IRC log of #cubox of Fri 24 Mar 2017. All times are in CET < Back to index

08:28 topi` jnettlet: are there any instructions on how to PROM the ethernet/wlan macs?
08:31 topi` I bet through some magic /sys/fsl_otp/ entry
08:31 topi` there is a header "Blowing fuses to program the unit MAC address"
08:31 topi` but it doesn't state whether this is for the Eth0 MAC or wlan0 MAC
08:31 topi` I suspect eth0
08:57 topi` jnettlet: I tried compiling libdrm and mesa from git, and I was able to get a HW accelerated kmscube working on my HB quad... this on the 4.11-rc1 kernel
08:58 topi` NV12 textures (for yuv->rgb conversion) work so it's a good starting point to decode h264 on the CPU and draw it to the YUV surface
09:31 jnettlet topi`, the fuses are only for the onboard adapter
09:31 jnettlet ethernet
10:55 topi` OK. what about the Wilink8 mac address? where is it stored?
10:56 jnettlet topi`, that is all self contained
10:57 topi` OK, so we shouldn't worry about the Wilink mac addresses
10:58 jnettlet nope, those come signed sealed and certified
10:59 auke- jnettlet: i'm trying to rebuild my cubox-i image with your latest changes in meta-solidrun-arm-imx6 but it seems like the KERNEL_DEVICETREE parameters aren't expanded: No rule to make target `arch/arm/boot/dts/imx6{q,dl}-{cubox-i,hummingboard,hummingboard2}.dtb'
11:00 jnettlet auke-, for fido or jethro?
11:01 auke- jethro
11:02 jnettlet you just did a git pull this AM?
11:03 auke- yes, latest commit is c73ff0f gstreamer-plugins-imx...
11:04 auke- but the KERNEL_DEVICETREE change is in the WiLink commit (184a905)
11:04 jnettlet yes, I just built new images this morning without an issue
11:05 jnettlet what image are you building?
11:06 jnettle 11:06 * jnettlet rebuilding the kernel now.
11:06 auke- I'm building a custom image. I will first clean the build directory and try again, maybe that helps.
11:07 jnettlet you can just do the kernel with. bitbake -c cleanall virtual/kernel && bitbake -c build virtual/kernel
11:07 auke- Unfortunately my build machine is slower than my CuBox-i4 :/
11:07 auke- okay, thanks
11:57 topi` i'm building "mpv" on a hummingboard quad :)
11:58 topi` the "waf" build system seems to truly use all 4 cores all the time, unlike Make
13:51 mhilt jnettlet and bencoh -- I'm having trouble getting capture/5460.c to build vs subdev/5460.c ... I'm working in Yocto, any idea how I get the capture driver to build?
14:13 wumpus topi`: hopefully it's not the build tool that uses all those cores but the compiler? :-)
16:23 auke- jnettlet: same error: make -j 1 imx6{q,dl}-{cubox-i,hummingboard,hummingboard2}.dtb ... No rule to make target `arch/arm/boot/dts/imx6{q,dl}-{cubox-i,hummingboard,hummingboard2}.dtb'. Stop
16:24 jnettlet auke- can you run bitbake -k -c build virtual/kernel -v and then send me the output
16:25 jnettlet oh wait. that make command won't work
16:25 auke- jnettlet: could the gcc version cause the error? GCCVERSION was set to 4.9.% in order to build older versions of the kernel
16:25 jnettlet it shouldn't
16:27 jnettlet auke-, which layers are you pulling in?
16:27 auke- I'll try it one more time from an empty build directory
16:32 auke- jnettlet: besides meta-yocto i use the older meta-fsl-arm and meta-fsl-arm-extra layers (all jethro)
16:32 jnettlet auke-...oh that is probably the problem. you shouldn't need meta-fsl-arm-extra if you are using ours
16:37 auke- okay. It used to work with that layer (with meta-solidrun-arm-imx6 from december), but i'll try it without -extra.
16:38 jnettlet auke-, I wrote our layer to replace meta-fsl-arm-extra since it was not maintained properly and was causing users lots of issues because they weren't testing software versions
16:39 auke- ah, ok. thanks for the help