<MoeIcenowy>
but Olimex laptop seems to have only 1GiB DRAM and 4GiB eMMC...
BenG83 has quit [Ping timeout: 255 seconds]
Mr__Anderson has joined #linux-sunxi
chlorine has joined #linux-sunxi
BenG83 has joined #linux-sunxi
matthias_bgg has quit [Read error: Connection reset by peer]
matthias_bgg has joined #linux-sunxi
Wizzup has quit [Quit: brb]
Gerwin_J has joined #linux-sunxi
Wizzup has joined #linux-sunxi
chlorine_ has joined #linux-sunxi
<fest>
why don't any of them expose GPIOs?
chlorine has quit [Ping timeout: 252 seconds]
reev has quit [Read error: Connection reset by peer]
velly has quit [Ping timeout: 252 seconds]
chlorine_ has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
popolon has joined #linux-sunxi
reev has joined #linux-sunxi
reev has quit [Max SendQ exceeded]
<buZz>
probably because they hate us
reev has joined #linux-sunxi
velly has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
velly has quit [Max SendQ exceeded]
chlorine has joined #linux-sunxi
Worf has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
velly has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
velly has quit [Max SendQ exceeded]
chlorine has joined #linux-sunxi
velly has joined #linux-sunxi
paulk-elm_ has joined #linux-sunxi
chlorine has quit [Ping timeout: 240 seconds]
velly has quit [Max SendQ exceeded]
velly has joined #linux-sunxi
velly has quit [Client Quit]
paulk-elm has quit [Ping timeout: 240 seconds]
reev has quit [Read error: Connection reset by peer]
<MoeIcenowy>
wens: finally initial PLL_CPUX tweaking worked by gate PLL_CPUX when changing it
reev has joined #linux-sunxi
<wens>
good
<MoeIcenowy>
P.S. I wonder whether the restriction of M is needed on A33
<MoeIcenowy>
as A33 seems to used M in its BSP _tbl.c
<MoeIcenowy>
but from H3 on no M is used (H3/V3s/A64/H5/R40)
<wens>
if gating PLL when changing it (and reparenting away when it's gated) fixes the issue
<wens>
i see no reason not to use M
<wens>
though we could rework nkmp clock to prefer using multipliers
<MoeIcenowy>
wens: nope gating PLL is not the key
reev has quit [Ping timeout: 252 seconds]
<MoeIcenowy>
M shouldn't be used on H3+ SoCs
<MoeIcenowy>
see BSP tables
chlorine has joined #linux-sunxi
cptG_ has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
chlorine_ has joined #linux-sunxi
cptG has quit [Ping timeout: 240 seconds]
yann-kaelig has joined #linux-sunxi
reev has joined #linux-sunxi
<wens>
unless there is some proper explanation given, i would not blindly trust the BSP
<wens>
it is a reference, and just like any other, should be verified
<MoeIcenowy>
wens: and megous did verified it
<MoeIcenowy>
without looking at BSP and get the same result
<wens>
did his experiment also gate the pll between changes?
<wens>
switch to 24M, gate, set new rate, ungate, switch back
<MoeIcenowy>
seems to be no
<wens>
then can you try it?
<MoeIcenowy>
ok
reev has quit [Ping timeout: 252 seconds]
Gerwin_J has quit [Quit: Gerwin_J]
reev has joined #linux-sunxi
reev has quit [Max SendQ exceeded]
<wens>
the reason why using M can lead to hangs was explained in detail before, it's basically how PLLs are implemented
<MoeIcenowy>
?
<wens>
the dividers are likely counter type dividers after the PLL, i.e. post-dividers
<wens>
multipliers are dividers in the feedback loop, changing them would require the PLL to reacquire lock
<wens>
or if you change them too much, or it goes out of range, it just stops working
reev has joined #linux-sunxi
<wens>
gating it would likely reset it
reev has quit [Ping timeout: 240 seconds]
<MoeIcenowy>
OK this time by gating the clock when changing it seems to be working
<MoeIcenowy>
let me run for longer time
<msev->
tkaiser, can i please ask you here some things...this log2ram, can I use that to have a sqlite database inside RAM (with allocated maximum space of RAM that the database would take)? Also does this log2ram allow "flush" of this file to SD card at regular intervals say once per day?
<MoeIcenowy>
wens: how to model it in Linux kernel?
<wens>
MoeIcenowy: you can't really
<MoeIcenowy>
WHAT??!!
<wens>
MoeIcenowy: best i can think of is adding pre/post rate change notifiers to deal with it
<wens>
the temporary reparenting is also handled this way
<wens>
so the clk framework doesn't actually know about it
<MoeIcenowy>
and it will be a gaint work on CCU driver... and Linux 4.11-rc needs it (for A33)
<MoeIcenowy>
P.S. if PLL_CPUX is gated, we can even use P divider for clocks bigger than 288MHz and won't meet system hang
<wens>
at this late a time, it's easier to just revert whatever made this happen
<MoeIcenowy>
you mean revert DVFS support for A33?
<wens>
like disable cpufreq for now
<wens>
yeah
<wens>
the user manual specifically mentions P, maybe there's some other hardware limitation
<wens>
you are welcome to test it though
matthias_bgg has quit [Read error: No route to host]
<MoeIcenowy>
P.S. I've heard that BPi will produce a H5 version of M2+
chlorine has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
<tkaiser>
KotCzarny: BCM4908 is headless. Perfect choice for RPi 4.
<KotCzarny>
by the 2019 it will be already obsolete, won't it?
<MoeIcenowy>
Allwinner have announced 16nm processor in 2018...
<KotCzarny>
also, they make a noise about 'mate, desktop, etc', so funsies expected
chlorine has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
chlorine has quit [Read error: Connection reset by peer]
chlorine_ has joined #linux-sunxi
Worf has quit [Quit: Konversation terminated!]
<tkaiser>
MoeIcenowy: Will SinoVoip specialists take care of voltage regulation. There's a SY8113B on BPi M2+, they just forgot to make it toggle voltage.
