2011-04-04T00:22:03 *** sqm has joined #rtems 2011-04-04T01:02:44 *** projectgus has quit IRC 2011-04-04T02:22:55 *** core-ix has quit IRC 2011-04-04T02:31:36 *** sebhub has joined #rtems 2011-04-04T02:31:45 good morning 2011-04-04T02:57:51 hi sebhub 2011-04-04T03:01:31 *** core-ix has joined #rtems 2011-04-04T03:15:19 *** ivo__ has joined #rtems 2011-04-04T03:15:22 *** eddy has joined #rtems 2011-04-04T03:22:53 *** core-ix has quit IRC 2011-04-04T03:22:53 *** sebhub has quit IRC 2011-04-04T03:22:53 *** sqm has quit IRC 2011-04-04T03:22:53 *** lekernel_ has quit IRC 2011-04-04T03:22:55 *** methril has quit IRC 2011-04-04T03:22:57 *** kristianpaul has quit IRC 2011-04-04T03:23:06 *** kristianpaul has joined #rtems 2011-04-04T03:23:30 *** sebhub has joined #rtems 2011-04-04T03:24:00 *** eddy has quit IRC 2011-04-04T03:25:17 *** ivo__ is now known as core-ix 2011-04-04T03:30:05 *** lekernel_ has joined #rtems 2011-04-04T03:30:19 *** sqm has joined #rtems 2011-04-04T03:30:45 *** cdcs has joined #rtems 2011-04-04T03:32:10 *** methril has joined #rtems 2011-04-04T03:36:53 *** sqm has quit IRC 2011-04-04T03:42:02 *** sqm has joined #rtems 2011-04-04T04:02:47 *** gibson has joined #rtems 2011-04-04T04:02:54 *** gibson has joined #rtems 2011-04-04T04:12:29 *** verm__ has quit IRC 2011-04-04T04:40:15 *** eddy has joined #rtems 2011-04-04T04:40:29 *** sqm has quit IRC 2011-04-04T04:40:29 *** sebhub has quit IRC 2011-04-04T04:40:30 *** antgreen has quit IRC 2011-04-04T04:42:56 *** antgreen has joined #rtems 2011-04-04T04:45:04 *** sebhub has joined #rtems 2011-04-04T05:27:25 *** lekernel_ is now known as lekernel 2011-04-04T05:34:14 *** cdcs has quit IRC 2011-04-04T05:35:30 *** cdcs has joined #rtems 2011-04-04T06:05:59 *** titankiller has joined #rtems 2011-04-04T06:12:00 *** titankiller has quit IRC 2011-04-04T06:13:22 *** methril has quit IRC 2011-04-04T06:25:59 *** methril has joined #rtems 2011-04-04T06:37:34 *** lekernel has quit IRC 2011-04-04T06:37:34 *** methril has quit IRC 2011-04-04T06:37:59 *** methril has joined #rtems 2011-04-04T06:38:29 *** lekernel has joined #rtems 2011-04-04T07:54:54 *** xiangfu has joined #rtems 2011-04-04T08:16:20 *** gibson is now known as verm__ 2011-04-04T08:16:27 *** verm__ has joined #rtems 2011-04-04T08:23:13 *** zwj has joined #rtems 2011-04-04T08:59:30 *** DrJoel has joined #rtems 2011-04-04T08:59:30 *** ChanServ sets mode: +o DrJoel 2011-04-04T08:59:30 * DrJoel waves hello 2011-04-04T09:00:40 hi DrJoel 2011-04-04T09:03:33 hi 2011-04-04T09:06:45 *** eddy has quit IRC 2011-04-04T09:07:39 *** eddy has joined #rtems 2011-04-04T09:09:31 *** eddy has quit IRC 2011-04-04T09:12:05 *** sqm has joined #rtems 2011-04-04T09:15:01 good morning 2011-04-04T09:29:24 hi 2011-04-04T09:29:50 hey sebhub 2011-04-04T09:33:19 hello 2011-04-04T09:34:43 *** easwar has joined #rtems 2011-04-04T09:36:10 DrJoel: i got io-language vm ported, it works on i386-rtems4.11 2011-04-04T09:36:43 i got ncurses runnig too, i need it for my irssi-like ui idea 2011-04-04T09:37:18 but the inputs for ncurses does not work perfectly.. i must either fix it or disable it and use keyboard input from rtems shell 2011-04-04T09:37:41 rtems sell is btw. missing PgUp and PgDown buttos which i would need.. but its not hard to add 2011-04-04T09:37:46 shell.. :D 2011-04-04T09:40:09 DrJoel, status update: I got RTEMS Hello World built and working yesterday, with some help from projectgus 2011-04-04T09:40:58 cool! So did you put yourself in the list of students doing proposals on the wiki? And.. send me the diff and screen schot 2011-04-04T09:40:59 shot 2011-04-04T09:42:14 nope, doing that now :) 2011-04-04T09:42:46 good night 2011-04-04T09:43:10 DrJoel: hi, have you review my proposal? 2011-04-04T09:44:14 DrJoel: who do you think i should invite to comment my proposal? 2011-04-04T09:44:14 not yet.. it was a pretty weekend here and we did yardwork :-D 2011-04-04T09:44:52 Anyone you like who is not a student this year. :-) 2011-04-04T09:44:55 hehe, now it is night monday in my time zone 2011-04-04T09:45:11 anyone on here want to help review ? 2011-04-04T09:45:40 my proposal is about Hypervisor for RTEMS 2011-04-04T09:46:01 i will be make my proposal be accessable for every one 2011-04-04T09:47:07 *** gedare has joined #rtems 2011-04-04T09:48:45 *** lj8175 has joined #rtems 2011-04-04T10:02:42 review what? 2011-04-04T10:05:54 students' gsoc project proposals.. 2011-04-04T10:08:14 ok, where? 2011-04-04T10:11:34 They are in google docs so we can comment and help correct grammar, etc. .. hold on and I will add you to the access list 2011-04-04T10:12:08 my email is not the same 2011-04-04T10:20:23 joel: is it possible to run the gcc test suite on the arm gdb simmulator? 2011-04-04T10:23:21 yes.. it should be.. I have been running on skyeye though. Shuldn't be difficult.. just a matter of adding a dejagnu file and a couple of lines to some scripts 2011-04-04T10:24:00 skyeye is also ok, is there something in rtems-testing? 2011-04-04T10:24:19 i want to test my arm eabi gcc patch 2011-04-04T10:25:26 yes.. the scripts are in rtems-testing/gcc. I recommend a double check that edb7312 works for you on your skyeye before running all the tests. :-D 2011-04-04T10:26:30 ok, i never run the gcc test suite before 2011-04-04T10:27:16 Thank you , I know how to run also :) 2011-04-04T10:28:40 the main scripts are test_driver which can loop over all targets and "do_one". You can use do_one to just build gcc for a particular language and test that. 2011-04-04T10:29:36 but you need to let the script build the gcc (w/newlib in source tree). The names of the build directories are pretty rigid so you can keep them around and avoid conflicts 2011-04-04T10:30:02 DrJoel, seems I'm an idiot and can't find directions on creating the diff screenshot like Vincent sent in 2011-04-04T10:30:12 zwj, have you talked with Tobias about htis project? 