eebrah|mobile has quit [Ping timeout: 240 seconds]
eebrah_ has joined #linux-sunxi
MasterChief101 has joined #linux-sunxi
MasterChief10 has quit [Ping timeout: 264 seconds]
eebrah_ has quit [Ping timeout: 240 seconds]
Renard has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
MasterChief101 has quit [Ping timeout: 250 seconds]
eebrah|mobile has joined #linux-sunxi
libcg has joined #linux-sunxi
MasterChief10 has joined #linux-sunxi
MasterChief10 has quit [Client Quit]
eebrah|mobile has quit [Ping timeout: 240 seconds]
gzamboni has quit [Read error: Connection reset by peer]
FR^2 has joined #linux-sunxi
eebrah|mobile has joined #linux-sunxi
astr has quit [Ping timeout: 255 seconds]
eebrah|mobile has quit [Ping timeout: 240 seconds]
astr has joined #linux-sunxi
xinj has joined #linux-sunxi
bgal has joined #linux-sunxi
BorgCuba has joined #linux-sunxi
<BorgCuba>
is it possible that early printk is not deactivated automatically after ttyS0 is up?
Renard has quit [Quit: Renard]
Renard has joined #linux-sunxi
Renard has quit [Remote host closed the connection]
formruga has joined #linux-sunxi
gzamboni has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
Black_Horseman has quit [Remote host closed the connection]
<oliv3r>
mnemoc: what about 3.0?
<oliv3r>
mnemoc: and what was the first 2.6 community kernel based off of, the ainol leak or the AW release
<mnemoc>
luke and #arm-netbook began trying to make sense out of ainol's release, pushed to the debian git
<mnemoc>
but nothing really happened until hipboi published the the .git in github
<oliv3r>
so we never really touched the ainol stuff
<oliv3r>
so if i say, ainol leaked, aw released, community started to work on the aw stuff; that'll be accurate enough
<mnemoc>
it gave us the needed hope to start the community
<mnemoc>
then 2.6.36 forwardning to 3.0 began, but the effort was trashed after qware's release
<oliv3r>
ok and 3.0?
<oliv3r>
qware released and aw followed which we picked up
<mnemoc>
please don't use the term leaked. GPL can't be leaked. it's liberated or released
<oliv3r>
well they leaked it from aw as aw didn't want it to be released (illegally)
<mnemoc>
qware and ainol honoured the GPL, they didn't LEAK anything. they made their duty
<oliv3r>
against AW wishes
<mnemoc>
sure
<mnemoc>
but it's not a leak
<oliv3r>
but i'll use the term liberated :p
<oliv3r>
i like it much better
<oliv3r>
and for 3.0? all the work started over again after the AW release?
<mnemoc>
the GPL code was been held hostage by AW
<oliv3r>
lol yeah
<mnemoc>
3.0-v2 began with qware's
<mnemoc>
from there we began the real work
<mnemoc>
that v2 became the 3.0 of linux-sunxi
<mnemoc>
since then the only aw code has been snippets extracted from this or that SDK released by different device vendors, specially olimex
<mnemoc>
and own development, and reimplementations
<oliv3r>
so when they finally started to officially release 3.0 sources, we where 'far ahead of them'
<mnemoc>
not sure if "far ahead" is the right term, but we had diverged
<mnemoc>
specially considering they keep per-soc trees, and we unified everything
<mnemoc>
per-soc trees even developed by different teams and inconsistent between each other, and with different versions of the drivers of their IPs
<oliv3r>
yeah
Renard has joined #linux-sunxi
<oliv3r>
mnemoc: you got started on a 3.14 hybrid LTS kernel right?
<mnemoc>
oliv3r: currently matches 3.15... but I haven't found time or energy to continue
<mnemoc>
and it's not hybrid yet. pure mainline
<oliv3r>
3.15? i thought we'd want to aim for 3.14 as that is an LTS version
<mnemoc>
it is 3.14 with all sunxi-ness of 3.15
<oliv3r>
ohh
<oliv3r>
so 3.14 + all sunxi patches backported
<oliv3r>
roger
formruga has quit [Quit: Konversation terminated!]
jemk has joined #linux-sunxi
<mnemoc>
next would be, keep backporting and testing, then improving mripard's babelfish to detect soc and convert all script.bin into dtb, adding a glue module emulating the legacy api, and bring in drivers not yet supported by mainline
<oliv3r>
lots of work still
<oliv3r>
if you can make that happen in the next 2 weeks I can use a hybrid kernel in my project! :)
<mnemoc>
i can't
<oliv3r>
busy with $work?
