**** BEGIN LOGGING AT Mon Oct 08 02:59:58 2012 Oct 08 03:09:09 infinity, heh that is nvidia for you, I'm not even quite sure what the difference between the T2 and T3 pack is yet Oct 08 07:38:59 lilstevie: the avp firmware is different, but that's also a different tarball ... Oct 08 07:44:53 marvin24, yeah I know the avp firmware is, but that is cause the avp is different in general from what we have seen Oct 08 07:45:47 marvin24, but the acceleration xorg driver and libs that you guys package up for the ac100 work just fine on the tf201 Oct 08 08:22:14 lilstevie: the original tar gz (for up to abi 12) contains a lot more Oct 08 08:22:22 e.g. kernel, bootloader, ... Oct 08 08:22:59 abi13 driver was released as a single tegra_drv.so (which is identical for tegra2/3) Oct 08 08:45:26 marvin24, oh yeah, bootloader and stuff could easily go in the same tarball though Oct 08 08:45:35 given size is not big Oct 08 09:20:02 maybe customers may get confused if they see more than one kernel ;-) Oct 08 09:44:20 maybe Oct 08 09:44:29 although that would be a little troubling :p Oct 08 13:28:29 infinity: my build is failing with ../sysdeps/unix/sysv/linux/init-first.c:102:1: sorry, unimplemented: function profiling Oct 08 13:28:49 and indeed it is unimplemented. Where do I turn it off for the arm64 build? Oct 08 13:29:05 (eglibc) Oct 08 13:33:12 aha debian/sysdeps/arm64.mk looks plausible. --disable-profile Oct 08 14:29:15 ogra_, is the openscenegraph ftbfs on arm the gl/gles problem? Oct 08 14:29:36 dunno, i didnt even know it ftbfs Oct 08 14:29:53 asking because a whole lot of things b-d on it Oct 08 14:30:00 (/me is on a special forces project until UDS ... so i didnt follow the ftbfs list) Oct 08 15:15:11 marvin24, what's the prospect of ac100 being all mainlined, any ETA? Oct 08 15:15:15 any blockers still? Oct 08 15:15:27 * janimo hopes it's there by 13.04 :) Oct 08 15:15:38 janimo: it is (all) mainlined for over a year now Oct 08 15:15:52 unfortunately, tegra2 support isn't ... Oct 08 15:16:10 I would also prefer to move to mainline for 13.04 Oct 08 15:16:31 we already lost supend with 3.1, so it shouldn't matter if we lose some more stuff ;-) Oct 08 15:17:06 marvin24, well I mean ac100 support as in working out of the box :) Oct 08 15:17:32 and your insight on where the tegra2 upstreaming stands Oct 08 15:18:30 * ogra_ doesnt think there is anything gfoing on wrt upstreaming the tegra2/3 bits Oct 08 15:19:36 especially in the light that they are mostly developed for android which uses snapshot kernels Oct 08 15:20:09 as long as android moves with that practice i doubt we'll see changes here Oct 08 15:20:39 janimo: it works out of the box if you add the drm driver Oct 08 15:21:15 ogra_: wrong, e.g. today lp2 support was added for tegra3 Oct 08 15:21:23 ah, nice Oct 08 15:21:35 I think tegra2 has it already Oct 08 15:21:38 what's lp2 Oct 08 15:21:46 well, they still build their binary drivers against a special kernel tag usually Oct 08 15:21:48 but I haven't looked lately Oct 08 15:22:19 and i doubt that will change in the short term Oct 08 15:22:40 marvin24, janimo's dream is to use mainline for ac100 (and other tegras) in 13.04 Oct 08 15:22:41 ogra_: tegra drm will likely come to staging in 3.8 Oct 08 15:23:17 when is 3.8 due ? Oct 08 15:23:21 and "they" already made some changes to mainline to get their propietary driver working Oct 08 15:23:26 6 mo probably Oct 08 15:23:26 early 20013 ? Oct 08 15:23:31 oops Oct 08 15:23:41 no, we are not that slow Oct 08 15:23:43 didnt mean to exaggerate that much :) Oct 08 15:23:44 ;-) Oct 08 15:24:16 so we will endup in a situation where both drivers are supported Oct 08 15:24:33 the open one likely without any 3d or video accel Oct 08 15:24:45 thats fine Oct 08 15:24:53 and thats how i use the closed one atm :) Oct 08 15:24:59 I hope I didn't look too deep into the glass Oct 08 15:25:03 on my ac100 at least Oct 08 15:26:51 yeah, O cam Oct 08 15:26:52 er Oct 08 15:27:38 I can't say I use the 3D and video accel on my prime either Oct 08 15:29:12 well, and what for :) Oct 08 15:29:20 there isnt really much thats built for GLES Oct 08 15:36:47 ogra_, yeah, and I really don't watch videos on my prime anyway Oct 08 16:05:13 ogra_: by any chance, did you try Q on your beagle? Oct 08 16:05:26 nope Oct 08 16:05:52 ok Oct 08 16:06:06 anyway, problem found and fix too Oct 08 16:06:13 i'll send a pull req tonight Oct 08 16:06:16 (after gym) Oct 08 16:06:27 ppisati, you rock ! Oct 08 16:06:38 will test with the fix then Oct 08 21:11:40 was the CONFIG_HIMEM issue on Pandaboard ever resolved? Oct 08 21:11:40 (i.e. random crashes when trying to do a kernel build if CONFIG_HIMEM was set) Oct 08 21:11:41 ogra@panda:~$ grep HIGHMEM /boot/config-3.5.0-211-omap4 Oct 08 21:11:41 # CONFIG_DEBUG_HIGHMEM is not set Oct 08 21:11:41 CONFIG_HIGHMEM=y Oct 08 21:11:41 we have it enabled since a while, so yes Oct 08 21:11:41 at least in 12.04 and 12.10 it should be fine Oct 08 21:11:41 do you recall what the issue root cause was? Oct 08 21:11:41 no, but there was a bug Oct 08 21:11:41 I'm working on another OMAP4 based system and it too has random crashes if HIMEM is enabled :-( Oct 08 21:11:41 running 3.6 Oct 08 21:11:42 bug 633227 Oct 08 21:11:42 Launchpad bug 633227 in linux-ti-omap4 "instabilities with highmem activated" [High,Fix released] https://launchpad.net/bugs/633227 Oct 08 21:11:43 we use 3.5 in 12.10, seems pretty solid Oct 08 21:11:43 but we dont use mainline. mind you Oct 08 21:11:45 hmmm . . . not really clear from that bug exactly what fixed the problem :-( Oct 08 21:17:14 What ever became of devicetree for arm? I'm reading articles covering the 3.7 kernel, and they don't even seem the same. Oct 08 21:57:30 sakoman_: I don't think we got to the point of finding exactly what was the patch that fixed the issue Oct 08 21:57:41 I believe that at some point it just worked as expected Oct 08 21:58:39 Hi, can someone give me a hint about how to debug linux kernel boot without JTAG? I was thinking in use early_printk and try to use usb debug Oct 08 23:58:20 tinti: DEBUG_LL is useful in early linux boot Oct 08 23:58:44 wookey: thanks :) Oct 08 23:58:47 maybe that's what's called early_printk these days Oct 08 23:58:58 it writes direct to serial port Oct 08 23:59:24 or probably USB these days, as no-one has serial ports anymore Oct 09 00:07:13 wookey: in fact it is for raspberry pi :) Oct 09 00:07:42 wookey: I think it is at least reasonable to implement it over gpio too **** ENDING LOGGING AT Tue Oct 09 02:59:59 2012