2011-04-04T10:30:18 this* 2011-04-04T10:30:36 any pointers? 2011-04-04T10:30:39 :( 2011-04-04T10:30:59 ok, thanks 2011-04-04T10:31:35 cdcs: yes, i have discussed with Tobias about this project 2011-04-04T10:31:50 he thinks its feasible? 2011-04-04T10:32:21 yeah, he think it is a good idea and he also give me some advise 2011-04-04T10:33:24 the main work of my propoal is to port linux to work with RTEMS which AIR do not implement 2011-04-04T10:33:55 yes, i have read your proposal 2011-04-04T10:34:33 thank you, and could you give me some comments 2011-04-04T10:34:40 you will need some time to get used to PMK 2011-04-04T10:35:24 PMK? 2011-04-04T10:36:14 Partition Management Kernel == Stripped version of RTEMS that works has a Hypervisor 2011-04-04T10:36:16 yeah, this is a really challenge, their development enviroment is different 2011-04-04T10:36:52 and there is also some toolchain patch for it 2011-04-04T10:37:23 so my first work is to be familar with the development enviroment 2011-04-04T10:37:48 it's not very different. You can compile and run the samples easily 2011-04-04T10:38:19 I want to make sure you understand that I really would prefer that their stripped RTEMS find its way as a variant in the main tree. 2011-04-04T10:38:40 Otherwise, this is more of less a divergent project that isn't directly beneficial to RTEMS 2011-04-04T10:39:16 cdcs: ok, do you have some document about how to build the development environment inculde build and debug, simulator 2011-04-04T10:40:04 DrJoel, If the license is compatible, merging with rtems will not be difficult 2011-04-04T10:41:00 debug, simulator == sparc-rtems4.8-gdb, tsim-leon 2011-04-04T10:42:52 tsim-leon uses evaluation version well? 2011-04-04T10:43:38 the build enviroment is similar to any rtems application; you just need some extra "exports xxx=xxxx\xxxx\" 2011-04-04T10:44:24 the evaluation version of tsim-leon3 is completly fine. 2011-04-04T10:44:43 zwj, so you are doing a AIR compatible hypervisor of a stripped version of RTEMS and make normal (or modified) RTEMS and gnu/linux run on it or what :)? 2011-04-04T10:48:52 ok.. stripped RTEMS which is hypervisor already exists? 2011-04-04T10:49:05 and it can run RTEMS? 2011-04-04T10:49:06 cdcs: OK.. just don't want divergence 2011-04-04T10:49:42 yes. AIR is an RTEMS based hypervisor that runs several RTEMS as partitions 2011-04-04T10:50:15 and this project is mainly about making linux run on AIR too ? 2011-04-04T10:50:36 and becaus AIR has RTEMS this is RTEMS project :) 2011-04-04T10:50:47 rokka: hypervisor use what AIR has developed, AIR also modified mult-RTEMS to run under this hypervisor, my work is to modify linux to run with RTEMS under this hypervisor 2011-04-04T10:51:10 yeh 2011-04-04T10:51:17 if it's merged to RTEMS main tree, it brings advantages to RTEMS 2011-04-04T10:51:35 i think its very cool project 2011-04-04T10:51:55 yeah, it can make RTEMS a configureable hypervisor 2011-04-04T10:52:28 do you know this ? https://sites.google.com/site/realtimexen/ 2011-04-04T10:52:32 yeah, i think so. and it could be a long-term project 2011-04-04T10:52:55 rokka, yeah i have hear about it 2011-04-04T10:53:09 *** sqm has quit IRC 2011-04-04T10:53:22 is it ARINC 653 compatible ? will it do same as this AIR ? 2011-04-04T10:53:48 i really don't know which are ARINC 653 requirements yet.. :) 2011-04-04T10:54:14 realtime Xen is doing some useful work, but Xen is not target at realtime. so there is some inapproporate to embedded system 2011-04-04T10:54:34 ok.. RT-XEN is only for PC? 2011-04-04T10:54:46 no it is not ARINC 653 compatiable 2011-04-04T10:54:46 zwj, as DrJoel said i think it is best for you to reserve some time for research how to merge PMK with RTEMS main tree. 2011-04-04T10:55:06 now it seems that RT-XEN is target at ARM 2011-04-04T10:55:49 *** xiangfu has quit IRC 2011-04-04T10:56:56 cdcs, yeah. this is also a work for me to do. if PMK is merged to RTEMS and make RTEMS configureable to hypervisor or native RTEMS is good 2011-04-04T10:58:13 *** sqm has joined #rtems 2011-04-04T10:58:36 i will contract with Tobias soon to make the code more early to dilever to me 2011-04-04T10:58:38 zwj: is the xen code a good example of how to get RTEMS hypervisor onto two architectures? 2011-04-04T10:58:41 which platform does this AIR support? 2011-04-04T10:59:39 rokka, sparc 2011-04-04T10:59:47 of course.. 2011-04-04T11:00:18 drjoel: Xen is only a hypervisor which is not the same as RTEMS. but i think linux is a good example for us 2011-04-04T11:00:48 but the mechanical issues at the architecture level should give hints as to how ARM support for AIR would be implemented 2011-04-04T11:00:49 soon i have my own leon3 evkit, i would like to test AIR :) 2011-04-04T11:01:25 because now linux has maintain a paravirt_ops which can make the same linux image work as paravirtured OS or native OS 2011-04-04T11:01:41 rokka, yes. it runs on real sparc HW :) 2011-04-04T11:02:35 hmm.. how does rtems support multiple cpu? 2011-04-04T11:03:17 or does multiprocessor mean multiple arch? 2011-04-04T11:04:15 * rokka is asking stupid questions today.. :) 2011-04-04T11:04:21 bye 2011-04-04T11:04:52 *** sebhub has quit IRC 2011-04-04T11:04:52 AIR add MMU support to RTEMS. i think this is main work to implement ARM AIR 2011-04-04T11:06:52 zwj: you may want to define the contribution first as merging PMK. Clear all licensing problems first 2011-04-04T11:07:12 Then show how to run Linux on RTEMS-PMK 2011-04-04T11:07:41 multiprocessor means multiple processors connected over a system bus 2011-04-04T11:07:46 look