_aeris_ has left #linux-sunxi ["Konversation terminated!"]
<wens>
i'm running debian, but thanks though :)
cptG has quit [Ping timeout: 250 seconds]
Gerwin_J has joined #linux-sunxi
mark5 has joined #linux-sunxi
IgorPec has joined #linux-sunxi
engideavr has quit [Quit: Konversation terminated!]
<mark5>
I'm having some problems with bluetooth stack/hardware. I'm running a script on a A10 board that continuously performs bluetooth inquiries, and after some time (say, 1 to 5 days) it exits because the hci0 device is 'down'. Hciconfig hci0 reset shows "Can't init device hci0: Connection timed out (110)". The only thing that resolves this is a reboot... I can recompile the kernel with debugging things enabled, but where do I start?
engideavr has joined #linux-sunxi
<Deskwizard>
mark5: anything in the logs about BT before it exits ?
<mark5>
Deskwizard: There doesn't seem to be anything useful, just "hci0 command tx timeout"
<Deskwizard>
mark5: kinda not usefull it takes that long... I was about to say try it on a regular computer see if it hangs after the same time, maybe its the adapter itself
<Deskwizard>
and a question of mine... Boot_1 vs Boot_2 ?
<Deskwizard>
v1 vs v2 I mean, sorry
<mark5>
Deskwizard: I'm not sure what you mean with v1/v2
<Deskwizard>
mark5: that was a question of my own, unrelated to your issue, sorry I tend to jump around a bit
<mark5>
Ah. :)
<Deskwizard>
but yeah... it can be 3 things basically (your issue)
<Deskwizard>
1) something with the hardware, host side
<Deskwizard>
2) the adapter
<Deskwizard>
3) your code is wrong
<Deskwizard>
I'm usually in category 3 myself lol
<Deskwizard>
* 1) hardware or software, like kernel side
<Deskwizard>
Uhm, even google comes up empty handed on the boot_v1.0 vs boot_v2.0 thing...
<KotCzarny>
according to this is has some 'HW' for media
<Rambler44>
So finally 28nm?
<Rambler44>
Because I've seen a lot of people lie about this. But seeing its 2 GHz over the 1 GHz I can see a slice in nm.
cptG_ is now known as cptG
<Rambler44>
I wonder if RISC ever will be any good for I/O operations?
apritzel1 has quit [Ping timeout: 248 seconds]
<Rambler44>
I can see why Allwinner can be good for a cheap NAS, but will these ever be good enough to handle it all. Like the A20 with SATA, now can you really make full advantage of that native port?
<Rambler44>
I just smell huge CPU usage when using I/O on these
<apritzel>
Rambler44: Allwinner is not for NAS, Allwinner makes chips for cheap tablets and set-top boxes
<apritzel>
I/O performance is not important, hence the SoCs simply are not designed for it
<Rambler44>
Yeah, but they are put into development boards, which some people want to use for NAS because they are cheap.
<apritzel>
that's another story ;-)
<Rambler44>
I almost cried when I saw Owncloud use Raspberry Pi 1 for a NAS
<KotCzarny>
:)
<Rambler44>
The abomination can't do more than 60Mbps on full load
<KotCzarny>
but works
<Rambler44>
hit it with encryption
<KotCzarny>
and someone probably can use it
<apritzel>
Rambler44: and what "RISC" were you referring to above?
<Rambler44>
and tadaa 16Mbps
cosm has joined #linux-sunxi
<Rambler44>
I was thinking that maybe RISC is worse than CISC for I/O
<apritzel>
there is no RISC vs. CISC anymore
<Rambler44>
My level of understanding only goes so far, I was thinking the instruction set or in so.
<Rambler44>
No?
<apritzel>
if you think of ARM vs. x86, there are ARM chips which provide huge I/O performance
<Rambler44>
Probably not Allwinner
<apritzel>
it's just that they don't cost 5 US$ ;-)
<apritzel>
well, you get what you pay for ...
