**** BEGIN LOGGING AT Sat May 24 02:59:58 2014 May 24 03:00:30 lol May 24 03:00:31 what did you expect? "Don't ask, just do this: ..." May 24 03:00:37 man, is it a full moon or something? May 24 03:00:39 lol the art of google first then ask is lost. May 24 03:00:48 Dan1: no, i'm like this every night May 24 03:00:58 im not going to sit here explaining the different unix apps and libraries to you because you want to install some unofficial image May 24 03:01:12 debian is the latest support image May 24 03:01:27 which image comes with the bbb preinstalled? May 24 03:01:36 debian May 24 03:01:39 debian...... May 24 03:01:41 the last batch anyway May 24 03:01:45 previously it was angstrom May 24 03:01:47 but angstrom is not good May 24 03:01:48 uname -a doens't say debian May 24 03:01:50 so no more angstrom May 24 03:01:59 that's because uname doesn't have anything to do with distribution May 24 03:02:26 it has on several of distributions i've used over the years... May 24 03:02:36 you might have seen distribution name in the kernel build May 24 03:02:36 but May 24 03:02:41 it does not provide any information on distribution May 24 03:02:59 root@beaglebone:~# uname -a May 24 03:02:59 Linux beaglebone 3.8.13-bone50 #1 SMP Tue May 13 13:24:52 UTC 2014 armv7l GNU/Linux May 24 03:02:59 root@beaglebone:~# lsb_release -a May 24 03:02:59 No LSB modules are available. May 24 03:02:59 Distributor ID: Debian May 24 03:03:00 then how do you determine the distrubution? May 24 03:03:00 Description: Debian GNU/Linux 7.5 (wheezy) May 24 03:03:02 Release: 7.5 May 24 03:03:03 Codename: wheezy May 24 03:03:13 ah. May 24 03:03:16 ok. thanks. May 24 03:05:02 Dan1: meteor storm night ;) May 24 03:05:07 too bad cloudy here May 24 03:05:13 or i'd setup my rig outside May 24 03:05:45 Debian GNU/Linux 7 May 24 03:05:45 BeagleBoard.org BeagleBone Debian Image 2014-04-23 May 24 03:05:45 Support/FAQ: http://elinux.org/Beagleboard:BeagleBoneBlack_Debian May 24 03:05:45 stdin: is not a tty May 24 03:05:45 LSB Version: :core-4.1-amd64:core-4.1-noarch:cxx-4.1-amd64:cxx-4.1-noarch:desktop-4.1-amd64:desktop-4.1-noarch:languages-4.1-amd64:languages-4.1-noarch:printing-4.1-amd64:printing-4.1-noarch May 24 03:05:45 Distributor ID: Fedora May 24 03:05:45 Description: Fedora release 20 (Heisenbug) May 24 03:05:46 Release: 20 May 24 03:05:46 Codename: Heisenbug May 24 03:06:00 wups. May 24 03:10:24 what is a good procedure for installing a library? should i cross-compile it on my big linux box, then just scp the .so's? May 24 03:11:09 or go through the pain of generating a rootfs with it and regenerating an image? May 24 03:11:34 that is, cross-compiling into a rootfs? May 24 03:11:55 specifically i'm looking at pocketsphinx May 24 03:12:34 or should i just apt-get it? May 24 03:20:01 ok... May 24 03:20:11 "This aptitude does not have Super Cow Powers." May 24 03:25:11 apt-get install the library May 24 03:25:53 otherwise youll need to read the cross compiler faq May 24 03:53:40 I need a faster memcpy for my application on the BBB, hints, anyone? I need to move 4MB of data in less than 20ms. May 24 04:17:37 A pru can theoretically do 1.9MB/20ms. Get two of them together and you're pretty close... May 24 04:34:22 foreverska: Unfortunately it's the PRUs that are generating data at this fast rate... So that's not an option. May 24 04:35:54 Lol I was kinda joking and my math was off anyhow. I forgot to SBCO after I LBCO May 24 04:36:36 Where are you storing the data? Dram? May 24 04:38:53 yeah, have two rings in which the PRU alternates May 24 04:41:38 I'm guessing local dram 0 and 1 May 24 04:46:02 Have you tried writing up to global directly. It will surely bring that bus to a halt for anything but the pru but you may get away with it. May 24 04:49:30 foreverska: Could you elaborate? May 24 04:51:47 ugh May 24 04:51:48 i hate ssrs May 24 04:51:51 pain in my ass May 24 04:52:52 solid state relays? May 24 04:53:54 Wait it fills 4MB in 20ms? May 24 04:54:40 yep May 24 04:54:53 And you want it to run indefinitely? May 24 04:57:31 not really, it's actually a logic analyzer with the PRUs, but I want to have a configurable amount of memory, say, 256MB May 24 04:59:12 