for...VME bus 2011-04-04T11:08:19 btw, why are LM32 and H8300 the only archs which still have a 4.5.2 toolchain instead of 4.6 in the RTEMS FTP? 2011-04-04T11:08:20 rokka: rtems current support is to put complete RTEMS images on separate processors, each act independently and communicate over shared bus 2011-04-04T11:09:09 rokka: the rtems-smp project is proceeding in parallel to mainline development to try to bridge the gap and allow 1 rtems image to span multiple cpus 2011-04-04T11:10:07 gedare: yeah. this should be confirmed otherwise all the contribution will not be progressed 2011-04-04T11:10:12 lekernel:I know lm32 doesn't build for 4.6.0.. probably h8 as well 2011-04-04T11:11:02 zwj: if you get the license cleared before closing your gsoc proposal, then merging PMK can be an intermediate step on your GSOC (not sure how hard it is, figure that from tobias) 2011-04-04T11:11:26 DrJoel: about merging the PMK to RTEMS tree, could you help to contract with Tobias to comfirm? 2011-04-04T11:11:31 zwj: smp progress is being merged as it is available. Jennifer and I have merged all RTEMS (cpukit) changes we have so far. We are working on leon3 and pc386 BSPs now .. to get the other cores out of reset and force them to make context switches by hand. 2011-04-04T11:11:34 IMO adding support to run Linux on AIR is more of a "bonus"...merging PMK itself is the real thing to do 2011-04-04T11:12:15 but right now it is in lgpl, which might be a concern depending on how it is built... 2011-04-04T11:12:17 zwj: email us both with the question of what can be done to get that into the RTEMS tree so it moves forward with RTEMS 2011-04-04T11:12:49 good morning drjoel :) 2011-04-04T11:12:52 well, almost noon there, and afternoon for me 2011-04-04T11:13:01 gedare: good morning.. 2011-04-04T11:13:34 i think licensing will not be an issue. But yes, you should merge it with rtems tree first 2011-04-04T11:14:16 ok, for the lgpl i will continue to email Tobias to confirm this question 2011-04-04T11:14:54 gedare: ok:) 2011-04-04T11:15:21 zwj: you might get the most headway if you ask if they will relicense it to use the rtems license directly. 2011-04-04T11:16:18 DrJoel: so amd64 cpu with multiple cores could work for rtems in near future? 2011-04-04T11:16:33 gedare: good suggestion.. and how do we get this code into the rtems tree 2011-04-04T11:16:44 and i do not think of linux as a bonus. Running linux alongside rtems, while maintaining real time capabilities is very useful 2011-04-04T11:17:13 DrJoel: the C compiler builds... the C++ compiler doesn't 2011-04-04T11:17:14 how can they communicate to each other? virtual ethernet? 2011-04-04T11:17:28 DrJoel: but the C++ compiler was already broken in 4.5.x so I'd say just disable it 2011-04-04T11:17:42 it built, but it produced broken code later on 2011-04-04T11:18:11 afaik only gcc 4.4 + lattice patches has a semi-acceptable C++ compiler 2011-04-04T11:20:39 lekernel: not building and not generating good code are different :) 2011-04-04T11:21:03 not building angers ralf... if no one uses it then it doesn't matter if it is broken 2011-04-04T11:21:18 oh, ok, then we can replace the c++ compiler with a "hello world" program 2011-04-04T11:21:20 easy :) 2011-04-04T11:21:44 it's just a matter of removing c++ in the --enable-languages list when building gcc 2011-04-04T11:22:01 bye . have to go to bed 2011-04-04T11:23:04 bye 2011-04-04T11:25:03 bye 2011-04-04T11:26:04 *** lj8175 has quit IRC 2011-04-04T11:43:39 is ncurses license (MIT) rtems compatible? 2011-04-04T11:44:01 im developing this irssi-like ncurses gui for rtems 2011-04-04T11:49:14 rokka: yes.. 2011-04-04T11:49:23 what version are you using? 2011-04-04T11:49:27 latest 2011-04-04T11:49:57 5.8? 2011-04-04T11:50:11 did it build easily or need patches? We have rtems-addon-packages and I think it has 5.3 .. would be good to update it 2011-04-04T11:50:23 some hacking:D 2011-04-04T11:50:37 hmmm... or even better just have instructions/patches on building it 2011-04-04T11:50:53 input from keyboard does not work with special keys like arrow.. even with keypad(sdtscr,TRUE); 2011-04-04T11:51:27 i can make wiki page if i get input working 2011-04-04T11:51:34 is that a problem in the pc386 bsp? 2011-04-04T11:51:39 i should check the previous patch 2011-04-04T11:51:47 i have tested only pc386 2011-04-04T11:52:06 same problem with qemu and native 2011-04-04T11:53:14 do you think its possible to have ncurses graphics work in pc386 textmode just after rtems has booted? 2011-04-04T11:53:26 of course it works with serial mode with linux terminal 2011-04-04T11:55:57 i mean vga text mode.. 2011-04-04T11:55:59 i guess 2011-04-04T11:57:05 hmm. i would also like to put ncurses terminal to telnet 2011-04-04T11:57:13 haven't tested it yet 2011-04-04T12:00:23 *** sqm has quit IRC 2011-04-04T12:02:47 may be some restriction on how the keyboard parser sees the chars to pass them up 2011-04-04T12:03:51 *** zwj has quit IRC 2011-04-04T12:05:39 could be 2011-04-04T12:06:18 i think its bad that ncurses doesn't have api to disable input, now its blocking it so i can't uses fgetc :( 2011-04-04T12:06:40 but of course i can add my on function for it 2011-04-04T12:06:44 own.. 2011-04-04T12:10:29 *** antgreen has quit IRC 2011-04-04T12:10:41 is it ok to use printf from different tasks? 2011-04-04T12:12:10 given the mess I get when I do that, I guess no 2011-04-04T12:13:37 it is ok but since you are sharing a single output device, lines tend to get mixed up. Using printk or stderr should improve thigns 2011-04-04T12:29:50 DrJoel: what would it take to support ncurses graphics on vga textmode? linux does that somehow.. 