Turl changed the topic of #linux-sunxi to: Allwinner/sunxi /development discussion - did you try looking at our wiki? https://linux-sunxi.org - Don't ask to ask. Just ask and wait! - https://github.com/linux-sunxi/ - Logs at http://irclog.whitequark.org/linux-sunxi
<danboid> I think it sees my remote as a keyboard of sorts?
vagrantc has joined #linux-sunxi
<danboid> Neither seem to work
vishnup has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
danboid has quit [Ping timeout: 260 seconds]
kaspter has joined #linux-sunxi
akaizen has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
nove has quit [Quit: nove]
mzki has quit [Quit: leaving]
<catphish> woo, i finally got my memory management working properly!
* apritzel is curious ... who was the culprit?
<catphish> apritzel: when setting TTBR0, i wasn't setting the memory cache type
<catphish> you have to tell TTBR0 what type of memory the table itself is held in
<apritzel> ah
<apritzel> so is it faster now?
<catphish> well, without that set, i was unable to invalidate the table lookup cache, which in turn meant it was using the table cached by u-boot
<catphish> which worked, because it was an identity mapping, but was slow
<catphish> (as far as i can tell anyway)
<catphish> now im using my own table, its very very fast
<apritzel> great!
<catphish> so how many days consultancy do i owe you for :)
<catphish> thanks a lot for the vast amount of help
<apritzel> so you spend the evening on fixing a few bits? ;-)
<catphish> 2 evenings, 2 bits :)
<apritzel> but learnt everything about the ARM VMSA ...
<catphish> certainly a small fraction of it :)
<catphish> come to think of it, i still dont invalidate the I and D caches, should probably do that too to prevent crashes down the line
<catphish> but it seems to be working well
<apritzel> pah, why being correct if you can be fast :-D
<catphish> now i have to adjust all my busy timing loops :)
<catphish> they're too fast
vagrantc has quit [Quit: leaving]
ganbold has quit [Quit: This computer has gone to sleep]
khuey is now known as khuey|away
catphish has quit [Quit: Leaving]
orly_owl has joined #linux-sunxi
cnxsoft has joined #linux-sunxi
hipboi has joined #linux-sunxi
fl_0 has quit [Ping timeout: 260 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
fl_0 has joined #linux-sunxi
orly_owl has quit [Ping timeout: 250 seconds]
ricardocrudo has quit [Ping timeout: 272 seconds]
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
akaizen has quit [Ping timeout: 260 seconds]
apritzel has quit [Ping timeout: 248 seconds]
akaizen has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
ninolein has quit [Ping timeout: 260 seconds]
akaizen has joined #linux-sunxi
ninolein has joined #linux-sunxi
akaizen_ has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
akaizen_ has quit [Read error: Connection reset by peer]
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
<wens> montjoie: you create a node under the pio / r_pio controller, list the pin you want to set, and the function to set it to, and the pull-up/down and drive strength
<wens> you then reference that node under the device that will be using it
hipboi_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 264 seconds]
kaspter has quit [Read error: Connection reset by peer]
FDCX has quit [Ping timeout: 250 seconds]
orly_owl has joined #linux-sunxi
codekipp1r has quit [Ping timeout: 276 seconds]
codekipper has joined #linux-sunxi
FDCX has joined #linux-sunxi
hipboi_ has quit [Quit: This computer has gone to sleep]
TheSeven has quit [Ping timeout: 240 seconds]
TheSeven has joined #linux-sunxi
p1u3sch1_ has quit [Ping timeout: 260 seconds]
p1u3sch1 has joined #linux-sunxi
p1u3sch1 has quit [Ping timeout: 260 seconds]
p1u3sch1 has joined #linux-sunxi
arossdotme has quit [Ping timeout: 260 seconds]
vagrantc has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
gzamboni has quit [Read error: Connection reset by peer]
vagrantc has quit [Quit: leaving]
Ueno_Otoko has joined #linux-sunxi
_stephan has joined #linux-sunxi
arossdotme has joined #linux-sunxi
vickycq has quit [Ping timeout: 276 seconds]
Andy-D has quit [Ping timeout: 276 seconds]
Andy-D has joined #linux-sunxi
vickycq has joined #linux-sunxi
hipboi has joined #linux-sunxi
akaizen has joined #linux-sunxi
IgorPec has joined #linux-sunxi
domidumont has joined #linux-sunxi
domidumont has quit [Remote host closed the connection]
premoboss has quit [Ping timeout: 272 seconds]
domidumont has joined #linux-sunxi
yann|work has quit [Ping timeout: 256 seconds]
domidumont1 has joined #linux-sunxi
hipboi has quit [Quit: This computer has gone to sleep]
domidumont1 has quit [Remote host closed the connection]
domidumont has quit [Read error: Connection reset by peer]
domidumont has joined #linux-sunxi
domidumont1 has joined #linux-sunxi
soderstrom has joined #linux-sunxi
domidumont has quit [Ping timeout: 246 seconds]
cnxsoft1 has joined #linux-sunxi
cnxsoft has quit [Ping timeout: 272 seconds]
cnxsoft1 is now known as cnxsoft
danboid has joined #linux-sunxi
_massi has joined #linux-sunxi
kaspter has joined #linux-sunxi
soderstrom has quit [Ping timeout: 255 seconds]
diego_r has joined #linux-sunxi
engideavr has quit [Quit: Konversation terminated!]
engideavr has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
orly_owl has quit [Quit: leaving]
orly_owl has joined #linux-sunxi
apritzel has joined #linux-sunxi
danboid has quit [Quit: Leaving]
apritzel has quit [Ping timeout: 248 seconds]
yann|work has joined #linux-sunxi
zumbi has quit [Quit: Lost terminal]
lemonzest has joined #linux-sunxi
reinforce has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
apritzel has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
hipboi has joined #linux-sunxi
vishnup has quit [Quit: Connection closed for inactivity]
iamfrankenstein has joined #linux-sunxi
hipboi has quit [Quit: This computer has gone to sleep]
hipboi has joined #linux-sunxi
hipboi has quit [Client Quit]
leio has joined #linux-sunxi
<ccaione> Wizzup: 2
cptG has joined #linux-sunxi
pekka_afk is now known as pekka_10
leio has quit [Quit: No Ping reply in 180 seconds.]
