<wens>
ijc: what about the fdt for driver model in u-boot? seems like it has to be compiled in?
afaerber_ is now known as afaerber
petr has quit [Ping timeout: 272 seconds]
<Turl>
wens: AW is supposedly sending me one
<Turl>
wens: they gave me a EMS tracking number which according to the track log left china 3 times (!)
<Turl>
so I dunno what's going on :)
<afaerber>
wens, any suggestion what files to look at for adding USB support to a80 dtsi?
petr has joined #linux-sunxi
<wens>
afaerber: for ehci/ohci, just dig through drivers/usb/sunxi_usb/{usbc,manager} i think
<wens>
to get the addresses, and phy stuff
<wens>
i looked at it a bit, and seems the same as on older socs
<wens>
but we still need gpio to toggle vbus regulators
<wens>
mripard: (wink)
<wens>
Turl: that's funny...
<Turl>
wens: are there any known post issues on there?
<Turl>
supposedly EMS is fast but not on china clearly :p
<wens>
i not aware of anyone using ems in china :|
ricardocrudo has joined #linux-sunxi
ricardocrudo has quit [Remote host closed the connection]
bertrik has quit [Remote host closed the connection]
gzamboni has quit [Read error: No route to host]
TheSeven has quit [Ping timeout: 272 seconds]
TheSeven has joined #linux-sunxi
<afaerber>
arch/arm/mach-sunxi/include/mach/sun9i/platform-sun9iw1p1.h turned out helpful for some addresses
<afaerber>
the usb code was calling into binary-only libusb300 to obtain otg base :/
ganbold_ has quit [Ping timeout: 255 seconds]
ganbold_ has joined #linux-sunxi
xeros has quit [Ping timeout: 260 seconds]
castor3 has quit [Ping timeout: 245 seconds]
castor3 has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Svetlana has joined #linux-sunxi
Svetlana has quit [Remote host closed the connection]
Svetlana has joined #linux-sunxi
leviathanch2 has joined #linux-sunxi
Svetlana has left #linux-sunxi ["до свидания"]
gzamboni has joined #linux-sunxi
bengal has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 260 seconds]
_massi has joined #linux-sunxi
Zboonet has quit [Remote host closed the connection]
Zboonet has joined #linux-sunxi
Andy-D has joined #linux-sunxi
leviathanch2 has joined #linux-sunxi
esperegu has joined #linux-sunxi
esperegu has quit [Ping timeout: 250 seconds]
<ijc>
wens: TBH I'm not completely sure, but yes it does look like it is mostly a compile time thing, at least right now. That still has benefits from a maintenance PoV and opens the door to using a separate dtb in the future.
<ijc>
At least for the sunxi patches it appears to be using the same DT as Linux -- some of the existing DM users in u-boot seem to have their own DT with their own bindings...
leviathanch2 has quit [Ping timeout: 246 seconds]
leviathanch2 has joined #linux-sunxi
deasy has joined #linux-sunxi
konradoo77 has joined #linux-sunxi
techn has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 244 seconds]
konradoo77 has quit [Ping timeout: 255 seconds]
leviathanch2 has joined #linux-sunxi
wenbin has quit [Ping timeout: 246 seconds]
tinti has quit [Read error: Connection reset by peer]
tinti has joined #linux-sunxi
<hramrach_>
slapin: well, form wayland is also missing features I want. or was last time I looked. And so X11. So if any display system gets fixed up I migrate to it and don't really care which one it will be
<hramrach_>
technically I do not *need* any features. I could use computers with DOS and direct hardware access and it sufficed to get some work done, too
<hramrach_>
when desperate you could just get >1 computer
<hramrach_>
for concurrent rendering in multiple apps or whatever
<hramrach_>
not that it's not rtevisited again with all that virtualization and virtual appliances
<hramrach_>
which means that in practace current display managers are no better than DOS :/
melonipoika has quit [Read error: Connection reset by peer]
<hramrach_>
of course, I could myself fix much of software problems if somebody paid me to work on making a decent operating system rather than fixing their problems with the broken-by-design operating systems ;-)
<hramrach_>
ssvb, slapin, ijc : anyway, how would you expect these exchangeable SD cards to work?
