ChanServ changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See http://linux-sunxi.org | https://github.com/linux-sunxi/ | Logs at http://irclog.whitequark.org/linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
anunnaki has quit [Remote host closed the connection]
tesingh has joined #linux-sunxi
torindel_ is now known as torindel
tesingh is now known as torqu3e
christopher has joined #linux-sunxi
christopher is now known as Guest91531
hramrach has quit [Ping timeout: 276 seconds]
libv has quit [Ping timeout: 248 seconds]
libv has joined #linux-sunxi
hramrach has joined #linux-sunxi
RaYmAn has quit [Ping timeout: 256 seconds]
Guest91531 is now known as anunnaki
RaYmAn has joined #linux-sunxi
torqu3e has quit [Ping timeout: 245 seconds]
rz2k has joined #linux-sunxi
rz2k has quit [Read error: Connection reset by peer]
rellla has joined #linux-sunxi
wingrime has joined #linux-sunxi
nicktime has joined #linux-sunxi
hansg has joined #linux-sunxi
<mripard> mnemoc: ok. You're free to comment on my patches :)
rz2k has joined #linux-sunxi
paulk-desktop has joined #linux-sunxi
<mnemoc> mripard: i'm kind of allergic to those MLs with big people...
<mnemoc> and my comment is likely to sound as a ranting against miniand and similar resellers
<mripard> well, no
<mripard> just send a mail saying "the reseller is <I forgot his name>"
<mripard> and that's it
<mripard> look at Turl, he didn't hesitate to tell me I did crap on my patches :)
<mnemoc> ok :)
<mripard> (and he was definitely right :))
<mnemoc> *g*
<mnemoc> mripard: replied
<mripard> thanks :)
<Turl> mripard: :P
<Turl> mripard: you can pick 54093a7 from my tree to get clock phandle support
<mripard> Turl: that was what I was able to ask you by mail
<mripard> s/able/about/
<Turl> :)
<mripard> Turl: what's your git tree again ?
<mripard> (I'll add it to my remote this time, I swear...)
<mripard> thanks
<mnemoc> Turl: any feedback from bitbucket regarding their crappy performance with bit repos?
<Turl> mripard: you're welcome :)
<Turl> mnemoc: they acknowledged the problem and told me they have internal issues tracking the slowness https://bitbucket.org/site/master/issue/6550/bitbucket-performs-badly-with-big-git
<Turl> mripard: please push a tag/branch to your repo after you send v2, so I can rebase the uart gate dt stuff over it
<mripard> Turl: done
<mripard> it's the sunxi-3.9-wip-uarts branch
<Turl> great, thanks
<mripard> I'll probably post it only tonight (in ~10h)
<mripard> but, I'll send it as is
<Turl> I probably won't be rebasing it until tomorrow or on the weekend, so no hurry
<gzamboni> mnemoc ping
<gzamboni> mnemoc can i /q ?
<mnemoc> gzamboni: sure
<theOzzieRat> mnemoc, what happened to stage/sunxi-3.0 branch?
<mnemoc> theOzzieRat: it was merged, and a new hasn't been spawned yet
<mnemoc> -r1 was tagged just after the merge
<mnemoc> will create a new stage branch for hramrach's nand partition code and other patches which are on-hold
<mnemoc> tonight most probably
<theOzzieRat> ah, ok. not that up on how all this works :(
<mnemoc> neither I am
<mnemoc> :)
<theOzzieRat> i posted 2 spi patches, and have another one ready and almost have a patch for spi full duplex done
<mnemoc> the stage branches are just a central place to test new changes. they aren't intended for massive usage, only for devs
<mnemoc> theOzzieRat: if i haven't merged those spi patches by the time you fix full duplex, please resubmit the whole serie
<mnemoc> theOzzieRat: also, it might be a good idea to see allwinner spi changes in the lichee/ branches of my github
<mnemoc> there are probably interested changes, and as you are already into sunxi-spi, you are perfect to review that part :)
<theOzzieRat> ok, will have a look. all my patches are incrememtal so it shouldn't be a major issue
<theOzzieRat> i have full duplex for short transfers working, just fighting with ddma for longer ones
<mnemoc> :)
<theOzzieRat> I also updated the fex guide earlier as discussed if you would like ot review
<theOzzieRat> ot=to
<mnemoc> theOzzieRat: mripard found the mA levels of the drive field. it's in the gpio document on mainline
<theOzzieRat> if you have a link I will update or ask mripard to do it?
