Black_Horseman has quit [Ping timeout: 264 seconds]
VargaD has joined #linux-sunxi
<Turl>
mripard: hey :)
indy has quit [Remote host closed the connection]
<libv>
Vanfanel: slow, but should pick up again tomorrow.
<libv>
i will spend ~6h on lcd bringup this evening
<Vanfanel>
hey, libv! thanks! :D
<wingrime>
libv:you finaly removed all translations ...
<swabbles>
window 6
<Vanfanel>
libv: should we expect a speed increse on GLES2 apps running on sunxi-kms versus the current fbdev solution? For example, the Cubieboard2 isn't much faster than a Raspberry Pi with current solution...
<libv>
Vanfanel: not for the binary mali driver, but perhaps for the open source driver
<Vanfanel>
should LIMA + sunxi-drm be faster?
<Vanfanel>
sunxi-kms, sorry
<libv>
s/should/could/
<libv>
Vanfanel: watch the fosdem talk/slides
<Vanfanel>
ah, yes, I was reading your slides yesterday regarding sunxi-kms
<Vanfanel>
but I didn't see a mention to lima+sunxi-kms being faster than current fbdev solution
<libv>
the lima talk is more important there
<wingrime>
libv: sunxi-kms currently are simple interface to disp code, not full stack?
<wingrime>
libv: as I currently remember
<libv>
wingrime: what do you mean with a full stack in light of a kms driver?
<wingrime>
libv: I mean, it just interface to old crap DISP code
<libv>
Vanfanel: the lima talk mentions the issues the lima driver has, and how, together with etnaviv, it is in a uniquely difficult position
<wingrime>
libv: not clean rewrite
<libv>
as mali and vivante, and pvr (which i and some others have successfully managed to avoid), are IP cores matched with many other blocks from many other vendors
<libv>
there the issue of integration is quite definitive with respect to performance
<libv>
and the lack of communities for anything but sunxi is really holding us back
<libv>
wingrime: it's a clean rewrite
<libv>
wingrime: but i have a massive cleanup of disp sitting about
<libv>
wingrime: this is explained in my kms talk
<wingrime>
libv: hm ... seems my information too outdated
<libv>
as for removing all translations, i had a few hours, but this was not proper development time
<wingrime>
libv: pvr are in modern allwinner socs and in sweet a80 .... thats suck
<wingrime>
libv: so looks like we can't avoid it
indy has joined #linux-sunxi
indy has quit [Excess Flood]
<libv>
wingrime: i will avoid it
<libv>
and i strongly encourage others to avoid it
<libv>
the imgtec way of working is quite muddled
<wingrime>
libv: for me, it just more one blob crap
<libv>
and, as a result, imgtec should be responsible for open sourcing itself
<wingrime>
libv: talking about internals, how much it crap?
indy has joined #linux-sunxi
<libv>
the engine is heavily microcode based
<wingrime>
libv: and what that reason allwinner choose it instead modern mail?
<libv>
that microcode is counted on for working around issues (this makes hw people lazy), and for speed
<libv>
and this microcode talks to the kernel directly
<libv>
with shared structures
<wingrime>
libv: so it VLIW ?
<libv>
but the microcode is closed, and is loaded from userspace
<libv>
whereas the kernel is gpled
<libv>
it is extremely rare to have both compatible
<wingrime>
libv: seems that problematc to RE instructions
shineworld has joined #linux-sunxi
<libv>
lkcl already has some documentation out on the USSE
<wingrime>
libv: how much mail differenet form that
<wingrime>
libv: as I understand it too complie shaders to microcode too
<wingrime>
like any modern gpu
<libv>
wingrime: find out about microcode/firmware
<libv>
wingrime: mali has some, but it hasn't been updated for years, and it is stored on chip
<libv>
wingrime: it is not changed for each revision or hw vendor or kernel version
<libv>
it plays a much lesser role than the pvr microcode
<libv>
and you are mixing up shader code with firmware
<libv>
even though in pvr the instruction set is, iirc, the same
<wingrime>
libv: how much mali pipeline fixed?
