**** BEGIN LOGGING AT Mon Mar 12 02:59:58 2012 Mar 12 03:03:48 Hi everyone, I am using beagleboard-xm rev c and my board has trouble detecting usb after the installation of pre built image from the online builder. Mar 12 03:03:55 here is my paste bin.. http://pastebin.com/mqQmF4E7 Mar 12 03:04:20 Would appreciate much if someone could help me with this.. thanks! Mar 12 03:09:36 or is there any angstrom linux 3.0 prebuilt image for me to install? Mar 12 03:32:56 are all the GPIOs on beagle xm's expansion port supposed to be both input and output? Mar 12 03:33:20 half of mine don't seem to work as output, and only 4 work as input Mar 12 03:35:48 W1N9Zr0: you're doing it wrong Mar 12 03:36:14 i only poked it with a multimeter D: Mar 12 04:29:13 prpplague, are flyswatter2 schematics open? Mar 12 06:28:31 hi Mar 12 06:28:43 where can I download the ksplash.ko sources? Mar 12 06:31:28 or how can I exchange the beagleboard logo + kernel log by a fullscreen picture? Mar 12 13:27:41 hello guys, *WARNING* beginner question coming Mar 12 13:27:55 * _av500_ braces Mar 12 13:28:21 I've installed ubuntu 11.10 in my beagleboard xm. I am trying to play audio but although I don't getany error I don't get any audio out of it... Mar 12 13:28:29 does anyone has any pointer? Mar 12 13:28:45 I've tried to use aplay, or alsa or python audio, or paython alsa... Mar 12 13:28:56 <_av500_> #ubuntu-arm might have pointers Mar 12 13:29:03 it appears like if it is playing... but I don't hear anything Mar 12 13:29:06 <_av500_> are you suing the official ubuntu or rcn? Mar 12 13:29:11 <_av500_> using Mar 12 13:29:13 oh... is this chat only for Angstrom? Mar 12 13:29:16 <_av500_> no Mar 12 13:29:29 <_av500_> but there is a dedicated chat for ubuntu on arm Mar 12 13:29:41 cool I didn't know that. Thanls Mar 12 13:29:45 thanks Mar 12 13:30:03 <_av500_> are you using official ubuntu or rcn? Mar 12 13:30:39 ehmm... not sure, I followed instructions in here: http://elinux.org/BeagleBoardUbuntu#Oneiric_11.10 Mar 12 13:30:48 is it officila? Mar 12 13:32:02 <_av500_> if you got the image from ubuntu.com then yes Mar 12 13:32:47 <_av500_> and that page mentions ubuntu-arm too Mar 12 13:32:56 <_av500_> tried amixer? Mar 12 13:33:03 just noticed: the link seems to point to rnc .... Mar 12 13:33:20 rcn-ee.net Mar 12 13:33:26 <_av500_> thats rcn's private build Mar 12 13:33:32 <_av500_> he is here sometimes Mar 12 13:33:36 <_av500_> not atm Mar 12 13:33:39 <_av500_> tried amixer? Mar 12 13:33:41 ah! Mar 12 13:33:48 no, I will try amixer Mar 12 13:34:07 and maybe an official build then. Thanks for clarifying this Mar 12 13:40:29 @av500 another question if you don't mind: I presume the RobertCNelson is another private build then (just found in the same poage). Point is: is the official ubuntu image not good enough or is just that people need different things? Mar 12 13:43:05 <_av500_> RobertCNelson = rcn Mar 12 13:43:13 <_av500_> he makes his own ubuntu build Mar 12 13:43:32 <_av500_> sincei dont use ubuntu, i have no idea Mar 12 13:43:43 right ... thanks Mar 12 14:09:10 hello Mar 12 14:09:26 Can this safely be ignored? : angstrom DOES NOT support libiconv because the eglibc provided iconv library is used Mar 12 14:10:28 <_av500_> i would ignore it Mar 12 14:10:59 thx Mar 12 14:12:03 <_av500_> but then, I am ignorant.... Mar 12 14:12:50 _av500_: nice to start the morning off with some truth Mar 12 14:12:55 * prpplague jokes with _av500_ Mar 12 14:14:08 Nothing wrong about being ignorant... Mar 12 14:25:54 Martron: thanks, the GOP presidential candidates should feel better knowing that Mar 12 15:01:09 Does the GOP still exist? :) Mar 12 15:01:26 group of pictures? Mar 12 15:02:12 heh Mar 12 15:02:19 over here in europe we're busy fixing our own fincancial problems :) Mar 12 15:02:50 *ahem* channel getting too real again :) Mar 12 15:03:05 but don't lets talk about politics...this channels deserves better :) Mar 12 15:03:29 <_av500_> these days a GOP can be larger than a GDP... Mar 12 15:03:30 lets not get into another #libertariancirclejerk talk :) Mar 12 15:04:08 this DST is really screwing me up.. Mar 12 15:04:24 if we only legalise all drugs, everything else will sort itself out Mar 12 15:04:27 <_av500_> try SRC Mar 12 15:04:39 screw GOP and GDP as long as we have the GPL Mar 12 15:04:54 <_av500_> all 