ChanServ changed the topic of #linux-rockchip to: Rockchip development discussion | Wiki at http://linux-rockchip.info | Logs at http://irclog.whitequark.org/linux-rockchip | ML at http://groups.google.com/group/linux-rockchip
UndrWater has joined #linux-rockchip
UndrWater has quit [Remote host closed the connection]
assember has joined #linux-rockchip
levd has joined #linux-rockchip
BorgCuba has quit [Ping timeout: 246 seconds]
vickycq has quit [Read error: No route to host]
naobsd has joined #linux-rockchip
Anarchy has joined #linux-rockchip
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 260 seconds]
levd1 is now known as levd
afaerber has quit [Ping timeout: 260 seconds]
cnxsoft has joined #linux-rockchip
afaerber has joined #linux-rockchip
hipboi has joined #linux-rockchip
assember has quit []
hipboi has quit [Ping timeout: 245 seconds]
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 260 seconds]
levd1 is now known as levd
hipboi has joined #linux-rockchip
<Anarchy> Alright I need some assistance. I used rkcrc to create a kernel.img to write to the kernel partition. I used dd to write it now when I attempt to boot up using ctrl+u it just dings at me and goes back to os verification off screen. Anyone have any ideals where I might have screwed up
<naobsd> Anarchy: I couldn't understand this part "when I attempt to boot up using ctrl+u it just dings at me and goes back to os verification off screen."
<naobsd> which device are you using?
<Anarchy> hisense chromebook
<naobsd> kernel.img format(rkcrc -k) is only for RK proprietary loader
<naobsd> sorry
<naobsd> kernel.img is only for RK proprietary loader for Android
<naobsd> chromebook uses different loader (unless you replace it)
hipboi has quit [Ping timeout: 240 seconds]
hipboi has joined #linux-rockchip
<Anarchy> naobsd, both useless as they just copy the kernel over, I will do some digging tomorrow and see what I can find.
<Anarchy> Thanks for the info, atleast gives me a starting point
<naobsd> I don't have any chromebook, so I'm not sure about it
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 260 seconds]
levd1 is now known as levd
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 255 seconds]
levd1 is now known as levd
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 245 seconds]
levd1 is now known as levd
levd has quit [Remote host closed the connection]
levd has joined #linux-rockchip
hipboi has quit [Ping timeout: 264 seconds]
hipboi has joined #linux-rockchip
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 255 seconds]
levd1 is now known as levd
naobsd has quit [Quit: naobsd]
hipboi has quit [Read error: Connection reset by peer]
cnxsoft1 has joined #linux-rockchip
cnxsoft has quit [Ping timeout: 260 seconds]
cnxsoft1 is now known as cnxsoft
hipboi has joined #linux-rockchip
ckeepax has quit [Ping timeout: 240 seconds]
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 245 seconds]
levd1 is now known as levd
kapouer has joined #linux-rockchip
kapouer has quit [Remote host closed the connection]
marcr has joined #linux-rockchip
<marcr> Hi guys
<marcr> Any news about nand controller driver?
<marcr> Can anyboty clarify if all rk3xx soc's has the same version of nandc ip? Because i see that there is doc in px2 trm for controller (v6.0.0 at least)
ckeepax has joined #linux-rockchip
<marcr> Does anybody started working on the driver ?
<marcr> I ask because i've read the docs 5 times now and i'm thinking about writing it myself. There are still parts i don't understand (ECC) but i'm getting closer ...
