cnxsoft changed the topic of #imx6-dongle to: i.MX6 HDMI Dongle Software Development. You can also join us at https://groups.google.com/forum/?fromgroups#!forum/imx6-dongle - IRC Logs - http://irclog.whitequark.org/imx6-dongle/
dshankar has quit [Ping timeout: 252 seconds]
dshankar has joined #imx6-dongle
<dshankar> JFYI I was able to get chromium running on Jessie
dshankar has quit [Client Quit]
cnxsoft has joined #imx6-dongle
rz2k has quit []
dailylinux has quit [Quit: Leaving]
fossxplorer has quit [Remote host closed the connection]
fossxplorer has joined #imx6-dongle
fossxplorer has quit [Changing host]
fossxplorer has joined #imx6-dongle
dshankar has joined #imx6-dongle
<dshankar> any tips on getting VPU acceleration in chromium?
<dshankar> specifically the hardware VP8 decoder. chromium is using software right now, no gpu/vpu
fossxplorer has quit [Ping timeout: 252 seconds]
rz2k has joined #imx6-dongle
Akaizen has quit [Remote host closed the connection]
alex_kag_ has joined #imx6-dongle
fossxplorer has joined #imx6-dongle
fossxplorer has joined #imx6-dongle
rz2k has quit []
cnxsoft has quit [Ping timeout: 246 seconds]
alex_kag has joined #imx6-dongle
alex_kag_ has quit [Ping timeout: 252 seconds]
hste_ has quit [Ping timeout: 272 seconds]
hste_ has joined #imx6-dongle
panda84kde has joined #imx6-dongle
cnxsoft has joined #imx6-dongle
fossxplorer has quit [Read error: Operation timed out]
cnxsoft has quit [Ping timeout: 246 seconds]
cnxsoft has joined #imx6-dongle
alex_kag has quit [Quit: Konversation terminated!]
alex_kag has joined #imx6-dongle
cnxsoft has quit [Quit: cnxsoft]
JBD1986 has quit [Quit: Leaving]
JBD1986 has joined #imx6-dongle
panda84kde has quit [Quit: Konversation terminated!]
Akaizen has joined #imx6-dongle
alex_kag has quit [Ping timeout: 252 seconds]
jas-hacks has joined #imx6-dongle
Akaizen has quit [Remote host closed the connection]
alex_kag has joined #imx6-dongle
<JBD1986> jas-hacks, I got glxgears running with hardware acceleration, but only if openbox is not running
<JBD1986> if openbox is running, starting glxgears will cause the desktop to hang, with xserver at 100%
<jas-hacks> JBD1986: good progress
<JBD1986> interestingly, sometimes if I start es2gears_x11, it will start and display while openbox is running, but if I interact with the window that es2gears is running in, it will hang similarly
<JBD1986> in most cases, starting es2gears will cause the same hang
<jas-hacks> look like xserver isn't compiled against the libdrm patch libdrm_2.4.49.bbappend
<jas-hacks> /s/look/looks
<JBD1986> jas-hacks, can you elaborate?
<JBD1986> jas-hacks, I will check
<JBD1986> grep reveals this:
<JBD1986> ./sources/meta-fsl-arm/recipes-graphics/drm/libdrm_2.4.46.bbappend:3:SRC_URI_append_mx6 = "file://drm-update-arm.patch"
<JBD1986> I will note that my version is slightly different, 2.4.46
<jas-hacks> that looks ok
<jas-hacks> you must be have built xorg-server using that libdrm
<JBD1986> afaik, I am using the stock dora meta-fsl-arm stuff.
<jas-hacks> xorg-server will be outside the fsl stuff
<JBD1986> ok, so it is possible that libdrm_(some version other than 2.4.46) is being used
<JBD1986> ?
