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
apritzel has quit [Ping timeout: 244 seconds]
iaglium has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
popolon has quit [Quit: WeeChat 1.3]
lynxis has quit [Quit: beaming to the mars]
Nacho has quit [Ping timeout: 272 seconds]
lynxis has joined #linux-sunxi
Da_Coynul has joined #linux-sunxi
Nacho has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
ninolein has quit [Ping timeout: 258 seconds]
ninolein has joined #linux-sunxi
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
kaspter has joined #linux-sunxi
cnxsoft has joined #linux-sunxi
<MoeIcenowy> mripard: I tried to use sun4i-ts on A33, but failed
<MoeIcenowy> and when will your I2S driver get merged?
<MoeIcenowy> (It seems that A33/A64's audiocodec used an internal I2S
<lennyraposo> quick question
<lennyraposo> has anyone used gcc 6 as of yet?
<MoeIcenowy> lennyraposo: on what purpose?
<MoeIcenowy> our distro is using gcc 6 for a few weeks
<lennyraposo> for compiling building kernel
<MoeIcenowy> (oh our cross toolchain is still 4.9
<MoeIcenowy> but our x86_64 kernel is built with gcc6
<lennyraposo> gotcha
<MoeIcenowy> Linux version 4.6.2-aosc-main (root@general) (gcc version 6.1.0 20160427 (AOSC OS, Core) (GCC) ) #1 SMP PREEMPT Mon Jun 13 13:36:52 CST 2016
<lennyraposo> kewl and the gang :)
<lennyraposo> well that answers my question for cisc
<lennyraposo> ahev you attempted on any arm builds?
<lennyraposo> just wondering
<MoeIcenowy> lennyraposo: maybe I should try it this afternoon (Asia/Hong_Kong)
<wens> lennyraposo: you can ask hans
<wens> lennyraposo: seems fedora uses the latest GCC, and they hit a bug with my uboot work :(
<MoeIcenowy> wens: :-)
<MoeIcenowy> wens: is there any newer version of axp209 battery driver patch?
<MoeIcenowy> I want make one for axp22x
<lennyraposo> thanks Moelcenowy
<wens> not that i'm aware of
<MoeIcenowy> wens: ah-oh
<MoeIcenowy> laga and bonbons are all not here...
Gerwin_J has quit [Quit: Gerwin_J]
kaspter has quit [Remote host closed the connection]
kaspter has joined #linux-sunxi
lennyraposo1 has joined #linux-sunxi
lennyraposo has quit [Read error: Connection reset by peer]
lennyraposo1 has quit [Ping timeout: 258 seconds]
kaspter has quit [Remote host closed the connection]
lennyraposo has joined #linux-sunxi
lennyraposo has quit [Client Quit]
lennyraposo has joined #linux-sunxi
fire2191 has quit [Read error: Connection reset by peer]
lennyraposo has quit [Client Quit]
lennyraposo has joined #linux-sunxi
tithrion_ has quit [Ping timeout: 240 seconds]
lennyraposo has quit [Quit: Leaving.]
lennyraposo has joined #linux-sunxi
_stephan has joined #linux-sunxi
<_stephan> hi. I got that PE0 pin bank running applying wens hint (LDO3 hat to be set to 2.8V). I have another problem...
<_stephan> I try to set up PE0 pin to be pull down for use as sysfs gpio. So I put this in the DTS: http://pastebin.com/7i47hdLU - but the pin is still floating
<_stephan> (so, with no voltage applied, it randomly has 0 or 1 in /sys/class/gpio/gpio128/value)
<_stephan> am I missing some detail in the DTS?
<_stephan> kernel version is 4.6.3
<wens> how did you set pull down?
<_stephan> wens http://pastebin.com/7i47hdLU here I pasted the block
tithrion_ has joined #linux-sunxi
<wens> for the pinmux to be applied, it has to be tied to a device
<wens> just adding a node under &pio does nothing
<_stephan> ah, thank you. do you have any advice, which device to tie it to, to make it usable through sysfs?
<_stephan> it's only used in userspace...
<wens> if you're just testing stuff yourself, you could tie it to any device that's enabled
<wens> for formal stuff that will be submitted, i don't know
<_stephan> okay. thanks you very much. for both (the regulator stuff and this :))
<_stephan> maybe I just use it in the pio device itself... that might be the cleanest thing to do.