<Rambler44>
But all of them are just the same old ARM core licensee stuff
<Rambler44>
Or have I mistaken to some point?
<apritzel>
show me a compete Intel board for 20-30 US$ which has a decent I/O performance
<Rambler44>
Like Allwinner top box can allow for native USB 2.0 ports and attach a 1Gbps PHY
<Rambler44>
Sounds impossible
<apritzel>
Rambler44: the core has not much to do with I/O performance
<apritzel>
it's the internal bus, really
nove has joined #linux-sunxi
<KotCzarny>
apritzel, so it's basically some internal bus getting stuffed? (regarding eth/sata bottleneck)
<apritzel>
sure
<Rambler44>
I seem to recall that mr tkaiser over here did some tests on H3
<apritzel>
the bandwidth of that is severely limited for cost reasons
<Rambler44>
Something 100 MB/s sequential read
<Rambler44>
Wouldn't that be able to handle at least a 100Mbps etg to the full extent.
<apritzel>
this is the idea of the ARM business: the vendors provide a SoC _taylored_ for some particular use case
<KotCzarny>
idea: all in one benchmark tool to add info on device pages
<apritzel>
and Allwinner taylors them to be cheap and good enough for tablets
<KotCzarny>
Rambler44: tkaiser test was 3x usb raid
<KotCzarny>
or something similar (combined performance)
<Rambler44>
And you also have architectural licensees, like Snapdragon, where they design their own cores.
<KotCzarny>
which single sata port on a10/a20 already reached (not on write tho)
<Rambler44>
Doesn't matter, even 20 MB/s should be able to handle the 100Mbps NIC right?
<Rambler44>
Its not like you want read performance beyond the capabilities of the NIC, well, depends on what you do.
<KotCzarny>
100mbps == ~12MB/s
kelvan has joined #linux-sunxi
<Rambler44>
Exactly
<Rambler44>
Well, I heard RPi 2 should do it, well, without encryption that is.
<Rambler44>
So for 100Mbps NAS there you go, perfect cheap choice.
<KotCzarny>
well, a20 should be fine for such use, even with encryption, but one shall do the tests first
<Rambler44>
I know, due to the nature of Allwinner support its probably a good choice to pick old ones and rely on community support.
<KotCzarny>
but my cpu is downclocked to ~800mhz
leio has joined #linux-sunxi
<Rambler44>
Is there even any point in picking quad core over dual core?
<KotCzarny>
pity hw crypto is broken :/
<apritzel>
Rambler44: if you care about NAS performance you should actually pick a NAS chip, like the Marvells
<KotCzarny>
rambler44: regarding core, 1 core bound to eth/usb, 1 core bound to sata/usb, if system has to do anything else, you should choose 4 cores
p1u3sch1 has quit [Ping timeout: 276 seconds]
<Rambler44>
Tell me where to buy that from a store that ships internationally for a reasonable amount and can bypass VAT of 25%, most likely nowhere. At that point of time I'd rather buy a dock specifically designed for NAS and suddenly the whole idea of low cost NAS has vanished beyond any small budget
<Rambler44>
Which is why some cheap crap companies like OPi managed to sell over 6000 units of whatever bad board.
<Rambler44>
They ship it with dirt cheap Chinese Registered Air Mail
<Rambler44>
and the amount goes beyond taxation, it's ridicilous.
<Rambler44>
It could be possible if any electronic brands actually take in the devices, but that's highly unlikely.
<Rambler44>
I should be happy they even get RPi 2 into stock.
<KotCzarny>
tme.eu has olimex boards
<Rambler44>
Which reminds me, anything from hardkernel would probably be in the same scenario fue to Korea.