3 of them Mar 12 15:05:12 mru: at least we'll be too high to care :) Mar 12 15:07:54 * mranostay hides from kkeller Mar 12 15:08:10 yo hello mranostay Mar 12 15:09:28 mranostay: seems you're not hiding very well Mar 12 15:09:49 * Crofton|work watches the Colbert report Mar 12 15:10:48 report what to whom? Mar 12 15:21:05 Martron: only in name does it exist Mar 12 15:38:50 guys, is there any ubuntu on beagleboard user in here? Mar 12 15:39:19 mik__: probably easier to ask over in #ubuntu-arm Mar 12 15:39:58 but is that a chat as well? av500 told me about it but I could not find it... let me google once again Mar 12 15:45:56 mik__: just do this command in your client: /join #ubuntu-arm Mar 12 15:46:59 thanks... I better find another client... I was trying to use pdgin Mar 12 15:47:20 mik__: it works in pidgin Mar 12 15:47:59 ah.. and server is: #ubuntu-arm I guess Mar 12 15:48:27 server is irc.freenode.net Mar 12 15:48:48 the channel is #ubuntu-arm Mar 12 15:49:44 is channel same thing than alias? sorry I am dumb... Mar 12 15:50:58 i think you can leave alias blank Mar 12 15:51:18 I don't find where to set the channel... grrr. Mar 12 15:51:20 you add channels after you connect to the network Mar 12 15:51:30 ah that explains Mar 12 15:52:18 I did it... Mar 12 15:52:30 thanks, I feel less idiotic... :-) Mar 12 15:54:46 does opkg upgrade also upgrade the kernel ? Mar 12 15:57:47 I am sure it updates a kernel module (I didi it for cdc-acm) but not sure about the whole kernel... Mar 12 15:59:11 mik__, Ah it looks like it does except the boot partition isnt mounted so I guess I'll have ot copy the files over Mar 12 16:19:25 prpplague: ping Mar 12 16:19:48 emeb: pong Mar 12 16:27:33 the beagleboard evaluation angstrom shows a beagleboard.com logo instead of a penguin Mar 12 16:27:46 when I want to exchange this picture, I have to recompile the kernel :( Mar 12 16:27:59 is there a way to insert an animation instead of this logo? Mar 12 16:58:08 Built a kernel with CONFIG_DAVINCI_EHRPWM=y but there is only ecap in /sys/class/pwm, any idea what could be the problem? Mar 12 17:27:56 sure is quiet in here Mar 12 17:28:25 * _av500_ throws a few chair around Mar 12 17:29:17 * LetoThe2nd cranks up an eläkeläiset record to max volume. Mar 12 17:34:44 _av500_: you forgot to yell "developers, developers, developers" Mar 12 17:35:11 oO( "humppa" ) Mar 12 17:36:45 * mranostay gets hit by _av500_'s flying chair Mar 12 17:36:51 * mranostay sues _av500_ Mar 12 17:36:59 * dm8tbr finds it entertaining, that the 'retirees' are more popular in germany than in their home country Mar 12 17:42:26 mru: what beer is that in your G+ photo? Mar 12 17:48:08 brewdog Mar 12 17:50:38 never heard of it Mar 12 17:50:44 * mranostay wonders if you can get in the States Mar 12 17:50:49 you should try it Mar 12 17:50:58 probably hard to get on your side of the pond Mar 12 17:52:23 heh a lot of IPAs I see Mar 12 17:52:48 I have a love/hate relationship with IPAs. Mar 12 17:53:04 love the good ones, hate the bad ones? Mar 12 17:53:30 a brew specifically designed to keep for long time on the boat ride from UK -> IN. Mar 12 17:53:42 no i love them all but they give me a lovely hangover the next day Mar 12 17:54:03 never had that problem with ipa Mar 12 17:54:07 emeb: yes i know the history. i think we are all functionally alcoholics in here :) Mar 12 17:54:19 s/ly// Mar 12 17:54:29 Hi All! I need help with the beaglebone, as I'm completely stuck. All I need is an any Linux distribution on Beagle that has a) userspace access to analog input and b) userspace access to SPI. Default Angstrom image does have access to ADC, but no SPI. When I try to rebuild kernel (or to use Ubuntu distro), I got SPI access, but no ADC. I'm by no means kernel developer, all I need is to develop application, and not to dive into kernel d Mar 12 17:54:41 Is there ANY distro that has both? Mar 12 17:54:59 build both into the kernel? Mar 12 17:55:47 kosta, let me check I think arch might have both Mar 12 17:55:55 Have no idea how to do this. Well, I was able to find out magic commands to build SPI into kernel, but I have no idea what to use for ADC Mar 12 17:56:41 jsabeaudry - thans, I've already tested Arch, it has none :( Mar 12 17:56:50 mru: so you guys have microbreweries too? Mar 12 17:58:10 kosta, Just check and arch