and there's just a 20ms window to copy data before the next buffer comes in May 24 05:00:20 So you've got it copied up into the OS at this point? May 24 05:01:24 yep, there's a 8 MB memmapped region using extram_pool_sz in the UIO kernel driver. I divided into two buffers of 4 MB each, and the PRU sends the app an interrupt after it fills each buffer May 24 05:02:02 so there's the 20ms window between buffer 1 ready and buffer 2 ready May 24 05:02:35 if I can't copy the 4MB fast enough, the mem will get overwritten May 24 05:03:49 Hmmm maybe you could pull some more performance writing a custom assembly version of memcpy May 24 05:05:00 I think so, I'll try that, or might need to move into kernel space if userspace isn't enough. May 24 05:05:48 packaged functions like that always have userfriendly checks that you can skip since you know what you're doing. May 24 05:07:39 k, I don't think I can pass a pointer to a malloc'ed region of mem to the PRU? May 24 05:08:31 foreverska: I'm actually writing directly from the PRUs onto the DDR memory. May 24 05:12:32 maybe... the very end of the L3 memory map is a 1GB block. That isn't RAM is it? May 24 05:13:59 I'm not quite sure about that though May 24 05:14:50 I'd be tempted to pass the pointer down either way and try and SBBO on it... see where it ends up. May 24 05:15:11 PRUs have a lot of amnesty in this system. May 24 05:16:20 yep, that is also possible, but how to get a handle to say, 256 MB of a contiguous chunk? May 24 05:16:52 Malloc a massive array? May 24 05:17:55 I've never tried mallocing that much memory at once. May 24 05:18:12 yep, that's my concern actually. May 24 05:20:08 I assume you could malloc all the way up to the rlimit though May 24 05:20:47 I was able to malloc 8 MB for initial testing where I found that memcpy 4MB takes 281ms :/ May 24 05:24:11 Well you don't technically need all 256MB... You just need a loop big enough that the PRUs don't catch it's tail May 24 05:30:11 I wonder how long it takes to write a byte to a fast SSD or the EMMC May 24 05:30:58 lol oh wait I just asked what's the write speed of an SSD. May 24 05:31:16 10MB/s if you're lucky, NM' May 24 05:44:23 Just unboxed my Beaglebone Black Rev C, and installed the 64 bit drivers on Windows 8, can't see the Beagle drive or open the default Beaglebone web page. My bone seems to boot ok, or at least the heartbeat led flashes after the 10 second boot sequence. Just don't see the USB teather drive. May 24 05:45:20 You'd have to wait another minute or so. May 24 05:45:58 I connected a monitor to the micro HDMI port, now when I boot I see Tux in the top left hand corner, and nothing else. May 24 05:46:16 Yeah tried waiting May 24 05:46:52 I never use HDMI but that sounds like a bad flash May 24 05:48:19 I guess I can flash a mirco SD card with some distro or other and try and boot from that. Before I go down that path, anyone got anything else I could try? May 24 05:49:09 you must have tried rebooting your BBB?\ May 24 05:49:13 Do you have a serial interface? May 24 05:51:24 yeah tried rebooting, and powering down and up using the power button May 24 05:55:49 tried loading the Mac Os drivers as well but no joy with the Mac either. May 24 05:56:49 Sounds like your board man May 24 05:57:19 oh well guess I'll just have to try flashing a distro onto a microSD then trying to boot from that May 24 05:57:50 if that fails, I'll have to try getting a replacement May 24 05:57:56 bummer huh May 24 06:01:11 I guess... I don't use the eMMC anymore personally May 24 06:01:29 same May 24 06:02:02 It's like she said: "It's just not big enough" May 24 06:02:05 i keep a factory image on it May 24 06:02:13 but then i have a 16gb high speed microsd May 24 06:03:16 exact same here UHS class 1 here May 24 09:43:10 hmm looks like the cause of my wifi dropping out might actually have been a wanky cheap chinese wifi module May 24 10:07:52 What binary is the BBB MLO configured to load by default? Can I make it launch a custom binary instead of U-Boot? May 24 10:08:33 check its source? May 24 10:15:16 m'kay :( May 24 10:18:50 varad: while you are at it, please write a proper documentation for MLO so that we can