leio has joined #linux-sunxi
leio has quit [Remote host closed the connection]
viccuad has joined #linux-sunxi
gzamboni has joined #linux-sunxi
matthias_bgg has quit [Quit: Leaving]
<KotCzarny> i must say that fbturbo is a must
<KotCzarny> i wonder why loboris and igorpec dont compile it by default
<oliv3r> wens: you worked a lot on the cpufreq stuff for sun[47]i right? have you noticed crashes when frequent scaling changes happen on sun7i?
leio has joined #linux-sunxi
<GeneralStupid> KotCzarny: what image are you using? (Orange Pi PC? )
bmeneg has quit [Ping timeout: 260 seconds]
<KotCzarny> gs: loboris jessie
<KotCzarny> with some fixes applied
zumbi has joined #linux-sunxi
<KotCzarny> but everything works so far
<KotCzarny> (well, apart from 100mbit, its stuck on 10, right i had to test different cable)
<KotCzarny> no dice, lets try diff. switch
<GeneralStupid> i have problems with my dvb-t stick (may be due to low current...) and problems with slow WiFi
<KotCzarny> have you reallocated irqs?
<KotCzarny> it might help with usb
<GeneralStupid> no, i dont. is there a sunxi page for that?
<KotCzarny> nope, unless you consider user pages
<GeneralStupid> i have 100Mbps uplink and i dont changed something there
<KotCzarny> tkaiser has a few hacks
<GeneralStupid> do you have a link?
<KotCzarny> i can even paste you my reallocs:
<GeneralStupid> that would be nice
<KotCzarny> i've put all usb on cpu1, but you can assign ports to diff. cpus
vishnup has joined #linux-sunxi
<KotCzarny> but seriously, would be nice to have kernel auto scheduling irqs
ikmaak has quit [Read error: Connection reset by peer]
bmeneg has joined #linux-sunxi
yann|work has quit [Ping timeout: 240 seconds]
montjoie has quit [Quit: Lost terminal]
ikmaak has joined #linux-sunxi
montjoie has joined #linux-sunxi
mmwolbrink has joined #linux-sunxi
<apritzel> KotCzarny: any chance you are looking for irqbalance? http://linux.die.net/man/1/irqbalance
yann|work has joined #linux-sunxi
vishnu_ has joined #linux-sunxi
enrico_ has joined #linux-sunxi
yann|work has quit [Ping timeout: 240 seconds]
naobsd has joined #linux-sunxi
viccuad has quit [Ping timeout: 260 seconds]
mzki has joined #linux-sunxi
yann|work has joined #linux-sunxi
<KotCzarny> apritzel: probably, but it doesnt seem to be enabled/working?
<apritzel> yeah, you have to enable it
<apritzel> just forgot how that worked, though ...
<GeneralStupid> kernel config?
<GeneralStupid> but ... comeon it should be fine to have wifi on one core and dvb on the other... four cores are more then enough :D
<apritzel> GeneralStupid: irqbalance is a userland tool, afaik it should be installed as a service in your distro
ricardocrudo has joined #linux-sunxi
<KotCzarny> gs: did you check how that dvbt stick behaves on diff irq?
<wens> oliv3r: not really, but i'm not stressing my boards either
viccuad has joined #linux-sunxi
<GeneralStupid> apritzel: not installed :D
<GeneralStupid> KotCzarny: no not right now. i will write an exam this evening, so i have to learn a bit :)
<GeneralStupid> i think i will try irqbalance on my second device
<KotCzarny> yeah 'exam' and 'irc'
<ssvb> wens, oliv3r: maybe some sort of a timing issue? caused by not waiting for the PLL lock or until the new voltage settles?
<ssvb> oliv3r: maybe you can try to disable voltage scaling as a test and check if this helps
<ssvb> oliv3r: btw, do you have a testcase for this problem?
<ssvb> apritzel: Cortex-A53 errata list is rather scary
<apritzel> ssvb: should see the internal version :-D
<apritzel> ssvb: anything in particular?