has both spidev2.0 and ain Mar 12 17:58:15 checked* Mar 12 17:58:32 Cool, do you have a link? Mar 12 17:58:39 right polarity and speed? Mar 12 17:59:03 mranostay: all the good stuff comes from smallish breweries, don't know what the limit is for 'micro' Mar 12 17:59:41 i think that is an american term Mar 12 17:59:43 Any idea which SPI mode you need kosta? Mar 12 17:59:55 I used instructions from archlinuxarm.org. When I build an image, it had no ains in platform/tsc. May be I was looking in a wrong place? Mar 12 18:00:07 since most of the crap people drink here comes from huge hundred thousand litre vats Mar 12 18:00:36 kosta, just download the prebuilt and update with pacman -Syu Mar 12 18:01:06 kosta: as soon as some actually answers when being asked the polarity/speed question, we can't provide spidev in the default kernel Mar 12 18:01:10 kosta, Here is in /sys/devices/platform/tsc/ain1 Mar 12 18:01:27 jsabeaudry - just any. I'm very novice, and got taks at work to communicate with an SPI device. With no docs or samples all that I was able to find is to read/write to /dev/spi*. Well, default Angstrom has not such files, i.e. no usedspace SPI Mar 12 18:01:28 mranostay: wikipedia has definitions of the term, go figure Mar 12 18:01:41 * mranostay changes the definition Mar 12 18:02:09 kosta, You will soon discover that just "communicate using SPI" is too vague and you need specs on the speed and mode as koen points out Mar 12 18:02:42 koen, Would it be possible to change the speed and polarity the the spidev device using a sysfs driver? Mar 12 18:03:00 if you extend the spidev driver, certainly Mar 12 18:03:32 that might be a graceful option for including it in the "standard" kernel Mar 12 18:03:38 * koen afk Mar 12 18:05:06 jsabeaudry, thank you - I'll try this. As for specs on speed and mode - well, one step at a time. First I need to at least see SPI device. After that, I'll start to worry about speed and mode. Actually, speed is non-essential Mar 12 18:06:00 As for mode - I know nothing about different modes yet. If you can give me link to FAQ, I'll be very thankful :) Mar 12 18:06:12 kosta, wikipedia SPI Mar 12 18:08:52 jsabeaudry: oh, this what you refer to. I thought you were talking about something else. Speed is 12Mhz, mode is 3 Mar 12 18:09:56 I was thinking about manual bitbanging, but that was too slow Mar 12 18:10:14 * mranostay holds back joke Mar 12 18:12:09 jsabeaudry - thank you for your help Mar 12 18:12:15 * kosta afk Mar 12 18:17:25 * GrueMaster is getting desparate. No networking on beagleXM rev B initially. Mar 12 18:18:17 Works with the angstrom validation image, but not with Ubuntu Oneiric or Precise kernels (which are very close to mainline). Mar 12 18:18:50 I have gotten it to work when I compile the smsc95xx module into the kernel, but not when it is loaded as a module. Mar 12 18:39:20 GrueMaster: did you missed "omap is totaly fucked up" thread? :P Mar 12 18:39:59 Must have. Mar 12 18:40:22 ynezz: guess i missed that one too Mar 12 18:41:14 linky poo? Mar 12 18:44:55 jsabeaudry: seems like you know good deal about SPI. May be I'm doing it wrong to begin with? I'm writing in C/C++, and in the end all I need is some function like writeToSPI(data, options). Doing it via file i/o seems like lots of overhead. May you can point me to right direction? The only example I was able to find in this regard was some kernel module example, and want to avoid that as my knowledge of Linux internals is close to zero Mar 12 18:46:35 BTW, similar function for analog input would be also nice:) I made it work via file i/o, but it seems too cumbersome approach Mar 12 18:46:38 prpplague: http://www.spinics.net/lists/linux-omap/msg65779.html Mar 12 18:46:42 ^ GrueMaster Mar 12 18:47:34 I've seen someone complaining about similar issues (module/not module) here also, but it was probably something else then ethernet Mar 12 18:47:47 * prpplague looks Mar 12 18:47:49 kosta: In order to do SPI from userspace (the right way), you'll need to attach the spidev driver to one of the SPI ports. This requires a kernel modification and re-build. Mar 12 18:47:50 Meh, that is kernel next. I am using 3.0 for Oneiric, 3.2 for Precise. Mar 12 18:48:08 well, it was rather meant as a joke :) Mar 12 18:48:18 kosta: there's a patch floating around, let me find... Mar 12 