<MoeIcenowy>
I hope so
<MoeIcenowy>
maybe they have already noticed that the H3 on BPi M2+ is too hot that the CPU cores do not want to work ;-)
<wens>
are the R and H BUs different?
OtakuNekoP has joined #linux-sunxi
<MoeIcenowy>
wens: of course
<MoeIcenowy>
but I've heard that BU5 (R) has already been broken
<MoeIcenowy>
it only survived for several months
<MoeIcenowy>
P.S. H is BU3
<wens>
broken? as in dissolved/gone?
lamer14918270960 has joined #linux-sunxi
<MoeIcenowy>
most of the people went to BU1
<MoeIcenowy>
and the rest to BU2 (BU2 is focusing on tablet/VR, usually known as A series chips)
terra854 has quit [Quit: Connection closed for inactivity]
tkaiser has quit [Ping timeout: 255 seconds]
<MoeIcenowy>
as what I've heard R16 went to BU1 and R40 went back to BU2
<MoeIcenowy>
(in fact BU5 used to be part of BU2)
<wens>
hmm, was BU5 that short lived? I think R8 was out for quite some time already?
<MoeIcenowy>
maybe R series will continue under BU1 ;-)
<wens>
still don't remember what BU1 is
<wens>
maybe we should put it in a wiki page :p
<KotCzarny>
:)
<MoeIcenowy>
in fact I also do not know well about BU1
chlorine_ has quit [Remote host closed the connection]
perr has joined #linux-sunxi
DullTube has quit [Quit: Leaving]
chlorine has joined #linux-sunxi
chlorine has quit [Read error: Connection reset by peer]
chlorine_ has joined #linux-sunxi
terra854 has joined #linux-sunxi
chlorine_ has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
chlorine has quit [Read error: Connection reset by peer]
chlorine has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
Ultrasauce has joined #linux-sunxi
dh1tw has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
OtakuNekoP has quit [Ping timeout: 255 seconds]
OtakuNekoP has joined #linux-sunxi
dh1tw has joined #linux-sunxi
perr has quit [Remote host closed the connection]
OtakuNekoP has quit [Ping timeout: 255 seconds]
coco3431 has joined #linux-sunxi
dh1tw has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
cnxsoft has quit [Remote host closed the connection]
ganbold has joined #linux-sunxi
dh1tw has joined #linux-sunxi
michael_ has joined #linux-sunxi
<michael_>
Servus, i had successfully dumped the firmware from an cheap A33 tablet (ibowin P130; 1200x800; 1GB RAM ), but i could not get the script.bin from RAM
<michael_>
the script.bin i got, can not be read by sunxi-bin2fex
<michael_>
E: script.bin:1: data must follow a section.
<KotCzarny>
get newer sunxi-tools
<michael_>
i use 1.4.2
<MoeIcenowy>
try git
<michael_>
will try
<tuxillo>
hi, anybody has bought one of those orange pis with cortex-a5 and gsm connectivity?
<KotCzarny>
nope. and good luck having any support on them
<TheLinuxBug>
Yeah while Xunlong makes half decent boards their software support is often more than lacking..
<KotCzarny>
uh oh, 513 orders on them.. o.O
<TheLinuxBug>
hehehe bunch of people saw IoT and 2G and bought some
<TheLinuxBug>
wonder how easy it is to actually integrate sim card and get network connectitivy
<TheLinuxBug>
connectivity*
<KotCzarny>
what do you mean by 'integrate sim card' ?
<KotCzarny>
minipcie gsm modem?
<KotCzarny>
you would have to route antennas somehow
<TheLinuxBug>
as in how the hardware is actually used, obviously it goes in a sim card slot so its not like its difficult to install the card, but I am curion what it takes to actually gain connectivity
<TheLinuxBug>
ohh I thought he was refereing to that new OrangePi with 2g
<KotCzarny>
software, obviously
<KotCzarny>
and if modem chip is a blob, you're f*cked
<KotCzarny>
tuxillo, there is hardly standardized api for voice/data
<tuxillo>
oh hmm
<KotCzarny>
for simple commands or sms, it should be doable via AT commands
Nacho has quit [Remote host closed the connection]
<TheLinuxBug>
what sucks is I think they are trying to discontinue most of the 2g networks in USA here in the not to distant future so I am not sure the value of that board in the US in long term.
<TheLinuxBug>
By January 2017, AT&T's 2G network will be fully retired
<TheLinuxBug>
which is the main 2G GSM network in US
<TheLinuxBug>
maybe sprint will still work, not sure
<KotCzarny>
wasnt t-mobile using 2G as well?
<TheLinuxBug>
well sprint/tmobile are pretty much the same and I was under the impression they were 3G+ only
<TheLinuxBug>
but I could be wrong
Nacho has joined #linux-sunxi
<KotCzarny>
north america is only one of the markets
<KotCzarny>
and not the biggest one too
<tuxillo>
2G will be retired in 2025 in Spain, so I'm golden :)
<tuxillo>
well if the bloody thing works
<KotCzarny>
tuxillo, ever tried xunlong provided os images? ;)
<tuxillo>
hmmm, I think I did for the opi0
<tuxillo>
but they sucked
<tuxillo>
why do you ask?
<tuxillo>
hehe
<KotCzarny>
while this things shouldnt overheat even in worst conditions, software stability might still be a thing
<KotCzarny>
*thing
IgorPec has quit [Ping timeout: 240 seconds]
<TheLinuxBug>
KotCzarny: sure just was commenting that my personal usecase for that board would probably be limirted by 2g availability, in Europe and Asia this isn't the same concern.
<TheLinuxBug>
limited*
<KotCzarny>
TheLinuxBug: yeah, i bet even central/south america would support 2g for some years still
matthias_bgg has quit [Read error: Connection reset by peer]
<TheLinuxBug>
whats funny is the main CDMA carrier here said they wont stop 2g till 2021
<TheLinuxBug>
was considering going looks like may be a few interesting talks
yann has quit [Ping timeout: 240 seconds]
<lvrp16>
are you guys talking about the old spreadtrum rda chips?
