**** BEGIN LOGGING AT Tue Sep 18 02:59:59 2012 Sep 18 03:39:08 hmmm Sep 18 05:39:21 <_av500_> +1 Sep 18 05:55:10 -2 Sep 18 07:42:49 Hello. Sep 18 07:43:25 * av500 feels the buzz building up Sep 18 07:43:29 If I connect a LCD to the Beagleboard, will I still have use of the DVI port ? Sep 18 07:50:08 Hello ? any takers ? If I connect a LCD to the Beagleboard, will I still have use of the DVI port ? Sep 18 07:51:43 mik_: do you mean at the same time ? Sep 18 07:51:44 nope Sep 18 07:51:51 same pins used Sep 18 07:51:56 the TRM and or SRM would have told you Sep 18 07:52:12 Correct. Looking at the schematic I suspect the answer is kind-a. Sep 18 07:52:25 Correct. Sep 18 07:52:35 we passed? Sep 18 07:52:36 At a minimum, I'd be stuck with the same resolution. Sep 18 07:52:41 yes Sep 18 07:52:45 I passed. =) Sep 18 07:52:51 * LetoThe2nd hands av500 a cookie. Sep 18 07:52:56 if you find a LCD that takes exactly the same timing as the TFP chip Sep 18 07:53:08 yep... it's a non-starter. =) Sep 18 07:53:43 Thanks av500. Sep 18 07:54:03 Since I have your attention, I could use your opinion on the software side. Sep 18 07:54:18 or ...perhaps I'm really asking a chip question. Sep 18 07:54:32 av500: "which software will make everything automagically work?" Sep 18 07:54:59 Does the scalar on BeagleBoard allow me to render 720P using the DVI port while I'm using the S-VIDEO port ? Sep 18 07:55:18 Sorry.. That's wasn't clear. Sep 18 07:55:46 Will the scalar allow me to mirror the 720P image rendered from the DVI port on to the S-VIDEO port ? Sep 18 07:56:06 That's still unclear but, possibly less unclear. =) Sep 18 07:59:00 it's indeed medium unclear Sep 18 07:59:48 Are you suggesting you are unclear as to if the scalar will support this use case or are you referring to my wording of the question ? =) Sep 18 07:59:58 not clear about that Sep 18 08:00:26 :) Sep 18 08:00:36 I forgot all about OMAP3 DSS Sep 18 08:01:21 omap3 dss cannot scale the GUI plane, only the video planes Sep 18 08:01:34 and you have 1 GUI and 2 VIDEO Sep 18 08:02:01 you could maybe use one video plane in RGB mode from the 720p frame and scale that to s-video Sep 18 08:02:53 but on omap3 all the scaling restrictions are complex Sep 18 08:02:59 I need the GUI mirrored. Sep 18 08:03:16 yes, read what I wrote Sep 18 08:03:59 I read it but, likely didn't understand what you wrote. Does it matter that I'm using Android and am stuck to a single, fb0 for desktop rendering ? Sep 18 08:04:19 not really Sep 18 08:04:50 Ohhh.... Sep 18 08:05:05 that response shoudl have been a hmmm.... Sep 18 08:05:16 My network-connection dies silently if I do bit transfers under high cpu-load. Turning of turbo in sms95xx didn't help, but doubling my vm.min_free_kbytes to 32768 *seems* to. Anyone with similar experiences? Sep 18 08:05:27 eh, not bit transfers, but BIG transfers Sep 18 08:05:46 and not turning of, but turning OFF Sep 18 08:06:17 * av500 hands tasslehoff some coffee :) Sep 18 08:06:46 * LetoThe2nd tries to figure out if a tasslehoff is looking for treedom. Sep 18 08:06:51 tasslehoff: that vm min free has been cargo cult for years Sep 18 08:07:13 I have a fb0.I use overlay0 and manager0 for DVI. Sep 18 08:07:15 noone can explain to my why that works and if it should be the default :) Sep 18 08:07:40 I mirror fb0 also on to overlay1 and use manager1 for S-VIDEO. Sep 18 08:08:11 av500... are you saying that I can use manager2 to render fb0 out on to DVI ? Sep 18 08:09:57 ? Sep 18 08:10:05 you have overlay0 already on DVI, no? Sep 18 08:10:10 koen: heh, matches my research so far :) Sep 18 08:10:19 you *cannot* scale overlay0 Sep 18 08:10:22 Yes. Sep 18 08:10:44 I'm confused by the term "video plane". Does "video plane" correlate to overlay or to manager ? Sep 18 08:10:51 overlay Sep 18 08:10:55 manager does not exist Sep 18 08:11:00 got it. Sep 18 08:11:04 its just a concept Sep 18 08:11:20 Thanks! I didn't get that from the docs. Sep 18 08:11:24 so, you could point overlay1 or 2 to the same 720p buffer Sep 18 08:11:38 and scale that down for svideo Sep 18 08:11:54 Okay. I haven't tried that. Sep 18 08:12:15 but I dont know what downscaling restrictions exist for that Sep 18 08:13:14 I hear ya. I have a go / no-go requirement. 