**** BEGIN LOGGING AT Thu Sep 20 02:59:58 2012 Sep 20 06:51:43 koen: Angstrom in oe-core should work fine with networkmanager instead of connman? Sep 20 07:31:30 tasslehoff: yes, angstrom supports ifupdown, connman and nm Sep 20 07:31:43 tasslehoff: it's just that images that need to boot fast don't use nm, it's too slow Sep 20 07:33:12 koen: ok. I'm having some issues where the network connection suddenly disappears, and I wonder if it may be connman-related. Sep 20 07:40:34 mranostay: thanks for reminding me my rss feed was broken Sep 20 07:48:36 tasslehoff: which plateform, and kernel version ? Sep 20 07:53:14 bizulk: our own beagleC3-like board with 2.6.39 Sep 20 08:34:55 http://i.imgur.com/8JpPY.png Sep 20 08:35:00 TI has free samples! Sep 20 08:46:44 my xM reboot event with the Angstrom Demo Images. (there is something wrong with the usb or ethernet over usb). Does someone have a clue ? Sep 20 08:48:20 Hi at all, Sep 20 08:49:55 hm. if I only have a wired network interface that should be brought up with static or dhcp, is there a reason to use anything more fancy then ifupdown? Sep 20 08:50:20 no Sep 20 08:50:21 I want a service with console when my beaglebone boot, but the same console is use for shell and my service, how can I start my service with consola? Sep 20 08:52:18 console ? you mean serial ? Sep 20 08:53:31 when my beaglebone start my service write in console, but if push any key of my keyboard, my service doesn't detect it, instead of appear shell Sep 20 08:54:07 Yes, serial console, I use ttyO0 in my beaglebone Sep 20 08:58:37 you must disable your console on sérial (or just let it for boot time), or use another serial line (USB to SERIAL for example) Sep 20 09:29:27 usb reboot : that should be my power suppluy Sep 20 09:56:50 hi beagle, i am using am3358evm kit for development.First of all , how to debug kernel messages.I couldnt find any logs from J12 USART. Am i debugging properly ? Sep 20 09:57:47 we would not know, nobody here has an am3358evm kit Sep 20 09:58:41 Am i in correct channel??? Sep 20 09:58:59 you are in the beagle and beaglebone channel Sep 20 09:59:36 which is the correct place to ask about am3358? Sep 20 09:59:57 the chip or the evm? Sep 20 10:00:24 evm Sep 20 10:00:38 I have evm Sep 20 10:00:58 console is definetely on J12 Sep 20 10:01:06 how you debugging it? Sep 20 10:01:35 can you see anything on J12? Sep 20 10:02:24 i will confirm once here Sep 20 12:14:08 koen: I threw out nm and conman. To make /etc/init.d/networking run again, what tricks must I pull? I've removed the 'networking.server -> /dev/null' symlink in systemd-compat-units.. Sep 20 12:16:20 tasslehoff: cbrake has a neat udev based solution for that Sep 20 12:22:57 * tasslehoff hopes neatness will flow from irc Sep 20 12:23:56 koen: so the issue is binding together network'y stuff with the devices and drivers being found/initialized? Sep 20 12:24:52 yes Sep 20 12:25:09 when e.g. eth0 appears it will do an ifup on that if Sep 20 12:25:35 tasslehoff: he described it on the systemd mailinglist, but I haven't seen the complete implementation yet Sep 20 12:26:42 koen: ok. for now I can just do a hack to make sure I call ifup sufficiently long after the drivers are initialized. thanks. Sep 20 14:36:58 mru: wrt tms470, does it compile cleanly now? Sep 20 14:37:12 not quite Sep 20 14:37:22 it's getting there Sep 20 14:44:17 the latest version somewhat annoyingly renamed all the executables, some of them clashing with ones from armcc Sep 20 14:44:27 so you can't have both in your PATH any more Sep 20 14:46:28 gee Sep 20 14:47:05 actually, I didn't try with the TI tools first Sep 20 14:47:18 maybe armcc invokes its support stuff with full path Sep 20 15:34:06 hi all Sep 20 15:53:46 hi. Still on tidspbridge tests.... with Sep 20 15:54:20 .3.6-rc6 when I load the tidspbridge drivers (modprobe with no arguments), cat /proc/interrupts returns : 25: 2 INTC OMAP DISPC, omapdss_dsi.0 Sep 20 15:54:21 28: 0 INTC DspBridge iommu fault Sep 20 15:54:21 36: 0 INTC dsp_wdt Sep 20 15:54:35 pastebin Sep 20 15:54:49 oops sorry. Sep 20 15:56:54 http://pastebin.com/QgZAab32 Sep 20 15:57:11 I got an error when loading baseimage.dof Sep 20 16:08:55 Is the mem option still mandatory to get this module work ? Sep 20 16:10:19 mem? Sep 20 16:11:35 av500: uboot mem option to reserve memory for IPC communication Sep 20 16:12:07 * av500 forgot everything about dsp bridge Sep 20 16:12:43 no mem= needed Sep 20 16:13:37 first rule of dsp club Sep 20 16:13:58 av500: ok, but syslink is for omap4 (i'm working on omap3530). dslink is not in main kernel tree. Sep 20 16:14:26 mranostay: and the seconde rule ? :D Sep 20 16:14:32 isn't it rpmsg now? Sep 20 16:15:24 not on omap3 Sep 20 16:15:41 bizulk: i did not say you should use dsplink Sep 20 16:15:44 or syslink Sep 20 16:15:52 i just said that I forgot about dspbridge Sep 20 16:16:00 mranostay: see http://omappedia.org/wiki/Category:RPMsg (it's for omap4 and beyond) Sep 20 16:16:01 of course, omap3 is forgotten and abandoned Sep 20 16:16:03 so I checked my sources Sep 20 16:16:37 bizulk: the android ports for omap3, do they use dspbridge or link? Sep 20 16:17:10 av500: it seems to be dspbridge Sep 20 16:18:04 av500: it is based on "openmax" project that is based on dspbridge (http://omappedia.org/wiki/OMX_Android) Sep 20 16:19:28 hahahahhaa Sep 20 16:19:31 openmax Sep 20 16:19:56 bizulk: yes, android uses openmax Sep 20 16:20:00 that is not ti specific Sep 20 16:21:06 av500: I mean that omappedia/openmax implémentation on omap3 uses tidspbridge. Sep 20 16:21:40 there is gst-dsp alternative, still using the same stuff ... but you may know about that Sep 20 16:21:56 or gstreamer-ti Sep 20 16:22:24 that has 100% less felipec Sep 20 16:26:03 ahhahahahhahah Sep 20 16:33:45 opemax is always vendor specific hahah Sep 20 16:36:07 it's one of those standards that's actually not a standard at all Sep 20 16:36:24 because the standard functionality is useless so everybody does what they need as vendor extensions Sep 20 16:36:28 and do it differently Sep 20 16:57:25 koen: can you be my proxy? Sep 20 17:00:12 well. I'm still stuck with the same ch..t :( . But that's enough for today Sep 20 17:12:23 XorA|gone: yes Sep 20 17:29:33 I just had an idea for a troll Sep 20 17:29:53 walk into an apple store and complain about some made-up problem with "your iphone" Sep 20 17:30:13 when the support dude(tte) asks to see, pull out a samsung phone Sep 20 17:31:15 when they immediately point out the "incorrect" hardware, say something along the lines of "but I heard in the news that samsung phones were the same thing" Sep 20 17:32:13 awesome Sep 20 17:34:17 or get your hands on an actual iphone, Sep 20 17:34:19 then go to the shop and say you ordered it online, but when you opened the box there was a samsung inside Sep 20 17:34:24 show them the iphone Sep 20 17:34:29 but insist that it is a samsung Sep 20 17:49:22 koen: thanks, Ill send you the form before meeting Sep 20 19:07:38 one of these days i'm going to take out a patent on chewing foood Sep 20 19:08:58 then we'll have to find some food that doesn't need chewing Sep 20 19:08:59 like beer Sep 20 19:10:11 beer++ Sep 20 19:11:16 <3 beer Sep 20 19:14:18 stout has sufficed as a food at various times Sep 20 19:14:22 one of these days you'll be eatting only food that doesn't require chewing Sep 20 19:30:18 doppelbocks are the original "liquid bread" Sep 20 19:30:39 funny to see even the wikipedia entry use the slang term Sep 20 19:32:09 ds2: you mean when it's time for http://sv.wikipedia.org/wiki/%C3%84ttestupa Sep 20 19:34:07 