<Gumboot>
What the world clearly needs is a CPU on a board with a wee bit of flash and half a dozen properly-powered USB ports. No display. No audio. No GPIO. Just attach what you need via USB at extra cost.
<xenoxaos>
until you run out of bandwidth
<Gumboot>
Attach more bandwidth!
<xenoxaos>
with duct tape and hot glue
<Gumboot>
Actually, you probably could. You could daisy-chain them so that you get more CPUs with more bandwidth communicating back just the essentials.
lerc_ has joined #arm-netbook
<xenoxaos>
the programming logistics on that would be hell
<Gumboot>
It's for education!
lerc has quit [Ping timeout: 240 seconds]
<xenoxaos>
But I don't want to go to sodomy school.
<Gumboot>
You could even connect them in a ring.
<Gumboot>
I would like to disassociate that last comment from the sodomy reference, if possible.
<xenoxaos>
nope
<Gumboot>
Fine. I think there's a word for it, but I'm not going to google it.
<xenoxaos>
!g ring of sodomy
<xenoxaos>
oh...there's a picture....
mnemoc has quit [Excess Flood]
mnemoc has joined #arm-netbook
RITRedbeard has quit [Ping timeout: 245 seconds]
tzafrir_laptop has quit [Ping timeout: 260 seconds]
RITRedbeard has joined #arm-netbook
stefanro1 has joined #arm-netbook
stefanro has quit [Ping timeout: 260 seconds]
focus_well has quit [Ping timeout: 245 seconds]
* hno
will likely to token ring/bus networking in the near future. Thought that was dead.
focus_well has joined #arm-netbook
tuliom has quit [Quit: Konversation terminated!]
alcides has quit [Quit: There's no place like 127.0.0.1]
tinti has joined #arm-netbook
<Marex>
hno: token ring is splendid ... one ring to rule them all (tm) ;-)
<Marex>
btw. mail.google.com is down? nice ;-)
<Marex>
not anymore ... weird
tinti has quit [Quit: Leaving]
freakazoid0223 has joined #arm-netbook
tzafrir_laptop has joined #arm-netbook
avernos has quit [Ping timeout: 246 seconds]
marcan has quit [Read error: Operation timed out]
avernos has joined #arm-netbook
marcan has joined #arm-netbook
markatto has quit [Ping timeout: 252 seconds]
grevaillot has quit [Ping timeout: 240 seconds]
grevaillot has joined #arm-netbook
markatto has joined #arm-netbook
<Mehhh>
Does anyone here have a lapdock?
popolon has quit [Quit: Quitte]
mSquare has joined #arm-netbook
<xxiao>
how can I avoid the fxxx windows all together
<xxiao>
can't recall how to get its env variable, echo $VAR1?
* xxiao
uses windows 10 hours per year
freakazoid0223 has quit [Read error: Connection reset by peer]
<Marex>
xxiao: to do taxes ? ;-)
<Marex>
wasnt it %VAR ?
<xxiao>
Marex: that's right
<xxiao>
it's something like %VAR%? anyway i managed to get it working, JAVA_HOME and PATH both need to be set manually
<HACKhalo2>
heh
Quarx has joined #arm-netbook
jquip has joined #arm-netbook
jquip has quit [Read error: Connection reset by peer]
<libv>
well, check what numBuffers and bufferMask are doing
<libv>
erm, you do want to fix things properly
<Turl>
yeah I know
<Turl>
was just a PoC :)
<Turl>
libv: numBuffers is 2
<libv>
something has changed
<libv>
in surfaceflinger or gralloc
<libv>
but do review the life of numBuffers and bufferMask
<libv>
through logging
<Turl>
other platforms have supported triple buffering for a while I think
<libv>
well, it could be a bug in the handling of numBuffers/bufferMask between allocs and frees
<libv>
or it could be an actual shortage of memory
<libv>
memory assigned to the mali
<hno>
libv, the Allwinner logical nand driver is not submittable without anyone understanding the logical layer and stepping up to document it and accept being a maintainer.
<Turl>
hardcoding numBuffers to 3 on gralloc seems to work
<hno>
libv, but we will likely have a well functioning MTD driver.
<Turl>
I need to check where does numBuffers come from
<libv>
i do not know how fb responds to that though
<libv>
might be driver dependant
Mazon has quit [Quit: meh]
<libv>
if both the samsung and your allwinner return that, then you might get away with it
<libv>
the change to services/surfaceflinger/Layer.cpp is rather scary
<Turl>
if they made a makefile define switch to change it it must be not that scary
<libv>
you would expect the hw framebuffer module to feed that back
<libv>
and not be hardcoded somewhere up top
<libv>
it's clearly dynamic for surfaceflinger, but these defines make it semi-hardcoded
<libv>
and the hw has no say in it
popolon has joined #arm-netbook
SouL_ has joined #arm-netbook
<Turl>
they also have a NUM_FRAMEBUFFER_SURFACE_BUFFERS
SouL_ has left #arm-netbook [#arm-netbook]
svp has joined #arm-netbook
ssspiff has joined #arm-netbook
sspiff has quit [Ping timeout: 250 seconds]
vgrade_ has joined #arm-netbook
Mazon has joined #arm-netbook
tinti has joined #arm-netbook
focus_well has quit [Quit: Leaving]
<Gumboot>
It really is tempting to set up an rpi hate site and fill it full of adverts and a couple of unkind remarks. Can you make money on adverts just being displayed, or do they have to be clicked?