720P on DVI and NTSC on S-VIDEO. If it don't go then restrictions are too restrictive. =) Sep 18 08:13:38 Thanks again av500. Sep 18 08:18:02 well, try it out, its quite simple Sep 18 08:26:46 koen: is it assumed that it *does* help? certainly seems so for my particular case. Sep 18 08:32:39 tasslehoff: people generally say "hey, that worked" Sep 18 08:32:54 tasslehoff: but as I said, noone could tell if it's safe to make the default Sep 18 08:33:20 it really works? as-in no more usb network issues? Sep 18 08:33:28 or just "less" issues? Sep 18 08:40:07 av500: less Sep 18 08:40:16 av500: seems to be less Sep 18 08:41:25 is 'usb network issues' a common thing in our ecosystems? Sep 18 08:44:31 "EHCI has issues" is common Sep 18 08:44:53 its sometimes masked by "MUSB has issues" Sep 18 08:45:28 on 3530, we had EHCI crap out when outputting 720p over HDMI Sep 18 08:45:38 we should all just swithc back to OHCI Sep 18 08:46:06 doing any kind of usb power management on EHCI is totall hell Sep 18 08:47:51 av500: ya. we're sticking with 2.6.39, since power management and general stability has been good there (until now) Sep 18 08:50:48 av500: overlay manager does exist, it's the HW block that composes overlays and outputs the data according to the video timings to an output. Sep 18 08:51:25 tomba: right Sep 18 08:52:12 tomba: i'm still too much used to pre- DSS2 code :) Sep 18 08:52:30 feel free to suggest a better name, I have the current one =) Sep 18 08:53:41 av500: what do you mean by masked by "MUSB has issues"? We have the networking (Lan9500) on MUSB. On high load (and ~1.2MB/s on EHCI), it's the network driver that silently dies. Sep 18 08:54:23 https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap4/+bug/746137 Sep 18 08:54:28 tasslehoff, like there ? Sep 18 08:56:02 setting vm.min_free_kbytes = 8192 works around this in ubuntu (and we have a yet unapplied kernel patch to actually set the right defaults) Sep 18 08:58:48 oh, i just noticed the backlog (was scrolled far offscreen, sorry ... ) Sep 18 09:00:10 ogra_: I have two issues. one is where it silently just stops working, the other is a "NETDEV WATCHDOG". Increasing vm.min_free_kbytes seems to make the first one happen less frequently. The latter is so rare that it's really hard to say. Sep 18 09:00:13 note that this bug isnt arch specific at all, i have reports from a bunch of x86 users with USB NICs Sep 18 09:01:03 so, what is the root cause? Sep 18 09:02:07 not enough ram that gets initially allocated apparently Sep 18 09:02:26 how much ram? Sep 18 09:02:58 i think the default is around 2k, not sure, check with sysctl :) Sep 18 09:07:55 cooloney in #ubuntu-kernel worked on the kernel side fix btw ... if you want more info talk to him Sep 18 09:08:46 btw, that lp page looks totally cluttered, it needs a redesign :) Sep 18 09:10:49 haha Sep 18 09:11:11 lp isnt developed anymore, its considered done and went into maintenance mode Sep 18 09:11:46 patches accepted indeed if you want to dive into python Sep 18 09:14:38 ogra_: thanks for the info. Sep 18 10:10:52 av500, fyi http://kernel.ubuntu.com/git?p=ppisati/ubuntu-quantal.git;a=commit;h=86b83747713545234c28e5347c6f0e5efb652332 and http://kernel.ubuntu.com/git?p=ppisati/ubuntu-quantal.git;a=commit;h=89b16e837180b215b667affdd4227220827f8bb4 Sep 18 10:11:46 oh, and http://kernel.ubuntu.com/git?p=ppisati/ubuntu-quantal.git;a=commit;h=2d7b257c8fee397ec7e91b002bd0d664ebb2f726 might also be related Sep 18 11:10:54 ogra_: does those patches seem to fix the problem completely? Sep 18 11:35:12 tasslehoff, on panda they do Sep 18 11:35:39 but our omap kernel is build from mainline so patching isnt as easy as in the dedicated omap4 branch Sep 18 11:36:44 ogra_: ok. I'll try adding it to my good old kernel and see if my omap3 feels better afterwards Sep 18 11:36:58 it should Sep 18 11:37:20 you can also just try the sysctl workaround ... its the same just in a different place Sep 18 15:55:08 I compiled 3.6.0-rc6, with arm2012, for beaglboard. Now it seems the "smsc95xx" --resets-- and even sometimes causes a reboot Sep 18 15:55:16 beagleboard xM Sep 18 15:55:52 this did not happen with rc4 Sep 18 17:13:32 Hmmm Sep 18 20:29:01 QUIET DOWN people!! i'm trying to sleep... :) Sep 18 21:18:14 djlewis: difficult to sleep with people going in and out nonstop Sep 18 21:20:34 turn on deep sleep Sep 18 21:20:42 then people will stop going in and out! Sep 18 21:26:02 ds2, there's errata preventing entering deep sleep Sep 18 21:26:14 will fix in next spin! **** ENDING LOGGING AT Wed Sep 19 02:59:58 2012