<theOzzieRat> for most purposes i suspect <default> is optimal
<theOzzieRat> mnemoc, which branch on your githib in particular has spi changes? I can't find them.
<mnemoc> liche/a*-dev branches *might* have spi improvements/fixes
<theOzzieRat> ok, had a look and no changes to spi_sunxi.c for 8 months, so I think there isn't much going on at the moment. I also see Lorenzo Nava's irq patch didn't get merged.
<theOzzieRat> mripard, thanks. I will add to the fex guide in the morning, its damn close to sleep time here.
<theOzzieRat> I read somewhere that <deafault> for that means 1
<mripard> probably, I'm definitely not a fex expert :)
<mnemoc> theOzzieRat: -1
<mnemoc> which means, "don't change"
<theOzzieRat> -1 means deafault
<mnemoc> or 0xffffffff
<theOzzieRat> the whole concept of "don't change" doesn't work, as you have to ask "don't change from what"? There has to be a default setting.
<mnemoc> keep SoC defaults
<mnemoc> if you can that default by setting it to <default> and then use the `pio -m` tool to see the current value in the register
<theOzzieRat> so back to hoping the Allwinner datasheet is somewhat accurate, which it isn't.
<mnemoc> `pio -m` will tell you the truth. the datasheet might
<theOzzieRat> ok, pio -m is new to me
<mnemoc> from sunxi-tools, running native it will work directly via /dev/mem
<mnemoc> to see the full list of defaults you can start in fel, dump the register, and pass the dump to pio
<mnemoc> sunxi-tool has a fel-gpio thing to help on that
<mnemoc> it was used to test the first a13-olinuxinos
<theOzzieRat> sounds like sonething to look at tomorrow, almost midnight here.
<mnemoc> good night :)
<mnemoc> once you get the defaults, you can extend [[A10/PIO]] accordingly
<theOzzieRat> night dude :)
ganbold_ has joined #linux-sunxi
<hramrach> mnemoc: did some led trigger that can light a led based on IO to *any* block device
<wingrime> mnemoc: witch section in fex used for wifi
<hramrach> tested some earlier version and it works reasonably
<wingrime> ?
<hramrach> wingrime: what device?
<hramrach> there is no WiFi on A10 so no fex for it
<wingrime> my a13 tablet rtl8188e IC
<hramrach> is that USB ic? other if ic?
<wingrime> sdio
<hramrach> then one of the mmc sections presumably
<hramrach> do you have the original fex?
<wingrime> so where connection wiring configured ?
<hramrach> I don't have such device so I would not know
<wingrime> I don't think it enough build module from it
<hramrach> that's a huge patch
<wingrime> that wifi driver is huge
<hramrach> see rtw_drv_init in sdio_intf.c
<hramrach> that's where you pass a sdio if to the wifi driver
<hramrach> the caller of this function defines what sdio if is used
<hramrach> it's passed into sdio_register_driver
<hramrach> so presumably generic kernel sdio layer handles that
<hramrach> does the thing build?
<wingrime> not tryed yet
<hramrach> that patch touches unrelated stuff like disp, btw
<wingrime> I just try copy driver folder and patch Kconfig / Makefile
<wingrime> I hope it can be merget mainline
<hramrach> 7  drivers/video/sunxi/hdmi/hdmi_core.c
<hramrach> 0 additions, 7 deletions not shown
<wingrime> I have other device and now I wanna add wifi and ts support
<wingrime> not perfect at all
<hramrach> merging something yhis huge upstream will be very hard
<hramrach> 200klocs
<wingrime> but I have this IC, there is two cases my fork/ mpstream patches
<wingrime> and now new strange code style demands
<hramrach> well, making it work is the first step
<wingrime> http://pastie.org/6411676 lsmod form a13
<wingrime> some drivers realy strange
<hramrach> does the driver nevere finish loading?
