**** BEGIN LOGGING AT Wed Apr 10 02:59:58 2013 Apr 10 03:01:07 ds2: emulate that tool on the phone and the board requests the things needed ? Apr 10 03:02:50 ungureanuvlad: it starts there Apr 10 03:02:57 it does this - Apr 10 03:03:13 if the processor's sys_boot pins are configured right and it is a cold start, it will make a request over one of the USB ports Apr 10 03:03:27 that tool response to it and downloads MLO into the onboard SRAM Apr 10 03:03:57 MLO (aka Xloader) can then request U-boot which cna then request the kernel Apr 10 03:04:11 from that point you can either setup an NFS server and do NFS root or force a ramdisk or... Apr 10 03:05:12 ds2: ok makes more sense now Apr 10 03:05:47 ds2: there seems to be a few cnc's on ebay with 20krpm spindles Apr 10 03:07:38 prpplague: if their runout spec is decent, it might do okay Apr 10 03:07:56 supposely you can also do FPC's with a fast enough spindle Apr 10 03:07:58 1 Apr 10 03:08:20 ungureanuvlad: did you read the boot section of the TRM? think it is the last chapter Apr 10 03:11:32 ds2: http://www.ebay.com/itm/New-3-Axis-H80mm-3040-CNC-Router-Engraver-PCBs-Routing-Drilling-Machine-/290874709579?pt=LH_DefaultDomain_0&hash=item43b97bd64b Apr 10 03:11:43 ds2: no, i am searching now for it. Apr 10 03:16:00 intersting... no run out specs Apr 10 03:16:27 of course this one asks for more $$ Apr 10 03:19:02 ds2: ok so the sys_boot pins define the order of the boot. and sys_boot5 needs to be 1 for peripheral booting Apr 10 03:19:36 ungureanuvlad: continue reading... Apr 10 03:19:47 ds2: yea not that much more Apr 10 03:20:25 ds2: says it uses g-code, but i gather it would take me a few weeks to figure out how to get it talking/working Apr 10 03:20:25 could go either way.. fun or frustrating Apr 10 03:20:38 prpplague: g-code is good... eagle can generate that Apr 10 03:20:52 but if you want to bypass it... there is a decent chance it will have a step/dir interface Apr 10 03:20:58 that is easy to drive Apr 10 03:21:36 ds2: yea, just need to do some research Apr 10 03:21:42 ds2: thanks for the feedback Apr 10 03:25:01 ds2: okok so rom code tells to download stuff or not Apr 10 03:26:19 ds2: romcode needs to find usb as 1st bootable interface and the pins need to be set as having usb first, to have a match Apr 10 03:27:17 ungureanuvlad: yep Apr 10 03:27:33 G code is easy Apr 10 03:27:46 G00 X Y Apr 10 03:27:57 if you care about speed, G01 X Y F Apr 10 03:28:10 there is one for arcs and that's most of it Apr 10 03:29:47 ds2: what do they mean by "wait for WD reset" ? especially wd ? Apr 10 03:31:43 WD == watchdog Apr 10 03:31:51 forgot if that is enabled by default.. it might be Apr 10 03:33:50 ds2: ok so on the phone the ASISC ID interpreter needs to be done Apr 10 03:34:04 pusb does that Apr 10 03:34:09 so you can leverage that code Apr 10 03:34:47 and the 1st stuff that is sent is MLO Apr 10 03:34:57 greats. Apr 10 03:35:11 crap adafruit's protocape fails the DRC Apr 10 03:35:16 or any code you want... just needs to be hashed like MLO Apr 10 03:35:23 how i can install the driver of "EasyCap" device. Apr 10 03:35:57 maybe in the olden days, there were no DRC ;) Apr 10 03:36:16 ds2: hashed, meaning to have the right stuff in the header right ? Apr 10 03:36:45 ungureanuvlad: yeah... it is basically a degenerate signing... look for the signGP tool Apr 10 03:36:59 under certain conditions (not relevant to us), it is real signing Apr 10 03:37:22 ds2: i read smth about it, it makes some operations on the first 4 bytes...did not read it fully Apr 10 03:37:26 somebody for help? Apr 10 03:40:16 ds2: yep...but if i can send