don't speak SV Sep 20 19:34:22 the article is only available in sv and fi Sep 20 19:35:21 "ättestupa" in swedish folklore is a cliff where in the old days they'd chuck old people to get rid of them Sep 20 19:35:33 oh Sep 20 19:35:49 modern historians believe it to be a myth Sep 20 19:36:49 wait til they find the pile of bones in some fjord Sep 20 20:43:44 hmm, two people agree on an sd card brand Sep 20 20:43:50 hell just froze over Sep 20 20:45:06 Crofton: a few weeks? Sep 20 20:45:13 Crofton: to detect a chipmunk? Sep 20 20:45:34 Crofton: bill gave the cards to me to test :) Sep 20 20:53:50 heh Sep 20 20:55:56 someone in open source agreed? Sep 20 20:56:06 that really is a moment for celebration :-D Sep 20 20:59:23 I don't think that's justified Sep 20 21:00:35 we are bit banging rs232 from an avr :) Sep 20 21:04:10 I used to have to bitbang SPI on a PIC, due to the clock not actually being synchronous :-( Sep 20 21:06:45 PIC or it didn't happen Sep 20 21:28:29 So maybe my google-fu isn't what it should be, but what is an MPU, as used by ARM, and how does it differ from an MMU? Sep 20 21:28:56 memory protection unit Sep 20 21:29:05 indeed, that's the acronym Sep 20 21:29:11 but what is it? Sep 20 21:29:13 the -r and -m cores have those Sep 20 21:29:22 it's what the name says Sep 20 21:29:28 how is it different from an MMU? Sep 20 21:29:35 doesn't an mmu offer memory protection? Sep 20 21:29:36 you can define regions with different permissions Sep 20 21:29:40 but there's no remapping Sep 20 21:29:48 ah, ok Sep 20 21:30:03 so it's for lower-end parts, likely with no OS running Sep 20 21:30:07 it's very useful for write-protecting things that should be read-only Sep 20 21:30:29 uclinux can run, after a fashion, on such systems Sep 20 21:31:18 if the hardware supports user/kernel privilege separation, it lets you isolate processes like on a full mmu system Sep 20 21:31:27 just without the flexibility of virtual addresses Sep 20 21:31:33 yeah Sep 20 21:31:40 so you need to be careful to avoid memory fragmentation Sep 20 21:32:12 so that's not a bad deal then, for systems that are light on memory or processes or both, or can have it efficiently managed manually. Sep 20 21:33:07 So you're a NEON guy.... is there any NEON in Android? Sep 20 21:51:20 there's some neon here and there Sep 20 21:51:47 I don't work directly on android Sep 20 22:03:07 How do I use the c64x dsp on this gizmo Sep 20 22:03:16 The word syslink scares me Sep 20 22:03:21 with difficulty Sep 20 22:04:28 I've used code composer with a TMS320c6x, but I just used a USB jtag to program Sep 20 22:05:59 Specifically this one: http://www.ti.com/tool/tmdsdsk6713 Sep 20 22:14:51 Oh well, I'll attempt later when I'm more awake Sep 20 22:45:40 Mmmmmmmm MPUs Sep 20 23:09:15 well this meeting end... Sep 20 23:09:22 *will Sep 20 23:09:53 the door is right there... Sep 20 23:22:37 mru: well i do need my paycheque Sep 20 23:25:53 mranostay: need alt? Sep 20 23:26:07 mranostay: fake a phone call, furrow your brow like it's important, plug your other ear and walk out. Sep 20 23:26:23 In fact, what's your number... I'll give you a real phone call Sep 20 23:27:36 awww man... it's not on your linkedin profile :) Sep 20 23:28:19 heh Sep 20 23:29:56 heh Sep 20 23:40:22 you know what I'd like (and feel free to tell me where it is....)