<TheLinuxBug>
I believe thats whats on t hat OPi device..
<michael_>
MoeIcenowy: same error with bin2fex from git version. in the sunxi-wiki there are two different versions for retrieving the script.bin with sunxi-fel
<lvrp16>
i heard xunlong got them for really cheap
<MoeIcenowy>
I think for newer SoCs you should retrieve the script.bin via /dev/mem under Linux
<TheLinuxBug>
lvrp16: im sure they are, because half the markets are shutting down 2G it sounds like
JohnDoe_71Rus has joined #linux-sunxi
<MoeIcenowy>
maybe 2G will still survive for a long time in "the third world"
chlorine has quit [Remote host closed the connection]
<TheLinuxBug>
im sure like Pakitsan, India and parts of Europe will be a while behind on that, yeah
<lvrp16>
i doubt it, 2g consumes so much spectrum
<lvrp16>
at the rate prices for electronics and phones are dropping, i'm pretty sure 2g will be gone ahead of schedule than behind
<TheLinuxBug>
even in US while they say they are shutting it down they seem to be selling the towers off to third parties, not actually shutting them down..
<michael_>
MoeIcenowy: sunxi-script_extractor from sunxi-tools?
<TheLinuxBug>
as there still seems to be some 2g carriers as I pointed out
<MoeIcenowy>
michael_: I don't know
<MoeIcenowy>
usually I do this by dd
<lvrp16>
any interest in r40 in raspberry pi form factor you guys think?
<KotCzarny>
pair it with good sata port multiplier and it will be something new
<KotCzarny>
i mean, onboard sata pm
<lvrp16>
i'm thinking of building an r40 with sata where the camera slot is on the pi 3
Ntemis has joined #linux-sunxi
chlorine has joined #linux-sunxi
<KotCzarny>
at this moment there are plenty of boards, unless you make it special in some way, it would be hard to stick out
<lvrp16>
any boards other than the bpi m2u?
<KotCzarny>
also, somehow a20/r40 based boards are crazy expensive, do you know the cause? are there any licenses to pay or some other reason?
<lvrp16>
because they sell like hotcakes
<lvrp16>
no other reason ;)
<KotCzarny>
you mean a20/r40?
<KotCzarny>
or other thingies?
<lvrp16>
yes, sata is very unique
<lvrp16>
europeans love the a20/r40. americans love the s905x.
<MoeIcenowy>
some a20 boards have really rich features ;-)
<MoeIcenowy>
so they're expensive
<MoeIcenowy>
e.g. Cubietruck
<KotCzarny>
MoeIcenowy: but even plain bpi-m1 (the original and spawns) are in the 50usd range
<KotCzarny>
which is crazy when comparing to h3/a64/h5
<lvrp16>
trust me, it's the sata
<MoeIcenowy>
On taobao BPi-M1 is sold at ￥209
<KotCzarny>
H6 promises usb3, which should eat sata performance wise
<KotCzarny>
and minipcie
chlorine has quit [Remote host closed the connection]
<MoeIcenowy>
P.S. according to TL Lim said H6 will only have one-lane PCI-E
<KotCzarny>
(and by sata, i mean the one from a20/r40)
popolon has quit [Ping timeout: 252 seconds]
<lvrp16>
does tl lim still work for pine64?
<MoeIcenowy>
KotCzarny: ￥209 is equal to $30.25 according to my GNOME Calculator
techping has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
<MoeIcenowy>
lvrp16: of course ;-)
<KotCzarny>
MoeIcenowy: still crazy and a lot more than it gives back
<MoeIcenowy>
yes...
<MoeIcenowy>
only for the crappy internal SATA...
<lvrp16>
usb 3.0 will not replace sata though
<MoeIcenowy>
(P.S. I've heard people believe on internal SATA more than USB3/PCI-E...
<lvrp16>
usb 3 packetization is huge overhead for a53
<KotCzarny>
lvrp16: uasp?
<lvrp16>
for pc it's not a big deal
<MoeIcenowy>
P.S. in fact even on A20 USB2 is proven to have better random io performance than native SATA (of course with UASP)
<lvrp16>
uasp solves part of the problem
<lvrp16>
but uasp is far weaker than sata 3.0
<KotCzarny>
we are comparing to sata on a20/r40
fkluknav has quit [Ping timeout: 245 seconds]
Mr__Anderson has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
<lvrp16>
if only amd didn't abandon heirofalcon
<lvrp16>
8x sata 3.0
<lvrp16>
not to mention 10gb ethernet lol
Andy-D has joined #linux-sunxi
chlorine has quit [Remote host closed the connection]
tkaiser has quit [Ping timeout: 240 seconds]
popolon has joined #linux-sunxi
chlorine has joined #linux-sunxi
fkluknav has joined #linux-sunxi
<tuxillo>
but this opi 2g for IoT isnt even listed in orangepi.org right?
<tuxillo>
at least I don't see it
chomwitt has quit [Ping timeout: 246 seconds]
IgorPec has joined #linux-sunxi
<MoeIcenowy>
it's listed in downloadresources
<tuxillo>
i know
<tuxillo>
does this even look serious business at all?