<Rambler44>
due*
p1u3sch1 has joined #linux-sunxi
<Rambler44>
But they are Polish, my guess is that since I go outside EU, into the EU economical zone I wonder if they can use the treaty to even import it or I'd be subject to double taxation (which is quite fun)
<KotCzarny>
poland and norway are probably in shengen zone, which means no taxation
<Rambler44>
Yea, we are.
<KotCzarny>
also, olimex is the brand from europe
<KotCzarny>
(hungary afair)
<Rambler44>
I just know I have to pay tax at toll customs, I don't know yet what they will tax me from Poland but our country has some weird rules.
<KotCzarny>
ask your taxation office?
<Rambler44>
As far as I know any electronics would be taxed with a taxation fee.
<cosm>
KotCzarny: Bulgaria
<KotCzarny>
right.
<cosm>
Rambler44: I keep hearing people complaining about OPi hardware. What's the deal with that, I didn't notice anything sketchy with mine?
<KotCzarny>
cosm: for one, 1.6 cpu marketing scam
<KotCzarny>
1.6ghz
<KotCzarny>
a lot of people had troubles when they ran official images
<Rambler44>
cosm there is a chance they don't bother QC and it can come faulty and whatnot to save money.
<KotCzarny>
another thing, my usb devices sometimes dont get detected on boot (reconnecting fixes it)
<Rambler44>
Sometimes lucky sometimes not
<Rambler44>
It's like buying the best cheap IMEI from Xiaomi
<Rambler44>
Some hybrid earbuds sound bad others are good
<Rambler44>
Sorry IEM*
<Rambler44>
$50 board with €16 shipping would only result in $18 tax though, maybe I'm making too big a fuss about nothing.
<Rambler44>
thinking about it, OPi is cheaper than the tax...
<Rambler44>
KotCzarny I think even the forum images
<Rambler44>
I talked to a friend and he immediately discovered the overvolting and changed it
<cosm>
I see, thanks
<cosm>
I'm running a headless system with an image I've built myself, so I guess I didn't have the chance to run into most of those issues
<Rambler44>
By that I mean this concept in particular "continuously encrypts and decrypts itself so that it is impossible to crack."
vagrantc has joined #linux-sunxi
cosm_ is now known as cosm
bonbons has joined #linux-sunxi
<Deskwizard>
o.O
<Deskwizard>
impossible today, yeah okay ;) hehe
<Deskwizard>
Also, next time, full quote.
<Deskwizard>
Denuvo Software Solutions has stated that the technology "does not continuously encrypt and decrypt any data on storage media. To do so would be of no benefit in terms of security or performance." The company has not revealed how Denuvo Anti-Tamper works
<Deskwizard>
you forgot that part didnt you ;)
<Deskwizard>
I call BS.
<Rambler44>
:P You thought it would use HDD?
<Rambler44>
It probably does something in RAM
<Rambler44>
There is some kind of key crystallographic engine in there
<Deskwizard>
last sentence sums it out for me :P
<Rambler44>
Well, they sell this to companies so of course.
<Rambler44>
It's like HDCP
<Deskwizard>
no its not
<Rambler44>
HDMI 2.0 was cracked before it even came out because of a loophole that allows to convert HDMI 2.0 to older ones.
<Deskwizard>
lol nice, I didnt know that
<Rambler44>
It's there to make sure it's safe for the copyright companies and therefore they sued
<Rambler44>
But they were in the legal limits xD
<Deskwizard>
nice to hear, for once
<Deskwizard>
ftw
<Deskwizard>
(for the win, not the other one...)
<Rambler44>
Not that I care, I just hate when DRM screws with other aspects of something.
<Rambler44>
For example my old GPU does not go well with HDCP
<Rambler44>
It can miss handshakes each half hour or so
<Rambler44>
I managed to throw together a hacky driver that avoids that
<Rambler44>
Too off-topic, as an extra note that's why I buy games from gog.com if I ever fancy playing something.