<apritzel> ssvb: btw: your libjpeg-turbo NEON stuff is seriously mind-boggling ;-)
<apritzel> wasn't aware that the aarch64 support is that sophisticated
leio has quit [Read error: Connection reset by peer]
<ssvb> apritzel: can you read the REVIDR bits? Cortex-A53 r0p4 still has two CatA errata: 835769 and 843419
<apritzel> ssvb: sure, but not right now (I left the board at home to get some work done here ;-)
<ssvb> ok, there is no hurry
leio has joined #linux-sunxi
mmwolbrink has quit [Ping timeout: 250 seconds]
Ueno_Otoko has quit [Ping timeout: 260 seconds]
<apritzel> ssvb: I noticed that Allwinner added a compiler switch in their BSP kernel Makefile for a certain errata
<apritzel> (which is for 843419, so I guess the SoC has at least this)
mmwolbrink has joined #linux-sunxi
mmwolbrink has quit [Remote host closed the connection]
ssvb has quit [Remote host closed the connection]
afaerber has quit [Quit: Ex-Chat]
<vishnu_> Received my Pine64 :)
afaerber has joined #linux-sunxi
<apritzel> vishnu_: enjoy! Seems like the dd'able image has bootdelay set to 0 in u-boot, so you can't get the prompt easily
<apritzel> (the older image I used as a base had "3" in there)
<apritzel> you may be able to fix this by changing the variable in the u-boot env partition
<vishnu_> okie
lynxis has quit [Remote host closed the connection]
lynxis has joined #linux-sunxi
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<apritzel> vishnu_: the u-boot environment data is checksum'ed, though, so you have to recreate it with tools/mkenvimage.c from the u-boot tree
cnxsoft1 has joined #linux-sunxi
engidea-vr has joined #linux-sunxi
enrico__ has joined #linux-sunxi
Turl has quit [Ping timeout: 260 seconds]
Nacho__ has joined #linux-sunxi
techn_ has joined #linux-sunxi
menomc has joined #linux-sunxi
wigyori_ has joined #linux-sunxi
GeneralS1upid has joined #linux-sunxi
egbert has quit [Disconnected by services]
swabbles has joined #linux-sunxi
vbmithr_ has joined #linux-sunxi
lynxis_ has joined #linux-sunxi
egbert has joined #linux-sunxi
Wizzup_ has joined #linux-sunxi
LostSoul_ has joined #linux-sunxi
plaes_ has joined #linux-sunxi
tkoskine_ has joined #linux-sunxi
orly_owl_ has joined #linux-sunxi
tomcheng86 has joined #linux-sunxi
ccaione_ has joined #linux-sunxi
aep_ has joined #linux-sunxi
jelle- has joined #linux-sunxi
joedj_ has joined #linux-sunxi
specing_ has joined #linux-sunxi
fredy_ has joined #linux-sunxi
Turl has joined #linux-sunxi
lynxis has quit [*.net *.split]
ricardocrudo has quit [*.net *.split]
aep has quit [*.net *.split]
fredy has quit [*.net *.split]
jelly has quit [*.net *.split]
SJRvanSchaik has quit [*.net *.split]
Wizzup has quit [*.net *.split]
LostSoul has quit [*.net *.split]
vpeter has quit [*.net *.split]
jelle has quit [*.net *.split]
vbmithr has quit [*.net *.split]
atenart has quit [*.net *.split]
tkoskine has quit [*.net *.split]
wens has quit [*.net *.split]
Jacmet has quit [*.net *.split]
techn has quit [*.net *.split]
camh has quit [*.net *.split]
yann|work has quit [*.net *.split]
enrico_ has quit [*.net *.split]
paulk-collins has quit [*.net *.split]
orly_owl has quit [*.net *.split]
engideavr has quit [*.net *.split]
cnxsoft has quit [*.net *.split]
Nacho_ has quit [*.net *.split]
ccaione has quit [*.net *.split]
bbrezill1 has quit [*.net *.split]
joedj has quit [*.net *.split]
book` has quit [*.net *.split]
tomcheng76 has quit [*.net *.split]
wigyori has quit [*.net *.split]
specing has quit [*.net *.split]
GeneralStupid has quit [*.net *.split]
rellla has quit [*.net *.split]
mnemoc has quit [*.net *.split]
plaes has quit [*.net *.split]
cnxsoft1 is now known as cnxsoft
joedj_ is now known as joedj
ccaione_ is now known as ccaione
ccaione has quit [Changing host]
ccaione has joined #linux-sunxi
book` has joined #linux-sunxi
specing_ is now known as specing
jelle- is now known as jelle
jelle has quit [Changing host]
jelle has joined #linux-sunxi
wigyori_ is now known as wigyori
atenart has joined #linux-sunxi
Jacmet has joined #linux-sunxi
wens has joined #linux-sunxi
vpeter has joined #linux-sunxi
sajimon has quit [Ping timeout: 272 seconds]
sajimon has joined #linux-sunxi
ricardocrudo has joined #linux-sunxi
bbrezill1 has joined #linux-sunxi
yann|work has joined #linux-sunxi
camh has joined #linux-sunxi
vishnu_ has quit [Remote host closed the connection]
tkoskine_ is now known as tkoskine
Ueno_Otoko has joined #linux-sunxi
vishnu_ has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
jelly-home has joined #linux-sunxi
tomboy64 has quit [Quit: Uhh ... gotta go.]
tomboy64 has joined #linux-sunxi
cnxsoft has quit [Quit: cnxsoft]
ricardocrudo has quit [Remote host closed the connection]
_stephan has quit [Quit: Ex-Chat]
fredy_ is now known as fredy
fredy is now known as Guest21129
ricardocrudo has joined #linux-sunxi
jjwerner has joined #linux-sunxi
tgaz has quit [Ping timeout: 256 seconds]
tgaz has joined #linux-sunxi
<jjwerner> morning gentleman
<jjwerner> aptritzel: i looked at the regulator both in mainline kernel and sunxi bsp
<apritzel> jjwerner: does the BSP code access the regulator directly? Or do they call arisc?
<jjwerner> sunxi has added a whole bunch of crud in /power/sxp in their kernel branch instead of adding regulator where it should be. I will look into more details of getting regulator code in place, and how it could / will be accessed
<apritzel> "has added a whole bunch of crud": welcome to Allwinner's BSP ;-)
<KotCzarny> hmm, fbturbo looks funny on opipc
<KotCzarny> while in the browser its ok, moving desktop windows/scrolling gtk lists 'forgets' to redraw
<jjwerner> i believe directly, but I haven't spent enough time understanding the code yet... day job calls
<jjwerner> I saw the mentions of the dd image. Have you posted it somewhere online?