point the next guy to it instead of the source May 24 10:50:59 there is documentation if you look May 24 10:51:20 i ve seen and used it May 24 11:09:35 found it in the code, it is done here: http://code.metager.de/source/xref/denx/u-boot/common/spl/spl.c#223 using http://code.metager.de/source/xref/denx/u-boot/common/spl/spl.c#64. It looks for an image signature and if not found, just assumes it to be u-boot.bin. May 24 11:12:14 so I could replace the default u-boot.bin with something else with the same name and that would work, I guess May 24 15:16:14 anyone know if kernel normally has iommu function enabled on beagle? May 24 15:16:28 or is it needed? May 24 15:21:54 dogrocket .. I think you might have to enable it with 'stock' kernels ..not sure about RCN's .. he usually enables everything lol May 24 15:27:38 yeah... thats what im doing cleaning out things he puts in :) May 24 15:28:46 lol yeah I spoke to him about that .. ppl gonna complain either way he said lol May 24 15:53:28 im getting a kernel build error... and i dont think i touched this May 24 15:54:17 drivers/gpu/drm/i2c/tda998x_audio_drv.o: In function `init_module': May 24 15:54:17 tda998x_audio_drv.c:(.init.text+0x0): multiple definition of `init_module' May 24 15:54:17 drivers/gpu/drm/i2c/tda998x_drv.o:tda998x_drv.c:(.init.text+0x0): first defined here May 24 15:54:42 i have no idea what to do May 24 15:55:16 make clean May 24 15:55:23 you might have residuals somewhere May 24 15:55:30 make distclean if you wanna be totally sure. May 24 15:56:00 ok thx May 24 15:56:10 that would be my #1 May 24 15:57:11 make clean in KERNEL? May 24 15:59:30 same place you do ya main 'make' yeah May 24 15:59:43 it should clean out any old '.o' object files May 24 16:06:24 Abhishek_: what are you going to do with the data once it gets memcopied? May 24 16:06:34 and did you ever solve your 4MB/20 ms problem? May 24 16:06:39 just curious May 24 16:07:25 yates: It's a GSoC project with BeagleBoard this year, I'm building a logic analyzer. May 24 16:08:09 are you writing the pru code in assembly? May 24 16:08:45 yep May 24 16:09:14 would it be possible for me to have a look at your code/project? May 24 16:09:48 i'd really liek to see how you're doing this. May 24 16:09:54 http://elinux.org/BeagleBoard/GSoC/2014_Projects May 24 16:10:13 thanks! May 24 16:12:30 yw May 24 16:13:16 Abhishek_: can't you set up a DMA to transfer this data fast? May 24 16:14:00 how many bits/hz you hopin for? May 24 16:14:11 could be a very popular project May 24 16:14:34 vermit: That's all in the link I gave, I made a video presentation, May 24 16:14:43 ok will take a look ty May 24 16:15:07 veremit: If I can copy fast enough then 100MHz May 24 16:17:42 Are the libraries bundled in Debian 7 compiled with VFP hard-float or NEON hard-float? May 24 16:18:20 (Debian 7 bundled with BeagleBones now) May 24 16:21:42 Abhishek_: DMA? May 24 16:22:48 The issue isn't in PRU-DRAM transfers (which work fine without DMA as well), it's only for DRAM-DRAM copy as there are two 4 MB buffers from the PRU, and data has to be removed within 20ms May 24 16:24:25 I'm considering alternative approaches, though I'd stick with this for the time being May 24 16:25:02 Abhishek_: yes, and can't you set up a DMA from DRAM-DRAM? wouldn't that be faster than spinning processor cycles copying? May 24 16:26:22 It should be but even better, there should be a way for doing away with memcpy altogether. May 24 16:32:03 o,O May 24 16:32:08 hi SoCo :p May 24 16:52:29 Abhishek_ use elfread to find out out yourself May 24 16:52:37 or readelf May 24 16:52:50 okay readelf May 24 17:00:56 Thanks woglinde! May 24 17:03:51 anyone know if its possible to set up i2s/i2c audio without involving the i2s that comes from the hdmi chip? May 24 17:04:14 i would like to disable the hdmi chip altogether for testing May 24 17:04:48 woglinde: If I have an assembly code using NEON (an optimized memcpy implementation), can that be linked with VFP code? May 24 17:04:48 cape_mgr.disable=BB-HDMI etc as per all the public posts on the internet :P~ May 24 17:05:22 how about if i just disable hdmi in kernel build? May 24 17:05:48 you what? May 24 17:09:30 my