2013-08-19T01:53:02 *** sebhub has joined #rtems 2013-08-19T01:54:38 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T01:55:06 *** sebhub has joined #rtems 2013-08-19T01:58:01 *** sebhub has joined #rtems 2013-08-19T02:03:35 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:04:15 *** sebhub has joined #rtems 2013-08-19T02:11:50 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:15:11 *** sebhub has joined #rtems 2013-08-19T02:34:08 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:36:01 *** sebhub has joined #rtems 2013-08-19T02:40:09 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:41:59 *** sebhub has joined #rtems 2013-08-19T02:44:55 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:45:49 *** cafi has joined #rtems 2013-08-19T02:46:06 hi 2013-08-19T02:46:48 *** sebhub has joined #rtems 2013-08-19T02:47:44 i have an issue when I'm compiling rtems for arm : rtems_stm32f4/cpukit/score/cpu/arm/armv7m-context-restore.c:29:3: error: impossible constraint in ‘asm’ 2013-08-19T02:48:14 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:48:57 has someone a good idea how to solve this ? I use gcc 4.8.0 newlib 1.20. 0 2013-08-19T02:50:07 *** sebhub has joined #rtems 2013-08-19T02:53:18 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T02:59:59 *** sebhub has joined #rtems 2013-08-19T03:00:03 good morning 2013-08-19T04:04:35 *** sebhub has quit IRC (Ping timeout: 245 seconds) 2013-08-19T04:11:58 *** arvind_khadri has joined #rtems 2013-08-19T04:53:48 *** sebhub has joined #rtems 2013-08-19T05:06:35 hello sebhub 2013-08-19T05:12:35 *** tictoc has joined #rtems 2013-08-19T06:18:22 *** cafi has quit IRC (Quit: Page closed) 2013-08-19T07:25:34 *** arvind_khadri has quit IRC (Remote host closed the connection) 2013-08-19T07:25:52 *** cafi has joined #rtems 2013-08-19T07:27:06 *** arvind_khadri has joined #rtems 2013-08-19T07:43:16 sebhub, hi 2013-08-19T08:06:06 *** MegaAlex|away has quit IRC (Ping timeout: 276 seconds) 2013-08-19T08:07:26 *** MegaAlex|away has joined #rtems 2013-08-19T08:08:54 hi chris 2013-08-19T08:21:06 hello 2013-08-19T08:26:14 does anyone know how solve "error: impossible constrants asm " in cpukit/score/cpu/arm/armv7m-context-restore? 2013-08-19T08:36:59 *** MegaAlex|away has quit IRC (Ping timeout: 268 seconds) 2013-08-19T08:38:27 *** MegaAlex|away has joined #rtems 2013-08-19T08:41:23 this looks like a tool chain problem 2013-08-19T08:41:38 which tool and rtems version to you use? 2013-08-19T08:55:32 *** arvind_khadri has quit IRC (Remote host closed the connection) 2013-08-19T08:57:39 rtmes 4.11 2013-08-19T08:58:41 tools newlib 1.20.0 gcc 4.8.0 bin utils 2.23.2 gdb 7.6 2013-08-19T09:04:18 I tried even with gcc 4.8.1 RC-20130527 and same result 2013-08-19T09:17:41 *** MegaAlex|away has quit IRC (Ping timeout: 268 seconds) 2013-08-19T09:22:17 which BSP do you use and what is your configure command line 2013-08-19T09:22:44 please use the RSB to build the tools 2013-08-19T09:38:12 *** DrJoel has joined #rtems 2013-08-19T09:38:12 *** DrJoel has joined #rtems 2013-08-19T09:38:12 *** ChanServ sets mode: +o DrJoel 2013-08-19T09:38:27 sebhub: Do you have a minute for a quick question/ 2013-08-19T09:40:30 hi joel 2013-08-19T09:40:50 yes, i have time 2013-08-19T09:42:23 hey.. I am finishing up slides for next week at the Embedded OS Workshop in Toulouse. Is the new SMP Priority Scheduler you have posted using the same ready set structures as the Deterministic Priority Scheduler? 