any code then can i just send uboot and then request the kernel? Apr 10 03:40:35 ungureanuvlad: if you can strip u-boot down to size, sure Apr 10 03:40:44 at that point, DRAM is not available Apr 10 03:42:05 ds2: yes rom code does not know nothing about external memory right ? Apr 10 03:42:09 one of the jobs of MLO is to setup the RAM timing Apr 10 03:42:12 correct Apr 10 03:42:26 ds2: yep Apr 10 03:42:29 there is suppose to be a way to give a table of timings to the ROM in one of the headers Apr 10 03:42:43 for that you need to search the ML... my memory on that is hazy Apr 10 03:43:50 ds2: so if i can strip u-boot to fit the internal memory and give the table of timings then i can use the dram to do the rest of the stuff Apr 10 03:44:03 yes Apr 10 03:44:09 there is a project to try to do that Apr 10 03:44:21 but I don't know the URL off hand Apr 10 03:44:23 ds2: i need to say this project is really nice Apr 10 03:45:33 if you get it working, it isn't too hard to do it for the classic and the xM boards Apr 10 03:46:04 in the specs of the project is says that it needs to run on the bone 1st i thinl Apr 10 03:46:13 think* Apr 10 03:46:15 yes Apr 10 03:46:39 now i am trying to get an overview of the timeline Apr 10 03:46:45 getting it for another board is largely supporting another asic ID + a new MLO Apr 10 03:46:45 how to put each part and where Apr 10 03:47:30 value here is you can reuse most of it for board bring ups Apr 10 03:47:51 unbricking boards ? Apr 10 03:48:07 do you know how to wrestle control of the UDC from android? Apr 10 03:48:23 if this works, the board is not bricked... so strictly speaking, that is a misnomer Apr 10 03:48:25 but sure Apr 10 03:50:05 n/m Apr 10 03:50:10 host controller I mean Apr 10 03:50:13 mhmh not kinda, i don't have xp with android that much. i have a nexus 7 and a samsung galaxy but more than installing some custom roms on them and tweaking not so much. i did some apps in java for my phone but i don't think it is related Apr 10 03:51:36 s/i don't think/surely is not Apr 10 03:52:11 that is the other half of getting to work on android Apr 10 03:53:54 so...to access the usb port on the phone is a requirement to be rooted ? Apr 10 03:54:27 usually depending on your definition of rooted of course ;) Apr 10 03:55:11 ok so on the phone can you give me as an overview what needs to be done ? Apr 10 03:55:39 setup as a driver to handle the chip and response to it Apr 10 03:55:48 to do that you need root (typicallY) Apr 10 03:55:57 but android likes to get in the way Apr 10 03:56:23 so mainly you take over the usb port Apr 10 03:56:30 with all its functions Apr 10 03:56:34 no Apr 10 03:56:41 just need to be a driver Apr 10 03:57:02 gadget, yes... but I mixed that up with something else.. hence the 'n/m' Apr 10 03:57:06 now i'm feeling bad i did not take the operating systems lab and left it for next year Apr 10 03:57:10 you just need to be a usb driver Apr 10 03:57:37 bah.. it isn't that big of a deal Apr 10 03:57:44 I never took any of that "STUFF" Apr 10 03:58:22 so can this be done using the android api ? Apr 10 03:58:28 nope Apr 10 03:58:30 Linux api Apr 10 03:58:44 unless that libusb referred to in the description provides an android api Apr 10 03:58:57 <--- not a userland person Apr 10 03:59:08 and android is nothing but userland crap Apr 10 03:59:30 ok Apr 10 03:59:43 so on the phone be driver listen for codes and supply the files requested Apr 10 04:06:35 ds2: do you think there can be incompatibility between many android versions ? Apr 10 04:06:53 ungureanuvlad: it comes down to the kernel used Apr 10 04:07:16 * mranostay pokes