<marcr> ... so if someone have some info in this topic please share :) ... i'll be grateful
levd1 has joined #linux-rockchip
<aborche> :) everybody sleeps :)
<aborche> please share a link :)
levd has quit [Ping timeout: 255 seconds]
levd1 is now known as levd
<aborche> huh. i got rk903v, and test it. tooooo hooooot for normal work :(
gb_master has joined #linux-rockchip
gb_master has quit [Client Quit]
<mmind00> marcr: the nand IPs are largely similar ... but I don't think anybody has looked into a real nand-driver yet (or has plans to do so)
naobsd has joined #linux-rockchip
<marcr> mmind00: thanks
<marcr> mmind00: maybe rockchip will release docs for newer versions of nand controler
<naobsd> I think chipspark rk3288 TRM also has NAND section
srao_ has joined #linux-rockchip
<naobsd> it's incomplete TRM but it has "Chapter 9 NandC"
mcan_ has joined #linux-rockchip
srao has quit [*.net *.split]
mcan has quit [*.net *.split]
_jungh4ns has quit [*.net *.split]
srao_ is now known as srao
_junghans_ has joined #linux-rockchip
<marcr> naobsd: thank's ... i missed that
<marcr> rk3288 has some higher version (v6.2.2)
karlp_ has joined #linux-rockchip
johnnyr_ has joined #linux-rockchip
naobsd has quit [Quit: naobsd]
_junghans_ is now known as _jungh4ns
ganbold has quit [*.net *.split]
johnnyr has quit [*.net *.split]
karlp has quit [*.net *.split]
Kruppe has quit [*.net *.split]
levd has quit [Ping timeout: 245 seconds]
<rperier> mmind00: you mean that nand IP between rk3x and rk3288 should work in the same manner ?
<mmind00> rperier: largely yes ... aka similar, but haven't compared in-depth
<rperier> ok
<rperier> that's a good point...
<sjoerd> rperier: any more luck with getting network going on your firefly's u-boot ?
<rperier> sjoerd: no, and I have exactly the same problem with my rock2's u-boot
<rperier> however, the rest works perfectly fine
<sjoerd> Oh really? thats interesting
<rperier> yes... I wonder if it might have something to do with the link speed...
<rperier> because I have a link at 100/Full...
<sjoerd> I did test with 100mbit/s at some point but usually use gbit
<rperier> (instead of 1000)
<sjoerd> The only differnt should be the clock configuration afaik
<sjoerd> did you try with a proper switch in teh end or only cross-link?
<rperier> yes, because both will use rgmii... (I mean, it is electrically wired... I cannot really change right?)
<rperier> I only tried with a cross-link, no switch
<sjoerd> Any change you could try that, i'd love that datapoint as well
<rperier> it works fine from linux
* sjoerd also checks what the other kernelci stuff uses
<sjoerd> ah the other kernleci lab with a rock2 and my u-boot branch also use gigabit networking
<sjoerd> rperier: I know, i might be overly worried here about the cross-link stuff indeed :)
<sjoerd> rperier: getting nandc going on old rk3xxx would be really quite cool, that's an enabler for so many devices
BorgCuba has joined #linux-rockchip
* sjoerd adds check with a cross-link to his todo
Ueno_Otoko has joined #linux-rockchip
<rperier> sjoerd: if I can help, feel free to ping me, I have my boards at home, but I can test things during the evening
<sjoerd> rperier: whats the git sha1 you tested with ?
<sjoerd> I'd like to retest that one as well just to be sure :)
Luke-Jr has quit [Excess Flood]
Luke-Jr has joined #linux-rockchip
gb_master has joined #linux-rockchip
Luke-Jr has quit [Read error: Connection reset by peer]
Luke-Jr has joined #linux-rockchip
<rperier> sjoerd: 3c66d84628936819bf186509909a80419dc19564
<rperier> yeah that's this one, the unique commit with "WIP" in the commit log :p
<rperier> (that's because my recipe in yocto is not committed yet into meta-rockchip, but I am sure that it is this one)
gb_master has quit [Remote host closed the connection]
jusched has joined #linux-rockchip
marcr has quit [Quit: Page closed]
<BorgCuba> rperier, whats the status of the mainline kernel for rk3xxx? Has usb-device mode been fixed already?
Ueno_Otoko has quit [Ping timeout: 264 seconds]
<BorgCuba> more specifically I meant for rk3188.
Ueno_Otoko has joined #linux-rockchip
<rperier> I did not re-test recently, I just know that there are some pending fixes about usbphy and usb in rockchip. Last time I had a look at it, it was supported on old socs no? (in fact, I was a bit disconnected these days, because I had new jobs opportunities...)
<BorgCuba> when I last tested it (about 3.18) it was not working
<BorgCuba> something related allocating fifo buffers failed
<rperier> you should retry with a recent kernel, lastest mainline is 4.3 now :)
<BorgCuba> Indeed, hopefully I find my dts file again ... for me it always takes some time to remember all that after a few months of absence
<aborche> rperier, and where this latest mainline located ?
