2015-03-16T00:22:36 *** Hesham has joined #rtems 2015-03-16T00:51:17 *** kiwichris has quit IRC () 2015-03-16T00:51:22 *** DrJoel_ has quit IRC (Ping timeout: 240 seconds) 2015-03-16T00:51:36 *** DrJoel_ has joined #rtems 2015-03-16T00:54:26 *** zoso has joined #rtems 2015-03-16T00:56:25 *** zoso has quit IRC (Client Quit) 2015-03-16T00:56:42 *** zoso has joined #rtems 2015-03-16T00:59:17 *** monstr has joined #rtems 2015-03-16T01:55:19 *** andrei_ has joined #rtems 2015-03-16T02:56:03 *** Hesham has quit IRC (Ping timeout: 264 seconds) 2015-03-16T03:29:54 [trac] #2307/enhancement (v:4.11) created by sebastian.huber (Improved watchdog implementation for SMP) http://devel.rtems.org/ticket/2307 2015-03-16T03:32:42 [trac] #2271/enhancement (accepted) updated by sebastian.huber (empty comment) http://devel.rtems.org/ticket/2271 2015-03-16T03:33:17 *** Hesham has joined #rtems 2015-03-16T03:33:36 [trac] #2307/enhancement (new) updated by sebastian.huber (empty comment) http://devel.rtems.org/ticket/2307 2015-03-16T03:39:34 [trac] GSoC/2015 edited by yangqiao (empty comment) http://devel.rtems.org/wiki/GSoC/2015 2015-03-16T03:44:43 Hello 2015-03-16T03:47:46 I understood that the Condition Variables project involves implementing condition variables in supercore 2015-03-16T03:48:00 then developing a "wrapper" in classic api 2015-03-16T03:48:09 to hide both the implementation score and pthreads 2015-03-16T03:48:14 but I don't understand something 2015-03-16T03:48:51 pthreads are linux specific 2015-03-16T03:49:18 and it's POSIX 2015-03-16T03:50:19 what is "classic API" ? 2015-03-16T03:51:10 it's the API based on RTEID or did I misunderstand? 2015-03-16T04:04:23 andrei_: What do you mean by "pthreads are linux specific"? 2015-03-16T04:07:02 that pthreads are used mostly by unix systems, hence, by unix applications 2015-03-16T04:08:39 Yeah but not specific to "Linux", it can run on any platform that gives an implementation for POSIX. 2015-03-16T04:09:33 I suppose that's true 2015-03-16T04:09:36 Classic API is the RTEMS API. For example if you want to create a task you would call "rtems_task_create" 2015-03-16T04:09:58 yeah, so my understanding is that currently RTEMS has only pthreads support 2015-03-16T04:10:04 for cv 2015-03-16T04:10:35 it would need the same for RTEMS API and a wrapper (like a function with an ifdef in it, in a simplified description) that you call from anywhere 2015-03-16T04:10:40 RTEMS POSIX implementation is wrapping some existing Classic API functions 2015-03-16T04:11:10 so for pthread_create, it calls rtems_task_create in turn. 2015-03-16T04:15:08 *** starprince has joined #rtems 2015-03-16T04:16:02 Not the same RTEMS API, some POSIX functions may not have corresponding RTEMS functions, which need a custom implementation and/or workarounds 2015-03-16T04:16:06 Hesham: https://devel.rtems.org/wiki/Developer/Projects/Open/Condition_Variables 2015-03-16T04:16:14 Achieve this by refactoring the POSIX condvar to a supercore implementation 2015-03-16T04:16:51 maybe I'm wrong, but my understanding is that I have to wrap to score from POSIX, not other way around 2015-03-16T04:17:33 also, thanks a lot for helping me out ! 2015-03-16T04:20:11 Honestly I don't understand the specific details of how the project should go, you may want to wait for RTEMS experts to reply :) 2015-03-16T04:21:04 yeah, no problem, thanks :) 2015-03-16T04:21:27 Welcome, good luck with the proposal ;) 2015-03-16T04:21:41 I ll also keep searching, I got (imo) a good general idea, just stuck with specifics 2015-03-16T04:21:45 likewise 2015-03-16T04:26:35 *** namste has joined #rtems 2015-03-16T04:28:39 *** chipsMonstret has joined #rtems 2015-03-16T04:41:30 *** namste has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-16T04:47:47 *** siddharth7 has joined #rtems 2015-03-16T05:30:41 *** chipsMonstret has quit IRC (Quit: My Mac has gone to sleep. ZZZzzz…) 2015-03-16T06:26:02 *** ragunath has joined #rtems 2015-03-16T07:13:40 ketul: yes i did 2015-03-16T07:29:23 *** siddharth7 has quit IRC (Read error: Connection reset by peer) 2015-03-16T07:52:59 [git] dc5e5f4 by Sebastian Huber: [tmtests/tmfine01: New test] http://devel.rtems.org/changeset/dc5e5f4/rtems 2015-03-16T08:00:19 *** DrJoel has quit