<mnemoc>
and out of energy
<libv>
it's been 2 years since sunxi was formed
<libv>
2 years is about right to lose drive, especially if other events get in the way
<oliv3r>
mnemoc: what happened to your energy
<oliv3r>
libv: i have plenty of drive! i'm using an a20 for our next product, wriiting a book! just no time to code :(
<oliv3r>
libv: btw, eva got back to me; she appologized and said she'd get back to me soon, my mail got into her junk
<oliv3r>
which i find odd
<libv>
oliv3r: heh, you have had encouraging events happening to you, and it is amazing that your work is interested in sunxi hw
<libv>
oliv3r: if you do not have that, and things continue like that, you would've lost drive quite a while ago
<libv>
as for eva, that's bullshit
<oliv3r>
libv: well i made them interested in sunxi hardware
<libv>
what next, her dog ate my GPL email(s)?
<oliv3r>
libv: i decided what HW platform to use :D
<oliv3r>
they where even thinking of using a pi :(
<oliv3r>
libv: yeah like she'd check her junk folder for messages
<libv>
again, that was mostly luck, you were there at a time when they were in need of such a device, and you gave them the better option
<oliv3r>
that is true so
<oliv3r>
libv: what happened to your drive and motivation?
<libv>
i have been doing lima since march 2011
<oliv3r>
too long now?
<libv>
lima was one of the reasons why i dropped what now looks like a perfectly good job at SuSE, and decided to stick with codethink
<libv>
lima has been mostly my spare time
<libv>
and codethink only used it to make some noise, but wasn't really interested to run with it properly
<libv>
so i feel like i got used, to some extent, as i let myself be used
<libv>
now, nobody will back lima, in a way that makes this liveable
<oliv3r>
that totally sucks
<oliv3r>
would you work for a bounty?
<libv>
and the position i have had to put myself in to make open arm gpu drivers happen makes me even less desirable as an employee
<libv>
and all of that is before i count in my previous 8 years of graphics driver development, where i have been doing similar things, with broadly similar results
<oliv3r>
how much $ would be required for you to keep working on lima stuff?
<libv>
i should actually be happy with lima, as this time round, it is clear that it was me, on my own, and people could not naysay me or trash me for it (which they happily have done for things like modesetting and freeing ati)
<oliv3r>
you are not the only one that is proud
<oliv3r>
there are many that are proud of what you have accomplished so far!
<libv>
anyway, i fully understand that mnemoc is out of energy
<libv>
2ys of sunxi, 2.5ys of working on the platform, helping to form cubietech, etc
<oliv3r>
yeah time for a little break I suppose :(
astr has quit [Ping timeout: 245 seconds]
<libv>
given mnemoc his personal situation, and the fact that his job offers no chance of working on sunxi, i fully understand it
<oliv3r>
libv: speaking of coding, i want to forwardport the libnand driver from the lichee u-boot to sunxi-boot. think that could work?
<mnemoc>
finishing the mtd driver makes more sense imo
<libv>
oliv3r: yes, but... would you really want to do that knowing that there are much newer libnands out there?
<libv>
mnemoc: indeed
<libv>
especially since oliv3r will need it for its own installations and not for being compatible with android or somesuch
<libv>
s/its/his/
<BorgCuba>
hi, is the lima+mesa code available?
<mnemoc>
i suppose it's also possible to make a backward compatible "nand" layer/partition table over the mtd driver
<libv>
BorgCuba: no.
<libv>
BorgCuba: thanks for playing.
<BorgCuba>
okay
nabblet has joined #linux-sunxi
<oliv3r>
mnemoc: finishing the mtd driver requires a LOT more work and knowledge, forwardporting the libnand driver is copy/paste more or less
<oliv3r>
the new bbrezillon based driver is highly experimental
<mnemoc>
product-wise, sure.
