rellla changed the topic of #linux-sunxi to: Allwinner/sunxi /development discussion - did you try looking at our wiki? - Don't ask to ask. Just ask and wait! - - Logs at - *only registered users can talk*
<catphish> i'm trying to understand the boot0 process, but failing at the first hurdle, there seems to be a header that should be present to tell the boot rom how to load it, but the format of my uboot doesn't seem to match the docs :(
<catphish> specifically, i got as far as "header size" and it seems to contain the word "SPL" instead
<buZz> yeah i think 'SPL' is the common name for that blob
<buZz> i dont think anyone reversed its operation ever? not sure though
<catphish> but the docs say there should be "header size" there
<catphish> unless i'm mis-reading
<catphish> docs here:
<catphish> right
<catphish> but this doesn't match my u-boot, which i was expecting it to
<catphish> u-boot willfully ignores the spec :)
NeuroScr has quit [Quit: NeuroScr]
catphish has quit [Quit: Leaving]
sasrubix has quit [Quit: Leaving]
megi has quit [Ping timeout: 250 seconds]
kaspter has quit [Ping timeout: 272 seconds]
victhor has quit [Remote host closed the connection]
NeuroScr has joined #linux-sunxi
kaspter has joined #linux-sunxi
<wens> BenG83: the bus gate is always connected to the bus :)
<wens> BenG83: it would mean that the THS also takes an extra "sampling" or module clock
<wens> BenG83: the bus gate is simply used to access the registers, not for the module to run
gumblex has joined #linux-sunxi
gumblex_ has quit [Ping timeout: 250 seconds]
Andy-D has quit [Ping timeout: 246 seconds]
dddddd has quit [Remote host closed the connection]
valkyr1e has quit [Quit: Bye.]
valkyr1e has joined #linux-sunxi
leviathanch has joined #linux-sunxi
paulliu has joined #linux-sunxi
sasrubix has joined #linux-sunxi
sasrubix has quit [Quit: Leaving]
sasrubix has joined #linux-sunxi
IgorPec has joined #linux-sunxi
woodjrx_ has joined #linux-sunxi
woodjrx_ has quit [Remote host closed the connection]
<sasrubix> Hi people. I'm setting up to compile a new kernel 3.4.x. Will it literally take days to compile on the A10 machine? Maybe I should do it on the PC..
<KotCzarny> wrong channel :P
<KotCzarny> also, why do you need 3.4?
<sasrubix> wrong channel? Hmm. As for 3.4, I wanted to first build a kernel very similar to the one it came with.
<KotCzarny> this channel is about mainlining sunxi
<\x> >3.4.x
<\x> >new
<\x> wow
<KotCzarny> anyway
<KotCzarny> check out the wiki
<KotCzarny> it provides lots of historical hints too
<sasrubix> the wiki is what sent me here.. as such I figured this channel would accomodate some questions regarding the A10 board I'm adding to the wiki..
<sasrubix> (er, not questions specific to the board, but to my adding it to the wiki.)
<sasrubix> anyhow, thank you. I'll figure it out. :)
merlin1991 has quit [Remote host closed the connection]
kelvan has quit [Read error: Connection reset by peer]
<[TheBug]> sasrubix: It will take a while directly on the A10, if you have another machine to cross-compile on it will be much faster, the wiki includes hints as to how to cross-compile if you need
merlin1991 has joined #linux-sunxi
kelvan has joined #linux-sunxi
<[TheBug]> sasrubix: if you do have some specific question or problem and mention it, someone may help but mostly the goal is working on mainline here -- dont let people scare you off though ;p
<sasrubix> TheBug - thank you for the hints. Nope, I don't need help with the cross-compile; was merely wondering if it would take forever on the A10 :)
<[TheBug]> heheh was playing with my Cubieboard A10 this weekend and was reminded of its sheer speed, lol
<[TheBug]> was using it to test hard drives, even SATA is a bit slower than I remember
<[TheBug]> hehe
<sasrubix> I do have another machine, much faster. yes, the speed is overwhelming ;)
<sasrubix> fwiw I'm in this IRC from my A10 desktop. Works great in general, for a little tiny thing with one core.
<[TheBug]> sasrubix: if you ever feel masochistic enough I made several Android images for the A10 that are on the Wiki lol
<sasrubix> Nice! I shall check them out (probably already seen the iages somewhere there, but there are SO MANY pages Im trying to osmosify into my brain... I have about 13 years in Linux, but this is my first xperience on an ARM machine. Lots to research.
f0xx has joined #linux-sunxi
<[TheBug]> sasrubix: listed at bottom of
<[TheBug]> sasrubix: ahh yeah, A10 is where I satrted, then moved to A20, then H3 :D just a warning, it gets addictive ;p
<[TheBug]> sasrubix: at this point I have > 10 H3 devices, 2x A20, 2x A10, RPi sitting next to me lol
<[TheBug]> not to mention various others
<sasrubix> LOL. There's a Pi here too, being worked on near me. It's perhaps what first interested me in the little tiny computers.