IRC (Quit: Leaving) 2015-03-16T08:20:31 *** antgreen` has quit IRC (Ping timeout: 244 seconds) 2015-03-16T08:20:58 *** antgreen has quit IRC (Ping timeout: 265 seconds) 2015-03-16T08:31:37 *** DrJoel_ has quit IRC (Quit: Hard work pays off in the future, laziness pays off now) 2015-03-16T08:35:39 *** Davidbrcz has joined #rtems 2015-03-16T08:51:54 *** zoso has quit IRC (Quit: Leaving) 2015-03-16T08:59:16 *** gedare has joined #rtems 2015-03-16T08:59:16 *** ChanServ sets mode: +o gedare 2015-03-16T09:01:25 *** andrei_ has quit IRC (Ping timeout: 246 seconds) 2015-03-16T09:02:28 gedare: hi gedare 2015-03-16T09:02:51 hi ragunath 2015-03-16T09:03:10 gedare: is this the latest document on rtems i2c? http://rtemsaio.googlecode.com/svn/trunk/rtems/cpukit/libi2c/README_libi2c 2015-03-16T09:03:49 no that is some old stuff there. 2015-03-16T09:04:25 gedare: thank god i asked .... is there any new one available? 2015-03-16T09:04:43 so, the code is located under https://git.rtems.org/rtems/tree/cpukit/dev 2015-03-16T09:04:58 i don't know if there is documentation available yet 2015-03-16T09:06:09 why are you using rtemsaio.googlecode.com? 2015-03-16T09:06:13 something you found? 2015-03-16T09:06:50 gedare: i wanted to know abt the i2c interfaces in rtems .. in google i found this 2015-03-16T09:07:24 gedare: it seems old code still remains in the code base 2015-03-16T09:07:52 there is still the libi2c, because some may still use it 2015-03-16T09:08:02 but the dev/i2c is the newer, better i2c approach 2015-03-16T09:08:13 we should eventually try to get rid of libi2c i guess. 2015-03-16T09:10:07 gedare: in the mail threads i saw a discussion abt a new i2c interface .. wanted to know which one it is 2015-03-16T09:10:32 yeah, it is the cpukit/dev/i2c one. 2015-03-16T09:10:45 gedare: cool .... i will check it 2015-03-16T09:10:47 always ask if you don't know. ;0 2015-03-16T09:11:16 gedare: definitely :) 2015-03-16T09:29:24 *** Davidbrcz has quit IRC (Ping timeout: 245 seconds) 2015-03-16T09:36:18 [trac] Test_with_fgl.png (Test event send/receive with fine grained locking.) attached to #2273 by sebastian.huber http://devel.rtems.org/ticket/2273 2015-03-16T09:36:39 [trac] test_with_giant_lock.png (Test event send/receive with Giant lock.) attached to #2273 by sebastian.huber http://devel.rtems.org/ticket/2273 2015-03-16T09:44:47 [git] 71260b4 by Alan Cudmore: [preliminary Raspberry Pi Model 2 support This patch adds a BSP variant for the Raspberry Pi 2. You can build both variants by configuring with the option --enable-rtemsbsp="raspberrypi2 raspberrypi" For the current BSP, the only change was the peripheral register base address and the compiler options. The 2015-03-16T09:44:47 raspberrypi/make/custom rules were re-factored...] http://devel.rtems.org/changeset/71260b4/rtems 2015-03-16T09:45:02 *** DrJoel has joined #rtems 2015-03-16T09:45:02 *** DrJoel has joined #rtems 2015-03-16T09:45:02 *** ChanServ sets mode: +o DrJoel 2015-03-16T09:46:27 hi DrJoel 2015-03-16T09:49:09 good morning 2015-03-16T09:49:26 hmm... claas' uni email address bounced. :( 2015-03-16T09:49:51 i thought it might. 2015-03-16T09:50:06 [trac] #2273/enhancement (accepted) updated by sebastian.huber (The new test [dc5e5f44485b8f9709da767f79595d2fa8aff74d/rtems] shows that the ...) http://devel.rtems.org/ticket/2273 2015-03-16T09:52:43 the sparc64 warnings center on a couple of printf formats which probably should be using PRIxxx constants and what I am guessing is a place where the fatal error processing path uses uint32_t instead of uintptr_t 2015-03-16T09:53:01 ../../../../../../rtems/c/src/../../cpukit/libcsupport/src/__assert.c: In function '__assert_func': 2015-03-16T09:53:01 ../../../../../../rtems/c/src/../../cpukit/libcsupport/src/__assert.c:52:43: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] 2015-03-16T09:53:01 rtems_fatal( RTEMS_FATAL_SOURCE_ASSERT, (rtems_fatal_code) &assert_context ); 2015-03-16T09:53:10 *** ragunath is now known as ragunath|away 2015-03-16T09:53:11 yup that sounds about right 2015-03-16T09:53:13 There is a similar warning in some of the tests 2015-03-16T09:53:28 if you send me a list i can chip away at it i hope 2015-03-16T09:53:41 i'm losing energy already though so i think i'm