<libv>
quite a lot, as is true for most 3d engines
<libv>
with pvr though, a lot less is fixed, meaning that a lot of the housekeeping is done by the microcode
wingrime1 has joined #linux-sunxi
wingrime2 has joined #linux-sunxi
wingrime1 has quit [Read error: Connection reset by peer]
wingrime2 has quit [Read error: Connection reset by peer]
wingrime has quit [Read error: Connection reset by peer]
wingrime has joined #linux-sunxi
wingrime has quit [Client Quit]
dlan^ has quit [Quit: leaving]
Gerwin_J has quit [Ping timeout: 252 seconds]
<mnemoc>
moin
mkutsevol has left #linux-sunxi [#linux-sunxi]
mkutsevol has joined #linux-sunxi
<Turl>
moin mnemoc
HeHoPMaJIeH has quit [Remote host closed the connection]
cajg has quit [Ping timeout: 240 seconds]
grafgustav has quit [Quit: leaving]
kuldeepdhaka has quit [Ping timeout: 252 seconds]
Skaag has quit [Ping timeout: 255 seconds]
Vanfanel has quit [Quit: Lost terminal]
awafaa_ is now known as awafaa
cajg has joined #linux-sunxi
Dodger78 has joined #linux-sunxi
moral has joined #linux-sunxi
moral has quit [Client Quit]
<Dodger78>
hi , is root at boot.cmd where the kernel is or where the distri is ?
deasy has joined #linux-sunxi
<Dodger78>
im trying to run a 3.14 kernel on cubieez / cubietruck
<mripard>
Dodger78: it's supposed to be the partition where your system is
kuldeepdhaka has joined #linux-sunxi
<Dodger78>
im only booting blind with hansg kernel config. i can hear the hard disk spinup and so, but no LEDs no ping ... hansg told me the config is correct
<Dodger78>
damn i need a USB UART thing to debug this
<Dodger78>
noone has a cubietruck and a kernel uImage that works for me ?
y0g1 has quit [Ping timeout: 255 seconds]
<Dodger78>
im also using hansg boot.cmd
souther has quit [Ping timeout: 265 seconds]
<Dodger78>
but i havent updated the u-boot environment from cubieez 1.0, is that the fault ?
arky has joined #linux-sunxi
<mripard>
Dodger78: I don't know, that might, but we'll need logs to know :(
<mripard>
that way, you actually describe what's connected to the bus
<mripard>
which is what you want in the DT
<mripard>
and you get spidev
<mripard>
which is what you want as a user
<mripard>
and if eventually, you have a brand new driver for this
<mripard>
you just remove the compatible from spidev, merge your driver
<mripard>
and you don't even have to change the DT
FR^2 has quit [Quit: Connection reset by peer]
<mrnuke>
I understand how you want it done. What still fuzzy on, is the fuzz about spidev. I thought spidev sits on top of whatever spi_master driver is available
<mrnuke>
so, I thought spidev is a generic userland interface, and comes after you write your platform SPI driver
<mrnuke>
or is the platform SPI driver expected to provide that interface in the long run?
<mrnuke>
(And I may be using terrible terminology)
<Turl>
mrnuke: from what I gather, spidev is kind of a handy hack, and in the long run you should write a real driver to interact with the spi device
<mrnuke>
my end goal is to have a spidev node to use with flashrom
<mrnuke>
flashrom is this handy tool that can program SPI NOR flash chips
uRandomMM has joined #linux-sunxi
morfoh_ is now known as morfoh
Fusing has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
Quarx has quit []
bertrik has joined #linux-sunxi
wingrime has joined #linux-sunxi
<wingrime>
mnemoc: ping
<wingrime>
anyone does known how stay in irc with bad mobile connection
<Turl>
wingrime: hi
<Turl>
wingrime: run client on a server
TheSeven has quit [Read error: Connection reset by peer]
<Turl>
wingrime: eg irssi on screen on a server
<memleak>
does HDMI output work in mainline kernel? by work I mean can it initialize a 1080p display?
<wingrime>
Turl: heh
<Turl>
wingrime: then connect to the server via mosh
<Turl>
wingrime: you can also use bouncer like znc
<Turl>
wingrime: and normal client on your pc
<memleak>
if so how does it do it without drivers/video/sunxi/hdmi?
<wingrime>
....
<Turl>
memleak: no, no video at all
<memleak>
oh ok thanks
<wingrime>
Turl: thanks
<wingrime>
Turl: I should find solution with android
<mrnuke>
the one you didn't believe I was gonna get
<memleak>
i was referring to your ubuntu comment
<memleak>
and yes the cigars that you payed for
<memleak>
(not won)
<mrnuke>
I won at an suction
<mrnuke>
make that "auction
<memleak>
o_rly_owl.png
<mrnuke>
Well, I guess suction is also an applicable term
<memleak>
you won it at a vacuum..