18:48:48 kosta: https://groups.google.com/forum/#!topic/beagleboard/B3akyoyjwG4 Mar 12 18:48:49 I think what you saw earlier was [06:28:23] I've installed ubuntu 11.10 in my beagleboard xm. I am trying to play audio but although I don't getany error I don't get any audio out of it... Mar 12 18:49:23 kosta: that's for the bone, there's a similar one for the beagleboard. Which are you using? Mar 12 18:49:34 I was chatting with him on #ubuntu-arm. Appears our daily ubuntu-server image has...issues. I'm debugging it now. Mar 12 18:52:03 GrueMaster: mixer settings correct? :) Mar 12 18:52:58 On my system in Ubuntu, I don't even get a sound card. /proc/asound/cards is empty. Desktop image shows dummy device. Mar 12 18:53:29 If I compile the audio drivers into the kernel, then it works (with some minor tweaking of the mixers). Mar 12 18:54:21 But I am not sure if my beagleXM is faulty or possibly different from production boards. Mar 12 18:55:16 koen: ping Mar 12 18:57:04 kosta, I'm pretty new to all this stuff myself, I would say the spidev approach looks like the simplest to me too Mar 12 18:57:33 koen: ping Mar 12 18:57:35 ? Mar 12 18:57:39 whoops Mar 12 18:57:47 saw prpplague already did. Mar 12 18:58:28 GrueMaster: nope, it was tasselhoff I think and it was like 2 weeks ago Mar 12 19:00:43 Oh, no I missed that bit. But the link you gave me is for kernel-next, not 3.2. And I don't have scrollback going that far back. Mar 12 19:02:45 Is koen's repo still the defacto standard for the beaglebone kernel? Mar 12 19:08:02 alan_o: I'm using beaglebone Mar 12 19:09:37 And this is the thing, I'm trying to avoid kernel patches. First of all, I have no idea how to do that :(, and second, beagle is out there for 3 years now, and still this is problem? I'm sure it could make it's way into main kernel already, why everybody should suffer? Mar 12 19:09:56 ^and go though the same exercise all the the time? Mar 12 19:11:12 jsabeaudry: what do mean by "spidev approach"? Access via file? Mar 12 19:11:41 kosta: you can use file I/O on the spidev device. Mar 12 19:11:55 kosta, Yes, access using a driver named "spidev" that will expose a device file typically in /dev/ Mar 12 19:12:19 Like /dev/spi* ? Yes, that what I meant: it is simple, but too inefficient Mar 12 19:12:21 kosta: http://www.kernel.org/doc/Documentation/spi/spidev_test.c Mar 12 19:12:43 kosta, How is it inefficient? Mar 12 19:13:33 jsabeaudry: well, you write file, then process should read the file, parse it, call driver....blah-blah-blah - too many intermediate steps Mar 12 19:14:26 kosta: it will do just fine, trust me. you only other option is to write a kernel driver Mar 12 19:14:57 So it is either file i/o or kernel driver, right? Mar 12 19:15:13 I was thingking may there is some other way that I just don't know about Mar 12 19:15:22 kosta, Your assumptions are false, communicating with a char device does not involve what you think it does Mar 12 19:16:05 kosta: yes Mar 12 19:16:21 kosta: which device are you talking to? Mar 12 19:16:22 jsabeaudry: I'll take your word for that :) Again, my knowledge of Linux internal is zero, so my assumption might be wrong Mar 12 19:16:35 dwery: bone Mar 12 19:17:11 kosta, http://en.wikipedia.org/wiki/Device_file Mar 12 19:17:21 kosta, Is this part of a school project? Mar 12 19:17:22 kosta: and the other device? Mar 12 19:18:09 kosta: yes, what jsabeaudry said. It's not a file like a file on a filesystem that contains text or something. It's a special file which talks directly to a kernel driver. In this case spidev. Mar 12 19:20:00 I see, so it is more like memory map, right? Mar 12 19:20:02 kosta, try reading the entire contents of the file /dev/urandom, let me know if it fits on your hard drive Mar 12 19:20:16 kosta: no. It's not a memory map Mar 12 19:20:27 Stream? Mar 12 19:20:40 kosta: (however there is a /dev/mem device which _is_ a memory map) Mar 12 19:21:06 kosta: it's an interface Mar 12 19:21:07 jsabeaudry: it fits in my /dev/null Mar 12 19:21:13 Did you read the wikipedia link above? Mar 12 19:21:20 mru, ;) Mar 12 19:21:23 Yes, briefly Mar 12 19:21:35 Looks like a stream Mar 12 19:21:56 alan_o: there are other memory-mapped dev files Mar 12 19:22:14 like /dev/fb Mar 12 19:22:59 kosta: It's basically a direct interface to a driver (in this case a character