2011-03-07T00:11:15 *** lcpfnyvc has joined #rtems 2011-03-07T01:31:22 *** kiwichris has joined #rtems 2011-03-07T01:32:30 *** sebhub has joined #rtems 2011-03-07T01:32:55 sebhub, hi 2011-03-07T01:33:20 good morning 2011-03-07T01:33:49 well, good afternoon or night ;-) 2011-03-07T01:35:08 Evening here. Just cooking the evening meal at the moment 2011-03-07T01:35:32 We cannot check for NULL in the HeapFree because of the region free call. 2011-03-07T01:35:39 We have to move it to the Workspace free 2011-03-07T01:38:04 I also have another little chains related bit of code I wanted to pass by you for comment. 2011-03-07T01:40:54 In the run time load I have a chain of global symbols for the object module. When you remove an object file I need to extract the nodes before I free the memory. 2011-03-07T01:41:35 I could now ever fail in the loading and during the symbol table loading so I want this piece of code: 2011-03-07T01:42:04 for (sym = 0, gsym = obj->global_table; sym < globals; ++sym, ++gsym) 2011-03-07T01:42:04 gsym->node.next = gsym->node.previous = &gsym->node; 2011-03-07T01:42:24 This little node trick means the extact works for a node not linked to a list. 2011-03-07T01:42:40 I was wondering if this is a good addition to the API. 2011-03-07T01:50:52 you call this code during the load? 2011-03-07T01:53:57 we should check for NULL in _Heap_Free() and modify rtems_region_return_segment() accordingly 2011-03-07T02:29:03 *** core-ix has quit IRC 2011-03-07T03:13:02 *** core-ix has joined #rtems 2011-03-07T04:06:43 *** arvind_khadri has quit IRC 2011-03-07T04:25:09 *** arvind_khadri has joined #rtems 2011-03-07T04:30:06 *** core-ix has quit IRC 2011-03-07T04:33:14 *** core-ix has joined #rtems 2011-03-07T04:51:50 *** arvind_khadri has quit IRC 2011-03-07T05:08:39 *** arvind_khadri has joined #rtems 2011-03-07T05:08:39 *** arvind_khadri has joined #rtems 2011-03-07T05:15:20 *** arvind_khadri has quit IRC 2011-03-07T05:15:58 *** arvind_khadri has joined #rtems 2011-03-07T05:22:02 *** arvind_khadri has quit IRC 2011-03-07T05:54:32 *** mmu_man has joined #rtems 2011-03-07T06:02:13 *** arvind_khadri has joined #rtems 2011-03-07T06:02:13 *** arvind_khadri has joined #rtems 2011-03-07T06:23:53 *** antgreen has joined #rtems 2011-03-07T06:57:37 *** arvind_khadri has quit IRC 2011-03-07T07:25:16 *** th_d has joined #rtems 2011-03-07T08:24:44 *** lcpfnyvc has quit IRC 2011-03-07T08:31:13 *** zwj has joined #rtems 2011-03-07T08:45:17 *** madrazr has joined #rtems 2011-03-07T09:22:37 *** dr__house has joined #rtems 2011-03-07T09:22:37 *** dr__house has joined #rtems 2011-03-07T09:28:02 *** antgreen` has joined #rtems 2011-03-07T09:30:08 *** antgreen has quit IRC 2011-03-07T09:31:26 *** antgreen` has quit IRC 2011-03-07T09:31:42 *** antgreen` has joined #rtems 2011-03-07T09:50:39 *** bswimley has joined #rtems 2011-03-07T10:09:49 *** peerst has joined #rtems 2011-03-07T10:09:56 hi 2011-03-07T10:12:11 hi 2011-03-07T10:16:41 *** th_d has quit IRC 2011-03-07T10:26:05 *** sebhub has quit IRC 2011-03-07T10:30:30 *** zwj has quit IRC 2011-03-07T11:20:33 *** dr__house is now known as vattam 2011-03-07T11:20:49 *** vattam is now known as dr__house 2011-03-07T11:31:42 *** dr__house is now known as Purple_ 2011-03-07T11:31:49 *** Purple_ is now known as dr__house 2011-03-07T12:33:46 *** DrJoel has joined #rtems 2011-03-07T12:33:46 *** ChanServ sets mode: +o DrJoel 2011-03-07T13:05:19 *** dr__house has quit IRC 2011-03-07T13:09:41 *** methril has joined #rtems 2011-03-07T13:16:11 *** kristianpaul has quit IRC 2011-03-07T13:18:09 kiwichris: Are you awake? 2011-03-07T13:30:00 *** dr__house has joined #rtems 2011-03-07T13:30:00 *** dr__house has joined #rtems 2011-03-07T13:41:44 *** methril has quit IRC 2011-03-07T13:52:54 *** madrazr has quit IRC 2011-03-07T13:54:05 *** methril has joined #rtems 2011-03-07T13:54:40 *** madrazr has joined #rtems 2011-03-07T13:57:39 *** antgreen` has quit IRC 2011-03-07T14:36:09 *** madrazr has quit IRC 2011-03-07T15:09:38 *** dr__house has quit IRC 2011-03-07T15:56:50 Hi Joel 2011-03-07T15:58:08 hey chris... 2011-03-07T16:06:03 hi 2011-03-07T16:11:58 *** kiwichris has quit IRC 2011-03-07T16:15:35 hey Fallenou 2011-03-07T16:16:49 nothing personal anyone.. just tired and heading home 2011-03-07T16:16:49 *** DrJoel has quit IRC 2011-03-07T16:24:05 *** projectgus has joined #rtems 2011-03-07T16:27:05 *** dr__house has joined #rtems 2011-03-07T16:27:05 *** dr__house has joined #rtems 2011-03-07T16:30:06 *** antgreen has joined #rtems 2011-03-07T16:32:24 *** antgreen has quit IRC 2011-03-07T16:37:27 *** bswimley has quit IRC 2011-03-07T17:14:37 *** methril has quit IRC 2011-03-07T17:28:12 *** kiwichris has joined #rtems 2011-03-07T17:29:39 *** dr__house has quit IRC 2011-03-07T17:44:20 *** dr__house has joined #rtems 2011-03-07T17:45:51 *** mmu_man has quit IRC 2011-03-07T20:30:32 *** kristianpaul has joined #rtems 2011-03-07T20:30:32 *** kristianpaul has joined #rtems 2011-03-07T20:33:12 *** kristianpaul has quit IRC 2011-03-07T20:33:33 *** kristianpaul has joined #rtems 2011-03-07T20:33:33 *** kristianpaul has joined #rtems 2011-03-07T21:03:36 *** dr__house has quit IRC 2011-03-07T21:50:15 *** kiwichris has quit IRC 2011-03-07T22:27:14 *** kiwichris has joined #rtems 2011-03-07T22:41:24 *** kiwichris has quit IRC 2011-03-07T22:46:40 *** kiwichris has joined #rtems 2011-03-07T23:49:59 *** mmu_man has joined #rtems 2011-03-08T00:20:06 *** projectgus has quit IRC 2011-03-08T00:43:02 *** mmu_man has quit IRC 2011-03-08T01:34:18 *** sebhub has joined #rtems 2011-03-08T01:34:32 good morning 2011-03-08T03:25:19 sebhub, hi 2011-03-08T03:27:17 hi 2011-03-08T03:28:46 Ok to chat NIOS ? 2011-03-08T03:29:02 that would be nice ;-) 2011-03-08T03:29:43 The tools build but you need to add RTEMS to them. 2011-03-08T03:29:45 we want to use a nios system to control a ship engine 2011-03-08T03:29:58 It is a nice device. 2011-03-08T03:30:27 How will you manage the RTEMS code base ? 2011-03-08T03:31:00 the first thing we have to get ready is the tool chain 2011-03-08T03:31:13 afterwards i will try to build the existing nios bsp 2011-03-08T04:57:42 *** mmu_man has joined #rtems 2011-03-08T05:15:07 *** mmu_man has quit IRC 2011-03-08T05:46:40 *** kiwichris has quit IRC 2011-03-08T06:33:38 *** core-ix has quit IRC 2011-03-08T06:33:55 *** peerst has quit IRC 2011-03-08T06:33:56 *** verm__ has quit IRC 2011-03-08T06:33:56 *** agrier has quit IRC 2011-03-08T06:35:12 *** verm__ has joined #rtems 2011-03-08T06:39:23 *** agrier has joined #rtems 2011-03-08T06:59:46 *** verm__ has quit IRC 2011-03-08T06:59:46 *** verm__ has joined #rtems 2011-03-08T08:31:32 *** zwj has joined #rtems 2011-03-08T08:39:35 *** dr__house has joined #rtems 2011-03-08T09:22:07 *** wenjie has joined #rtems 2011-03-08T09:22:17 *** zwj has quit IRC 2011-03-08T09:43:14 posix signal expert online?:) 2011-03-08T09:59:47 *** DrJoel has joined #rtems 2011-03-08T09:59:47 *** ChanServ sets mode: +o DrJoel 2011-03-08T10:01:06 *** dr__house has quit IRC 2011-03-08T10:26:59 DrJoel: im trying posix signals again:) 2011-03-08T10:27:17 :) 2011-03-08T10:33:18 each task can register their own posix signal handlers? 2011-03-08T10:35:27 yes.. but you have to use pthread_kill (not kill) to direct to a specific thread 2011-03-08T10:35:34 ah..:D 2011-03-08T10:43:22 still not working, anything else different from kill(getpid(), SIGUSR1) 2011-03-08T10:44:11 *** wenjie has quit IRC 2011-03-08T10:45:33 can i install handler same way? 2011-03-08T10:47:21 ah.. pthread_sigmask 2011-03-08T10:48:44 *** sebhub has quit IRC 2011-03-08T10:57:10 :< 2011-03-08T10:57:12 still no.. 2011-03-08T10:58:18 check out psxsignal06. It is only 128 lines long so should be pretty clear to show the steps 2011-03-08T10:58:36 im checking psxtest04 2011-03-08T10:59:19 imm 2011-03-08T10:59:20 hmm 2011-03-08T10:59:32 is it new test.. last i have is psxsignal05 2011-03-08T11:00:03 doesn't matter psxsignal04 .. 03/init.c with USE_REAL_TIME_SIGNALS_THREAD) should be good example 2011-03-08T11:01:30 *** methril has joined #rtems 2011-03-08T11:28:33 pthread_kill says 2011-03-08T11:28:47 kill(getpid(), SIGUSR1) works well 2011-03-08T11:29:46 how are you calling it? 2011-03-08T11:31:02 pthread_kill(thread_pid, SIGUSR1); 2011-03-08T11:34:09 where did you set thread_pid? mmmm.. you don't use a pid.. you use a pthread_t .. pthread_self() 2011-03-08T11:41:37 pthread_t thread_pid; 2011-03-08T11:41:43 thread_pid = pthread_self(); 2011-03-08T11:42:16 value is same than with the task when i list task in shell with "task" 2011-03-08T11:43:43 BAD_NAME is ESRCH (I think) and only happens because the id->TCB lookup failed. Can you trace into pthread_kill() and see what it is happening? 