<rm>
sadly I didn't save complete commit diff from 'before'
torindel has quit [Ping timeout: 240 seconds]
ZaEarl has joined #linux-sunxi
ZaEarl has quit [Ping timeout: 245 seconds]
<mnemoc>
rm: the stage branches are rebased
<mnemoc>
rm: they are intended for devs to test new changes, not to users....
<mnemoc>
if a change is good for users, it has to be moved out of the stage
<mnemoc>
well... not out. merged to the stable branch
n01 has joined #linux-sunxi
<mnemoc>
stage/ will still have it, but below it's base
<wingrime>
mnemoc: dma It like depedency hell
<mnemoc>
do we need to make the clocks and irq "civilized" first?
<wingrime>
you still not acepted pm unification
<wingrime>
and I now working on dma unfication
<wingrime>
and only clocks are pending in mach
<mnemoc>
the dsl in my new apartment still works like crap... it's syncing at 3M (when syncs at all) while it should be 16-18M :<
<wingrime>
mnemoc: sunxi make some HW dst/src bits different on sun4i/sun5i
<mnemoc>
using the phone as modem atm. will merge the pm thing soon
<wingrime>
so i fight with depedeny hell
<wingrime>
mnemoc: pm 100% same , there must be no-problem with it
<wingrime>
if my patch to outdated hramrach made same
<wingrime>
*too
<wingrime>
hramrach made simular patch becose he did't noticed this one
rellla has joined #linux-sunxi
<wingrime>
I wonder That I didn't forget something for dma
<rm>
mnemoc, I want to build kernels from stage
<rm>
tell me what do I use as a version string then
<rm>
date/time?
<rm>
I thought commit checksums were stable
<mnemoc>
rm: rebase makes new commits
<mnemoc>
timestamp is probably the only way
shineworld has joined #linux-sunxi
<mnemoc>
rm: can you send patches to improve our defconfigs?
vicenteH has quit [Read error: Connection reset by peer]
paulk-desktop has joined #linux-sunxi
<shineworld>
Can script.bin to be used to define display geometry (I'm thinking to Screen DPI, or X DPI, Y DPI) ? Actually I've used ro.sf.lcd_density=120 to match my lvds display and avoid the SystemUI exception in loading StatusBar but at boot geometry recovery by SurfaceFlinger from driver are wrong: Mar 31 23:40:14 <shineworld>I/gralloc ( 88): xres = 800 px
<ssvb>
wingrime: you just change wrong code to a bit different wrong code :)
<oliv3r>
mripard_: hm, the header says ioread? is a 'generic' interface for doing new-style memory-mapped or PIO access. e.g. isntead of read[bwl]
<wingrime>
ssvb: dma code is crap
<wingrime>
ssvb: but we need unify it at all
<oliv3r>
wingrime: fair point
<ssvb>
wingrime: the needed cache clean and invalidate operations are different for reads and writes
<mripard_>
oliv3r: hmmm, I don't quite see how it relates to memory mapping, once the memory has been mapped, it's mapped, ioread or not.
<wingrime>
ssvb: At first I must implement new finity-state machine
<mripard_>
but ioread makes endianness conversions, so, do whatever you want :)
<ssvb>
wingrime: maybe we can just reuse some dma framework from the mainline?
<oliv3r>
mripard_: i'm just learning :p and found that comment in iomap.h which implied 'read[bwl] is old, use ioread' :)
<wingrime>
ssvb: yes, but i still see only cache helper
<wingrime>
ssvb: and more interesting, try use dma_mapper as replacment for invalidation
<wingrime>
ssvb: also sound code did't do any flush
<oliv3r>
wingrime: sound is also quite a mess :p
<paulk-desktop>
seems like my a13 tablet touchscreen driver is missing from the linux-sunxi tree
<paulk-desktop>
it's elan_ktf2k
<ssvb>
wingrime: ok, just try not to reinvent wheels unless this is really necessary :)
<wingrime>
ssvb: I seeking good example
<wingrime>
ssvb: for drvier base
<wingrime>
ssvb: using kernel framweork means break wemac ,ether, nand, sound
<wingrime>
ssvb: I uart also have full HW dma support
<ssvb>
wingrime: is it difficult to change wemac, ether, nand, sound to use the standard dma framework?