<mnemoc>
at least google ads pays for both, obviusly differently
<Gumboot>
Maybe I really should, then. I wonder how much I can make.
<Gumboot>
And if I did, what should I do with the money?
<mnemoc>
buy rpis
jquip has joined #arm-netbook
<Gumboot>
If the DSP toolchain makes any progress I'd be tempted.
<Gumboot>
It's only the ARM which is completely useless.
tzafrir_laptop has quit [Ping timeout: 245 seconds]
<RaYmAn>
useful commit messages there ;)
<libv>
yeah :(
<Gumboot>
Nevermind the commit messages.... does it work?
<Gumboot>
I probably shouldn't look. It'll drive me mental knowing that they've got little mistakes here and there which I can't correct.
<Gumboot>
Does the bootloader include the GPU driver, or is that uploaded at a later stage?
<mnemoc>
i've seen commit messages about needing to go for coffee or to the grocery store :|
<libv>
Gumboot: the 2mb blob which is later contains that i guess
<libv>
loaded later even
<Gumboot>
Well that's only the same size as the startup elf file.
<Gumboot>
I hope these binutils use the same elf metadata as is already used in the existing elfs. It would be easy to just start from scratch and forget about compatibility.
avernos has quit [Ping timeout: 246 seconds]
focus_well has joined #arm-netbook
ssspiff has quit [Remote host closed the connection]
sspiff has joined #arm-netbook
avernos has joined #arm-netbook
arete74 has joined #arm-netbook
tzafrir_laptop has joined #arm-netbook
jquip has quit [Ping timeout: 245 seconds]
jquip has joined #arm-netbook
arete74 has quit [Ping timeout: 256 seconds]
RITRedbeard_ has joined #arm-netbook
madmalkav has joined #arm-netbook
hipboi has left #arm-netbook [#arm-netbook]
itamar_ has joined #arm-netbook
Gumboot has quit [Ping timeout: 245 seconds]
Workboot has quit [Ping timeout: 248 seconds]
SouL_ has joined #arm-netbook
SouL_ has left #arm-netbook [#arm-netbook]
SouL_ has joined #arm-netbook
SouL_ has left #arm-netbook [#arm-netbook]
Gumboot has joined #arm-netbook
ibrah has quit [Ping timeout: 255 seconds]
SouL_ has joined #arm-netbook
SouL_ has left #arm-netbook [#arm-netbook]
cat_x301 has quit [Ping timeout: 256 seconds]
vinifm has joined #arm-netbook
cat_x301 has joined #arm-netbook
vinifm has quit [Quit: Ex-Chat]
svp has quit [Quit: Leaving.]
<slapin>
do anybody knows how I can link custom section to .rodata without using linker script? this is userspace program, so I don't want to mess with linker scripts for it
antonl has quit [Ping timeout: 248 seconds]
Jef91 has joined #arm-netbook
hg_5_ has joined #arm-netbook
hg_5_ has quit [Client Quit]
hg_5 has joined #arm-netbook
Gumboot has quit [Ping timeout: 255 seconds]
Gumboot has joined #arm-netbook
<oliv3r>
if i have a 4 Byte variable, and i do 3 << 30; that can't be right, correct?
<mnemoc>
it's fine
<oliv3r>
won't it overflow?
<oliv3r>
oh wait, no
<oliv3r>
i'm stupid
<mnemoc>
3 is 0b11
<oliv3r>
yeah
<oliv3r>
i'm stupid :)
<mnemoc>
yes :)
<oliv3r>
... thanks :p
<oliv3r>
i was thinking of 3 being 0b111
<oliv3r>
i had that still in my calculator, from an earlier 7 << 0
<oliv3r>
:p
<mnemoc>
anyhow, unsigneds don't overflow in C
<mnemoc>
7<<30 == 3<<30
<oliv3r>
what about chip registers :p
<mnemoc>
should be the same
<oliv3r>
still, it COULD cause bugs!
<oliv3r>
crashes, fires, skynet's self awareness!
<mnemoc>
take the ref guide and see the shift instruction you are using
<mnemoc>
the excess could be sent to another register
<mnemoc>
but i'm not an asm guy
<oliv3r>
oh, nah, i just thought i found a bug in a header file :p
<mnemoc>
nope :)
<oliv3r>
i'm wikiing things :p
<mnemoc>
which controller?