2011-03-08T11:44:17 i have no debugger possibility but maybe i add debug messages to it 2011-03-08T11:46:21 is it possible that thread_pid is not set when signal sent? 2011-03-08T11:46:30 because of preemption among two tasks 2011-03-08T11:46:44 If this is a simple exapmle, you can use sparc/sis to have a nice debugger 2011-03-08T11:47:42 i print the value just before i use it so it should be valid? 2011-03-08T11:48:48 yes 2011-03-08T11:48:57 i did that 2011-03-08T11:49:00 and it is valid 2011-03-08T11:49:58 ls 2011-03-08T11:53:50 DrJoel: it returns rtems_set_errno_and_return_minus_one( ESRCH ); 2011-03-08T11:54:15 that means it did not find the id. What value do you have for the ID? 2011-03-08T11:54:19 Did the pthread_create actually work? 2011-03-08T11:54:28 im not using pthread 2011-03-08T11:54:33 rtems task 2011-03-08T11:56:57 Did the task_create work? 2011-03-08T11:57:25 at least i can printf from the task :) and install handlers without error 2011-03-08T11:57:26 Do you have a small self-conftained example? I can probably quickly telly ou what is wrong 2011-03-08T12:00:11 I am sure this is something very simple. I just can't see it.. you can try the old paste.org or whatever that is 2011-03-08T12:00:20 sure.. wait a moment 2011-03-08T12:03:02 http://pastebin.com/UnHRYfpi 2011-03-08T12:03:49 then i call pthread_kill(thread_pid, SIGUSR1); from shell command 2011-03-08T12:04:26 its not straight compiling code :l 2011-03-08T12:04:32 missing init 2011-03-08T12:04:35 and stuff 2011-03-08T12:05:53 Don't see the pthread_kill call 2011-03-08T12:06:11 yep.. i call it from another thread (shell) 2011-03-08T12:07:03 it get pid from extern volatile pthread_t thread_pid; 2011-03-08T12:08:46 http://pastebin.com/73hprNtp 2011-03-08T12:22:09 pthread_kill SIGUSR1 to 0x0a010006=167837702 2011-03-08T12:22:09 [cmd_test] error at pthread_kill: 2011-03-08T12:22:27 *** mmu_man has joined #rtems 2011-03-08T12:22:36 0a010006 THREAD 10 DELAY:Wisig P:T:nA NONE 2011-03-08T12:22:54 from "task" 2011-03-08T12:27:09 Hmmm.. looks ok .. can you put together a n app with it for me to look at? Doesn't have to be small.. just BSP independent so I can run it on sparc/sis and debug it for you 2011-03-08T12:28:37 ok.. i make it single .c file my development enviroment is quite different :) i use Qt Creator as IDE :) 2011-03-08T12:38:22 Cool. I have used that also. If yuo have RTEMS plugins, howto, etc for Qt Creator, an entry in the wiki would be appreciated. 2011-03-08T13:05:50 http://pastebin.com/WWpJRpGe 2011-03-08T13:06:18 at least it compile with my dev env :D 2011-03-08T13:06:23 and runs too 2011-03-08T13:06:45 you have to command "test pthread_kill" on shell 2011-03-08T13:10:56 ok.. 2011-03-08T13:11:32 give me a minute 2011-03-08T13:30:44 still looking at it.. having to take care of other things at the same time.. 2011-03-08T13:36:10 *** mmu_man has quit IRC 2011-03-08T13:45:47 *** bswimley has joined #rtems 2011-03-08T13:54:25 *** core-ix has joined #rtems 2011-03-08T13:55:10 DrJoel: ok :) 2011-03-08T13:58:12 you found a bug... :( 2011-03-08T13:58:46 really? 2011-03-08T13:58:50 Since Classic API (rtem_task_XXX) and POSIX threads (pthread) and ITRON tasks are all supposed to be interchangeable, you have to call a "get" routine to convert the id into a pointer that is able to look across all APIs 2011-03-08T13:59:23 yes! And I am going to have to add a test for this. Can I cut your example down? 2011-03-08T13:59:34 sure 2011-03-08T14:00:49 hold on.. I am filing a PR 2011-03-08T14:02:35 https://www.rtems.org/bugzilla/show_bug.cgi?id=1759 2011-03-08T14:03:08 /tmp/pr1759.diff 2011-03-08T14:03:17 Grrr... wrong window.. attachign proposed patch to pr 2011-03-08T14:03:21 :) 2011-03-08T14:04:02 Let me test it .. hold on 2011-03-08T14:06:52 [/] # test pthread_kill 2011-03-08T14:06:52 pthread_kill SIGUSR1 to 0x0a010002=167837698 2011-03-08T14:06:52 Signal_info: 25 caught by 0x0A010002 si_signo= 4 si_code= 34006032 value= 34099688 2011-03-08T14:07:00 nice! 2011-03-08T14:08:20 i can test it tomorrow.. i already left from work for today.. its 22:07 and im going to sauna =) 2011-03-08T14:08:43 :D My hacked version of your test works now 2011-03-08T14:09:06 Night .. thanks for the bug 2011-03-08T14:09:52 thanks for the fix and fast help :) 2011-03-08T14:12:41 this irc channel is really great for support :) 2011-03-08T14:14:40 :) 2011-03-08T14:14:50 What version are you using? 2011-03-08T14:15:10 4.10 branch from CVS 2011-03-08T14:15:33 ok.. I was sure it is impacted.. need to see how far back this goes for legacy support 2011-03-08T14:15:51 We actually still cut releases from 4.6 for those who have it in products and don't want to upgrade 2011-03-08T14:16:28 DrJoel: is there some documentation about memory allocator ? 2011-03-08T14:16:35 do you have a kmalloc() ? 2011-03-08T14:16:51 kernel malloc? 2011-03-08T14:16:54 yep 2011-03-08T14:17:27 There is only one type of memory in RTEMS. no protection.. So malloc is OK 2011-03-08T14:17:46 Btw. will it be possible in future for rtems i386 versio to support virtualization with RT-XEN? https://sites.google.com/site/realtimexen/ i would really much like to use PC power for doing realtime stuff with rtems :) 2011-03-08T14:17:57 ok so I can use normal malloc() inside the kernel code 2011-03-08T14:18:21 no "sleeping" problem about it ? I can use it even in interrupt ? 2011-03-08T14:18:53 rokka_: should be.. haven't investigated though. There was a port to L4 so feasible. If someone pays, I will code. :-D 2011-03-08T14:19:11 Fallenou: not inside interrupt.. where do you need it? 2011-03-08T14:19:19 was just asking 2011-03-08T14:19:25 actually I don't need it inside interrupt :) 2011-03-08T14:19:38 it's to allocate rx buffers for ethernet driver 2011-03-08T14:19:50 will put that inside the initialization function of the drivr 2011-03-08T14:19:52 good .. that would be bad form.. inside RTEMS proper we use from RTEMS Workspace. and that is carefully accounted for 2011-03-08T14:19:54 driver* 2011-03-08T14:20:02 DrJoel: hmm.. im not familiar with L4, whats that :) 2011-03-08T14:20:19 malloc is fine.. you may have to use an aligned malloc variant 2011-03-08T14:20:23 L4 is another hypervisor 2011-03-08T14:20:33 xi yang did the work as I recall as part of a thesis 2011-03-08T14:20:39 what kind of trouble can be caused by using malloc inside an interrupt top half handler ? 2011-03-08T14:20:49 (yes sure i have to align the data to 4 bytes boundaries) 2011-03-08T14:21:39 if "top half" == task, then none except delay in processing next interrupt since malloc can compete with other users of the heap 2011-03-08T14:21:58 top half = irq handler itself 2011-03-08T14:22:05 4 is OK.. malloc is guaranteed to be correctly aligned for a double by standard. So 4 or 8 byte aligned on every target 2011-03-08T14:22:15 ok great :) 2011-03-08T14:22:42 so a real low-level ISR... don't malloc there. It locks a mutex and can block. 2011-03-08T14:23:39 ok 2011-03-08T14:23:42 that's what I thought 2011-03-08T14:24:09 :) 2011-03-08T14:24:15 Hmm. is it possible for RTEMS running on hypervisor to access PCI-devices, USB or other hardware? 2011-03-08T14:25:34 I would guess so. If VMware and Virtualbox count, they can run pc386 BSP 2011-03-08T14:25:45 I didn't know you had a blog DrJoel ! good news, I'm bookmarking it :) 2011-03-08T14:27:10 ok. i must test RTEMS on pc.. :) 2011-03-08T14:27:32 :) It is discouraging not to see comments .. blogging without readers sucks 2011-03-08T14:28:12 i would like to use RTEMS to control CNC machine or just servo motors for start 2011-03-08T14:28:27 rokka_: running tests on qemu with pc386 BSP or sparc/sis is really easy and free... Your bug was easy to track down in sis since that is built into gdb. 2011-03-08T14:29:38 i see:) 2011-03-08T14:31:31 Do you ever seer Peer Stryinger on here? 2011-03-08T14:31:40 he wants an RTEMS/PLC 2011-03-08T14:32:33 i386 PLC or something else? 2011-03-08T14:33:19 I don't know that he has identified a good free package to base things on. He is in the Munich area so it sounds like you are on the same timezone. :) 2011-03-08T14:34:07 okey.. :) sounds interesting 2011-03-08T14:34:38 I think so. I want more veritical stacks for RTEMS 2011-03-08T14:36:25 meening what.. :)? 2011-03-08T14:36:29 meaning.. 2011-03-08T14:37:35 graphics toolkit, cnc, plc, scientific controller, etc.. more add-ons in a particular domain or related area 2011-03-08T14:38:05 the graphics toolkit isn't in bad shape but could be better.. Microwindows/nano-X plus jpeg, png, tiff, T1 font and TrueType support 2011-03-08T14:38:40 okay :) 2011-03-08T14:39:13 hmm.. does the i386 bsp support multiple cores? 2011-03-08T14:40:29 not yet.. SMP is under construction.. 