<jas-hacks> yes possible
<JBD1986> hmmm
<JBD1986> justin@sfrubserlusergt:~/development/tc3-detos/detos$ find ./sources -name "libdrm*"
<JBD1986> ./sources/meta-oss/recipes-graphics/drm/libdrm_2.4.49.bb
<JBD1986> ./sources/meta-fsl-arm/recipes-graphics/drm/libdrm_2.4.46.bbappend
<JBD1986> ./sources/meta-oss/recipes-graphics/drm/libdrm
<JBD1986> ./sources/meta-fsl-arm/recipes-graphics/drm/libdrm
<JBD1986> ./sources/meta-dm814x-fx1/recipes-graphics/drm/libdrm
<JBD1986> ./sources/meta-dm814x-fx1/recipes-graphics/drm/libdrm/files/libdrm.patch
<JBD1986> ./sources/meta-dm814x-fx1/recipes-graphics/drm/libdrm/libdrm_2.4.23-ti.bb
<JBD1986> ./sources/meta-ti/recipes-graphics/drm/libdrm_2.4.41.bb
<JBD1986> ./sources/meta-ti/recipes-graphics/drm/libdrm-2.4.41
<JBD1986> ./sources/openembedded-core/meta/recipes-graphics/drm/libdrm_git.bb
<jas-hacks> unfortunately I don't use yocto that often
<JBD1986> ./sources/openembedded-core/meta/recipes-graphics/drm/libdrm.inc
<JBD1986> ./sources/openembedded-core/meta/recipes-graphics/drm/libdrm_2.4.46.bb
<JBD1986> ./sources/openembedded-core/meta/recipes-graphics/drm/libdrm
<JBD1986> yeah ill set a preferred version
<JBD1986> if there isn't one already
<jas-hacks> is there a libdrm dependency in the xorg-server recipe
<JBD1986> doesn't look like it
<JBD1986> I checked the bbappend
<JBD1986> none in there either
<JBD1986> and actually, it looks like 2.4.49 built
<JBD1986> from buildhistory
<jas-hacks> search for xf86drm.h in the source tree and see if it is patched
<JBD1986> i'm fairly confident it's not being patched
<JBD1986> I can include that patch for 2.4.49
<jas-hacks> also rebuilt xserver
<JBD1986> rebuilding the image
<JBD1986> will probably take a while
<jas-hacks> I thought all the X stuff was working under Yocto
<JBD1986> we just upgraded from dylan to yocto. And this is my fault if that's the issue
<JBD1986> meta-fsl-arm has that patch with 2.4.46
<JBD1986> we add 2.4.49 in our own layer
<JBD1986> but there is no arm patch
<jas-hacks> JBD1986: What you building with Yocto?
<JBD1986> yeah
<JBD1986> sorry dylan to dora*
<JBD1986> brain slip
<JBD1986> but yes, it is yocto
<JBD1986> jas-hacks, I'm leaving for the day. I'll let you know how the rebuild goes
<jas-hacks> no problems
<dshankar> jas-hacks: did you ever get VPU acceleration in chromium working in your older builds?
<dshankar> or just GPU for webgl etc.?
<jas-hacks> dshankar: I think the only thing that was working was egl for chromium rendering
<dshankar> ah ok
<jas-hacks> dshankar: I think WebGL might work now with new GPU libs, did you try?
<dshankar> i haven't tried, but I noticed it was up on chrome://gpu
<dshankar> really need the hw vp8dec though
<dshankar> gstreamer does it - tested with the sintel 1080p file and CPU was at ~1% :D
<dshankar> chromium was frying the CPU. burnt my finger touching the heatsink haha
bfederau has quit [Remote host closed the connection]
<jas-hacks> chromium was using sw rendering
<dshankar> yup
<jas-hacks> another option would be to see if totem works in chromium
newell has joined #imx6-dongle
<newell> I am new to yocto and wondering if anyone in here might be able to help me as I am having issues getting console output for core-image-base and core-image-minimal when I build them for the imx6dl system
<newell> curious if anyone might be able to help me
<newell> I am using the MCIMXDL-SDP board
<newell> i.MX6 Sabre SDP
bfederau has joined #imx6-dongle
<jas-hacks> newell: does the image boot?
<newell> nothing comes out of the console...so at this point I would have to say no
<newell> I have followed all the instructions and not sure where to go at this point
<jas-hacks> you get output from uboot?
<newell> nope
<newell> nothing
<newell> I am connected to the console with 115200 8N1 and all flow control off
<jas-hacks> is the uboot built from yocto?
<newell> Ah that is probably the problem.
<newell> I am using the *.sdcard image and dd'ing it to the SD card
<WarheadsSE> which contains uboot
<WarheadsSE> but if it is booting from SPI/NOR u-boot and not the SD card
<WarheadsSE> ..
<newell> where would I adjust that in the configuration?
* newell is a newbie to yocto and freescale system
<WarheadsSE> thats a freescale board thing
<newell> oh you mean the switch on the board...that is set to SD card
<newell> I just don't know what u-boot parameters are since I can't even get u-boot up on the console
<WarheadsSE> then it should, in theory, be loading that uboot
<newell> hmmm
<newell> I even used /dev/zero to dd the whole SD card before dd'ing the image
<newell> I thought for sure I would be getting console output...but I am getting nothing
<WarheadsSE> well, have you checked any of the other UART pinouts?
<jas-hacks> Can you test with a prebuilt uboot for the SabreSD?
<newell> jas-hacks, from freescale or does yocto have one?
<newell> the pre-built stuff from freescale works fine
<newell> and I was also able to compile my own kernel with LTIB and Ubuntu rootfs and everything worked fine as well
<jas-hacks> newell: that goods, so can't you try that uboot against the yocto image?