<aborche> :)
<aborche> on which git ? :)
<aborche> BorgCuba can you share your dts ? :)
<aborche> for testing ofcourse
soimless has quit [Ping timeout: 265 seconds]
<aborche> current 3.18.25 cannot boot with radxarock-lite config for 3.18.11. some error about ext4 partition
soimless has joined #linux-rockchip
<BorgCuba> aborche, sure, when I find it
<BorgCuba> btw, I have radxa lite here too
<BorgCuba> aborche, are you sure the ext4 problem is related to the dts?
<aborche> yesterday we make a development board from cx919 fake - D908. and now i see all log data
<BorgCuba> what are your kernel parameters and how does your sd-card look like?
<aborche> BorgCuba, i don't know. but 3.0.36+ boot at this sd normally
<BorgCuba> make sure you have something like "root=/dev/mmcblk0pN" (N = your partition) and "rootwait" in your kernel parameters
<BorgCuba> looks like I deleted all my local linux-next copies
<aborche> ;(
<aborche> BorgCuba, this parameters exist in parameter.img
<BorgCuba> are you sure that the mainline kernel reads these parameters? I would try to hardcode them into the kernel (make menuconfig)
<BorgCuba> wait, I think uboot does this
<aborche> where i can get rk3188-box.dts for clean testing ?
<rperier> aborche: torvards's tree
<rperier> or linux-stable on git.kernel.org
<aborche> torvalds ?
<BorgCuba> or linux-next (kernel.org)
<aborche> ok. i will try.
<aborche> as i see only rk3188-radxarock.dts included :( without box version
<rperier> what do you mean by box version ? are you talking about pro vs. non-pro?
<rperier> tv box perhaps?
<aborche> no. i have unnamed rk3188 board. modified radxa kernel with corrected voltager work is ok. but nothing more kernels i can start on it.
<aborche> voltager = voltages
<aborche> i need clean version for rk3188 board, where radxa platform changes is not included
cnxsoft has quit [Remote host closed the connection]
<aborche> clean config i mean
<aborche> dts
<mrueg> w00t.
<aborche> =)
<mmind00> mrueg: that patch seems to be the talk of the day :-D
<mrueg> mmind00: commit message is golden :D
<mmind00> mrueg: yep
<aborche> guys how i can convert back decompiled dtb tree ? with pin aliases and etc ?
<mmind00> aborche: you cannot fully restore phandles from a compiled dtb ... to get a somewhat-similar source to your dtb, "dtc" is still the program to use
<aborche> i know, but values which i get from compiled dtb from kernel image differs with values which i get when compile my tree
<aborche> and i can't understand where is the problem ? different version of kernel or different version of includes
<aborche> i think dtb tree have cross compatibility between kernels. or not ?
<aborche> as example slitaz image as they say, succesfully booted on radxa rock and screen work is ok
<aborche> and at radxa forum julien run 3.10 at radxa rock with worked hdmi
<aborche> and i don't understand - how ? :(
<rperier> everytime I read "dtc" (for the devicetree compiler), it disturbs me... do you know, what does it stand for in french ? that's an acronym... like "in your ****" but in french... xD
<rperier> (I am serious, I know that dtc is not french at all, but you know... sometimes you find funny name like this one :D)
<rperier> anyway... ^^
Ueno_Otoko has quit [Ping timeout: 245 seconds]
jusched has quit [Ping timeout: 245 seconds]
<aborche> =)
Kruppe has joined #linux-rockchip
ganbold has joined #linux-rockchip
karlp_ is now known as karlp
tlwoerner has joined #linux-rockchip
cyteen_ has joined #linux-rockchip
cyteen has quit [Ping timeout: 272 seconds]
Ueno_Otoko has joined #linux-rockchip
bbelos has joined #linux-rockchip
<aborche> no luck :(
<aborche> also. kernel must use gpt partition or no ?
gb_master has joined #linux-rockchip
<aborche> current kernel, rockchip_defconfig config, rk3188-radxarock.dts
<mmind00> aborche: it already detected the one partition, so part-table should be ok already ... is your partition on the sd-card actually ext4-formatted?