kernel build is having errors in hdmi build, i need to test an i2s codec, might as well just disable it May 24 17:09:51 err hdmi link section May 24 17:10:59 i2s go out mcasp pins regardless of hdmi enable? May 24 17:12:40 indeed May 24 17:16:03 there's some trick to using the mcasp because the hdmi framer May 24 17:16:12 its beendone .. google is your friend :) May 24 17:18:23 ok, will google May 24 17:18:52 seen something in my crazy wild travels across the net .. but wouldn't be able to put my finger on it May 24 17:19:05 your kernel compile troubles are a different matter to getting I2S up, though, i would suggest ;) May 24 17:30:56 How am I supposed to specify “gpmc,sync-read” in the device tree? I tried “gpmc,sync-read;” and “gpmc,sync-read = <1>;” but in both cases the bit is not set on GPMC_CONFIG1 May 24 18:07:52 When can an NEON SIMD-based asm memcpy cause a segfault? May 24 18:36:24 when your code has a bug? May 24 19:13:05 When the manual says "Do not drive any gpio pins until SYSRESET goes high", does that mean "do not drive any pins to 3.3V"? Can I drive them low at power-on? May 24 19:13:40 gtp123 .. you don't want any current whatsoever goin to the beagle at power-on May 24 19:13:53 or it will get fried. May 24 19:14:25 if that means low or high .. just be careful :) May 24 19:15:24 Ok. I have my four inputs AND'd with SYSRESET at the moment. Don't know if I should replace the AND gate with a three-state buffer. May 24 19:15:44 personally, would err on the side of cautino and go tri-state May 24 19:16:17 use an output from the board (even +5v) to run enable pin May 24 19:17:04 K, will do. Thanks veremit. May 24 19:17:28 a few ppl have killed beagles with erroneous power on pins .. self included lol May 24 19:18:59 I've already killed inputs on one, and the entire P9 header on another. My boss has been patient up to now, but I don't want to take anymore chances :P May 24 19:21:17 lol sounds strangely familiar .. lol May 24 19:27:22 yeah dont provide voltage to any pins until the board is live May 24 19:41:29 the kernel instructions mention using 3.13 or 3.8 kernels whats difference? May 24 20:24:27 dogrocket: depends on what you do, but one of the big diff is that dt overlays are not ported to 3.13 May 24 20:25:16 what the who May 24 20:25:26 yea no capemgr in 3.9+ May 24 20:58:40 Hey there, I'm purchased my first BeagleBone a few days ago. Somehow I'm not able to connect via browser. Can someone help? May 24 21:00:06 gmek: which OS is your PC running? May 24 21:00:29 Arch Linux May 24 21:01:06 did your arch linux find the network interface and assign an IP? May 24 21:01:47 Followed the "Getting started" guide, I have no ethernet cable, so I tried using the usb cable May 24 21:04:10 yes, usb gives you usb-ethernet May 24 21:05:53 how do i check if this usb-ethernet works, never used this before May 24 21:06:19 google 'usb gadget arch' because you'll need the kernel drivers May 24 21:06:59 I'd expect it to have cdc-ether built May 24 21:07:24 ifconfig -a .. see if you get a usb0 ethernet up May 24 21:07:26 check if that gets loaded and if you have an additional interface and if it has an ip May 24 21:10:01 could be one of a number of gadget modules .. had to do something weird last time I tried .. it was under one of the composite devices May 24 21:10:10 because it also acts as a storage device May 24 21:10:15 and a serial device too May 24 21:10:41 so I've done this before May 24 21:10:45 cdc_ether works May 24 21:10:56 but you should use one of the composite ones instead May 24 21:10:59 I forget the name of it May 24 21:11:10 that way you get all of the features instead of just ethernet May 24 21:11:17 jpfau .. sounds right May 24 21:11:24 ifconfig doesn't exist in newer arch builds, ip link instead does not give me an additional interface May 24 21:11:45 then you can try lsmod .. but I'm guessing its not there May 24 21:13:52 no its not May 24 21:14:05 or wait, maybe it was you needed a composite driver on the BBB but you can load the the modules individually on the host May 24 21:14:14 regardless, you can start by doing modprobe cdc_ether May 24 21:15:16 how long should it take to write a 1GB image