<[TheBug]> :D
<wens> [TheBug]: what do you use them for
AneoX_ has joined #linux-sunxi
<sasrubix> [TheBug]: Upon first boot fresh out of the box, it hung and my NAND got corrupted, not sure how; so I'm running it from the SDcard ever since. I plan to fix the NAND if for no reason than to see if it's faster than running from SDcard..
<[TheBug]> various things
<[TheBug]> lol
<[TheBug]> lost for dev of H3Droid (H3 devices)
<[TheBug]> 1 RPi runs Kodi off in the background, Have a n A20 running as a NAS, use A10s mostly for hard drive review for new raid drives before they go in the raid
<[TheBug]> others have had many various uses over time
<[TheBug]> hehe
AneoX has quit [Ping timeout: 246 seconds]
<[TheBug]> even have an Vim 1 (Amlogic) in the mix
<[TheBug]> sasrubix: ya NAND is a nice bump up on those boards, I have one Cubieboard A10 with NAND and its nice :D
<[TheBug]> for a long time I ran both a NAS and an Android connected to old CRT for video on A10
<[TheBug]> in the garage
<[TheBug]> was decent enough for older videos
<[TheBug]> thus the Android images for A10 that I created
<[TheBug]> s/lost/lots *
<sasrubix> Yes I just went there, noted they are Android images. I plan to keep actual Linux on this thing though :)
<[TheBug]> hehe
AneoX has joined #linux-sunxi
<sasrubix> But you've inspired me to move the NAND fix up the list of hings to do.
<sasrubix> *things
<[TheBug]> well its what inspired H3Droid on H3devices and me harassing KotCzarny till he agreed to help me ;p
<sasrubix> :)
AneoX_ has quit [Ping timeout: 246 seconds]
<sasrubix> Plus p, if I create a defconfig and DTB for this (eventually) they may end up as candidates for mainline kernel? Even if there are nearly none of this board anywhere, the chipset is everywhere.
<[TheBug]> Actually there were a lot of A10 based boards made, just not as popular now with faster boards available
<[TheBug]> Olimex, Cubieboard, PcDuino few others
<sasrubix> Yes, I've read anecdotally there are "hundreds" of different boards.
kaspter has quit [Ping timeout: 272 seconds]
<[TheBug]> sasrubix: well, I am off to do some other things for now, have fun and a good day! :D
<sasrubix> Cheers, & thanks !!
selfbg has joined #linux-sunxi
reinforce has joined #linux-sunxi
kaspter has joined #linux-sunxi
nuuuciano__ has quit [Ping timeout: 245 seconds]
valkyr1e has quit [Quit: Bye.]
airwind has joined #linux-sunxi
valkyr1e has joined #linux-sunxi
valkyr1e has quit [Client Quit]
valkyr1e has joined #linux-sunxi
leviathanch has quit [Remote host closed the connection]
<KotCzarny> sasrubix: yeah, that's what this channel is about, just keep in mind 3.4.x is treated as best forgotten legacy
<KotCzarny> only useful for android and/or getting things from nand before reformatting it
<KotCzarny> because quite a lot of things already work with mainline
valkyr1e has quit [Quit: Bye.]
valkyr1e has joined #linux-sunxi
f0xx has quit [Ping timeout: 244 seconds]
<wens> most new devices come with emmc anyway
<KotCzarny> wens, are there any plans to support legacy nand format or it's definitely put into WONTFIX ?
<KotCzarny> it could be nice to have ability to run recovery kernel from sdcard
f0xx has joined #linux-sunxi
<KotCzarny> nand page on wiki talks about work already being submitted for reviews, but not updated since then
valkyr1e has quit [Quit: Bye.]
valkyr1e has joined #linux-sunxi
<wens> IIRC legacy nand format has proprietary FTL implementation
<wens> on Linux your only option is to use UBIFS, which doesn't support MLC flash
<wens> so it's basically dead ATM
<KotCzarny> by prorpietary you mean closed and needing to be REd ?
