<afaerber>
is anyone else working on the U-Boot side of A80 already?
<philippe_fouquet>
I try to use the last sunxi-3.4 kernel with android and I have a error on gralloc
<philippe_fouquet>
It is not vesion compatile
<afaerber>
wens, you aware of this? [ 1.261496] sun9i-usb-phy a02800.phy: failed to create PHY
<mripard>
afaerber: I'm not aware of anyone working on u-boot for the A80
<mripard>
afaerber: but it's good that you're working on this :)
<afaerber>
mripard, well, I don't know U-Boot sources much, so this is poking in the dark based on A23 patches... some guidance e.g. whether we can reuse the sun6i clk code would be helpful
<mripard>
ijc: ^ :)
<philippe_fouquet>
I find the source for gralloc but not for libEGL_mali.so libGLESv1_CM_mali.so libGLESv2_mali.so libMali.so
<philippe_fouquet>
were Can I find it
<afaerber>
philippe_fouquet, they are not public
<afaerber>
that's why there's the lima project ... unfortunately a bit stalled at the moment
<wens>
afaerber: hmm... i just rebased a bunch of stuff, so it might not be working :/
<philippe_fouquet>
afaerber: they no have comatible version for sunxi-3.4.104
<wens>
afaerber: it's unlikely you can reuse any code in u-boot for a80
<wens>
all the offsets are different
<afaerber>
wens, ok
<wens>
on the other hand, most of the basic stuff that would be needed in u-boot is already in mainline :)
<wens>
mainline kernel that is
<wens>
should be a good reference
<afaerber>
wens, what are you working on next? my next point of interest would be the gmac
<wens>
gmac seems to need some unknown regulator (some output of axp), which requires rsb, which requires prcm and r_pio
<mripard>
RSB would be nice to tackle
<afaerber>
is pcrm different, too?
<afaerber>
*prcm
<wens>
low bits of prcm are the same
<wens>
high bits are unknown, see my prcm branch :)
<wens>
or the patches i posted
<wens>
mripard: i have some notes on RSB, just haven't gotten around to implementing them :|
<mripard>
maybe we can just use the A23 to do the RSB bringup ?
<wens>
that's the plan i guess
<mripard>
they can't be possibly be changing the pmic bus every new SoC
FDCX has joined #linux-sunxi
akaizen has joined #linux-sunxi
traeak has joined #linux-sunxi
traeak has left #linux-sunxi [#linux-sunxi]
<hramrach_>
ricardocrudo: as I understand it if you manage to build a livesuit image for your device you can use phoenixcard tool to turn in into a SD card image
<hramrach_>
huh, wrong scroll position
crudo has joined #linux-sunxi
crudo is now known as Guest98746
Guest98746 has quit [Client Quit]
<hramrach_>
anyway, that SD card image you write to a SD card and when the boards boots it it flashes the internal nand
<wens>
mripard: i hope they dont
ricardocrudo has quit [Remote host closed the connection]
ricardocrudo has joined #linux-sunxi
akaizen has quit [Ping timeout: 256 seconds]
<mripard>
wens: bbrezillon spent a lot of time to have this p2wi driver in, and it's close to useless now :/
<wens>
:(
<bbrezillon>
mripard: what ?
<wens>
well i can use it as a sample to work with, so all is not lost
<mripard>
bbrezillon: well, p2wi is not used and probably won't be used aside from the A31
<mripard>
since they changed for RSB in the A23 and A80
<bbrezillon>
mripard: yes, but it's mandatory on A31 SoCs/borads
<bbrezillon>
so it's not completely useless ;-)
<mripard>
bbrezillon: are we even using it these days ? :)
<wens>
we aren't, because the axp221 driver isn't in yet
<hramrach_>
RSB? oh that weird bus
paulk-collins has joined #linux-sunxi
<RzR>
hi
<RzR>
I managed to get A80 Optimus board
<hramrach_>
nice
<RzR>
is anyone interested about hacking on this platform ?
<RzR>
yea but I am unsure I'll have dedicated time for it
<RzR>
it's shipped w/ android
<hramrach_>
yes, pretty much everyone other than me around atm
<bbrezillon>
wens, mripard: okay, my bad, I gave up on this driver :-(
<hramrach_>
so when you have time stop by and maybe there will be something interesting to try
<mripard>
bbrezillon: it's ok. Just push it to github and we'll deal with it :)
<hramrach_>
I personally find a80 a mix of features and misfeatures which is not favourable at the price point
<wens>
bbrezillon: i'd say focus on nand, axp is easier to pick up :)
<mripard>
wens: +1 :)
ZZambia has joined #linux-sunxi
akaizen has joined #linux-sunxi
<hramrach_>
how much is known about the nand controller sofar? does it seem similar to the previous sunxi stuff?
