hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
akaizen_ has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
popolon has quit [Quit: Quitte]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
robb83 has joined #linux-sunxi
robb83 has quit [Client Quit]
arete74 has quit [Ping timeout: 246 seconds]
arete74 has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
hipboi has joined #linux-sunxi
uwe_ has quit [Read error: Operation timed out]
uwe_ has joined #linux-sunxi
Turl has quit [Ping timeout: 240 seconds]
Turl has joined #linux-sunxi
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
robb83 has joined #linux-sunxi
robb83 has quit [Quit: - A hand crafted IRC client]
Ehsand has joined #linux-sunxi
rellla has joined #linux-sunxi
massi has joined #linux-sunxi
<oliv3r> gooood mornin' all
rellla2 has joined #linux-sunxi
rm has quit [Quit: ZNC -]
rellla has quit [Ping timeout: 264 seconds]
rm has joined #linux-sunxi
rm has quit [Quit: ZNC -]
rm has joined #linux-sunxi
n01 has joined #linux-sunxi
<hno> morning
rellla2 is now known as rellla
<oliv3r> g'day
<mnemoc> moin
<oliv3r> hno: the wiki is strange
<oliv3r> hno: A10 & A13 boots the SPL loader from block 8. This then loads actual u-boot from block 40 onwards, counted in 1KB blocks. Replace /dev/sdX with the device name of your media
<oliv3r> but then 2 lines lower If using v2013.07 or earlier then the procedure is slightly different
<oliv3r> but both section talk about block 40
<oliv3r> so I think (maybe under a special header) seek=32 for old
<oliv3r> also, old uses u-boot.bin (at seek=40), but 'new' isn't mentioned at all (which uses u-boot.img)
<oliv3r> default boot action lists 'watchdog 0' but i think that command is long gone
popolon has joined #linux-sunxi
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
rellla2 has joined #linux-sunxi
<oliv3r> hno: also, possibly unrelated, but maybe make a mention in your very last chapter (about SD card layout) that when using a GUID the maximum amount of partitions is 15, due to the BROM starting to read at 8k?
ZetaNeta has quit [Ping timeout: 260 seconds]
<hno> oliv3r I am not familiar with GUID partition table requirements.
<hno> is anyone even using GUID partitions on SD cards?
<oliv3r> has a nice image
<oliv3r> you don't have to know about GUID, except for that little image which explains pretty much all
<oliv3r> or is a nice summary :)
<oliv3r> what I don't understand is why 8k offset was choosen; just such an arbitrary offset. MBR + partition table is only 512 bytes, 4k offset maybe ideal for 4k sector drives (that we'd not even use on flash or mmc) or 64k for a fully blown GUID
<oliv3r> 8k is so ... random
<oliv3r> also, the 'size' in the storage map is wrong, the start appears to be right
<hno> oliv3r, I don't know why 8KB. But it appears that is the only offset BROM tries to load the boot image from.
ZetaNeta has joined #linux-sunxi
<oliv3r> yeah, i was just curious behind the thought-process or reasoning
<hno> what is wrong in size?
<hno> or better yet, just fix what is wrong.
<oliv3r> it says start 8, size 24k, but then start 40 for the next, making it not align with the size
<hno> SPL size should be 32 KB now.
<oliv3r> while 24 (23.5) is the current size, do we want to list the maximally available 'space' or the current size there?
<oliv3r> hno: exactly!
<hno> in layout it's the size of the reserved region.
<oliv3r> hno: 32 KiB (with a mention that the SPL itself cannot be larger then 30k due to hardware limits?)
<oliv3r> anyhow, i've never edited a github wiki so no clue :p
<hno> just push edit button, or check out the wiki git repo and edit with whatever tool you like.
<oliv3r> ok, i'll fix it :)
<oliv3r> i'll also reword How to make a bootable sunxi SD-card
<oliv3r> hno: should be good now
<oliv3r> i left watchdog section in for now
panda84kde has joined #linux-sunxi
<oliv3r> hno: have you pushed your changes to sunxi yet?
<oliv3r> the seek=40 changes?
<oliv3r> i dont' see them yet
<oliv3r> nvm, i'm looking wrong
<oliv3r> what's this combined u-boot + spl image for?
CoSM0 has quit []
vicenteH has joined #linux-sunxi
Black_Horseman has quit [Ping timeout: 264 seconds]
merbanan has joined #linux-sunxi
<oliv3r> hno: +#define CONFIG_SPL_MAX_SIZE 0x4000 /* 16 KiB */ what is this limit for?
<Seppoz> hno :ping
<oliv3r> Seppoz: are you trying the new u-boot?