<hramrach_>
when you boot from nand the SPL is loaded from reserved, braindead formatted part of the nand. This SPL needs to contain dram parameters or read them from easily reachable location because it has not much memory space to work with until dram is initialized
<hramrach_>
if you boot from SD the SPL is loaded from a reserved area in the SD and currently contains its own copy of the dram parameters
<hramrach_>
if you were to build a SD card bootable in any device this SPL either needs support for both mmc to read u-boot from mmc and nand (whichever part the parameters are stored) to read dram parameters
<hramrach_>
or you could just load u-boot from nand ad pass it some flag that it has been loaded from mc and should mount root from mmc but read everything including kernel(+dtb) from nand
<mripard>
not only the DRAM parameters, but also the muxing, and which MMC/NAND controller to retrieve data
<hramrach_>
once you have u-boot running in dram it can presumably read a script.bin or dtb from anywhere
<mripard>
So I really can't see a completely generic image, without a device specific SPL, with hardcoded fragments or an embedded DTB
<hramrach_>
the boot media must be in the same place on all device or eGON would not be able to boot from there, anyway
<hramrach_>
so the only device that needs to be configured differently before a dtb is loaded is dram
<ijc>
hramrach_: ssvb has a plan (and even patches) for a generic SPL/DRAM setup thing with some common baseline of stuff autodetected. I'm not sure how the whole story fits together though, you'd have to ask him...
<hramrach_>
you could technically initialize dram at low speed/bus width, load some more code into dram, search for device-specific dram parameters, restart dram (trash content)
<hramrach_>
then you could store dram parameters in /boot and opt for loading those from nand
<libv>
hramrach_: ijc is right, and you should know about this
<hramrach_>
you could like standardize on separate /boot which is already supported in most distros and just mount it from nand on this unuversal SD card
bertrik has joined #linux-sunxi
FunkyPenguin has quit [Ping timeout: 245 seconds]
FunkyPenguin has joined #linux-sunxi
F1skr has joined #linux-sunxi
kaspter has quit [Ping timeout: 260 seconds]
<deasy>
yup, someone know what is the memphis chip on olimex board ?
<deasy>
lol sorry i'm fool
<deasy>
it's the ram !
leviathanch2 has quit [Ping timeout: 245 seconds]
lauri has quit [Ping timeout: 260 seconds]
viccuad has joined #linux-sunxi
konradoo77 has joined #linux-sunxi
_massi has quit [Remote host closed the connection]
Zboonet has quit [Remote host closed the connection]
philippe_fouquet has joined #linux-sunxi
FunkyPenguin has quit [Ping timeout: 244 seconds]
FunkyPenguin has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 240 seconds]
konradoo77 has joined #linux-sunxi
Seppoz has joined #linux-sunxi
tinti has quit [Remote host closed the connection]
Seppoz has quit [Remote host closed the connection]
Quarx has joined #linux-sunxi
Black_Horseman has quit [Ping timeout: 246 seconds]
\\Mr_C\\ has joined #linux-sunxi
Andy-D has quit [Remote host closed the connection]
leviathanch2 has joined #linux-sunxi
philippe_fouquet has quit [Remote host closed the connection]
<hno>
hramrach_, libv, yes, and yes the "generic SPL" defaults to sub-optimal settings that should work on most boards. Far from all parameters can be autodetected.
<hno>
The intention is to use this for installers, which then switches to more proper settings after installation.
philippe_fouquet has joined #linux-sunxi
philippe_fouquet has quit [Remote host closed the connection]
castor3 has quit [Quit: WeeChat 0.4.3]
nove has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 250 seconds]
Zboonet has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 255 seconds]
Andy-D has joined #linux-sunxi
Quarx has quit [Remote host closed the connection]
popolon has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
melonipoika has joined #linux-sunxi
leviathanch2 has joined #linux-sunxi
konradoo77 has joined #linux-sunxi
paulk-aldrin has joined #linux-sunxi
<afaerber>
mripard, looks like you didn't push after applying wens' a80 series last night? :)
afaerber_ has joined #linux-sunxi
afaerber has quit [Ping timeout: 245 seconds]
philippe_fouquet has joined #linux-sunxi
hipboi_ has quit [Read error: Connection timed out]
hipboi_ has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 250 seconds]
konradoo77 has joined #linux-sunxi
astr has quit [Ping timeout: 250 seconds]
astr has joined #linux-sunxi
Quarx has joined #linux-sunxi
souther has quit [Remote host closed the connection]
souther has joined #linux-sunxi
philippe_fouquet has quit [Remote host closed the connection]
hlavery1 has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
\\Mr_C\\ has quit [Quit: .]