ka6sox Apr 10 04:07:41 mranostay: back from PDX? Apr 10 04:07:51 yeah have been Apr 10 04:08:22 should JTAG be buffer from both sides? Apr 10 04:08:25 *buffed Apr 10 04:08:30 *buffered grrr Apr 10 04:08:56 ds2: so after all booting stuff is done we will run linux from dram right, or i am completely wrong? Apr 10 04:08:59 buffered? Apr 10 04:09:12 ungureanuvlad: you are correct Apr 10 04:09:42 and if a mmc is present our linux can access it and read/write from it, right ? Apr 10 04:09:51 yes Apr 10 04:11:08 ok i have another question, supposedly uBoot is now on the board got the zImage and now needs to get the filesystem, how exactly is that done? do u have a link where i can read ? Apr 10 04:11:33 donno of any links Apr 10 04:11:37 other then source Apr 10 04:13:57 ds2: U-Boot> fatload usb 0:1 0xC0700000 uImage like this the request is Apr 10 04:15:50 what board is that? Apr 10 04:16:06 haven't seen a U-boot that supports that on the TI ARM's I looked at Apr 10 04:16:09 ds2: yes buffered Apr 10 04:16:29 ds2: it is just generic uboot command http://processors.wiki.ti.com/index.php/Booting_Linux_kernel_using_U-Boot#USB_Storage Apr 10 04:16:36 mranostay: not sure what you mean by 2 sides Apr 10 04:17:15 ungureanuvlad: ah... donno much about that... look like it is loading from a USB mass storage device? Apr 10 04:17:32 ds2: well the host JTAG debugger is 3.3v logic Apr 10 04:17:37 ds2: yes, exactly just copies file at some specific place in memory Apr 10 04:17:47 target could be be 1v8, 3v3 or 5v Apr 10 04:17:59 mranostay: that is just 1 level of buffering Apr 10 04:18:03 aka level xlation Apr 10 04:20:11 mranostay, who? what? Apr 10 04:21:19 ds2: so with u-boot i get both kernel and filesystem and then i run the kernel which does the rest of the job having the filesystem given as an argument ? Apr 10 04:21:43 sort of Apr 10 04:21:55 things become messy once you get to u-boot since there is so many options tehre Apr 10 04:22:56 on how to get things in the dram? Apr 10 04:24:21 MLO reads U-boot Apr 10 04:24:34 u-boot can read from some media into DRAM Apr 10 04:24:39 or you can kermit it over Apr 10 04:25:26 mranostay, what do you mean "buffer from both side" Apr 10 04:27:12 ds2: mhmhm Apr 10 04:28:16 is this channel logged ? Apr 10 04:28:58 sssh...we din't want to wake the trolls Apr 10 04:28:59 yes Apr 10 04:29:05 too late Apr 10 04:29:58 ya, there is my logger Apr 10 04:31:58 ds2: so the main start of the project should be the android driver as i see it then after this is working move to work on the board part Apr 10 04:41:26 ds2: sorry to bug you with this, but if u can provide the answer for the following question i would be really grateful . thank you in advance ! If your project is successfully completed, what will its impact be on the BeagleBoard.org community? need answer from one SoC admin and from a community user. Apr 10 05:37:53 jkridner__: right in time for the build :) Apr 10 05:38:20 sweet. I'm on to testing panto's pwm patches. will you have those in the build too? Apr 10 05:39:01 no Apr 10 05:39:08 waiting for someone to test them first Apr 10 06:10:49 morning Apr 10 06:23:23 hi tsjsieb Apr 10 06:23:28 hi tasslehoff Apr 10 06:23:38 hi woglinde :) Apr 10 06:27:11 hi woglinde Apr 10 06:30:47 [kernel] koenkooi pushed 1 new commit to 3.8: http://git.io/J56TBg Apr 10 06:30:47 kernel/3.8 66f856e Koen Kooi: 3.8: add pwm-test patches so people can well, test them Apr 10 06:30:56 jkridner__: your patches are in, a build using those has been posted Apr 10 06:31:25 jkridner__: they are also being uploaded to the feeds, so opkg upgrade will work Apr 10 06:37:19 