<Seppoz> no im trying to figure out what exactly is new
<oliv3r> the addressing has changed
<oliv3r> the maximum size allowed has changed
<oliv3r> before, you used to do dd=u-boot.bin of=/dev/sdX size=1024 seek=32
<oliv3r> now you have to do dd if=u-boot.IMG of=/dev/sdX size=1024 seek=40
<oliv3r> note the IMG (hence caps) and the new seek=*40*
<oliv3r> hno: i think you broke the BSP, let me figure out how to fix it ;)
<oliv3r> hmm, maybe it's not the BSP that's broken?
<oliv3r> make[1]: *** No rule to make target `/silo/build/sunxi-bsp/build/cubieboard2-u-boot/spl/sunxi-spl.bin', needed by `/silo/build/sunxi-bsp/build/cubieboard2-u-boot/u-boot-sunxi-with-spl.bin'. Stop.
<oliv3r> make Cubieboard2 CROSS_COMPILE=arm-pc-linux-gnueabi- O=/silo/build/sunxi-bsp/build/cubieboard2-u-boot/
<oliv3r> mripard: ping
mcbrick has quit [Quit: Leaving]
<Seppoz> sorry was on the phone
rellla2 has quit [Quit: Nettalk6 -]
<Seppoz> oliv3r: ok by why?
<Seppoz> what was changed in buoot?
<Seppoz> *uboot
<oliv3r> the size
<oliv3r> we found out that the SPL can be larger, since we where hitting the limits almost allready, we changed the sizes to be ready for future expansions
<oliv3r> and since the BROM can read 30k, why not use 30k
<mnemoc> greed....
<oliv3r> lol
<oliv3r> no, but it means we stablize the disklayout more :)
deasy has joined #linux-sunxi
geecko has joined #linux-sunxi
<libv> oliv3r: i worked for nokia, on the N9
<oliv3r> ahh i wasnt sure anymore :p
<Turl> libv: the N9 ads are awesome
<oliv3r> Turl: can I steal 5 minutes of your time?
<Turl> sure
<oliv3r> i'm reviewing it myself now, but i suck at reviewing my own stuff
<oliv3r> cause to me, it's all fine
<Turl> didn't you say you supported A20 too?
<Turl> need to fix description :p
<Turl> CPU's -> my eyes :p
<Turl> need to split the dt bits to separate patches too
<oliv3r> didn't I do that?
<oliv3r> 1 patch per dt? nobody complained about that before (sun4i sun5i)
<oliv3r> but ok, 3 patches for each dt
<Turl> well you could do one dt I guess
<Turl> but it shouldn't be in the same one as code
<Turl> L72, do you really need the cast?
<Turl> same L109
<Turl> I think L110 is unnecesary (devm_kfree), but confirm with someone else
<Turl> the point of using devm is that you don't need to clean right?
<Turl> bbl
<oliv3r> bye bye :)
<oliv3r> and thanks
<oliv3r> and there's 2 patches :)
<oliv3r> 1 for DT, 1 for the rest
<Turl> np :) I'll have a more thorough look later
<Turl> well your link is just 1 :P
<oliv3r> yeah but it's 2 seperate commits :)
<oliv3r> or should be, now i'm doubting myself :p
<Turl> well github says it's 1 ;)
<oliv3r> yeah
<oliv3r> ok that's stupid
<oliv3r> i have 2 .patch files
<Turl> and for 4month ago
<oliv3r> i created with git format
<Turl> sure it's that? :P
<Turl> from*
<oliv3r> i'll check content, but yeah i noticed that too
<oliv3r> but i created a new branch
<oliv3r> based from maxime's branch
<oliv3r> i then did git am my patches atop of tha
<oliv3r> i'll redo it
<oliv3r> content is from very recent though
<oliv3r> so i don't know what git's deal is
<oliv3r> as to L72, doesn't platform_get_Drvdata return a void *
<oliv3r> so i would say it's atleast clean to do the cast?
<oliv3r> Turl: commit was somehow gone bad, here's the proper log
<oliv3r> that linked hash may have been old/wrong
<Turl> void * casts to any * without needing an explicit cast
<oliv3r> ok i'll drop the cast then
<oliv3r> i still find it aids in readability :p
<Turl> it's dangerous tho
<oliv3r> it can be, i agree there
<Turl> explicit casts can be evil :P
<oliv3r> i thoguht you where bbl-ing :)
<Turl> yeah, needed to pick a few things still :P
<Turl> now I'm gone for real
wingrime has joined #linux-sunxi
wingrime has quit [Ping timeout: 245 seconds]
tinti has joined #linux-sunxi
rellla has quit [Ping timeout: 264 seconds]
wingrime has joined #linux-sunxi
<mnemoc> will temporarily shutdown the sunxi server (2 minutes) to get some extra partitions mounted..... damn lxc
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<mnemoc> up
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
ganbold_ has joined #linux-sunxi
akaizen_ has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
<wingrime> mnemoc: why mount additional partitions requires reboot?
rz2k has joined #linux-sunxi
<mnemoc> wingrime: because the /dev/... don't exist within the container
<mnemoc> and lxc doesn't have a tool to run things in the guest's namespace as vserver does
<oliv3r> you mount the partition UNDER lxc?