<apritzel> jjwerner: not yet, but getting very close
<apritzel> I spent the whole evening yesterday deciphering their loading process
<apritzel> I think I got it now, just need to wrap things up
p1u3sch1 has quit [Ping timeout: 256 seconds]
<jjwerner> cool, my board is is like 20 miles away. I should get it tomorrow
<apritzel> nice, I keep you posted ...
p1u3sch1 has joined #linux-sunxi
IgorPec has quit [Ping timeout: 272 seconds]
<jjwerner> I will keep chipping at the regulator code, have to catch wens and ask him about his patch for axp809 I rather go through the "read the spec, look at other implementation, adjust to different hardware and write code for mainline" than figure wtf they are trying to do in allwinner bsp ;)
rellla has joined #linux-sunxi
plaes_ is now known as plaes
plaes has quit [Changing host]
plaes has joined #linux-sunxi
<apritzel> yeah, I'd suggest to just look at their code if you need to look something up which isn't properly documented or doesn't work as expected
<apritzel> otherwise it may harm your sanity ;-)
<jjwerner> what is this sanity you're talking about ? :)
<apritzel> ;-)
ricardocrudo has quit [Remote host closed the connection]
apritzel has quit [Ping timeout: 248 seconds]
p1u3sch1 has quit [Ping timeout: 240 seconds]
p1u3sch1 has joined #linux-sunxi
IgorPec has joined #linux-sunxi
<wens> we have the datasheet for the axp803, so one can just adapt the axp drivers in mainline for it
<wens> assuming we have access to the rsb controller that is
IgorPec has quit [Ping timeout: 265 seconds]
<jjwerner> hey wens
camh has quit [Ping timeout: 260 seconds]
camh has joined #linux-sunxi
<jjwerner> apritzel mentioned your patches for axp809
<jjwerner> i looked at the specs and that one is the closest to axp803
domidumont1 has quit [Read error: Connection reset by peer]
apritzel has joined #linux-sunxi
<jjwerner> what would limit our access to rsb? lack of support or the EL1/EL3 shenenigans?
tomboy64 has quit [Quit: Uhh ... gotta go.]
JohnDoe_71Rus has joined #linux-sunxi
IgorPec has joined #linux-sunxi
kelvan has quit [Read error: Connection reset by peer]
Ueno_Otoko has quit [Ping timeout: 255 seconds]
kelvan has joined #linux-sunxi
tomboy64 has joined #linux-sunxi
<wens> the "secure peripherals controller" controls what peripherals are available to non-secure
<wens> plus the arisc firmware also uses rsb, and you don't want to step on it
<jjwerner> ok, so in bsp there is option to use two-wire-interface or arisc, i have to look at kconfig of running image to see how the kernel was configured, but I do expect it to be arisc
akaizen has joined #linux-sunxi
<wens> it seems arisc is loaded by u-boot, so we don't have much of a choice if we use stock u-boot from the bsp
<wens> or just leave it till later :|
akaizen has quit [Remote host closed the connection]
<jjwerner> so let me get it straight, we can get mfd data and regulator code in, but while arisc is initialized we won't be able to talk with the regulator without sunxi arisc code?
yann|work has quit [Ping timeout: 255 seconds]
<wens> you can, but arisc is running beside you, doing god knows what
<KotCzarny> 'hey, we put a soc in your soc'
<wens> on armv7 we're not using arisc
<wens> though mripard has been playing with it recently
<apritzel> jjwerner: of course their kernel has no /proc/config.gz, if you were thinking about that
<apritzel> and I agree with wens we shouldn't tinker with the PMIC as long as the arisc controls it
<wens> it's probably somewhere is the bsp
<apritzel> but as the bsp does not match what's in the image ...
<wens> :/
<wens> anyway, since the mfd/regulator drivers just use regmap
<wens> we could write a simple arisc-rsb driver, or port the one the bsp has (if any)
<apritzel> wens: arisc-rsb? you mean to send requests to the arisc?
<apritzel> AFAIK this does not use rsb, but some kind of mailbox
<wens> apritzel: yeah
<apritzel> if I remember the BSP code correctly
<wens> apritzel: i meant send commands to arisc using the msgbox
<apritzel> right
<wens> iirc it has a few msg types for rsb access
<wens> i'd rather not do a whole new regulator driver on top of the arisc pmic stuff
<apritzel> ah, you mean we should actually pass commands to the AXP, just not using the RSB directly, but instead using the msgbox?
<jjwerner> oh of course, why would it have config.gz :/
<jjwerner> wens: but using the msgbox would still require us to use their arisc code
<jjwerner> ?
<wens> apritzel: correct
<wens> jjwerner: yes, but since it's loaded by their u-boot... unless we have some other bootloader we can use...
<apritzel> wens: but actually programming the AXP would kind of defeat the purpose of having it separately in the first place, right?
<apritzel> as I would have imagined that this prevents any driver from doing stupid things like applying 3V to the CPU or the like
<wens> apritzel: that's what regulator constraints are for
<wens> they are set in the dt
<jjwerner> also that would also step away from clean regulator model in the mainline kernel?
<apritzel> ideally I would imagine to have a rather abstract interface, which says: power that USB controller
<apritzel> without having to know the voltage it requires or how to encode this
<apritzel> and regulator constraints are pure software
<wens> apritzel: i don't think the firmware is that smart
<apritzel> that doesn't prevent you from having a kernel bug frying your board, isn't it?
<apritzel> wens: right, I was forgetting ;-)
<wens> that's why one has to use a proper dt that matches your board :)
<jjwerner> yikes, looks like i'm trying to bite something bigger than I can chew :)
<apritzel> jjwerner: no worries, we are here to help you
<KotCzarny> bite in smaller chunks?
<apritzel> jjwerner: can you try to find out:
<apritzel> a) how the BSP kernel talks to the arisc
<apritzel> (using some kind of msgbox, I guess)
<wens> the interface code should be somewhere like drivers/arisc/
<apritzel> b) what commands it sends to the arisc
<wens> apritzel: in armv7, the axp drivers from the bsp just read/write the AXP through arisc
<wens> arisc does not do anything fancy
<apritzel> now that's useful ;-)
p1u3sch1 has quit [Ping timeout: 240 seconds]
<jjwerner> yup, found that already actually following drivers/power/axp/axp-rw.c -> axp-rw.h -> include/linux/arisc/arisc.h -> drivers/arisc/arisc.c to guess what: arisc_sun50iw1p1.code binary file
<jjwerner> ...
<apritzel> but that should be the OpenRISC binary, which we don't need to know, right?