<tuxillo>
looks like*
<lvrp16>
it's called orange pi i96
<lvrp16>
based on the linaro 96boards iot spec
<lvrp16>
i got one from a conference a few days ago
<tuxillo>
I mean you go to their site, lick on model
<tuxillo>
*click
<tuxillo>
damn stupid keyboard
<lvrp16>
xunlong never updates their website
<tuxillo>
sigh
<lvrp16>
or at least 3 years after the product launches
<lvrp16>
getting them to update their site is like pulling teeth
<tuxillo>
if they are busy making boards that's not so bad :P
<lvrp16>
how do you think they get the price so low
<tuxillo>
not sure I want to know
<lvrp16>
basically the bom + $1
<lvrp16>
they do a really good job of keeping costs down but some of the design methods for their boards are a little older
<lvrp16>
they could cut costs further if they updated their designs
<tuxillo>
in any case if this 2g thing works I'm going to do my little crappy solar project much more easily :D
MikeyG has quit [Remote host closed the connection]
<lvrp16>
idk the details of linaro's deal with xunlong, i doubt they will provide kernel support
tkaiser has joined #linux-sunxi
<lvrp16>
are rda chips in mainline? i don't even know where to find the bsp
chlorine has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
yann has joined #linux-sunxi
<tuxillo>
hehe almost every comment you read on opis you see people mocking xunlong distros
<tuxillo>
:D
Nacho_ has joined #linux-sunxi
<lvrp16>
i asked steven if i could take over that site
<lvrp16>
and handle the downloads for him
<lvrp16>
he didn't say yes
Nacho has quit [Ping timeout: 255 seconds]
<lvrp16>
sometimes hard to work with
<tuxillo>
who?
chlorine has quit [Ping timeout: 255 seconds]
<lvrp16>
steven, the head of xunlong
<lvrp16>
he is a character i tell you
<tuxillo>
you have direct contact with that guy?
<lvrp16>
shenzhen is full of eccentric people
<lvrp16>
yeah i visit his office once in a while
<tuxillo>
hmm ok
<KotCzarny>
lvrp16: but he got lucky and probably making some money on allwinner sbcs
<lvrp16>
that's a long long story
<KotCzarny>
still, his margins are low, so no money for anything else
paulk-elm has quit [Read error: Connection reset by peer]
<lvrp16>
he ships a LOT of boards
<lvrp16>
more than any1 else in shenzhen in this space
<KotCzarny>
does he sell via other channels than aliexpress?
fkluknav has quit [Ping timeout: 245 seconds]
<lvrp16>
yes, orange pi boards are in a lot of products for various markets
<KotCzarny>
ho, hum.
yann has quit [Ping timeout: 260 seconds]
<lvrp16>
indian market is big for him
<miasma>
any idea who sells more boards, xunlong or the raspberry pi guys
<lvrp16>
raspberry pi by far
<tuxillo>
I can see tkaiser was involved in the comments on a article in cnx-software.com
<lvrp16>
raspberry pi has 10x the volume of xunlong
<tkaiser>
tuxillo: Take it seriously, the OS image is crappy :)
reinforce has joined #linux-sunxi
<tuxillo>
tkaiser: I do. I'm glad I only ordered one for now. it looks a very nice board for my purposes
<tkaiser>
lvrp16: Are you kidding?
<lvrp16>
tkaiser: no why?
<tkaiser>
lvrp16: Only ten times more Raspberries sold than Oranges sounds weird.
<lvrp16>
raspberry pi is 150k a month?
<tuxillo>
tkaiser: so no source code still released?
<lvrp16>
steven sells maybe 10-30k?
<tuxillo>
excuse my ignorance but this is a one-guy effort?
<lvrp16>
orange pi has like 30 people?
<tkaiser>
lvrp16: Well, the figures on Aliexpress are much lower.
<lvrp16>
20 of which are handling logistics probably
<lvrp16>
he make a lot of custom stuff, fangankehu
massi_ has joined #linux-sunxi
massi_ has quit [Remote host closed the connection]
<tkaiser>
tuxillo: Better hunt for an RDA SDK/BSP yourself. I've no idea, was just curious whether 'xunlong software quality' improved
fkluknav has joined #linux-sunxi
MikeyG has joined #linux-sunxi
tkaiser has quit [Ping timeout: 255 seconds]
<miasma>
interesting, so the rpi sales aren't that much bigger than the rest of the boards together
<lvrp16>
miasma, rpi sales are bigger than the rest combined
matthias_bgg has quit [Quit: Leaving]
<KotCzarny>
miasma, but their profit margins are probably bigger
<miasma>
and marketing
wzyy2 has quit [Ping timeout: 240 seconds]
<wens>
heard bpi stuff is pricier because they use better components and have higher quality standards
<wens>
don't know if that's actually true or not
<KotCzarny>
o.o
komunista has joined #linux-sunxi
<KotCzarny>
that would be great, if they didnt b0rk the designs at the same time..
<wens>
the M2U seems ok
<wens>
haven't checked out the M2M
<wens>
no time for new boards lately
<KotCzarny>
didnt m2u have some performance problems?
<MoeIcenowy>
wens: how to do the gating hack of PLL_CPUX...
<MoeIcenowy>
I cannot think out a better solution
<wens>
MoeIcenowy: i'm thinking new flag in ccu_common, and check for said flag in the set_rate callback
<MoeIcenowy>
as I don't think it can really ensure when PLL is gated the CPUX won't be switched back
<MoeIcenowy>
if CPUX got switched back when we still didn't ungate PLL it's a disaster
<wens>
MoeIcenowy: i guess that is left up to the driver to be careful
<MoeIcenowy>
CPUX and PLL_CPUX are too critical
<wens>
if you want i can try to whip up something tomorrow
<MoeIcenowy>
thanks ;-)
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
paulk-collins has joined #linux-sunxi
lennyraposo has quit [Quit: Leaving.]
michael_ has quit [Quit: Page closed]
chlorine has joined #linux-sunxi
<lvrp16>
KotCzarny: Raspberry Pi is a money loser for everyone involved.
<KotCzarny>
oh?