<wingrime> mnemoc: we use VPS ?
<oliv3r> e.g. /path/to/lxcontainers/vm1/home
<oliv3r> where vm1 would have its root mounted (if at all)?
<oliv3r> so then i understand you have to stop the container, mount the partition and start the vm
<mnemoc> oliv3r: the config file tells where is /, the rest is an fstab in the host
<mnemoc> oliv3r: stop the container, edit fstab, start the container
<oliv3r> ah right
<oliv3r> i have so much to do, i need a todo list so i dont' keep forgetting it
<oliv3r> and maybe use it to force myself to actually do it
<mnemoc> wingrime: it's isolated, but it doesn't have resource restrictions
<wingrime> mnemoc: there is anything more that runs on same server?
<mnemoc> my build server
<mnemoc> and some other minor usage web apps
* wingrime can't figure whats difference between lxc and chroot
<mnemoc> chroot can see the world's processes, lxc/vserver/vz doesn't
<mnemoc> capabilities and resources can also be limited. and you can't escape the jail... as it can be done with chroot
<wingrime> why not xen
<mnemoc> each guest has it's own pid 1
<mnemoc> more expensive?
<mnemoc> containers don't virtualize or emulate, and run the same one kernel
<mnemoc> vserver/lxc/vz is basically a secure chroot with it's own pid 1
<mnemoc> very lightweight
<mnemoc> but lxc is still a child
<wingrime> xen more perefed if you buy VPS
<oliv3r> can you 'suspend' an lxc like you'd suspend a KVM?
<oliv3r> xen is shit
<wingrime> oliv3r: xen can't be overselled so easy
<oliv3r> it's all cooperatized
<wingrime> oliv3r: vz can
<mnemoc> oliv3r: don't think so..... but you can stop it :p
<oliv3r> hmm, one of the things i like baout my KVM remote desktop is, that I can save/resume it whenver i shutdown/reboot my server
<oliv3r> tecnically it should be possible imo :)
<mnemoc> the key advantage of lxc is that it's mainlined and comes enabled by default in my package-manager controlled kernel :p
<wingrime> mnemoc: not sure that hardware virtualizatoin and paravirtualizet hw have much speed impact
<wingrime> mnemoc: kvm also mainlined
<mnemoc> i use kvm to test work, to see how things boot. but i don't need emulation to isolate a web app
<oliv3r> yeah that's why i'm slwoly switching from kvm to lxc
<oliv3r> but not having suspend ability kinda really sucks
<wingrime> there is anyone who tryed digital ocean or amazon aws?
<mnemoc> if you want to suspend a service, just stop it
<oliv3r> i also have a few kvm'ed hosts that run some things. If i need to reboot my server, no problem, they continue to run after a reboot
<wingrime> oliv3r: also kvm must be workable on a20
<oliv3r> wingrime: i think so :)
<mnemoc> allwinner's a7 has virtualization?
<oliv3r> mnemoc: so if i stop the process, does its memory get written to disk, and when I start it again, it reads the memory and resumes from where it left off/
<oliv3r> mnemoc: of course
<oliv3r> mnemoc: A7 == slow(A15)
<oliv3r> so it has kvm possibility :)
<mnemoc> sloooooooooowly
<wingrime> mnemoc: At least we can try emualte gameboy with it
<mnemoc> :)
<mnemoc> or a c64
<wingrime> cool
<wingrime> android in qemu
<oliv3r> but i'm not sure we are that far with a20 yet :)
rellla has joined #linux-sunxi
<oliv3r> but with kvm on arm
<oliv3r> running linux and a small KVM ith android next to it is cool
<oliv3r> but .... you need cubietruck at the least; as it has 2gb :)\
<buZz> was cubietruck going for the a31?
<buZz> or a20?
<wingrime> oliv3r: with kvm we can use android and linux sametime
<oliv3r> yeah, but you need RAM
<wingrime> oliv3r: also, ve can 512+512
<mnemoc> iirc there will be two cubietrucks, one with a20 (and sata) and one with a31
<buZz> ah cool
<buZz> a31 will not have sata?
<wingrime> oliv3r: also will be nice get to mali in conteiner
<mnemoc> a31 is for tablets, like the a13
<buZz> ah :(
<buZz> boooo
<oliv3r> i don't care much for a31 :p
<mnemoc> it's good for android people....
vicenteH has quit [Ping timeout: 240 seconds]
<wingrime> I still wait some thing with 4xCortex a15 and mali Txxx
<oliv3r> because they don't care about blobs
<oliv3r> and i say, let them swallow on their belovd blobs!
<oliv3r> drown too
* mnemoc wants a chromebook pixel with exynos5 octa!