<aborche> yep. this partition boot is ok when kernel is 3.0.36+
tlwoerner has quit [Quit: Leaving]
<aborche> http://pastebin.com/P8uQ2gPd and this log when i try to use ramdisk
<aborche> compiled with mkbootimg
<aborche> sorry. this log
Ueno_Otoko has quit [Ping timeout: 246 seconds]
<aborche> maybe i should use boot option with partition name ? without direct set to first partition ?
hipboi has quit [Ping timeout: 255 seconds]
<aborche> o! maybe my u-boot 2014.01 is too old for linux-next kernels ?
Anarchy has quit [Ping timeout: 260 seconds]
sunxi_fan has quit [Ping timeout: 240 seconds]
hipboi has joined #linux-rockchip
sunxi_fan has joined #linux-rockchip
Anarchy has joined #linux-rockchip
Anarchy has quit [Changing host]
Anarchy has joined #linux-rockchip
gb_master has quit [Remote host closed the connection]
gb_master has joined #linux-rockchip
gb_master has quit [Remote host closed the connection]
<BorgCuba> aborche, all three logs you posted are from your 4.4 kernel
zhecka has joined #linux-rockchip
soimless has quit [Ping timeout: 276 seconds]
zhecka is now known as aborche_
<aborche_> BorgCuba, i cannot boot any other kernel instead 3.0.36+ at my board.
<via> are there any issues with building an upstream linux kernel like 4.3 for the c201 chromebook (rk3288)? e.g. can i just use the stock config (plus make oldconfig) and the speedy dts?
<BorgCuba> aborche, I think I will try it too ...
soimless has joined #linux-rockchip
<aborche_> BorgCuba, thx. i'm trying get a worked version of any kernel last two week. and without any positive result :(
<aborche_> i found image from Slitaz, and get dtb file from it, but boot is unsuccesful :( I think many problems linked with dts file. freq, voltage, interface speeds
<aborche_> but without any worked image i cannot make some steps for migrating system to this kernel and test it.
<BorgCuba> btw, I found an old dts file on my hdd
<BorgCuba> here it is: http://pastebin.com/5q5AAxkQ
<aborche_> hm. this config content similar as rk3188-radxarock.dts
<BorgCuba> aborche, did you use a defconfig to configure the kernel?
JohnDoe_71Rus has joined #linux-rockchip
<BorgCuba> is there no support for the act8846 in 4.4-rc6?
<BorgCuba> I can see TI TPSxxxxx but no act8846?
<BorgCuba> hmm, is it maybe compatible to act8865?
<mmind00> via: depends on what level of functionality you need ... aka the eDP driver (needed for the internal display) is not yet upstream, but sitting on the mailinglists still
<mmind00> mmind00: so with current mainline you'd get hdmi-output only
<mmind00> via: ^^ not sure, why I addressed that to me ;-)
<via> mmind00: hmm. i washoping enabling the rockchip drm would allow the use of the framebuffer
<via> so that doesn't work?
<via> thats kind of a dealbreaker
<mmind00> via: the low-level stuff works, just the outputs are still a bit limited (aka hdmi)
<via> damn
<mmind00> via: if you really need edp, I guess I'd point you to my somewhat-stable branch: https://github.com/mmind/linux-rockchip/commits/devel/somewhat-stable
<mmind00> which includes the pending edp stuff + a rk3288_veyron_defconfig
<via> are there any specific commits i could merge?
<via> for the edp
<via> or i suppose i could just use that
<mmind00> via: if you want to stay close to mainline: https://github.com/mmind/linux-rockchip/commits/tmp/analogixdp_v10.1-veyron
<via> thank you, i'll give that a shot
<via> doesthe 4.2somewhat-stable tag have edp support?
<mmind00> via: that too, but still with the old edp implementation ... both branches linked above do use the new dp driver, that will hopefully make it into mainline soo
<via> ok
<via> so i guess you keep the somewhat-stable branch up to date, andi could just occasionally pull/build?
<mmind00> via: bleeding edge is more like it ... that is my dev-/integration-branch but where I try to stay true to its name ;-)
<via> cool. yeah,i only expect so much stablity from a setup like this..., but i guess more testing is better anyway
<mmind00> via: also I keep the gpu-support in there (which cannot go into mainline at all currently)
<via> rockchip-drm?