definitely coming down with something 2015-03-16T09:55:15 :( don't let it get worse. If you need to go to the doctor, go. 2015-03-16T09:55:59 But I will admit to doing some great work in a hotel room while too sick to work. with only the BBC news in English, TV wasn 2015-03-16T09:56:05 wasn't interesting 2015-03-16T09:58:01 oh its not bad right now. 2015-03-16T09:58:14 heh 2015-03-16T10:02:03 Stay off of Fox News. LOL 2015-03-16T10:02:49 i don't have cable anymore 2015-03-16T10:03:02 i "cut the cord" about a year ago 2015-03-16T10:03:21 spinning up sparc64 tools 2015-03-16T10:03:35 Then watch South Park .. I watched the freenium games one this weekend with one of my sons. We were introduced to the new character Canadian Satan 2015-03-16T10:03:56 Thanks on the sparc64. I suspect some of those will be general issues which fix some of the 16 bit target issues. 2015-03-16T10:04:21 I am also hoping that the discussion I started on newlib@ will result in more of the 16 bit target warnings disappearing 2015-03-16T10:04:43 rm-rf h8300 2015-03-16T10:04:45 ;-) 2015-03-16T10:05:05 i kid 2015-03-16T10:10:02 After 4.11, I think some targets will be shot. I do not kid. 2015-03-16T10:10:11 m32r is definitely near the top of my list. 2015-03-16T10:10:24 m32c and h8300 are interesting as functional 16 bit targets 2015-03-16T10:13:58 I am trying to clean up odd edges so when gdb and RSB issues resolved, we know state better and can decide to branch. 2015-03-16T10:14:43 I also want to tag all tickets that are warning related with key word "warning" and have a report for those. My warning script is starting to ignore those with tickets and the -Wcpp deprecated ones 2015-03-16T10:21:26 *** ragunath|away is now known as ragunath 2015-03-16T10:26:47 *** dr__house|away is now known as dr__house 2015-03-16T10:30:24 *** Davidbrcz has joined #rtems 2015-03-16T10:44:17 *** ketul has joined #rtems 2015-03-16T10:54:34 [trac] new_bintime_without.png () attached to #2271 by AlexKrEB http://devel.rtems.org/ticket/2271 2015-03-16T10:54:37 beng-nl: hi you there? 2015-03-16T10:54:45 [trac] new_bintime_with.png () attached to #2271 by AlexKrEB http://devel.rtems.org/ticket/2271 2015-03-16T10:54:55 [trac] new_uptime_with.png () attached to #2271 by AlexKrEB http://devel.rtems.org/ticket/2271 2015-03-16T10:55:05 [trac] new_uptime_without.png () attached to #2271 by AlexKrEB http://devel.rtems.org/ticket/2271 2015-03-16T10:57:05 DrJoel: i think sometimes ptr size is long long. 2015-03-16T10:57:16 anyway there isn't a guarantee a ptr fits in long 2015-03-16T10:57:24 thus we have the uintptr_t tyep 2015-03-16T10:58:24 Yep. I think it is unfortunately a very complicated issue. The logic in newlib/configure.in determines if it is long long, long, or int. But with multilib's, you can't know for sure at configure time 2015-03-16T10:58:37 The question is how to do this at cpp time 2015-03-16T10:58:41 right. 2015-03-16T10:58:54 ok just making sure you have the 'long long' case in tyour head too 2015-03-16T10:59:13 in my head i'm debating: dayquil or napquil? 2015-03-16T10:59:14 ;) 2015-03-16T10:59:44 A nap sounds nice :) 2015-03-16T11:18:07 [trac] #2271/enhancement (accepted) updated by AlexKrEB (Plots with hardware periphery (NGMP) These plots as well as those in the upcoming ...) http://devel.rtems.org/ticket/2271 2015-03-16T11:20:00 [trac] #2271/enhancement (accepted) updated by gedare (Can you please scale-down the sizes of the images they don't need to be so large to ...) http://devel.rtems.org/ticket/2271 2015-03-16T11:23:50 *** acer has joined #rtems 2015-03-16T11:24:07 DrJoel: ../../../../../../../../../rtems/c/src/../../cpukit/libfs/src/nfsclient/src/sock_mbuf.c:157:1: warning: no previous prototype for 'sendto_nocpy' [-Wmissing-prototypes] 2015-03-16T11:24:20 i don't see it in the log you sent me, but its showing in my build 2015-03-16T11:24:44 or did you --disable-networking 2015-03-16T11:27:58 *** acer has quit IRC (Client Quit) 2015-03-16T11:28:14 I disabled networking. 2015-03-16T11:28:21 gotcha. 