<wingrime> what?
<hramrach> 8188eu 695991 1 - Loading 0xbf0d1000
<wingrime> funny
<wingrime> wait
<wingrime> I try make lsmod again
<hramrach> kernel messages?
<wingrime> this is a stock android
<wingrime> wiat I make dmesg
<wingrime> all normal)) I just make lsmod too early
<wingrime> all "Live"
<hramrach> did you even boot Linux on the tablet?
<wingrime> debian works quick
<wingrime> No wifi no ts
<hramrach> good so far
<hramrach> so does the driver build?
<hramrach> and where did you get the driver you added?
<wingrime> not so quick.....
paulk-desktop has quit [Quit: Ex-Chat]
nicktime has quit [Quit: Leaving.]
<wingrime> oh "eu" is usb "es" SDIO
<wingrime> it simpler than I think
<wingrime> arm-linux-gnueabihf-ld: drivers/gpu/mali/mali/linux/mali_osk_memory.o: invalid string offset 4293984260 >= 229 for section `.strtab'
<wingrime> non regular bug
<wingrime> hramrach: builds ok
<hramrach> then you can try loading the driver and see how it crashes ;-)
<wingrime> look like it connected to USB
<hramrach> you could see it with lsusb then
<wingrime> android not have lsusb wait debian
<wingrime> hramach: .... it works
<wingrime> hramach: yes it usb connected
<mnemoc> wingrime: running the make again should solve that
<mnemoc> wingrime: it's common. but no idea what causes it
<wingrime> mnemoc: 8188eu works but I have no idea how to push it upstream
theborger has joined #linux-sunxi
<mnemoc> we have tons of those not-upstreamable rtl junk
<mnemoc> upstream uses rtlwifi
<wingrime> mnemoc: is there any sence to make patch and push to ML? (I WILL not fix style for such BIG thing)
<mnemoc> wingrime: linux-wireless is probably the right place for that question
<wingrime> mnemoc: I talking about linux-sunxi ML
<mnemoc> from `https://github.com/amery/git-import-help` run `x_file.sh` over all the files you want to include
<mnemoc> something like git ls -zoim --exclude-standard drivers/net/wireless | xargs -0r ../git-import-help/x_file.sh
ZaEarl_ has quit [Ping timeout: 240 seconds]
<mnemoc> `../git-import-help/x_ls.sh drivers/net/wireless` should do a similar pre-commit cleanup
<mnemoc> bbl
<mnemoc> it's a very basic sanitization routine, but at least it will save us from vomiting when seeing your commit
<wingrime> I try
<hramrach> mnemoc: the driver is obviously not correctly indented to say hte least
<hramrach> also 8188[cd] is handled by 8192cu or somesuch
<hramrach> so probably poking somebody who has some idea how that driver works might give some results
<hramrach> as in just adding a device ID or a small patch if the hardware is actually similar
<wingrime> 8192cu not worked
<wingrime> it loads but not see
<wingrime> any interface
<hramrach> it is not ttying to probe the card because it has an ID not known to that driver
<hramrach> *trying
<wingrime> maybe
<wingrime> it will be great if it will works
<Turl> it's more likely your usbc is powered off
<Turl> and not the driver's fault
<Turl> turn it on on script.bin and try again :)
<wingrime> Turl: this driver loads and work
<hramrach> no, it's more likelly the hardware is too new
<Turl> yes, the rtlNNNN from realtek itself have allwinner glue to turn on the usbc on load
<Turl> the rtlwifi from mainline doesn't
<hramrach> that's also possible
<mnemoc> hramrach: that script will 1) fix ^M, 2) reencode into utf-8, 3) chmod -x, 4) fix obvious whitespace issues, 5) add firmwares and other forbidden files to the stage as well
<mnemoc> the usbc issue can be solved changing the initial state in script.bin
<Turl> that's the glue
<mnemoc> changing the .fex is more civilized :p
<wingrime> mnemoc: "git ls -zoim" what did you mean ? "git ls" not exists ....