? I'd like to be able to jump to an item in menuconfig from the search results. Sep 20 23:41:18 maybe even be able to walk down through the dependencies of a module I found from search and turn it on, right from there. Sep 21 00:13:56 alan_o: I'd like the search to show the thing I'm looking for at the top Sep 21 00:14:35 yeah, that too. Sep 21 00:15:17 but I guess that brings is back to the old #pragma dwim Sep 21 00:16:00 I think that's what Java was supposed to have. Sep 21 00:16:24 and instead we got org.springframework.aop.framework.AbstractSingletonProxyFactoryBean Sep 21 00:17:22 wtf is that? :) Sep 21 00:17:35 people code in java? Sep 21 00:17:40 http://static.springsource.org/spring/docs/2.5.x/api/org/springframework/aop/framework/AbstractSingletonProxyFactoryBean.html Sep 21 00:17:51 mranostay: no, people don't code in java Sep 21 00:17:57 they make frameworks Sep 21 00:17:57 It would be nice if there was some way for drivers to specify what parts they match. Too often you're grepping for 2515, and you have to know that the driver is 251x, and sometimes 2x15, etc. If there were some regex or something that could go in the descriptions, or another field or something. Typically, _I_ can find what I'm looking for in there, but I'd have a hard time explaining it to someone else other than saying things like "be s Sep 21 00:17:58 mart about it." Sep 21 00:17:59 touche Sep 21 00:18:07 hah Sep 21 00:18:26 You know, there's this idea among some, that if you want to make stuff cross platform, you have to do it in Java. Sep 21 00:18:28 cpu_is_34xx() Sep 21 00:18:28 have to Sep 21 00:18:42 or, it's the best way, or right way or whatever Sep 21 00:18:49 Maybe in 1997 Sep 21 00:19:40 back when I was TA-ing java at university, I'd try to show the students how to do some of the things in other languages too Sep 21 00:19:52 At Microchip for example, all their new tools are in Java. Their IDE is Netbeans. Freescale's IDE is eclipse. Installs a JVM for you as part of it, nice.... Sep 21 00:20:02 my favourite was a page-long example that I replaced with a single line of perl Sep 21 00:20:03 how'd they respond to that? Sep 21 00:20:08 hah Sep 21 00:20:22 a proper line, <80 chars Sep 21 00:21:01 java is cross-platform as long you have _exactly_ the right jvm available Sep 21 00:21:11 There was no PERL at my university, that's for sure. They valued doing things in fake languages and for fake processors as to not color our judgement of things we'd encounter in the real world. Sep 21 00:21:22 so the only way to run anything remotely complex is to ship the right jvm for the only target you care about (windows) alongside it Sep 21 00:21:26 believe me, you don't have to sell me on the pitfalls of Java :) Sep 21 00:21:37 yes, exactly that Sep 21 00:21:50 and every program has a JVM, and you have the JVM installed many times on your HD Sep 21 00:22:00 and _that_ is the only way to do it "right" Sep 21 00:22:02 we ran a mix of DEC Alpha and Sun machines Sep 21 00:22:17 CS ran Sun and EE ran Alpha Sep 21 00:22:36 That's efficient Sep 21 00:22:46 We had Sun machines. Wrote Java and C on them. I loved going in there and using CDE Sep 21 00:22:52 they all played nicely together with kerberos and afs Sep 21 00:23:11 the admins were the sort that if you needed something, they'd install it for you Sep 21 00:24:01 I had GNOME whatever version at my apartment, and it was good, but there was something about CDE. Mostly I think it was because I couldn't get it at my house. Nothing makes you want something like being told you can't have it :) Sep 21 00:24:07 Good admins, that must have been nice Sep 21 00:24:22 Ours were ok I suppose. Quotas was our problem. Sep 21 00:24:29 I even got them to install some of my patches Sep 21 00:24:43 I got the staff quota :) Sep 21 00:24:49 so I survived Sep 21 00:25:08 yeah, I found out that I could get it bumped up to like 7M if I told them I wanted StarOffice. Sep 21 00:25:23 So it installed a bunch of user data in ~ iirc, so the quota had to be higher Sep 21 00:25:50 so then I removed staroffice (to get the space back), and my quota dropped to 5... seemingly automatically. Sep 21 00:25:53 when I started in 99 the student quota was 50MB iirc Sep 21 00:26:08 Never figured that one