gm koen Apr 10 06:41:12 hey woglinde Apr 10 06:45:17 Can anyone assist me switching off swap and any other writing to SD card that is strictly speaking not needed Apr 10 06:46:56 The reason I ask is that after 8 months continues use my SD card is corrupt. I need to test over longer periods. Apr 10 07:08:09 <_av500_> swapoff Apr 10 07:13:22 with the SD card in the board for both xM and bone swap is off by default Apr 10 07:18:14 hi! Apr 10 07:30:17 <_av500_> lo! Apr 10 07:31:16 hi! Apr 10 07:31:57 <_av500_> lo! Apr 10 07:33:46 lo! Apr 10 07:33:56 * koen consults the manchester encoding cheat sheet Apr 10 07:34:58 shannan Apr 10 07:59:37 <_av500_> shaman Apr 10 08:07:17 panto: pong Apr 10 08:10:15 i'm still struggling with the pwm stuff on the beaglebone. i'd like to use ehrpwm1a and ehrpwm1b (later maybe ecap, too) from a kernel module which i'll write, i think it'll be based on the geiger cape... but i'm not sure how to initialize both pwm channels there. may the pwms property appear multiple times? how will the names be mapped? Apr 10 08:24:37 oh it seems to be just pwms = <&ehrpwm1 0 500000 0>,<&ehrpwm1 1 500000 0>; and two names... i guess, that's too easy... Apr 10 08:33:44 JIHAD dudes! Apr 10 08:38:03 kothJi Apr 10 08:38:25 rotti again so easy? Apr 10 08:38:41 hm shannon it was Apr 10 09:33:06 Dear all, I want to build kernel for the beaglebone. I took the latest source from the official git repository. I couldn't find instructions on how to build it. I am not using any specific build environment. I have the am335x TI SDK installed on my host machine. Apr 10 09:33:43 I don't think you need that SDK Apr 10 09:34:04 undless you want to use the TI kernel Apr 10 09:34:12 Ok I understand Apr 10 09:34:27 I use beablebone kernel taken from git repository and patches applied Apr 10 09:35:04 "I couldn't find instructions on how to build it" Apr 10 09:35:12 have you looked at the README ? Apr 10 09:35:56 yes the README.md file Apr 10 09:36:04 I know, out of the box thinking, reading the README Apr 10 09:36:43 Hari_: the README.md has the 'usage' section Apr 10 09:36:47 and gosh darnit Apr 10 09:36:47 yes Apr 10 09:36:53 instructions in that section Apr 10 09:36:59 woglinde: yep, at least it seems so... i still have to find out if it really works the way i want it to, but it looks good Apr 10 09:37:30 I am sorry I am not sure which README you are talking about Apr 10 09:38:12 the README.md that's in the "atest source from the official git repository" Apr 10 09:38:42 exact steps, with extreme prejudice Apr 10 09:38:42 Hari_: which version do you want to use? Apr 10 09:39:27 I hava a Rev A6 board with me Apr 10 09:39:32 if a text editor is too hard to use, you can view it online: https://github.com/beagleboard/kernel/tree/3.8 Apr 10 09:40:13 Thanks Koen Apr 10 09:40:26 I am able to see the instructions from the link you shared Apr 10 09:40:43 Hari_: so why couldn't you find it in your checkout? Apr 10 09:41:44 The Readme.md file that I have has only first few lines Apr 10 09:42:05 I am not sure what has happened Apr 10 09:42:14 and after you checkout the right branch, what does the README.md say? Apr 10 09:42:50 besides that Apr 10 09:42:55 it's a standard linux kernel Apr 10 09:43:04 it shouldn't need instructions Apr 10 09:44:18 I am new to this Apr 10 09:45:22 Hari_: if you're new to this, you're better off following http://www.angstrom-distribution.org/building-angstrom and http://www.slimlogic.co.uk/2011/05/openembeddedangstrom-kernel-workflow/ Apr 10 09:46:28 I saw the first link Apr 10 09:46:56 It was using OE. I just wanted to build the kernel wihtout OE or