>
Nowadays this kind of guide-like introductions are less and less
<apritzel>
netchip: "waiting" for the next best thing never really works out in computing (because then you wait forever)
<KotCzarny>
true that
<apritzel>
netchip: you buy what seems to be the best exactly when you need it
<KotCzarny>
good enough [tm] is all you can get
<KotCzarny>
and im happy with my bpi-r1, bpi-m1, two opipc and one opi+2e
<KotCzarny>
:)
<MoeIcenowy>
good enough... I remembered the talk about Chabuduo “差不多” this afternoon in UTC+8
<apritzel>
MoeIcenowy: was just thinking the same ;-)
<netchip>
apritzel: but I don't need it :p it's just a gadget to mess around with
<netchip>
I have a RPi 1 B right now
<netchip>
And I don't want a RPi 3 because it lacks uboot
<MoeIcenowy>
I do have a RPI3 now, but I never unpacked it...
<netchip>
The Olimex A20 is a serious candidate for its great community support and SATA
<KotCzarny>
yup
<KotCzarny>
sata is nice
<KotCzarny>
unique too
<KotCzarny>
because its real sata port
<KotCzarny>
and not crappy usb bridge
<netchip>
MoeIcenowy, I have a RPi 1 and I have only used it for an IRC bouncer, and for resurrecting my father's NAS (Netgear RN102) yesterday
<netchip>
KotCzarny, exactly. Might be handy in the future
<apritzel>
netchip: there is U-Boot support for the RPi3
<apritzel>
it's just that the RPi foundation does not use it :-(
<netchip>
apritzel, oh, I didn't see it... Derp
<apritzel>
netchip: $ ls configs/rpi_3_*
<netchip>
I'll check it when I come home, I'm now on my phone and am going to my work in a bit :)
popolon has joined #linux-sunxi
<netchip>
I guess I'll order a Pine64+... The drawbacks seem mainly software related, and as mainline Linux support is added, those problems will fade away I think?
<KotCzarny>
depends on your needs
<netchip>
They aren't graphics related
<netchip>
For that I could always buy a RPi 3
<netchip>
I2C, SPI, GPIO, UART and well maintained kernel code/uboot code is what I'm buying it for
<netchip>
Anyway I have to go
<netchip>
Will comeback in a few hours :)
<netchip>
Thx everyone!
<buZz>
pi3 is proprietary poop
<buZz>
like all raspis
<buZz>
imho more toxic than AW hardware
<netchip>
Ah
<KotCzarny>
a20 socs are usually paired with batteries
<apritzel>
buZz: what I find annoying about RPi is that the official kernel (which virtually every one uses) is not really picking up upstream contributions
<buZz>
apritzel: its just a waste of time and effort, imho
<apritzel>
buZz: if you browse their kernel repo you will find many places where their original code coexists with the upstreamed version, but they ignore the latter
<buZz>
fuckers
<buZz>
lets just burn boredcom's offices
<apritzel>
buZz: I wouldn't go that far ;-)
<buZz>
ok, just pour cement in the exhaust of all their cars
<apritzel>
but it's annoying and goes against the good upstream reputation that they enjoy
<buZz>
which good reputation?
<buZz>
first i hear about it
<apritzel>
well, right, not upstream, but OpenSource reputation
<buZz>
they arent opensource
<apritzel>
gfx?
<buZz>
well, their kernel is , but i doubt its by their choice
gianMOD has quit [Remote host closed the connection]
<apritzel>
they are not bad and they are improving
<buZz>
they dont even have a distro
<buZz>
'raspbian' is community built, not by them
gianMOD has joined #linux-sunxi
<apritzel>
buZz: a board vendor shouldn't have a specific distro, really
<apritzel>
it's a users choice to pick a distro
my123_ has joined #linux-sunxi
<buZz>
well sure, but this isnt a board vendor nor do their users think they have choice
<buZz>
their product is a 'community'
<buZz>
with proprietary hardware that gets no specs or docs or source
<KotCzarny>
but has toys to play with
<KotCzarny>
codebase too
<jelle>
isn't the rpi2 almost mainline
my123 has quit [Ping timeout: 245 seconds]
<buZz>
jelle: if you dont want graphics
<buZz>
jelle: or audio
<jelle>
buZz: there is a foss driver for it in the works I think
<apritzel>
jelle: yes, but as mentioned they don't use it ...
<buZz>
nope, just for the 3D
<buZz>
you wont get 2D
<jelle>
apritzel: I hope they will... yeah
<jelle>
well benefit of the AW design is the easy boot support too I guess
<Amit_T>
But Broadcom is no more Broadcom, Its Avago now .
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<buZz>
avago is part of broadcom
<buZz>
> Broadcom Limited (formerly Avago Technologies) is an American designer, developer and global supplier of products
<silviop>
My tablet not boot direct from sdcard obviously(only installed android work), what can do for deeper debug ?
<silviop>
I see that there is a early uart debug option in .config but ask parameters that i don't know.