2011-03-08T14:40:49 I can get a pc386 with multicore initialized but we don't have scheduler or lockiing in place yet 2011-03-08T14:41:19 so can i start rtems with modern pc.. like intel 2 core? 2011-03-08T14:41:34 after bios 2011-03-08T14:41:46 grub bootloader or something?:) 2011-03-08T14:43:08 maybe i should just read wiki first and then google and then ask here :D 2011-03-08T14:51:51 sorry someone walked in. 2011-03-08T14:51:53 yes.. 2011-03-08T14:52:26 on qemu, for automated testing, I use a virtual floppy image which has grub and always boots test.exe from C: 2011-03-08T14:52:43 on a real embebdded PC for testing, we boot over the network 2011-03-08T14:53:11 on a real embedded PC for testing, you would boot from flash and use grub or something similar 2011-03-08T14:53:28 but for now, on a multi-core PC, you are only using the first core. 2011-03-08T14:54:10 Jennifer and I are slowing making progress on SMP support. There is a PR now to add --enable-smp which turns on allocation of an array of information "per cpu" 2011-03-08T14:56:02 Next patches add first cut at SMP BSP support for LEON3 and PC386.. can bring the systems out of reset as multi-core .. up through init.. but since no way to schedule threads on the secondary cores, the system dies at the end of init 2011-03-08T14:56:16 getting closer.. just a matter of (funded) time on keyboard to finish it.. 2011-03-08T15:40:17 *** methril has quit IRC 2011-03-08T16:01:32 *** bswimley has quit IRC 2011-03-08T16:08:16 *** dr__house has joined #rtems 2011-03-08T16:08:16 *** dr__house has joined #rtems 2011-03-08T16:18:55 *** projectgus has joined #rtems 2011-03-08T16:24:37 okay :) 2011-03-08T16:30:15 rokka_: since you are still there (although probably not awake).. pr1759 is committed 2011-03-08T16:32:09 nice 2011-03-08T16:32:58 thanks for the report 2011-03-08T16:35:50 *** mmu_man has joined #rtems 2011-03-08T16:38:32 you're welcome 2011-03-08T16:41:36 *** dr__house` has joined #rtems 2011-03-08T16:41:36 *** dr__house` has joined #rtems 2011-03-08T16:41:48 *** dr__house has quit IRC 2011-03-08T16:50:36 *** DrJoel has quit IRC 2011-03-08T17:35:32 *** dr__house` has quit IRC 2011-03-08T17:56:05 *** rokka_ is now known as rokka 2011-03-08T18:03:42 *** mmu_man has quit IRC 2011-03-08T18:05:42 *** projectgus has quit IRC 2011-03-08T18:13:45 *** projectgus has joined #rtems 2011-03-08T19:05:56 *** dr__house has joined #rtems 2011-03-08T19:05:56 *** dr__house has joined #rtems 2011-03-08T19:31:59 *** kristianpaul has quit IRC 2011-03-08T19:32:57 *** kristianpaul has joined #rtems 2011-03-08T19:40:46 *** kristianpaul has quit IRC 2011-03-08T19:46:12 *** kristianpaul has joined #rtems 2011-03-08T19:46:12 *** kristianpaul has joined #rtems 2011-03-08T22:21:44 *** dr__house has quit IRC 2011-03-08T22:22:11 *** dr__house has joined #rtems 2011-03-08T22:22:11 *** dr__house has joined #rtems 2011-03-08T22:54:38 *** dr__house has quit IRC 2011-03-09T00:03:36 *** projectgus has quit IRC 2011-03-09T00:05:23 *** lcpfnyvc has joined #rtems 2011-03-09T00:26:10 *** lcpfnyvc has quit IRC 2011-03-09T00:26:20 *** lcpfnyvc has joined #rtems 2011-03-09T00:27:33 *** lcpfnyvc has joined #rtems 2011-03-09T00:27:53 *** lcpfnyvc has joined #rtems 2011-03-09T00:40:12 morning 2011-03-09T01:16:53 *** mmu_man has joined #rtems 2011-03-09T02:07:16 *** mmu_man has quit IRC 2011-03-09T03:58:08 *** core-ix has quit IRC 2011-03-09T04:39:38 *** core-ix has joined #rtems 2011-03-09T05:39:25 *** antgreen has joined #rtems 2011-03-09T05:54:33 *** mmu_man has joined #rtems 2011-03-09T06:22:33 *** sebhub has joined #rtems 2011-03-09T06:22:42 hi 2011-03-09T06:56:23 *** zwj has joined #rtems 2011-03-09T08:06:58 *** mmu_man has quit IRC 2011-03-09T08:27:50 *** mmu_man has joined #rtems 2011-03-09T08:41:18 *** zwj has quit IRC 2011-03-09T08:45:21 *** DrJoel has joined #rtems 2011-03-09T08:45:21 *** ChanServ sets mode: +o DrJoel 2011-03-09T08:49:35 hi, joel 2011-03-09T08:49:43 hey sebhub .. is peter in the office today? 2011-03-09T08:50:18 yes 2011-03-09T08:50:25 can you ask him to Skype me? 2011-03-09T08:50:50 one moment, i ask him 2011-03-09T08:51:33 FYI posted coverage results overnight for erc32.. down to 5 ranges with 2 in heap.. http://www.rtems.org/ftp/pub/rtems/people/joel/coverage/erc32/erc32-Ospd-20110308-1907/uncovered.html 2011-03-09T08:52:10 you did you catch the open ranges in the timer server? 2011-03-09T08:52:50 can peter call you tomorrow? 2011-03-09T08:52:56 apparently at -Os.. there was also a bug in the coverage analysis .. 2011-03-09T08:52:59 sure.. 2011-03-09T08:53:11 at which time? 2011-03-09T08:54:08 I might be on as early as 7:30am but probably 8:15am is safer. 2011-03-09T08:54:38 ok, he calls you tomorrow 2011-03-09T08:55:09 thanks.. just trying to get details of class worked out.. can bring family on airmiles so need to book early. 2011-03-09T08:55:38 the QorIQ bsp is finished, i only need an official tool chain for a commit 2011-03-09T08:55:43 Timer server still has cases at -O2.. but not as -Os... one step at a time 2011-03-09T08:55:46 YEAH!!! 2011-03-09T08:55:54 DrJoel: signal sending works now in my app.. thanks again :) 2011-03-09T08:56:15 Is the arm thumb2 patch ready to merge? If Ralf is going to be spinning tools, we might as well catch both 2011-03-09T08:56:31 rokka: Thanks for the bug.. your code turned into psxclassic01 2011-03-09T08:56:44 hm, it is a complex change to switch to the new ABI 2011-03-09T08:56:55 :) 2011-03-09T08:57:31 i already changed all arm linker command files as a preparation step 2011-03-09T08:58:08 well.. we have to start sometime.. and it is the CVS head.. plus it is the only way we can pick up newer ARM models 2011-03-09T08:58:49 ok, i try to find some time to build a new arm compiler and test the new ABI 2011-03-09T08:59:08 it is a big change, not just a new multi-lib 2011-03-09T09:00:38 ok.. I can always test the patch against gcc-svn so we can compare current test results with and without the patch 2011-03-09T09:05:23 joel, can you please send peter your skype contact info 2011-03-09T09:06:00 ok.. thanks 2011-03-09T09:06:43 *** dr__house has joined #rtems 2011-03-09T09:06:43 *** dr__house has joined #rtems 2011-03-09T09:17:18 http://www.rtems.org/ftp/pub/rtems/people/joel/coverage/erc32/erc32-Ospd-20110308-1907/annotated.html#range12 2011-03-09T09:17:32 it seems to be impossible to reach this code 2011-03-09T09:17:53 i think about removeing this aling up statement 2011-03-09T09:18:56 hm, ok, there is a test case for this 2011-03-09T09:24:46 The ones remaining tend to not be easy. :( 2011-03-09T09:25:46 We just keep nibbling so we can say 100% instruction coverage at -Os for these directories ... and grow the set of directories and/or broaden to -O2. 2011-03-09T09:29:03 *** mmu_man has quit IRC 2011-03-09T09:30:12 ok, i have a test case 2011-03-09T09:31:39 *** bswimley has joined #rtems 2011-03-09T09:38:41 *** lcpfnyvc has quit IRC 2011-03-09T09:43:23 bye 2011-03-09T09:43:26 *** sebhub has quit IRC 2011-03-09T09:52:20 *** mmu_man has joined #rtems 2011-03-09T10:22:34 Dr Joel: hellow 2011-03-09T10:22:41 I mean hello! 2011-03-09T10:30:23 bswimley: howdy! 2011-03-09T10:35:51 Dr Joel: Do you know what the status of the Xilinx microblaze RTEMS support is at this time? Is it actively being supported? 2011-03-09T10:36:35 I know people are interested in it 2011-03-09T10:37:10 That's another option we're considering, and I was just wondering what the status of the port was. 2011-03-09T10:37:30 Is it in the tree at all? 2011-03-09T10:37:49 Doesn't exist at the moment. Most likely target to appear next IMO 2011-03-09T10:39:34 hmm - I may dig around and see what I can find. I think FreeRTOS is supported by the Xilinx SDK. 2011-03-09T10:40:13 I hope that doesn't mean you would choose that over RTEMS :-D 2011-03-09T10:40:44 I know what you mean. I'd like to go with RTEMS but development costs on this project are tight. 2011-03-09T10:41:23 I may dig around and see what I can find on microblaze support. That would be a rockin' combination IMO. 2011-03-09T10:42:08 I really don't know of another architecture right now of much interest without heading into DSPs 2011-03-09T10:51:28 hmm. 600-800us sound like a bit too much for signal to reach task handler? or is it? 2011-03-09T10:51:44 56MHz arm7tdmi 2011-03-09T10:54:29 hold on.. watching discovery land for last time .. 3 minutes to touchdown 2011-03-09T10:55:12 thanks for tip! 2011-03-09T10:58:12 nice landing 2011-03-09T10:58:20 sigh... only one mission left... 2011-03-09T10:59:02 oar is across Interstate from space and rocket center.. on way to work every day I see the shuttle Pathfinder and a Saturn v 2011-03-09T10:59:13 :-D 2011-03-09T10:59:29 take a high res photo for us 2011-03-09T11:00:40 How about Google Maps 2011-03-09T11:00:48 ;) 2011-03-09T11:01:02 address? 2011-03-09T11:01:56 http://maps.google.com/maps?f=q&source=s_q&hl=en&geocode=&q=OAR+Corporation+Huntsville&aq=&sll=37.0625,-95.677068&sspn=46.946584,68.115234&ie=UTF8&hq=OAR+Corporation&hnear=Huntsville,+Madison,+Alabama&ll=34.711403,-86.654245&spn=0.005768,0.008315&t=h&z=17 2011-03-09T11:02:18 That view shows pathfinder on lower right and saturn v replica's shadow 2011-03-09T11:02:34 the building to the left of the shadow has the real one 2011-03-09T11:03:18 that time doesn't sound too much out of line depending on how the code is setup 2011-03-09T11:03:55 ok.. but i miss the hard deadline of 450us 2011-03-09T11:03:58 POSIX signal has to be sent, unblock receiver, context switch to receiver and then dispatch signal 2011-03-09T11:04:07 it might be a little long though 2011-03-09T11:04:15 Do you need to interrupt a blocking sleep? 