device). When you call write() on the /dev/spidev* node, it calls the write() funciton in the spidev driver. What that write() (driver) function does is up to the driver, and varies from driver to driver. In the case of spidev, it will clock data out the SPI port. Mar 12 19:23:32 mru: true, some device drivers are memory mapped. It's basically all up to the driver, how it's appropriate for it to be implemented Mar 12 19:23:44 alan_o: thank you! Now that is very clear explanation Mar 12 19:24:18 Actually, this is part-time job project. I'm totaly noob in linux/embedded, coming from many years of strictly Win env :) Mar 12 19:24:36 since spi devices are datagram oriented (each "packet" is delimited by chip select going high to low and back again, for most devices, yes there are exceptions), spidev can be thought of as that kind of device. Mar 12 19:25:26 kosta: in the spidev example program, look at the transfer function. This will do a bi-directional transfer (which is what SPI does all the time). Mar 12 19:26:21 kosta: since read/write aren't appropriate for this, it uses an ioctl(), so that it can pass both an input and output buffer. I think that kind of paradigm will probably be your workhorse from userspace. Mar 12 19:28:28 alan_o: yeap, this is exactly what I was going to do. I was just wondering how efficient it would be. Now with your excellent explanation about how it interfaces with the driver, I don't worry anymore Mar 12 19:29:34 kosta: also, in default configuration, d0 is MISO and d1 is MOSI. That took me a long time to figure out and I still don't quite know why or how. (I should have just tried it). Mar 12 19:30:41 thank you, alan_o! 5 minute talk saved me week of reading docs :) Mar 12 19:31:06 which reminds me that I wanted to ask why and how that works.... Any takers here? d0 and d1 can both be set to MOSI or MISO on the am335x. However, I can find nothing in the kernel which sets these, and according to the TRM, it shows that the power-on-reset values of the register bits that set the direction both default to output, so I can't see how it works at all (but it does). Mar 12 19:32:09 kosta: No problem. There's still docs to read :) but I find that just getting pointed in the right direction is a huge time saver. Mar 12 19:33:59 alan_o: I noticed that the default d0/d1 setting was inverted wrto the bone trm manual, at least on the revision I read at that time Mar 12 19:34:51 alan_o: maybe uboot does it? Mar 12 19:35:13 jay6981: Good point. I didn't look there. Mar 12 19:36:00 dwery: I read someone else on the mailing list use the term inverted wrt the am335x TRM, but the am335x TRM has examples which show both ways (and states clearly that the CPU will do both) Mar 12 19:36:14 dwery: not sure if I have the bone trm Mar 12 19:36:27 I have a BONE_SRM.pdf Mar 12 19:36:41 yes, that one Mar 12 19:37:03 it is configurable but the kernel interface does not yet support it Mar 12 19:39:28 dwery: Mine doesn't have directions in it at all. Just says d0 and d1 Mar 12 19:39:35 The word MOSI doesn't appear Mar 12 19:39:53 uhm.. right Mar 12 19:40:18 let me find my notes.. Mar 12 19:40:24 try simo Mar 12 19:41:55 W1N9Zr0: not that either Mar 12 19:43:05 dwery: can't find them. I think it was inverted wrto the supposed default configuration in some PDF Mar 12 19:43:16 * dwery talks to himself Mar 12 19:58:00 heh Mar 12 20:00:15 alan_o, Could they possibly be controlled by the pin mux? the one set with IEN will be used in input? Mar 12 20:03:56 jsabeaudry: it's controlled by McSPI Mar 12 20:06:18 bbl Mar 12 20:07:34 *yawn* Mar 12 20:07:38 !@$@#@!$!@# timechange Mar 12 20:07:56 <_av500_> ds2: so, you have a 9to5 job or what? Mar 12 20:08:14 * djlewis thinks ds2 makes his own hours Mar 12 20:08:31 <_av500_> seems not Mar 12 20:08:34 I have a 0100-0900 sleep pattern that is getting disrupted :P Mar 12 20:08:50 pesky stores not staying open at the right hours Mar 12 20:09:00 <_av500_> you shop from 0900 to 0100? Mar 12 20:09:21 no Mar 12 20:09:28 anyways, it is a disruptive change Mar 12 20:10:20 * djlewis tries to stick with his own timezone rules (no change). Mar 12 20:10:48 if I could just get the world to cooperate Mar 12 20:10:53 <_av500_> ds2: ipad disruptive or steam engine disruptive? Mar 12 20:11:10 <_av500_> or are we talking wheel dimensions? Mar 