<wingrime>
ssvb: It difficult test that sitty codebase
<wingrime>
*shiity
<wingrime>
*shitty
<oliv3r>
mripard_: heh; checkpatchs complains to include linux/io.h instead of asm/io.h :)
ganbold__ has joined #linux-sunxi
<wingrime>
ssvb: we need do cache invalidate in more selective way
<wingrime>
that can me great performance impovment
rundfunk has quit [Quit: leaving]
<rellla>
oliv3r?
eebrah_ has quit [Ping timeout: 256 seconds]
eebrah has quit [Ping timeout: 240 seconds]
<oliv3r>
rellla: hey, what is the allwinner xbmc status?
<oliv3r>
rellla: your image still valid? is it only a rootfs? does it work in combination with our 'hwpack'?
<oliv3r>
wingrime: you responded before I could. THough I see it took me 25 minutes to reply. ouch. anyway, I don't think it'll be backported. the code to support sid (for soc-detect) is currently in the kernel allready
<rellla>
oliv3r: never tried the hwpacks. last week i set up my cubieboard with debian rootfs and new sunxi-3.0.
<rellla>
xbmc from my repo run as expected. not better than half a year ago, but even not worse.
<n01>
guys, in which branch do we have sunxi.dtsi?
<oliv3r>
rellla: well the hwpack only contains closed binaries (mali, cedarx) the kernel and a script.bin
<oliv3r>
rellla: the theory is, you have a 'clean' rootfs, without kernel or closed source binaries or modules. you'd simply extract the hwpack and you have the missing bits :)
<rellla>
i know. but i bundled this things on my own.
<oliv3r>
n01: you probaly have to do git remote add mripard git://github.com/mripard/linux.git
<oliv3r>
rellla: do you still have a precompiled binary?
<oliv3r>
do you have a more recent compilation of that?
<oliv3r>
i guess i need to put that ontop of an existing rootfs
<rellla>
yes. "stable" Frodo. as far as you can say "stable" ;)
<oliv3r>
rellla: this ontop of X? or just the framebuff
<rellla>
fb!
<oliv3r>
rellla: ok; well there's a session/XBMC.desktop :)
<n01>
oliv3r: are u able to boot already a kernel with dt?
<oliv3r>
n01: mripard_ can, I don't have uart so if I could, i can't see anything
<rellla>
this november version should work in a way, too - you want to test the fb_scaler patch?
<n01>
great
<oliv3r>
rellla: nah, just want to see if I can get it to run at all
<rellla>
the binary is crosscompiled with debian/sid-rootfs. putting it on top of that rootfs should work.
<rellla>
hopefully.
* rellla
will update xbmca10-binary as soon as he boots up his dev-pc...
<oliv3r>
rellla: awesome :)
<oliv3r>
rellla: even better if you'd make it compatible with hwpacks :p
<oliv3r>
rellla: i'll find a stable sid-rootfs
<oliv3r>
and uptodate
<oliv3r>
ah, i just build my own ;)
<rellla>
maybe it's even better if you build xbmc at your own, too. you surely have some toolchain there...
<oliv3r>
rellla: ideally, we expand the sunxi-bsp to build xbmc ;)
<oliv3r>
rellla: but I know xbmc will need libs etc
<rellla>
oliv3r: afaik you have to install the libs in addition to the rootfs. don't know, if my vanilla rootfs at dl.linux-sunxi.org/rellla is including them.
<rellla>
best way would be a "dpkg -i" or "apt-get install"
<rellla>
in the future
<rellla>
but as long there is no progress with allwinner/cedarx, xbmc-development will stay stopped as it is for a few months.