onto a microsd? May 24 21:15:42 certainly less than 5 minutes, no? May 24 21:15:54 giev it 10 :D May 24 21:16:03 yates: what speed class is your SD card? May 24 21:16:15 Class 4 means it caps out at 4MB/s, in theory May 24 21:16:21 jpfau: i don't know May 24 21:16:48 but yeah writing to SD cards is usually pretty slow May 24 21:16:59 5 minutes for 1GB actually sounds about right May 24 21:17:17 depending on the SD card May 24 21:17:48 ok, i'll wait... May 24 21:17:56 * yates digs the other microsd out of the trash May 24 21:18:01 Ok guys, I'm not I didn't get all that, will spend some time googeling and return if I fail again May 24 21:18:10 thank you very much anyway May 24 21:18:13 .ww May 24 21:18:15 whoops May 24 21:18:37 gmek: googling. sheesh - learn to spell, ok? May 24 21:18:48 don't expect anything to be quick with the beagle stuff :p~ May 24 21:19:13 this ain't octo-core 24ghz computing here ... May 24 21:19:18 :) May 24 21:19:24 veremit: well, this is a Kodak micro sd reader on a 3.9 GHz i7-4820.... May 24 21:19:26 no native speaker here, I'm happy if you get me May 24 21:19:36 we're back to amiga standards lol May 24 21:19:39 gmek: lol - just joking. go in peace. May 24 21:19:44 think loading your games from tape ;) May 24 21:20:02 veremit: hey, i used the Radio Shack TRS-80 back in the day. May 24 21:20:03 yates: neither the reader nor the computer's specs are going to be the limiting factor here May 24 21:20:09 it's gonna be the card itself that's slow May 24 21:20:30 yeah, figured. May 24 21:20:37 the advantage of the 21st centruy .. its its a miniature .. lol May 24 21:22:57 i'm confused about the internal vs external MMC on the bbb. May 24 21:23:27 does booting off the external drive automatically copy the image onto the internal? May 24 21:24:09 http://beagleboard.org/latest-images/ May 24 21:24:37 or is there something in the eMMC image there that does this when you boot from external? May 24 21:24:43 there ^^^^^ May 24 21:25:51 does booting off the external drive automatically copy the image onto the internal? May 24 21:25:54 if you use the flasher image May 24 21:25:56 it will flash the emmc May 24 21:25:57 from sd May 24 21:26:07 emmc is the flash storage on the bb May 24 21:26:44 what if i don't want to flash the emmc? what image should i use? May 24 21:27:08 the first one? May 24 21:27:20 "BeagleBone" May 24 21:27:27 i guess, yeah, that's pretty obvious. May 24 21:28:06 ok, but after flashing the eMMC image onto my internal drive, i'm having a problem where it boots fine, i can ssh in, but then it shuts itself down after 5 or 10 minuts.. May 24 21:28:40 after rebooting off the internal eMMC, taht is. May 24 21:29:37 anyone seen this? May 24 21:30:26 ? May 24 21:31:34 no answers would be appreciated. May 24 21:31:42 "no" answers, i.e. May 24 21:31:49 as well as yes. May 24 21:32:09 the regular sd image, not the flasher May 24 21:32:22 one says flasher May 24 21:32:25 one says May 24 21:32:28 (Runs on BeagleBone Black as well without flashing the eMMC) May 24 21:32:34 i think its pretty self explanatory May 24 21:32:44 http://debian.beagleboard.org/images/bone-debian-7.5-2014-05-14-2gb.img.xz May 24 21:32:48 install that onto sd May 24 21:32:51 it will not flash emmc May 24 21:32:54 and will run entirely off sd May 24 21:32:59 right. May 24 21:33:31 but when i used the flasher (eMMC) image, after flashing onto the internal, then booting from the internal, it shuts itself down after 5-10 minues May 24 21:34:47 sounds like you have another problem May 24 21:35:09 you removed the sd card when booting from internal, yes? May 24 21:35:53 no May 24 21:36:16 i thought if you apply power without holding the3 switch it will just boot from internal and ignore the external May 24 21:43:01 is that not true? May 24 21:43:48 yates: it depends on uenv.txt actaully May 24 21:44:52 and the bootloader :) May 24 21:45:09 yes. and the default behaviour has changed over time.. it’s a really confusing thing to google May 24 22:01:50 its shutting down May 24 22:01:53 because you keep booting the flasher May 24 22:01:55 and its falshing every time May 24 22:01:57 then turning off May 24 22:02:00 take the sd card out