12 20:18:31 * alan_o just figured out who ds2 is. Mar 12 20:19:12 wtf? :) Mar 12 20:19:12 ds2: we spoke at ELC after your presentation. Mar 12 20:19:19 oh okay Mar 12 20:19:27 Well, and by figured out I mean, right click and looked at the name :) Mar 12 20:19:35 alan_o: signal11? Mar 12 20:19:38 yes Mar 12 20:19:43 ah, okay =) Mar 12 20:20:13 I don't exists. I am a figment of your imagination Mar 12 20:21:27 (white rabbit) Mar 12 20:22:04 <_av500_> ds2: you dont exist, but in a specific time zone... Mar 12 20:42:03 anyone know what the batterylife on a beaglejuice would be for a fairly inactive beagboard booted up into the OS Mar 12 20:51:13 Hey- what are the different memory maps for the Cortex A8? Mar 12 20:51:23 <_av500_> ? Mar 12 20:51:27 are those different states? Mar 12 20:51:44 <_av500_> ?? Mar 12 20:51:58 i see L3, L4_WKUP, and L4_PER Mar 12 20:52:12 just different power domains Mar 12 20:52:37 0. Mar 12 20:53:34 okay Mar 12 20:54:03 for those regions specified in L3 but not L4_WKUP, does the L3 meaning apply? Mar 12 20:56:05 oh, i see Mar 12 20:57:08 the L4s are subsets of L3 Mar 12 20:59:50 Are there reasons besides power consumption that would explain why someone would not want to have the PWM clocks running? Mar 12 21:00:00 EMI Mar 12 21:00:07 thermal Mar 12 21:00:39 <_av500_> power is reason enough Mar 12 21:00:56 I'd argue EMI is a stronger reason Mar 12 21:01:22 <_av500_> yes, but you like to argue Mar 12 21:01:23 the EMI folks get a bit irritated when you take out radio coverage for miles on end ;) Mar 12 21:01:32 Interesting, what would you think about changing the default? Mar 12 21:01:54 <_av500_> ds2: so that is why all the phones cannot sold with lcd backlight any more.... Mar 12 21:02:02 <_av500_> due to the pwm emi... Mar 12 21:02:04 exactly ;) Mar 12 21:02:15 <_av500_> pixelqi will make a killing Mar 12 21:02:33 <_av500_> and candle stick makers Mar 12 21:02:47 really? how bad is the EMI from the PWM clocks? Mar 12 21:03:28 <_av500_> depedns on the thinkness of the copper tape Mar 12 21:06:18 okay, so memory from 0x00000000 to 0x1FFFFFFF on a BeagleBone is the GPMC- what's the GPMC actually mapped to Mar 12 21:06:27 <_av500_> ? Mar 12 21:06:43 gpmc is another subsystem on the soc Mar 12 21:06:52 right Mar 12 21:07:04 yes, it is Mar 12 21:07:11 on the beaglebone, what does the GPMC use as the storage Mar 12 21:07:13 nothing? Mar 12 21:07:17 <_av500_> whatever you attach Mar 12 21:07:25 <_av500_> ram, rom Mar 12 21:07:30 <_av500_> core memory Mar 12 21:07:34 NAND Mar 12 21:07:35 <_av500_> carrier pidgeons Mar 12 21:07:39 heh Mar 12 21:07:43 NOR Mar 12 21:07:47 <_av500_> XOR Mar 12 21:07:57 but on the stock beaglebone, no cape Mar 12 21:08:06 <_av500_> nothing Mar 12 21:08:09 it's mapped to nothing and shouldn't be used. Mar 12 21:08:14 okay, that's what I thought Mar 12 21:08:17 XMPPwocky, by definition its OUTSIDE. Mar 12 21:08:29 <_av500_> its like an ISA or PCI bus with nothing plugged in Mar 12 21:08:49 waiting for somebody to plug in their GUS Mar 12 21:09:00 but wait, then how the heck do I do interrupts? aren't the vectors supposed to be at 0x00000000? Mar 12 21:09:04 heh Mar 12 21:09:38 or am I supposed to set up virtual memory first Mar 12 21:09:51 <_av500_> you use the linux kernel irq sussystem :) Mar 12 21:09:53 <_av500_> sub* Mar 12 21:10:09 <_av500_> and the IRQ vectors can be moved I think Mar 12 21:10:17 not linux Mar 12 21:10:27 <_av500_> still the irq base can be moved I think Mar 12 21:10:49 there's an a8 register called VBAR you can set Mar 12 21:11:20 <_av500_> CP15 Vector Base Address Register Mar 12 21:11:27 yep Mar 12 21:12:05 that's an actual register, not a memory-mapped region, right? Mar 12 21:12:14 correct Mar 12 21:12:20 hooray Mar 12 21:12:24 i think I've got what I need Mar 12 21:12:26 thanks a lot Mar 12 21:12:47 what are you implementing? Mar 12 21:12:52 you might have noticed that exceptions don't work in uboot Mar 12 21:13:03 jay6981: yeah, I'm using OpenOCD flash_image now Mar 12 21:13:44 ka6sox: writing a toy OS in assembly to teach me about ARMv7 Mar 12 21:13:53 kewl Mar 12 21:13:56 have fun. Mar 12 21:14:33 you too Mar 12 21:14:37 wait. er. << Mar 12 21:17:33 Greetings. Short question - how to avoid ext3 scan on BB without batter for RTC? I think I've seen some info when I was searching on