Angstrom Apr 10 09:50:28 Can I use the am335x evm toolchain arm-arago-linux-gnueabi-gcc Apr 10 09:51:36 no Apr 10 09:51:39 that one is broken Apr 10 09:52:09 Angstrom gives you a working toolchain, use that Apr 10 09:52:26 Hari_: if you want to use arago stuff, you need to ask TI for support, not here Apr 10 09:52:32 this isn't a TI support channel Apr 10 09:52:58 ok Apr 10 09:53:10 which toolchain I need to use for beaglebone Apr 10 09:53:36 the one oe builds Apr 10 09:53:44 ok Apr 10 09:53:56 So I have to use OE...there is no other way. Is it so ?? Apr 10 09:53:58 or download linaro toolchain Apr 10 09:54:28 if you are making seriouse developement use oe Apr 10 09:54:53 Ok.. Apr 10 09:54:58 s/are making/want to make/ Apr 10 09:57:04 thank you all Apr 10 09:57:56 Hari_: send koen and woglinde some beer Apr 10 09:58:06 sure Apr 10 10:02:02 Hari_, denix will talk about arago, but he is asleep. I hope :) Apr 10 10:15:01 <_av500_> one does not simply talk about arago here Apr 10 10:18:49 hey Apr 10 10:23:07 troll Apr 10 10:36:41 raspberry PI Apr 10 10:40:01 is there any ETA for the new beaglebone yet? Apr 10 10:43:42 * ogra_` throws ubuntu CDs at mrpackethead Apr 10 10:48:04 why does linux got wdt time out in seconds and not mseconds Apr 10 10:48:23 sigh, i my trolling was too lame Apr 10 10:48:24 :-) Apr 10 10:48:31 Apr 10 10:50:16 rsv: what timeout? Apr 10 10:50:22 watchdog? Apr 10 10:50:32 yes Apr 10 10:50:39 what would you like it to be? Apr 10 10:51:26 i know the timeout should not be too low Apr 10 10:51:38 why do you want sub-second granularity? Apr 10 10:51:43 I don't see any point in that Apr 10 10:51:50 that is correct Apr 10 10:52:09 i was thinking why watchdog docs have it in seconds. so i got that doubt Apr 10 10:52:10 and setting it less than 1s would be suicide anyway Apr 10 10:52:23 linux is not an rtos Apr 10 10:52:55 so what is the minimum time that we can guantee that a feeding the watchdog will happen Apr 10 10:53:04 there are no guarantees Apr 10 10:53:09 it's not an rtos Apr 10 10:53:23 not ever 400ms? Apr 10 10:53:29 or 500ms Apr 10 10:53:39 500ms would probably work most of the time Apr 10 10:54:03 even 100ms would _mostly_ work Apr 10 10:54:29 but you don't want a spurious watchdog reset just because you had bad luck with the timing of some events Apr 10 10:56:10 yes, in linux we cannot say this thread can run every 30-40ms Apr 10 10:56:21 with resonably Apr 10 10:56:26 ? Apr 10 10:57:12 timing on that scale works well enough for things like video playback Apr 10 10:57:22 but there are occasionally glitches Apr 10 10:58:28 okay in the case of wdt if there is occasional glitch it is fatal that i understood Apr 10 10:59:03 the watchdog is for catching a lockup from which the system will not recover Apr 10 11:04:57 got it Apr 10 11:06:10 default watchdog for a lot of devices is in the 2 minutes range Apr 10 11:08:52 rsv, whatst eh watch dog for Apr 10 11:09:16 mrpackethead snacks Apr 10 11:09:29 mrpackethead: yeah! Apr 10 11:10:49 mru: i got the doubt when i went through: "Documentation/watchdog/watchdog-kernel-api.txt" in which the timeout is seconds Apr 10 11:12:07 so my thought process was what is the min period linux can guantee that someone can feed the watchdog, Apr 10 11:12:39 o.O Apr 10 11:12:47 i think i got your point as well, Apr 10 11:12:58 _no guarantees Apr 10 11:13:15 that it should be long enough so that it can catch fatal conditions Apr 10 11:17:41 tsv what you wanna do with watchdog then? Apr 10 11:19:03 dogfooding :) Apr 10 11:21:00 gm ogra Apr 10 11:21:08 morning :) Apr 10 11:22:41 woglinde: knowledge gathering Apr 10 11:25:44 