TheSeven has quit [Ping timeout: 272 seconds]
[7] has joined #linux-sunxi
kaspter has joined #linux-sunxi
IgorPec has joined #linux-sunxi
fredy has quit [Excess Flood]
IgorPec has quit [Ping timeout: 244 seconds]
fredy has joined #linux-sunxi
Mr__Anderson has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
IgorPec10 has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
tithrion_ has quit [Remote host closed the connection]
tithrion_ has joined #linux-sunxi
IgorPec10 has quit [Ping timeout: 258 seconds]
jernej has quit [Ping timeout: 240 seconds]
IgorPec has joined #linux-sunxi
vagrantc has joined #linux-sunxi
dearfibonacci has joined #linux-sunxi
<_stephan> wens, thank you, it works fine now.
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
Nyuutwo has quit [Ping timeout: 240 seconds]
Nyuutwo has joined #linux-sunxi
ricardocrudo has joined #linux-sunxi
Mr__Anderson has quit [Remote host closed the connection]
apritzel has joined #linux-sunxi
apritzel has quit [Ping timeout: 244 seconds]
ricardocrudo has quit [Remote host closed the connection]
yann|work has joined #linux-sunxi
phipli has joined #linux-sunxi
Amit_T has joined #linux-sunxi
phipli has quit [Ping timeout: 244 seconds]
vagrantc has quit [Quit: leaving]
afaerber has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
sW`` is now known as sW`
dearfibonacci has quit [Quit: Leaving]
dearfibonacci has joined #linux-sunxi
JohnDoe_71Rus has quit [Read error: Connection reset by peer]
JohnDoe_71Rus has joined #linux-sunxi
popolon has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
<speakman> wens: Thank you. Although I've been through the Input section in kernel config many times, I still missed the CONFIG_EVDEV. Woops. :)
dearfibonacci has quit [Read error: Connection reset by peer]
dearfibonacci has joined #linux-sunxi
dearfibonacci has quit [Read error: Connection reset by peer]
dearfibonacci has joined #linux-sunxi
mnr has joined #linux-sunxi
Mr__Anderson has joined #linux-sunxi
<speakman> I still don't see how I can set LCD parameters to enable LVDS output.
premoboss has joined #linux-sunxi
lemonzest has joined #linux-sunxi
<speakman> Is it up to U-boot to initialize the frame buffer?
<MoeIcenowy> speakman: which chip
<speakman> MoeIcenowy: A20
<MoeIcenowy> I have used mainline linux on a Viewsonic 133Q Tablet
<NiteHawk> speakman: for "simplefb" yes
<MoeIcenowy> which has a 13.3" LVDS
<MoeIcenowy> you can easily configure a LVDS in u-boot config
<MoeIcenowy> CONFIG_VIDEO_LCD_IF_LVDS=y
<MoeIcenowy> CONFIG_VIDEO_LCD_PANEL_LVDS=y
<MoeIcenowy> Then write CONFIG_VIDEO_LCD_MODE according to script.fex
<speakman> MoeIcenowy: Thanks! Sounds great :)
<speakman> Do you guys know if this can be altered within u-boot environment variables?
<speakman> I have a couple of different displays I need to use
<MoeIcenowy> speakman: Please prepare different u-boot
pekka10 has quit [Ping timeout: 260 seconds]
<wens> speakman: no, these are compile time settings
IgorPec10 has joined #linux-sunxi
pekka10 has joined #linux-sunxi
<speakman> Ok, I have to figure something out then. :)
tithrion_ has quit [Ping timeout: 240 seconds]
tithrion_ has joined #linux-sunxi
enrico_ has joined #linux-sunxi
IgorPec11 has joined #linux-sunxi
IgorPec10 has quit [Ping timeout: 252 seconds]
IgorPec11 has quit [Quit: Nettalk6 - www.ntalk.de]
IgorPec11 has joined #linux-sunxi
IgorPec11 has quit [Client Quit]
Nyuutwo has quit [Ping timeout: 240 seconds]
The_Loko has joined #linux-sunxi
Nyuutwo has joined #linux-sunxi
massi has joined #linux-sunxi
kelvan has quit [Remote host closed the connection]
Da_Coynul has joined #linux-sunxi
premoboss has quit [Quit: Sto andando via]
kelvan has joined #linux-sunxi
normalform has joined #linux-sunxi
normalform has quit [Ping timeout: 252 seconds]
iaglium has quit [Quit: Bed Time]
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
arossdotme has joined #linux-sunxi
arossdotme-planb has quit [Ping timeout: 246 seconds]
ricardocrudo has joined #linux-sunxi
kaspter has quit [Ping timeout: 246 seconds]
enrico_ has quit [Quit: Bye]
mnemoc has quit [Read error: Connection reset by peer]
reinforce has joined #linux-sunxi
Amit_T has quit [Ping timeout: 240 seconds]
tithrion_ has quit [Ping timeout: 240 seconds]
<oliv3r> how do i interrupt the default sunxi bootloader, I recall there was some secret key combo to stop it
<speakman> now I have my /dev/input/ directory with eventX device files. But that makes me thing; how are things supposed to use input devices if not through /dev/input/eventX ?