u-boot but I couldn't locate it now. Mar 12 21:21:15 so, you'd just use c0, c1, cFOO anywhere you'd normally use a register? Mar 12 21:21:32 <_av500_> ixtec: turn off the scanin fstab? Mar 12 21:21:37 ixtec: tune2fs -i0 -c0 /dev/foo Mar 12 21:21:40 there's a special instruction to change the CP registers Mar 12 21:21:52 check the ARM ARM Mar 12 21:21:57 mcr and mrc Mar 12 21:22:51 the syntax sucks but the cortex a8 trm is nice enough to give you the exact insns you need Mar 12 21:23:17 how would you have the syntax? Mar 12 21:23:17 http://www.digitaljournal.com/article/321059 Mar 12 21:23:32 If they turned that thing up, it'd be a death ray Mar 12 21:24:45 SilicaGel, it turns out that DEW are very hard to do....and getting sufficent power to "kill" is tough unless you are a squirrel. Mar 12 21:24:56 Oh man, I do hate squirrels Mar 12 21:25:12 I doubt this can track one fast enough. Mar 12 21:25:19 crap Mar 12 21:25:22 * SilicaGel goes back to the drawing board Mar 12 21:25:27 ka6sox: so use higher power Mar 12 21:25:41 bring your own Nuke Plant with you? Mar 12 21:25:43 :D Mar 12 21:26:04 nuclear tranzistor? Mar 12 21:26:15 i was thinking of strapping three or so microwave ovens to my hood, removing the doors, and jamming the switch so they still operate Mar 12 21:26:16 you don't want to be in front of that...because its very uncomfortable. Mar 12 21:26:33 mrc: thanx, I was fixed about mounting the superblock in the future but I see that probably was not an issue. Mar 12 21:26:47 ynezz, yup Mar 12 21:27:02 today's equivelent of a Flux Capacitor. Mar 12 21:27:26 XMPPwocky: that sounds fun. I ported FreeRTOS to the specific NXP ARM7TMDI I was using and it was pretty fun. Mar 12 21:27:39 what, iTimeMachines? Mar 12 21:28:12 ynezz, mostly as a power "source" Mar 12 21:29:36 ka6sox: 40 kJ should be enough to cook a squirrel Mar 12 21:29:54 maybe someone could point me to some solution for ttyS2 vs ttyO2, can I remove "system" ttyS0-3 to allow omap's ttyO0-3 be called ttyS0-3? I've changed this in kernel file but now I can't register ttyS since they are "already there". Mar 12 21:29:57 conservative estimate Mar 12 21:30:17 <_av500_> ixtec: there should be no system ttyS Mar 12 21:30:19 what's adding them, some udev rule? Mar 12 21:30:22 <_av500_> only omap ones Mar 12 21:30:29 <_av500_> or you mean in /dev? Mar 12 21:30:47 <_av500_> any why not just use ttyOx? Mar 12 21:30:48 no they are there on my build as well. ttyS0 through S3. Mar 12 21:30:57 mru, depend on how well done you like your squirrel. Mar 12 21:31:04 dead, no? Mar 12 21:31:09 wasn't that the idea? Mar 12 21:31:09 dead. Mar 12 21:31:15 <_av500_> medium dead? Mar 12 21:31:20 <_av500_> or well dead Mar 12 21:31:23 grateful dead Mar 12 21:31:29 some day I *will* have a rocky and bullwinkle sandwich Mar 12 21:31:31 _av500: there are apps created to run with ttySx with the older kernel. Mar 12 21:31:38 <_av500_> grated dead Mar 12 21:31:43 chain mail! Mar 12 21:31:44 :D Mar 12 21:31:49 _av500: I'd prefer not to force someone to change it. Mar 12 21:31:58 <_av500_> ixtec: s/ttyS/ttyO/ Mar 12 21:32:14 <_av500_> sed, poor mans hexedit Mar 12 21:32:15 hardcoding serial device names is stupid Mar 12 21:32:17 <_av500_> +' Mar 12 21:32:24 no use for that Mar 12 21:32:41 I would not want to be less than 50' from that...but a squirrel @ 75meters would be gone I suspect. Mar 12 21:32:41 to figure this out you have to figure out what device driver adds ttyS to it Mar 12 21:32:45 <_av500_> ln -s ttySx ttyOx Mar 12 21:32:47 then you have to compile it out Mar 12 21:32:52 because that device Mar 12 21:32:57 it comes from udev Mar 12 21:33:01 the *other* thing you could do is Mar 12 21:33:05 mru, only good for console and early work. Mar 12 21:33:06 screw with udev Mar 12 21:33:10 <_av500_> normally it either standard OR omap serial drivers Mar 12 21:33:14 udev is evil Mar 12 21:33:16 ka6sox: a studio photo flash discharges a few kJ in a few milliseconds Mar 12 21:33:16 <_av500_> not both Mar 12 21:33:21 no it's not Mar 12 21:33:32 it's just really really hard to figure out at first :) Mar 12 21:33:33 so power-wise, 10 of those would be sufficient Mar 12 21:33:35 static /dev entries are fine for most uses Mar 12 21:33:40 mru, yes, but you would have to be in "contact" with it to have an issue. Mar 12 21:33:46 it