<KotCzarny> or there is a way to patch mainline to be able to read legacy nand
<wens> closed
<wens> and linux won't accept FTLs either
<KotCzarny> shucks.
valkyr1e has quit [Quit: Bye.]
tllim has quit [Read error: Connection reset by peer]
valkyr1e has joined #linux-sunxi
paulk-leonov has joined #linux-sunxi
NeuroScr has quit [Quit: NeuroScr]
yann has quit [Ping timeout: 240 seconds]
dddddd has joined #linux-sunxi
IgorPec has quit [Quit: Nettalk6 -]
nixdork has quit [Ping timeout: 252 seconds]
<Net147> what's the benefit of the pll7 fractional mode for 270 MHz and 297 MHz over integer mode on A20?
<wens> unknown lol
msimpson has joined #linux-sunxi
<Net147> I notice legacy would prefer to use fractional mode while mainline uses integer mode for those frequencies
<KotCzarny> maybe fractional allows higher parent clock for other system?
nixdork has joined #linux-sunxi
<Net147> I am testing 3840x2160 HDMI output on A20 and managed to get it to work at 15 Hz but wondering what is stopping it from working at higher refresh rates such as 24 or 30 Hz.
<Net147> any suggestions?
<wens> the UM says it only does HDMI 1.3 up to 1080p60
clemens3 has joined #linux-sunxi
<Net147> the datasheet says HDMI 1.4 transmitter with HDCP
<wens> conflicting docs :/
victhor has joined #linux-sunxi
yann has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
NeuroScr has joined #linux-sunxi
<\x> Net147: reduced blanking mode.
<Net147> \x: which one?
megi has joined #linux-sunxi
<Net147> wens: what would it take to get the frontend running on A20?
<Net147> wens: I was thinking sun4i_frontend.c was just missing a compatible for it but I can't see anything in drm.debug that indicates the frontend is actually working
<wens> paulk-leonov: ^
<paulk-leonov> Net147, IIRC there is a debug message with KMS or DRIVER debug
<paulk-leonov> it might be called "video layer"
<paulk-leonov> Net147, also the compatible was added for the A20 in the next drm-misc
<paulk-leonov> with the required quirks
<Net147> paulk-leonov: okay, thanks. I will check it out.
IgorPec has joined #linux-sunxi
clemens3 has quit [Ping timeout: 246 seconds]
<Net147> paulk-leonov: is the frontend always used after those changes instead of going directly to backend? or is there some conditions for it to be used?
NeuroScr has quit [Quit: NeuroScr]
clemens3_ has joined #linux-sunxi
<paulk-leonov> Net147, ah wait, the patches are actually still staging for A20 support and YUV support
<paulk-leonov> Net147, either way, it's only used when the backend can't handle the requested setup
<Net147> paulk-leonov: the reason I am asking is because in it indicates that using the hardware scaler can solve issues with memory bandwidth starvation when using high resolution and high refresh rates
<Net147> paulk-leonov: even when the scaling is 1:1
<paulk-leonov> Net147, I see, well in mainline the scaler is simply bypassed if there is no scaling involved (and no need for the frontend otherwise)
<\x> i dont know if itll help
<\x> but i use it on my desktop to oc monitors
<Net147> paulk-leonov: to force to use scaler even for 1:1 I would need to modify sun4i_backend_plane_uses_scaler?
<paulk-leonov> Net147, yes
Andy-D has joined #linux-sunxi
dddddd has quit [Remote host closed the connection]
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
netlynx has joined #linux-sunxi
<wens> Net147: IIRC the memory bandwidth limit was only on the A10
NeuroScr has joined #linux-sunxi
popolon has joined #linux-sunxi
leviathanch has joined #linux-sunxi
BenG83 has quit [Ping timeout: 250 seconds]
NeuroScr has quit [Quit: NeuroScr]
<Net147> wens: it says it occurs on A20 in X with GLES accelerated Mali applications too
The_Loko has joined #linux-sunxi
<wens> hmm...
fkluknav has quit [Remote host closed the connection]
fkluknav has joined #linux-sunxi
AneoX has quit [Ping timeout: 246 seconds]
AneoX has joined #linux-sunxi
BenG83 has joined #linux-sunxi
catphish has joined #linux-sunxi
<catphish> morning
BenG83 has quit [Remote host closed the connection]
dizzuhen has joined #linux-sunxi
dizzuhen has quit [Client Quit]
NeuroScr has joined #linux-sunxi
matthias_bgg has quit [Ping timeout: 240 seconds]
f0xx has quit [Ping timeout: 246 seconds]
BenG83 has joined #linux-sunxi
f0xx has joined #linux-sunxi
BenG83 has quit [Quit: Leaving]
NeuroScr has quit [Quit: NeuroScr]
tilpner has quit [Disconnected by services]
tilpner has joined #linux-sunxi
tilpner has quit [Remote host closed the connection]
chomwitt has joined #linux-sunxi
airwind has quit [Quit: airwind]
Gerwin_J has joined #linux-sunxi
Putti has joined #linux-sunxi
Gerwin_J has quit [Ping timeout: 264 seconds]
NeuroScr has joined #linux-sunxi
Andy-D has quit [Ping timeout: 246 seconds]
Andy-D has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
<plaes> any idea about this u-boot issue: Error binding driver 'sunxi-musb': -96
<plaes> latest u-boot for Lime2-eMMC board
Asara has quit [Ping timeout: 244 seconds]
JohnDoe_71Rus has joined #linux-sunxi
<plaes> could it be something regarding with the DM stuff that is going on?