2011-04-04T12:35:34 not know how it works... I assume they print a special character which just needs to be in the font we use for vgacons 2011-04-04T12:40:07 *** easwar has quit IRC 2011-04-04T12:44:37 *** easwar has joined #rtems 2011-04-04T12:44:37 *** easwar has joined #rtems 2011-04-04T12:50:29 *** easwar has left #rtems 2011-04-04T12:54:42 *** cdcs has quit IRC 2011-04-04T12:58:02 DrJoel: ok 2011-04-04T13:22:19 gedare: still awake? 2011-04-04T13:32:27 drjoel: hi 2011-04-04T13:32:48 drjoel: just working on my thesis proposal. a lot of writing. 2011-04-04T13:34:00 :D just wanted to say that I am pushing the red/black tree code in.. running tests now 2011-04-04T13:34:14 thanks for reading the draft.. 2011-04-04T13:34:45 cool. I always meant to try constructing a test case that I knew hit all of the cases of red-black tree..but I never did 2011-04-04T13:35:11 yep, glad to 2011-04-04T13:35:26 fyi, i plan to go to FSW 2011-04-04T13:35:55 saw that .. it is a nice workshop .. great presentations.. 2011-04-04T13:36:09 well with a coverage run we will see what your sp test misses 2011-04-04T13:36:14 yeah, i hope to learn something and schmooze potential employers ;) 2011-04-04T13:37:06 :D 2011-04-04T14:08:55 cool. now i have to decide if i should work on retargeting my edf, or see what the person on the ml comes up with. 2011-04-04T14:09:05 i like the latter approach /lazy 2011-04-04T14:09:54 lol... 2011-04-04T14:15:29 good evening :) 2011-04-04T14:15:53 DrJoel: how are you? 2011-04-04T14:17:33 doing well .. would be even better if my code didn't go off in the weeds 2011-04-04T14:18:50 hm? :) 2011-04-04T14:19:27 *** mw1 is now known as mwalle 2011-04-04T14:20:42 trying to debug get the second core on an x86 to go to its idle thread. 2011-04-04T14:25:48 on real hardware? 2011-04-04T14:28:52 DrJoel: do you know somebody who maintains gdb 2011-04-04T14:30:25 mwalle: qemu.. which is some of the problem I think.. it runs a LONG time on one cpu before switching to the other 2011-04-04T14:30:37 mwalle: not personally.. just mailing list.. bug or time to submit? 2011-04-04T14:31:09 bug :) 2011-04-04T14:31:43 i'm filing a bug atm, still waiting for registration mail 2011-04-04T14:34:27 just email the list.. they are pretty responsive 2011-04-04T14:35:22 even for lm32 bugs? 2011-04-04T14:39:05 gdb and lm32 bugs? :) 2011-04-04T14:42:34 *** cdcs has joined #rtems 2011-04-04T14:50:15 If you have a fix, I can champion it through. cc me on the PR 2011-04-04T14:56:08 *** antgreen has joined #rtems 2011-04-04T14:58:36 gedare: http://www.rtems.org/ftp/pub/rtems/people/joel/coverage/erc32/erc32.html#Ospd 2011-04-04T14:59:31 Added 6 or 7 cases apparently with rbtree 2011-04-04T15:17:01 *** antgreen has quit IRC 2011-04-04T15:35:00 that's not awful. 2011-04-04T15:35:45 will have to be looked at...pretty big jump in the size of the binary too 2011-04-04T15:35:53 well, compared to the history there. 2011-04-04T15:38:34 it is the sum of all objects under analysis. So just a few more test cases. :-D 2011-04-04T15:38:46 * DrJoel waves bye.. stormy here and need to get home 2011-04-04T15:38:50 *** DrJoel has quit IRC 2011-04-04T15:39:02 ok. ciao 2011-04-04T15:39:08 *** gedare has quit IRC 2011-04-04T15:56:28 *** core-ix has quit IRC 2011-04-04T15:59:31 *** kuzew has quit IRC 2011-04-04T15:59:33 *** dimonov has quit IRC 2011-04-04T16:02:13 *** kuzew has joined #rtems 2011-04-04T16:02:13 *** dimonov has joined #rtems 2011-04-04T16:04:32 *** madrazr has joined #rtems 2011-04-04T16:24:55 *** core-ix has joined #rtems 2011-04-04T16:54:12 *** cdcs has quit IRC 2011-04-04T17:15:02 *** DrJoel has joined #rtems 2011-04-04T17:15:02 *** DrJoel has joined #rtems 2011-04-04T17:15:02 *** ChanServ sets mode: +o DrJoel 2011-04-04T17:15:09 *** DrJoel has quit IRC 2011-04-04T17:17:44 *** projectgus has joined #rtems 2011-04-04T18:34:22 *** madrazr has quit IRC 2011-04-04T19:29:25 *** kiwichris has joined #rtems 2011-04-04T19:32:02 *** methril has quit IRC 2011-04-04T19:38:56 *** methril has joined #rtems 2011-04-04T19:59:01 *** lekernel has quit IRC 2011-04-04T20:10:59 *** lekernel has joined #rtems 2011-04-04T20:24:13 *** xiangfu has joined #rtems 2011-04-04T21:34:55 *** lj8175 has joined #rtems 2011-04-04T22:15:29 *** zwj has joined #rtems 2011-04-04T23:37:12 *** Robbie0909 has joined #rtems 2011-04-04T23:58:44 *** zwj has quit IRC 2011-04-05T00:22:55 *** xiangfu has quit IRC 2011-04-05T00:23:29 *** xiangfu has joined #rtems 2011-04-05T00:50:51 *** Robbie0909 has quit IRC 2011-04-05T00:55:42 *** Robbie0909 has joined #rtems 2011-04-05T01:14:56 *** sqm has joined #rtems 2011-04-05T01:32:57 *** sebhub has joined #rtems 2011-04-05T01:39:02 *** projectgus has quit IRC 2011-04-05T02:07:53 *** lj8175 has quit IRC 2011-04-05T02:17:07 *** core-ix has quit IRC 2011-04-05T02:17:35 *** sqm has quit IRC 2011-04-05T02:23:43 *** sqm has joined #rtems 2011-04-05T02:58:20 *** core-ix has joined #rtems 2011-04-05T03:10:25 *** arvind_khadri has joined #rtems 2011-04-05T03:10:25 *** arvind_khadri has joined #rtems 2011-04-05T03:27:29 *** dqb has joined #rtems 2011-04-05T03:36:00 *** dqb has quit IRC 2011-04-05T03:36:25 *** dqb has joined #rtems 2011-04-05T03:37:58 *** cdcs has joined #rtems 2011-04-05T04:17:06 *** th_d has joined #rtems 2011-04-05T04:19:44 *** sqm has quit IRC 2011-04-05T04:21:06 *** sqm has joined #rtems 2011-04-05T04:30:41 *** sqm has quit IRC 2011-04-05T04:33:37 *** Robbie0909 has quit IRC 2011-04-05T04:37:24 *** sqm has joined #rtems 2011-04-05T04:43:53 *** sqm has quit IRC 2011-04-05T04:49:54 *** sqm has joined #rtems 2011-04-05T05:13:59 *** th_d has quit IRC 2011-04-05T05:15:12 sqm, hi 2011-04-05T05:22:03 sqm, are you about ? 2011-04-05T05:26:06 *** Robbie0909 has joined #rtems 2011-04-05T05:36:38 o hi 2011-04-05T05:37:20 sqm, hi 2011-04-05T05:37:43 kiwichris: hi chris 2011-04-05T05:38:22 I have another project underway to add runtime linking. 