after you flash it May 24 22:02:00 lol May 24 22:10:28 lmao .. plausible :D May 24 22:10:44 thats classic :D May 24 22:16:32 hmm seems my wifi issues were sorted by chanfing the firmware and turning of power managment May 24 22:21:05 well I'm not surprised about power management rofl .. May 24 22:21:09 firmware .. well .. May 24 22:21:19 always something to check May 24 22:24:38 its been stable for most of the day now but apparently these Relink chips are the worst to get running May 24 22:26:03 if you got that damned .. rt8192u yes .. May 24 22:26:28 and everything these days has 'em May 24 22:26:42 so you need like a 802.11g card from the 2000s lol May 24 22:26:52 will work flawlessly I bet May 24 22:27:29 i think i just spoke to soon :( May 24 22:29:36 Bus 001 Device 002: ID 148f:5370 Ralink Technology, Corp. RT5370 Wireless Adapter May 24 22:29:38 good shit May 24 22:29:41 nie and stable May 24 22:29:44 $10 May 24 22:29:59 http://www.amazon.com/Panda-Ultra-Wireless-Adapter-150Mbps/dp/B00762YNMG May 24 22:30:06 thats the chip i am using on arch May 24 22:30:17 works good in deb May 24 22:30:20 RT5370 May 24 22:31:05 still think a wire into that beautiful RJ45 socket with 10/100 capability straight into the am3359 is the shit though :D May 24 22:31:26 never dropped a bit yet .. well ok maybe one or two :D May 24 22:31:37 the 100mb cost me <$20 May 24 22:31:42 switch++ May 24 22:31:55 and with my AP I got a LAN :D May 24 22:32:32 yeah i might yet resort to that as this is for a home automation system so the box that will house all the parts will be in a cupboard and i was trying to avoid routing an cat5 cable to it but in reality there is no reason not to , pluss i will get better throughput on a wired connection May 24 22:32:45 absolutely May 24 22:40:07 the trendnet gigabit switches are cheap and god May 24 22:40:55 went for d-links .. 100mb for here, 1gb for work .. don't need the speed saved a few $ for here :) May 24 22:44:34 think they were slightly discounted at the time May 24 22:47:02 all my beagle needs to do is be net accessable and comunicate with a few arduinos and a gsm module for wich im using the uarts anyway May 24 22:47:29 yup ideal May 24 22:48:25 once i get some level converters as i know i can do uart level shifting using resistors but i dont like that idea May 24 22:49:41 beagle output at 3.3 is not an issue as both the arduinos and the gsm module will accept that but the 5v out from them the beagle wont like May 24 22:59:02 ah yea .. some 74-series will be fine May 24 23:01:42 oh and i need to get a bidirectional one as i forgot to say im adding a DS1307 rtc with bat backup to the I2C and again its 3.3 verses 5 v May 24 23:02:15 it won't run at 3.3? May 24 23:02:37 its a premade module built for 5v May 24 23:02:38 bi-dir ones much more expensive .. unless you can switch it with gpio May 24 23:03:35 hmmm im not sure if the DS1307 will work on 3.3 i would have to check the datasheet but if it can i could mod the module to use 3.3 May 24 23:03:41 you might find you got some soldering ;) May 24 23:04:01 I got a ds1307 or -5, definitely some -2's May 24 23:04:11 veremit, the trendnet gigabits are like $20 =] May 24 23:04:17 i dont mind soldering , i come from the arduino world lol so im used to it May 24 23:04:18 been running em throughout the house, run great May 24 23:04:31 nyt .. cool .. wonder if they exist over here .. May 24 23:04:41 wheres there? May 24 23:04:47 used netgear ones at work for ages until we discovered they dropped 1/2 the packets May 24 23:05:02 so I bought a rackmount switch .. no problems there now :D May 24 23:05:03 uk May 24 23:05:06 ah yeah im sure May 24 23:05:12 amazon should have May 24 23:05:20 some trend stuff comes over May 24 23:05:23 depends on vendors May 24 23:05:24 anyways my ft gig is neteng May 24 23:05:40 and these havent pissed me off yet May 24 23:05:42 which speaks volumes lol May 24 23:05:57 and im used to working w/ gear that costs more than my house May 24 23:06:04 lol May 24 23:06:29 theyre dumb switches, but they work well and throughput is delivered May 24 23:06:31 don't call that cardboard box ya house mate ... :P~ May 24 23:07:07 no catchy-firey