2013-08-19T09:42:40 http://www.sigops-france.fr/Main/EWiLi13 2013-08-19T09:43:08 yes, and it uses several inline functions from schedulerpriorityimpl.h 2013-08-19T09:43:23 so we virtually no copy and paste 2013-08-19T09:44:03 I could tell that there was little cut and paste but the comments were sparse and I couldn't tell what it actually did under the hood. 2013-08-19T09:44:15 Is it fair to say it is the same O(XXX) as the DPS? 2013-08-19T09:44:23 What additional overhead does SMP add? 2013-08-19T09:45:11 on smp you have also the scheduled chain which has a maximum of processor count elements 2013-08-19T09:45:21 *** cafi has quit IRC (Ping timeout: 250 seconds) 2013-08-19T09:45:48 a thread is scheduled if it is ready and has an allocated processor, e.g it is an heir or executing 2013-08-19T09:46:02 scheduled threads are on the scheduled chain 2013-08-19T09:47:02 it is O(1) on SMP since the processor count and the priority count is constant 2013-08-19T09:48:11 it is not O(1) if you use a variable processor or priority count 2013-08-19T09:50:06 the main function of the scheduler is _Scheduler_SMP_Enqueue_ordered() 2013-08-19T09:51:54 OK. Then I can safely claim in good conscious that it is deterministic like the uniprocessor DPS is? 2013-08-19T09:52:21 yes 2013-08-19T09:52:41 Great! I wanted a slide on it but didn't want to risk saying something incorrect. 2013-08-19T09:54:00 is is_transient instead of is_in_the_air ok for you? 2013-08-19T09:54:11 The official name is "Priority SMP Scheduler" 2013-08-19T09:54:52 what is the name of the single processor one? 2013-08-19T09:54:54 I need to read and think a bit. I just got into the office. On top of everythign else going on, I got called in for jury duty. I went in this morning just long enought to get myself excused from serving on a case this week. 2013-08-19T09:55:03 Deterministic Priority Scheduler 2013-08-19T09:55:34 ok, then it should be Deterministic Priority SMP Scheduler 2013-08-19T09:55:49 OK. The slide will reflect that. 2013-08-19T09:57:44 I had a few slides on Application Challenges which listed things like disabling preemption not being good enough, disable ISR not covering all cores, can't assume highest priority task is only thing running, ... Do you have any ideas to add to that? 2013-08-19T09:58:06 task variables don't work 2013-08-19T09:58:08 And I ended with a Research Areas/Open Topics... anything you want to pop up on that 2013-08-19T09:58:20 Task variables.. that's the one that I didn't remember .. 2013-08-19T09:58:26 i write currently a list of topics 2013-08-19T09:58:56 We need to have guidance for those moving applications to SMP. It breaks a number of assumptions. 2013-08-19T09:59:23 yes, i think the new score assertions can help here 2013-08-19T10:00:01 Yes but I also want some documentation so people can realize the issues before they put code on hardware 2013-08-19T10:00:26 one big issue are the rtems chains, they use ISR disable/enable 2013-08-19T10:00:30 Is SMP working on QorIQ 2013-08-19T10:00:45 yes, its one of my development systems 2013-08-19T10:00:52 i use it for benchmarks 2013-08-19T10:00:58 I thought so from the code but better to ask and be sure 2013-08-19T10:01:15 Have you run on a higher CPU than the 1020? 2013-08-19T10:01:26 no 2013-08-19T10:01:37 next year we will have a powerpc with 24 cores 2013-08-19T10:02:15 On FACE, I hear folks talking about the P4080. That seems to be the CPU avionics folks are looking at for next generation systems. 2013-08-19T10:02:18 system and application traceing will be a number one issue 2013-08-19T10:02:19 24 is a LOT 2013-08-19T10:03:10 yes, the QorIQ is targeting this area 2013-08-19T10:04:36 There is a general concern that SMP analysis for flight systems is not completely solved but given enough constraints, it can be analyzed. But it is inevitable that soon, they will only be able to get SMP systems so they might as well face the music. 