<NiteHawk> oliv3r: usually you have a BOOTDELAY default of 2 seconds - if you press a key / send any character over the UART console, the autoboot will stop
<oliv3r> your speaking of u-boot :)
<oliv3r> i'm talking about boot0/boot1
<oliv3r> old skool crappy allwinnner stuff
<oliv3r> :)
<oliv3r> boot1 actually, boot0 is still functional
<NiteHawk> ah, okay :)
<NiteHawk> wasn't that '0' or '1' for boot0?
<oliv3r> that's what i thought
<oliv3r> but can't find it on our wiki anymore
<oliv3r> i'm smashing my head that I didn't document it when i wrote the piece about the BROM :)
<mripard> 2, on the uart iirc
Amit_T has joined #linux-sunxi
tithrion_ has joined #linux-sunxi
<NiteHawk> oliv3r: we mention it on the FEL page: http://linux-sunxi.org/FEL#Through_serial_console
IgorPec has quit [Ping timeout: 246 seconds]
<MoeIcenowy> mripard: what's your I2S driver based?
<mripard> what do you mean?
<KotCzarny> manual or re or other driver
<MoeIcenowy> I2S driver (Maxime Ripard (mripard)) patch-v2
<MoeIcenowy> what kernel version is it based?
<montjoie> jmcneill: thanks for the BIT(2), I confirm the speed gain, at least on 100Mbit, will test Giga later
<mripard> 4,7
<MoeIcenowy> mripard: thanks
<montjoie> jmcneill: for 100Mbit it made transmi from 84 to 94Mbit/s
<MoeIcenowy> and should I split my sun8iw3-ths patch set to pll2 patch and ths patches?
<MoeIcenowy> (I want to now make an A33 audio codec driver
Gerwin_J has quit [Quit: Gerwin_J]
vagrantc has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
Gerwin_J has quit [Client Quit]
fire219 has joined #linux-sunxi
arossdotme-planb has joined #linux-sunxi
arossdotme has quit [Ping timeout: 250 seconds]
matthias_bgg has quit [Ping timeout: 272 seconds]
arossdotme has joined #linux-sunxi
arossdotme-planb has quit [Ping timeout: 264 seconds]
_stephan has quit [Quit: Ex-Chat]
<jmcneill> montjoie: great!
matthias_bgg has joined #linux-sunxi
arossdotme-planb has joined #linux-sunxi
arossdotme has quit [Ping timeout: 258 seconds]
<wens> oliv3r: 2 for fel mode, s to drop to a shell
Amit_T has quit [Ping timeout: 240 seconds]
arossdotme has joined #linux-sunxi
<oliv3r> s ah of course :)
BenG83 has joined #linux-sunxi
<oliv3r> well it turned out the power button was damaged :p
<oliv3r> so the fix was some soldering tin
arossdotme-planb has quit [Ping timeout: 276 seconds]
<BenG83> \o
<oliv3r> poop; i'm trying to boot the SPL on an A10s, but it appears that support is not yet perfect here; panic("Could not determine boot source\n");
<oliv3r> which is supprising to me actually
<oliv3r> spl can be read successfully from mmc0, so why not the full u-boot ...
<jmcneill> montjoie: I implemented polling last night and it didn't make any measurable improvements in rx performance :/
arossdotme-planb has joined #linux-sunxi
<montjoie> jmcneill: what's your performance for 100 and 1G ?
<jmcneill> haven't tried 100, only 1G on bpi-m3
<jmcneill> close to 400tx, 450rx
<jmcneill> using iperf3 -Z, tcp with no special tuning
arossdotme has quit [Ping timeout: 240 seconds]
<montjoie> I will try after work with your tip, but without polling I was stuck with 300/400
<jmcneill> are you using that tx interrupt mitigation trick i mentioned?