won't be hand-held, but certainly "portable" Mar 12 21:33:50 okay, so the low bits of VBAR are reserved: is it safe to overwrite those, or should I read-update-write? Mar 12 21:33:58 well then edit /dev/rc to rm /dev/ttyS0 ; ln -s /dev/ttyO0 /dev/ttyS0 Mar 12 21:33:59 :-) Mar 12 21:34:01 <_av500_> ds2: even android does dynamics devs these days :) Mar 12 21:34:10 I mean that's as "static" as you're going to get on a system with devtmpfs Mar 12 21:34:13 ka6sox: so we need to devise a better deliver mechanism than light Mar 12 21:34:15 there is no "mknod and it stays" any more. Mar 12 21:34:15 just use the MLRS equivelent of a Taser? Mar 12 21:34:16 _av500_: it has been doing it their way for a long time Mar 12 21:34:37 <_av500_> ds2: pre-HC seemed to use mknod a lot Mar 12 21:34:46 <_av500_> now it seems they do listen to uevents Mar 12 21:35:02 well, lasers would do I suppose Mar 12 21:35:03 _av500_: IIRC, they had it mixed into a .c file Mar 12 21:35:05 Ok, good idea - I'll look for compiling them out. Mar 12 21:35:08 that was in GB Mar 12 21:35:12 <_av500_> ds2: that too Mar 12 21:35:25 doesn't change the fact that it is evil Mar 12 21:35:28 I don't know what you'd compile out though, I'm looking at it now Mar 12 21:35:34 uboot? Mar 12 21:35:34 and the ICS "udev" is f'ed half the time Mar 12 21:35:40 or udev? Mar 12 21:35:41 <_av500_> ds2: of course, but we dont need to debate facts :) Mar 12 21:35:46 CONFIG_SERIAL_8250=y Mar 12 21:35:50 CONFIG_SERIAL_8250_RUNTIME_UARTS=4 Mar 12 21:35:54 <_av500_> why? Mar 12 21:36:01 I think that you want to get rid of 8250 support Mar 12 21:36:02 <_av500_> there is not 8250 in the omap Mar 12 21:36:09 I know there isn't Mar 12 21:36:10 I don't see udev taking care of ttyS in my case. Mar 12 21:36:15 <_av500_> might hook one up on gpmc though Mar 12 21:36:17 So why is there an 8250 driver compiled into my stock angstrom kernel? Mar 12 21:36:21 _av500_: just venting. have to keep a busybox with a mknod to debug with Mar 12 21:36:28 <_av500_> ds2: dont we all Mar 12 21:36:48 <_av500_> adding a 8250 might be usefull as it has that scratch byte to store my password Mar 12 21:36:52 What I just pasted is out of /proc/config.gz on 3.2.0+ Mar 12 21:37:05 the UARTs in the OMAP are fun Mar 12 21:37:14 why do you need to store your password in the uart, can't your emember "hunter2" ? Mar 12 21:37:15 toggle the wrong bit and you get non stop interrupts Mar 12 21:37:32 maybe this would be enough? CONFIG_SERIAL_8250_RUNTIME_UARTS=4 Mar 12 21:37:33 <_av500_> NSI, non-stoppable interrupt Mar 12 21:37:38 changing it to "0" Mar 12 21:37:42 ixtec: it might Mar 12 21:37:46 <_av500_> try Mar 12 21:37:53 <_av500_> it wont blow up Mar 12 21:37:58 XMPPwocky: they're "reserved" because it has to be aligned Mar 12 21:38:14 ixtec: if that gets rid of them, then you could add a rule to /etc/rules.d/something that detects when ttyO* gets added and does like {SYMLINK}+=/dev/ttyS{num} Mar 12 21:38:22 jay6981: aligned to what? Mar 12 21:38:23 that's not even close to being right by the way. It's something *similar* to that. Mar 12 21:39:05 word boundry Mar 12 21:39:33 SilicaGel: I wonder that If I won't have them - omap ones will be properly registered. I'll check it now. Mar 12 21:39:58 jay6981: so it's fine to overwrite? Mar 12 21:40:11 i'd just write zeros Mar 12 21:40:14 the omap ones will be properly registered, but udev has a facility for adding symlinks, you should use that Mar 12 21:40:15 yep Mar 12 21:40:18 thanks again Mar 12 21:40:19 and leave th ereal device where it is Mar 12 21:40:44 I really don't recommend you prevent it from removing the /dev/ttyO* ... and I also don't recommend adding a second device node either. It think using SYMLINK+= is the way to go here. Mar 12 21:40:53 s/removing/adding/ Mar 12 21:41:15 just patch the kernel Mar 12 21:41:21 ugh! no! Mar 12 21:41:22 revert the DMA support for the UART Mar 12 21:41:28 so much easier Mar 12 21:41:35 Patch the kernel to chagne a device name! haha Mar 12 21:42:08 so I should make a section in my linkscript just for the interrupt vectors? Mar 12 21:42:15 I find tha tif one takes the time to learn udev, it really grows on you Mar 12 21:42:32 it's just really dfificult if you don't know how to use udevadm to test Mar 12 21:42:58 <_av500_> XMPPwocky: sounds like a plan Mar 12 21:42:58