2013-08-19T10:04:47 They favor PowerPC's currently. 2013-08-19T10:05:56 yes, the free lunch is over 2013-08-19T10:06:50 :) 2013-08-19T10:07:39 It is interesting seeing how the ARINC 653 RTOSes are approaching the world of SMP. Very controlled.. but everything in ARINC 653 is statically defined offline and now they are just extending this idea to scheduling multiple cores. 2013-08-19T10:10:26 Any other plugs for future work you want to throw in the end of this presentation? 2013-08-19T10:10:46 I have 60 slides for a 2 hour keynote including question time. 2013-08-19T10:10:56 when do you need this info? 2013-08-19T10:12:13 In time to get in the slides before Wed or Thursday. This is for http://www.sigops-france.fr/Main/EWiLi13 and it is being held at the same time as the French Real-Time Summer School ETR 2013. So an opportunity. 2013-08-19T10:13:08 If ideas show up late, I can probably hack them in later but at some point I will have to pack and there is the travel time. 2013-08-19T10:13:15 ok, i hope that i have enough time tomorrow 2013-08-19T10:13:55 Doesn't have to be much. This is more of a concluding thing .. what's left to do? To entice anyone interested to pitch in 2013-08-19T10:14:00 i have to leave now, cu 2013-08-19T10:15:06 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-19T11:01:32 *** monstr__ has quit IRC (Remote host closed the connection) 2013-08-19T11:33:19 *** tictoc has quit IRC (Quit: ChatZilla 0.9.90.1 [Firefox 23.0/20130730113002]) 2013-08-19T18:23:14 *** DrJoel has quit IRC (Quit: IceChat - Keeping PC's cool since 2000) 2013-08-19T20:15:53 *** the9a3eedi has joined #rtems 2013-08-19T23:35:52 *** monstr_ has joined #rtems 2013-08-20T00:45:53 *** kiwichris_ has joined #rtems 2013-08-20T00:48:51 *** kiwichris has quit IRC (Ping timeout: 268 seconds) 2013-08-20T02:05:19 *** sebhub has joined #rtems 2013-08-20T02:08:46 good morning 2013-08-20T02:10:22 *** arvind_khadri has joined #rtems 2013-08-20T02:12:08 *** tictoc has joined #rtems 2013-08-20T02:15:09 *** arvind_khadri has quit IRC (Ping timeout: 264 seconds) 2013-08-20T02:18:54 *** arvind_khadri has joined #rtems 2013-08-20T04:11:33 Hey, I have a question to real-time programming in general when using asynchronous POSIX or RTEMS signals. How would you usually exchange data in a safe way between the signal handler and the target thread (or ASR / task respectively) without atomic operations? 2013-08-20T04:14:43 i never use signals 2013-08-20T04:15:39 for posix i would look at the posix documentation or a standard handbook 2013-08-20T04:15:55 in rtems i would use events, semaphores, message queues 2013-08-20T04:16:13 posix documentation doesn't have anything useful 2013-08-20T04:17:44 I was trying to figure out if there was any advantage in using async signal handlers over synchronous signals / events or message queues 2013-08-20T04:18:49 but thanks, I will likely resort to only using these mechanisms as well 2013-08-20T04:19:49 you can use sig_atomic_t for flag variables 2013-08-20T04:22:19 oh ok, I wasn't aware 2013-08-20T04:27:01 you cannot do read-modify-write, but you have atomic store and load 2013-08-20T04:27:49 ok that would have been my next question :) 2013-08-20T04:30:49 C11 supports now atomic operations 2013-08-20T04:31:44 on rtems some work is required to support them 2013-08-20T04:32:45 yes I read the discussion on the mailing list regarding the GSOC project from last year, however as I could gather from the atomic.h it is only enabled for smp mode atm 2013-08-20T04:48:31 *** phipse has joined #rtems 2013-08-20T04:48:38 hi 2013-08-20T04:56:56 sebhub: regarding the i386 patch: the macros in question are i386 interrupt macros, _CPU_ISR_Set_level and _CPU_Fatal_halt; I guess these are macros for performance reasons, do you think it makes sense to replace them with functions? 