<montjoie> no
<jmcneill> that should help too
<montjoie> certainly, I await for test it, but my bpim3 is down
<montjoie> need to finish my mini datacenter with relay for powering up
<jmcneill> eager to hear your results
<jonkerj> I'm thinking about a similar rig :-)
<jonkerj> was thinking about combining an USB relay, some 4.0/1.7mm barrel plugs and a 5V power brick for my orange pi testing rig
pietrushnic`away has joined #linux-sunxi
pietrushnic`away is now known as pietrushnic
jstein_ has joined #linux-sunxi
jstein_ is now known as jstein
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
premoboss has joined #linux-sunxi
Amit_T has joined #linux-sunxi
arossdotme-planb has quit [Ping timeout: 276 seconds]
arossdotme-planb has joined #linux-sunxi
<MoeIcenowy> mripard: sun8iw3 thermal sensor seems to be working okay after replacing pll2 to a dummy pll
IgorPec11 has joined #linux-sunxi
dearfibonacci has quit [Quit: Leaving]
enrico_ has joined #linux-sunxi
IgorPec11 has quit [Ping timeout: 272 seconds]
cnxsoft has quit [Quit: cnxsoft]
IgorPec11 has joined #linux-sunxi
scream has joined #linux-sunxi
<mripard> MoeIcenowy: please don't rush things this way
<MoeIcenowy> mripard: ?
<MoeIcenowy> what?
<MoeIcenowy> (I cannot understand this
<MoeIcenowy> (I know little about kernel developing...
<mripard> you waited less than 10 minutes between a mail asking a question
premoboss has quit [Quit: Sto andando via]
<mripard> and a new version of patches
<mripard> which doesn't even address anything, since you didn't have any review on it
<MoeIcenowy> oh sorry
enrico_ has quit [Quit: Bye]
afaerber has quit [Quit: Ex-Chat]
IgorPec11 has quit [Ping timeout: 260 seconds]
<mripard> MoeIcenowy: usually, wait for a few days since the last review before sending a new version
<mripard> and if you got no reply for like 10 days, you can resend it
IgorPec11 has joined #linux-sunxi
IgorPec11 has quit [Client Quit]
afaerber has joined #linux-sunxi
<lvrp16> is it possible for 4k resolution on the orange pi?
nove has joined #linux-sunxi
afaerber has quit [Ping timeout: 250 seconds]
afaerber has joined #linux-sunxi
The_Loko has quit [Ping timeout: 258 seconds]
apritzel has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
zuikis has joined #linux-sunxi
The_Loko has joined #linux-sunxi
vagrantc has quit [Quit: leaving]
Gerwin_J has joined #linux-sunxi
vagrantc has joined #linux-sunxi
apritzel has quit [Ping timeout: 244 seconds]
jernej has joined #linux-sunxi
cptG has joined #linux-sunxi
cptG_ has quit [Ping timeout: 272 seconds]
phipli has joined #linux-sunxi
vagrantc has quit [Quit: leaving]
pekka10 has quit [Ping timeout: 264 seconds]
massi has quit [Quit: Leaving]
Gerwin_J has quit [Quit: Gerwin_J]
BenG83 has quit [Ping timeout: 258 seconds]
matthias_bgg has quit [Quit: Leaving]
apritzel has joined #linux-sunxi
popolon has quit [Quit: WeeChat 1.3]
<Wizzup> ssvb: also works with the winbond, making a more clear picture now
apritzel has quit [Ping timeout: 244 seconds]
tithrion_ has quit [Ping timeout: 276 seconds]
IgorPec has joined #linux-sunxi
BenG83 has joined #linux-sunxi
vagrantc has joined #linux-sunxi
Ultrasauce has joined #linux-sunxi
Mr__Anderson has quit [Remote host closed the connection]
ricardocrudo has quit [Read error: Connection reset by peer]
ricardocrudo_ has joined #linux-sunxi
BenG83 has quit [Quit: Leaving]
yann|work has quit [Ping timeout: 240 seconds]
apritzel has joined #linux-sunxi
ricardocrudo_ has quit [Remote host closed the connection]
arossdotme has joined #linux-sunxi
vagrantc has quit [Quit: leaving]
iamfrankenstein has quit [Quit: iamfrankenstein]
arossdotme has quit [Quit: Ex-Chat]
arossdotme-planb has quit [Quit: Ex-Chat]
arossdotme has joined #linux-sunxi
yann|work has joined #linux-sunxi
staplr has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
vagrantc has joined #linux-sunxi
staplr has quit [Ping timeout: 252 seconds]
staplr has joined #linux-sunxi
plaes has quit [Ping timeout: 244 seconds]
plaes has joined #linux-sunxi
staplr has quit [Ping timeout: 258 seconds]
<montjoie> jmcneill: I wrote performance numbers on http://linux-sunxi.org/Sun8i_emac, but I got strange underperformance for RX
<jmcneill> I made some more adjustments, now I get 455tx/455rx
<jmcneill> without polling
<jmcneill> default ahb2 clock source is AHB1 clock, which is 100MHz
<jmcneill> changing to PLL_PERIPH/2 gives 300MHz and takes me from ~400/400 to 450/450
The_Loko has quit [Quit: Leaving]
<jmcneill> oh and I guess default U-Boot CPU frequency is 984MHz, running at 1.2GHz gives me 500Mbps
<jmcneill> 1.8GHz -> 590Mbps
<montjoie> For the moment I try to not overclock, since thermal control is not ready, but yeah perf could be far better
Mr__Anderson has joined #linux-sunxi
<montjoie> but I do not understand, last time I got 700Mbits/s for RX
<jmcneill> I just wanted to rule out ahb as the bottleneck
<jmcneill> before changing AHB2 source I wasn't maxing out the CPU
<jmcneill> Not mentioned in the A83T user manual, but the A64 user manual says for AHB2 "Its default value is 300Mhz."