out. Must have been some setting somewhere to reclaim it if I wasn't using it. Sep 21 00:26:09 that increased to considerably more over the next few years Sep 21 00:26:28 yeah, this was in 99 also :) Sep 21 00:26:32 it was stupid Sep 21 00:26:49 you could have created a large file with a filesystem inside Sep 21 00:27:03 or at least some kind of archive format Sep 21 00:27:19 oh yeah, if I had known the quota was going to come back down. didn't know it would do that till after it did it :) Sep 21 00:27:51 I installed VNC on that system, so I could get to it from home. Sep 21 00:28:05 eventually they plugged that hole. Sep 21 00:28:19 I ran X over ssh Sep 21 00:28:29 yeah, this was before I had ssh. Sep 21 00:28:34 before the patent expired Sep 21 00:28:35 worked a treat on the 10Mbps ethernet connection I had at home Sep 21 00:28:54 I had dialup, vnc worked pretty well, surprisingly Sep 21 00:29:02 are we talking about abusing student accounts? :) Sep 21 00:29:11 no...... Sep 21 00:29:18 mranostay: technically, mine was a staff account Sep 21 00:29:20 I barely abused mine, more like made it usable Sep 21 00:29:42 heh i still have student email they haven't turned off :) Sep 21 00:29:45 and the admins were well aware of most things I did Sep 21 00:29:53 I think I could have gotten ssh and built it myself, but at that time I wasn't interested, and therefore didn't know all of what it could do. Sep 21 00:29:56 mranostay: same here Sep 21 00:30:03 hah Sep 21 00:30:09 mru: i use that to abuse the samples system :) Sep 21 00:30:22 I have several aliases there that still work actually Sep 21 01:00:41 *sigh* Sep 21 01:01:03 how come..... vendors can't give you a proper defconfig? Sep 21 01:01:25 This one in particular, gives me a defconfig that builds a kernel that won't fit in the kernel portion of the nand..... Sep 21 01:01:44 it's significantly larger than the pre-built kernel you can download from their website Sep 21 01:02:23 defconfigs are overrated Sep 21 01:02:35 just spend the 30mins and go through the entire config yourself Sep 21 01:02:56 if they turned stuff you needed off you'd be upsettoo mostlikely Sep 21 01:03:10 know thy system. Sep 21 01:03:28 I guess... but they give me a kernel, prebuilt, that boots the board, then they give me source, without a config that can build that kernel that they shipped me. Sep 21 01:03:46 /proc/config.gz? Sep 21 01:03:59 oh yeah, I can go digging for it... but that's not the point :) Sep 21 01:04:05 really? Sep 21 01:04:13 you can run a script to pull it out Sep 21 01:04:26 oh yeah that's true. Sep 21 01:04:37 iirc, that script doesn't work all the time Sep 21 01:06:24 of course usually i get a vendor 2.6.30 kernel or something god awefule old Sep 21 01:06:35 at least it is a 2.6 kernel Sep 21 01:06:41 2.6.33.20.1 for this one Sep 21 01:06:48 what chip? Sep 21 01:07:16 Sure, I can bring a new kernel on, make my own config, etc, but I figured I'd start with known working configuration and then break stuff one at a time. Sep 21 01:07:38 It's an at91sam9620 Sep 21 01:07:49 oh those. Sep 21 01:07:59 iirc, not all of that is upstream Sep 21 01:08:18 I determined on quick look at the diffs that I can probably just pull the board file out and put it into a mainline kernel. Sep 21 01:08:38 the newer cortex based AT91's are upstreamed? Sep 21 01:08:44 * ds2 looks Sep 21 01:09:12 they have significant differences between theirs and the mainline of that version, but I determined a lot of it to not matter for this chip, as if they applied some generic patch from atmel a couple years ago. Sep 21 01:09:32 oh indeed they are Sep 21 01:09:45 Their support people were less than useless when asked questions. Sep 21 01:09:56 hehe Sep 21 01:10:09 maybe the problem is my worldview.... I believe that when you buy