2013-08-20T05:02:14 isr disable/enable is critical, but the _CPU_ISR_Set_level is not performance critical 2013-08-20T05:04:08 fatal halt should also be not performance critical, so I replace the two and think about the isr stuff 2013-08-20T05:06:32 you have to replace _CPU_ISR_Disable/Enable() ? 2013-08-20T05:10:52 yes, a virtual machine is not allowed to change the hardware interrupt flags. 2013-08-20T05:12:44 I could transform them to functions and use inline? 2013-08-20T05:19:48 inline doesn't help, since you need the definition in the header file 2013-08-20T05:20:08 who is your mentor? 2013-08-20T05:21:46 joel, claudio, julien 2013-08-20T05:22:23 and Gedare reviewed the first version of the patch 2013-08-20T05:24:59 ok, i personally don't care about i386 performance; using functions for ISR disable/enable has some problems: 1. it makes a lot of leaf functions non-leaf functions, 2. there is the function call overhead 2013-08-20T05:26:58 we have some options: 1. use functions for ISR disable/enable and life with the performance penalty, or 2. use a new configure option and conditional compilation, or 3. use a new GCC i386 multilib with a special built in define, or 4. use a new architecture name 2013-08-20T05:31:27 I have chosen no.4 for L4RTEMS but I think, I would dismiss no.3, as this is just another site to care for, so no.2 would be the best I think. 2013-08-20T05:31:56 But I don't know how much work it is to get the new option in the configure scripts. 2013-08-20T05:32:25 technically its not a big deal, but new configure options are a political issue 2013-08-20T05:32:54 i would notify your mentors about this problem 2013-08-20T05:35:03 We have these USE_COM1_AS_CONSOLE and BSP_PRESS_KEY_FOR_RESET options, if we just add USE_VIRTUAL_I386 and default it to 0, then nobody is harmed? What's the difference of caps options anyway? 2013-08-20T05:36:07 for the cpukit you need a different configure support 2013-08-20T05:36:16 have to leave now for some hours 2013-08-20T05:36:31 ok, thanks 2013-08-20T05:43:10 *** sebhub has quit IRC (Ping timeout: 246 seconds) 2013-08-20T08:27:02 *** phipse has quit IRC (Quit: leaving) 2013-08-20T08:36:58 *** sebhub has joined #rtems 2013-08-20T10:06:02 *** monstr_ has quit IRC (Remote host closed the connection) 2013-08-20T10:25:26 *** arvind_khadri has quit IRC (Ping timeout: 240 seconds) 2013-08-20T10:30:57 *** sebhub has quit IRC (Remote host closed the connection) 2013-08-20T10:44:05 *** MegaAlex|away has joined #rtems 2013-08-20T10:52:18 *** tictoc has quit IRC (Quit: ChatZilla 0.9.90.1 [Firefox 23.0/20130730113002]) 2013-08-20T12:58:14 *** antgreen has quit IRC (Ping timeout: 240 seconds) 2013-08-20T15:51:02 *** antgreen has joined #rtems 2013-08-20T23:34:30 *** monstr__ has joined #rtems 2013-08-21T01:04:27 *** Shubham has joined #rtems 2013-08-21T01:23:49 *** freenix has joined #rtems 2013-08-21T01:46:54 *** sebhub has joined #rtems 2013-08-21T01:47:05 good morning 2013-08-21T02:48:08 *** arvind_khadri has joined #rtems 2013-08-21T02:53:45 *** Shubham has quit IRC (Ping timeout: 256 seconds) 2013-08-21T03:37:29 *** beng-nl has quit IRC (Ping timeout: 241 seconds) 2013-08-21T04:05:22 *** beng-nl has joined #rtems 2013-08-21T04:14:26 *** Shubham has