2011-03-09T11:04:34 POSIX signals are really not real-time no matter how they are named. Horribly complex code to handle them 2011-03-09T11:04:44 :l 2011-03-09T11:05:00 Better to have a dedicated task and either send it an event (rtems_event_send/receive) or a message via classic api 2011-03-09T11:05:03 i would like to dedicate all SPI-bus stuff to one task.. 2011-03-09T11:05:35 and i thought that signal is faster than msg :l 2011-03-09T11:06:13 what should the task to while waiting for event / msg? 2011-03-09T11:06:21 Messages work well. We get ~12-16us on a PowerPC MPC5200 (200 Mhz) complete processing time from discrete input change interrupt to task (in Ada) receiving it. That requires an interrupt dispatch plus context switch with FPU save/restore 2011-03-09T11:06:40 that sound much better 2011-03-09T11:06:50 task could be blocked if you want fast response to the input 2011-03-09T11:07:14 hmm.. doesn't sleep block? 2011-03-09T11:07:46 Yes but you have to interrupt the block and dispatch the signal. The event/message don't have that overhead. 2011-03-09T11:07:56 Could your time have a print in it? Seems too short for that though 2011-03-09T11:08:11 look at examples-v2/led for some event and message examples 2011-03-09T11:08:15 ok 2011-03-09T11:08:48 idea is that whatever the task is doing it must discard that and handle the event/msg/signal 2011-03-09T11:09:06 in much less time than 450us 2011-03-09T11:09:44 event is a little lighter than message but it all depends on how much data you need. Events don't queue.. messages queue 2011-03-09T11:10:03 and the signal/msg/event is send from ISR which reacts to a digital input 2011-03-09T11:12:03 ASR sounded nice solution but it could not interrupt sleep :l 2011-03-09T11:12:09 Same thing I did on pcmmio (winsystems multi IO board). ISR decides which pin changed and sends message with timestamp 2011-03-09T11:12:34 Signals are usually not right for this... events or messages usually turn out to be nice 2011-03-09T11:22:55 mm.. 2011-03-09T11:23:17 200us for resuming task from suspend 2011-03-09T11:24:26 that sounds too high 2011-03-09T11:25:41 tm05:rtems_task_suspend: calling task 96 2011-03-09T11:25:41 tm06:rtems_task_suspend: returns to caller 34 2011-03-09T11:25:45 tm05:rtems_task_resume: task readied -- preempts caller 92 2011-03-09T11:25:46 tm06:rtems_task_resume: task readied -- returns to caller 35 2011-03-09T11:26:30 on sparc/erc32 at either 14 Mhz.. 2011-03-09T11:26:35 im resuming it from another thread 2011-03-09T11:27:20 not either.. at 14Mhz 2011-03-09T11:27:34 im not sure what is your result:) 2011-03-09T11:28:42 resume preempts means a thread calls resume and it is preempted by another thread immediately which stops the timer 2011-03-09T11:29:08 suspend calling task starts timer, calls suspend(self), switches to other task which stops timer 2011-03-09T11:29:32 task readied -- returns to caller 35... same as other case except task resumed is lower priority and we don't switch 2011-03-09T11:30:09 how many time does it take? 2011-03-09T11:30:14 how much.. 2011-03-09T11:30:31 oh.. those are microseconds 2011-03-09T11:30:54 your CPU is 3 times faster and the times are 2 times longer.. so...??? 2011-03-09T11:31:14 i see.. something is wrong?:l 2011-03-09T11:31:25 16-bit bus? .. high wait states? not timing what you think you are timing? timer not calibrated? 2011-03-09T11:31:30 sometimes the diff is only 100us.. 2011-03-09T11:31:53 what is your timing source? 2011-03-09T11:32:22 okay.. my time measuring method is a bit stupid:D rtems_and clock_get_ticks_since_boot() 2011-03-09T11:33:19 then it has a granularity of a clock tick.. which much be programmed to 100 usecs? 2011-03-09T11:33:30 its 50us now 2011-03-09T11:33:42 so it sometimes resumes in 50-100us 2011-03-09T11:33:50 mostly over 3 2011-03-09T11:33:56 3=150us 2011-03-09T11:34:18 ok.. probably reasonable... but this method has overhead of interrupting clock ticks and isn't that accurate. 2011-03-09T11:35:08 for small times, the timer driver in the bsp is useful. It sometimes uses the same hw as the clock tick so sometimes can't be used at the same time .. but you need to find a counter of some sort to get a more accurate time 2011-03-09T11:35:33 on powerpc, we use time base register (TBR) and TSC (cycle counter) on pentium and above. 2011-03-09T11:35:53 you can probably read the counter for the clock if you like 2011-03-09T11:36:35 for small times, you need high granularity and low overhead .. clock_get_ticks_since_boot() is low overhead (ignoring clock tick isr), high granularity 2011-03-09T11:37:27 getting TOD (if bsp has nanoseconds since last tick) is high granularity and can have greater time between clock ticks 2011-03-09T11:39:31 so what was the funktion you recommend? 2011-03-09T11:40:34 ill check the bsp code-> 2011-03-09T11:41:39 rtems_clock_get_uptime()... and the BSP's clock driver has to have the get nanoseconds since last tick handler. It is optional 2011-03-09T11:41:55 lpc24xx bsp 2011-03-09T11:42:00 There are a couple of tests (nsecs) and another to test the nanoseconds support in examples-v2 2011-03-09T11:42:44 static uint32_t lpc_clock_nanoseconds_since_last_tick(void) 2011-03-09T11:42:45 { 2011-03-09T11:42:45 uint64_t clock = LPC_CLOCK_REFERENCE; 2011-03-09T11:42:45 uint64_t clicks = lpc_clock->tc; 2011-03-09T11:42:45 uint64_t ns = (clicks * 1000000000) / clock; 2011-03-09T11:42:45 return (uint32_t) ns; 2011-03-09T11:42:47 }\ 2011-03-09T11:42:50 so you are set to go. 2011-03-09T11:43:13 cool.. i'll try it -> 2011-03-09T11:44:00 Run nanosecond_tick_wrap to make sure it behaves correctly if the counter runs under zero while we are reading the TOD... we realized recently that some BSPs were succeptible to being confused if the counter went below 0 .. 2011-03-09T11:44:44 basically a clock interrupt is generated whiel we have interrupts disabled getting the tod... sparc BSPs handle it now.. other bsps may or may not be succeptibel depending on hardware 2011-03-09T11:50:48 okay its saying about 190us 2011-03-09T11:51:14 but that is less than my deadline;) 2011-03-09T11:51:36 but it would be nice to reach in like 20us 2011-03-09T11:51:43 react.. 2011-03-09T11:55:00 hmm how can this processor measure 1ns if its self runs 56Mhz? doesnt it require 1GHz clock 2011-03-09T11:56:05 eint0: 294 372087708 2011-03-09T11:56:05 signal: 294 372278159 2011-03-09T11:56:05 diff: 190 2011-03-09T11:56:05 eint0: 300 51912708 2011-03-09T11:56:05 signal: 300 52104079 2011-03-09T11:56:07 diff: 191 2011-03-09T11:56:10 eint0: 301 913846510 2011-03-09T11:56:12 signal: 301 913801979 2011-03-09T11:56:15 diff: -44 2011-03-09T11:56:17 bug?:D 2011-03-09T11:56:20 what went wrong.. 2011-03-09T11:56:47 numbers are seconds.. nanoseconds 2011-03-09T11:56:54 diff is microseconds 2011-03-09T12:04:08 it is based upon the clock frequency and the counter... may not be able to measure 1ns.. may actually jump by N... I know the erc32 BSP only has a granularity of microseconds 2011-03-09T12:04:26 so you always have 3 zeroes on your nanoseconds value 2011-03-09T12:05:54 the -44 is what I was talking about .. see examples-v2/misc/ for a test. Might also want to post to the mailing list if you don't want to tackle it yourself.. I think Sebastian Huber can fix it. But isn't hard to address if you can understand what the counter is doing.. basically if an interrupt is pending, you have to add a full clock tick 2011-03-09T12:06:03 check sparc/erc32/clock code for a simple template 2011-03-09T12:06:30 ok thanks.. :) i check that tomorrow 2011-03-09T12:06:51 2011-03-04 Joel Sherrill 2011-03-09T12:07:24 PR 1748/bsps 2011-03-09T12:07:24 * clock/ckinit.c: When the clock tick generates an interrupt WHILE we 2011-03-09T12:07:24 have interrupts disabled doing a get TOD or uptime, the get 2011-03-09T12:07:24 nanoseconds handler was returning a bogusly large number. 2011-03-09T12:07:24 Very recent behavioural discovery which may or may not impact a clock driver's nanoseconds since last tick handler depending on the hardware 2011-03-09T12:32:14 *** mmu_man has quit IRC 2011-03-09T13:47:39 *** DrJoel has quit IRC 2011-03-09T13:50:43 *** DrJoel has joined #rtems 2011-03-09T13:50:43 *** ChanServ sets mode: +o DrJoel 2011-03-09T14:21:31 *** core-ix has quit IRC 2011-03-09T14:51:35 *** mmu_man has joined #rtems 2011-03-09T15:11:03 *** core-ix has joined #rtems 2011-03-09T15:20:44 *** demute has joined #rtems 2011-03-09T15:21:22 * DrJoel notes that it is quite quiet in here... 2011-03-09T15:28:07 hello people, first time here... :) I have just compiled the cross-compiler and I am on my way to compiler rtems-4.10. I don't have a BSP package yet, can I install rtems anyway? Or rather, how does the edit-compile-debug cycle look like for BSP development? 