**** BEGIN LOGGING AT Fri Jun 16 03:00:03 2017 Jun 16 11:12:39 koen: around? Jun 16 11:14:31 I've noticed you added MAGICK_HOME in https://git.linaro.org/openembedded/meta-linaro.git/commit/meta-optee/recipes-security/optee/optee-os_git.bb?h=pyro&id=9758c6dc36fe74829798a7f2278fdfa91ce9f6a3 but still it seems to fail with message that there isn't imagemagick-dev installed on host, any idea what went wrong since then? Jun 16 11:20:11 JaMa: It seems to depend on the host distro :( Jun 16 11:20:30 that commit works on fedora, but not on debian Jun 16 11:21:33 I haven't looked at it in a while, maybe the new optee version introduces different issues... Jun 16 11:22:59 ok, I'll try to dig around it Jun 16 11:23:34 I'm seeing it on morty (even when the link pointed to pyro branch) Jun 16 11:25:27 is there a way to get log/debug output when booting a raspberrypi? Jun 16 11:25:50 khem: I have reverted back to 835b705, just before gcc7 commit, and armv5 kernel boots Jun 16 11:26:03 i only get the 4 raspberries and at some point the login prompt.. not really helpful to debug boot issues Jun 16 11:26:29 i tried adding earlyprintk to the bootargs, but doesn't make a difference Jun 16 11:37:53 koen: it seems to be caused by different imagemagick version we have in OE and wand expects sysroots/x86_64-linux/usr/lib/libMagickWand-7.Q16HDRI.so and wand checks for: Jun 16 11:37:56 versions = '', '-6', '-Q16', '-Q8', '-6.Q16' Jun 16 11:37:58 options = '', 'HDRI', 'HDRI-2' Jun 16 11:38:06 so missing -7.Q16 version Jun 16 11:38:40 that would explain a lot of the issues Jun 16 11:39:30 koen: and it cannot be easily just added to the list: AttributeError: .../sysroots/x86_64-linux/usr/lib/libMagickWand-7.Q16HDRI.so: undefined symbol: MagickGetImageMatteColor Jun 16 11:39:43 checking if there is newer python-wand version compatible with our imagemagick Jun 16 11:41:48 nope :/ https://github.com/dahlia/wand/blob/4fe2c6ba9cb0d4105361cea6e9e9e83116080473/wand/api.py#L58 Jun 16 11:42:00 it pisses me off we only need wand to create a bitmap font Jun 16 11:42:15 which optee will paint pixel-by-pixel at runtime Jun 16 11:42:36 "Graphical display seems slow" ORLY Jun 16 11:45:08 would it make sense to pregenerate the bitmap fonts instead of depending on older imagemagick? It was also discussed more than year ago in https://stackoverflow.com/questions/37011291/python-wand-image-is-not-recognized Jun 16 12:12:07 khem: tested with 4.4.8 and 4.4.72 Jun 16 13:57:16 koen: I've sent one fix for optee-client to openembedded@lists.linaro.org (waiting for moderator approval) and I've also noticed that optee-os triggers ld.gold to fail with "internal error in set_address, at ../../gold/output.h:322" so we're forcing gcc.mk to use BFD Jun 16 13:57:45 maybe it doesn't fail like this with linaro version of binutils :) Jun 16 14:13:12 koen: also gcc-7.1 from latest oe-core doesn't recognize '-mno-apcs-float' added by optee-os Jun 16 14:14:19 what a surprise Jun 16 14:14:42 if you want examples on how to write gcc version specific code, look no further than optee Jun 16 14:27:14 :) Jun 16 14:43:42 this is driving me nuts.. i can't make any boot messages show up before the login prompt. by now i'm passing "dwc_otg.lpm_enable=0 root=/dev/mmcblk0p2 rootfstype=ext4 rootwait earlyprintk systemd.log_level=debug systemd.log_target=kmsg log_buf_len=1M printk.devkmsg=on verbose systemd.journald.forward_to_console=yes TTYPath=/dev/tty0 console=ttyS0" but nothing seems to have any effect Jun 16 14:44:41 i just see the four raspberries and a few secs later get the login prompt.. there's not a single line of kernel or systemd output what's currently going on. it makes troubleshooting boot failures incredibly hard Jun 16 14:46:19 and i'm running out of ideas what to google for.. **** ENDING LOGGING AT Sat Jun 17 03:00:01 2017