<NiteHawk>
silviop: the "libusb usb_bulk_send error -7" may not neccessarily indicate an error, iirc. with an sdcard breakout, you might have to adjust the UART settings in U-Boot to get usable output on the serial console. you might also have to prevent U-Boot from trying to detect/setup an actual card (mmc0), as that's unavailable / reused for the uart
<silviop>
do you know .config values ? I changed only (UART0 on MicroSD breakout board)
<silviop>
i suppose that it disable all interfering things
DullTube has joined #linux-sunxi
<Net147>
I got Mali r3p2 framebuffer driver working on mainline
<NiteHawk>
silviop: yes, that should do. unfortunately, i'm not experienced with using sdcard breakout
<KotCzarny>
is sdcard in slot or out?
<silviop>
no only breakout without sdcard (i boot from usb/fel)
<ssvb>
The EULA used to be a major pain. And the availability of suitable blobs with suitable licenses has not been fully resolved yet.
<ssvb>
It's also one of the reasons why we had a half-decent Mali drivers support at least since 2013, but had to force the users to install everything manually at their own risk via the step by step instructions at https://linux-sunxi.org/Mali_binary_driver
<Net147>
ssvb: so can r3p2 be used in products?
<ssvb>
If ARM had their EULA and blobs distribution logistics in a better shape, then ARM Linux distributions could have had the Mali userland blobs packaged since a very long time ago
orly_owl has quit [Ping timeout: 260 seconds]
<ssvb>
Net147: no, r3p2 has license issues
<ssvb>
one issue is that we even did not receive any EULA notice at all, and another issue is that the r3p2 EULA (as used by ODROID and the others) used to forbid blobs redistribution, which is a no-go
<ssvb>
now the EULA is fixed since r6p0, but we are yet to receive the Mali blobs with this new license for our hardware
<ssvb>
Pine64 received the r6p0 blobs from Allwinner without EULA, in the very same way as we had it with the r3p2 blobs a long time ago
gianMOD has quit [Remote host closed the connection]
<ssvb>
but I don't feel like even wasting any of my time debunking the bullshit from this dude
<Net147>
which part of the EULA in particular forbids redistribution?
<ssvb>
have you read it?
The_Loko has joined #linux-sunxi
Worf has quit [Quit: Konversation terminated!]
gianMOD has joined #linux-sunxi
<Net147>
I am looking at the odroid r5p1 license
<ssvb>
try to search the keyword "distribute" in the EULA text
<wens>
mripard: :)
<Net147>
"2.2.5 You will not delete or in any manner alter the copyright, trademark or other proprietary rights notices appearing on the Software or documentation as delivered or distributed by"
<ssvb>
"Hardkernel hereby grants Licensee a limited, non-exclusive, non-transferable, irrevocable, restricted right to use the Software in object code form only on the Computer System solely for the purpose of developing application software for Hardkernel devices using the Samsung Exynos 5422 System-on-a-Chip"
<ssvb>
"non-transferable" and "using the Samsung Exynos 5422 System-on-a-Chip" make it worthless for us
<Net147>
okay but the r3p2 blob is from cubietech, not hardkernel
<ssvb>
would you be happy if it had the EULA which stated "non-transferable", "solely for the purpose of developing application software" and "Allwinner A20 System-on-a-Chip"?
<ssvb>
Linux distributions still would have no right to package it
<ssvb>
only the r6p0 blob EULA is somewhat sane
<Net147>
I am referring to the use case of someone developing an embedded product that includes the blob in the filesystem.
<ssvb>
well, I'm not a lawyer myself, but you can probably hire a lawyer who could provide consulting on this matter...
<ssvb>
my interpretation is that you are not allowed to distribute any mali blobs older than r6p0 to other people, unless you have some other agreement with ARM
<Net147>
thanks for the info
<Net147>
hopefully we don't have to wait too long for suitable blobs
<ssvb>
BTW, if ODROID has the r6p0 blobs with a permissive EULA, then we could use them too
<mripard>
ssvb: some r6p0 had the restrictive license too
<mripard>
it changed halfway through it
cnxsoft has quit [Quit: cnxsoft]
<Net147>
I don't think they updated beyond Mali r5p0 for their now discontinued boards with Mali-400
gianMOD has quit [Remote host closed the connection]
<Net147>
though I guess it wouldn't hurt to try contacting them
florianH has quit [Quit: Connection closed for inactivity]
gianMOD has joined #linux-sunxi
<Net147>
but they may be restricted to Samsung only
gianMOD has quit [Client Quit]
reinforce has quit [Quit: Leaving.]
The_Loko has quit [Quit: Leaving]
mozzwald has quit [Quit: leaving]
gianMOD has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
The_Loko has joined #linux-sunxi
mozzwald has joined #linux-sunxi
psychedup74 has joined #linux-sunxi
gianMOD has quit [Remote host closed the connection]
apritzel1 has joined #linux-sunxi
jernej has joined #linux-sunxi
florianH has joined #linux-sunxi
<KotCzarny>
i wonder what is the reason not to free the old devices
gianMOD has joined #linux-sunxi
gianMOD has quit [Ping timeout: 250 seconds]
<terra854>
Money i guess
Amit_T has quit [Ping timeout: 244 seconds]
Amit_T has joined #linux-sunxi
Putti has joined #linux-sunxi
reinforce has joined #linux-sunxi
psychedup74 has quit [Quit: leaving]
leviathanch_ has quit [Remote host closed the connection]