<via> or something else
<mmind00> via: nope, the mali stuff
<via> ah. does that require firmware outside of this?
<mmind00> via: a userspace binary libGLES* + x-server
<mmind00> via: see mali-driver and xf86-video-armsoc repos on github
<via> ok, thanks
<via> wasn't there an effort to make an open soruce mali driver?
<via> lima
<via> well that project isn't looking like its gotten a lot of work
<mmind00> also, lima was for the mali utgard series (200, 400) ... there was once a presentation about an unpublished driver called "tamil", for the mali midgards (like the T764 in the rk3288)
<Wizzup> it got a lot of work, it is not visible, via
<via> hmm, i don't know what to think about that, unreleased open source code? that whole project seems to have some political issues, at least thats what turns up when googling
<via> either way, i'll stick with the blobs for the moment
c0d3z3r0 has quit [Excess Flood]
c0d3z3r0 has joined #linux-rockchip
JohnDoe_71Rus has quit [Quit: KVIrc 4.2.0 Equilibrium http://www.kvirc.net/]
<aborche_> BorgCuba, rockchip_default or radxarock_default
<BorgCuba> aborche_, I cannot find this in 4.4-rc6
<aborche_> only from rockchip.fr
<BorgCuba> quite old
<BorgCuba> I'll try with that defconfig
gb_master has joined #linux-rockchip
<via> just build and installed a kernel from that branch and i don't appear to have a framebuffer. this happens to me a lot <_<
<via> mmind00: looking at the kernel log, it definitely detects the framebuffer and says its setting up a 1024x768 fb, but i have a black screen... any idea what im doing wrong?
<mmind00> via: you should also see messages about the hdmi + dp around that area ... what devicetree did you use?
<via> rk3288-veyron-speedy
<via> it does seem to detect both hdmi and dp
<via> i'll try to get this into a paste
<BorgCuba> aborche_, I didnt even get that far. Here is my bootlog: http://pastebin.com/sRdgu0M0
<BorgCuba> mmind00, have you recently (maybe 4.4-rc6) tried to boot the kernel on your radxa board?
<mmind00> via: I guess time to update my defconfig to reflect the new driver, please add
<mmind00> CONFIG_ROCKCHIP_ANALOGIX_DP=y
<mmind00> CONFIG_DRM_ANALOGIX_DP=y
<mmind00> CONFIG_PHY_ROCKCHIP_DP=y
<via> ok
<mmind00> BorgCuba: nope, anything broken?
<via> mmind00: er,those are already set
<aborche_> BorgCuba, kernel boots without any error ? did you get start system ? initrd or mount partition ?
<mmind00> via: that is strange then
<BorgCuba> I am not sure (maybe the defconfig I used is outdated) but it stops at some point without any error
<BorgCuba> also no tty switching message
<mmind00> via: ah sorry, I see it in the log as well now
<via> yeah, it references dp, just lacking edid data
<BorgCuba> mmind00, this is my bootlog: http://pastebin.com/sRdgu0M0
<via> i doubt this is relevent, but should console=tty1 for the linux cmdline?
<aborche_> BorgCuba, you use radxa board or tv stick ?
<BorgCuba> radxa lite board
<BorgCuba> aborche_, what hardware do you use?
<mmind00> via: can you try increasing the startup-delay in the panel_regulator node (say to 200ms ... value 200000), just to check that the panel doesn't take to long to start
<aborche_> D809 stick
<aborche_> similar radxa rock lite board without rk616 and rk1000
<BorgCuba> then it also has the act8846
<aborche_> rk3188t, rtl8723bs, it66121, act8846
<via> ok, increasing from 50000 to 200000
<aborche_> BorgCuba, yes it is
<BorgCuba> indeed, only a different hdmi trasnmitter
<via> mmind00: success
<aborche_> BorgCuba, radxa kernel 3.0.36+ boot and work perfectly with some vcc changes
<mmind00> via: \o/
<via> thank you, now i'll see about making everything else work
<BorgCuba> aborche_, yes 3.0.36+ works for me as well
<mmind00> via: if you have time, can you try some intermediate steps like 100, 150?
<BorgCuba> but its a total mess
<via> mmind00: sure
<aborche_> this is my fork from radxa with minor changes
<mmind00> via: I have reports for speedy being able to do the 50ms, so I'd like to find some sort of minimum for everybody ;-)
<BorgCuba> aborche_, what do you want to do with your device?