2011-04-05T05:38:26 brb 2011-04-05T05:41:03 *** eddy has joined #rtems 2011-04-05T05:41:53 *** sqm has quit IRC 2011-04-05T05:43:26 *** sebhub has quit IRC 2011-04-05T05:43:31 *** eddy has left #rtems 2011-04-05T05:43:36 *** eddy has joined #rtems 2011-04-05T05:43:46 *** eddy has left #rtems 2011-04-05T05:44:14 *** sqm has joined #rtems 2011-04-05T05:44:23 *** Robbie0909 has quit IRC 2011-04-05T05:45:45 sqm, with the MMU objects will we need a way to reference them like we do a semaphore or task ? 2011-04-05T05:45:48 kiwichris: did you say something? 2011-04-05T05:46:32 I was wanting to chat about the GSoC MMU application. 2011-04-05T05:46:39 *** sebhub has joined #rtems 2011-04-05T05:47:57 I think it will be a accessory used by tasks 2011-04-05T05:49:25 but maybe considering it as an object would be more useful 2011-04-05T05:49:44 like partition I guess 2011-04-05T05:49:48 It would sit in the Classic API better. 2011-04-05T05:50:23 Yes. Then you could have other tasks join the memory region and gain access. 2011-04-05T05:51:31 *** Robbie0909 has joined #rtems 2011-04-05T05:52:37 I wonder is it a good idea to add this to the Partition and Region? 2011-04-05T05:53:26 sqm, that is an interesting idea. 2011-04-05T05:53:35 or add this as another chioce for memory manage? 2011-04-05T05:53:54 For example you create a region and then make a call to set attributes ? 2011-04-05T05:55:19 yes , this should be more useful than just add another APIs Iguess 2011-04-05T05:55:27 I like this idea. If we can avoid a new API and just extend an existing one it makes things much simpler. The existing API has done all the hard work. 2011-04-05T05:57:02 It means you can concentrate on the difficult task of linking a single region to more than one task and having a low overhead context switch 2011-04-05T05:57:44 And I think a new API just with memory protecting function looks not so useful 2011-04-05T05:57:46 sebhub, how does an application see if a file system is busy in Linux ? 2011-04-05T05:58:33 I agree. It makes sense to me to use the existing API and set attributes. If the target does not have MMU support the call fails. 2011-04-05T06:02:12 you get an error status depending on the system call 2011-04-05T06:02:29 an umount call ? 2011-04-05T06:02:46 depends on the flags 2011-04-05T06:03:21 Like the do everything but do not actually do the umount ? 2011-04-05T06:04:11 yes, for example 2011-04-05T06:05:54 Why not just add a flags to the unmount op handler to indicate a dry run and let the file system report an error if busy and ok if not. 2011-04-05T06:07:25 the unmount api is not the problem 2011-04-05T06:07:46 there is no way to figure out if a file system is in use in rtems 2011-04-05T06:07:47 What is ? 2011-04-05T06:08:19 sry, meal is ready 2011-04-05T06:29:25 *** cdcs has quit IRC 2011-04-05T06:37:46 back 2011-04-05T06:44:03 *** cdcs has joined #rtems 2011-04-05T06:51:05 *** core-ix has quit IRC 2011-04-05T06:55:01 If a file system accepts an unmount when it knows it has files open then that is a bug in the file system. 2011-04-05T07:04:17 yes, it is a bug or flaw in the rtems file system layer 2011-04-05T07:04:41 the particular file system cannot solve this problem 2011-04-05T07:05:06 For iop it can, for the pathloc no. 2011-04-05T07:05:32 But only the pathloc in the NFS hold data. 2011-04-05T07:05:40 from the heap 2011-04-05T07:06:59 the problem is that the root node has not reference count 2011-04-05T07:07:20 in the IMFS ? 2011-04-05T07:08:03 yes 2011-04-05T07:08:47 ok and why is this a problem ? 2011-04-05T07:11:41 suppose you call eval path (node producer) and land up in e.g. msdos_eval_path() 2011-04-05T07:11:45 now you block 2011-04-05T07:11:53 now you unmount 2011-04-05T07:12:07 you return to msdos_eval_path() and use freed memory 2011-04-05T07:12:42 Yes the threading issue is one which is difficult to solve with out a lot of work. 2011-04-05T07:13:08 In this case the file system should lock itself. 2011-04-05T07:13:26 it was quite hard to figure out this problem, because im my case the 2011-04-05T07:13:27 sc = rtems_semaphore_obtain(fs_info->vol_sema, RTEMS_WAIT, 2011-04-05T07:13:29 MSDOS_VOLUME_SEMAPHORE_TIMEOUT); 2011-04-05T07:13:53 was ok after the resume since another task created a semaphore with the same id 2011-04-05T07:14:06 I remember you having the problem. 2011-04-05T07:14:24 the file system cannot do anything here 2011-04-05T07:14:32 it will see freed memory 2011-04-05T07:15:40 it is at least an imfs issue 2011-04-05T07:15:51 *** lj8175 has joined #rtems 2011-04-05T07:15:53 maybe also a libio issue 2011-04-05T07:16:16 Who's memory is being referenced ? MSDOS ? 2011-04-05T07:16:38 yes, via pathloc 2011-04-05T07:17:18 What is the call that enters msdos_eval_path ? 2011-04-05T07:18:10 rtems_filesystem_evaluate_path() 2011-04-05T07:18:28 From say mknod ? 2011-04-05T07:18:39 or open 2011-04-05T07:20:12 So the issue is the handlers is set in the pathloc and then called with an invalid pathlog 2011-04-05T07:20:35 because it has gone. 2011-04-05T07:20:50 The MSDOS file system. 2011-04-05T07:21:25 no, at the time of the call all is valid 2011-04-05T07:21:44 if you block inside the call and unmount, then you have the problem 2011-04-05T07:22:09 Yes that is what I mean. The reference is the handlers to the now gone file system in pathloc. 2011-04-05T07:22:31 I am wanting to find the memory referenced. 2011-04-05T07:24:15 *** Cfen has joined #rtems 2011-04-05T07:37:04 *** Robbie0909 has joined #rtems 2011-04-05T08:00:17 *** methril has quit IRC 2011-04-05T08:05:34 the freed memory is referenced by: msdos_fs_info_t *fs_info = pathloc->mt_entry->fs_info; (msdos_eval.c line 81) 2011-04-05T08:05:52 Ah yes of course. Thanks. 