<jjwerner> i don't know yet ;)
<jjwerner> still learning here
p1u3sch1 has joined #linux-sunxi
<apritzel> I would imagine that a proper arisc implementation would provide a) naming abstraction, so you say USB1 instead of register 0x4738
MaDMaLKaV has quit [Read error: Connection reset by peer]
<apritzel> and b) constraints, preventing you from setting stupid voltages, for instance
<wens> apritzel: nope, doesn't have that
<apritzel> a) would allow to use the same driver for multiple SoCs
<wens> the drivers under drivers/power/axp/ read/write to the AXP using the functions in axp-rw.h
<apritzel> oh dear, how stupid
<wens> which indirectly accesses AXP through arisc
<apritzel> this renders the whole exercise kind of pointless
<jjwerner> ^^ wens, also i see constraints in the axp*.c files
<wens> apritzel: i assume by "proper implementation" you mean the firmware knows the layout of the board, like what regulator outputs supply what, and knows voltage constraints
<apritzel> exactly
<wens> do other vendors do that?
<apritzel> I know that ARM provides something along the lines, called SCP
<apritzel> but have actually no deeper insight into this
<wens> ah, i think it was just recently mainlined
cptG_ has joined #linux-sunxi
<wens> or in the process of
<apritzel> yes
<wens> was kind of wondering what it is (and all the other acronyms appearing on lakml)
<apritzel> but I have to check what it exactly covers
<apritzel> actually Allwinner uses the SCP term also (can find it in the bootlogs)
<wens> don't know about arm64, for arm it seems everyone has there own external PMICs, with regulator drivers
<wens> no smart firmware
MaDMaLKaV has joined #linux-sunxi
<wens> firmware seems to cover cpu up/down stuff mostly
<apritzel> indeed, at least for the embedded boards
<apritzel> but as ARM64 is also about servers, there is some hope that we can avoid this here
<wens> servers also would have pmics, no?
<apritzel> yes, but maybe not exposed to the OS
<apritzel> but instead handled by the firmware or a management processor
<wens> reminds me of x86-world BMCs....
<apritzel> exactly
cptG has quit [Ping timeout: 264 seconds]
* wens $works as a sw engineer / IT specialist
<jjwerner> apritzel: should I keep on barking on this tree or you have something else to do?
reinforce has quit [Quit: Leaving.]
<wens> apritzel: i actually dread some of those...
<apritzel> jjwerner: well, we need to find a solution to access the PMIC
<apritzel> I guess we need this for the Ethernet PHY
<wens> yup
<apritzel> looking at the LEDs on my board the firmware/u-boot does _not_ power it on
<wens> if we can get away with not running arisc, then it's just a matter of getting rsb working, and adding axp803 support to the axp mfd and regulator drivers
avph has quit [Ping timeout: 255 seconds]
<wens> if not, maybe an arisc-based rsb driver
<apritzel> yeah, it the arisc does only what you say, we could go that way
<wens> i think arisc is there for deep sleep modes, something we don't support in mainline
<apritzel> I guess we can avoid the arisc being loaded, but their u-boot depends on it
paulk-collins has joined #linux-sunxi
<apritzel> I replaced their DT last night, and u-boot stopped complaining about missing an arisc node
<wens> i.e. you power down the main cluster, and have arisc wait for interrupts on either the power button or something else to bring it back up
<apritzel> I thought that the PMIC alone can do that?
<apritzel> at least the power button case
<vishnu_> wens: for sdio wifi on, even after configuring PL3 as interrupts in brcm, it doesn't show under /proc/interrupt, is it expected ?
<vishnu_> I do not see irq error as before from brcm driver
avph has joined #linux-sunxi
nove has joined #linux-sunxi
<wens> vishnu_: it should be there
<wens> apritzel: yeah, but then you go through the whole boot process
<apritzel> good point
<wens> and pmic power on requires a longer press than just a click, which you normally do for phones/tablets
<apritzel> confirmed ;-)
<apritzel> just held down the power button for multiple seconds and it turned off
<apritzel> afk
<wens> that's the force power off feature :p
apritzel has quit [Ping timeout: 248 seconds]
ricardocrudo has joined #linux-sunxi
<wens> fyi arisc and scp is loaded in boot0, see u-boot-2014.07/sunxi_spl/boot0/main/boot0_main.c
<wens> and arch/arm/cpu/armv7/sun50iw1p1/spl/fip_common.c
vishnu_ has quit [Remote host closed the connection]
vishnu_ has joined #linux-sunxi
ricardocrudo has quit [Remote host closed the connection]
<wens> there's also complete source code for boot0, fes1 (boot0 for fel mode), and even secure brom
<vishnu_> :)
<vishnu_> dram init ?