o.O Apr 10 11:34:31 Hello. I was wondering if the knowledge of Linux compulsory in order to apply for BeagleBoard(through GSoC)? Apr 10 11:35:00 Koushik depends Apr 10 11:35:16 if you want to run some other os and can find a mentor for it, that's fine Apr 10 11:35:19 Koushik what you wanna do Apr 10 11:36:58 Although some projects lists requirements as C/C++, I have to take a Qualification Test in order to apply for it. The test is to create a statically-linked ARM Linux "hello world" style executable.. Apr 10 11:37:11 I'm not sure where to start learning it.. Apr 10 11:39:37 what would you like to do as your project? Apr 10 11:41:37 I'm more interested in Arduino-related projects... Apr 10 11:41:53 <_av500_> then probably beagleboard gsoc is not for you Apr 10 11:42:05 beagleboard is not arduino Apr 10 11:42:07 Although i've heard of Beagleboards, i've never worked on it.. Apr 10 11:42:22 <_av500_> rather APOC - Arduino Phorum of Code Apr 10 11:42:52 <_av500_> Koushik: any linux knowledge at all? Apr 10 11:43:33 Yes, But there are projects for implmenting arduino and starterware o/s library for accessing the hardware Apr 10 11:43:40 _av500_ no :-( Apr 10 11:43:59 Koushik than better not Apr 10 11:44:16 you can not learn all the stuff in three four months Apr 10 11:44:26 to deliver good quality Apr 10 11:44:44 <_av500_> Koushik: ah yes, this one is not really linux oriented Apr 10 11:44:53 you could but it would take a lot of your time Apr 10 11:45:12 hm ah right the starterware Apr 10 11:45:35 @_av500_ : are you a GSoC admin or only jkridner is ? Apr 10 11:45:58 already trying to vote me off the island? Apr 10 11:45:59 <_av500_> admin no Apr 10 11:46:11 <_av500_> ungureanuvlad: why? Apr 10 11:46:20 he jkridner is still here Apr 10 11:46:23 <_av500_> jkridner: show of force? Apr 10 11:46:29 ;-) Apr 10 11:46:55 <_av500_> jkridner: so congrats on getting us accepted :) Apr 10 11:47:04 _av500_: i need to get feedback from the community regarding the project i want to apply. Android based boot host Apr 10 11:47:15 thanks! Apr 10 11:47:31 would love to see some of your thoughts on this android boot thing put on the wiki. :) Apr 10 11:47:38 <_av500_> :) Apr 10 11:47:57 <_av500_> is there pizza? Apr 10 11:48:01 * mru has no idea what "android boot host" might mean Apr 10 11:48:29 <_av500_> mru: boot a bb off an android phone Apr 10 11:48:30 av500 wait for wmat Apr 10 11:48:50 wtf Apr 10 11:48:52 if someone is willing to share an answer with me here is the full question http://pastebin.com/YR4RDxXB . thanks in advance :) Apr 10 11:50:32 well, so answer it Apr 10 11:51:13 you haven't even explained what you want to do Apr 10 11:52:43 as discussed with ds2 the whole point of the project is to boot a board from a peripherial device like an android phone which will analyze the ASIC IDs that are sent by the board and supply the right stages of the boot process. Apr 10 11:53:05 usb boot? Apr 10 11:53:09 and in the end will supply the filesystem Apr 10 11:53:21 yes, usb but from some android device Apr 10 11:53:44 what's the use case for that? Apr 10 11:53:45 so I could use my TV? Apr 10 11:54:49 you can use this system for board bring ups, like test them Apr 10 11:55:10 @ynezz if the program that will be made for android works on your tv sure, why not Apr 10 11:55:33 why would you be using an android device for that? Apr 10 11:55:35 mru, you can download the image via your phone and boot the beagle off it ... so you dont need a PC ... just a brain interface later to actually use the beagle once booted Apr 10 11:55:36 why not a real computer? Apr 10 11:55:54 cheaper then phone