scream has quit [Remote host closed the connection]
<ssvb> jmcneill: what kind of AHB2 clock setup is used by the kernel from the Allwinner's SDK on A83T?
<jmcneill> not sure
<ssvb> IMHO it would be a bit more reliable than just pulling numbers from the manual for a different SoC and hoping for the best
<jmcneill> Sure. There is no divisor and only two clock source options for AHB2 (on both SoCs)
<jmcneill> clk = clk_register_fixed_factor(NULL, "ahb2", "pll_periph", 0, 1, 2);
<jmcneill> I'm not familiar with Linux but does that mean it uses pll_periph for parent of ahb2?
orly_owl_ has joined #linux-sunxi
jstein_ has joined #linux-sunxi
jstein is now known as Guest85790
vagrantc has quit [Quit: leaving]
orly_owl has quit [Ping timeout: 246 seconds]
jstein_ is now known as jstein
<apritzel> ssvb, agraf: I think pointing to the SRAM C usage patch in U-Boot as the culprit for breakage on the Pine64 was a bit premature
Guest85790 has quit [Ping timeout: 252 seconds]
<apritzel> I bisected again, and in contrast to last time, where it pointed me to the SRAM C patch, it now tells me that TINY_PRINTF is to blame
<apritzel> and indeed removing it from the Kconfig fixes it for me
<apritzel> that seems to affect the SPL only
<apritzel> so I guess the SRAM C was just helping because my old SPL was just not compatible with that change
<apritzel> (you have to compile both the SPL and the proper U-Boot either with or without the patch)
<ssvb> apritzel: was it a mistake when doing bisecting? or does reverting the SRAM C usage patch instead of TINY_PRINTF also help?
<apritzel> ssvb: I am also wondering about that
<apritzel> this time I bisected over the SPL
<apritzel> last time over the real U-Boot
<ssvb> could you please confirm it?
<ssvb> SRAM C is a weird thing
<apritzel> so I guess last time I just used the _old_ SPL (before the SRAM C patch), which was using the old addresses
<apritzel> and AFAICT this change has to be on both SPL and proper U-Boot, right?
<apritzel> also I think agraf was using the SPL from my old branch yesterda
<apritzel> y
<ssvb> I think that it is only expected to affect the SPL
<apritzel> ssvb: but let me double check that ...
<ssvb> initially I was trying to move the stack to SRAM C because the SPL with linked libdram is larger than 24K, and we still need some space for the stack
<ssvb> apritzel: but with AHB1 clocked at 200 MHz, the SPL was mysteriously deadlocking
<ssvb> apritzel: I tried to do simple read/write tests in SRAM C but these tests could not detect any problems
<ssvb> apritzel: and considering that there is also the libdram blob, debugging this all was not a very attractive idea
<ssvb> apritzel: then I tried to use SRAM C as a temporary buffer for the FEL based SPI flasher and could see that changing the AHB1 clock speed seems to affect the reliability
<ssvb> apritzel: downclocking AHB1 to 100 MHz also resolved the "SPL stack in SRAM C problem" for me, which used to be a strange mystery before
<ssvb> apritzel: but if you can observe something weird on your board, then some problem with SRAM C might be still there
zuikis has left #linux-sunxi [#linux-sunxi]
<ssvb> Wizzup: yes, it should be compatible
<jemk> apritzel: ssvb: don't know if this is relevant, but i think maybe sram C is still used for VE as it was in A10/A20, it only isn't documented anymore.