<KotCzarny>
o.O
chlorine has quit [Read error: Connection reset by peer]
<KotCzarny>
then i do not understand what is it's purpose
<igraltist>
for long term, i running a rpi 1 since over 3 years none stop and no reason to change the board
<MoeIcenowy>
bcm283x is not an ARM SoC
<MoeIcenowy>
it's a Broadcom (R) VideoCore IV MCU with some ARM co-processors
<MoeIcenowy>
;-)
chlorine has quit [Ping timeout: 240 seconds]
<KotCzarny>
;)
<lvrp16>
that article is wrong, 2gb ddr would instantly put the pi 3 in 60 territory
<MoeIcenowy>
I think the 2GiB DDR version of RPi will lost the compatibility with the old VideoCore IV cores ;-)
<lvrp16>
2GB Pi 4 is not happening ever
<lvrp16>
unless they do a POP package
chomwitt has joined #linux-sunxi
<KotCzarny>
they talk about different broadcom soc
<lvrp16>
you can't route enough lines on the PCB for 2GB
<lvrp16>
the CSI connector on the Pi will block it
<lvrp16>
unless they redesign the spec
<KotCzarny>
since they hit the roadblock, they will probably be forced to do that
cloud-e has quit [Ping timeout: 240 seconds]
<lvrp16>
yeah but having a solution in 2019 leaves someone to destroy them in 2017 and 2018
<lvrp16>
2 years is a large window
<KotCzarny>
yup
<TheLinuxBug>
lvrp16: Tell steven next you see him I want my Orange Pi PC Plus Set he refuses to ship to me which I ordered and paid him for but he decided not to actually ship me with the order... the main reason I have stopped ordering boards from Xunlong (considering I now have a pretty fair amount of them is because of this)
<lvrp16>
did he refund you?
<TheLinuxBug>
nope
<lvrp16>
did he provide a tracking #?
<TheLinuxBug>
currently waiting to see if Aliexpress will actualy do anything useful for me
<TheLinuxBug>
ohh he did all kinds of fun
<TheLinuxBug>
pretended he was gonna ship it again
<TheLinuxBug>
made me pay again for shipping
<lvrp16>
was the first order lost during shipping?
<TheLinuxBug>
then told me "it was returned to him" (he never shipped it you can see it on the tracking)
<TheLinuxBug>
nope the set was ommited from the first order
<KotCzarny>
lvrp16: incomplete shipment
<TheLinuxBug>
I let them know about it and such and he pretended they were gonna ship me it then didn't
<lvrp16>
that happens, we sell boards and sometimes customers slip through the cracks
<TheLinuxBug>
well this has been going for 4 months now
<TheLinuxBug>
my original order I believe was 12/20/2016
<lvrp16>
you are dealing with some lowly paid worker
<lvrp16>
with half decent english skills
<TheLinuxBug>
basically kept my money and never shipped it
<lvrp16>
what was in the set?
<KotCzarny>
plot thickens. so, someone is making money on incomplete orders at xunlong?
<TheLinuxBug>
yep.. and Aliexpress is only marginally more helpful
<lvrp16>
lol, don't blow it out of proportion
<lvrp16>
it's a $20 transaction
<TheLinuxBug>
Orange Pi PC Plus SET13 : Orange Pi PC Plus + Transparent ABS Case+ Power Supply+ 16GB Class 10 SD Card Beyond Raspberry Pi
<TheLinuxBug>
[Transaction Screenshot]
<TheLinuxBug>
$ 33.99 X1
sztibi88_ has joined #linux-sunxi
<KotCzarny>
sure, but 0.1% of incomplete orders with 10k orders a month is real money for 'lowly paid worker'
<TheLinuxBug>
81136176206599 is the order number
fkluknav has quit [Ping timeout: 245 seconds]
<TheLinuxBug>
34$ + shipping costs
<lvrp16>
i wouldn't bother him with a 30 dollar transaction, i'll check my stock later and send you one for free lol
<KotCzarny>
lvrp16: still, if he has some bad worker that could give him bad reputation
<lvrp16>
it happens once in a while
<lvrp16>
that's the difference when you buy something with 5% margin vs something with 300% margin
<lvrp16>
the guy makes $500 a month if that
<lvrp16>
take it easy lol
<lvrp16>
500 x 6.9 is 3500 rmb, the fulfillment people probably make 3000
<lvrp16>
the shipping carriers out of china are problematic sometimes
<lvrp16>
next time, demand a refund if they haven't shipped and reorder
<MikeyG>
they're refusing to ship parts now?
<lvrp16>
because when you go through customer service, there's a high probability it will fall through the cracks
<TheLinuxBug>
lvrp16: the issue to me though is
<lvrp16>
no, probably just an one off gone wrong in fulfillment
<TheLinuxBug>
I intended to buy probably another 5-6 units here
<TheLinuxBug>
and intended to make him more money
<TheLinuxBug>
but if he can't even send me a complete order
<TheLinuxBug>
no way I am gonna order more
<lvrp16>
he made $3 off your $33 order
<lvrp16>
if you buy 6 units, it's $18
<TheLinuxBug>
the order I made was over 100$
<lvrp16>
of profit
<TheLinuxBug>
that was only 1 item
<MikeyG>
too many alternatives for them to not care about customers :(
<MikeyG>
almost everything i bought opz/rpz's for now am using esp32 for
<TheLinuxBug>
well then maybe they need to raise prices enough to be able to take care of their customers better, I don't know :/
<lvrp16>
then they have no competitive edge
<KotCzarny>
lvrp16: it could be more if seller starts getting bad reviews because of those 'cracks'
<lvrp16>
yeah but people who buy $20 open source boards will not care between 4 stars and 5 stars
<KotCzarny>
it's potential money they could lose
<lvrp16>
it's not like you are paying $5000 a development board
<lvrp16>
i'm speaking from a business perspective
<lvrp16>
not that you guys aren't right
<TheLinuxBug>
I guess my point is I have almost every H3 Xunlong board sitting in front of me and my intentions were to buy more for projects, I can understand if I am that person buying a single unti and something happens, but after buying like 3 units first time and 1/3 doesn't show up and then I have issues getting it, doesn't make me want to buy anymore.
<lvrp16>
and you are absolutely right, but when you run a business and manage people who may not be perfect, it's hard to control every little aspect of your operation
chlorine_ has quit [Remote host closed the connection]
chlorine has joined #linux-sunxi
<lvrp16>
and more often than you think, what you pointed to will happen, and it's not necessary that steven if purposefully doing it
<MikeyG>
i mean it doesn't compare to a full fledged linux machine (opz at $7) but .. what were you really going to do with that "power" anyway?