<oliv3r>
i cannot in good concioness use that in our product, even though a mainline kernel should and would probably be better
<oliv3r>
so with being stuck on 3.4, i'm stuck with libnand
<mnemoc>
iirc libv even REed parts of the blob to extract new data for the table
<libv>
yup
<oliv3r>
being stuck with libnand means either boot0/boot1 or a forwardported u-boot+libnand
<libv>
hence eva not having her gpl complaint emails eaten by her dog would've been really helpful
<oliv3r>
I did ask her for that :)
<oliv3r>
the only advantage is that we have pretty much known nand, as olimex supplies our boards
eebrah|mobile has joined #linux-sunxi
<mnemoc>
oliv3r: wasn't your mail also eaten by her dog?
eebrah|mobile has quit [Client Quit]
<oliv3r>
btw, Benn even contacted me about th ebook :)
<oliv3r>
mnemoc: yeah but she got back to me and appologized
<oliv3r>
so now i'm just going to be patient
astr has joined #linux-sunxi
<mnemoc>
oliv3r: can you poke Benn to get some info about the A80 and some cb8 prototypes for the mainlining team?
zeRez has joined #linux-sunxi
<mnemoc>
CT is close to AW. the latest charbax video from AW shows the cb8 board
oliv3r_phone has joined #linux-sunxi
<oliv3r_phone>
mnemoc but a80 iz dirty powervrness :(
<mnemoc>
shit happens
<mnemoc>
even as headless it's a very nice SoC, and it's a sunxi
<BorgCuba>
btw, I think i read on olimex that the a23 is in tqfp package, i guess this is not true?
<oliv3r_phone>
yeh and it will work as a basis for a40 im sure
<libv>
BorgCuba: bga.
<BorgCuba>
yes, I thought so. thanks for clearing this up libv
<libv>
BorgCuba: look at the wiki, there's a photo there
<BorgCuba>
At least I havent found any kernel sources on their page
oliv3r_phone2 has joined #linux-sunxi
oliv3r_phone has quit [Read error: Connection reset by peer]
ricardocrudo has joined #linux-sunxi
<oliv3r_phone2>
bad wifi in the toilet :(
<BorgCuba>
thick air?
xinj has quit [Quit: xinj]
<libv>
oliv3r: that's use case #1 for wifi at my place :p
xinj has joined #linux-sunxi
<oliv3r_phone2>
haha yeah. i use my phone THE most there :)
<libv>
the first thing i did when i moved to nue was get a special stool to rest the laptop on. last time i went to ikea, i noticed that they were still selling them :)
tomcheng76 has joined #linux-sunxi
<libv>
al bundy is an old school man, grabbing the newspaper and tucking it under his arm and then casually walking off
<libv>
when i walk past with my laptop in my hand, my other half knows where i am heading :)
<oliv3r_phone2>
LOL
<oliv3r_phone2>
so true
<oliv3r_phone2>
though i dont think id get away with a laptop stand in the loo
<libv>
grounds for divorce :p
<libv>
i've even taken a laptop and a tablet attached to otg in :)
<BorgCuba>
I guess I have to try qgl then ;)
nabblet has quit [Quit: leaving]
<oliv3r>
libv: lol I have done that, and similar things :p
<oliv3r>
but a permanent stand ...
<libv>
it's just a small stool
<oliv3r>
i use my lap mostly :)
avsm has joined #linux-sunxi
xinj has quit [Read error: Connection reset by peer]
<BorgCuba>
libv, why does liblimare depend on "__mali_compile_essl_shader"? I thought it is a full replacement for the binary libs?
xinj has joined #linux-sunxi
<wens>
oliv3r: careful not to drop your phone :p
<oliv3r>
haha
<oliv3r>
and flush it :p
<wens>
ouch
<wens>
dropping it is bad enough
y0g1 has quit [Read error: Connection reset by peer]
libcg has quit [Remote host closed the connection]
<wens>
just think of it as a staging tree over lts, maintained by the linux foundation
<oliv3r>
ah ok
<oliv3r>
so there'll be LTS, and LTSi then?