Apr 10 11:56:26 I'm not convinced Apr 10 11:57:02 pessimist ... it will bring you the brain interface for free ! Apr 10 11:57:08 to expand the software on the pc is not really though, making it work on an android device is more as a challange Apr 10 11:57:11 maybe if android some day becomes a viable "desktop" system complete with dev tools Apr 10 11:57:15 <_av500_> mru: mind that this project was proposed by jkridner Apr 10 11:57:20 but until then, this is just a gimmick Apr 10 11:57:34 _av500_: unpossible, it doesn't have nodejs Apr 10 11:57:36 lol Apr 10 11:57:50 sure it does. ;-) Apr 10 11:57:56 mru: heh, I was about to say the same :) Apr 10 11:58:19 jkridner: what's the latest word from the cloud9 people? Apr 10 11:58:31 AWOL Apr 10 11:58:35 i'll send a ping Apr 10 11:59:54 sorry.. was AFK.. I think I won't be of much help in this project :-( .. But thanks for the feedback! Apr 10 12:00:44 <_av500_> Koushik: we did not say that Apr 10 12:00:54 <_av500_> dont giveup so fast Apr 10 12:02:48 jkridner: everything would be fixed it they merged https://github.com/ajaxorg/o3/pull/26 Apr 10 12:03:13 jkridner: but cloud9 deps seem to be abandonware Apr 10 12:03:23 <_av500_> +1 Apr 10 12:03:29 No. Not blaming anyone. It's just that the topics are quite advanced for my skillset.. Apr 10 12:03:30 jkridner: alternatively we can apply https://github.com/desaintmartin/cloud9/commit/627f16aeef3212f07b6e1c3e218de38dc97a19ca Apr 10 12:04:49 that is a bit of a long path. forking the whole thing and making it sane might be the approach, but let me ping them a couple of more times first. Apr 10 12:05:00 they were helpful in giving me some debug processes. Apr 10 12:05:21 maybe I should bring you into the conversation? Apr 10 12:06:04 They are in Amsterdam. :-) Apr 10 12:06:06 koen: wow, the lack of clue in that pull req is astounding Apr 10 12:11:05 jkridner: I've been looking at that Apr 10 12:12:07 jkridner: looks like most c9 developers now work for strongloop Apr 10 12:12:20 <_av500_> but maybe its finished Apr 10 12:12:27 <_av500_> like the kernel after 3.8 Apr 10 12:14:14 TI kernel is finished...don't know about your hippy kernel Apr 10 12:17:50 <_av500_> 2.4 is done since long Apr 10 12:23:03 we've got a project to backport am335x to .96pl4 Apr 10 12:40:48 <_av500_> mdp: why not minix then? Apr 10 12:41:11 <_av500_> am335x deserves better than a toy os Apr 10 12:41:13 indeed, why not? Apr 10 12:41:35 at least it has a team of qualified academics working on it Apr 10 12:49:52 mdp: what could possibly go wrong with that? Apr 10 12:52:54 it's guaranteed to have no gotos Apr 10 13:31:24 re Apr 10 13:59:53 http://pastebin.com/bbq3szLZ (top 60 files in the minix source tree containing goto's) Apr 10 14:01:05 this is after running the build so lots of gnu gcc stuff. of course the kernel is way smaller then gcc so we would need to get a goto per line of code sys/ufs/ufs/ufs_vnops.c is quite good at that :P Apr 10 14:05:23 hehe Apr 10 14:08:15 http://blog.regehr.org/archives/894 was also a nice read Apr 10 14:08:30 about "good" usage of goto in the linux kernel Apr 10 14:08:35 how can i use gpios from a kernel module? is gpio = of_get_named_gpio() enough so that i can use gpio_set_value() or is something missing? Apr 10 14:08:45 regehr is one of the few academics with a clue about the real world Apr 10 14:10:39 the funny thing about the article is that he assumes that if there is a goto "out" there probably also should be a goto out2 (to prevent nesting) but the number don't really show that Apr 10 14:11:13 often that's not needed because free(NULL) is harmless Apr 10 14:16:18 