<newell> jas-hacks, okay so burn that u-boot to the sd card instead...yeah will try that
<WarheadsSE> burn the .sdcard, then burn the other uboot over top.
<newell> k
<newell> WarheadsSE, and to do just the uboot I should do something like: https://community.freescale.com/docs/DOC-94989
<newell> sudo dd if=u-boot-imx6qsabresd-v2012.10-r3.imx of=/dev/sdX bs=512 seek=2
<WarheadsSE> roughly
<newell> well the other u-boot is padded I believe so probably need to throw a skip=2 in there
<WarheadsSE> if it is padded, it is padded that 2
<newell> WarheadsSE, the command I used when getting uboot to work with LTIB build was:
<newell> $ sudo dd if=u-boot-mx6dl-sabresd.bin
<newell> of=/dev/sdb bs=512 seek=2 skip=2 conv=fsync
<newell> So my thought was to use the exact same command
<jas-hacks> newell: the ".imx" doesn't need padding
<newell> okay so no skip=2 then?
<jas-hacks> dd if=xxxx.imx bs=1k seek=1 of=/dev/sdx
<newell> jas-hacks, yeah but I am using the other uboot that is not the imx uboot from yocto build
<newell> I was going to use the uboot from LTIB build right?
<newell> okay will just follow what you wrote
<jas-hacks> then for .bin use padding
<newell> ah okay ;)
<newell> that worked but now it can't find the kernel image....probably because of my uboot parameters
<newell> I guess I can burn the image and rootfs individually instead of using the .sdcard
* newell is back to burning
<newell> anyone here enabled the RT PREEMPT patch on the yocto kernel before? Trying to figure that out.
<newell> That was the main motiviation for us trying yocto out
<newell> I was having issues getting the rt-patch to apply properly with freescale's linux kernel and bsp
<jas-hacks> newell: so the patches wouldn't apply easily to the fsl kernel
<newell> not the one that came with the LTIB
<newell> when I was reading the yocto documentation it seems like they actually try and make sure that the rt patch is compatible and there is even a configuration option you can pick to enable it (haven't tried it yet).
<newell> First I wanted to get the vanilla yocto building and booting
<newell> Then I am going to try and tackle that next
fossxplorer has joined #imx6-dongle
<jas-hacks> Give there is a patch set for 3.0.36, you could try applying it manually
<newell> by manually do you just mean the normal patch -p1 < ../patchname.patch ?
<jas-hacks> yeah
<newell> that is what I did
<newell> applying it manually fails as there are .rej and .orig files that are created
<newell> i.e. the entire patch set is not applied
<newell> when I go to compile the kernel after doing that....build fails
<newell> whereas without the patchset the build succeeds
<WarheadsSE> well yeah, you would have to fix the rejects..
<newell> yeah did that too...still didn't build ;)
<jas-hacks> what's the reasoning for a RT kernel
<newell> we make an industrial controller that needs irq interrupt latency on the order of 15 microseconds
<jas-hacks> you don't offload to a mcu
<newell> We have used mcus in the past but want to go to a processor for our upcoming product.
<newell> The irq is for handing Ethernet frames of our own protocol
jas-hacks has left #imx6-dongle [#imx6-dongle]
fossxplorer has quit [Quit: Leaving]
fossxplorer has joined #imx6-dongle
fossxplorer has quit [Read error: Connection reset by peer]
fossxplorer has joined #imx6-dongle
fossxplorer has joined #imx6-dongle
fossxplorer has quit [Changing host]
Akaizen has joined #imx6-dongle
<WarheadsSE> And you chose imx6 over something like the sitara that has 2 PRU onboard?
<newell> I didn't choose the board...the hardware guys did that
<newell> WarheadsSE, probably has to do with cost and what they figured would be the best processor to have on our product
<WarheadsSE> they I assume being a team of EEs
<newell> yes
<newell> and software managment
<newell> This decision was made before I came onboard
<newell> ;)
<newell> otavio, do you have any idea if anyone has gotten an fsl kernel to build with rt patch?
bfederau has quit [Remote host closed the connection]
bfederau has joined #imx6-dongle
fossxplorer has quit [Ping timeout: 252 seconds]
rz2k has joined #imx6-dongle
<otavio> newell: I am not aware of someone; do it in 3.10.17 shouldn't be hard thought
Akaizen has quit [Remote host closed the connection]
Akaizen has joined #imx6-dongle
<newell> otavio, so your advice, if I choose to blaze down that path is to use 3.10.17 instead of 3.0.35?
<newell> Is imx6qsabresd
<otavio> sure
<newell> imx6qsabresd supported by 3.10.17 as well?
<otavio> Going to be announced soon
<newell> okay awesome, thanks for help
<otavio> the 3.10.17 is available
<otavio> the 3.10.17-1.0.0-beta is available
<otavio> and we're working on merging those to community BSP