2011-04-05T08:07:10 I think the solution is to move the locks out of the file systems and into the RTEMS FS layer. 2011-04-05T08:07:48 The pathlocs need to be reference counted and chained to the mt_entry 2011-04-05T08:08:26 yes, something like this, unfortunately i had no time to solve this issue 2011-04-05T08:08:45 I do not either. It is not a small task. 2011-04-05T08:08:47 its on my list for a year now :-( 2011-04-05T08:09:20 I have the rtems-ld (linker) to work on first. 2011-04-05T08:11:01 I am off. bye 2011-04-05T08:11:03 *** kiwichris has quit IRC 2011-04-05T08:12:48 *** methril has joined #rtems 2011-04-05T08:18:09 *** zwj has joined #rtems 2011-04-05T08:26:57 *** sqm has quit IRC 2011-04-05T08:27:30 *** sqm has joined #rtems 2011-04-05T08:35:01 *** arvind_khadri has quit IRC 2011-04-05T08:37:49 *** sqm has quit IRC 2011-04-05T08:42:04 *** sqm has joined #rtems 2011-04-05T08:48:23 *** DrJoel has joined #rtems 2011-04-05T08:48:24 *** ChanServ sets mode: +o DrJoel 2011-04-05T08:50:30 hi joel, when did you run the gcc test suite on arm the last time? 2011-04-05T08:50:51 === gcc Summary === 2011-04-05T08:50:52 # of expected passes 43613 2011-04-05T08:50:54 # of unexpected failures 15916 2011-04-05T08:50:55 # of unexpected successes 8 2011-04-05T08:50:57 # of expected failures 181 2011-04-05T08:50:58 # of unresolved testcases 11127 2011-04-05T08:51:00 # of unsupported tests 1124 2011-04-05T08:51:01 /home/sh/b-arm-gcc/gcc/xgcc version 4.6.0 (GCC) 2011-04-05T08:51:03 is this normal? 2011-04-05T08:51:49 http://www.rtems.org/pipermail/rtems-tooltestresults/2011-March/000479.html 2011-04-05T08:51:56 # of expected passes 70537 2011-04-05T08:51:56 # of unexpected failures 2297 2011-04-05T08:51:56 # of unexpected successes 8 2011-04-05T08:51:56 # of expected failures 186 2011-04-05T08:51:56 # of unresolved testcases 136 2011-04-05T08:51:57 # of unsupported tests 1104 2011-04-05T08:52:26 hm, nearly 20000 more test cases 2011-04-05T08:52:34 But I did a run overnight and the message to the toolresults list bounced due to being too large. So not sounding good.. maybe a recent breakage 2011-04-05T08:52:36 30000 2011-04-05T08:53:03 /users/joel/test-gcc/b-arm-gcc/gcc/xgcc version 4.6.0 20110304 (experimental) [trunk revision 170678] (GCC) 2011-04-05T08:53:13 with my arm eabi switch patch i have: 2011-04-05T08:53:29 === gcc Summary === 2011-04-05T08:53:30 # of expected passes 72429 2011-04-05T08:53:32 # of unexpected failures 200 2011-04-05T08:53:33 # of unexpected successes 7 2011-04-05T08:53:35 # of expected failures 183 2011-04-05T08:53:36 # of unresolved testcases 138 2011-04-05T08:53:38 # of unsupported tests 1103 2011-04-05T08:53:39 /home/sh/b-arm-gcc/gcc/xgcc version 4.6.0 (GCC) 2011-04-05T08:53:48 Wow! That's better than what I got overnight 2011-04-05T08:53:50 # of expected passes 43795 2011-04-05T08:53:50 # of unexpected failures 15993 2011-04-05T08:53:51 # of unexpected successes 8 2011-04-05T08:53:51 # of expected failures 185 2011-04-05T08:53:53 # of unresolved testcases 11143 2011-04-05T08:53:55 # of unsupported tests 1125 2011-04-05T08:54:13 /home2/joel/build/b-arm-gcc/gcc/xgcc version 4.6.1 20110329 (prerelease) (GCC) 2011-04-05T08:54:20 ok, so not a local gcc test running issue 2011-04-05T08:54:43 something got worse in the past month :( 2011-04-05T08:54:55 Do you want me to try your patch and see how it goes? 2011-04-05T08:55:13 yes 2011-04-05T08:55:21 i posted it to the gcc mailing list 2011-04-05T08:55:36 your script is pretty cool 2011-04-05T08:56:13 there are a lot of -flto problems it seems 2011-04-05T08:58:14 hi DrJoel 2011-04-05T08:58:48 joel: no need to try the patch 2011-04-05T08:59:08 *** sqm has quit IRC 2011-04-05T08:59:20 hey dqb 2011-04-05T08:59:34 i want to update the tcp/ip stack for rtems 2011-04-05T08:59:51 as a gsoc project? 2011-04-05T09:00:07 yes,sebhub 2011-04-05T09:00:26 what's your advice? 2011-04-05T09:00:50 is it suitable for gsoc project? 2011-04-05T09:01:02 it is challenge 2011-04-05T09:01:27 it depends on your freebsd 8 knowledge 2011-04-05T09:01:41 provided we choose freebsd 8 as the update candidate 2011-04-05T09:04:16 i have read rtems stack code for about one year,and 2011-04-05T09:07:06 *** sqm has joined #rtems 2011-04-05T09:08:08 DrJoel: I just talked with chris. I think make mmu context used by partition and region would be more useful for rtems. So I am updating my proposal. Any suggestion? 2011-04-05T09:09:26 rtems stack is porting from bsd 4.4,if i want to port freebsd 8 ,how big diffrence between them? 2011-04-05T09:10:05 sqm: that sounds like a very reasonable use. Did you see the code from aanjhan's gsoc on a previous year? It would be good to build on 2011-04-05T09:11:22 yes. I have the code and have viewed it 2011-04-05T09:11:42 i have read two papers of FreeBSD, and know that it has do much optimization for SMP 2011-04-05T09:12:12 did you use freebsd somehow? 