<Deskwizard>
last game I bought was Duke Nukem Forever, I think that says it all really.
dev1990 has joined #linux-sunxi
Rambler44 has quit [Quit: Page closed]
apritzel has quit [Ping timeout: 248 seconds]
<cosm>
KotCzarny: so at higher DRAM frequencies, the cube in lima-memtester is not spinning, but the LED keeps blinking for over 10 minutes
<KotCzarny>
hrm?
<KotCzarny>
ssvb: ^ bug in your script?
<KotCzarny>
cosm, my guess is that test fails when cube stops spinning
<ssvb>
cosm: do you have a log from the serial console?
_massi has quit [Quit: Leaving]
enrico_ has quit [Quit: Bye]
<cosm>
ssvb, wow, it's been a while
<cosm>
let me find a serial adapter
<ssvb>
it is possible that the script itself has crashed before it had a chance to update the LED status
<ssvb>
is this reproducible or just happened once?
<cosm>
ssvb: it's reproducible
<cosm>
but the script crashing makes sense
<cosm>
because at lower frequencies it hangs as expected :)
<ssvb>
hmm, either way, it will probably not light the red led either
<ssvb>
"hangs as expected"?
<cosm>
the led stops blinking when the cube stops spinning
<KotCzarny>
ssvb, i think you should update led status every few s manually
<KotCzarny>
(though i've thought you were doing it that way)
<ssvb>
cosm: do you mean that the test already fails gracefully at some lower clock frequencies, but you still keep trying to increase the DRAM clock speed even more to see what happens?
<ssvb>
KotCzarny: the kernel code is doing the LED blinking animation
IgorPec has joined #linux-sunxi
<cosm>
ssvb, I've started at 768MHz, which I was expecting to fail
<cosm>
but the LED kept blinking
<ssvb>
KotCzarny: and as I already said, the users are expected to wait until the red led lights up, which should be impossible if the script has crashed :-)
<KotCzarny>
:)
<KotCzarny>
explains, still, results in unexpected state as mentioned above ;)
<ssvb>
cosm: we are running code on a potentially (very) unreliable hardware, so we can't really guarantee anything
<cosm>
I was expecting the LED to be switched every X iterations of a loop writing, reading and checking the contents of the memory
<cosm>
until it fails
mzki has quit [Quit: leaving]
<cosm>
but nevertheless, this explains it
<cosm>
what are the serial port settings, I'm not getting anything?
lerc has quit [Quit: No Ping reply in 180 seconds.]
lerc has joined #linux-sunxi
atsampson has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
bonbons has quit [Ping timeout: 250 seconds]
afaerber has joined #linux-sunxi
bonbons has joined #linux-sunxi
nove has quit [Quit: nove]
lemonzest has quit [Quit: Leaving]
paulk-collins has quit [Quit: Quitte]
bonbons has quit [Quit: Leaving]
KotCzarny has joined #linux-sunxi
interrobangd has joined #linux-sunxi
<cosm>
KotCzarny: it failed after more than an hour at 672
interrobangd has quit [Read error: Connection reset by peer]
lerc has quit [Quit: No Ping reply in 180 seconds.]
lerc has joined #linux-sunxi
lerc has quit [Quit: No Ping reply in 180 seconds.]
lerc has joined #linux-sunxi
iamfrankenstein has quit [Quit: iamfrankenstein]
<Deskwizard>
my chenglish isnt too good, just a quick question... does that mean "enable the serial uart debug" or "use that uart port for serial debug" ? > ;cpus_uart_debug_used --r_uart whether used for cpus debugging
mosterta has joined #linux-sunxi
mcan has quit [Changing host]
mcan has joined #linux-sunxi
naobsd has joined #linux-sunxi
ricardocrudo has quit [Remote host closed the connection]
uwe_ has quit [Ping timeout: 248 seconds]
uwe_ has joined #linux-sunxi
interrobangd has joined #linux-sunxi
interrobangd has quit [Remote host closed the connection]