<memleak>
space?
<mrnuke>
No, I do not have humidor space for them. I don't know where I'm going to keep all of them
<memleak>
XD
<mrnuke>
I'm in the process of ordering a custom-made hardwood humidor BTW
<memleak>
hey who was that guy again that talked about you during a conference?
<memleak>
something about you being crazy?
<memleak>
because whoever said that was spot on too XD
<memleak>
if you simply smoked your cigars faster (like me) you wouldnt need a humidor.
<VargaD>
Hi, I just migrated my system from cubieboard 1 to cubietruck, I'm using stock kernel config (make ... sun7i_defconfig), my ssh session stalls and timeouts many! times
<memleak>
VargaD, did you configure /etc/hosts correctly?
<VargaD>
I ssh from my desktop to the cubietruck
<VargaD>
I have never seen similar before the network topology doesn't changed it work fine with cubieboard
<VargaD>
when the openssh stalls I can't connect, but after I can connect again
<VargaD>
I can' see anything special in the logs just timeout
<Turl>
VargaD: maybe you have a bad network cable between CT and network equipment
<memleak>
I'd definitely take a look at /etc/hosts.allow and /etc/hosts.deny
<VargaD>
maybe, but the apache and other services works just fine while the openssh stalled
<VargaD>
ok I check it
<memleak>
you gotta be sure all dem permissions are config otay odderwize it may not work
<VargaD>
I don't have those files, It's a gentoo btw
<Turl>
memleak: <insert link to english do you speak it meme>
<memleak>
i ahpoepoejize my english not vry good...
shineworld has joined #linux-sunxi
<VargaD>
are you sure it can be a permission problem? I can loging, I even have pair of authentication keys, so I can login without password, it worked fine for about a year
<VargaD>
I can still login and it works fine
<memleak>
ohhhhh
<VargaD>
... until it stalls
<memleak>
ohhhhhhhhhhhhhhhhhhhhhhhh
<VargaD>
the I get timeout
<memleak>
OHHHHHHHHHHHHHHHHHH!!!!!!!!!!!
<VargaD>
I just updated openssh to 6.5_p1-r1 which is the newest in gentoo but still have the same problem
<memleak>
hmmmmm
<VargaD>
strange pulseuadio works just fine
<VargaD>
I redirect sound from my desktop
<VargaD>
It seems other services works fine
<memleak>
I'd say listen to Turl
<Turl>
VargaD: raw audio? or compressed?
<Turl>
VargaD: maybe you are just starving the bandwidth
<memleak>
for example are you using lame / mpg123 / mpg321 etc
<VargaD>
honestly, I don't know what does pulseaudio sends when I change output
<Turl>
VargaD: it's probably uncompressed
<Turl>
VargaD: can you disable it and see if ssh still stalls?
<memleak>
Turl, by uncompressed do you mean wav or like...?
<VargaD>
probably, it works fine on local network
<Turl>
memleak: yeah, wavlike
<Turl>
memleak: it's rtp I think
<memleak>
ah ok
<memleak>
rtp?
<memleak>
real time processing?
<Turl>
memleak: can easily kill a reasonably sized pipe
<memleak>
ah
<Turl>
memleak: "The Real-time Transport Protocol (RTP) defines a standardized packet format for delivering audio and video over IP networks."
<memleak>
ahhhhhhhhhh
<VargaD>
unfortunately ssh is the most important sevice on cubietruck
<memleak>
VargaD, tried using a diff client such as SSH-LIKE such as VNC. windows 7 remote desktop feature, VSSH?
<VargaD>
I'm also sure the cubietruck can't overheat as I have a 12V old AMD cpu cooler on it
<memleak>
(windows 7 remote desktop thing was a joke)
<memleak>
vssh though would be worth a shot i guess
<Turl>
osh
<Turl>
mosh*
pirea has joined #linux-sunxi
<VargaD>
I try an other ssh server
<memleak>
mosh, bosh, bosch, bosch fuel injectos and O2 sensors
<memleak>
O_o
<memleak>
oh hey everyone its st pattys day, i forgot