<aborche_> BorgCuba, a big problem with 3.0.36+ - old mali drivers and very slowly render in chrome and nwjs
<via> well, 80 ms works
<via> i'll keep going down and report back
<aborche_> BorgCuba, i need fast Xorg screen driver with video hardware acceleration
<BorgCuba> aborche_, I am aiming for retroarch on framebuffer (with mali)
<BorgCuba> got a pretty bad glmark2-es score like maybe 40 or at least < 100 while the orangepi-pc I recently received has a score of about 480pts (only MP2)
<aborche_> BorgCuba, olegk0 finished yesterday work with vdpau driver
<BorgCuba> nice
<aborche_> but only for old kernel
<mmind00> BorgCuba: can you apply https://bpaste.net/show/de072f8bfcda and check if it stops at some specific clock?
<BorgCuba> mmind00, I'll try
<aborche_> i think someone can fork their work into stable kernel.
<via> mmind00: 70 ms doesn't work for me, 80 does
<mmind00> via: no need to check in 5ms steps ;-)
<mmind00> if 80ms works for you, we'll just go to 100ms and be done with it
<via> sounds good to me, thank you
<via> fwiw, scroll speed is dramatically slower than it was with the chromeos kernel
<BorgCuba> mmind00, this is the bootlog with your patch applied: http://pastebin.com/cCujmkgR
gb_master has quit [Remote host closed the connection]
<mmind00> BorgCuba: :-( ... so it doesn't look like a clock is at fault
tlwoerner has joined #linux-rockchip
<ssvb> BorgCuba: about "got a pretty bad glmark2-es score like maybe 40 or at least < 100"
<via> now i just need to get libreboot on here
<ssvb> BorgCuba: it looks like the rockchip kernel might have problems with the buffer address validation code in MALI_IOC_MEM_MAP_EXT
<ssvb> BorgCuba: this probably may give you some ideas - https://www.mail-archive.com/linux-sunxi%40googlegroups.com/msg14756.html
<ssvb> BorgCuba: do you observe a 100% cpu usage?
<BorgCuba> ssvb, thanks! I'll read it
<BorgCuba> ssvb, I havent checked this
<ssvb> BorgCuba: just start "htop" while running glmark2-es2, and you will probably see it
<ssvb> BorgCuba: and if you have a profiler (perf), then you can also run "perf top" and probably see that all the time is wasted in memcpy
* ssvb does not have any rockchip hardware and is just lurking here
<BorgCuba> okay, I'll try. have to build htop first
<ssvb> build? what kind of distro are you using?
akaizen has quit [Read error: Connection reset by peer]
* aborche_ have mele a3700(allwinner a10) and d809.
akaizen has joined #linux-rockchip
Anarchy has quit [Ping timeout: 246 seconds]
<BorgCuba> ssvb, buildroot
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-rockchip
<ssvb> BorgCuba: ok
<BorgCuba> ssvb, indeed, the first core goes up to 100%
<ssvb> BorgCuba: if you can patch the kernel, then you can use an easy hack, something like this - https://github.com/ssvb/linux-sunxi/commit/2586c38ac17a918a71002bbd74357ca3f01cacb6#diff-e5a9a17e24cf93028716f25f00e5cc40R72
<ssvb> BorgCuba: just make sure that "mali_mem_validation_check" function returns _MALI_OSK_ERR_OK and this should fix your FPS
<via> mmind00: is unaccelerated fb expected?
<mmind00> via: yep
<BorgCuba> ssvb, I'll try to get that patch integrated
akaizen has quit [Ping timeout: 265 seconds]
<BorgCuba> old glmark2-es score: 40
<BorgCuba> ssvb, so far I only changed "mali_mem_validation_check" to always return _MALI_OSK_ERR_OK;
<BorgCuba> from the first fps values I can see it has not changed
<BorgCuba> core 1 still goes up to 100%
<BorgCuba> mostly at slightly above 90%
<ssvb> BorgCuba: yes, always returning _MALI_OSK_ERR_OK would help if it was this particular problem
<ssvb> BorgCuba: then it's probably a good idea to run a profiler and check what's going on
<ssvb> BorgCuba: I'm sorry that my advice was not very useful, I have no other ideas
<BorgCuba> I am always thankful for any advice
<BorgCuba> thanks ssvb!