<oliv3r>
not bad
ricardocrudo has quit [Ping timeout: 245 seconds]
<wens>
LTSI's purpose is to get vendors aligned with LTS, while having a relaxed submit process for some of their needs
<wens>
like backporting a driver/subsystem from the latest release, or just staging a driver from vendor tree (like the staging area in mainline)
<wens>
also mentioned was a common test suite for kernel features (though not available atm)
ogi has joined #linux-sunxi
<heffer>
i just received a new board and it hangs just after "DRAM" in u-boot. i used the latest git clone from u-boot-sunxi and the dram values from a10-meminfo, so i'd assume those are right. any hints on where to start looking?
<heffer>
right. i have this board on my desk for like 90 minutes
<heffer>
not exactly
<libv>
follow that howto, in the order as is listed.
<libv>
if you don't, there is no point in us trying to help you, or you even asking for help
<heffer>
okay. thank you so far.
zeRez has quit []
physis has joined #linux-sunxi
prahal has joined #linux-sunxi
<astr>
I can't get the lcd to work on my lime
<astr>
I did have a opps of the lime was powered up and I connected to the lcd to the wrong pin set
prahal has quit [Quit: prahal]
<astr>
the 3v power is slightly less 2.somethingV the 5v is only messuring 0.50v
<astr>
have I fried it?
prahal has joined #linux-sunxi
oliv3r_phone2 has quit [Ping timeout: 240 seconds]
<bonbons>
ccaione: any idea where the axp20x device tree patches are on their path to mainline? Looking at 3.16-rc5 I don't see them...
rinni has joined #linux-sunxi
<bonbons>
I'm looking at assembling power_supply subdrivers for your axp20x MFD driver
froese has joined #linux-sunxi
<rinni>
I have a small questiion concerning the mempry addresses in boot.cmd: how do I convert them to MB so I know my kernel and my initramfs will not overlap?
<rinni>
right now I have http://sprunge.us/XVGN with kernel and intramfs ~2.5MB - I'm on a Cubieboard2 btw
<rinni>
coudln't find any documentation about it :(
oliv3r_phone has joined #linux-sunxi
<froese>
they are just memory addresses in hexadecimal.
<froese>
drop the rightmost 5 digits to convert from byte to megabyte
<rinni>
froese: OK, thanks!
<rinni>
froese: Just to make sure I got it right - anexample from my boot.cmd: 0x45700000 - 0x43000000 = 0x02700000 = 27MB, correct?
<FergusL>
hi, should I use the same cross-compile toolchain used for the kernel to build packages for cubieboard ?
<oliv3r>
don't have to, but possible sure
xavia has joined #linux-sunxi
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 250 seconds]
avsm has joined #linux-sunxi
<BorgCuba>
what about gnueabi gnueabihf and probably elf, are they compatible?
<BorgCuba>
I am currently using gnueabi toolchain and libMali gnueabihf
Black_Horseman has joined #linux-sunxi
linkmauve1 has joined #linux-sunxi
libv_ is now known as libv
<Black_Horseman>
Hell -o
<Turl>
BorgCuba: you cannot use a gnueabihf binary if the rest of the userspace you have is gnueabi, and viceversa
mihu has joined #linux-sunxi
avsm has quit [Quit: Leaving.]
<libv>
no hf?
* libv
looks at a watch that he doesn't wear anyway
<libv>
in 2014?
<libv>
BorgCuba: did you ever even manage to install the binaries?
<libv>
is this why you are now trying to play with lima?
<BorgCuba>
no
<BorgCuba>
what do you mean by installing the binaries?
<BorgCuba>
I am playing with lima out of curiosity I would say
<ssvb>
jemk: thanks for sharing the new tpr3 scan data for your cubietruck
<BorgCuba>
Turl, thats probably the reason why that program I compiled says nothing
<ssvb>
jemk: you can probably also try zq values 0x1c, 0x3c, 0x2b, 0x2d (the immediate neighbors of 0x2c)
souther has quit [K-Lined]
<libv>
BorgCuba: then you can use any binary
<BorgCuba>
ssvb, my a13 board is now running (with only one cable attached - g_ether). thanks again for the help last time!