<MikeyG>
run systemd?
<lvrp16>
MikeyG: i think embedded is trending towards enterprise rather than the reverse.
<MoeIcenowy>
wens: is the linux clk notifier mechanism enough for gating PLL_CPUX before changing it?
|Jeroen| has joined #linux-sunxi
pg12 has joined #linux-sunxi
chlorine has quit [Ping timeout: 255 seconds]
<MikeyG>
lvrp16: not sure i get what you mean, you mean embedded is moving more toward running full OS stacks? cos i wouldn't call anything Xunlong has produced "enterprise" .. thats like calling sewer water a luxury beverage
<MikeyG>
i mean 75% of the reason i own orange pis at all is cos they only kinda work and there are cool kernel hackers working on getting the hardware fleshed out and working... which again.. is the opposite of enterprise
<lvrp16>
MikeyG: haha, yes, reusable hardware agnostic software. nothing to do with software.
<lvrp16>
hardware*
<TheLinuxBug>
lvrp16: welp I appriciate you listening to my rant about it :) ... it would just be nice to have it show up (or my money) and restore my faith in Xunlong ;p Up until this issue with them I had been pretty happy and I have OPi PC, OPi PC Plus (purchased from banggood after they couldn't deliver), OPi Plus 2E, OPi PC2 and had I not had issues I would have ordered a Win, Pi Zero and 2 more Plus 2Es already but... yeah. (shrug)
<MikeyG>
lvrp16: then yes i agree with you
<scelestic>
TheLinuxBug: open an aliexpress dispute, if you are 100% sure it was never shipped they can see it as well
<MikeyG>
but there are abstractions.. mongoose, nodemcu, etc.. for other, cheaper, more reliable kits.
<MikeyG>
personally im a fan of the unikernel / just-enough-distro approach. works well in the hypervisor and in the IoT.
<TheLinuxBug>
done that and waiting on Ali, but this is now the second time I have gone around with them... waiting.... maybe if the starts align they will decide in about 8 days to return my money, but based on my experience so far, I am betting I won't get my money back.
<MikeyG>
the more code you load that you dont *need* the more needless attack surface you have.
<TheLinuxBug>
scelestic: they conned me by pretending to ship the board again in a seperate order and made it take long enough for my dispute protection to expire so now I am at the mercy of Aliexpress and whether or not they decide to give my money back.
massi has quit [Remote host closed the connection]
nove has joined #linux-sunxi
<lvrp16>
MikeyG: security and maintainability is kind of the bugger that kills the unikernel approach. one of the great forthcoming things I want to play with is RISC V MCUs.
<MikeyG>
lvrp16: don't you think the tooling has matured enough to make those non issues? i donno.. i guess not everone has a i7 6700k.
<scelestic>
i didnt want to get into this conversation but i've done many order on aliexpress and others, even had a few non deliveries and even shipping auto-completes on aliexpress, it's just my experience that it works out 99% of the time
<MikeyG>
and OpenELEC/LibreELEC didn't have too much trouble being nimble.
<lvrp16>
MikeyG: i think having a full blown stack will reduce code vulnerability in the long run that is critical to IoT because it enables continuous updating logistics.
<lvrp16>
MikeyG: you just have to worry about your application and all of the layers below are managed by other people.
<TheLinuxBug>
scelestic: of course I could just be a dick and dispute the charge with my card provider if I do have issues, but since I do appriciate the products I got I was trying not to screw them out of the additional funds by doing a full chargeback which would include more than the item they failed to provide. If it really comes to that though, I am still undecided on how I will proceed.
<TheLinuxBug>
This is the only reason I actually went through Ali instead of just disputing it outright.
<MikeyG>
so you envision a future where OS updates are a mandatory service that just happens
<lvrp16>
MikeyG: yes, i don't yet see a way around it for connected devices.
<MikeyG>
who decides what deps are managed and by who? who decides when to break the ABI?
<MikeyG>
and you can use the VM argument but then we can just examine the Java R&D timeline over the last 20 years.
<MikeyG>
java updates never broke anything. ever.
gzamboni has joined #linux-sunxi
<MikeyG>
maybe you see something i don't, and i gotta say i kinda like your version of the future lvrp16, but i don't think it's a better approach in the short term than building single purpose stacks with tiny attack surfaces.
<lvrp16>
MikeyG: i'm not advocating the VM argument. I only advocate for having permanent logistics to deliver updates and having separate layers.
<MikeyG>
lvrp16: who do you trust to deliver those?
<MikeyG>
if it's theo de raadt then im in.
<lvrp16>
there's a range of alternatives to choose from, diversity is healthy
<lvrp16>
uniform systems are not healthy systems
<lvrp16>
for example, there will be numerous competing vendors offering libraries or stacks for mcus and socs
<lvrp16>
that implement security
<MikeyG>
so from an application developer's perspective, the more environments it can run on the better, even if that application is an embedded one and would likely be coupled with its own hardware
<lvrp16>
and updating of that security
<MikeyG>
this is true.
<MikeyG>
there already are
<MikeyG>
and ofc a lot of the responsibility for security rests with the app developer
<MikeyG>
i think im just saying that MOST of the responsibility for security should rest with the app developer. but i guess that's not a popular view.
<lvrp16>
what i mean is this, the app developer always has to be aware of the stock and is responsibile for understanding the security constraints of the layers below
<lvrp16>
however, he should not be responsible for a tcp handling bug in the kernel
<KotCzarny>
lvrp16: who should be responsible for them then? product vendor?
<MikeyG>
but what you're describing, to me, sounds like shifting sands.
<lvrp16>
and he should not worry about the delivery of a patch to the kernel to his system
<MikeyG>
i think it will be a mix of both of our visions actually.
<lvrp16>
the responsibility falls on a community that manages the specific implementation of that system
<MikeyG>
like.. honestly.. i dont "trust" any linux stack at this point... i probably haven't in years. freebsd maybe, openbsd yes.