<BorgCuba> glmark2 has finished: 40pts (as before)
<BorgCuba> ssvb, what do you use for profiling "-pg" and gprof?
<BorgCuba> let me see what buildroot has to offer
<ssvb> BorgCuba: gprof can't be used for profiling the kernel
<aborche_> BorgCuba, which config you use for your board ?
<aborche_> system boot is ok ? or halt ?
<aborche_> guys i have a question. in linux-next kernel operation with ramdisk(initrd) has been changed or no ? why kernel cannot find attached ramdrive ?
akaizen has joined #linux-rockchip
<aborche_> and unsuccessful boot can be fired if i use extended ext4 partition instead primary partition ?
<BorgCuba> aborche_, I used the defconfig from rockchip.fr (you pointed me there)
<BorgCuba> and the radxa dts file that comes with 4.4-rc6
<aborche_> hm. also i'm found at radxa forum fresh config for radxa rock lite from hipboi.
<aborche_> can you check your system with it ?
<BorgCuba> sure
<aborche_> a big tar file contains config for radxa-lite and broken dts. dts file with k31811 contains correct device tree
<aborche_> at my d809 i cannot boot, kernel panic fired when kernel trying to boot
<via> im having trouble with the rockchip sound now...but i'm not sure how to debug this. if i use mplayer with -ao alsa (i copied ove rthe ucm profiles from chromeos), output just hangs
<via> i can show the mixer controls though
<aborche_> i think problem in pulseaudio. when you change X session, pulseaudio is not stopped by default, and when Xsession started - another copy of pulseaudio started too
<mmind00> via: yep that is the current state of affairs ... on the chromeos-partition supposedly is something called an ucm-profile for alsa ... but I don't think anybody has looked into that yet
<via> mmind00: yeah, i copied those over as per the debian instructions
<via> so... hmm
aborche_ has quit [Remote host closed the connection]
ckeepax has quit [Ping timeout: 240 seconds]
<BorgCuba> ssvb, I just ran the profiler on glmark2, here are my results: http://pastebin.com/2EbiBs6C
<BorgCuba> also the performance of glmark2 dropped by 50% after I enabled perf_events in the kernel, is this to be expected?
<ssvb> BorgCuba: you need symbols for glibc to get the function names, but most likely 0x71c08 is memcpy
<ssvb> when you are running a profiler, some performance drop is expected
<ssvb> but just enabling perf_events in the kernel should not degrade anything
<ssvb> BorgCuba: is this a framebuffer flavour of the mali blob? or x11?
<BorgCuba> fbdev r3p2
<ssvb> same as on sunxi?
<BorgCuba> on the opi pc I am using r4p0 without ump and its much faster (more the 10 times compared to the rockchip)
<BorgCuba> ssvb, yes the libs/blobs are from sunxi git
<ssvb> that's very strange, there still must be something wrong in the kernel because the blob apparently can't map the framebuffer memory directly
<ssvb> are you sure that the "mali_mem_validation_check" patch was really applied when you tested it?
<ssvb> also does the rockchip framebuffer driver correctly report the physical address of the framebuffer in "smem_start"?
<BorgCuba> ssvb, I am quite sure (changed it, built it, copied it to my rootfs, insmod mali.ko)
<BorgCuba> ssvb, for your last question: I am not sure
<ssvb> this particular change is in the mali driver (mali.ko), so I'm assuming that you copied both the kernel and the mali module, right?
<BorgCuba> well, at first I didnt change the kernel, so I only copied mali.ko
<BorgCuba> after building the kernel with perf events I also rebuild the mali module (new kernel & new mali.ko)
<ssvb> you can add some debugging prints to the MALI_IOC_MEM_MAP_EXT ioctl just to see whether it gets called at all
<BorgCuba> yes, I'll do that
akaizen has quit []
akaizen has joined #linux-rockchip
akaizen has quit []
gb_master has joined #linux-rockchip
akaizen has joined #linux-rockchip
<BorgCuba> weird, its seems like I have broken it completely
gb_master has quit [Remote host closed the connection]