joined #rtems 2013-08-21T04:48:20 *** sebhub has quit IRC (Ping timeout: 245 seconds) 2013-08-21T07:17:43 *** sebhub has joined #rtems 2013-08-21T07:31:15 *** sebhub has quit IRC (Ping timeout: 260 seconds) 2013-08-21T07:53:21 *** gedare has joined #rtems 2013-08-21T08:47:27 *** tictoc has joined #rtems 2013-08-21T08:51:30 *** Shubham has quit IRC (Ping timeout: 264 seconds) 2013-08-21T09:39:44 *** freenix has quit IRC (Remote host closed the connection) 2013-08-21T10:05:56 *** arvind_khadri has quit IRC (Remote host closed the connection) 2013-08-21T10:25:14 *** gedare has quit IRC (Quit: Leaving) 2013-08-21T10:35:15 *** antgreen has quit IRC (Read error: Operation timed out) 2013-08-21T10:49:53 *** monstr__ has quit IRC (Remote host closed the connection) 2013-08-21T11:06:27 *** tictoc has quit IRC (Quit: ChatZilla 0.9.90.1 [Firefox 23.0.1/20130814063812]) 2013-08-21T11:35:04 *** antgreen has joined #rtems 2013-08-21T11:50:13 *** antgreen has quit IRC (Ping timeout: 246 seconds) 2013-08-21T12:12:47 *** S_Somani has joined #rtems 2013-08-21T13:47:26 *** gedare has joined #rtems 2013-08-21T14:02:54 *** S_Somani has left #rtems 2013-08-21T14:05:23 *** gedare has quit IRC (Remote host closed the connection) 2013-08-21T14:05:27 *** Sebastian_Huber has joined #rtems 2013-08-21T14:36:07 *** MegaAlex|away has quit IRC (Ping timeout: 246 seconds) 2013-08-21T14:37:07 *** MegaAlex|away has joined #rtems 2013-08-21T15:03:36 *** antgreen has joined #rtems 2013-08-21T15:19:05 *** Sebastian_Huber has quit IRC (Quit: ChatZilla 0.9.90.1 [Firefox 22.0/20130618035212]) 2013-08-21T15:44:18 *** MegaAlex|away has quit IRC (Ping timeout: 264 seconds) 2013-08-21T15:45:07 *** MegaAlex|away has joined #rtems 2013-08-21T20:01:38 *** freenix has joined #rtems 2013-08-21T23:20:58 *** monstr__ has joined #rtems 2013-08-22T00:58:13 *** kiwichris_ has quit IRC (Quit: This computer has gone to sleep) 2013-08-22T01:20:31 *** kiwichris has joined #rtems 2013-08-22T01:40:53 *** kiwichris has quit IRC (Quit: This computer has gone to sleep) 2013-08-22T01:44:46 *** sebhub has joined #rtems 2013-08-22T01:45:01 good morning 2013-08-22T02:31:40 *** kiwichris has joined #rtems 2013-08-22T02:36:33 morning 2013-08-22T02:43:39 sebhub, hi 2013-08-22T02:44:27 hi chris 2013-08-22T02:48:32 I have found the leak in confdefs.h 2013-08-22T02:48:51 And I have a patch 2013-08-22T02:49:38 hello's workspace was 214 bytes too small with the pthread_once patch 2013-08-22T02:50:10 its good that the size estimate is so close 2013-08-22T02:50:46 Yeah it is. It suprised me how tight the fit ir 2013-08-22T02:50:48 is 2013-08-22T02:51:15 hello is back again 2013-08-22T02:59:03 sebhub, I am not sure the mmu setup and the interrupt mapping stuff is working 2013-08-22T02:59:33 if I have my hack when running from gdb it works and so I have tried booting from the sd card and it fails 2013-08-22T02:59:54 When I use Ric's BSP set up it was working. 2013-08-22T03:00:09 I need to head off for the evening; been a long day 2013-08-22T03:00:17 cu 2013-08-22T03:12:55 *** arvind_khadri has joined #rtems 2013-08-22T04:18:21 *** the9a3eedi has quit IRC (Ping timeout: 268 seconds) 2013-08-22T05:49:40 *** freenix has quit IRC (Quit: Leaving) 2013-08-22T06:14:06 *** beng-nl has quit IRC (Ping timeout: 256 seconds) 2013-08-22T06:35:27 *** beng-nl has joined #rtems 2013-08-22T06:42:09 *** arvind_khadri has quit IRC (Ping timeout: 276 seconds) 2013-08-22T06:54:44 *** arvind_khadri has joined #rtems 2013-08-22T06:56:45 *** arvind_khadri has joined #rtems 2013-08-22T06:57:48 *** arvind_khadri has joined #rtems 2013-08-22T06:58:58 *** arvind_khadri has joined #rtems 2013-08-22T07:02:09 *** arvind_khadri has joined #rtems 2013-08-22T07:03:41 *** arvind_khadri has quit IRC (Max SendQ exceeded) 2013-08-22T07:05:32 *** arvind_khadri has joined #rtems 2013-08-22T07:06:47 *** arvind_khadri has joined #rtems 2013-08-22T07:11:30 *** arvind_khadri has quit IRC (Ping timeout: 256 seconds) 2013-08-22T07:14:16 *** MegaAlex|away has quit IRC (Ping timeout: 264 seconds) 2013-08-22T07:15:21 *** MegaAlex|away has joined #rtems 2013-08-22T07:20:16 *** MegaAlex|away has quit IRC (Ping timeout: 268 seconds) 2013-08-22T07:20:52 *** MegaAlex|away has joined #rtems 2013-08-22T07:23:40 *** sebhub has quit IRC (Read error: Operation timed out) 2013-08-22T07:24:24 *** sebhub has joined #rtems 2013-08-22T07:24:28 *** arvind_khadri has joined #rtems 2013-08-22T07:27:42 *** MegaAlex|away has quit IRC (Ping timeout: 264 seconds) 2013-08-22T07:29:54 *** MegaAlex|away has joined #rtems 2013-08-22T07:33:03 *** antgreen has quit IRC (Ping timeout: 245 seconds) 2013-08-22T08:39:20 *** antgreen has joined #rtems 2013-08-22T09:29:02 *** arvind_khadri has quit IRC (Ping timeout: 240 seconds) 2013-08-22T09:45:46 *** arvind_khadri has joined #rtems 2013-08-22T09:59:38 *** sebhub has quit IRC (Ping timeout: 256 seconds) 2013-08-22T10:01:30 *** arvind_khadri has quit IRC (Ping timeout: 256 seconds) 2013-08-22T10:04:25 *** S_Somani has joined #rtems 2013-08-22T11:07:21 *** S_Somani has left #rtems 2013-08-22T11:08:32 *** monstr__ has quit IRC (Read error: Operation timed out) 2013-08-22T11:27:42 *** antgreen has quit IRC (Ping timeout: 264 seconds) 2013-08-22T11:41:43 *** antgreen has joined #rtems 2013-08-22T12:30:47 *** S_Somani has joined #rtems 2013-08-22T12:47:55 *** antgreen has quit IRC (Ping timeout: 264 seconds) 2013-08-22T13:03:47 *** antgreen has joined #rtems 2013-08-22T13:49:39 *** S_Somani has left #rtems 2013-08-22T16:16:35 *** antgreen has quit IRC (Quit: Leaving) 2013-08-22T18:57:27 *** freenix has joined #rtems 2013-08-22T19:03:01 *** the9a3eedi has joined #rtems 2013-08-23T00:10:57 *** monstr__ has joined #rtems 2013-08-23T00:49:54 *** MegaAlex|away has quit IRC (Ping timeout: 264 seconds) 2013-08-23T00:51:32 *** MegaAlex|away has joined #rtems 2013-08-23T01:28:20 *** sebhub has joined #rtems 2013-08-23T01:33:01 good morning 2013-08-23T02:12:23 dhananjay, around ? 2013-08-23T02:15:19 sebhub, sent you a few patches that add the wfe instruction to some halt or wait loops. This places the processor in a low power state and so nice to use if we can. 2013-08-23T02:16:13 kiwichris: Was on a znc session so missed it. 2013-08-23T02:16:17 Back now 2013-08-23T02:16:36 ah ok; just sent a respose. 2013-08-23T02:16:54 How many subcommands do you have left to port ? 2013-08-23T02:17:37 I have all the index commands. 2013-08-23T02:17:46 Working on time of the day now 2013-08-23T02:18:02 (Just started) 2013-08-23T02:18:09 https://github.com/dbalan/rtems-gdb/wiki/Subcommands 2013-08-23T02:18:18 What I am after is a list of the commands you need to complete and then an estimate of the time to complete 2013-08-23T02:18:31 okay 2013-08-23T02:19:10 Any completed ? 2013-08-23T02:20:21 You can put this in an email to. 2013-08-23T02:20:40 I will compile a list and mail it 2013-08-23T02:21:14 Fantastic. 2013-08-23T02:22:00 kiwichris: I am on IST btw. 2013-08-23T02:22:10 GMT +0530 2013-08-23T02:23:01 Ok so 1pm. You should be able to get here most weekdays during your morning. 