2015-03-16T11:28:25 That's a sticky mess itself. :( 2015-03-16T11:28:38 yeah not worth touching with a 10' pole. 2015-03-16T11:28:56 I didn't want to go there until the very last, if at all. 2015-03-16T11:42:51 * DrJoel pokes students.. remember to apply to the GSoC site and, if eligible, also the SOCIS site 2015-03-16T11:45:13 yes we need to let students know, it is most important to get "something" uploaded, and it can be updated before the deadline 2015-03-16T11:45:32 DrJoel: sure DrJoel ...link opens 2 hours from now 2015-03-16T11:45:48 Thanks. 2015-03-16T11:46:11 gedare: politely asked the person who has the code alan referred to for a relicense. 2015-03-16T11:46:24 He can only say no 2015-03-16T11:46:25 looking into it 2015-03-16T11:46:35 there are two related projects there 2015-03-16T11:46:47 For Pi? 2015-03-16T11:47:18 I was just referring to the code base he thinks would be useful to reuse. It is under GPL V3 so I asked if they would relicense it for use with RTEMS 2015-03-16T11:48:07 ok looks like it is just one github account, proably just one person, so it is at least feasible 2015-03-16T11:48:19 trying to find contact info 2015-03-16T11:49:05 i don't see an obvious contact 2015-03-16T11:49:07 I found it and emailed him. Has a consulting website. Already done. 2015-03-16T11:49:23 http://www.maccasoft.com/contact/ 2015-03-16T11:49:41 I started to cc more than Alan but didn't want to overwhelm him 2015-03-16T11:49:56 ok. no that is good. 2015-03-16T11:50:34 i think that is someone else, maybe 2015-03-16T11:50:51 but he probably knows the other 2015-03-16T11:51:21 I thought the github link at the bottom of the contact page went to the same code 2015-03-16T11:52:23 *** andrei has joined #rtems 2015-03-16T11:52:36 Hello! 2015-03-16T11:52:59 hi 2015-03-16T11:57:07 Sorry, I don't really have a backlog since I switch computers from work to home (now) 2015-03-16T11:57:14 about conditon variables gsoc project 2015-03-16T11:58:15 rtems classic api needs to offer the same functionality as there currently exists posix. All this should reside in supercore 2015-03-16T11:58:18 is that correct? 2015-03-16T11:58:44 It's a really dull question, but I didn't find an answer I could understand 2015-03-16T11:58:57 why would an application choose to use classic api instead of posix one 2015-03-16T12:00:45 it's correct. some aspects of classic api are better suited to a real-time system. 2015-03-16T12:01:20 you might also investigate if there are alternative condvar implementations or interfaces that are better suited for real-time, that might offer an alternative to posix 2015-03-16T12:02:35 is it desired to have an interface, much like thread_create(param1, ... ) 2015-03-16T12:02:41 where param1 can be posix or classic? 2015-03-16T12:03:12 to make the difference between them abstract 2015-03-16T12:03:47 i don't quite follow.. 2015-03-16T12:04:02 sorry, I'll link from project ideas 2015-03-16T12:04:43 ok great. 2015-03-16T12:04:51 " . Achieve this by refactoring the POSIX condvar to a supercore implementation, then create a basic condition variable for classic API " 2015-03-16T12:05:26 what I understood from this is that we should provide an aditional layer, so you wouldn't use a separate API for condvar 2015-03-16T12:05:33 but rather choose a diferent parameter 2015-03-16T12:05:33 When will the ESA application period open? 2015-03-16T12:05:35 i think if you look at semaphores then you might get a good idea 2015-03-16T12:05:45 let me find some links for you 2015-03-16T12:05:46 barriers is likely simpler 2015-03-16T12:05:51 ok 2015-03-16T12:06:06 or message queues. Classic semaphores wrap a couple of score obejcts 2015-03-16T12:07:36 andrei: See https://git.rtems.org/rtems/tree/cpukit/rtems/include/rtems/rtems/barrier.h and https://git.rtems.org/rtems/tree/cpukit/posix/include/rtems/posix/barrier.h and https://git.rtems.org/rtems/tree/cpukit/score/include/rtems/score/corebarrier.h to start 2015-03-16T12:07:54 the first is the "Classic" api for barriers, the second is the "Posix", adn the third is the score layer 2015-03-16T12:08:01 classic and posix are logically parallel to each other 2015-03-16T12:08:08 [trac] GSoC/2015 edited by gtament (empty comment) http://devel.rtems.org/wiki/GSoC/2015 2015-03-16T12:10:16 which we used to have a picture for but now i can't find it. 2015-03-16T12:11:21 awesome! I didn't really manage to think of any incremental patch/feature/fix I could do to make sure I got the hang of this, maybe after looking over that code I'll have a better idea. If you do, please let me know. Thank you ! 