2011-03-09T15:31:27 Hi and welcome! 2011-03-09T15:31:27 do you have a cpu architecture/board in mind? 2011-03-09T15:33:16 hello DrJoel! Yes, the board is called kb9202 (from KwikByte) it has an at91rm9200 processor. 2011-03-09T15:34:12 thanks to google, I see that is a nice board. :) 2011-03-09T15:36:11 Is it close to the Cogent CSB337? It also uses an at91rm9200. 2011-03-09T15:36:31 The normal path is to find something that is close and see exactly how little work you have to do. :) 2011-03-09T15:37:00 we have a concept called "bsp variant" where you can compile a single BSP under different names and tweak parameters... 2011-03-09T15:37:04 (late answer) yeah, it's nice, if you google on my name, manne tallmarken, you will see the actual project too, just click on the first hit and then the arm based amplifier 2011-03-09T15:38:12 so what actually happens when I compile rtems? does it make object files for a bunch of stuff which later on will be linked in my application? 2011-03-09T15:38:22 cool project 2011-03-09T15:38:31 thanks :) 2011-03-09T15:39:07 in general terms, RTEMS itself is one library (librtemscpu.a) with your BSP (librtemsbsp.a) and your application code.. all are statically linked together. 2011-03-09T15:39:33 make install installs the two libraries, BSP linkcmds, etc and you can then use it to build whatever app you want outside the autoconf'ed tree 2011-03-09T15:40:00 ok 2011-03-09T15:40:36 To get a handle on things, install sparc-rtems and try out sis (SPARC Instruction Simulator) built into gdb. That way you have something free and easy to use and step around in. If you are trying APIs, wanting to trace something that works, get a feel for things, etc it is a nice way to learn 2011-03-09T15:41:12 for GSOC, most students use either that .. or qemu with the i386/pc386 BSP 2011-03-09T15:41:12 ok, thanks for the tip 2011-03-09T15:42:02 but I think the csb337 BSP is very close to what you have 2011-03-09T15:42:38 yes I think so too. So what is the minimum requirements for a bsp? initialization code? 2011-03-09T15:42:42 nothing more? 2011-03-09T15:43:43 If you want to run the tests, you need some start code, polled console output, clock tick, linker script, and clean shutdown. I think the h8300/h8sim is one of the smallest BSPs. 2011-03-09T15:44:08 I think all fo the BSPs for real HW support all of what I listed 2011-03-09T15:44:35 you can't timeout or delay without a clock tick 2011-03-09T15:45:33 Dr Joel: Can you build other targets with the virtual box images you published (coldfire, etc.) 2011-03-09T15:46:50 DrJoel: that sounds great, pretty easy start. Got to go to bed, thanks for help! 2011-03-09T15:48:12 no problem.. feel free to join the rtems-users mailing list and ask questions. Lots of people can answer it there 2011-03-09T15:52:37 will do that... so, I just sent a mail to rtems-users-subscribe@rtems.com. See you 2011-03-09T15:52:44 *** bswimley_ has joined #rtems 2011-03-09T15:53:56 *** demute has quit IRC 2011-03-09T15:54:31 *** methril has joined #rtems 2011-03-09T15:56:04 *** bswimley has quit IRC 2011-03-09T16:17:13 *** dr__house has quit IRC 2011-03-09T16:30:08 *** johntramp has joined #rtems 2011-03-09T16:32:05 *** johntramp has left #rtems 2011-03-09T16:33:48 *** projectgus has joined #rtems 2011-03-09T16:34:17 hey gus how's your project going? 2011-03-09T16:44:14 hey DrJoel 2011-03-09T16:44:19 which one? ;) 2011-03-09T17:42:46 *** bswimley_ has quit IRC 2011-03-09T18:01:19 *** mmu_man has quit IRC 2011-03-09T19:10:52 *** kiwichris has joined #rtems 2011-03-09T19:11:02 * kiwichris waves to Joel 2011-03-09T19:27:18 *** bswimley has joined #rtems 2011-03-09T19:27:41 *** bswimley has left #rtems 2011-03-09T19:50:05 *** methril_ has joined #rtems 2011-03-09T19:51:29 *** methril has quit IRC 2011-03-09T19:52:31 *** methril_ is now known as mehtril 2011-03-09T19:52:50 *** mehtril is now known as methril 2011-03-09T20:01:44 *** dr__house has joined #rtems 2011-03-09T20:13:12 *** dr__house has quit IRC 2011-03-09T22:03:29 *** kiwichris has quit IRC 2011-03-09T22:36:57 *** kiwichris has joined #rtems 2011-03-09T23:03:48 *** methril has quit IRC 2011-03-09T23:14:08 *** methril has joined #rtems 2011-03-10T00:08:06 *** projectgus has quit IRC 2011-03-10T00:21:47 *** lcpfnyvc has joined #rtems 2011-03-10T00:21:59 morning 2011-03-10T01:26:08 *** kiwichris has quit IRC 2011-03-10T02:13:09 *** mmu_man has joined #rtems 2011-03-10T02:58:05 *** mmu_man has quit IRC 2011-03-10T03:30:42 *** mmu_man has joined #rtems 2011-03-10T03:43:11 *** core-ix has quit IRC 2011-03-10T04:01:25 *** tuxmaniac is now known as aanjhan 2011-03-10T04:12:40 *** methril_ has joined #rtems 2011-03-10T04:14:15 *** methril has quit IRC 2011-03-10T04:22:51 *** core-ix has joined #rtems 2011-03-10T04:25:55 *** methril__ has joined #rtems 2011-03-10T04:26:51 *** methril_ has quit IRC 2011-03-10T04:49:05 hello 2011-03-10T04:59:19 hello 2011-03-10T05:06:33 *** mmu_man has quit IRC 2011-03-10T05:32:43 *** mmu_man has joined #rtems 2011-03-10T05:39:55 *** methril_ has joined #rtems 2011-03-10T05:40:39 *** methril__ has quit IRC 2011-03-10T06:15:41 *** methril__ has joined #rtems 2011-03-10T06:16:39 *** methril_ has quit IRC 2011-03-10T07:16:08 *** mmu_man has quit IRC 2011-03-10T07:17:35 * DrJoel is really here now 2011-03-10T07:29:29 *** zwj has joined #rtems 2011-03-10T07:37:35 early..:) 2011-03-10T07:44:38 Yes. My youngest son likes to go into school early on Thursday and Friday for the anime club. LOL 2011-03-10T07:45:11 haha:) 2011-03-10T07:45:53 It is hard to believe my youngest (of 4) will be 17 this summer. Daughter will be 22 in May and should graduate college in December. 2011-03-10T07:46:34 :) do you feel time flows faster and faster every year?-) 2011-03-10T07:47:18 yes.. and the Boy Scout swimming test gets harder and harder to pass.. even though it doesn't change. I managed to pass it last night. So good again for another year 2011-03-10T07:49:00 does it test only speed? 2011-03-10T07:49:35 no speed.. just distance and style.. really only 100m but seems much farther.. 2011-03-10T07:52:32 i can swim maybe couple kilometers and dive 25m :-) 2011-03-10T07:57:20 wow! obviously you are in better shape than I am. 2011-03-10T07:58:10 depends of style..:D 2011-03-10T07:58:48 front crawl is difficult :l 2011-03-10T07:59:34 never was the most athletic person.. :( 2011-03-10T08:00:19 swimming is my only sport hobby 2011-03-10T08:01:20 I was geocaching a lot before a health scare last year. The meds I was put on just knocked me on my butt.. 2011-03-10T08:01:21 hmm.. besides rare orienteering on summer 2011-03-10T08:03:22 :l oh.. 2011-03-10T08:06:49 well the meds are lower now but I gained 20 pounds (13Kg?) I didn't need and am going to have to make a point of getting more active 2011-03-10T08:07:36 have you tried low carb diet? 2011-03-10T08:07:42 or lower.. 2011-03-10T08:07:52 not zero carb.. :) 2011-03-10T08:09:54 *** mmu_man has joined #rtems 2011-03-10T08:37:37 rokka: trying .. but bread is sooo good 2011-03-10T08:40:44 yes it is;) 2011-03-10T08:41:31 and beer... 2011-03-10T08:45:36 200 us to receive msg 2011-03-10T08:46:14 slower than resuming task :o 2011-03-10T08:46:24 from task to task? 2011-03-10T08:46:30 from ISR to task 2011-03-10T08:47:00 still seems high... 2011-03-10T08:47:11 next.. i'll try event 2011-03-10T08:47:28 I would expect about 1/4 that.. are you still measuring with number of clock ticks? 2011-03-10T08:47:51 rtems_clock_get_uptime 2011-03-10T08:49:02 hmm.. does -Os vs -O2 make difference? 2011-03-10T08:50:07 i had.. 200 MICROSECONDS_PER_TICK 2011-03-10T08:50:11 :D 2011-03-10T08:51:01 -Os usually produces ~10% smaller executables... speed seems to be the same when the primary speed issue is number of instructions executed 2011-03-10T08:51:20 now it is 50us with 50 MICROSECONDS_PER_TICK 2011-03-10T08:51:23 nice! 2011-03-10T08:51:39 but it varies.. O_O 2011-03-10T08:51:43 sometimes 150us 2011-03-10T08:51:56 of course.. how do clock ticks hit 2011-03-10T08:52:15 how can i know what is worst case 2011-03-10T08:52:20 try using benchmark_timer_initialize and benchmark_timer_read.. it uses the timer 2011-03-10T08:52:48 ok 2011-03-10T08:53:07 Your granularity is so high and the clock tick overhead is seriously impacting the measurement. 