<megi> it might be something with the last DTS Linux->u-boot sync? I had some issues with that. But not with usb, but I don't use usb in u-boot.
dddddd has joined #linux-sunxi
<megi> I had to add some new compatibles to u-boot driver, that didn't know about compatible strings used in Linux DTS files
<plaes> A20 board here
<megi> but that was PWM driver
NeuroScr has quit [Quit: NeuroScr]
\\Mr-C\\ has joined #linux-sunxi
matthias_bgg has quit [Ping timeout: 250 seconds]
\\Mr_C\\ has quit [Ping timeout: 268 seconds]
nuuuciano__ has joined #linux-sunxi
selfbg has quit [Quit: selfbg]
reinforce has quit [Quit: Leaving.]
sunshavi has quit [Remote host closed the connection]
matthias_bgg has joined #linux-sunxi
sunshavi has joined #linux-sunxi
msimpson has quit [Quit: Leaving]
\\Mr-C\\ has quit [Quit: (Read error: Connection reset by beer)]
AneoX_ has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
AneoX has quit [Ping timeout: 250 seconds]
reinforce has joined #linux-sunxi
leviathanch has quit [Remote host closed the connection]
qeed has quit [Quit: Leaving]
<megi> there's a small quirk in the sunxi-drm, where when using overlay planes + zpos, sometimes (rarely) the primary plane flashes over everything else for like one frame during atomic modeset
<megi> probably some ordering issue with register updates, I guess
<megi> coinciding with page flip or whatever
<megi> anyway, v4l2+dmabuf+drm is awesome. I get 60fps from camera scaled to fullscreen display with 0.7% cpu load on A83T tablet :)
m3Lith1 has joined #linux-sunxi
tllim has joined #linux-sunxi
m3Lith1 has quit [Remote host closed the connection]
<megi> mainline CSI driver doesn't work, though
<plaes> o_O
<megi> it just returns empty buffers
tilpner has joined #linux-sunxi
NeuroScr has joined #linux-sunxi
<megi> I have to use my old CSI driver for capture to work
<megi> I suspect buffer management, which seems to be the only relevant difference between the two drivers
<megi> mainline driver tries to be too smart not to lose the first frame
lurchi__ has joined #linux-sunxi
aalm has quit [Ping timeout: 240 seconds]
lurchi__ has quit [Ping timeout: 252 seconds]
BenG83 has joined #linux-sunxi
chomwitt has quit [Ping timeout: 268 seconds]
lurchi__ has joined #linux-sunxi
NeuroScr has quit [Quit: NeuroScr]
\\Mr_C\\ has quit [Quit: (Read error: Connection reset by beer)]
iamfrankenstein has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
qeed has joined #linux-sunxi
<BenG83> is anyone coming to 35C3 / CCC this year?