<wens> not just dram init, everything
<wens> apritzel: (when you're back) could you post the boot logs again somewhere? preferably on the wiki
akaizen has joined #linux-sunxi
reinforce has joined #linux-sunxi
khuey|away is now known as khuey
domidumont has joined #linux-sunxi
orly_owl_ has quit [Quit: leaving]
orly_owl has joined #linux-sunxi
vishnup has quit [Quit: Connection closed for inactivity]
akaizen has quit [Remote host closed the connection]
iamfrankenstein1 has joined #linux-sunxi
mpmc- has joined #linux-sunxi
vbmithr has joined #linux-sunxi
oliv3r_ has joined #linux-sunxi
enrico__ has quit [Quit: Bye]
hehopmajieh__ has joined #linux-sunxi
Nacho_ has joined #linux-sunxi
dlan_ has joined #linux-sunxi
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
tkoskine_ has joined #linux-sunxi
lordlod_ has joined #linux-sunxi
andoma_ has joined #linux-sunxi
fl_0 has quit [Ping timeout: 240 seconds]
tgaz_ has joined #linux-sunxi
kelvan_ has joined #linux-sunxi
mcan- has joined #linux-sunxi
reinforce has quit [*.net *.split]
avph has quit [*.net *.split]
kelvan has quit [*.net *.split]
tgaz has quit [*.net *.split]
tkoskine has quit [*.net *.split]
vbmithr_ has quit [*.net *.split]
swabbles has quit [*.net *.split]
GeneralS1upid has quit [*.net *.split]
Nacho__ has quit [*.net *.split]
iamfrankenstein has quit [*.net *.split]
oliv3r has quit [*.net *.split]
HeHoPMaJIeH has quit [*.net *.split]
mpmc_ has quit [*.net *.split]
VargaD has quit [*.net *.split]
focus has quit [*.net *.split]
RzR has quit [*.net *.split]
andoma has quit [*.net *.split]
fest has quit [*.net *.split]
Rondom has quit [*.net *.split]
dlan has quit [*.net *.split]
lordlod has quit [*.net *.split]
mcan_ has quit [*.net *.split]
tgaz_ is now known as tgaz
iamfrankenstein1 is now known as iamfrankenstein
GeneralStupid has joined #linux-sunxi
ZrZ has joined #linux-sunxi
VargaD has joined #linux-sunxi
vagrantc has joined #linux-sunxi
fest has joined #linux-sunxi
avph has joined #linux-sunxi
Rondom has joined #linux-sunxi
reinforce has joined #linux-sunxi
vagrantc has quit [Changing host]
vagrantc has joined #linux-sunxi
SJRvanSchaik has joined #linux-sunxi
fl_0 has joined #linux-sunxi
focus has joined #linux-sunxi
ZrZ is now known as rZr
vagrantc has quit [*.net *.split]
fest has quit [*.net *.split]
reinforce has quit [*.net *.split]
avph has quit [*.net *.split]
Rondom has quit [*.net *.split]
igraltist has quit [Remote host closed the connection]
<jjwerner> or are you asking about something else?
vagrantc has joined #linux-sunxi
fest has joined #linux-sunxi
avph has joined #linux-sunxi
Rondom has joined #linux-sunxi
reinforce has joined #linux-sunxi
matthias_bgg has left #linux-sunxi ["Leaving"]
_massi has quit [Remote host closed the connection]
Netlynx has joined #linux-sunxi
ssvb has joined #linux-sunxi
vagrantc has quit [*.net *.split]
fest has quit [*.net *.split]
reinforce has quit [*.net *.split]
avph has quit [*.net *.split]
Rondom has quit [*.net *.split]
vagrantc has joined #linux-sunxi
fest has joined #linux-sunxi
avph has joined #linux-sunxi
Rondom has joined #linux-sunxi
reinforce has joined #linux-sunxi
jjwerner has quit [Changing host]
jjwerner has joined #linux-sunxi
jelly-home is now known as jelly
vagrantc has quit [*.net *.split]
fest has quit [*.net *.split]
reinforce has quit [*.net *.split]
avph has quit [*.net *.split]
Rondom has quit [*.net *.split]
khuey is now known as khuey|away
<ssvb> wens: isn't the a64 dram init code currently provided in binary only libdram library?
<ssvb> wens: there is a subdirectory with a collection of dram init sources for various SoC types, but none of them seems to match the code used in the boot0 blob
vishnup has joined #linux-sunxi
vishnu_ has quit [Ping timeout: 260 seconds]
<ssvb> my pine64 board is already somewhere in Helsinki, according to the tracking system
<ssvb> maybe I will get it in a few days :-)
khuey|away is now known as khuey
<oliv3r_> ssvb: nice one!
fl_0 has quit [Ping timeout: 260 seconds]
fl_0 has joined #linux-sunxi
afaerber has quit [Quit: Ex-Chat]
IgorPec has quit [Quit: Nettalk6 - www.ntalk.de]
akaizen has joined #linux-sunxi
khuey is now known as khuey|away
mosterta has joined #linux-sunxi
domidumont has quit [Ping timeout: 246 seconds]
interrobangd has joined #linux-sunxi
mnr has joined #linux-sunxi
premoboss has joined #linux-sunxi
vagrantc has joined #linux-sunxi
avph has joined #linux-sunxi
Rondom has joined #linux-sunxi
fest has joined #linux-sunxi
reinforce has joined #linux-sunxi
gaby_ has joined #linux-sunxi
interrobangd has quit [*.net *.split]
mnr has quit [*.net *.split]
premoboss has quit [*.net *.split]
vagrantc has quit [*.net *.split]
fest has quit [*.net *.split]
reinforce has quit [*.net *.split]
avph has quit [*.net *.split]
Rondom has quit [*.net *.split]
vishnup has quit [Quit: Leaving]
andoma_ is now known as andoma
<adj_> ssvb, who cares about dram initialization!
<adj_> totally unimportant
interrobangd has joined #linux-sunxi
mnr has joined #linux-sunxi
premoboss has joined #linux-sunxi
vagrantc has joined #linux-sunxi
avph has joined #linux-sunxi
Rondom has joined #linux-sunxi
fest has joined #linux-sunxi
reinforce has joined #linux-sunxi
<ssvb> adj_: well, some people would prefer having completely open source software running on the board, without any mandatory blobs
<adj_> I tried to be sarcastic
<adj_> I don't really understand the interest on supporting every soc developed by allwinner
<adj_> we already have more or less good support for A10, A13 and A20
<jelle> but you want support for intel's latest proc too right? ;)
<adj_> intel and amd cpus are in other league in performance and capabilities
<KotCzarny> adj_: but that's only thanks to the sunxi guys doing work
<KotCzarny> without, imagine having only a10
<KotCzarny> and even that only partially
<adj_> that's what I tried to say
afaerber has joined #linux-sunxi
JohnDoe_71Rus has quit [Quit: KVIrc KVIrc Equilibrium 4.2.0, revision: 42021, sources date: 20120701, built on: 2013-10-21 12:25:22 UTC 42021 http://www.kvirc.net/]
<adj_> there are newer socs, some buggy, other without many capabilities of a20, or with even more secret gpus
<adj_> and performance of that newer socs are not much greater of a20
<KotCzarny> well, a64 promises quite a bigger jump
<adj_> and of course at least an order of magnitude slower than intel or amd
interrobangd has quit [*.net *.split]
mnr has quit [*.net *.split]
premoboss has quit [*.net *.split]
vagrantc has quit [*.net *.split]
fest has quit [*.net *.split]
reinforce has quit [*.net *.split]
avph has quit [*.net *.split]
Rondom has quit [*.net *.split]
<adj_> a64 memory bandwidth is low
<adj_> and max memory also low afaik
<adj_> any benchmark showing big speed advantage?