<ZZambia>
hi there
<hramrach_>
ehlo
<ZZambia>
anybody working on OB A80?
<mripard>
hramrach_: the NAND controller of the A80 ?
<hramrach_>
of the unsupported chips in general
<wens>
did anyone actually get an optimus board with nand instead of emmc?
<hramrach_>
heh, good point
<hramrach_>
maybe will need a tablet for that
<hramrach_>
but my a23 tablet has nand
<wens>
the pcduino3 seems to ship nand
<mripard>
and for the A31/A23/A33, I don't think anyone looked into it yet
<hramrach_>
hehe, optimus shipping with emmc neatly solves the problem
<wens>
afaerber: i don't see "sun9i-usb-phy a02800.phy: failed to create PHY"
<ZZambia>
anybody got news about GPU SDK for Linux on A80?
<ZZambia>
now testing the CT4-derived lubuntu image but it looks not supporting any accelerated graphics
<RzR>
ZZambia: is this a SDcard image ?
<hramrach_>
well, from what I heard the GPU is so crappy the even with SDK it won't work
<hramrach_>
well, I am not sure this is an issue with pvr or this particular version of pvr
<wens>
mripard: you think we should do smp? or just wait for psci
<hramrach_>
but the SDK gives you a lib which exposes a GL interface
<ZZambia>
hramrach_: it should also offer opencl too
<hramrach_>
but to use the GPU successfully you need to avoid doing certain things
<ZZambia>
hramrach_: at least standing on imagination's specs
<hramrach_>
like call some calls in particular order which is valid GL but will crash the HW or SW
<ZZambia>
hramrach_: where did ou got news about SDK?
<ZZambia>
holy crap
<hramrach_>
on Mali you can crash the SW lib but at least the HW is solid
<ZZambia>
Mali is much more straightforward than PVR
<hramrach_>
large part of CPU time spent Radeon drivers is for checking that the commands you sent will not crash the HW
<hramrach_>
sometimes it crashes regardless
<hramrach_>
I don't suppose that will be different for pvr
<hramrach_>
ZZambia: it's not like I have seen a SDK but that's what a SDK does
<mripard>
wens: I have no idea how big little chips are usually supported
<hramrach_>
it gives you a lib that emulates some standard API on top of the hardware. With some level of success
<mripard>
but I was thinking it could be one of the tasks of babelfish to implement psci
<wens>
mripard: there are 2 implementations under arch/arm iirc
<ZZambia>
hramrach_: but is there any idea of binary blobs somewhere?
<mripard>
wens: exynos and what ? :)
<hramrach_>
pvr does not give non-binary SDKs so far
<hramrach_>
s/pvr/imation
<wens>
mripard: i remember seeing 2, maybe the other was the generic stuff :|
<mripard>
wens: what I'd like to see is: 1) having the mainline kernel at least booting on Allwinner's bootloader. In a degraded mode, so without SMP, without virtualization, but still at least boot. 2) having babelfish as a fallback solution to boot a kernel with PSCI, in HYP, while u-boot is catching up (A31, A23, A80), 3) having u-boot that does all that eventualy for the older platforms (A10 -> A20)
<ZZambia>
now grepping the fs, keep you posted
<mripard>
that would need some major effort on babelfish, but I think it could be done
<wens>
arch_timer on sun7i is a major blocker
<mripard>
yeah, and it can be solved.
<mripard>
that's the only thing that prevent the A20 to boot on allwinner's kernel iirc
<wens>
fixing arch_timer to fail when it should?
philippe_fouquet has quit [Remote host closed the connection]
<mripard>
that, or to set the frequency in the DT so that it won't crash and actually work
viccuad has joined #linux-sunxi
ZZambia has quit [Quit: leaving]
_massi has joined #linux-sunxi
<hramrach_>
do we even need bootability on aw bootloader on a20?
<mripard>
hramrach_: it's also about rather old u-boot-sunxi build
<mripard>
and a silent kernel crash is never fun to debug.
<hramrach_>
that is not
<mripard>
especially for newcomers, it's not a really nice experience
<hramrach_>
but it does not need to boot. it can just tell the user to go away. like the 64bit kernel on 32bit machines
<mripard>
I'm not saying that we have to be feature complete. Like I said, you wouldn't have SMP, simplefb (it it ever gets merged), virtualization, etc.
<mripard>
well, the thing is that it can boot very easily, so why not supporting it?
<mripard>
if it was a pain, sure
<mripard>
but it's not.
<hramrach_>
ok, it's fine then
<hramrach_>
is mtd working or what is missing?
ZZambia has joined #linux-sunxi
<hramrach_>
it still says WIP on the mainlining page but not what is missing
<hramrach_>
ok, I try the patches on cubieboard I guess. there nand is not readable anymore, anyway
paulk-collins has quit [Ping timeout: 265 seconds]