lurchi__ has quit [Ping timeout: 252 seconds]
yann has quit [Ping timeout: 240 seconds]
kaspter has quit [Read error: Connection reset by peer]
kaspter has joined #linux-sunxi
catphish has quit [Quit: Leaving]
anarsoul|2 has joined #linux-sunxi
scelestic has quit [Quit: leaving]
scelestic has joined #linux-sunxi
sasrubix has quit [Ping timeout: 250 seconds]
Andy-D has quit [Ping timeout: 246 seconds]
Andy-D has joined #linux-sunxi
random_yanek has quit [Ping timeout: 250 seconds]
\\Mr-C\\ has joined #linux-sunxi
\\Mr_C\\ has quit [Ping timeout: 245 seconds]
catphish has joined #linux-sunxi
random_yanek has joined #linux-sunxi
aalm has joined #linux-sunxi
hanni76 has joined #linux-sunxi
matthias_bgg has quit [Quit: Leaving]
NeuroScr has joined #linux-sunxi
NeuroScr has quit [Client Quit]
dddddd has quit [Ping timeout: 250 seconds]
yann has joined #linux-sunxi
random_yanek has quit [Quit: random_yanek]
random_yanek has joined #linux-sunxi
random_yanek has quit [Max SendQ exceeded]
random_yanek has joined #linux-sunxi
Mandragor has quit [Ping timeout: 250 seconds]
Mandragor has joined #linux-sunxi
RichardG867 has joined #linux-sunxi
RichardG867 has quit [Read error: Connection reset by peer]
RichardG867 has joined #linux-sunxi
\\Mr-C\\ has quit [Quit: (Read error: Connection reset by beer)]
JohnDoe_71Rus has quit [Quit: KVIrc 5.0.0 Aria]
\\Mr_C\\ has joined #linux-sunxi
megi has quit [Quit: WeeChat 2.3]
netlynx has quit [Quit: Ex-Chat]
iamfrankenstein has quit [Quit: iamfrankenstein]
IgorPec has quit [Ping timeout: 240 seconds]
f0xx has quit [Ping timeout: 250 seconds]
pgreco has quit [Ping timeout: 250 seconds]
lurchi__ has joined #linux-sunxi
pgreco has joined #linux-sunxi
hanni76 has quit [Quit: Leaving]
\\Mr_C\\ has quit [Quit: (Read error: Connection reset by beer)]
The_Loko has quit [Remote host closed the connection]
dddddd has joined #linux-sunxi
<catphish> when i run "sunxi-fel spl file" it executes "file" on my sunxi but then just hangs, is it waiting for something to happen?
<KotCzarny> when you execute something
<KotCzarny> you exit fel-stack, no?
<catphish> i do not, in fact all my code does is turn on an LED and stop
<KotCzarny> but do you return where it was?
<KotCzarny> think of it as fel-stack running on cpu
<KotCzarny> you run some program on cpu, you end your program without returning anywhere
<catphish> perhaps it does more than i think it does, i thought it would just upload my code at 0x0, execute it, and stop caring
<KotCzarny> if you want nanny usb-otg, think of using openrisc as a loader/communicator
<KotCzarny> or just write your own
paulliu has quit [Quit: Leaving.]
<catphish> so "sunxi-fel" is waiting for me to return to somewhere? or it wants the code it executes to take control of the usb-otg?
<catphish> if there's somewhere to be returned to, perhaps i'm misunderstanding fel :)
apritzel has joined #linux-sunxi
<KotCzarny> ahm, so your question is not if code runs, but why sunxi-fel spl file doesnt return to shell instantly?
<KotCzarny> does it time out ?
<catphish> that'c correct, the code runs, but the sunxi-fel doesn't return to the shell until some time later, when it times out
<catphish> "usb_bulk_send() ERROR -7: Operation timed out"
<KotCzarny> so yeah, most likely it expects some answer from fel stack
<KotCzarny> in case your command didnt run or something
<KotCzarny> might be a bug too
<apritzel> catphish, does your code do a "bx lr" after it has finished? With LR and SP being the same as when you entered?
<catphish> apritzel: no, my code just turns on an LED then halts
<apritzel> that's how U-Boot's SPL hands back return to FEL
<apritzel> catphish, your own assembly code?
<catphish> apritzel: yeah, right now just literally a few lines to check i could boot successfully
<catphish> so u-boot does some init, and then returns control to the boot ROM?
<catphish> and then, i guess FEL uploads some more code after that
<apritzel> catphish: if it detects being loaded via FEL, it hands back control to the boot ROM, by restoring as much of the state as possible and returning
<catphish> thanks, looks like i can work around this problem by using "sunxi-fel write 0 boot; sunxi-fel exe 0" instead
<catphish> has the same effect but doesn't expect anything to happen next
bonbons has joined #linux-sunxi
<apritzel> catphish:;a=blob;f=arch/arm/cpu/armv7/sunxi/fel_utils.S
bonbons_ has quit [Ping timeout: 246 seconds]
<apritzel> catphish: sure you can, but if you didn't touch LR or SP, then just a "bx lr" instead of a wfi loop sounds simple enough
<catphish> being able to arbitrarily jump back to FEL might actually be extremely useful for me to debug
<apritzel> catphish: another dirty trick: the FEL routine starts at 0x20 (for 64-bit SoCs)
<apritzel> just branch there enters FEL mode
Putti has quit [Ping timeout: 246 seconds]
<apritzel> that's 0xffff0020 for older SoCs
<catphish> i'm on an A20, i think it starts near 0xffff0000
<catphish> ah, yeah, that's the one, thanks
IgorPec has joined #linux-sunxi
dev1990 has joined #linux-sunxi
<catphish> one more thing: am i right in thinking that the boot rom will load eGON.BT0 from an SD card?