<oliv3r>
yeah, cause I was being stupid
<oliv3r>
sec, i'll save my intermediate work
<oliv3r>
htt://linux-sunxi.org/A10/NFC
<mnemoc>
afaik the NFC is identical in all sunxi socs
<oliv3r>
then we need a better naming convention, since we started with A10/registers
<oliv3r>
i just went from there!
<mnemoc>
that's fine, but I believe A10/NFC should redirect to just NFC
<mnemoc>
but it can be moved later
<oliv3r>
yep
<oliv3r>
I think a lof of devices and controllers are identical
<mnemoc>
do you want to learn SVG? :)
<mnemoc>
tables to describe registers suck badly :<
<oliv3r>
er, not really :( why
<oliv3r>
ok, NOW you say this?
<oliv3r>
you have any clue how much time i've spent writing register tables?
<mnemoc>
a lot
<mnemoc>
:)
<oliv3r>
:p
<oliv3r>
what did you have in mind?
<mnemoc>
an image :)
<oliv3r>
obviously :p how much would it be different though?
<oliv3r>
I know a table isn't ideal
<oliv3r>
I wonder how an image would be more usefull
<mnemoc>
adds visual meaning to the bit ranges
<oliv3r>
do you have an example of such an image?
<oliv3r>
'interactive' memory map would be way cooler :p
<oliv3r>
I can try to do it for the next controller
<oliv3r>
yes
<oliv3r>
bfree: exactly that
<oliv3r>
mnemoc: in either case, we'd need a good template for the registers to begin with
<mnemoc>
hopefully a bit larger so the text can be read :p
<mnemoc>
oliv3r: yes, next controller, always forward
<oliv3r>
well the older stuff will have to be adapted at some point
<bfree>
that could be doable on the fly in javascript from the html tables in the wiki output ;) but if you have a lot of bits in there where is all the text going to go and is it really going to be much of an improvement?
<oliv3r>
if you can whip up a template + example, i'll use it on the next controller :)
<oliv3r>
you'd only have bit + name
<oliv3r>
or even just bitnumber, as you can look up the bit number in the table (from template) below
madmalkav has quit [Read error: Connection reset by peer]
<mnemoc>
there is a wip/sunxi-next/i2c there too, but cleanup only, no porting yet
<mripard>
yes, I'm having a hard time too to get how to map those completely random pins supported in A10, A10s and A13, and how to map them usefuly to pinctrl
<mripard>
ah
<mripard>
the Pio IPs are different from A10 to A13 ?
<mnemoc>
no, they are the same
<mnemoc>
i have separated .c for the statics
<mnemoc>
because groups and features difer between socs... obviusly
<mripard>
but what about the aw1626 and aw1623
<mripard>
?
<mnemoc>
aw1623 is the so called A10
<mnemoc>
aw1626 is A13
<mripard>
aaah, I see
<mnemoc>
"chip-id" vs. marketing name
<mnemoc>
the chip-id lives inside the header of the BROM
<mnemoc>
I was doing it detection based in 3.4... until I realized modern arm-linux hates detection, and everything needs to be hardcoded in the .dts
mSquare has quit [Quit: Leaving.]
<xxiao>
hopefully u-boot etc will be able to generate dts automatically
<mnemoc>
eventually we will be able to turn an script.bin into dts
<xxiao>
actually for usb, pci or most probe-able bus it's still detected dynamically
<mnemoc>
but we need to model it manually first :p
<xxiao>
dts just assign interrupts and some basic resources for those buses
<xxiao>
but for misc bus/interfaces you have to hand-code them
SouL_ has joined #arm-netbook
SouL_ has left #arm-netbook [#arm-netbook]
cat_x301 has quit [Ping timeout: 246 seconds]
SouL_ has joined #arm-netbook
madmalkav has joined #arm-netbook
SouL_ has left #arm-netbook [#arm-netbook]
<slapin>
hno: ping
<slapin>
hno: hi!
Workboot has joined #arm-netbook
<L84Supper>
who was looking for the PCB breakout for the a10 bga?
<L84Supper>
then again if you really need help with the BGA breakout I highly doubt you'll be able to design a board that works well
<focus_well>
L84Supper: yum yum! that be me looking for those bga tracks!!!
<mnemoc>
i have zero intention in designing a board, ever
<mnemoc>
just want to be sure no sunxi related knowledge gets lost
<focus_well>
L84Supper: a couple of questions
<focus_well>
1. The inner BGA pads and some of the other bga pads have thick tracks.
<focus_well>
Will those tracks have issues with solder freezing?
<focus_well>
Is it better off to have min track width to the pads to prevent that?