2013-08-23T02:23:13 Please feel free to email or catch me here 2013-08-23T02:25:41 I will. 2013-08-23T02:49:26 *** freenix has quit IRC (Ping timeout: 264 seconds) 2013-08-23T02:53:26 *** freenix has joined #rtems 2013-08-23T03:02:44 freenix, hi 2013-08-23T03:03:02 hi,kiwichris 2013-08-23T03:03:23 Did you use the rtld-file.cpp archive code or elftools ? 2013-08-23T03:04:00 for rtl-host? 2013-08-23T03:04:07 yeah 2013-08-23T03:04:17 no 2013-08-23T03:04:31 I have not checked the patch as you can tell :) 2013-08-23T03:04:49 what did you use ? 2013-08-23T03:05:12 sorry. i did not find rtld-file.cpp . I used the archive writer in rld-outputter.cpp 2013-08-23T03:06:22 Ah ok that is just me not remembering where the code is 2013-08-23T03:06:25 Nice 2013-08-23T03:07:23 I reused the archive function in rld-outputter.cpp to create the ra file. 2013-08-23T03:07:52 I am rewritting the patch. 2013-08-23T03:08:12 Ok. 2013-08-23T03:08:18 The archive is in the files namespace 2013-08-23T03:08:48 yeah. 2013-08-23T03:09:22 And that is working ok ? 2013-08-23T03:09:58 yeah. Just for short name of object files, because I have not implement long name support. 2013-08-23T03:10:24 Now I am using the binutils implements to update the ra file. 2013-08-23T03:10:46 use a tmp file 2013-08-23T03:11:12 the files::archive::create supports extended file names. 2013-08-23T03:12:03 yeah. I found it. But now if I want to append or replace file in the created ra file, I have to update the extended file names table. 2013-08-23T03:12:31 Just as what I mailed you, how to insert the long names in the middle of the ra file. 2013-08-23T03:13:04 I want to use what binutils implemented, using a tmp file, using the mkstemp/mktemp syscall. 2013-08-23T03:13:29 And replace the old ra file with the new created tmp file. 2013-08-23T03:13:57 Why create a new archive in a tmp dir and then move to the new location ? 2013-08-23T03:14:18 It should be easy. 2013-08-23T03:14:30 (hehe always is when it is not yourself) 2013-08-23T03:14:44 Just create it in the dir where the old ra locates 2013-08-23T03:15:20 Check the liberty dir for this the mktemp stuff. 2013-08-23T03:15:24 yeah. easy.:) 2013-08-23T03:15:28 ok. 2013-08-23T03:15:43 It has a portable mktmp implementation 2013-08-23T03:15:53 ok. 2013-08-23T03:16:01 It is used in the execute classes to run processes and catch the output 2013-08-23T03:16:26 include the code of mktmp in rtl-host? 2013-08-23T03:17:49 No, it is present already. Look in libibery/make-temp-file.c 2013-08-23T03:18:24 Thanks. see it. 2013-08-23T03:20:46 If you open the existing archive you will get a list of objects. If you ammend the list to be your new list you can create a new archive 2013-08-23T03:21:07 There should be no need to pull the archive apart and put together again 2013-08-23T03:23:11 yeah. 2013-08-23T03:25:43 my current implementation is first collect the objects in the ra file, then iterate the objects in `dependents`(which is a set of objects to generate rap file) to see which is from an archive file and put is into the object list of ra file. 2013-08-23T03:27:27 The dependence should be handled for you. 2013-08-23T03:27:45 Oh I think see what you mean 2013-08-23T03:29:15 yeah. the dependence was handled, and i just use it as an input parameter 2013-08-23T03:33:32 Great 2013-08-23T03:33:56 Is the convert performance from ELF to RAP ok ? 2013-08-23T03:35:11 performance? the time cosumed? 2013-08-23T03:36:13 yeah 2013-08-23T03:36:29 Have you convertered libc for example ? 2013-08-23T03:36:51 if not pass --runtime-lib to rtems-ld, all is the same as before. 2013-08-23T03:37:12 you mean converting libc.a to a single RAP file? 2013-08-23T03:37:45 all the objects in libc.a will be merged into a single RAP file? 2013-08-23T03:38:25 no libc.a -> libc.ra 2013-08-23T03:38:49 Did we discuss a separate tool to do this ? 