2011-04-05T09:12:57 i have much experience on linux kernel, but less on freebsd 2011-04-05T09:13:18 there is also IPv6, altq, firewalls, wireless networks, different media access layers, cryptography stuff, routing improvements etc. 2011-04-05T09:13:53 DrJoel: As I have to do the project on simulator. PSIM is a good choice. I can reuse the exception handle part from the old project I gues 2011-04-05T09:14:34 yes, it's really a challenge 2011-04-05T09:14:58 sqm: good.. I have your proposal in a window and will try to read it this morning 2011-04-05T09:15:32 i did the usb stack port from freebsd 8 to rtems, this approach is also suitable for a network stack update 2011-04-05T09:16:02 i have already done some basic work, but you need a lot of time to finish this 2011-04-05T09:16:14 and it is not clear if anybody can aggree with this approach 2011-04-05T09:16:26 but can I port some basic function, such as IPv4, routing etc? 2011-04-05T09:16:43 and we would like a clean "bsd adapter kit" so the functionality in about a dozen .h files is available in RTEMS. So the code comes over cleanly .. minimal mods to BSD code 2011-04-05T09:17:24 sebhub: I think we agree we want the kit and layer on top of that. Minimal mods to BSD code. Easy to update. The devil is in the details 2011-04-05T09:17:34 the usb port meets this demand 2011-04-05T09:17:48 although some people disagree here 2011-04-05T09:19:28 *** xiangfu has quit IRC 2011-04-05T09:19:46 a start may be to integrate the usb stack, this is easier than a complete network stack update 2011-04-05T09:19:50 I could be wrong but I think there is some concern over old tcpip code and new usb code 2011-04-05T09:19:54 yep.. agree on first task 2011-04-05T09:21:16 there is only a problem if you use usb and old tcp/ip in one file 2011-04-05T09:26:03 DrJoel: I also know there is another choice for porting - DragonflyBSD , what about this one? 2011-04-05T09:28:06 *** gedare has joined #rtems 2011-04-05T09:28:37 sebhub got the USB stack from FreeBSD (I think) and it would probably be simpler to get the TCP/IP stack from there. Dragonfly is an option but differences between the *BSDs would just increase the pain 2011-04-05T09:29:33 yes, the usb stack is from freebsd 8, i don't see what is wrong with the freebsd 8 network stack 2011-04-05T09:29:52 ok,i got it,thanks 2011-04-05T09:34:08 *** Cfen has quit IRC 2011-04-05T09:34:51 networking stack should can come from newer vesion of freebsd 2011-04-05T09:35:03 i looked at this 2-3 mo. ago... it is not mergeable directly 2011-04-05T09:35:35 but updating the network stack = good 2011-04-05T09:35:36 yes, merging is not an option from my point of view 2011-04-05T09:36:06 the current rtems network stack is excellent, but it lacks some features 2011-04-05T09:36:13 it is not 64-bit clean 2011-04-05T09:36:27 yes, 64-bit is a feature ;-) 2011-04-05T09:36:30 that's the top of my list, not many people care :P 2011-04-05T09:36:59 you should have a look at the usb stack port 2011-04-05T09:37:12 btw vxworks is really excited about having a fully 64-bit RTOS 2011-04-05T09:37:26 basically this approach works for the network stack also 2011-04-05T09:37:41 I was going to do that sometime... it seems interesting 2011-04-05T09:38:05 oh i was wondering, do you know anything about realview pbx arm? 2011-04-05T09:38:40 a 64-bit single process system seems to be a huge chance for memory corruption 2011-04-05T09:39:20 heh 2011-04-05T09:40:19 realview pbx arm? no idea 2011-04-05T09:40:48 ok.. i guess it is some line of development boards, pbx is one of them.. RealView 2011-04-05T09:41:20 Just curious. 2011-04-05T09:41:41 I ran into a simulator that supports ARM but works best with the RealView PBX configuration of Linux 2011-04-05T09:41:52 Haven't had time to poke at it with RTEMS 2011-04-05T09:43:18 *** xiangfu has joined #rtems 2011-04-05T09:43:37 i am about to update the arm tool chain to support the cortex, currently i wait for feedback on the gcc list 2011-04-05T09:45:07 gedare: thanks for your advice 2011-04-05T09:46:10 sqm: you're welcome. good luck! 2011-04-05T09:46:19 null trap should be a target for my project. It is very useful 2011-04-05T09:46:45 I think it is important to focus on providing useful abstractions of common MMU operations... 2011-04-05T09:47:13 And not worry about how tasks will use them, since applications are built in the "one process" framework right now you don't want to do anything to break that 2011-04-05T09:47:49 *** xiangfu has quit IRC 2011-04-05T09:47:52 So that without an application doing anything, the kernel will just map the memory range of the application and never touch the MMU again 2011-04-05T09:48:38 you mean providing the common operations as classic API ? 2011-04-05T09:49:06 I'm not sure where the best place to put them would be... 2011-04-05T09:49:20 Something you'll have to kind of think on and figure 2011-04-05T09:49:38 But classic API is the right place for access by user tasks 2011-04-05T09:50:16 the functionality itself should probably be somewhere like libcpu... and the generic kernel interface would be in score if the kernel needed to use MMU services (unlikely) 2011-04-05T09:50:43 it depends on how tightly the MMU implementations are to the system... 2011-04-05T09:51:05 if they are board-specific, then libbsp will be where to implement... umm 2011-04-05T09:51:32 look at how cpu and bsp specific functionality is currently implemented, clock driver is a good one to look at 2011-04-05T09:51:45 each bsp can define a clock driver, and rtems knows how to access it 2011-04-05T09:52:16 I think this might be not proper to provide this APIs directly to users. They can be used by other memory manager like partition and region 2011-04-05T09:52:50 ok 2011-04-05T09:54:17 then implement the full MMU interface at score level 2011-04-05T09:54:20 sqm: I disagree.. i think the user should explicitly have to set MMU context .. 2011-04-05T09:54:33 I can see both approaches being useful 2011-04-05T09:54:53 better to provide it as an indepedent feature and then let users find uses for it 2011-04-05T09:54:58 directly by address? 2011-04-05T09:55:13 directly by interface 2011-04-05T09:55:23 rtems_mmu_set_context(...) or whatever 2011-04-05T09:56:42 and maybe provide configure options to enable mmu protection of partitions/regions.. 