<focus_well>
2. What is the bga pad size?
<focus_well>
3. What is the min track width used in the design between the bga pads?
<focus_well>
4. What is the hole size for vias used in the design between bga rows of
<focus_well>
pads?
<focus_well>
5. What is the overall via size used in the design for vias between the
<focus_well>
bga rows of pads?
<focus_well>
From some other conversation I remember
<focus_well>
the tracks the engineers used were 5 mils with 5 mil clearance.
<focus_well>
The vias are 16 mils.
<focus_well>
The drills were around 5 or 6 mils.
<focus_well>
The companies I know makes PCB with these limits
<focus_well>
min track width 0.075mm
<focus_well>
min hole size 0.15mm.
<L84Supper>
I'd suggest that you spend a few more years learning high speed PCB design
<L84Supper>
there is no way around it
<L84Supper>
or you end up with the usual junk
<focus_well>
L84Supper: I design lots of PCBs these days
<focus_well>
All working :)
<L84Supper>
then you should be able to answer this yourself :)
<focus_well>
The info request is specific to the A10 bga
<focus_well>
L84Supper: those diagrams were very good - thanks!
* slapin
looks at working board, and smiles, very, very evilly, and puts it into junk bin for parts
* slapin
wants to learn BGA PCB design a cheap way - any pointers?
<focus_well>
L84Supper: I have info here and there, but rather I use 100% certain data. But I have allow 2-3 redesigns to get it right - so I'm not in a hurry to get it right first time.
<L84Supper>
those pics are from a magic board, that just happens to work due to good luck
<focus_well>
The board i want is ideally torture tested and still working
<drachensun>
my last bga design had to be 4 mil clearance, there was no way to route it at 5 mils, the pads had something close to 12 mils between them, so 4 on each side and 4 for the track and that was it
<drachensun>
there are places that can do tighter design clearances, you just have to find them
<focus_well>
dranchensun: keep talking I pick up conversation in a few hours.. bye
<drachensun>
heh I think that was all I had to say about it. I ran into the same thing, I wanted to keep it at 5 mil because I can get cheap prototypes at that size
<drachensun>
but it just wasn't possible and that was an Atmel with a lot fewer pins than an A10
<slapin>
drachensun: any pointers on cheap BGA prototypes (6-8 layers)>
cat_x301 has joined #arm-netbook
<slapin>
?
<L84Supper>
sometimes you can use 5/5mil for most of the board and only go down to 4/4 or even 3 just between BGA pads
RITRedbeard_ has quit [Ping timeout: 245 seconds]
RITRedbeard_ has joined #arm-netbook
ZaEarl has joined #arm-netbook
Jef91 has joined #arm-netbook
Jef91 has quit [Client Quit]
Jef91 has joined #arm-netbook
<jquip>
is the cdc-acm driver a part of the kernel defconfig??
<mnemoc>
maybe ranting about a different input device?
<drachensun>
ok, I got a little mixed up, on my system with the mali drivers and my screwed up xorg conf from trying to get xrandr to rotate the screen, I have the same message and currently the touch doesn't work
<drachensun>
on my other system it works
<drachensun>
and the Xorg.0.log looks like you would expect "using input driver 'evdev' for 'ft5x_ts'
<techn>
drachensun: thanks.. now to track why that doesn't happen :p
<drachensun>
give me a sec and I'll post what logs and conf files I can from the working system
<drachensun>
Ok, the only difference in the conf files seems to be that 99-mali400.conf from the mali driver and my additions to it
<libv>
CaCtus491: if you get the ones without wifi, you can use that usb connection to connect a 3g module
tuliom has joined #arm-netbook
<libv>
CaCtus491: feel free to contact olimex or tom cubie to find out how much they charge for adapting their designs to your needs and for providing the limited run you need
<focus_it>
CaCtus491: olimex IRC channel is #olimex on freenode
<libv>
but i personally consider them quite on-topic in here too :)
<libv>
in as far as linux-sunxi is on-topic in the eoma channel ;)
alcides has joined #arm-netbook
alcides has joined #arm-netbook
alcides has quit [Changing host]
<libv>
olinuxino has two uarts exposed directly
<libv>
olimex will not be adverse to designing a 3g submodule if the revenue is there
<CaCtus491>
excellent
specing has joined #arm-netbook
<CaCtus491>
the A13 OLinuXino looks pretty close to what I need
<CaCtus491>
...the 3G stuff will need to be 'integrated' in that I need to have the solution in a single extruded aluminium case
<CaCtus491>
hmm, Olimex...
<CaCtus491>
interesting, this project is to replace an older LPC2148 based design from a few years back
<CaCtus491>
I seem to remember purchasing some LPC2148 stuff from them :)
<libv>
is a cortex a8 really what you need for your project?
<CaCtus491>
heh, yeah, new revision will have many new additional features that simply aren't possible witht the existing hardware, including a LCD for UI