<lvrp16>
you can config out most of the crap you don't want
<lvrp16>
isn't that the whole point?
<lvrp16>
i don't feel that freebsd is any more secure than linux.
<lvrp16>
being more conscious of security does not improve overall security
<scelestic>
TheLinuxBug: i don't see how you're being a dick if you are certain they're pulling your leg but i don't feel comfortable continues such discussion
<MikeyG>
but i don't want redhat or canonical updating my kernel
<lvrp16>
freebsd is slow in hardware support, that constrains the user to more legacy hardware which may have unfixable bugs
<MikeyG>
lvrp16: install ubuntu server. ps aux | wc -l ... then do it with openbsd
<lvrp16>
MikeyG: thats because the infrastructure for updating is not correct, that exactly highlights what i mean.
<lvrp16>
there's ppas
<lvrp16>
you can always debootstrap debian
<lvrp16>
ubuntu includes a lot of unnecessary garbage by default
<MikeyG>
i know you can get debian very small.. especially if you're paravirtualizing it
<lvrp16>
for their "world view"
<MikeyG>
i haven't installed straight debian in a while.. but i will give it a try.
<KotCzarny>
debian also features crap called systemd nowadays..
<MikeyG>
if your distro doesn't at least have the option to turn off sendmail by default
<MikeyG>
like i start up and i see dbus and avahi and 400 systemd processes
<MikeyG>
doing god knows what
<lvrp16>
systemd is OK, just don't know why now every debian package has to support it over others
<MikeyG>
im exaggerating but
<KotCzarny>
lvrp16: it's not in the sense it tries to depend on plethora of things and requires others to depend on it, removing ability to trim os to your needs
<KotCzarny>
unix started as 'do one thing but do it perfect' building blocks of utils
<lvrp16>
thats what i mean, i think the mistake is not on the part of systemd, but on debian
<KotCzarny>
debian's dependency hell is separate problem
<KotCzarny>
;)
<lvrp16>
for embracing the model of interfacing with systemd in such a pervasive way
<MikeyG>
its not too hard to have your application's startup process handle SIGCHLD and inherit orphans. if pid 1 was gonna be systemd anyway it may as well be your application
<MikeyG>
few nonblock waitpids on a timer
<KotCzarny>
uh, hell no.
<MikeyG>
why you say hell no?
<KotCzarny>
systemd is a big cow that's full of holes
<MikeyG>
so is my application
<MikeyG>
lol.. well i hope it isn't.. but i have to believe it is or i am looking at it wrong
<KotCzarny>
but if you have perfect init process that can restart your app, then that's better, no?
<MikeyG>
what im arguing for is having my app *BE* init.
<MikeyG>
as well as my app
<KotCzarny>
see my 'perfect little building blocks' argument
<MikeyG>
or at least serve as init for the kernel
<KotCzarny>
once your app grows, so the possibility of introducing bugs
<lvrp16>
the next generation kernel after Linux will have to deal with probable and not exact systems
<lvrp16>
Linux is an very ancient beast.
<MikeyG>
yeah.. i can't believe linux didn't tackle GPUs harder than they did
<MikeyG>
like the GPU should be in the regular scheduler by now
<KotCzarny>
maybe because it was more of a headless server thingie than desktop replacement?
<MikeyG>
that assumes that GPUs are only good for graphics ;(
<MikeyG>
i mean once upon a time fpu's were co-processors i387 for example
<lvrp16>
GPUs are still what I consider fix function hardware
<MikeyG>
i see GPUs in the same vein
<KotCzarny>
cuda and friends are only quite recent invention
<MikeyG>
cuda and opencl are the wrong approach
<MikeyG>
they depend on proprietary software blobs
<lvrp16>
future systems with have mesh or 3d topology, means of discovery, consensus, etc
<KotCzarny>
yup, but they only got invented after hw evolved
<MikeyG>
cuda more so than opencl.
<MikeyG>
lvrp16: exactly.. now we see that GPUs are much more than ways to play counterstrike
<KotCzarny>
and earlier gpu was just simple 3d accel, fast, but limited in functionality
<MikeyG>
they're the key to machine learning
<lvrp16>
the dnn stuff is still fix function hardware imho
<MikeyG>
had linux at least worked to bring GPUs into the equation.. even with a 50% performance loss vs. vendor blobs
<MikeyG>
just.. changed the architecture to make room for them
<MikeyG>
we'd be so much further along
<MikeyG>
whatever replaces linux should just obtain and consume resources
<MikeyG>
crawl the network for nodes that respond to signed requests to "assimilate" and spit out lists of their devices/capabilities
<KotCzarny>
hehe, b0rg
<lvrp16>
hahaha
<MikeyG>
the os should be able to take over routers, switches, IoT devices, etc
<lvrp16>
i was just going to say that
<KotCzarny>
resistance is futile, you will be assimilated!
<MikeyG>
and make them perform the task they're most suited to
<MikeyG>
i was thinking of borg lol
<KotCzarny>
btw. closest thing to your idea is trojan/botnet network
<KotCzarny>
;)
* lvrp16
starts startrek music
<KotCzarny>
so, no bright future there
<MikeyG>
im talking in a constructive way.
<MikeyG>
lol
<KotCzarny>
rememeber you are dealing with human products
<KotCzarny>
no way future is 100% white
* lvrp16
seven of mine ;)
<MikeyG>
plus.. in my scenario.. if you obtained my private key
<MikeyG>
you'd be able to control all my hardware
<MikeyG>
single point of total failure
<MikeyG>
but this has been the funnest conversation i've had in a long time
<MikeyG>
lol thanks for being a bunch of smart dudes hanging out somewhere
<MikeyG>
(or gals)
<KotCzarny>
seriously, people were trying no-security approach, then bad guys come and subvert/steal resources
<MikeyG>
mine isn't a no security approach.
<KotCzarny>
humans are the weakest links nowadays
<MikeyG>
could use the blockchain to back up / state that these compute resources belong to me
<MikeyG>
humans have been the weakest link since forever.