2013-08-23T03:40:44 Yeah. But if libc.ra is an archive of elf object files, is there a need to convert libc.a to libc.ra? 2013-08-23T03:42:30 I originally want to convert libc.a to a single RAP file. I may misunderstand your mean. 2013-08-23T03:42:43 We get libc.a from the tools. There maybe a case where a user wants to put the whole of libc.a on a target and a libc.ra maybe smaller. 2013-08-23T03:43:51 A single RAP file is all merged together while a libc.ra is all the files available in the RAP format where at runtime you can load what ever ones you wish. 2013-08-23T03:44:04 If a single RAP file you need to memory to load any bit 2013-08-23T03:44:13 to -> the 2013-08-23T03:45:09 ie a RAP file is a single load and libc will use a lot of memory in the target 2013-08-23T03:45:15 libc.a -> libc.ra means to convert each elf object files of libc.a to rap files and merged the rap files into libc.ra? 2013-08-23T03:45:23 yes 2013-08-23T03:45:47 libc.a:printf.o -> libc.ra:printf.rap 2013-08-23T03:45:51 etc etc 2013-08-23T03:45:55 Got it. 2013-08-23T03:45:59 excellent 2013-08-23T03:46:41 *** sebhub has quit IRC (Ping timeout: 245 seconds) 2013-08-23T03:46:45 Then the runtime-lib option passed to rtems-ld, the librt.ra should also be archiver of rap files? 2013-08-23T03:46:50 The linker variations allows a user to create a single RAP image with everything. In this case no extra image would be loaded. 2013-08-23T03:46:52 *** sebhub2 has joined #rtems 2013-08-23T03:47:15 No rtems-ld works with ELF files but creates RAP files. 2013-08-23T03:47:32 I do not think we have the ability to link against RAP files. 2013-08-23T03:47:41 can we do that ? 2013-08-23T03:47:49 or could be do that ? 2013-08-23T03:47:54 be -> we 2013-08-23T03:48:35 The 2nd variation is rtems-ld create an app image and also a runtime-lib 2013-08-23T03:48:53 this is the case of more than app being loaded with a shared runtime-lib 2013-08-23T03:49:16 there is a case where a user converts all runtime libs by hand and loads the target 2013-08-23T03:49:44 ok. 2013-08-23T03:50:00 Some targets have plenty of storage and some do not. 2013-08-23T03:50:37 The convert everything option might useful when developing application while production is the rtems-ld version 2013-08-23T03:50:40 The new tool is to create a archive of rap files. The new option to rtems-ld is to collect elf object files and place them into a archive of elf object files? 2013-08-23T03:50:49 yes 2013-08-23T03:51:01 But both is with the .ra suffix. 2013-08-23T03:51:03 same c++ code behind being used in different ways 2013-08-23T03:51:06 yes 2013-08-23T03:51:12 ok. I Got it. 2013-08-23T03:51:15 nice 2013-08-23T03:51:28 this is rather neat stuff 2013-08-23T03:51:53 yeah. 2013-08-23T03:52:26 and waf is nice the way is cleanly handles the separate tools from the same code base 2013-08-23T03:53:21 You know we have a large user manual to put together on how to use this stuff to create applications :) 2013-08-23T03:54:17 yeah. 2013-08-23T03:55:15 I have not implemented the new tool. I misunderstool your mean, and just converted libc.a to a single rap file. After the new option part, I'll implement the new tool. 2013-08-23T03:57:43 misunderstood 2013-08-23T04:19:46 *** the9a3eedi has quit IRC (Read error: Connection reset by peer) 2013-08-23T06:00:38 *** MegaAlex|away has quit IRC (Ping timeout: 240 seconds) 2013-08-23T06:01:37 *** MegaAlex|away has joined #rtems 2013-08-23T06:41:50 *** MegaAlex|away has quit IRC (Ping timeout: 240 seconds) 2013-08-23T06:44:38 *** MegaAlex|away