2015-03-16T12:12:38 https://docs.rtems.org/doxygen/cpukit/html/structPOSIX__Mutex__Control.html might help. 2015-03-16T12:13:09 https://docs.rtems.org/doxygen/cpukit/html/structPOSIX__Barrier__Control.html 2015-03-16T12:13:14 [git] b4d1f08 by Gedare Bloom: [rfs: cast minor to uintptr_t to truncate explicitly] http://devel.rtems.org/changeset/b4d1f08/rtems 2015-03-16T12:13:14 [git] 1ff449c by Gedare Bloom: [rtems: make error codes uintptr_t] http://devel.rtems.org/changeset/1ff449c/rtems 2015-03-16T12:13:14 [git] ed152a0 by Gedare Bloom: [capture: fix warnings in print] http://devel.rtems.org/changeset/ed152a0/rtems 2015-03-16T12:13:14 [git] ce4e691 by Gedare Bloom: [shell: fix printf warnings] http://devel.rtems.org/changeset/ce4e691/rtems 2015-03-16T12:13:24 DrJoel: pull and re-build 2015-03-16T12:13:34 https://docs.rtems.org/doxygen/cpukit/html/group__ClassicBarrier.html 2015-03-16T12:13:44 there's still a warning in hexdump.c about wint_t vs wchar * 2015-03-16T12:13:55 Classic API and POSIX API barriers are objects wrapped around SuperCore barriers 2015-03-16T12:14:53 Yep. We have had discussions about that. Apparently that uses a FreeBSD printf() format and doesn't follow the POSIX std. 2015-03-16T12:14:57 https://devel.rtems.org/wiki/TBR/UserManual/RTEMS_Software_Architecture 2015-03-16T12:14:59 there it is 2015-03-16T12:15:26 DrJoel: maybe youc an write something up about layering, and link to it from https://devel.rtems.org/wiki/Developer/Coding/Conventions#Layering 2015-03-16T12:15:28 https://devel.rtems.org/wiki/TBR/UserManual/RTEMS_Software_Architecture this helps a lot ! 2015-03-16T12:15:28 :) 2015-03-16T12:16:07 so rtems kernel is not monolithic? 2015-03-16T12:16:16 it is, but it is modular 2015-03-16T12:16:47 https://linuxprograms.wordpress.com/2008/03/07/c-printing-data-types-using-printf-short-wchar_t-long-double/ 2015-03-16T12:17:09 Someone mentioned that you had to put them in an array of wint_t and print the first one 2015-03-16T12:17:24 Now really sure how to do it but that's the only non-networking warning on all targets in cpukit/ 2015-03-16T12:18:10 actually, that looks like a bug 2015-03-16T12:18:44 https://git.rtems.org/rtems/tree/cpukit/libmisc/shell/hexdump-conv.c#n149 2015-03-16T12:19:18 wc_tmp is a pointer 2015-03-16T12:19:50 so should be wc_tmp[0]? My attempted fix is wrong :( 2015-03-16T12:19:51 it should be wc_tmp[0], but i don't quite follow the reason to have this ifdef'd on rtems anyway 2015-03-16T12:20:20 or does wchar_t not print correctly as %lc? 2015-03-16T12:20:32 this looks weird. 2015-03-16T12:20:40 (void)printf("%*s%lc", pad, "", wc); 2015-03-16T12:20:42 should just wrok 2015-03-16T12:20:54 if you get a fix, push it 2015-03-16T12:21:25 i'm going to try removing the ifdef 2015-03-16T12:21:29 it doesn't make sense to me. 2015-03-16T12:22:15 it compiles clean for me without the ifdef 2015-03-16T12:22:31 i suspect if there is a problem, it is because of newlib/gcc definition of wchar_t 2015-03-16T12:23:31 [git] a62220f by Gedare Bloom: [shell: remove rtems ifdef, upstream is correct] http://devel.rtems.org/changeset/a62220f/rtems 2015-03-16T12:25:06 ok that should only leave deprecated warnings I think... 2015-03-16T12:25:37 do you remember what targets had a problem in hexdump that caused you to make that work-around? 2015-03-16T12:26:08 i'm about to check out for the day i think to go rest. 2015-03-16T12:35:53 No. I will start a build of all BSPs again after I pull your ccmmit 2015-03-16T12:36:15 go get some rest 2015-03-16T12:36:57 yup. see you later 2015-03-16T12:40:36 take it easy. Whiskey is your friend :) 2015-03-16T12:41:49 *** gedare has quit IRC (Ping timeout: 264 seconds) 2015-03-16T12:46:00 DrJoel: want to ask that beng-nl wrote about "Booting without uboot" 2015-03-16T12:46:17 so what is the need for that ? 2015-03-16T12:46:53 actually rtems in under GPLv2+ right? 