<KotCzarny> not yet, i guess they dont have OS written yet, hehe
apritzel has joined #linux-sunxi
<adj_> don't get my words wrong, I really appreciate everyone that has supported allwinner socs
<adj_> and i use it regularly
<adj_> and everyone works on whatever they want, but it's really unbelievable that even something as important as dram initialization is not supported
<adj_> for a company, allwinner, that lives thanks to open source software running on their hardware
<KotCzarny> adj: nope, they are tablet company
<adj_> with their own propietary operanting system?
<KotCzarny> s/company/chips/
<adj_> i thought that that tablets run mostly linux and other open source software
<KotCzarny> one can hardly call 'droid open source
<adj_> why?
<KotCzarny> most apps are binary
<adj_> are you arguing that android is mostly proprietary software?
khuey|away is now known as khuey
<jelle> aren't most drivers blobs too?
<KotCzarny> apart from os apps, how many of the rest you have access to their source?
<adj_> blobs are a small part of an android system
<KotCzarny> small but strategically placed control points
<jelle> iirc nexus 5x even has some blob drivers
<adj_> at least nexus 4 has many more blobs than most devices from allwinner or samsung
<adj_> google devices are not specially open source friendly
akaizen has quit [Remote host closed the connection]
mnr has joined #linux-sunxi
Netlynx has quit [Quit: Leaving]
akaizen has joined #linux-sunxi
jjwerner has quit [Ping timeout: 252 seconds]
mnr has quit [Quit: leaving]
vagrantc has joined #linux-sunxi
fest has joined #linux-sunxi
premoboss has joined #linux-sunxi
Rondom has joined #linux-sunxi
interrobangd has joined #linux-sunxi
yann|work has joined #linux-sunxi
p1u3sch1 has quit [Ping timeout: 240 seconds]
p1u3sch1 has joined #linux-sunxi
Wizzup_ has quit [Quit: Reconnecting]
Wizzup has joined #linux-sunxi
avph has joined #linux-sunxi
yann|work has quit [Ping timeout: 260 seconds]
Worf has joined #linux-sunxi
azend|vps has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
soderstrom has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
azend|vps has joined #linux-sunxi
avph has quit [*.net *.split]
avph has joined #linux-sunxi
akaizen has joined #linux-sunxi
avph has quit [*.net *.split]
iamfrankenstein has quit [Quit: iamfrankenstein]
nove has quit [Quit: nove]
dlan_ is now known as dlan
<ssvb> adj_: about "a64 memory bandwidth is low" - how did you measure it?
paulk-collins has quit [Remote host closed the connection]
<ssvb> adj_: with 672 MHz DRAM clock speed, it should be in fact pretty good
<ssvb> adj_: not top notch of course, but don't forget that we have relatively slow in-order Cortex-A53 cores running at a relatively low clock speed
tgaz has quit [Quit: Over and out]
soderstrom has quit [Ping timeout: 256 seconds]
tgaz has joined #linux-sunxi
leio has quit [Read error: Connection reset by peer]
akaizen has quit [Remote host closed the connection]
<adj_> ssvb, you know more about allwinner dram bandwidth than i do, but 672MHz means only about 20-25% more bandwidth than a20
khuey is now known as khuey|away
<apritzel> adj_: as you were complaining about DRAM init not being supported by Allwinner:
<apritzel> in fact it is supported, in their boot0
<apritzel> and it shouldn't really be the issue of Linux or some community hackers to do this
<apritzel> the only problem is:
<apritzel> their firmware is meant to just run Android, and they don't care about providing a clean interface to the OS
<apritzel> as they are a tablet company, the OS comes with the device
<apritzel> big difference to PCs
<adj_> yes?
<adj_> sure?
<apritzel> no, the customer is choosing it
<apritzel> yes
<adj_> in PCs the OS also comes with the device
topi` has quit [Ping timeout: 260 seconds]
<apritzel> I can put an Ubuntu CD in any PC and off it goes
topi` has joined #linux-sunxi
<apritzel> I am not talking about that Windows being shipped by default
<apritzel> no, it's an embedded device
<adj_> how many PCs in the world comes with a OS selected by the user?
<apritzel> it's about the general idea that it's a computer
<apritzel> where software is bundled with it
<adj_> a tablet is a computer
<adj_> also is in a PC
<apritzel> you are not supposed to put anything else on it
<adj_> you are wrong
<Worf> just because they choose poorly ... also enough come without any software bundled
<adj_> apritzel, do you mean that software can't be installed on tablets?
<apritzel> it's not meant to be (apart from apps, but that's userland)
<adj_> every android device that i have own allows me to write my own programs and installed on it
<adj_> apps are computer programs
<adj_> just marketing name
<adj_> most PCs are not intended for the OS to be replaced by the user
<apritzel> but they have a very clear interface to do so
<apritzel> standardized and protected
<adj_> and not even guaranteed to be compatible with next versions of the same OS
<apritzel> you can boot them from USB, CD, hard disk
<adj_> just wait a little bit
<apritzel> BIOS? it's the same for more than 30 years now
<ssvb> apritzel: have you found the a64 dram init sources in the SDK?