rz2k has joined #linux-sunxi
ricardocrudo has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 260 seconds]
jelly-home has quit [Excess Flood]
jelly-home has joined #linux-sunxi
Zboonet has quit [Remote host closed the connection]
viccuad has quit [Read error: Connection reset by peer]
melonipoika has quit [Remote host closed the connection]
Zboonet has joined #linux-sunxi
Andy-D has quit [Ping timeout: 272 seconds]
ninolein_ has quit [Ping timeout: 245 seconds]
ninolein has joined #linux-sunxi
Andy-D has joined #linux-sunxi
konradoo87 has joined #linux-sunxi
<ssvb>
hno: the dram parameters, which can't be autodetected are all anyway rather poorly configured by the vendors
leviathanch2 has quit [Ping timeout: 255 seconds]
konradoo77 has quit [Ping timeout: 260 seconds]
<ssvb>
hno: the only real concern is the VDD-DLL voltage, which is set to 1.2V when AXP209 (or other PMIC) is not explicitly initialized by u-boot, so we need to keep the dram clock speeds low for generic settings
Gerwin_J has joined #linux-sunxi
<ssvb>
hramrach_: I'll roll out a feature complete demo in a few days, stay tuned :-)
<ssvb>
paulk-aldrin: is it a FEL button or something else?
jmcneill has joined #linux-sunxi
<paulk-aldrin>
well, definitely a push button
<paulk-aldrin>
no idea what it does
<paulk-aldrin>
I'll figure it out shortly :)
<ssvb>
paulk-aldrin: ok, thanks
<jmcneill>
I have NetBSD running on the A31! But I still don't quite understand the boot process and it's bugging me.
<ssvb>
paulk-aldrin: also just out of curiosity, it says "Inet k100c: Same case, different board", but Yones_Toptech_BD1078 dimensions specify 9mm vs. 12mm for Inet_k100c
<paulk-aldrin>
ssvb, fair point
<hramrach_>
NetBSD on everything ;-)
<paulk-aldrin>
let me get a straight book cover and sort this out
<jmcneill>
Usually not until the part is discontinued :p
Akagi201_ has quit [Read error: Connection reset by peer]
Akagi201 has joined #linux-sunxi
<jmcneill>
ssvb: as far as the boot process, I haven't discovered much other than instructions on how to use the sdk tools to build a bootable card
<jmcneill>
ssvb: i'd really like to be able to build one without them though -- like we do for the A10 and A20 boards
<hramrach_>
would require to port u-boot to a31
<hramrach_>
if you have a31 sdk with u-boot you may have source of something that runs on a31
<jmcneill>
yeah the u-boot sources are in there, but it's.. different
<hramrach_>
yeah, that's the problem
<hramrach_>
also reading back dram controller settings so that you can write them back in your own SPL is probably not solved for anything past a20
<jmcneill>
yeah, i don't mind so much having to rely on the awin boot0 and boot1, as we don't distribute any of the bootloaders anyway
<ssvb>
jmcneill, hramrach_: yes, u-boot still needs proper spl for a31, with the code to initialize dram
<jmcneill>
but just to be able to say -- this is the process to make a bootable card
<ssvb>
wens and hansg tried to do something about it, but not sure if they made any progress
<hramrach_>
this is the process to make a bootable SD card .. in case your A31 device happens to boot with dram settings used in this SDK
<jmcneill>
one device at a time :)
konradoo87 has quit [Ping timeout: 246 seconds]
<ssvb>
I have an a31s tablet, which is a bit of more challenge than a devboard to do any development on it
<jmcneill>
I bet
<ssvb>
but I will try to tame it eventually
<jmcneill>
At least you (hopefully) get to start with kernel sources.