<oliv3r> i just want money to buy food :(
<mnemoc> i didn't say I wanted to PAY for that chromebook :p
<mnemoc> i just want it, like a child
<oliv3r> oh
<oliv3r> yeah, i want a google chromebook pixel too!
* mnemoc hopes to see more laptops with decent screens this xmas
<mnemoc> 1366x768 is a joke
<mnemoc> how many years stuck with that junk?
<mnemoc> phones have better resolution than most laptops
<oliv3r> heh, yeah
<oliv3r> my laptop has 1440x1050
* wingrime untill thay sell everything
<oliv3r> but that laptop is from 2004 :p
<oliv3r> (which I happily use)
<oliv3r> my gf's 2007 laptop has 1400x900, horrible
<mnemoc> look at today's catalog. most are still 1366x768
<oliv3r> it's pathetic really
<oliv3r> but people want a laptop for 200 USD
<oliv3r> and don't care about resolution
<n01> I have a retina display that rocks ;)
<oliv3r> heck, here at work, they change beautiful 1920x1200 24" monitors to 1024x768 becaues it's better to read for their old eyes
<mnemoc> 1k laptops also have 1366x768
<mnemoc> oliv3r: o_O
<mripard> oliv3r: pong
<oliv3r> mripard: hi!
<oliv3r> mripard: can you give a quick wif on the top 2 commits here:
<mripard> wif?
<oliv3r> mripard: I was gonna send it off as is, unless you spot something that needs changing
<oliv3r> quite smell
<oliv3r> quick*
<oliv3r> a short sniff
<oliv3r> a spiffy glance
<oliv3r> :)
<mripard> just send it :)
<oliv3r> well there's new a20 support
<oliv3r> but, okay, hopefully i don't have to do a v7!
atiti has joined #linux-sunxi
\\Mr_C\\ has quit []
atiti has quit [Ping timeout: 245 seconds]
<wingrime> libv: I readed some news about lima, what a hell with ARM ?
<oliv3r> wingrime: lol I think that's old news, but yeah, stupid bastards :S
fredy has quit [Ping timeout: 245 seconds]
<mnemoc> wingrime: there is a post in libv's blog
<mnemoc> not THAT old i think, but sad anyway
<oliv3r> 4th of june no?
<mnemoc> true :(
<mnemoc> old
<wingrime> oliv3r: easy shut up ARM, just not aporve arm64 patches
<oliv3r> haha, won't help us
<oliv3r> better way is to just have an awesome mainlined driver and just give them the middle finger
<oliv3r> once no linux distro will use the mali drivers, we won
<oliv3r> as for android, well, that's a nother story :(
<wingrime> oliv3r: greg can do
<wingrime> oliv3r: linux is major market point of arm
<oliv3r> yep it is
stekern has quit [Ping timeout: 276 seconds]
<wingrime> middle finger can be usefull
stekern has joined #linux-sunxi
<wingrime> oliv3r: also, I don't like jemk devel style
<oliv3r> wingrime: it's PoC, it doesn't matter :)
<wingrime> oliv3r: it workable
<wingrime> oliv3r: vpdau plugin
<oliv3r> oh very nice
<oliv3r> but it'll have to be re-written when we have a new kernel driver
<wingrime> oliv3r: we need disp driver firstly
<oliv3r> also to be legally safe it'll have to be rewritten anyway
<oliv3r> yeah i dont' know if that'll happen any time soon
<oliv3r> i don't have the technical knowledge to do that :)
<wingrime> oliv3r: sata looks not issue
<oliv3r> allready started on that :p
<wingrime> oliv3r: but I need pll/clock/gate in mainline firstly
hramrach has quit [Ping timeout: 240 seconds]
<oliv3r> i think clocks got merged allready, might hit 3.11
popolon has quit [Quit: Quitte]
\\Mr_C\\ has joined #linux-sunxi
Ehsand has quit [Quit: Leaving.]
<wingrime> oliv3r: but it required for most driver
<wingrime> *s
dragonn has joined #linux-sunxi
<oliv3r> yeah
<wingrime> oliv3r: axp also can be in my plans
<oliv3r> yeah, but big big driver
<wingrime> oliv3r: not realy big, but complex
<wingrime> oliv3r: it have own gpio and IRQs on board , regulators , reboot/shutdown support , battary control
<oliv3r> yeah it's like a mini kernel ;)
hramrach has joined #linux-sunxi
gzamboni has quit [Ping timeout: 276 seconds]
fredy has joined #linux-sunxi
<oliv3r> crap my '--to' failed :S
gzamboni has joined #linux-sunxi
wingrime has quit [Ping timeout: 256 seconds]
wingrime has joined #linux-sunxi
<Turl> oliv3r: btw prefix on patch subject would be nice
<Turl> like 'ARM: sunxi: dt: Add sunxi sid to dts blah blah'
<oliv3r> Turl: touche!