<pmpp> catphish: it will
<catphish> at an offset of 8KiB, it's not working, but hopefully i'm not far off, maybe i just didn't quite populate the header fully enough :)
<pmpp> 8K + some bytes if you are speaking of the marker
<apritzel> catphish: you need the "eGON.BT0" magic at 4 bytes, the size (512 bytes aligned) at 16 bytes and the 32-bit checksum over that whole size at +12
<apritzel> mksunxiboot does this for you
<catphish> apritzel: ah, my size is only 4 byte aligned
<catphish> the rest is correct (and sunxi-fel confirms the checksum), i'll check the length alignment now
<catphish> apritzel: i just used mksunxiboot, works perfectly :)
<catphish> thanks! i think my problem was just the length alignment, but the tool is great
Xalius_Ph has joined #linux-sunxi
BenG83 has quit [Quit: Leaving]
<catphish> thank you once again smart humans :)
sasrubix has joined #linux-sunxi
<catphish> fun observation: in its default state, the a20 runs really cool, in the state u-boot leaves it, it gets hotter than the sun, i assume this is what happens when you start enabling clocks
<KotCzarny> or setting cpu voltage up
<KotCzarny> required for higher freqs
<catphish> makes sense
<KotCzarny> but yeah, the more components tick, the more heat is generated
IgorPec has quit [Ping timeout: 250 seconds]
<catphish> i guess my first priority needs to be getting the dram working, i imagine this will be moderately complicated, off i go to read again :)
<KotCzarny> you are up for a rough ride
<KotCzarny> :>
<apritzel> VERY rough
<apritzel> and reading won't help you much
<apritzel> catphish: your best bet is to take the U-Boot code
<catphish> it looked like someone documented the dram unofficially
<apritzel> or libdram if you don't care about licensing
<catphish> but i guess copying u-boot will be the only way to actually get started
<apritzel> catphish, unless you want to miss Christmas: yes
<catphish> unfortunately, i'm sad, and this will probably be my idea of a fun christmas
indy has quit [Quit: ZNC -]
<catphish> :)
<KotCzarny> beware, you might end up sadder
<catphish> lol
<apritzel> catphish: yeah, it sounds like fun at the beginning, but the DRAM controller is so huge and so little documented, you will end up frustrated
<apritzel> catphish: believe me!
<catphish> that's how i've ended up with much of this, don't know how i'd have got anywhere without the linux-sunxi code and docs
<catphish> is this project's work mostly based on observing the nonfree u-boot fork?
<KotCzarny> there were some help from aw long time ago
<KotCzarny> then the linux guys at aw got fired or reallocated
<KotCzarny> then there was another attempt at aw opening manuals
<KotCzarny> in the meantime there was a lot of RE
<apritzel> catphish, for the interesting parts like booting, SMP bringup and DRAM, it's mostly looking at some dodgy code, yes
<catphish> to be honest i'm confused by the whole situation, how is it that linux can be the only OS that realistically runs on their platform, yet they're so uninvolved in it
<KotCzarny> volume
<apritzel> catphish, not Linux, really, but Android
<KotCzarny> most of their chips end up in cheapo android products
<catphish> isn't android just linux+google stuff?
indy has joined #linux-sunxi
<KotCzarny> sure, but then, BLOBS
<catphish> same low leven stuff no?
<KotCzarny> beware the blobs
<catphish> is it really cheaper for them to make blobs than just contribute code?
<apritzel> catphish, technically it is related, but mindset wise it's not
<KotCzarny> seems so
<apritzel> catphish: they just need seemingly working code
<apritzel> which is a big difference from upstream quality
<catphish> i'm so confused by the whole thing, i know hobby osdev isn't their market, but it baffles me that they wouldn't want linux to properly support their hardware (and avoid the need to publish new blobs for every new android version)
<apritzel> so you just #ifdef 0 parts that are in the way and ship it
<apritzel> catphish, but from their POV they support Linux
<KotCzarny> catphish: coming back with 'if you dont know what its all about, it's about money'
<catphish> i suppose they just want to show their customers android running, and they get sales
<KotCzarny> must be cheaper for them that way
<catphish> the other thing that troubles me is that this is the same (worse even) as every other (affordable) arm SoC maker
<KotCzarny> worst offenders are gpu parts included in those socs
<catphish> yeah, i actually find it odd that even arm don't document the gpu publicly
<catphish> with them it must be more than a cost thing
<catphish> since their docs are usually excellent
<apritzel> catphish, as KotCzarny said: it's all about money ;-)
<apritzel> there is no real market for enthusiast running Linux on dev boards
<catphish> shame really, but if i cant a cheap SoC, i can't much complain
<apritzel> the people paying money need Android 3D working yesterday
<KotCzarny> wasnt there some riscv opensourced platforms?