<jemk> at least on h3 it is, as it only works there after the ve clock gate is enabled
IgorPec has quit [Ping timeout: 264 seconds]
<ssvb> jemk: this is interesting, thanks
reinforce has quit [Quit: Leaving.]
reinforce has joined #linux-sunxi
<ssvb> so can we also enable SRAM C on H3 in the bootloader if we poke some VE configuration bits?
jemk_ has joined #linux-sunxi
jemk has quit [Remote host closed the connection]
<Wizzup> ssvb: ok, it's on my x200 - trying to flash libreboot
<jemk_> ssvb: yes, by setting bit0 in 0x01c20064, but i don't know if it is safe
jemk_ is now known as jemk
<ssvb> I always thought that the SRAM C in the H3 manual was a copy/paste error from some other SoC documentation :)
<jemk> also the mapping bits in sysctrl reg0 still work, but the manual doesn't mention them anymore
<ssvb> it's good to know this, and also testing/confirming it with A64 may be a good idea
<ssvb> we actually currently only see 108K of SRAM C instead of 160K on A64, and there may be some knobs responsible for controlling this
<ssvb> Wizzup: are you trying to use the sunxi-fel tool as a general purpose SPI flasher to hack your x86 laptop firmware? :-)
<ssvb> Wizzup: just in case, it's best to be a bit careful with it, this code is still experimental
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria http://www.kvirc.net/]
<jemk> ssvb: on A20 only the first 62kB worked, the rest was strange, readonly, sometimes only one byte per word and such stuff, some bytes even changed themself
tlwoerner has quit [Ping timeout: 276 seconds]
<KotCzarny> magic
matthias_bgg has joined #linux-sunxi
<nove> it is guessed that those part of sdram (or all) represents the internal state of the video engine
<ssvb> jemk, apritzel: if the SRAM C is owned by the video engine, then it not the right place for the ATF firmware either way
<jemk> the lower part is the area for various tables (huffman, scaling lists,...) on a20, i haven't checked it anywhere else
<ssvb> we could try to use it as a temporary storage for the stack or the heap early at the boot time (before the DRAM is initialized), but then the VE state probably needs to be properly configured too
<ssvb> jemk: are these tables getting written there by the CPU during video decode?
<jemk> on a20 the whole ve had to be clocked, on h3 it seems to be enough to enable the bus gate
<Wizzup> ssvb: acknowledged. using the goodfet now
<Wizzup> (since some time)
<Wizzup> but since/if they are the same spi flash with same protocol, why would it now work? ;)
<jemk> ssvb: yes, through the two sram regs (0x?e0/0x?e4)
<nove> (in jpeg encoding, the quantization coefficients writing into the fifo register, can be seen in some (readonly) place of the sdram)
<ssvb> jemk: I mean is the CPU ever addressing this data in SRAM C directly? because right now it looks like when the AHB1 clock speed is "too fast", then data corruption is happening
<jemk> ssvb: not in normal operation
<jemk> it can, but it doesn't
Shirasaka-Hazumi has quit [Ping timeout: 250 seconds]
<jemk> looks like a64 still maps some sram to ve
Shirasaka-Hazumi has joined #linux-sunxi
<ssvb> jemk: "bootmode bit"?
<ssvb> is this "cedar_devp->sram_bass_vir" register documented somewhere?
<apritzel> ssvb: the plot thickens for tiny_printf
<apritzel> with USE_TINY_PRINTF: DRAM Type = <hangs>
<ssvb> aha, this makes sense :-)
<apritzel> without: DRAM Type = 3 (2:DDR2,3:DDR3,6:LPDDR2,7:LPDDR3 ... boots on fine
<apritzel> this is called from libdram
phipli has quit [Ping timeout: 244 seconds]
<jemk> ssvb: it is system control (0x01c00000), but the manual doesn't say anything about these regs
popolon has joined #linux-sunxi
<ssvb> apritzel: hmm, strings says "DRAM Type = %d (2:DDR2,3:DDR3,6:LPDDR2,7:LPDDR3)" and this does not look like something that is expected to upset tinyprintf
<ssvb> apritzel: still if you revert the stack move patch, does it start working even with tinyprintf?