<oliv3r> yes, i should have done that
<Turl> you still have that devm_kfree
<Turl> mripard: is it needed? ^
<Turl> I thought the point of devm_* was that all the frees and such were automagic
<oliv3r> well it IS a devm_* kfree
geecko_ has joined #linux-sunxi
<mripard> yeah, but the point is that you don't need it
<oliv3r> crap!
geecko has quit [Ping timeout: 248 seconds]
<WarheadsSE> psst. The mali driver sucks.
<Turl> oh you sent v5 oliv3r
<Turl> oliv3r: btw, for v6 include the v6 on all the [PATCH...] tags, not just on the cover letter
<Turl> makes it easier to spot
Offshore has joined #linux-sunxi
Offshore has quit [Client Quit]
stekern has quit [Read error: Operation timed out]
dragonn has quit [Ping timeout: 245 seconds]
stekern has joined #linux-sunxi
gzamboni has quit [Ping timeout: 245 seconds]
gzamboni has joined #linux-sunxi
Soru has joined #linux-sunxi
n01 has quit [Ping timeout: 248 seconds]
geecko_ has quit [Quit: Quitte]
ykchavan has joined #linux-sunxi
xtofury has joined #linux-sunxi
wingrime has quit [Read error: Operation timed out]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<Turl> mripard: the fixme on oliv3r's patch refers to the sysfs race that needs to be fixed post-greg's patches
<hno> oliv3r, still problems?
robb83 has joined #linux-sunxi
<hno> oliv3r, the 16KB SPL is th FEL version.
naobsd has joined #linux-sunxi
<mripard> Turl: and we can expect everyone reading this chunk of code to know that, right ? :)
wingrime has joined #linux-sunxi
rellla has quit [Quit: Nettalk6 -]
<xtofury> Ok red led issue solved, nail polish it is.. lol.. but now I have a new annoyance that I just simply do not get. There are two internal storage partitions, and the one for apps is too small seeing as if I wanted to store anything else I'd use sata or I'd use extsd. Where do I find the part of the source that can change the size of these partitions, so that the end compiled image winds
<xtofury> up giving my apps more breathing room, and, is there any reason for mnt/sdcard at all or can I get rid of it entirely?
<Turl> mripard: the /* proper way */ comment next to the group thing kinda gives a hint ;)
boycottg00gle has joined #linux-sunxi
ganbold_ has quit [Remote host closed the connection]
<xtofury> looks like it is under /bootable/diskinstaller
tinti has quit [Ping timeout: 256 seconds]
tinti has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
rz2k has quit [Read error: Connection reset by peer]
boycottg00gle has quit [Remote host closed the connection]
gzamboni has quit [Ping timeout: 245 seconds]
arete74_ has joined #linux-sunxi
arete74 has quit [Ping timeout: 264 seconds]
panda84kde has quit [Quit: Konversation terminated!]
notmart has quit [Quit: notmart terminated!]
WarheadsSE has quit [Ping timeout: 268 seconds]
WarheadsSE has joined #linux-sunxi
gzamboni has joined #linux-sunxi
massi has quit [Remote host closed the connection]
<oliv3r> hno: i saw hansg's patch drop by; but i'm not using _FEL
<oliv3r> Turl: dunnow how to add the v on all patches, since i manually edit the cover letter; ohh wasn't there some trick to modify the [PATCH] think, yeah i remember you talking about
<oliv3r> mripard: i don't understand why it's sun4i-sid, but not sun7i-sid
ykchavan has quit [Read error: No route to host]
ykchavan has joined #linux-sunxi
<oliv3r> mripard: I'm fine with changing it, but how was I supposed to know! I Or was it supposed to be sun4i-a10-sid too but it kinda happened this way?
<oliv3r> hno: with your latest commit (hansg) it builds again
<wingrime> Tsvetan: are you get last SDK ?
<wingrime> hno: #if 1
geecko has joined #linux-sunxi
<geecko> hi all
jemk has joined #linux-sunxi
leowt has joined #linux-sunxi
BJfreeman has quit [Read error: Connection reset by peer]
_BJFreeman has joined #linux-sunxi
<robb83> anyone know, how to process /dev/input/event2 touchscreen data?
_BJFreeman is now known as BJfreeman
jabjoe1 has joined #linux-sunxi
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
<jabjoe1> Hi, I've been working on a driver for the gslx680 touch screen my A13 tablet has. It's now working well enough for my uses.
<jabjoe1> I've pushed the code to:
<oliv3r> jabjoe1: you can do a PR on the ML or even better, submit a patch for review on the ML would be great
<jabjoe1> Ok, I'll do that tonight.
_BJFreeman has joined #linux-sunxi
BJfreeman is now known as Guest98998
_BJFreeman is now known as BJfreeman
Guest98998 has quit [Read error: Connection reset by peer]
<hno> wingrime, I have left an #if 1 somewhere?