<apritzel> KotCzarny: how does this help with the GPU?
<catphish> my only real requirement for this hobby project is hdmi output, i've not found anything at all with arm, hdmi, and proper docs
<catphish> a20 is the closest
<KotCzarny> apritzel: evolution and hope for the future if it catches (although it probably wont)
<apritzel> yeah, funny enough due to Allwinner providing so little the actual support in mainline is probably among the best
<KotCzarny> hdmi is a blob too in aw sources
<catphish> KotCzarny: well at least the released a20 manual covers hdmi and DE properly
<catphish> i've not found another affordable SoC with that much documented
<apritzel> KotCzarny, point is: the RISCV core might be free, but if you think about it the ARM *core* is not the problem, it's all the rest
<karlp> it's always abot the periphs :)
<catphish> nb. by sad, i mean "Unfashionable; socially inadequate or undesirable."
<karlp> the cpu ISA is just a gcc flag.
<catphish> not tha bad kind :)
<KotCzarny> apritzel: but again, if riscv people have open mindset, and arm doesnt, who should be supported
<catphish> i wish someone would build a fully open source SoC with modern functionality, but i don't suppose it would ever cover its costs, and wouldn't be useful when running on fpga
<catphish> plus without 3d graphics, i doubt anyone would be very interested
<apritzel> KotCzarny, feel free to support the development of your dream SoC with millions of dollars ;-)
<buZz> :D
<apritzel> until then you just pick up what the market gives you
<KotCzarny> mmm, dolla dolla bills
<catphish> i have no idea how much it costs to develop an arm soc
<catphish> i always assumed the cost was the fab rather than the design, since arm already did the hard work
<apritzel> catphish: too much
<apritzel> you need some well paying customer to sponsor this
<catphish> do we actually know how much?
<karlp> KotCzarny: riscv is again just the core though, we'll see the the same garbage with the periphs.
<karlp> all it changes is less royalties to the chip vendor.
<catphish> google says tens of millions
<apritzel> karlp: indeed, RISCV makes fun of other ISAa, but this doesn't really matter
\\Mr_C\\ has joined #linux-sunxi
<catphish> maybe a future me will make a SoC on top of a zynq (or similar fpga + arm core)
<buZz> i think i remember a SoC maker that lets you drag&drop to design one
<karlp> catphish: hah, because fpgas are so much more open :)
<karlp> nice one :)
<apritzel> buZz: time to market, cheap, fast. Pick one.
* karlp laughs
<catphish> well back to what's realistic i guess :)
<pmpp> 6502 ftw
* buZz clicks something up
<buZz> SHIP IT
<karlp> pmpp: back when you had no periphs, and they were all extrenal right? :)
<catphish> can't we just beg AW to release nice peripheral docs?
<catphish> they already did plenty
<pmpp> karlp: yeah plug and pray
<apritzel> catphish, what's really missing doc-wise?
<catphish> well the obvious one (that started this conversation) is the dram controller
<apritzel> catphish, the community has actually filled most of the gaps
<catphish> i also lack information about the DE2.0 in the h3
<KotCzarny> some docs about de2/3 are uploaded by aw to wiki
<apritzel> catphish, but H3 has quite good HDMI support in both U-Boot and Linux, no?
<catphish> anyway, between whet they publish, what the community has documented, and reading the linux code, it's definitely possible
<catphish> i'm very thankful to the community here
<catphish> apritzel: i believe so, i'll need to go through the uboot/linux code once i have my h3 chip, hopefully i can piece it together, as other people clearly have :)
<catphish> part of the problem is that i'm spoiled by ST documentation, when you want to do something with an STM32, they literally walk you through it
<buZz> didnt ST make some linux socs too?
<buZz> hmm, seems they just do MCUs
<catphish> i believe it's only MCUs, at least right now
<catphish> their arm mcus are amazing (and cheap)
<catphish> but nothing fast enough for hdmi
<buZz> just bitbang some VGA through some r2r's then :P
<apritzel> wasn't there some MMU capable STM32 SoC lately, gaining Linux support?
<catphish> sounds plausible
<apritzel> "STM32MP1 is a microprocessor designed by STMicroelectronics, based on a dual Arm Cortex-A7."
<karlp> are tehre any new details on that yet?
<catphish> well that's sexy
<karlp> people have been running linux on f4/f7/h7 for years.
<buZz> hm? that model doesnt give a hit on
<karlp> buZz: exactly,
<karlp> but there's a dts committed for it :)
<karlp> so it's.... somewhere.
<buZz> just folklore
<buZz> :P
<KotCzarny> lol
<catphish> i's quite enjoy an IC with a framebuffer and an hdmi output, then i could attach any processor to it
<karlp> ttps:// and search for stm32...