<hramrach_>
we do have sources but without the dram settings we are kind of stuck
<hramrach_>
same for a23
<jmcneill>
ah
<ssvb>
hramrach_: we do have enough sources for the dram code
<hramrach_>
ok, I can boot a kernel on a23 with some random SPL blob and it crashes during early init and I have no idea if it is due to kernel bug or wrong dram settings
<hramrach_>
yes, we supposedly have code that sets up the parameters but not the parameters
<jmcneill>
ssvb: I noticed you didn't use the word "readable" to describe that code.
<hramrach_>
it's fine, just all comments are in Chinese, usually in a mix of different encodings :p
<jmcneill>
I was thinking more -- lots of magic numbers
<ssvb>
jmcneill: the rockchip code is a bit more readable than allwinner's, and checking it should help a lot
<hramrach_>
and it sets up some registers that are completely undocumented
<ssvb>
hramrach_: true, a23 is newer than a31, and we have less information about it
<hramrach_>
does rk use similar memory controller?
* hramrach_
wonders where the Chinese buy this stuff
montjoie[home] has quit [Ping timeout: 272 seconds]
montjoie[home] has joined #linux-sunxi
<ssvb>
hramrach_: the names of many hardware registers are exactly the same, and their relative arrangement in the address space almost perfectly matches rk3288
<ssvb>
hramrach_, jmcneill: so the magic constants and bitfields can be deciphered
bertrik has quit [Ping timeout: 272 seconds]
<ssvb>
hramrach_: Synopsys DesignWare is the primary suspect as the vendor of the DRAM controller IP :-)
<jmcneill>
ssvb: i don't see in that rhombus-tech tree where dram_sun6i.c is used
<ssvb>
jmcneill: that's more like a starting point
<jmcneill>
understood
<ssvb>
it would be surely nice to have full A31 support in u-boot v2015.01
konradoo87 has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 255 seconds]
<ssvb>
libv: would "Tablets with HDMI port" category in the wiki make sense?
<ssvb>
"Devices with HDMI port" already exists, but development boards and TV boxes tend to have HDMI port anyway, while for tablets it is less common
<hramrach_>
maybe. It's not like I want that on AW tablets but somebody might
zeRez has joined #linux-sunxi
lerc has quit [Ping timeout: 250 seconds]
<ssvb>
libv, hramrach_: ok, I have added this new category, without editing too much just in case if somebody does not like it (so that we have less stuff to revert)
<ssvb>
overall, a few more categories may be useful to have: "DC plug", "USB host" and "real FEL button"
<ssvb>
but in fact I just want something like an aggregate category "this tablet may work as a desktop PC replacement" for the buying guide :-)
skaag has quit [Ping timeout: 244 seconds]
<jmcneill>
ah the u-boot-sunxi uboot actually works, great
<jmcneill>
i just had never tried it, was using the one from the sdk
wickwire has joined #linux-sunxi
<ssvb>
the 'next' branch from the former is on the way to be mainlined, so you may try to test it too
konradoo87 has quit [Ping timeout: 245 seconds]
<hramrach_>
ssvb: for 'this tablet may work as a desktop replacement' look away from AW
<hramrach_>
but if you were to use a tablet as desktop replacement you probably want a custom dock which will do the charging, USB connection, and HDMI connection
<hramrach_>
and for that 1 usb and 1 hdmi suffices
<hramrach_>
extra ports are nice to have but more importantly the tablets are just not powerful enough
skaag has joined #linux-sunxi
<hramrach_>
you can use the tablet as single purpose device - like upload som movies to it and play them back, one by one
<hramrach_>
or read some books on it, one by one
<hramrach_>
eg. not do research that involves opening multiple books, end even visiting web pages with (gasp) JavaScript
<jmcneill>
ssvb: I'll definitely be keeping an eye for any changes :)
<hramrach_>
looking at the u-boot sources the a23 have some dram definitions in standby only. did not find the SPL code. Either way, it seems to have dual phy and crapload of registers which are not even named but just numbered
<hramrach_>
and stfw for some random defines from the headers seems to give Indian pop songs and Vietnamese porn :s
<ssvb>
hramrach_: please add any references to any a23 u-boot or boot0 dram code to the wiki
<ssvb>
hramrach_: are the define names *that* much weird?