<apritzel> no, but haven't looked too deeply
<apritzel> I just mananged to convince the Pine U-boot to load my DTB directly from the SD
<apritzel> as they ignore the fdt addr
<apritzel> and instead always use the hardcoded DT they link with u-boot
<apritzel> so you cannot just load a new DT somewhere, it will always pass the internal address to the kernel
<ssvb> well, we'll try to do something about the bootloader eventually
<apritzel> great, isn't it?
<apritzel> so I just overwrite it ;-)
<apritzel> but their DT is pretty big (68K)
naobsd has quit [Quit: naobsd]
<apritzel> sure
<apritzel> but let's first get the Linux support rolling
<apritzel> I am about to finish an image file
<ssvb> yes, it's good that the current Allwinner's bootloader is usable for kernel development
<apritzel> next step would be to replace U-Boot
<apritzel> because the one on there is so crippled
<apritzel> but I managed to use it anyway: I can create my own "partition table"
<apritzel> which is actually their NAND partitioning scheme
<adj_> apritzel, the IBM PC's BIOS was not intended to be a standard interface for other OS to boot
<apritzel> located at exactly 20MB into the SD card
<adj_> it as a proprietary software intended to boot PC DOS
<adj_> it became an standard because reverse engineering
<adj_> just the same as the community does this days with tablets
<adj_> and even then the IBM PC was a personal computer, even with an undocumented boot process
<apritzel> adj_: I guess we could argue about that for ages ;-)
<apritzel> what I actually wanted to say initially is that if Allwinner wouldn't provided such a crippled u-boot we would be pretty happy already
<adj_> yes, because you have been convinced by marketing shit about tablets not being computers, even if the first tablets where regular PCs
<apritzel> as said I am fine with Allwinner providing a binary blob to initialise their hardware intially
<adj_> yes, that's the same that PCs do
<adj_> and you don't say they are embedded devices
<apritzel> adj_: I know that they are computers, just tell that the other 90% of the people
<adj_> so the excuse, not a computer, but embedded device is non sense
<apritzel> it's not an excuse, I don't like it too, i'ts just their point of view
<apritzel> they create the OS that runs with their hardware, and the OS is part of the product
<adj_> they don't create the OS that comes with the hardware
<apritzel> that's embedded
<ssvb> apritzel: a binary blob means that we can't have mainline U-Boot, which can be just configured, compiled and work out of the box
<apritzel> ssvb: I just meant keeping boot0, which does the DRAM init, loads u-boot and starts it
<ssvb> combining free software with blobs causes some practical inconveniences, is it not just an "ideological" thing
<apritzel> I know, and I am not arguing this
<apritzel> I am just pragmatical here
<apritzel> replacing boot0 sounds a lot more complicated than replacing u-boot
<apritzel> and u-boot is the only thing I really care about
<apritzel> in the Pine64 case we just to deal with the arisc still
<ssvb> having to rely on a collection of board-specific boot0 blobs is inconvenient and error prone
<apritzel> sure, in the long run I agree
<apritzel> I'd rather just run Linux now, and then start freeing the system afterwards
<adj_> that SOCs initially run embedded ROM is inevitable, but that after that basic hardware configuration like dram are done by blobs is very inconvenient at least
<apritzel> adj_: I agree
akaizen has joined #linux-sunxi
<apritzel> it should be done in ROM, but that's quite risky
<apritzel> I can see that they don't want to do that
<ssvb> DRAM controller needs board specific configuration (delays and impedance)
<apritzel> yeah, and we shouldn't be bothered with this, frankly
<apritzel> the board vendor should work it out and ship it with the board
<ssvb> the big PC computer cave more advanced DRAM controllers which can autoconfigure themselves, and also the DIMM modules have SPD (some configuration data stored in EEPROM)
<apritzel> exactly, DRAM training is quite rocket science, and I am glad that the BIOS does it for me
akaizen has quit [Remote host closed the connection]
<apritzel> that's what I mean: the BIOS does all the board specific stuff and presents me a usable system
<apritzel> and the BIOS in on the board
<ssvb> the point is that ARM boards are cheap, that's why the initialization is a bit more complex and needs some "magic" precomputed board specific settings
<apritzel> if those ARM boards would have just enough flash storage to hold the DT and the initial boot loader ...
<apritzel> ideally there would be some flash in the SoC
<ssvb> well, DT is not quite stable and keeps changing
<ssvb> IMHO a better solution is to store just the DT file name in flash or EEPROM
<ssvb> then the DT file can be read from the SD card or other boot media
<ssvb> in fact, this is more or less how the sunxi boards are booting now
<apritzel> if the on-SoC storage would be flash, the DT could be upgraded like with a BIOS update
<ssvb> currently there is a board specific U-Boot, and this board specific U-Boot knows the DTB file name and exports it as an environment variable
<ssvb> if we had a stable DT format, then sure
<apritzel> I am very much for stable DTs
<apritzel> we should move the boot/dts directory out of the kernel
<apritzel> and really accept the DTs as hardware description
<apritzel> or even better: we should make all the peripherals self discoverable
<apritzel> well, "we" can't do this, but the SoC vendors could
<ssvb> this increases hardware cost
<ssvb> would it be possible to do this and still sell boards with a $15 price tag?
<apritzel> well, if every IP block would have something like vendor/device ID as the first registers, that would help already a lot
<apritzel> we could scan the device range
<ssvb> wasn't ARM trying to advocate something like this and failed?
<apritzel> the IOMEM range for any device is so huge compared to what they actually use, just one R/O register with an ID shouldn't really matter
<apritzel> because ARM has no real power to advocate something
<ssvb> yes, a lot of hardware vendors need to discuss this and come to an agreement
<apritzel> I see that this will not happen ;-)
<ssvb> exactly
<apritzel> alone the discussion phase would take ages ...
<apritzel> ssvb: where could I eventually load an image to?
<ssvb> apritzel: I don't know, maybe something like google drive?
viccuad has quit [Quit: WeeChat 1.3]
<apritzel> sounds like an idea, but I usually try to avoid Google services
<apritzel> do you know how dl.linux-sunxi.org works?