xalius has joined #linux-sunxi
xalius has quit [Client Quit]
<ssvb> apritzel: a potentially unreliable SRAM area used for the stack may cause all kind of problems
<apritzel> no, reverting your patch doesn't help anymore
<apritzel> that only fixed something because SPL didn't have it, but U-Boot had it
<apritzel> so reverting it in U-Boot just put both binaries back to the same state
<apritzel> just put a printf %d in front of the libdram call: it hangs as well
<apritzel> so it's not related to some libdram pecularities
Mr__Anderson has quit [Remote host closed the connection]
nove has quit [Quit: nove]
mnr has quit [Quit: leaving]
tlwoerner has joined #linux-sunxi
Nacho has quit [Remote host closed the connection]
Nacho has joined #linux-sunxi
matthias_bgg has quit [Quit: Leaving]
iamfrankenstein1 has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 258 seconds]
iamfrankenstein1 is now known as iamfrankenstein
<apritzel> ssvb: looks like a linking problem
afaerber has quit [Quit: Ex-Chat]
<apritzel> div_out tries to access "zs", which gets address 4ff80104 in the binary
<ssvb> apritzel: hmm, something like the .bss section in DRAM weirdness?
popolon has quit [Quit: WeeChat 1.3]
<ssvb> apritzel: the sunxi-common.h header has #define CONFIG_SPL_BSS_START_ADDR0x4ff80000
Da_Coynul has joined #linux-sunxi
<apritzel> yes, zs is in .bss
<ssvb> I think that this shit was invented as a workaround for some poorly written code, which uses large global buffers (was it FAT filesystem support?)
<ssvb> but sunxi does not have a large .bss section
<apritzel> yes, readelf u-boot-spl.bin confirms this: .bss NOBITS 4ff80000 020000 000108
<ssvb> apritzel: that's a good catch, potentially every board is in danger after TINY_PRINTF has been enabled for sunxi
paulk-collins has quit [Quit: Leaving]
<ssvb> and this landed only a couple of weeks until the v2016.07 release - http://git.denx.de/?p=u-boot.git;a=commit;h=8c7d22965dc7bf3abcd3bdcc28164c378525f445
<apritzel> so BSS wasn't used before DRAM got initialized before tiny_printf?
<ssvb> U-Boot had some very weird rules about it, maybe Tartarus can explain
<ssvb> yeah, this probably originates from http://lists.denx.de/pipermail/u-boot/2011-February/086611.html
<apritzel> yes, initialising the two .bss variables in tiny-printf to 1 and thus moving them out of BSS fixes the problem
<ssvb> and instead of fixing the FAT bastard, somebody thought that it is a very cool and creative idea to place the .bss section in DRAM and try to make sure that nobody uses the data allocated in .bss before the DRAM is initialized
<apritzel> are there boards using FAT from SPL?
<apritzel> RPi?
<apritzel> ssvb: yes, just another prove that those assumptions _will_ break sooner or later
<ssvb> I think that the sunxi linker script was just copy/pasted from omap
<apritzel> any suggestion for better .BSS offset?
<apritzel> maybe the broken SRAM C region ;-)
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<ssvb> it's probably best to drop this .bss DRAM hack altogether, we are saving a lot more by just enabling TINY_PRINTF
<ssvb> the SRAM C section is not available on every sunxi board
<apritzel> yeah, was just a joke (because it reads as zero initially ;-)
<apritzel> I meant the region after 108K
<ssvb> :)
<apritzel> anyway, in my build .bss is now 256 Bytes
<apritzel> with tiny-printf enabled it's 264 Bytes
<ssvb> anyway, we are saving X bytes by enabling TINY_PRINTF and wasting Y bytes (~256) by dropping the BSS DRAM hack
<ssvb> X is much bigger than Y, and IMHO this is a good justification
<apritzel> indeed
<apritzel> so remove this symbol and let the linker figure it out?
<apritzel> or does U-Boot depend on this symbol?
Da_Coynul has joined #linux-sunxi
<ssvb> just patch the linker script http://git.denx.de/?p=u-boot.git;a=blob;f=arch/arm/cpu/armv7/sunxi/u-boot-spl.lds
<ssvb> we don't really need this "MEMORY { .sdram : ORIGIN = CONFIG_SPL_BSS_START_ADDR, LENGTH = CONFIG_SPL_BSS_MAX_SIZE }" thing
<apritzel> ah, I see
<apritzel> yes
afaerber has joined #linux-sunxi
<ssvb> and we probably need to do this fast, otherwise the v2016.07 release will be potentially broken for many sunxi boards
<apritzel> works
<apritzel> will send a patch ASAP
<apritzel> ssvb: thanks very much for the input!
<ssvb> apritzel: you are welcome
<apritzel> don't the other boards print anything during DRAM init?