Soru has quit [Ping timeout: 245 seconds]
<hno> right...
xtofury has left #linux-sunxi [#linux-sunxi]
<hno> fixed
Soru has joined #linux-sunxi
<wingrime> hno: can our uboot be mainlined ?
<mnemoc> ssvb: should I mirror sunxifb from your github or you intend to move it to the common repo? :p
<ssvb> mnemoc: what is the common repo?
<mnemoc> gh/linux-sunxi
<mnemoc> I'm making a but as r/o mirror of the stuff
<wingrime> mnemoc: nice
<mnemoc> and wanted to add your sunxifb there
<wingrime> mnemoc: don't forget code search and paste
<mnemoc> wingrime: the TODO is huge :p
<wingrime> cool
<mnemoc> but with away (moved to .cn) at least it will be easier to setup a mailman
<oliv3r> wingrime: OF COURSE ,WHY NOT
<oliv3r> wingrime: sorry caps, but yeah it can :)
<mnemoc> by code search I assume lxr. for pastes, personally i'm pretty happy with
<oliv3r> mnemoc: what do you mean, away
<geecko> what is modules.builtin?
<oliv3r> mnemoc: sprunge doesn't sometimes fit
<mnemoc> oliv3r: not hosted at sunxi's server
<oliv3r> mnemoc: oh really, why did they move
<geecko> got that when integrating mali drivers.
<mnemoc> oliv3r: the great firewall banned them
<mnemoc> weirdly, not our wiki. only their blog
<oliv3r> mnemoc: ohh that sucks
<ssvb> mnemoc: I think you can set up a read-only mirror for sunxifb for now
<oliv3r> i did notice the site beoming very slow
<mnemoc> ssvb: from your ssvb or after moving it to gh/linux-sunxi ? :p
<wingrime> mnemoc: why banned?
<oliv3r> wingrime: do not question the great firewall of china
<ssvb> mnemoc: sigh, I still need to do something about
<mnemoc> they simply decided to block it
<wingrime> oliv3r: our gov want setup something simular
leowt has quit [Quit: leowt]
<mnemoc> ssvb: if you move it, the issues will go with it ;-)
<mnemoc> ssvb: then you can make a personal fork for toying
<wingrime> oliv3r: some ISP do , some not
<wingrime> oliv3r: my not
<oliv3r> wingrime: great firewall of putin?
<hno> wingrime, it's one of my goals.
<wingrime> oliv3r: ))
<mnemoc> in the name of child porn!
<oliv3r> for the children!
<wingrime> yeax
<ssvb> mnemoc: I mean, it will be probably eventually moved to
<mnemoc> ok
<wingrime> oliv3r: tv saying that pirated totaly blocked
<wingrime> oliv3r: but I not even noticed
<oliv3r> yeah
<oliv3r> propaganda, its sad really
<wingrime> oliv3r: some providers realy do,
<oliv3r> we only hear negative news t hese days from russia
<mnemoc> the best thing about the uk's wall is that it will be implemented by huawei. same as china's, iran's and vietnam's
<wingrime> oliv3r: I have bought VPN for that case
<mnemoc> experts in.... child protection!
<wingrime> oliv3r: but not use it
<oliv3r> lol; atleast you know your in good company
<oliv3r> yeah i wanna get a .at and .usa vps (for other things) but also will use it as vpn gateway
<mnemoc> wingrime: part of the great wall features is that your connection gets resets if they detect you are trying to skip it
<wingrime> mnemoc: there many ways to avoid
<wingrime> mnemoc: tor
<wingrime> mnemoc: i2p
<wingrime> mnemoc: ssh
<wingrime> mnemoc: ...
<oliv3r> if i have more income, i'll host a tor exit node with a few VPS'es that'll allow it
<mnemoc> yes, and there are many ways to punish you for using them
<oliv3r> sattelite internet :)
<mnemoc> =)
<hno> $$$
<wingrime> mnemoc: but most probles , that thay ban for IP , that means that can ban many innocent sites
<oliv3r> over here, it's not too expensive anymore
<wingrime> oliv3r: with up-link ?
<oliv3r> here, they offer tripple play via sat
<wingrime> oliv3r: I also heared that in iran sat dishes not allowed
<oliv3r> but i watch tv via sat!
<oliv3r> yeah, then again; i wouldn't wanna live in a country like that
<hno> oliv3r, all such solutions I have seen uses some wired media as uplink.