<mnemoc> wingrime: meh. personal aliases :<
<mnemoc> wingrime: use x_ls.sh and change `git ls` for `git ls-files`
<mnemoc> correcting the scripts now to not rely in my aliases
<hramrach> neither is really civilized
<hramrach> and not powering the port by default probably saves power when the wifi is unused
<mnemoc> change to x_ls.sh/x_file.sh pushed
<mnemoc> need to leave now. back in 2h
<mnemoc> i tend to forget (by default) git doesn't have st,ci,co,cp,ls commands
hansg has quit [Quit: Leaving]
<hramrach> wingrime: the import tool appears to work: https://github.com/hramrach/linux-sunxi/tree/rtl8188eu
<hramrach> but I don't think it would prevent vomitting when looking at the code
vinifm has joined #linux-sunxi
paulk-desktop has joined #linux-sunxi
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
<wingrime> hramrach:send it to ML ))) thanks
rz2k has quit []
rellla has joined #linux-sunxi
rellla has quit [Client Quit]
vinifm has quit [Quit: Saindo]
<wingrime> there is anyway suspend debian
<wingrime> ?
<mnemoc> hramrach: :)
<wingrime> why raster framebuffer so fast ?
<hramrach> I changed the size of a kernel struc. need new modules /o\
<hramrach> now I get segfault on boot :s
<hramrach> maybe I did not check for null somewhere
ganbold_ has quit [Remote host closed the connection]
<wingrime> mnemoc: I send new version if touchscreen patch
<wingrime> please review ))
<wingrime> techn__: you here ?
<wingrime> http://pastie.org/6414346 genus code at drivers/mmc/sunxi-host/host_op.c:585!
<wingrime> we crash when no sdio section in fex
<mnemoc> I wrote another API to access the fex <https://github.com/linux-sunxi/linux-sunxi/blob/sunxi-3.0/arch/arm/plat-sunxi/include/plat/script.h> it should make the driver's code cleaner... I think
<wingrime> menmoc: I remove BUG_ON and try handle situation
<mnemoc> ok
<mnemoc> wingrime: you seem to have added an } too much in ctp_free_platform_resource() :)
<mnemoc> second chunk
<wingrime> you can fix it ?
<wingrime> or i have to remail ?
<wingrime> i just forget remove barket as script recomendation
<techn__> wingrime:
<techn__> looks good
<techn__> one comment in addion that bracket.. static variables are initialized 0 automaticly
<wingrime> I know
<wingrime> but reload driver
<techn__> shouldn't it unload binary then?
<wingrime> statics are global ?
<wingrime> as i remember
<techn__> inside that binary.. atleast in user space :p
<mnemoc> "global" and "reference" aren't C terms :)
<wingrime> zero variables is a "C" good tone ))
<wingrime> C not garante that nuinitialized variable will 0
<wingrime> mnemoc: I have to remail or not ?
<mnemoc> i can apply it and fix the bracket and static = 0 tonight after spawning the new stage
<wingrime> mnemoc:ok , but are you very that statics zeroed after rmmod/insmod ?
<mnemoc> 98% sure :p
<wingrime> mnemoc: I send fix for BUG_ON, now it simply try load driver and fails, no mmc remount ,
rellla has joined #linux-sunxi
rellla has quit [Remote host closed the connection]
wingrime has quit [Ping timeout: 240 seconds]
focus_it has joined #linux-sunxi
torqu3e has joined #linux-sunxi
<hramrach> how do I tell where the kernel crashed?
<hramrach> I have the vmlinux and the oops but when I disassemble the function I have no idea wth is that doing
<mnemoc> paste the begining of the oops
paulk-desktop has quit [Quit: Ex-Chat]
focus_it has quit [Quit: Leaving]
ZaEarl_ has joined #linux-sunxi
mab has joined #linux-sunxi
<Turl> mripard: ping
gzamboni_ has joined #linux-sunxi
gzamboni has quit [Ping timeout: 245 seconds]