3-4 ISRs 2011-03-10T08:53:29 which header file? 2011-03-10T08:53:31 Uses lpc24xx_timer ... don't know what numbers it reports.. could be cycles or times 2011-03-10T08:54:03 rtems/btimer.h 2011-03-10T08:54:51 cvs/rtems$ find | grep btimer 2011-03-10T08:54:55 no results 2011-03-10T08:55:45 hmmm.. I was on head... 2011-03-10T08:55:47 int benchmark_timer_read( void ) 2011-03-10T08:55:52 void benchmark_timer_initialize( void ) 2011-03-10T08:55:56 is the entire interface 2011-03-10T08:56:06 int benchmark_timer_read( void ) 2011-03-10T08:56:06 { 2011-03-10T08:56:06 return (T0TC/(LPC22xx_Fpclk/1000000)); 2011-03-10T08:56:47 c/src/lib/libcpu/arm/lpc22xx/timer has the hw interface code.. don't know the details but guessing reporting in usecs 2011-03-10T08:59:09 *** dr__house has joined #rtems 2011-03-10T09:01:54 *** dr__house has quit IRC 2011-03-10T09:01:59 *** dr__house has joined #rtems 2011-03-10T09:01:59 *** dr__house has joined #rtems 2011-03-10T09:03:08 *** antgreen has quit IRC 2011-03-10T09:03:55 *** antgreen has joined #rtems 2011-03-10T09:04:02 lets see.. 2011-03-10T09:05:55 *** antgreen has quit IRC 2011-03-10T09:07:01 it reports.. 100x more accurately than usec 2011-03-10T09:08:58 *** antgreen has joined #rtems 2011-03-10T09:10:23 *** dr__house` has joined #rtems 2011-03-10T09:10:24 *** dr__house` has joined #rtems 2011-03-10T09:10:46 *** dr__house has quit IRC 2011-03-10T09:11:26 :D now for a real number 2011-03-10T09:12:02 FWIW the time should be near constant given the same size message.. nothing but constant time operations in blocking and unblocking (w/o timeouts) 2011-03-10T09:12:10 eint0: 2139082703 2011-03-10T09:12:11 signal: 2139093899 2011-03-10T09:12:11 diff: 11196 2011-03-10T09:12:11 eint0: 2210384464 2011-03-10T09:12:11 signal: 2210393392 2011-03-10T09:12:13 diff: 8928 2011-03-10T09:12:15 eint0: 2252479639 2011-03-10T09:12:18 signal: 2252488574 2011-03-10T09:12:20 diff: 8935 2011-03-10T09:12:46 signal means moment when msg is received 2011-03-10T09:13:30 so ~8.9 - 11.2 useconds? with the slower case likely being the out of cache time 2011-03-10T09:13:30 that sounds much more like it. 2011-03-10T09:13:39 may bump up when ISR happens during measurement but that's the real time for the code in question 2011-03-10T09:14:37 your board has ~3x times faster clock than erc32 and more modern CPU. The times reflect that. 2011-03-10T09:16:15 no.. its more like 89 - 112 2011-03-10T09:16:39 or is it.. :l 2011-03-10T09:17:13 good question for sebastian since he was involved in writing that driver. A factor of 10 is an easy mistake to make reading that type of code. 2011-03-10T09:17:39 I really would expect the lower number but if your board is 16-bit RAM and high wait states, it could take a hit 2011-03-10T09:20:27 okay 2011-03-10T09:21:45 ~ 100us < 450us so this could work 2011-03-10T09:31:56 yes.. how high a rate are the interrupts coming? 2011-03-10T09:32:46 20ms 2011-03-10T09:33:43 ohh.. no problem.. a reasonable clock tick would be 1 millisecond or 500 microseconds and the overhead from that would be low. 2011-03-10T09:34:23 Just need to keep task receiving from ISR high priority and preemption on. :D 2011-03-10T09:36:33 it has highest priority and preemption:) 2011-03-10T09:36:44 priority = 10 2011-03-10T09:36:52 next is UI with 50 2011-03-10T09:40:10 *** bswimley has joined #rtems 2011-03-10T09:40:47 hey bswimley .. 2011-03-10T09:41:00 rokka: good.. should be a very sustainable rate 2011-03-10T09:41:23 Dr Joel: hi 2011-03-10T09:41:53 sent you some email.. hopefully second one in a bit 2011-03-10T09:42:06 Got the 1st one. Did you see my note regarding the VM image? 2011-03-10T09:45:19 know... 2011-03-10T09:45:22 is it broken? 2011-03-10T09:45:58 It is for me. THe 4.10 VM works fine, the 4.11 says it can't read the OVF file and throws an error. 2011-03-10T09:46:34 hmmm.. need to check the md5sum again.. takes a long time to transfer 2011-03-10T09:46:43 And download! 2011-03-10T09:47:47 I can continue on with the 4.10 VM for now, but if you get a chance to thow up the 4.11 again, that would be wonderful. 2011-03-10T09:50:11 *** core-ix has quit IRC 2011-03-10T09:50:14 It took me two days to get them uploaded.. :( 2011-03-10T09:50:34 Yuk! 2011-03-10T09:51:17 Could you throw it up on something like dropbox temporarily? I could give it a try and let you know if it works. 2011-03-10T09:51:46 Or FTP? 2011-03-10T09:52:21 the two days was mostly figuring out sourceforge and having a storm here ... 2011-03-10T09:52:27 checking md5sum's now 2011-03-10T09:53:17 hmm.. i tested benchmark timer with addiotional sleep(1) 2011-03-10T09:53:23 result 57629623 2011-03-10T09:53:34 my cpu is 57.6MHz 2011-03-10T09:53:45 that added a second 2011-03-10T09:53:54 yeah.. :) 2011-03-10T09:53:58 so it gives clock cycles 2011-03-10T09:56:51 *** core-ix has joined #rtems 2011-03-10T09:57:04 * @brief Returns current standard timer value in CPU clocks. 2011-03-10T09:57:04 * 2011-03-10T09:57:04 * This function uses Timer 1. 2011-03-10T09:57:04 */ 2011-03-10T09:57:04 static inline unsigned lpc24xx_timer(void) 2011-03-10T09:57:06 { 2011-03-10T09:57:09 return T1TC; 2011-03-10T09:57:11 } 2011-03-10T09:57:14 :D 2011-03-10T09:59:53 bswimley: md5sum for file on my disk matches sourceforge. Have you checked yours? 2011-03-10T10:00:57 No, what's the best way to do that under windows? 2011-03-10T10:01:19 bye.. im off to do some milling with our self built cnc :) thanks for help again 2011-03-10T10:03:51 bswimley: I used cygwin :-D 2011-03-10T10:04:03 at least check the size. 2011-03-10T10:04:15 rokka: good night.. glad the performance is looking better.. did we decide it was about 10 usecs? 2011-03-10T10:05:05 I'll check and get back to you. The size looked about right last nite, but I'll confirm for sure. 2011-03-10T10:05:35 it seems to be 200us now..:l 2011-03-10T10:05:49 with benchmark timer? 2011-03-10T10:05:53 yep.. 2011-03-10T10:06:22 :( tomorrow specs on your board.. you may want to run the tmtests as a reference 2011-03-10T10:06:23 i hope i did the math right .. us = clockcycles * 10 / 576 2011-03-10T10:06:40 good idea! 2011-03-10T10:21:55 *** core-ix has quit IRC 2011-03-10T10:25:37 *** core-ix has joined #rtems 2011-03-10T10:30:39 *** zwj has quit IRC 2011-03-10T10:30:42 Dr Joel: md5 matches on the Fedora14-RTEMS4.11.ova image 2011-03-10T10:34:52 and the import fails? 2011-03-10T10:35:16 Jennifer is needing it.. so she can try it. I may have to re-export it. :( 2011-03-10T10:36:40 Import fails. Could not read OVF file 'Fedora14-RTEMS4.11.ovf' (VERR_TAR_END_OF_FILE). 2011-03-10T10:37:46 I'm using virtual box 4.0.4. Just downloaded it yesterday. 2011-03-10T10:44:09 ok.. will try with Jennifer when she gets back.. may be something wrong with the exported file 2011-03-10T10:44:13 :< 2011-03-10T10:44:35 I agree 2011-03-10T10:44:50 Let me know. I can try a new version. 2011-03-10T11:12:05 *** mmu_man has quit IRC 2011-03-10T11:22:29 *** antgreen has quit IRC 2011-03-10T11:56:02 *** demute has joined #rtems 2011-03-10T11:57:14 *** dr__house` is now known as dr__house 2011-03-10T11:59:14 *** mmu_man has joined #rtems 2011-03-10T12:41:28 *** gedare has joined #rtems 2011-03-10T12:48:38 hi DrJoel 2011-03-10T12:50:05 Anyone in here with much/any experience with the Leon BSPs? 2011-03-10T12:50:22 hey gedare ... me? 2011-03-10T12:50:32 DrJoel: Yup, thought I'd ping the crowd too. :) 2011-03-10T12:51:17 hopefully someone else will pipe up.. I did the sparc port and erc32 BSP. Many hands since then.. :) 2011-03-10T12:51:24 DrJoel: I'm curious about using the Leon 2 or 3 for some research involving RTEMS and hardware 2011-03-10T12:51:48 DrJoel: I'm mostly interested in boards... I have an FPGA which I should be able to synth the core on to, but the functionality will be pretty limited 2011-03-10T12:52:17 Wondering if anyone can point me to some good (hopefully cheapish) boards that are known to work well with RTEMS, and have the soft/FPGA implementation of the Leon 2011-03-10T12:52:55 I haven't found anything that is cheap (to me) but the pender boards appear to be the recommended options.. I think they are in the $1-2K range.. 2011-03-10T12:53:52 yeah that's what I was concerned of :) I'll have to try normal FPGA first, it will probably be enough for me... hopefully not too hard to get RTEMS too bootstrap then 2011-03-10T12:54:42 okie, thanks for info 2011-03-10T13:35:55 in cpukit/posix/src/clockgetres.c there's a case switch with CLOCK_PROCESS_CPUTIME and CLOCK_THREAD_CPUTIME. I get compile error because these aren't defined, where should they be defined? grep -r CLOCK_THREAD_CPUTIME gave no #defines at all... 2011-03-10T13:44:17 /opt/rtems-4.11/sparc-rtems4.11/include/time.h .. it is from newlib. Where did you get your tools? If you built them yourself, you left out the C library 2011-03-10T13:44:42 gedare: Ask on the mailing list. Someone will have a better answer. 2011-03-10T13:45:11 DrJoel: hehe, wops. Yes I built them myself. I guess I will have to do it again. 2011-03-10T13:46:17 when you build gcc .. you have to have newlib under the top gcc directory.. gcc-XXX/newlib 2011-03-10T13:52:38 bswimley: failed to import for Jennifer.. will re-export here 2011-03-10T13:53:22 I had that. I wonder why it did not work... anyway, I will try to do it again and see if it makes any difference, my $PATH was strange some time during the compilation, maybe some hookus pookus happend. 