<libv>
BorgCuba: provided it matches the eabi of the rest of the system
<libv>
the lima driver doesn't care which version of the binary compiler it is
<ssvb>
BorgCuba: you are welcome
<libv>
well, it does, but it is, or should be, smart enough
<BorgCuba>
libv, I compiled mali_compile an when I run it, it just says nothing when I think it should print usage()
<BorgCuba>
while the info program does work
<BorgCuba>
-> r3p0 with 1x GP and 1x PP
<libv>
yes, but that's kernel
<libv>
but there are probably no non hf binaries for that
* libv
digs around a bit
<Turl>
libv: iirc r3p0 has binaries for everything
<Turl>
why are the android blobs inside armel? o.O
<libv>
mali_compile does not care about the kernel
<oliv3r_phone>
on a chapter that covers hardware interfacing (in its most basic way) what would you guys want r expect to see?
<BorgCuba>
libv, my current understanding is that I only need these mali binaries for compiling the shaders, or am I wrong?
<Turl>
oliv3r_phone: what do you mean by that?
<BorgCuba>
oliv3r_phone, a gpio pin an a button?
<BorgCuba>
with pull-up/down
<oliv3r_phone>
exactly what bc says
<oliv3r_phone>
i got that covered i guess in its most basic form
<Turl>
ah
<Turl>
an LED? :p
<oliv3r_phone>
that too!
<Turl>
that's about it for basic stuff I guess
<BorgCuba>
one for input one for output
<oliv3r_phone>
well turl you know the audience, what else?
<oliv3r_phone>
i did mention spi, i2c etc, which is something to look into if theyd want to learn more
<Turl>
I dunno, other than those basic ones
<ssvb>
libv: how much code might be needed to setup hdmi and disp to implement a very simple framebuffer (edid or some fixed resolution) for u-boot?
<libv>
ssvb: no idea
<libv>
ssvb: 1kloc? 2?
<libv>
depends on what u-boot wants
rinni has quit [Quit: Page closed]
<jemk>
ssvb: lima-memtester is running since yesterday with this configuration without errors :) btw, how did you get to the 0x2c value? all defaults are around 7 for odt div
<libv>
BorgCuba: yes, lima mostly depends on the binary shader compiler, as it is two projects in one
<ssvb>
jemk: the default is 0x7b, I just tried changing both high and low bytes in different directions to see whether this improves reliability or degrades it
<ssvb>
jemk: yeah, this needs a lot of time for running that many tests
<ssvb>
jemk: I mean low and high zq 4-bit hex digits
<libv>
it's actually amazing
<libv>
the ratio of people getting told to NDH versus the actual device pages being created
<ssvb>
jemk: and I did not even test the whole space of possible values, so 0x2c might be not the best one, but just a local maximum
<jemk>
ssvb: I just think its strange from an electrical point of view that such different impedances work
<jemk>
ssvb: or i didn't understand correctly how these dividers are related to the actual impedance
<ssvb>
jemk: we don't know the exact impedance values are derived from that (the rk30xx manual provides some description, but is beating around the bush without giving the exact formula)
<ssvb>
jemk: yes, these are just the magic numbers
<ssvb>
jemk: but 240/2 ohm termination impedance and 240/12 for the output drive impedance are not exactly unreasonable, the DDR3 spec offers impedance configuration options in a similar ballpark
<ssvb>
jemk: and we don't know how the zero divisor is supposed to be interpreted
<ssvb>
jemk: on the other hand, the default 0x7b looks rather strange and performs poorly
<jemk>
ssvb: I'm not sure about how this all works with ddr3, but 240/7 as temination could be too much load, we have four chips on ct, thats around 8ohm. ...but while writing this i notices this porgrams the controller, not the dram chips, so only one
<ssvb>
jemk: for the impedance on the ddr3 chips side we have dram_emr1
<ssvb>
jemk: we are configuring impedance on both sides: zq - for the dram controller, emr1 - for the dram chip
<ssvb>
jemk: and I also wondered whether some of the settings are actually safe and whether they can damage the hardware
<jemk>
ssvb: the general rule for good hf lines is output = line = termination impedance
kz1 has quit [Ping timeout: 256 seconds]
avsm has joined #linux-sunxi
<ssvb>
jemk: yeah, but I guess that this is just a simplistic model, kind of like ignoring air resistance when calculating bullet trajectory, or something like this
<jemk>
ssvb: and with this in mind all these different values look strange to me
<ssvb>
jemk: why are various ddr3 papers recommending different impedance values, not equal to the typical 50 ohm impedance?