<karlp> there's there's 2 ts files for 32mp1 variants
<buZz> would be nice with commit history
<catphish> ST wrote the linux patches in february 2018, maybe the chip is on the way
<buZz> Sorry, we had to truncate this directory to 1,000 files. 857 entries were omitted from the list.
<karlp> buZz: pick a different web interface?
<apritzel> yeah, real people submit patches *before* silicon hits the market
<buZz> boo github
Mr__Anderson has quit [Remote host closed the connection]
<apritzel> buZz, I think I saw patches to split up this directory into per-vendor ones, as arm64 does
<karlp> catphish: you may want to look for hdmi transceiver solutions then,
<buZz> yeah its about time
<catphish> karlp: yeah that might be sane, anyway i'm really just learning here, so i'll keep playing with the sunxi for now and reading up about what else is available
<catphish> no real requirement
<fALSO> karlp, 22:27 < karlp> people have been running linux on f4/f7/h7 for years.
<fALSO> awesome, i didnt knew that, i have a stm32f4 discovery board, but it has little ram, so it probably doesnt run lunix, but im going to dig into that
<fALSO> nice!
<apritzel> but those are Cortex-M based, so no MMU and thus only ucLinux
<fALSO> i also saw some stm32 things on the kernel ;-)
<buZz> fALSO: yeah but its for a nonexisting chip :P
<catphish> anyway, back to my dram controller i guess :)
<apritzel> catphish: when you are finished with that in an hour or so, you can help out here:
<catphish> ha
<apritzel> ZynqMP has some good DRAM controller documentation online, and it's almost the same IP as in Allwinner
<KotCzarny> can i nit pick?
<KotCzarny> * (C) Copyright 2018 Arm Ltd.
<KotCzarny> * based on previous work by:
<KotCzarny> * (C) Copyright 2017 Icenowy Zheng <>
<KotCzarny> :>
<anarsoul|2> :D
<KotCzarny> but kudos for credits
<apritzel> KotCzarny, are you concerned about the copyrights?
<KotCzarny> nah
<KotCzarny> just looks funny
<apritzel> that's purely legal, meaning: who must sue the guys violating the GPL
<apritzel> if you want to do this, be my guest, meanwhile I leave it up to our lawyers
<KotCzarny> :)
<anarsoul|2> must? :)
<apritzel> what really counts is the commit log and git blame
<apritzel> and btw: adding your copyright if you made substantial contributions is totally standard
<apritzel> I saw people doing it for a few lines already
<catphish> by the way, how does rockchip compare al sunxi in terms of openness?
<apritzel> good question!
<catphish> i see they make up the majority of the cheap SoC market
<catphish> *compare with
<apritzel> they are better from 10,000 feet, but not really if you count in the linux-sunxi community
<anarsoul|2> well, there's no status matrix for rk
<anarsoul|2> so it's hard to compare
<apritzel> when it comes to DRAM, for instance, it's the same problem
<apritzel> the published manual says explicitly: "... part 1"
<anarsoul|2> apritzel: support for some chips is blobless, but it's worse than for aw
<anarsoul|2> e.g. rk3399 is blobless for ddr3/lpddr3, but it's not for lpddr4
\\Mr_C\\ has quit [Quit: (Read error: Connection reset by beer)]
\\Mr_C\\ has joined #linux-sunxi
<catphish> i see
NeuroScr has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
sasrubix has quit [Quit: Leaving]
<catphish> somewhat nontrivial
leviathanch has joined #linux-sunxi
<karlp> apritzel: welllll, uclinux _is_ linux these days.... but yeah, they're all running it on boards with external ram at least
<karlp> buZz: there's dts files and periph drivers for the cortex-m stm32s too, not just the mythical mp1
jbrown has quit [Ping timeout: 252 seconds]
<apritzel> karlp, "_is_ linux" is somewhat true for the *kernel* (even though the kernel config is probably quite different)
<buZz> karlp: nice but maybe just for inclusion on other boards as peripherals?
<apritzel> karlp, but AFAIK it's quite a different story for userland
<karlp> embedded artists will give you a bsp to "run linux" and busybox on an f4 and things. just go look for it :)
<karlp> you don't need the dts and the driver files to use it as a peripheral...
<karlp> is a reasonable starting point....
<karlp> emcraft uses the old 2.6 uclinunx stuff, but there's some newer post merge stuff around.
<karlp> plaes: speakingo f dram, that patch of yours that regressed for me on h3, what had that been tested on? it'sgenerateed 0 feedback on the uboot list, is anyone looking at it?