2015-03-16T12:47:41 It isn't about RTEMS licensing as much as it is people with real applications who do not want to ship any GPL v3 code with their system. If you can boot without Uboot, then you have no redistribution requirements at all 2015-03-16T12:55:37 *** Davidbrcz has quit IRC (Quit: Leaving) 2015-03-16T13:05:25 DrJoel, looking over rtems/include/rtems/rtems/barrier.h , and posix/include/rtems/posix/barrier.h 2015-03-16T13:05:31 as well as score/include/rtems/score/corebarrier.h 2015-03-16T13:05:59 we have "posix/include/rtems/posix/condvar.h" 2015-03-16T13:06:08 and we need the one in rtems and in score 2015-03-16T13:06:14 I think I cleared the confusion now 2015-03-16T13:07:12 The score version is intended to be the union of all capabilities needed by any public API version 2015-03-16T13:07:41 The API just adds an "Object" to it with Id, name and puts a facade or veneer on it depending on what you like to call that pattern 2015-03-16T13:07:55 But the "core" behavior is in the "score" 2015-03-16T13:08:56 I'm looking at score/src/corebarrier.c 2015-03-16T13:09:04 and there's just an initialization 2015-03-16T13:09:37 corebarrierimpl.h? 2015-03-16T13:09:55 There is one method per file. Makes it easier for ld to only pull what is needed from library 2015-03-16T13:10:24 clever ! 2015-03-16T13:10:28 $ find . -name "*barrier*" | wc -l 2015-03-16T13:10:29 29 2015-03-16T13:10:33 That's from cpukit/ 2015-03-16T13:11:43 Old school trick to care about code size in a library. If each .o file is essentially one node in the dependency graph, we can trust ld to not pull in anything it shoudlnt 2015-03-16T13:13:51 what you said makes sense, both barrier.h from posix and rtems include rtems/score/corebarrier.h 2015-03-16T13:15:13 message queue is another pretty straightforward example. posix rwlocks follows the pattern but there isn't a classic api rwlock 2015-03-16T13:17:33 afaik, in posix condvar you can assign/set priority to threads 2015-03-16T13:17:50 doesn't that guarantee that a thread always gets executed beneath a fixed time threshold ? 2015-03-16T13:18:11 [git] 674f8b1 by Jamie Iles: [IMFS: don't strlen() an unterminated string in IMFS_dir_read(). a43a3466 (IMFS: Implement variable length node names) introduced a changed to IMFS_jnode_t from being a null terminated string to a separate unterminated string and length. IMFS_dir_read() was still performing a strlen() on this unterminated string though and when 2015-03-16T13:18:11 doing an 'ls' in the fileio ...] http://devel.rtems.org/changeset/674f8b1/rtems 2015-03-16T13:31:26 [git] a2ba004 by Sebastian Huber: [bsp/mbx8xx: Fix warnings] http://devel.rtems.org/changeset/a2ba004/rtems 2015-03-16T13:34:58 So, DrJoel do you have any idea about how to do start for Booting without uboot. :) 2015-03-16T13:35:56 *** antgreen has joined #rtems 2015-03-16T13:37:55 You will probably need a jtag and have to follow how it is done on arm boards used in production system. 2015-03-16T13:38:10 Ask on the mailing list. Sebastian Huber probably can answer your questions 2015-03-16T13:38:29 You may end up needing to flash the RTEMS app in place of UBoot. 2015-03-16T13:38:35 beng-nl.. any thoughts? 2015-03-16T13:39:22 ok. :) 2015-03-16T13:41:29 DrJoel: there was a book "arm systems developer guide" which has a sample bootloader implementation 2015-03-16T13:41:52 sounds like a good reference in this case. Hopefully the code examples are licensed so you can use them 2015-03-16T13:43:10 DrJoel: in the book there is code references , the author explains the details ... but i cannot find the actual code anywhere 2015-03-16T13:43:37 DrJoel: they also have a simple os implementation also 2015-03-16T13:44:40 oh that book I referred during my class :) 2015-03-16T13:45:01 ketul: :) 2015-03-16T13:46:59 *** x10 has joined #rtems 2015-03-16T13:50:56 DrJoel: Is it needed to implement a bootloader? IIRC, RPi BSP is using Linux/grub and the user has to overwrite kernel.img with the RTEMS image. 2015-03-16T13:55:44 *** x10 has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-16T14:06:18 GSoC student applications/proposals are now being accepted. Fill in what you can now 2015-03-16T14:09:12 *** ankur has joined #rtems 2015-03-16T14:15:41 *** gedare has joined #rtems 2015-03-16T14:15:41 *** ChanServ sets mode: +o gedare 2015-03-16T14:21:33 yep DrJoel: started 2015-03-16T14:21:54 trust me.. better to edit it later than fight it on the last day of the application period 2015-03-16T14:23:54 *** siddharth7 has joined #rtems 2015-03-16T14:30:18 *** siddharth7 has quit IRC (Ping timeout: 272 seconds) 2015-03-16T14:36:06 *** monstr has quit IRC (Remote host closed the connection) 2015-03-16T14:37:39 *** ketul_ has joined #rtems 2015-03-16T14:38:28 *** ketul has quit IRC (Ping timeout: 246 seconds) 2015-03-16T14:46:44 DrJoel: typedef ... CPU_Uint32ptr; ? 