Hello guys! :) Apr 10 14:17:19 I was wondering if somebody could educate me a bit on how to do GPMC between an FPGA and the bone Apr 10 14:18:50 <_av500_> OMG: http://www.phoronix.com/scan.php?page=article&item=ubuntu1304_apple_osx1083&num=2 Apr 10 14:19:19 haha Apr 10 14:19:28 and sooo unbiased Apr 10 14:20:11 even on the pages where apple is clearly faster it says "results are mixed" Apr 10 14:20:20 (see page 4) Apr 10 14:20:54 <_av500_> ogra_: I guess Ubuntu failed the PhotoShop and AfterEffects benchmark at least Apr 10 14:20:55 phoronix has the advantage that you don't need to read it to know what it says Apr 10 14:20:56 Rotti, first you need to pass the return value from of_get_named_gpio() into devm_gpio_request_one() Apr 10 14:21:10 _av500_: photoshop should run in wine Apr 10 14:21:16 I recommend dev_* calls to automagically take care of clean up when your driver unloads Apr 10 14:21:22 <_av500_> mru: I nominate that last line for 26+26=54 Apr 10 14:21:29 s/dev_/devm_/ Apr 10 14:21:34 i mean, as an ubuntu guy i'm happy to see that ... Apr 10 14:22:01 _av500_: which line? Apr 10 14:22:06 but still ... it would be nice if it was anywhere close to scientific Apr 10 14:22:10 <_av500_> mru: mru > phoronix has the advantage that you don't need to read it to know what it says Apr 10 14:22:17 ++ Apr 10 14:25:32 mdp: thnks, i'll try that one... err = devm_gpio_request_one(pdev->dev, info->gpio_dir_left, 0, "gpio-dir-left"); would that one work? Apr 10 14:25:42 (not sure about flags argument) Apr 10 14:26:26 flags is important Apr 10 14:27:01 look at the GPIOF_* definitions Apr 10 14:27:26 if you are using one as an input you better have GPIOF_DIR_IN set, for example Apr 10 14:27:44 and before somebody asks..this directional stuff is 100% independent of pinmuxing concepts Apr 10 14:28:10 mdp: ah, so gpio_direction_[in,out]put is not needed? Apr 10 14:29:31 if you do *request_one() it will call it for you based on flags Apr 10 14:29:52 Hi mdp! :) I am back to ask more questions :) Apr 10 14:29:55 you can also pass GPIOF_INIT_HIGH|LOW to get an initial output too Apr 10 14:31:32 mdp: are you familiar with use of the GPMC on bone? I am reading the manual right now, and I am sure I'll have a host of questions to ask in a little while :) Apr 10 14:31:53 I think that would make my life easier rather than banging on the PRU Apr 10 14:32:34 Rotti, so to review it's "call of helper to get gpio namespace integer", "call request one with namespace integer and proper flags (and nice label)", and "do your gpio_set_value" Apr 10 14:33:21 Rotti, you don't need to release if you use devm_ it'll release the gpio resource as a part of devres framework when you are done with the driver Apr 10 14:33:47 Stefan_, hi, only at the level of playing around with it Apr 10 14:34:25 http://engineersofthecorn.blogspot.com/2012/06/faux-gpmc-interfacing-with-beaglebone.html Apr 10 14:34:39 ancient stuff that doesn't work on a current kernel Apr 10 14:35:22 mdp: well basically I would like to use it to push data to an FPGA Apr 10 14:35:56 mdp: and since the Xilinx tools no longer support asynchronous FIFOs, I can't just use the GPIO, I need something with a clock Apr 10 14:36:52 mdp: hence why I was looking into the PRU, but there I would have to write most of the stuff myself, whereas I am hoping with the GPMC it's more or less configure and just push data Apr 10 14:38:06 depends on whether you have the verilog to work with the gpmc cycles done Apr 10 14:38:47 i.e.are you using a mode where you need to handle muxed address/data cycles? Apr 10 14:39:01 depends on your application Apr 10 14:39:33