<apritzel> or only success messages? ;-)
<buZz> i'm gonna preorder that FSFcertified cpucard and a desktopcase , i think :P
<ssvb> LOL - "A20 Dual-Core ARM Cortex A7, 1.2 GHz"
<buZz> nobody posted that link here yet? tsk tsk
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<ssvb> why on earth so many Allwinner based boards advertise unrealistic overclocked CPU speed?
<apritzel> why an A20? Seems like they don't even use the only A20 "killer feature" (SATA)
reinforce has quit [Quit: Leaving.]
<ssvb> buZz: maybe it's a better idea to just buy some Olimex board with Allwinner A20 if you care about OSHW?
<buZz> ssvb: oh its not my first A20
<buZz> i'm just curious how fun EOMA68 could be, and wanna keep ppl that make fun stuff occupied and happy ;)
<ssvb> does anybody really think that Allwinner A20 supposedly clocked at 1.2GHz is somehow better than the same Allwinner A20, but clocked at the nominal 1GHz?
<buZz> no
<buZz> well, probably loads of people
<buZz> like, turbo buttons used to sell computers
<ssvb> yeah, it's probably better for their marketing
<buZz> ssvb: i guess what i care about most is people knowing about each others efforts, and ability to ride along on their productions
<MoeIcenowy> EOMA68 is FSF certified?
<MoeIcenowy> oh so free
<buZz> MoeIcenowy: thats what that link mentions behind the 'Libre CPU board'
<buZz> > Respects Your Freedom (RYF) Certification from the Free Software Foundation (currently in progress with no known blockers - a full refund will be available if certification is for some reason not granted)
<buZz> (that bet is also part of the reason i want to back this project :P )
merbanan has quit [Ping timeout: 264 seconds]
<ssvb> such marketing lies about the CPU clock frequency backfire in the end, contributing to the already existing "Allwinner makes buggy hardware" stereotypes
<MoeIcenowy> Will sunxi BROM be considered as some "non-free blob"?
<MoeIcenowy> or it should be considered as a part of hardware (SoC)?
<ssvb> MoeIcenowy: no, because it is not upgradeable
<MoeIcenowy> ssvb: so it's not blob?
<ssvb> let me search the old Raspberry Pi story about the FSF requirements
<buZz> raspi now has opensource bootloader :D
<buZz> or well -a- opensource bootloader
<apritzel> buZz: and the other one containing a tiny bit of GPL code ;-)
<buZz> so its in breach of license?
<MoeIcenowy> I think Replicant can be ported to sunxi device...
<MoeIcenowy> Thus RMS can finally have a tablet to use :-)
<apritzel> buZz: hard to prove, I guess
<apritzel> though they once published some code snippet which had the very same comments ...
tlwoerner has quit [Ping timeout: 250 seconds]
<apritzel> and the disassembly looks strikingly similar
<buZz> :)
<ssvb> MoeIcenowy: ok, I give up, but there was a story about the Raspberry Pi people actually seeking for FSF approval, and one of the possible solutions was to make their firmware non-upgradeable
* buZz finds the defendant; guilty
<ssvb> MoeIcenowy: then it can be treated as part of the hardware and be perfectly fine from the FSF point of view
<MoeIcenowy> ah-oh
<MoeIcenowy> interesting
<buZz> hmm
<ssvb> MoeIcenowy: FSF cares about the freedom to modify the code (if this is doable in principle), but if the code is inherently non-upgradeable then it is not an issue :-)
<ssvb> both the hardware maker and the end users have equal rights in this case and nobody is more privileged than the other
<buZz> thats excellent actually
Da_Coynul has joined #linux-sunxi
<apritzel> I am not sure people want to see this code anyway, will only cause eye strain ;-)
tlwoerner has joined #linux-sunxi
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<ssvb> here is a link to the FSF criteria - http://www.fsf.org/resources/hw/endorsement/criteria
<ssvb> "The exception applies to software delivered inside auxiliary and low-level processors and FPGAs, within which software installation is not intended after the user obtains the product."
<ssvb> of course, different people may have different interpretations of this
apritzel1 has joined #linux-sunxi
lemonzest has quit [Quit: Leaving]
<ssvb> the "secondary processor" concept is poorly formulated there, so it is a kind of loophole for the Raspberry Pi because their VPU core may be interpreted as "secondary"
<ssvb> now I'm not so sure how the Allwinner's boot ROM fits there, but I guess the arisc firmware still does not need to be open source according to the FSF criteria
apritzel1 has quit [Ping timeout: 244 seconds]