<MikeyG>
in 20 years jobs like "network architect" or "systems administrator" should not exist
<MikeyG>
if we coders do our jobs right anyway
<KotCzarny>
wrong, ai will code
<KotCzarny>
humans doing dev is still weakest link
<MikeyG>
well, we still have to get the AI coding
<MikeyG>
if we got the AI coding.. i imagine it'd create an OS much like the one i just described
<KotCzarny>
and i'm 99.9999% sure, even the best ai will have backdoors
<MikeyG>
thats a lot of sure-ness.
<MikeyG>
i for one wouldn't leave an on purpose backdoor in an AI
<MikeyG>
i'd close it up completely and set it free
<MikeyG>
we have free will why shouldn't it?
<MikeyG>
so you'd have to hope my code was bad.. and that my ai wasn't smart enough to patch itself
scream has joined #linux-sunxi
<MikeyG>
im sure most people would leave a backdoor tho. i wouldn't... just because it'd be more fun to watch the outcome
<KotCzarny>
many backdoors are also safety measures, just in case something goes wrong
<MikeyG>
oh .. and buildroot was the thing i was goign to mention earlier. like i really like that approach
<KotCzarny>
MikeyG: well, for example car ai locking itself and just accelerating due to ai going depressive
vickycq- has quit [Read error: Connection reset by peer]
<MikeyG>
i'd like to create a build system that spits out a hybrid buildroot distro/container .. .. one build command gets you an arm image, a docker container, an x86_64 os image
vickycq- has joined #linux-sunxi
<MikeyG>
and they all work the exact same way from the user's perspective
<MikeyG>
KotCzarny: yeah totally.. you want a backdoor into that.
<MikeyG>
i was thinking of a fully sentient AI..
<MikeyG>
a consciousness
<KotCzarny>
so, the ai thinks user should be terminated kind of way?
<MikeyG>
if the ai thinks that i should be terminated and has the means to do so, then doesn't that make me what "went wrong"
<MikeyG>
he'd just be fixing the bug.
<MikeyG>
its just a matter of perspective
<KotCzarny>
from user perspective it would be wrong
<MikeyG>
well i mean.. my perspective isn't really relevant at that point
<MikeyG>
since i'd be dead soon
<MikeyG>
im sure it had a better reason for killing me than i had for not dying.
<KotCzarny>
what if it's reasoning would be the result of some bug?
<MikeyG>
but if its a consciousness and has a bug.. is it a bug .. or a character flaw?
<MikeyG>
like.. we dont go around saying people are buggy
<KotCzarny>
we do
<KotCzarny>
we call it mental diseases
<MikeyG>
like.. in the .. they have an illness.. or mental illness sense
<KotCzarny>
some are fixable, some are not
<MikeyG>
then im not sure.. i suppose one would definitely want to say it had a mental illness if it killed me in cold blood.
<MikeyG>
well i'd hope so anyway
<MikeyG>
it definitely wouldn't have a high regard for human life.. but what's a human to say what a machine's morality should be?
IgorPec has quit [Ping timeout: 260 seconds]
<KotCzarny>
who knows which way ai's morality would go, if any
<KotCzarny>
especially the one that would evolve without human's direct input
<MikeyG>
it could just be in its nature.. if a lion kills a human who was walking around in the savannah
<MikeyG>
does that lion have mental illness
<KotCzarny>
if it goes about 'give me resources', then we are f*cked
<MikeyG>
or was it just hungry
<MikeyG>
maybe it killed me so it could assume my identity to gain control of my bank accounts and acquire resources
<MikeyG>
so it could expand
<MikeyG>
thats just hunger. it's like the lion then. not mental illness.
dh1tw has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
mzki has quit [Quit: leaving]
<MikeyG>
you ever see ex machina? that look in nathan's eyes when she kills him.
mzki has joined #linux-sunxi
<MikeyG>
he was proud and terrified
dh1tw has joined #linux-sunxi
<KotCzarny>
it's all grey. no black and white because we and our neighbours (be it a lion) have different morals/motivations/needs
<KotCzarny>
and AIs wouldnt be much different if they become independent and start to evolve
dh1tw has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
scream has quit [Remote host closed the connection]
reinforce has quit [Quit: Leaving.]
terra854 has quit [Quit: Connection closed for inactivity]
sztibi88_ has quit [Remote host closed the connection]
lkcl has quit [Ping timeout: 255 seconds]
xes__ has joined #linux-sunxi
xes_ has quit [Ping timeout: 240 seconds]
dh1tw has joined #linux-sunxi
victhor has joined #linux-sunxi
Mr__Anderson has joined #linux-sunxi
yann|work has joined #linux-sunxi
paulk-collins has quit [Quit: Leaving]
popolon has quit [Quit: WeeChat 1.7]
komunista has quit [Quit: Leaving.]
Mr__Anderson has quit [Quit: Leaving.]
xes__ has quit [Quit: WeeChat 1.6]
rotcpy has quit [Ping timeout: 260 seconds]
rotcpy has joined #linux-sunxi
rellla has quit [Ping timeout: 240 seconds]
rellla has joined #linux-sunxi
tkaiser has joined #linux-sunxi
rotcpy has quit [Ping timeout: 240 seconds]
rotcpy has joined #linux-sunxi
Andy-D_ has joined #linux-sunxi
<willmore>
KotCzarny, lagged a bit. yeah, that Pi4 'interview' is crap.
tkaiser has quit [Ping timeout: 255 seconds]
Andy-D has quit [Ping timeout: 260 seconds]
<willmore>
tkaiser, BCM4908 headless? Perfect, then they won't have to worry about not having an open source video driver (HDMI, GPU, or video decode). I can see the headlines now: "Rpi4 has the best open source support of any Rpi!!!11eleven!!"
<willmore>
KotCzarny, t-mobile is keeping 2G going for a few more years. 2019, I think? Which is practially forever for an IoT company. :)