<mnemoc> wingrime: .cn blocked and not, both using the same IP
<wingrime> mnemoc: yes, I know
<mnemoc> and it wasn't a resolving problem. connection simply failed
<wingrime> mnemoc: but our blockers is too crazy
<oliv3r> hno: here in NL; upstream goes via sat
<mnemoc> if you use tor you can be assumed to be doing something fishy.... like with bitcoins
<wingrime> mnemoc: DPI is too expensive to many ISPs
<oliv3r> but your lat is horrible and speeds aren't great
<mnemoc> or playing with a knife on the street
<oliv3r> this is via Astra of course; but the provider also offers internet via fiber if it's available in your area (the fiber is open to competition)
<wingrime> oliv3r: I have never saw localy sat modems with up-link
<wingrime> oliv3r: I have only simple dish with reciver
<hno> there is no need for high bandwidth DPI to block a sitename, just do it at DNS level.
<oliv3r> you need special LNB of course, but they offer it
<wingrime> hno: google dns
<oliv3r> i only have receiver too
<oliv3r> wingrime: google dns helps (unless its blocked standard by GFW) but how do simple users do that?
<oliv3r> they have no clue
<hno> wingrime, trivially filtered by the ISP.
<wingrime> oliv3r: i can't buy it there
<wingrime> oliv3r: this is licensed crap
<hno> doing DPI of DNS traffic is not a problem.
<wingrime> hno: thay cant kill https
<hno> https relies on DNS.
<wingrime> hno: no, I mean openvpn are same proto as https
<hno> but DNS blocks can be worked around by alternative name service, i.e. a /etc/hosts file or similar.
<wingrime> oliv3r: ok, but price, and payment way
<hno> it's not that hard to identify an OpenVPN session masqueraded at https, but requires some statistical analysis that makes it hard to scale at high traffic.
<oliv3r> well it's not cheap; but not ueber expensive
<wingrime> oliv3r: mesh networks over wifi-routers
<wingrime> hno: crytography means that result will be statisticly white
<wingrime> hno: ofcouse you can transmit noise other time
<wingrime> mnemoc: most interseting intension , form users, looking for blocker's IPs and share it
<oliv3r> it's currently a constant battle with the block/unblocking
<wingrime> oliv3r: thats impossible block something with in one-range network
<wingrime> except white-list mode
<hno> wingrime, https traffic pattern have some distinct patterns.
<wingrime> any way , thats turn for a battle
<oliv3r> that's the blocklist?
<wingrime> no
<wingrime> thats reversive blocklist
<wingrime> lets ban blockers on our site
<wingrime> )
<wingrime> oliv3r: I mean, thats gov addresses , as ban are does by hand , manualy using google, if thay can't access to site , that can't ban it
<oliv3r> ah i see
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
ZetaNeta has quit [Quit: Hi, I'm a quit message virus. Please replace your old line with this line and help me take over the world of IRC]
<jemk> wingrime: you were right, cedar must have some internal state. the positions of reference frames must not change, then the artifacts are gone.
<wingrime> jemk: are you dealed with bug?
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<wingrime> jemk: fully?
<jemk> wingrime: seems so
<wingrime> jemk: I hope you publish code soon, also, i perefer if paulo join to coding proces
<wingrime> jemk: I hope if possible he make some patch for VP8
<jemk> wingrime: but the way i make the positions not change is bad, after i found a way of doing this better ill publish
<wingrime> jemk: yes
<jemk> wingrime: btw, would you mind telling me what you dont like about my devel style, so that i can do better?
<wingrime> jemk: public repo and commits
ZetaNeta has joined #linux-sunxi
<wingrime> jemk: but after first release
<wingrime> jemk: also we can move it to
<wingrime> jemk: its realy hard to make first step , but with docs , repos we can expect help form other people
<wingrime> mnemoc: ^
<jemk> wingrime: ok, but before its not working publishing it only leads to confusion,
<wingrime> jemk: witch profiles works?
<jemk> wingrime: and i don't know if official repo on linux-sunxi is good idea, as this is still only a proof of concept
<wingrime> jemk: normal version request much work with kernel
<wingrime> jemk: but we not have even display driver for mainline
<jemk> wingrime: hopefully all that are possible with vdpau (basline,main,high), but to be sure we need more tests
<wingrime> jemk: fix image placement firstly
<jemk> wingrime: i still not know where some flags have to be set, simply cause i've not found any video using these flags
<wingrime> jemk: after publish we can expect bug reports
<wingrime> with samples
<jemk> wingrime: correct placement must be done by someone familar with x11, and i don't think it will work well without driver integration
<wingrime> ssvb: ^
<wingrime> jemk: ssvb familar
<wingrime> jemk: he maintrians sunxi-fb (now with XV)
<ssvb> wingrime: one part is xorg ddx driver (most likely dri2), another part is a client side library which provides vdpau api
<jemk> wingrime: i know, using it all the time. sadly xv isn't able to handle the tiled format of cedar
<libv> jemk: is that really an xvideo restriction?