<ssvb>
hramrach_: searching for some magic constants may also help, they may be default reset values
jmcneill has quit [Remote host closed the connection]
bengal has quit [Quit: Leaving]
FDCX has quit [Ping timeout: 260 seconds]
rz2k has quit []
bonbons has quit [Quit: Leaving]
FDCX has joined #linux-sunxi
hipboi_ has quit [Ping timeout: 245 seconds]
konradoo77 has quit [Ping timeout: 260 seconds]
konradoo77 has joined #linux-sunxi
FDCX has quit [Ping timeout: 260 seconds]
hipboi_ has joined #linux-sunxi
<wickwire>
hi guys, I've been looking at overlaying two windows on X11, with the A20 SoC (on a Cubieboard)
<wickwire>
something along the line of alpha blending
<wickwire>
one window has video, the other has graphics transparency (to be on top)
<wickwire>
does anyone have any suggestions or more knowledge about current limitations for doing this?
FDCX has joined #linux-sunxi
froese has joined #linux-sunxi
FDCX has quit [Ping timeout: 260 seconds]
FDCX has joined #linux-sunxi
FDCX has quit [Ping timeout: 260 seconds]
konradoo77 has quit [Ping timeout: 258 seconds]
<hramrach_>
wickwire: current limitation is that it does not work
<hramrach_>
presumably you could make it work with correct layer ordering and/or using offscreen buffer and compositing
<hramrach_>
also depends on the video source. if you do everything in software everything will work ;-)
<hramrach_>
ssvb wrote the X11 driver so I guess he would know how to fix the driver and Cedar to render hardware accelerated video correctly
konradoo77 has joined #linux-sunxi
<wickwire>
hramrach_: hmmm ok... well, I suppose it's not a new idea, since folks with cubieboards seem to go after media center solutions
<wickwire>
and since we've come to a point where accelerated video works with libvdpau-sunxi
<wickwire>
and graphics work with Qt
FDCX has joined #linux-sunxi
<hramrach_>
also there is already OSD support in libva so if you can render whatever you wanted on top as OSD it will work
<wickwire>
I thought about combining the two
<wickwire>
in a more "custom" way, such as having the accelerated video in the back and the graphics on top
<hramrach_>
you can combine the two. it just needs some work
<hramrach_>
the media center solutions render in fullscreen and avoid this problem
<wickwire>
and do you know if there's someone already working on this, or if I'm going at it alone, where would be best to start...? the window manager would have a part in this...?
<hramrach_>
probably the X11 driver and libva would have part in this
bertrik has quit [Remote host closed the connection]
<hramrach_>
the window manager should not know .. unless you wnat to hack every window manager out there with a Cedar specific hack
<wickwire>
well, I halted a bit on the Qt5 front using QtMultimedia, trying to combine Qt and Gstreamer and libvdpau-sunxi
<hramrach_>
when you do you will have semi-working solution with video always on top
<wickwire>
as QtMultimedia seems to not like handling the libvdpau elements even though my Gstreamer 0.10 has libvdpau-sunxi working
<hramrach_>
to have proper layering somebody has to figure out how to coordinate layers between libvdpau and X11
<wickwire>
so I thought about just having the separate video process underlying the graphics layer...
<wickwire>
ok I see
<wickwire>
I'm not that savvy on the matter, would like to give it a shot though
<wickwire>
and basically I got to X11 because libvdpau-sunxi requires it as far as I know
konradoo77 has quit [Ping timeout: 272 seconds]
FDCX has quit [Ping timeout: 240 seconds]
<wickwire>
and if not for accelerated video,
<wickwire>
then Qt integrates normally with GStreamer and I can do it the standard way with QtMultimedia, although for SD videos and with some extra stress on the CPU of course
<wickwire>
ok, thanks for your help,
<wickwire>
I'll keep on reading
<wickwire>
see if I can find something to hold on to for this
<hramrach_>
presumably what libvdpau uses as its fixed output could be adapted as XV driver and then X11 would know it has this layer and could stack it correctly
<hramrach_>
but I haven't looked at the code because fullscreen video is good enough, especially on 800x480 screen
<hramrach_>
my current grief with video decoding is that 10bit h264 cannot be decoded. It is possible to remove the extra bits and somebody wrote a proof of concept decoder using this on top of AMD video decode block but it uses so obscure interface I could not even find a media player to test it
pirea has joined #linux-sunxi
arokux2 has quit [Remote host closed the connection]