either! :D May 24 23:07:10 lol May 24 23:08:29 S1: BR-MLX-100Gx2-X 2-port 100GbE Module CARD_STATE_UP 2 0024.3884.7600 May 24 23:08:29 S2: Blocked for full height card in Slot 1 May 24 23:08:29 S3: NI-XMR-1Gx20-SFP 20-port 1GbE/100FX Module CARD_STATE_UP 20 0024.3884.7660 May 24 23:08:29 S4: BR-MLX-10Gx4-X 4-port 10GbE Module CARD_STATE_UP 4 0024.3884.7690 May 24 23:08:30 s5fs: BR-MLX-100Gx2-X 2-port 100GbE Module CARD_STATE_UP 2 0024.3884.76c0 May 24 23:08:31 S6: Blocked for full height card in Slot 5 May 24 23:08:32 S7: BR-MLX-100Gx2-X 2-port 100GbE Module CARD_STATE_UP 2 0024.3884.7720 May 24 23:08:33 S8: Blocked for full height card in Slot 7 May 24 23:08:34 S9: BR-MLX-10Gx8-X 8-port 10GbE (X) Module CARD_STATE_UP 8 0024.3884.7780 May 24 23:08:35 S10: BR-MLX-10Gx8-X 8-port 10GbE (X) Module CARD_STATE_UP 8 0024.3884.77b0 May 24 23:08:36 S11: NI-XMR-10Gx4 4-port 10GbE Module CARD_STATE_UP 4 0024.3884.77e0 May 24 23:08:37 S12: BR-MLX-10Gx8-X 8-port 10GbE (X) Module CARD_STATE_UP 8 0024.3884.7810 May 24 23:08:38 S13: BR-MLX-10Gx8-X 8-port 10GbE (X) Module CARD_STATE_UP 8 0024.3884.7840 May 24 23:08:39 S14: BR-MLX-10Gx8-X 8-port 10GbE (X) Module CARD_STATE_UP 8 0024.3884.7870 May 24 23:08:44 good times May 24 23:08:58 https://www.cdw.com/shop/products/BROCADE-MLXE-2PORT-100GBE-X-CFP-MOD/2499903.aspx May 24 23:09:00 those 100g cards May 24 23:09:01 without any optics May 24 23:09:02 lol May 24 23:10:14 and thats just one out of ~100 May 24 23:10:46 i want to punch brocade in the throat lately May 24 23:10:47 * nyt sighs May 24 23:11:01 fucking horrid codebase May 24 23:16:29 aww spam spam spam :D May 24 23:27:00 I've lent my bbb out so I can't check myself, but it may be worth examining what the values of the TLB lockdown registers are on the am335x... on the dm814x ROM locks 2 out of 32 dTLB entries and ditto with iTLB May 24 23:27:14 that's definitely going to have a performance impact on Linux (or any other HLOS) May 24 23:27:27 (and unlocking those entries is a simple matter of writing zero to those registers) May 24 23:34:53 nyt: cracked me up that i was autocompleted haha May 24 23:35:02 haha yeah sorr y;) May 24 23:44:02 wtf May 24 23:56:44 ? May 24 23:58:43 some register crap May 25 00:00:00 it's pretty crap that the ROM leaves TLB entries locked, yes May 25 00:00:18 (then again, from what I've seen the ROM *is* crap, so it fits) May 25 00:00:40 lol May 25 00:04:13 so submit a patch May 25 00:04:56 makes no sense unless it's confirmed that ROM on the am335x also does this... it's plausible given their close relationship, but should be checked first May 25 00:05:06 so submit a patch May 25 00:05:39 a patch for/to ? May 25 00:06:00 https://github.com/beagleboard/ May 25 00:07:09 a patch for/to ? May 25 00:07:12 :P May 25 00:10:32 like I said, I lent my bbb out so I can't check myself right now, I will when I get it back May 25 00:11:06 but I thought I'd mention it in case anyone cares... dunno, maybe some gamers trying to squeeze the last bit of performance out of their bbb ;-) May 25 00:11:55 (the instructions for reading/writing those registers are on page 177 of the cortex-A8 TRM) May 25 00:14:39 gamers on bbb!? May 25 00:14:45 hehe May 25 00:14:45 really does take all sorts lol May 25 00:25:46 hmm, it should actually even be testable in userspace May 25 00:27:27 http://pastebin.com/dZcxe1eV May 25 00:28:30 though it's possible that the number of available TLB entries isn't the only memory system limit that might show up as a speed-bump in this test May 25 00:30:44 but I think it is... it'll blow right through L1 cache but that's write-through anyway, while it shouldn't get near any L2 cache limits May 25 00:31:47 (this assumes linux isn't sneakily using 64 KB pages instead of 4 KB... I'm not that familiar with how linux does things on ARM) May 25 00:35:58 hitting the TLB limit should be pretty noticable: a memory store normally takes one or two cycles, while a TLB miss adds 24 cycles to that May 25 00:37:10 anyhow, nobody cares :P will shut up now May 25 00:38:05 lol May 25 00:47:20 its just dead in here on saturday May 25 00:48:27 it's even worse on the E2E forums... do people actually have lives or