2011-03-10T13:54:32 By the way, are you the joel I have seen in many comments all around the rtems source? :) 2011-03-10T13:55:32 Dr Joel: OK, I will look for the newly available export. Bummer. 2011-03-10T13:56:58 I will import it on another machine this time .. lesson learned 2011-03-10T13:57:18 demute: yep.. I am one of the original developers 2011-03-10T13:57:45 I also am the one who teaches the class ... next in Huntsville in April and Munich in May 2011-03-10T13:58:03 hint.. Munich is lovely in late May, early June LOL 2011-03-10T13:58:08 cool! 2011-03-10T13:59:22 :) 2011-03-10T14:45:15 *** antgreen has joined #rtems 2011-03-10T15:22:28 *** kiwichris has joined #rtems 2011-03-10T15:26:59 *** core-ix has quit IRC 2011-03-10T15:39:43 *** mmu_man has quit IRC 2011-03-10T15:54:09 *** DrJoel has quit IRC 2011-03-10T15:55:26 *** projectgus has joined #rtems 2011-03-10T16:26:41 *** gedare has quit IRC 2011-03-10T16:32:29 *** bswimley has quit IRC 2011-03-10T17:55:31 *** methril__ has quit IRC 2011-03-10T18:06:52 *** methril has joined #rtems 2011-03-10T18:36:41 *** demute has quit IRC 2011-03-10T21:09:18 *** dr__house has quit IRC 2011-03-10T21:42:15 *** kiwichris has quit IRC 2011-03-10T22:24:01 *** dr__house has joined #rtems 2011-03-10T22:24:02 *** dr__house has joined #rtems 2011-03-10T23:16:22 *** dr__house has quit IRC 2011-03-11T00:03:04 *** lcpfnyvc has quit IRC 2011-03-11T00:14:01 *** lcpfnyvc has joined #rtems 2011-03-11T00:16:41 *** projectgus has quit IRC 2011-03-11T02:54:46 *** mmu_man has joined #rtems 2011-03-11T03:25:47 *** mmu_man has quit IRC 2011-03-11T03:48:55 *** methril has quit IRC 2011-03-11T03:49:48 *** methril has joined #rtems 2011-03-11T03:51:38 *** methril has quit IRC 2011-03-11T03:54:08 *** methril has joined #rtems 2011-03-11T04:09:27 *** mmu_man has joined #rtems 2011-03-11T04:28:46 *** mmu_man has quit IRC 2011-03-11T05:00:24 *** mmu_man has joined #rtems 2011-03-11T05:13:29 *** mmu_man has quit IRC 2011-03-11T06:55:15 *** mmu_man has joined #rtems 2011-03-11T07:29:56 *** DrJoel has joined #rtems 2011-03-11T07:29:56 *** ChanServ sets mode: +o DrJoel 2011-03-11T08:00:07 *** antgreen has quit IRC 2011-03-11T08:03:07 *** sebhub has joined #rtems 2011-03-11T08:03:10 hi 2011-03-11T08:04:05 hey sebhub 2011-03-11T08:44:10 *** peerst has joined #rtems 2011-03-11T08:46:32 hi 2011-03-11T08:49:06 hey peerst .. 2011-03-11T08:54:24 hi peer 2011-03-11T08:55:07 *** dr__house has joined #rtems 2011-03-11T08:55:07 *** dr__house has joined #rtems 2011-03-11T09:16:51 *** zwj has joined #rtems 2011-03-11T09:25:20 *** bswimley has joined #rtems 2011-03-11T09:35:19 bswimley: ova is reported to be OK by a virtualpc user. VirtualBox apparently has a bug. .ova can be untar'ed but I am unsure as to how to use the disk image inside it after that. 2011-03-11T09:35:51 It may work fine with VMWare Player 2011-03-11T09:44:37 *** sebhub has quit IRC 2011-03-11T09:47:56 hello :) 2011-03-11T09:56:00 DrJoel: Hi, joel. About make rtems run HyperVisor, is there any news? 2011-03-11T09:57:51 Dr Joel: I can try VMWare. I saw the blurb about "untarring" as well but I was also unsure how to do that. Let me play around a bit and get back to you. 2011-03-11T10:00:46 someone (xi yang?) has it on L4 but nothing other than that that I am aware of 2011-03-11T10:05:22 yeah, i also saw some thing about rtems on l4, but what is the real value of rtems hypervisor? 2011-03-11T10:06:45 if rtems run on l4, it has nothing better thing to rtems, it just show L4 can support Virtualization 2011-03-11T10:09:20 I mean what's the technology direction of rtems virtualization. add a RT-VMM? or other methods? 2011-03-11T10:11:59 ARINC-653 is the ultimate goal. Whether that turns RTEMS itself into the Hypervisor or makes RTEMS a smart "client" inside multiple partitions so you can have separate RTEMS apps in different time/space partitions communicatins 2011-03-11T10:17:59 the first method make rtems itself into the Hypervisor is clean way if it is able to realize. the second method that need a seperate Hypervisor or like L4 to do it, if rtems can run l4 it is will be ok 2011-03-11T10:19:17 but the hypervisor needs to have arinc-653 capabilities or it doesn't address the requirements of IMO most people who would use a hypervisor in a real-time system 2011-03-11T10:21:04 IMO? 2011-03-11T10:23:59 Does rtems now have arinc-653 capabilities? 2011-03-11T10:25:00 In My Opinions 2011-03-11T10:26:25 I am not that familiar with the standard but the basics are space (e.g. memory protection (so no)) and time separation. We don't really have those. 2011-03-11T10:26:31 There are other capabilities like sync and communication between the "partitions" which could be grown from current capabilities 2011-03-11T10:26:56 aanjhan worked on MMU support so might have more insight 2011-03-11T10:30:22 yeah, the L4 hypervisor can support space partition, not sure time separation. i think make rtems into a RT-VMM is direction 2011-03-11T10:31:47 it would be a fun and interesting project. :-D 2011-03-11T10:34:49 yeah, recently i collect some resources i think it is able to realize. but there are some components that is diffcult 2011-03-11T10:35:54 If you have a game plan or even an outline of one, please be encouraged to grow the Wiki Open Projects page. SMP is a huge task but by breaking it into a series of smaller ones, it looks like 4.11 might actually be SMP 2011-03-11T10:36:17 Partition Management,Process Management,Time Management,partition Communication 2011-03-11T10:38:02 yeah, it is really. open source is really magic. who is the expert about this topic? 2011-03-11T10:38:19 i think i can get some help from them 2011-03-11T10:40:00 I would ask on the list. I don't know who is the list. 2011-03-11T10:40:17 If you have good references or know where a copy of the standard is, I am interested 2011-03-11T10:41:29 *** bswimley has quit IRC 2011-03-11T10:43:25 OK , sure. tomorrow i will classify them and discuss. now it is really later for me. 0:42 in my time zone, seen you 2011-03-11T10:43:37 good night.. have a nice weekend 2011-03-11T10:45:27 now i have run tests tm01-tm29 2011-03-11T10:45:58 how do they look? Do the numbers look reasonable? 2011-03-11T10:45:59 some of them don't go to end :l 2011-03-11T10:47:39 tests that don't end: 3 5 7 11 12 13 14 16 19 20 22 24 26 27 2011-03-11T10:48:22 http://pastebin.com/K01usVhU 2011-03-11T10:48:54 copied results to pastebin 2011-03-11T10:49:49 *** zwj has quit IRC 2011-03-11T10:50:53 rokka: is console interrupt driven? Looks like it since some output didn't make it out.. I think the tests worked, just quit before all output was out 2011-03-11T10:51:11 So we think they are cycles.. so 753 is how much time? 2011-03-11T10:52:15 13.07us? 2011-03-11T10:52:27 753/56.7=13.07 2011-03-11T10:54:33 oh.. 57.6 :) 2011-03-11T10:54:34 rtems_semaphore_obtain: not available -- caller blocks 158 (useconds) on 14 Mhz ERC32 2011-03-11T10:54:50 you have 5115/57.6 for same test 2011-03-11T10:55:07 90 usecs 2011-03-11T10:55:55 strange?:l 2011-03-11T10:56:54 90 vs 158 ... using clock ratio of 14/56.7 * 158 gives me about 39. So your times are about twice what one would expect extrapolating pure clock speed and ignoring architectural differences 2011-03-11T10:57:08 Does the board have a 16-bit CPU bus? 2011-03-11T10:57:13 think so 2011-03-11T10:57:27 that would likely explain it. 2011-03-11T10:57:51 *** mmu_man has quit IRC 2011-03-11T10:57:54 takes twice as long to fetch the code :-D 2011-03-11T10:58:00 so explainable 2011-03-11T10:59:50 ok 2011-03-11T10:59:54 so nothing to worry?:) 2011-03-11T11:00:16 I don't think so. Worth a double check with Sebastian but they look OK based on clock speed and bus width. 2011-03-11T11:20:38 DrJoel: Hello/hola :-) 2011-03-11T11:21:55 What is the position of rtems with respect about Bus mastering, i mean the os is suposed to be real-time. 2011-03-11T11:23:48 May be my question is not clear, my , real interest is find a recomendation for codying style related to a driver wich is a master in the bus 2011-03-11T11:24:55 hey kristianpaul .. I personally don't have much on that.. good question for the list.. in general, every time you schedule access to a resource you are potentially introducing a priority inversion. 