2015-03-16T14:47:31 any objections to using such in score/cpu.h to declare a type capable to store at least a 32-bit int or a pointer? 2015-03-16T14:48:07 on 16-bit targets it should be: typedef unsigned long CPU_Uint32ptr; 2015-03-16T14:48:17 on 32-bit targets it can just be uintptr_t 2015-03-16T14:48:33 and probably 64-bit targets that is good too 2015-03-16T14:49:15 I am ok with it. 2015-03-16T14:49:18 Feeling better? 2015-03-16T14:49:33 no.. i think i'm just tired, but too much caffeine earlier can't sleep 2015-03-16T14:50:59 :( 2015-03-16T14:52:27 *** Hesham has quit IRC (Ping timeout: 264 seconds) 2015-03-16T14:54:07 what are the 16-bit targets? 2015-03-16T14:55:10 um 2015-03-16T14:55:16 is there a way to check it in types.h insrtead.... 2015-03-16T14:56:24 hmmm 2015-03-16T14:56:26 yeah 2015-03-16T14:59:55 h8300 and m32c. m32c is weird because size_t can't represent all of address space. You can only have 64k offsets if I remember DJ Delorie's explanation 2015-03-16T15:00:01 can we do... 2015-03-16T15:00:27 hmmm 2015-03-16T15:05:04 #if CPU_SIZEOF_POINTER < 4typedef unsigned long CPU_Uint32ptr#elsetypedef uintptr_t CPU_Uint32ptr 2015-03-16T15:05:10 oops 2015-03-16T15:05:50 well, i can just do it manually too 2015-03-16T15:05:54 might be simpler. 2015-03-16T15:11:03 and avr, not that it ever worked... 2015-03-16T15:15:06 avr gcc just is barely buildable. :( 2015-03-16T15:22:05 DrJoel: want me to push my types.h commit before i'm done for the day? 2015-03-16T15:22:44 You can. What do you think it will fix? Should I kill the build I just started? 2015-03-16T15:22:52 yeah 2015-03-16T15:22:55 it touches every target 2015-03-16T15:23:00 GNU parallel was underutilizing the local machine and overloading the remote one. :( 2015-03-16T15:23:10 it should fix the same warning i fixed before about rtems_fatal_error 2015-03-16T15:23:17 but now it should be "safe" for 16-bit targets too 2015-03-16T15:24:30 [git] 46d3c6d by Gedare Bloom: [cpukit: add and use CPU_Uint32ptr type] http://devel.rtems.org/changeset/46d3c6d/rtems 2015-03-16T15:24:39 ok DrJoel 2015-03-16T15:25:15 i'm out 2015-03-16T15:30:28 *** gedare has quit IRC (Ping timeout: 272 seconds) 2015-03-16T15:49:16 verm__: are smtp connections throttled to rtems.org? I am sending a patch set of ~20 patches and it has taken over five minutes to get them sent 2015-03-16T15:52:11 *** ankur has quit IRC (Quit: Page closed) 2015-03-16T16:17:10 *** ketul_ has quit IRC (Ping timeout: 246 seconds) 2015-03-16T16:25:08 *** mkhoory has quit IRC (Read error: Connection reset by peer) 2015-03-16T16:25:54 *** mkhoory has joined #rtems 2015-03-16T16:33:32 * DrJoel is out of here 2015-03-16T16:33:33 *** DrJoel has quit IRC (Quit: He who laughs last, thinks slowest) 2015-03-16T16:52:29 *** ragunath has quit IRC () 2015-03-16T17:27:34 *** andrei has quit IRC (Remote host closed the connection) 2015-03-16T17:41:26 *** kiwichris has joined #rtems 2015-03-16T18:07:06 *** kiwichris has quit IRC (Read error: Connection reset by peer) 2015-03-16T18:07:09 *** kiwichri_ has joined #rtems 2015-03-16T20:26:23 [trac] #2271/enhancement (accepted) updated by chrisj (Replying to [comment:7 gedare]: > Can you please scale-down the sizes of the images ...) http://devel.rtems.org/ticket/2271 2015-03-16T21:12:25 [git] b117be8 by Chris Johns: [gdb/python: Update the support to a recent RTEMS.] http://devel.rtems.org/changeset/b117be8/rtems-tools 2015-03-16T21:26:48 [trac] Developer/Tools/RTEMS_GDB_Python created http://devel.rtems.org/wiki/Developer/Tools/RTEMS_GDB_Python 2015-03-16T22:00:10 [trac] #2308/enhancement (v:4.11) created by chrisj (Change uniprocessor INIT task mode to preempt.) http://devel.rtems.org/ticket/2308 2015-03-17T00:41:11 *** dr__house is now known as dr__house|away 2015-03-17T00:48:47 *** monstr has joined #rtems 2015-03-17T00:58:20 *** zoso has joined #rtems 2015-03-17T01:36:19 *** andrei_ has joined #rtems 2015-03-17T01:53:27 *** starprince has joined #rtems 2015-03-17T01:55:31 hi 2015-03-17T01:56:27 hello 2015-03-17T01:56:56 *** starprince has quit IRC (Client Quit) 2015-03-17T02:31:17 [git] 3e3684b by Sebastian Huber: [score: Ensure properties of CPU_Uint32ptr] http://devel.rtems.org/changeset/3e3684b/rtems 2015-03-17T02:55:24 [git] 093bfde by Sebastian Huber: [bsp/qoriq: Fix warning] http://devel.rtems.org/changeset/093bfde/rtems 2015-03-17T02:55:24 [git] 8e10f62 by Sebastian Huber: [bsp/qoriq: Add register blocks for T variants] http://devel.rtems.org/changeset/8e10f62/rtems 2015-03-17T03:01:13 *** kiwichri_ has quit IRC () 2015-03-17T03:14:48 *** Hesham has joined #rtems 2015-03-17T03:15:22 *** zoso has quit IRC (Ping timeout: 240 seconds) 2015-03-17T03:15:55 *** zoso has joined #rtems 2015-03-17T03:42:29 *** kiwichris has joined #rtems 2015-03-17T04:34:02 [git] 8c7eb00 by Sebastian Huber: [testsupport: Add worker setup handler Add rtems_test_parallel_get_task_id().] http://devel.rtems.org/changeset/8c7eb00/rtems 2015-03-17T05:01:08 *** bapu has joined #rtems 2015-03-17T05:06:59 *** bapu has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-17T05:07:24 *** bapu has joined #rtems 2015-03-17T05:15:37 *** bapu has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-17T05:16:16 *** bapu has joined #rtems 2015-03-17T05:29:16 [git] ac8339a by Sebastian Huber: [bsp/mpc55xx: Fix flash settings] http://devel.rtems.org/changeset/ac8339a/rtems 2015-03-17T05:38:02 [git] dfc5994 by Chris Johns: [elftoolchain: Add support for Cygwin.] http://devel.rtems.org/changeset/dfc5994/rtems-tools 2015-03-17T06:45:17 [trac] #2309/defect (v:4.11) created by perezgc (RSB get stuck building: expat-2.1.0-x86_64-w64-mingw32-1) http://devel.rtems.org/ticket/2309 2015-03-17T06:48:44 [trac] config.log (config.log) attached to #2309 by perezgc http://devel.rtems.org/ticket/2309 2015-03-17T07:10:03 *** zoso has quit IRC (Quit: Leaving) 2015-03-17T07:14:41 *** antgreen has quit IRC (Ping timeout: 250 seconds) 2015-03-17T07:28:29 *** bapu has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-17T07:29:07 *** bapu has joined #rtems 2015-03-17T07:34:43 *** bapu has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-17T07:35:20 *** bapu has joined #rtems 2015-03-17T07:55:07 *** bapu has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-17T07:59:45 *** bapu has joined #rtems 2015-03-17T08:11:58 *** andrei_ has quit IRC (Ping timeout: 246 seconds) 2015-03-17T08:21:58 *** ragunath has joined #rtems 2015-03-17T08:23:15 *** bapu has quit IRC (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client) 2015-03-17T09:18:12 *** monstr has quit IRC (Ping timeout: 245 seconds) 2015-03-17T09:26:20 [trac] #2310/defect (v:4.11) created by sebastian.huber (Race condition in _Thread_Change_priority()) http://devel.rtems.org/ticket/2310 2015-03-17T09:29:00 [trac] #2310/defect (new) updated by sebastian.huber (empty comment) http://devel.rtems.org/ticket/2310 2015-03-17T09:31:17 [trac] GSoC/2014 edited by Sched (empty comment) http://devel.rtems.org/wiki/GSoC/2014 2015-03-17T09:34:13 *** monstr has joined #rtems 2015-03-17T09:37:43 *** ragunath has quit IRC (Ping timeout: 246 seconds) 2015-03-17T09:41:43 [trac] #2310/defect (new) updated by joel.sherrill (Off the top of my head, I don't see a more expedient solution. It does negatively ...) http://devel.rtems.org/ticket/2310 2015-03-17T09:48:13 *** DrJoel has joined #rtems 2015-03-17T09:48:30 *** DrJoel has joined #rtems 2015-03-17T09:48:30 *** ChanServ sets mode: +o DrJoel 2015-03-17T10:00:14 [trac] GSoC/2015 edited by gtament (empty comment) http://devel.rtems.org/wiki/GSoC/2015 2015-03-17T10:03:56 *** andrei__ has joined #rtems 2015-03-17T10:04:07 *** andrei__ is now known as andrei_ 2015-03-17T10:04:19 *** andrei_ has joined #rtems 2015-03-17T10:05:31 [git] 4b7df86 by Joel Sherrill: [cpukit/libdl/rtl-obj-comp.c: Use correct printf() specification] http://devel.rtems.org/changeset/4b7df86/rtems 2015-03-17T10:05:31 [git] 0e6f915 by Joel Sherrill: [testsuites/samples/minimum/init.c: Notepads no longer need to be disabled They are disabled by default as they are deprecated.] http://devel.rtems.org/changeset/0e6f915/rtems 2015-03-17T10:05:31