something? ;) May 25 00:50:28 wtf?! lol May 25 00:50:57 im stuck writing code so trying not to become too distracted =[ May 25 00:53:58 hmm, if you succeed in not becoming too distracted, could you tell me how... I could use some of that May 25 00:54:52 yeah my boss would pay $$ if I was less distracted May 25 00:54:53 :) May 25 00:58:00 wow, there are apparently a whole bunch of people who actually read public pastes on pastebin May 25 00:58:08 (> 150 views) May 25 00:59:42 * zmatt just tried it just out of curiosity May 25 00:59:56 and... like.. WHAT... http://pastebin.com/CvsNiyfD May 25 01:00:53 oh right, the not-getting-distracted-thing May 25 01:01:41 hahaha May 25 01:49:11 i'm booting from an 8GB microSD but rootfs only shows about 1.6GB May 25 01:49:23 on a bbb May 25 01:49:41 does it somehow still use the internal eMMC for rootfs? May 25 01:50:10 badly partitionned? May 25 01:50:11 yates: how is your sdcard partitioned? May 25 01:50:19 fdisk -l /dev/of/my/sdcard May 25 01:51:23 look at /opt/tools/scripts for a resize sript May 25 01:51:43 ok, checking - had to reboot May 25 01:52:57 Disk /dev/mmcblk0p2 doesn't contain a valid partition table May 25 01:53:34 mmcblk0 May 25 01:54:03 ah May 25 01:54:30 is there a paste command under debian image? like fedora has fpaste? May 25 01:55:02 http://www.fpaste.org/104700/09828961/ May 25 01:55:29 i created the image the "easy way" according to here: http://downloads.angstrom-distribution.org/demo/beaglebone/ May 25 01:55:36 (e.g., just copied it) May 25 01:56:12 so i can run /opt/tools/scripts/resize-blah to resize it in a running system? May 25 01:57:00 there is no tools folder under /opt/ May 25 01:57:34 ah, you mean /opt/scripts/tools May 25 01:58:22 yates May 25 01:58:30 theres a script to resize May 25 01:58:31 in /opt May 25 01:58:36 or you can just fdisk May 25 01:58:38 delete the partition May 25 01:58:40 recreate full size May 25 01:58:42 then run resize2fs May 25 01:59:04 or /opt/scripts/tools/grow_partition.sh May 25 01:59:05 but May 25 01:59:09 youre running old ansgrom shit May 25 01:59:15 so who knows if thats even there May 25 01:59:29 i'm pretty sure i grabbed the debian. May 25 01:59:33 oh you said angstrom May 25 01:59:36 so May 25 01:59:37 assumed May 25 01:59:37 anyway May 25 01:59:40 the script is there May 25 01:59:43 yes. May 25 01:59:46 i've found it May 25 01:59:46 or just do the fdisk / resize2fs May 25 02:00:38 Why are we not just plugging the card back into the main computer and resizing it there? May 25 02:01:13 with gparted?! May 25 02:01:48 I like gparted. Not everything you do is supposed to make you feel like a hacker. May 25 02:02:02 love it May 25 02:02:08 sysrescuecd very useful too May 25 02:02:13 nyt: delete the partition? isn't that going to delete my image? May 25 02:02:21 just run the script May 25 02:02:50 the method i told you deletes the partition, then you create another one starting at the same point, and then reboot May 25 02:02:57 so the partition extends the entire device May 25 02:03:24 ah. May 25 02:03:40 then you can run resize2fs May 25 02:03:45 to adjust the fs May 25 02:03:48 the script does basically the same May 25 02:04:00 but gives you less chance to mess it up May 25 02:04:07 ok, now we're good. thanks. May 25 02:04:18 22%, versus 100%... May 25 02:05:21 is there a way to delete a botched aptitude install? it prematurely stopped since i overran my disk May 25 02:05:31 apt-get -f install May 25 02:05:33 should clean it up May 25 02:05:36 or dpkg --configure -a May 25 02:05:42 or apt-get install --reinstall whatever May 25 02:06:42 good. May 25 02:06:56 this is almost better than crack! May 25 02:08:31 You've got the wrong kind of crack May 25 02:09:35 foreverska: LOL. (not sure what you mean, but i'm afraid to ask...) May 25 02:46:02 I have a black and white LCD screen with 16 pins. Does anyone have a link to a good tutorial for hooking it up to a Beagle Bone Black? I'm interested in having it show the terminal screen May 25 02:49:11 datasheet for screen? May 25 02:49:44 hang on May 25 02:52:48 here's a link to a PDF: http://unitech-lcd-modules.com/pdf/Uc-161a.pdf **** ENDING LOGGING AT Sun May 25 02:59:58 2014