2011-03-11T11:25:01 Good discussion question for the list :-D 2011-03-11T11:25:07 ok :-) 2011-03-11T11:28:33 *** mmu_man has joined #rtems 2011-03-11T11:29:05 *** bswimley has joined #rtems 2011-03-11T11:30:29 *** core-ix has joined #rtems 2011-03-11T11:36:35 DrJoel: Im leaving for today. Have a nice weekend. :) 2011-03-11T11:37:22 bye! have a nice weekend 2011-03-11T11:38:02 thanks:) 2011-03-11T11:41:12 *** dr__house` has joined #rtems 2011-03-11T11:41:12 *** dr__house has quit IRC 2011-03-11T11:41:12 *** dr__house` has quit IRC 2011-03-11T11:41:12 *** dr__house` has joined #rtems 2011-03-11T11:41:30 *** dr__house` is now known as dr__house 2011-03-11T11:43:32 dr__house: back to 100% coverage on Ospd!! 2011-03-11T11:43:53 DrJoel: Wow! Awesome! :D 2011-03-11T11:44:29 Most of OsPd are from the aio GSoC project last summer .. need to ping him again. He has submitted more test code a couple of times 2011-03-11T11:44:48 DrJoel: oh cool 2011-03-11T11:45:10 how is school going for you? 2011-03-11T11:45:46 DrJoel: pretty hectic, the spring break was just for namesake. A couple of homeworks and a mid term coming up :P 2011-03-11T11:48:25 my son's spring break starts the 18th think. High school spring break is next month.. and shortened to get some snow days made up 2011-03-11T11:49:25 oh ok 2011-03-11T11:51:27 *** dr__house` has joined #rtems 2011-03-11T11:51:27 *** dr__house has quit IRC 2011-03-11T11:51:27 *** dr__house` has quit IRC 2011-03-11T11:51:28 *** dr__house` has joined #rtems 2011-03-11T11:51:55 *** dr__house` is now known as dr__house 2011-03-11T11:58:49 DrJoel: I used winRAR to untar the OVA into an ovf and a disk image and was able to get the RTEMS-4.11 image to work with virtualbox. 2011-03-11T11:59:24 Speaking of spring break - I'm gone next week for spring break with my family. 2011-03-11T12:01:27 bswimley: great! What did you do after you untar'ed it? 2011-03-11T12:02:27 I had the ovf and the vmdk image side by side in a directory. I just imported the ovf file with virtual box, it took a while to read the disk image, and away I went. 2011-03-11T12:03:01 You may want to see if Jennifer can make this work as well. 2011-03-11T12:03:03 ok. I tried to attach the disk image to a new VM. that is the difference 2011-03-11T12:03:25 OK 2011-03-11T12:03:30 I will Monday. Her fence was damaged in the recent storm here and her dogs are getting out. So she went home early today LOL 2011-03-11T12:03:54 big spring break plans? 2011-03-11T12:08:32 Going to Washington DC with the kids. They're old enough to appreciate it and we have some friends there. We leave tomorrow. How about you? 2011-03-11T12:10:14 my kids are now spread across different schools with different holidays. I suppose I will do something with them local but no big plans.. 2011-03-11T12:10:32 I have a couple years before that starts :) 2011-03-11T12:10:34 really hopign the Munich class in May comes together. 2011-03-11T12:11:10 I would love to go to that, but not this year. I really enjoyed the one time I was in Munich. 2011-03-11T12:11:12 Youngest is 16.. oldest will be 22. 2 colleges, 1 high school, and one not in school.. 2011-03-11T12:11:27 I love Munich. It is really a nice place. 2011-03-11T12:11:43 Wow. I have 15, 12, and 10. Freshman in HS is my oldest. 2011-03-11T12:12:06 I went to Munich for a trade show in November. Pretty cold when we were there. Still enjoyed it a lot. 2011-03-11T12:12:10 parenting changes as they get older.. and they start staying up later than you do. 2011-03-11T12:12:33 My bedtime comes pretty early... 2011-03-11T12:13:14 I've taught the class there 8 times I think. Each time I manage to squeeze out 1-2 days of sightseeing. If the class comes together this time, we are considering it as a family trip and taking a train to Prague for a couple of days 2011-03-11T12:13:48 when the class is in Oct, it is usually scheduled to be the week after Oktoberfest. So hotels are cheaper and if you come in a few days early, you can go. :) 2011-03-11T12:14:06 That's what I'd like to do! 2011-03-11T12:16:10 Thomas mentioned trying for a fall class.. will wait until this one is done before investigating 2011-03-11T12:20:46 I usually need PE credits. The Munich class would be a great opportunity. 2011-03-11T12:21:24 The tran to Prague would be great. I took one from Frankfurt to Rothenburg (sp?) 2011-03-11T12:23:31 My wife wants to see more than we have in the past :-D 2011-03-11T12:23:51 gotta run for a haircut! 2011-03-11T12:35:48 *** mmu_man has quit IRC 2011-03-11T13:05:30 *** dr__house` has joined #rtems 2011-03-11T13:05:31 *** dr__house` has joined #rtems 2011-03-11T13:07:29 *** dr__house has quit IRC 2011-03-11T13:19:19 *** core-ix has quit IRC 2011-03-11T13:26:53 *** mmu_man has joined #rtems 2011-03-11T13:31:46 *** bswimley has quit IRC 2011-03-11T13:56:19 hmm. 2011-03-11T13:56:31 ubuntu + qemu + i386 rtems 2011-03-11T13:56:40 easy+ 2011-03-11T13:56:41 ? 2011-03-11T13:58:24 yes dr__house` and aanjhan have done it :D 2011-03-11T13:58:25 you become responsible for building your own RTEMS tools. 2011-03-11T13:58:40 Ralf has mentioned you should be able to unarchive the RPMs for a comparable fedora/centos and use those but I dont know for sure 2011-03-11T13:59:03 *** dr__house` has quit IRC 2011-03-11T13:59:11 we have talked about a .deb format and respository for years but it hasn't gotten the push to bring it online 2011-03-11T14:04:56 ok.. but does i386 require something special 2011-03-11T14:05:10 for arm i must compile cross-compiler 2011-03-11T14:05:24 yes.. you always need the rtems tools.. special c library and other settings 2011-03-11T14:08:26 ok 2011-03-11T14:08:59 i'll try to adapt my arm help for building tools for i386 2011-03-11T14:09:55 should be able the same with arm->i368 and lpcxxx->pc38 2011-03-11T14:09:59 pc386 I mean 2011-03-11T14:10:19 DrJoel: aanjhan has done what? :P 2011-03-11T14:12:29 built tools for ubuntu :-D 2011-03-11T14:12:43 and mmu work .. two references while you were asleep... 2011-03-11T14:13:03 DrJoel: so did we submit the app for this year GSoC? 2011-03-11T14:13:05 aanjhan: if the munich class happens in May, any chance of you making a visit? 2011-03-11T14:13:11 DrJoel: sure 2011-03-11T14:13:21 Munich is 4hrs from Zurich. 2011-03-11T14:13:23 yes.. Chris reviewed it .. 2011-03-11T14:13:24 I hope the train :) 2011-03-11T14:13:30 hop* 2011-03-11T14:13:36 so you are still there... 2011-03-11T14:13:48 Michele is curious about a couple of days in Prague before the class 2011-03-11T14:13:50 sigh. Apparently you cant finish off PhD in 6 months :P: 2011-03-11T14:14:32 unless they lower the class requirements, and residency LOL 2011-03-11T14:14:46 hehe 2011-03-11T14:14:51 so easily 3 more years here 2011-03-11T14:15:00 or until my Prof kicks me out 2011-03-11T14:15:05 whichever is earlier :P 2011-03-11T14:15:45 rokka: I have had enough experience in getting things to work on Ubuntu. Just ask 2011-03-11T14:15:54 especially the toolchain 2011-03-11T14:23:03 * DrJoel waves bye.. calling it a day early .. 2011-03-11T14:23:06 *** DrJoel has quit IRC 2011-03-11T14:56:55 *** core-ix has joined #rtems 2011-03-11T15:25:04 *** dr__house has joined #rtems 2011-03-11T15:25:04 *** dr__house has joined #rtems 2011-03-11T15:35:13 hmm.. :l my i386 compiler tools failed to compile rtems.. 2011-03-11T15:35:20 something wrong with linking 2011-03-11T16:53:00 *** dr__house has quit IRC 2011-03-11T16:56:48 *** DrJoel has joined #rtems 2011-03-11T16:56:48 *** DrJoel has joined #rtems 2011-03-11T16:56:48 *** ChanServ sets mode: +o DrJoel 2011-03-11T16:56:59 *** DrJoel has quit IRC 2011-03-11T17:00:20 *** methril has quit IRC 2011-03-11T17:58:33 *** mmu_man has quit IRC 2011-03-11T23:34:45 *** core-ix has quit IRC 2011-03-12T00:04:46 *** kristianpaul has quit IRC 2011-03-12T00:16:35 *** kristianpaul has joined #rtems 2011-03-12T01:39:42 *** antgreen has joined #rtems 2011-03-12T04:44:31 *** core-ix has joined #rtems 2011-03-12T06:18:03 *** dr__house has joined #rtems 2011-03-12T07:50:03 *** mmu_man has joined #rtems 2011-03-12T08:19:09 *** methril has joined #rtems 2011-03-12T10:10:39 *** zwj has joined #rtems 2011-03-12T11:05:04 *** core-ix has quit IRC 2011-03-12T11:25:09 *** zwj has quit IRC 2011-03-12T15:37:04 *** mmu_man has quit IRC 2011-03-12T15:38:57 *** mmu_man has joined #rtems 2011-03-12T16:02:09 *** peerst has left #rtems 2011-03-12T17:02:24 *** dr__house has quit IRC 2011-03-12T18:12:46 *** mmu_man has quit IRC 2011-03-12T18:35:56 *** core-ix has joined #rtems 2011-03-12T19:39:19 *** dr__house has joined #rtems 2011-03-12T19:39:19 *** dr__house has joined #rtems 2011-03-13T01:29:24 *** dr__house has quit IRC 2011-03-13T07:13:40 *** methril has quit IRC 2011-03-13T07:15:18 *** methril has joined #rtems 2011-03-13T08:53:37 *** mmu_man has joined #rtems 2011-03-13T09:24:45 *** madrazr has joined #rtems 2011-03-13T09:49:20 *** aanjhan has quit IRC 2011-03-13T10:36:54 *** mmu_man has quit IRC 2011-03-13T10:48:55 *** madrazr has quit IRC 2011-03-13T12:09:58 hi ! 2011-03-13T12:10:05 I have a weird problem in my network driver 2011-03-13T12:10:12 I put a malloc(size) 2011-03-13T12:10:18 included stdlib.h 2011-03-13T12:10:40 compilation says : error : macro malloc requiers 3 arguments, you only provided one 2011-03-13T13:25:56 *** mmu_man has joined #rtems 2011-03-13T18:19:57 *** mmu_man has quit IRC 2011-03-13T20:44:45 *** dr__house has joined #rtems 2011-03-13T22:48:10 *** dr__house has quit IRC