<oliv3r>
rellla: aye :( 'they are working on omx support'
<oliv3r>
rellla: need more in the RE department :)
<rellla>
oliv3r: as long as they work on any support ...
<oliv3r>
rellla: :S
<oliv3r>
rellla: they are busy with A31 and A20 so could be the new driver will encompase those aswell
<wingrime>
What a hell "copyright" issue on ML, I added my copyright to new files that AW have no deal with
<wingrime>
it stupid add AW copyright to new files
<rellla>
oliv3r: i think we have to wait... so i'll try to compile a stable vdr 2.0 in the meanwhile. maybe i'll find some time to plug my cardreader to the cubieboard ;-)
<oliv3r>
wingrime: i think technically, you are 'copying' the file (renaming) and the contents is still AW copyright. But i'm no copyright guru :)
<oliv3r>
rellla: ;)
<wingrime>
oliv3r: with is GPL we can wipe AW copyright and write linux-sunxi
<oliv3r>
wingrime: I do not know :(
<oliv3r>
wingrime: in the end, It will not matter, since the dma code has to be rewritten anyway, and then it absolutly will be no aw copyright anymore
<wingrime>
oliv3r: this is a header I don't know how can defines be copyrighted stuff
<oliv3r>
wingrime: i don't know either :(
<oliv3r>
i think headers aren't copyrightable anyway
<wingrime>
oliv3r: dma HAS to be rewrited anyway but when, what is a guestion
<oliv3r>
wingrime: when you have time? :p
<wingrime>
oliv3r: There was old story with oracle thay make from protocol magic numbers a poem with copyright
<wingrime>
oliv3r: you must send "poem" to database to connect
<wingrime>
poem are copyrighted)))
<oliv3r>
wingrime: lol really? was the poem in the 'header'
<wingrime>
I see. I just need alloc non cached memory
<wingrime>
))
<wingrime>
that most funniest thing
<wingrime>
I broke nand , wifi and sound still works
<wingrime>
strange
gzamboni has joined #linux-sunxi
<oliv3r>
n01: also Turl is rumoured to work on a wdt for mainlining efforts
<oliv3r>
wingrime: oh appearantly matt porter (mdp) is working on DMA a bit :)
<n01>
oliv3r: yes I know that unfortunately he is not here to ask
<n01>
or at least he is afk
<oliv3r>
n01: easter today; probably around tomorrow again :)
<oliv3r>
n01: he's usually reasonably active here
<n01>
:) ok, I'll ask him tomorrow then ... in any case it is a good exercise (I'm really not familiar with dt and this is a good occasion to practice a bit)
shineworld has joined #linux-sunxi
<oliv3r>
n01: ;)
<oliv3r>
n01: any particular reason your after the wdt?
<oliv3r>
anybody familiar with deboostrap?
<n01>
oliv3r: not really, I'm not used to kernel space and the wdt driver seemed easy enough to start with
<bfree>
oliv3r: re debootstrap I'd say: /topic - Don't ask to ask. Just ask! ;)
<oliv3r>
n01: okay :) your trying to get it mainline ready?
<n01>
oliv3r: yep, also dt-ready
<n01>
but obviously if Turl is already working on it I'll leave it to him
<oliv3r>
n01: turl does have a plate full! :)
<oliv3r>
n01: i've just submitted my first attempt at the 'sid' driver, i think it is the world most simplest driver ;)
<oliv3r>
n01: parallel to that, i've started working on the PWM driver
<n01>
I have seen it and I'm jealous :P
<oliv3r>
n01: but this is all to get stuff into mainline; so 3.8+
<oliv3r>
n01: heh, we'll see; still haven't gotten any feedback :)
<mripard_>
oliv3r: where did you send it?
<oliv3r>
mripard_: sunxi ml
<oliv3r>
i'll send you a copy
<mripard_>
'k
<mripard_>
please cc me on this, I'm interested ;)
<oliv3r>
mripard_: heh; i'll gladly need your feedback
<oliv3r>
i'll send you my own version that's inlined with questions; and on v2 i'll directly CC you