<wingrime> jemk: currently we only need setup color key and hw overlay
<ssvb> jemk: adding a tiled format for XV is not difficult, the problem is that XV wants SHM memory buffer
<libv> or you could pass handles as the actual buffer
<jemk> sry, i don't know anything about x
<jemk> need your help there
<libv> the trouble is usually the xv implementations of the media players
<libv> if you adapt the media player and the driver, then pretty much anything can be done
<ssvb> it makes sense to have a look at vdpau implementation in radeon and nouveau
<ssvb> I believe on the xorg side it is dri2
<hno> jemk, I always publish keep my notes public. Haven't seen it confuse anyone yet.
<jemk> ssvb: but dri needs help of driver?
<jemk> hno: but non working code massively changing all the time?
<mnemoc> mirrors are ready, cgit tomorrow
<hno> jemk, same thing really.
<ssvb> jemk: if you can provide ump buffers with decoded frames, then dri2 is going to work just fine
<wingrime> mnemoc: please read backlog (our conv with jemk)
<ssvb> jemk: and ump buffers can be easily created for physically contiguous memory blocks (just like they are created for the framebuffer)
<mnemoc> he can move the stuff into our gh whenever he wants
<jemk> ssvb: buffers are from cedar kernel driver, don't know if it can be changed to ump easy
<mnemoc> then I add it to the mirror
<ssvb> jemk: you can wrap the whole cedar reserved area as an ump buffer, have a look at
<oliv3r> mnemoc: very cool
<ssvb> jemk: of course this is not acceptable for the mainline, but it is easy to do
<ssvb> jemk: the mainline kernel would need some ump replacement (dmabuf?)
<Turl> mnemoc: gh mirrors? yay :)
<techn_> ssvb: I think drm has that kind replacement?
<ssvb> jemk: you can add an ioctl to cedar kernel driver, which would do something similar and return ump secure_id
<ssvb> techn_: it is complicated, they are always inventing new things -
<wingrime> ssvb: disp can resize overlays by fraction number?
<ssvb> techn_, jemk: but I don't see any substantial difference between the ump secure id and dmabuf handle, they can be used in similar ways and abstracted by a thin wrapper
<ssvb> wingrime: overlays can be scaled to arbitrary size
<jemk> ssvb: ok, i don't realy understand how ump works, but i think i'll find out
<pitillo> hello, does someone know why this happens with cb2 when booting from sd? (kernel 3.4.43) [mmc-msg] sdc0 power off
<oliv3r> need a little more info there; or is that everything that is displayed with regards to sdc and mmc?
<ssvb> jemk: how does cedar allocate memory for the frames?
<ssvb> jemk: can we override this allocation and provide the physical addresses for the memory blocks?
<oliv3r> [ 4.327782] mmc0: mmc_sd_init_card() failure (err = -110)
<oliv3r> <6>[mmc-msg] found data error, need to send stop command
<oliv3r> pitillo: try a different SD card
<pitillo> oliv3r: thank you very much
<jemk> ssvb: i get the address and size pf reserved memory from kernel and use own allocator there
<jemk> ssvb: so everything should be possible, also using other allocator directly should be possible
<ssvb> jemk: is allocation done for each new frame?
<jemk> ssvb: whenever vdpau client creates new videosurface
<jemk> ssvb: mplayer e.g. only does this once till it has enough videosurfaces
<ssvb> jemk: ok, it might be a bit tricky, but still doable
<ssvb> jemk: we would need a way to know which frame exactly to make visible in a window, when the x server gets DRI2SwapBuffers request
<jemk> ssvb: ah, i missed one step, in real world videosurfaces are mixed by videomixer onto outputsurfaces which are shown then...
<ssvb> jemk: maybe some alternative method to set/query this information might be needed (or maybe not)
<jemk> ssvb: only my poc shortcuts this
<jemk> ssvb: but this should be done by disp/g2d in theory
<wingrime> jemk: color key
<ssvb> jemk: disp is managed by the x server (to that we have no glitches with the windows stacking order, their sizes, etc.), and g2d can be used by either client side vdpau or by the x server
<ssvb> jemk: for dri2 we just need to know when some buffer with pixels needs to be shown in a window
popolon has joined #linux-sunxi
Soru has quit [Ping timeout: 260 seconds]
<wingrime> jemk: also , I hope that your code will work on a13/a20.....
ZetaNeta has quit [Ping timeout: 245 seconds]
Soru has joined #linux-sunxi
BJfreeman has quit [Ping timeout: 268 seconds]
wingrime has quit [Ping timeout: 245 seconds]
ZetaNeta has joined #linux-sunxi
jemk has quit [Remote host closed the connection]
geecko has quit [Quit: Quitte]
vicenteH has joined #linux-sunxi
jabjoe1 has left #linux-sunxi [#linux-sunxi]
\\Mr_C\\ has quit []
akaizen has quit [Remote host closed the connection]
robb83 has quit [Quit: - A hand crafted IRC client]
\\Mr_C\\ has joined #linux-sunxi
popolon has quit [Quit: Quitte]
hipboi_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 256 seconds]