something that it should work in the best possible way. Sep 21 01:10:44 work is subjective Sep 21 01:10:46 Like when I bought my first beagleboard... and plugged it into the USB and watched it kernel panic. Sep 21 01:10:50 so is 'best' Sep 21 01:11:01 yeah Sep 21 01:11:47 when I came in this channel to ask about that beagle people said, "hey, be glad you've got _that_ much" when I told them that it panic'd immediately upon boot. Sep 21 01:12:13 panicing kernels is a good thing Sep 21 01:12:18 despite what it appears Sep 21 01:12:25 is this back to "good" is subjective Sep 21 01:12:26 ? Sep 21 01:12:39 well... Sep 21 01:12:49 I would be overjoyed if I saw the kernel panic Sep 21 01:12:58 you might have been that guy...... Sep 21 01:13:13 the more 'interesting' problem is when it doesn't panic and generates absolutely no output! Sep 21 01:13:29 yes, I was about to say Sep 21 01:13:41 inifinitely better than turning it on and it doing _nothing_ Sep 21 01:13:52 this applys to any platform Sep 21 01:14:00 if I can see a panic, I know where to look for problems. Sep 21 01:14:06 yeah Sep 21 01:15:01 ok, I'm going to have to retract my lamentation from before Sep 21 01:15:08 in the name of being fair and balanced..... Sep 21 01:15:21 it was my fault the kernel was too big Sep 21 01:15:27 I built in CAN Sep 21 01:15:39 heh Sep 21 01:15:39 went back to the vendor defconfig and it's small enough Sep 21 01:15:44 could be worse Sep 21 01:15:47 * alan_o eats crow Sep 21 01:15:52 could have been ur initramfs ;) Sep 21 01:16:14 plenty of space in the nand for that. Sep 21 01:16:24 eh? Sep 21 01:16:35 well... I mean the rootfs Sep 21 01:16:36 initramfs's are generally compiled into ur uImage/zImage/etc Sep 21 01:16:50 yeah, not doing that here. It's not an initramfs Sep 21 01:17:11 Do you do that typically? Sep 21 01:17:27 depends on what I need Sep 21 01:17:46 I read somewhere (maybe in Documentation/ ) that if you do that, your initramfs contents now have to be GPL Sep 21 01:18:23 my boundary is often at the 'it starts init so chuck it to the OE guy' line Sep 21 01:18:40 I wish I had an OE guy..... Sep 21 01:18:42 to get something small, the off the shelf stuff is GPLed so that is moot Sep 21 01:18:48 yeah Sep 21 01:19:08 but if you're going to put your whole system in the initramfs, which I've done before..... Sep 21 01:19:15 OE can be made to work. if folks insists.... Sep 21 01:20:07 true Sep 21 01:21:31 Actually, I have to retract that too... on those systems the proprietary part is on another partition (FAT32) Sep 21 01:22:00 I'm in the difficult position on that one of having to make something that non-Linux people (and non-embedded people) can configure/maintain. Sep 21 01:22:34 that's normal Sep 21 01:22:55 I don't have to like it :) Sep 21 01:23:04 that's again one of the things to route through the OE guy ;) Sep 21 01:23:24 yes, stuff to throw on the SEP pile Sep 21 01:23:35 don't mind writing BB files but need someone else to front them Sep 21 01:30:43 man... ok... looking at the history, I looked at the defconfig, then ran make menuconfig, but never actually made the defconfig. That's how I screwed it up. It wasn't CAN support. ugh... I have to pay more attention Sep 21 01:36:38 you actually use make to copy a file?! Sep 21 01:37:05 yeah, make ${board}_defconfig Sep 21 01:37:10 blah Sep 21 01:37:18 cp arch/arm/configs/foo .config Sep 21 01:37:20 much easier Sep 21 01:37:27 mine's less characters :) Sep 21 01:37:33 (but yours autocompletes) Sep 21 01:37:41 mine is less obfuscated Sep 21 02:21:25 I love me a hard-wired serial port. It just feels right. Sep 21 02:22:27 The address never changes. Always works. No udev-fu that works until you have an open handle to something that gets reconnected. **** ENDING LOGGING AT Fri Sep 21 02:59:58 2012