**** BEGIN LOGGING AT Sun Aug 14 02:59:58 2016 Aug 14 06:22:00 zmatt: I would need to refresh my memory, but I understood that WA has drawbacks. performance maybe. it was not recommended as a solution, if the WA can be made in some other way. Aug 14 06:46:06 WA ? Aug 14 06:46:36 workaround Aug 14 06:46:40 ah Aug 14 06:48:22 so we have a vague erratum with a workaround unvolving setting tw9 bits in a magica undocumented register. *and* there may ne undocumented consequnces to doing so Aug 14 06:48:48 it would really grace TI if they actually documented teg register Aug 14 06:50:11 I'm guessint noone has done benchmarks yet? Aug 14 06:50:33 I haven't seen locksp yet though, but iirc they requires pretty rare corcumstabces ti vegub wutg Aug 14 07:21:44 tomba: I'm happy though that new extemely hacky approach to rotating the fbdev (branch patch/tiler-fbdev2) sems to work working Aug 14 08:29:43 tomba: *to be working. it's really not how I want to solve itthough... Ishould probabl introduce a "default rotation" configured in DT Aug 14 08:30:06 for biw only fbdev is foolwed, DRM seems exactly what's going on Aug 14 08:30:20 nice typo'ing zmatt ;P Aug 14 08:30:42 wow that IS impressive Aug 14 08:31:17 I guess half-crouching at an oblique angle at a keyboard is not an optimal way to use it Aug 14 08:31:27 nope lol Aug 14 08:31:42 the oblique angle alone is bad enough I find Aug 14 08:32:30 tomba: but there are probably very few applications/libraries doing DRM modesetting, we could just manually patch them to honor the default rotation Aug 14 08:33:25 rotating everything transparent to userspace is a problem anyhow since the non-tiled omap_bo_new() just gets total size, not width/height like dumb buffers work Aug 14 09:04:50 zmatt: rotation is hard =). that's why advocated using the external rotation from the panel. it's easy. Aug 14 09:38:38 rotation isn't hard at all Aug 14 09:39:02 pretty much all problems have been with memory (mis)management Aug 14 09:39:34 once your fb is in tiled memory dss can happily can it out in any orientation, and that functionality already exists Aug 14 09:39:43 (and works) Aug 14 09:40:34 and the rotation chips comes with a free pile of major headache Aug 14 09:40:55 it cannot synchronize its output frame rate to its input frame rate Aug 14 09:41:30 i.e. TE Aug 14 09:41:33 AR Aug 14 09:42:12 it's really only meant to be used in DSI command mode... video mode is an afterthought and is basically just treated like inefficient use of command mode Aug 14 09:43:26 and to be fair, the only *hardware* issue I've actually encountered in TILER it its lack of wrap-mode bursts Aug 14 09:43:49 *lack of support for* Aug 14 09:43:55 which was just a bad design decision Aug 14 09:45:23 well, and there's the odd little observation that TILER appears to be optimized for a memory burst len of 4, while EMIF is only capable of a burst length of 8 Aug 14 09:45:45 not a huge deal but makes me wonder what went wrong there Aug 14 09:51:58 tomba: I wonder how hard it would be to have a way to make arbitrary framebuffer objects appear as fbdevs on (suficciently privileged) userspace request Aug 14 09:52:58 (although getting rid of legacy fbdev also seems appealing to me) Aug 14 12:40:30 Hi. Any ideas, how I can remove the `g_multi` Linux kernel module? Aug 14 12:40:33 http://paste.debian.net/789335/ Aug 14 12:41:18 https://libreboot.org/docs/install/bbb_ehci.html says it can be removed. Aug 14 12:41:39 I’d like to load the g_dbgp module. Aug 14 19:17:33 jkridner, is there an ETA for the Beaglebone Blue? Aug 14 21:56:13 Hi, i have a problem with booting BBB from uSD card. Nothing is happen when i try to boot like that. Card is SD 4GB formatted FAT with MLO from starterware and example led blinking from starterware Aug 14 23:35:19 jkridner, is there an ETA for the Beaglebone Blue? **** ENDING LOGGING AT Mon Aug 15 02:59:58 2016