<ssvb>
jemk: in any case, experiment always beats theory :)
<jemk>
ssvb: I'll have to read some things again, don't want to talk bullshit, I only had basic lecture about hf
<jemk>
ssvb: I know there where some papers about impedance in ddr3 systems, and it was very complex, so maybe you are right and experiment bears theory
<jemk>
*beats
ricardocrudo has joined #linux-sunxi
rz2k has joined #linux-sunxi
xinj has quit [Ping timeout: 264 seconds]
mihu has left #linux-sunxi [#linux-sunxi]
<ssvb>
jemk: btw, the write errors in the lane 0 have disappered in your updated tpr3 table
<ssvb>
jemk: I guess, this might indicate that these problems were impedance related
<ssvb>
jemk: actually, regarding the signal propagation speed and delay adjustments, I'm leaning to think that it can't be a significant problem at such short distances
<ssvb>
jemk: we have much shorter PCB tracks than the typical desktop systems, and also lower dram clock speeds
<jemk>
ssvb: well, at 600mhz this distance isnt that short anymore, it gets near the relevant lenghts
<jemk>
ssvb: and i guess common desktop systems are much better designed than our boards
<ssvb>
jemk: the signals in PCB are propagating at roughly half of the speed of light in vacuum
<ssvb>
jemk: that's roughly half a meter per one dram cycle at 600MHz
kz1 has joined #linux-sunxi
bgal has quit [Ping timeout: 240 seconds]
<jemk>
ssvb: ok, i meant relevant for impedance
<ssvb>
jemk: oops, I got the math wrong and it is ~25cm
<ssvb>
jemk: anyway, the length mismatch between different tracks can't be that large, or can it?
<ssvb>
jemk: yes, the experiments show that the impedance configuration seems to be the most important for the dram reliability in our case
<jemk>
ssvb: don't know, too late for that much math :p
<BorgCuba>
are you designing a board?
<jemk>
ssvb: I could even run with tpr3=0x0 now
<jemk>
BorgCuba: no, we only try to understand the dram controller configuration
<ssvb>
BorgCuba: just trying to find the best dram controller configuration options for the existing boards
<jemk>
ssvb: I'll try to find what i had again, but i had this ages ago, not sunxi related, and didn
<jemk>
't bother keeping it
FDCX has joined #linux-sunxi
<jemk>
ssvb: your pdf looks good too, lets see what it says. but at the moment i'm pretty happy with the ct performance. thanks for doing all this experiments, that must have took ages. i already get impatient at the second scan-tpr3 run ;)
<jemk>
i should better leave now, see you
jemk has quit [Quit: leaving]
sehraf has quit [Read error: Connection reset by peer]
avsm has quit [Quit: Leaving.]
FR^2 has quit [Quit: Leaving]
bonbons has quit [Quit: Leaving]
<BorgCuba>
I compiled sth using a gnueabihf tc and I build the kernel previously using gnueabi. now when I want to start that program I compiled it says : No such file or directory. Do I have to rebuild the kernel?
<BorgCuba>
using gnueabihf toolchain
Gerwin_J has joined #linux-sunxi
<ssvb>
BorgCuba: this looks like an ABI problem, try to run ldd on this program
<ssvb>
most likely it has a broken reference to ld-linux.so.3 or something like this
ricardocrudo has quit [Ping timeout: 240 seconds]
prahal has joined #linux-sunxi
prahal has quit [Quit: prahal]
<BorgCuba>
ssvb, sounds reasonable. the linker complains that it cannot find libUMP.so.3 although it should
<BorgCuba>
I checked both libs paths and headers (using readelf) and it seems okay
bertrik has quit [Remote host closed the connection]