2011-04-05T09:57:13 that is ok. 2011-04-05T09:57:14 though I don't see the use case clearly from here ;) 2011-04-05T09:57:30 brb .. expecting phone call 2011-04-05T09:58:04 *** xiangfu has joined #rtems 2011-04-05T10:00:08 but shat should users set this by? In the old project, mmu context is used directly through the address like 0x00008111. I think this kind interface is not necessary for users 2011-04-05T10:00:44 users should have an interface, probably in the classic API 2011-04-05T10:03:52 Users operate the mmu context by operating memory blocks in partition or region. so the interface could be used as a part of these managers 2011-04-05T10:04:18 users know best what kind of separation/use of memory their application uses 2011-04-05T10:04:24 and how they want to protect some regions from others 2011-04-05T10:05:10 *** Robbie0909 has quit IRC 2011-04-05T10:05:50 just think about how you can enable users to manually specify mappings..the same capabilities would be needed in any automated solution too 2011-04-05T10:07:41 sounds reasonable 2011-04-05T10:17:48 gedare: thanks for your advice again. gives me a lot of enlightenment 2011-04-05T10:18:29 okie, good luck 2011-04-05T10:19:22 *** lj8175 has quit IRC 2011-04-05T10:25:33 *** dqb has quit IRC 2011-04-05T10:27:40 *** sebhub has quit IRC 2011-04-05T10:57:19 *** kristianpaul has quit IRC 2011-04-05T10:59:26 *** sqm has quit IRC 2011-04-05T11:02:46 hey DrJoel 2011-04-05T11:22:07 Fallenou: hey.. on phone.. brb 2011-04-05T11:22:33 oh, sorry 2011-04-05T11:27:04 *** kristianpaul has joined #rtems 2011-04-05T11:29:27 *** xiangfu has quit IRC 2011-04-05T11:38:18 *** rokka has quit IRC 2011-04-05T11:38:22 *** rokka has joined #rtems 2011-04-05T11:49:19 *** zwj has quit IRC 2011-04-05T11:49:21 *** zelazny has joined #rtems 2011-04-05T11:49:28 *** zelazny has joined #rtems 2011-04-05T11:49:51 *** zelazny is now known as Guest8769 2011-04-05T11:50:18 *** verm__ has quit IRC 2011-04-05T11:55:22 *** Guest8769 is now known as verm__ 2011-04-05T11:55:30 *** verm__ has joined #rtems 2011-04-05T12:03:55 * DrJoel is back 2011-04-05T12:20:50 hey back DrJoel 2011-04-05T12:21:13 DrJoel: do you think it's a good idea to use ether_input() in an interrupt handler ? 2011-04-05T12:22:45 no.. the isr is support to wake up a task to pull data off the NIC and pass it to ether_input 2011-04-05T12:23:51 *** methril_work has joined #rtems 2011-04-05T12:24:13 what if getting the data off the NIC is really fast and we have everything 2011-04-05T12:24:25 would it be possible to call ether_input() from the isr ? 2011-04-05T12:24:46 to avoid using a rxDaemon 2011-04-05T12:25:01 and decrease latency 2011-04-05T12:25:05 the tcp/ip stack is only safe from tasks. 2011-04-05T12:25:12 ok 2011-04-05T12:25:42 May make sense to pull it off in ISR and hand it to task to enqueue. But doubt it 2011-04-05T12:26:00 that's what we do now ^^ 2011-04-05T12:26:40 may increase overall interrupt latency but not a big deal more than likely 2011-04-05T12:28:12 ok 2011-04-05T12:29:30 *** arvind_khadri has joined #rtems 2011-04-05T12:29:31 *** arvind_khadri has joined #rtems 2011-04-05T13:01:01 *** cdcs has quit IRC 2011-04-05T13:36:16 what's the proper way to get the length of a mbuf chain? m_length() is missing... 2011-04-05T13:37:32 btw it's for copying the complete contents of a the mbuf chain using m_copydata() later on 2011-04-05T13:39:46 *** madrazr has joined #rtems 2011-04-05T13:42:35 *** arvind_khadri has quit IRC 2011-04-05T13:45:49 *** lekernel has quit IRC 2011-04-05T13:50:54 *** hexene has joined #rtems 2011-04-05T13:53:09 DrJoel: I've posted the question on the ml 2011-04-05T13:57:56 *** lekernel has joined #rtems 2011-04-05T13:58:21 hexene: good place.. hopefully you will get an answer 2011-04-05T14:01:47 DrJoel: I suppose inlining the functions could work, but this cannot be forced. So what would you say to making the cache routines macros? 2011-04-05T14:02:07 that's ok for the cpu implementations. 2011-04-05T14:02:25 hexene: have you filled out an application yet at google-melange? Please go ahead so something is in place 2011-04-05T14:02:36 ok 2011-04-05T14:05:20 DrJoel: so can't the generic cache routines be made macros too? 2011-04-05T14:07:19 I don't think so.. normally rtems_ routines are available as symbols in the library. Would need to be discussed 2011-04-05T14:24:05 *** lekernel has quit IRC 2011-04-05T15:30:46 lets see what is in the ncurses rtems patch.. 2011-04-05T15:32:02 oh.. its not patch :l 2011-04-05T15:53:00 DrJoel: is ncurses-5.3 vanilla in rtems-addon-packages 2011-04-05T15:53:04 ? 2011-04-05T15:54:31 *** hexene has quit IRC 2011-04-05T15:54:58 I would have to do a diff... 2011-04-05T15:55:48 only change I see is added bfin to config.sub 2011-04-05T15:56:11 are you building it like the makefile/script in rtems-addon-packages? 2011-04-05T15:56:12 same here 2011-04-05T15:56:37 i didn't try to build it 2011-04-05T15:56:52 i just wanted to see if there is changes to make it compatible with rtems 2011-04-05T15:57:19 since i had to modify ncurses-5.8 to make it build with i386-rtems4.11 toolchain 2011-04-05T15:57:36 Wow! There is a long list of configure options it uses.. 2011-04-05T15:57:41 yes. 2011-04-05T15:57:44 cf_cv_type_of_bool=char CXX="$(CC)" TERM=xterm ./configure \ 2011-04-05T15:57:44 --host=$(RTEMS_CPU)-rtems \ 2011-04-05T15:57:45 --prefix=$(exec_prefix) \ 2011-04-05T15:57:45 --libdir=$(exec_prefix)/$(RTEMS_BSP)/lib \ 2011-04-05T15:57:45 --includedir=$(exec_prefix)/$(RTEMS_BSP)/lib/include \ 2011-04-05T15:57:45 --without-ada \ 2011-04-05T15:57:47 --without-progs \ 2011-04-05T15:57:49 --without-cxx-binding \ 2011-04-05T15:57:53 --disable-sigwinch \ 2011-04-05T15:57:55 --disable-database \ 2011-04-05T15:57:57 --with-fallbacks="$(TERMINALS)" 2011-04-05T15:57:59 make "TERM=xterm" "HOSTCC=gcc" "HOSTCCFLAGS=-I. -I../include" HOSTLDFLAGS="" 2011-04-05T15:58:11 Are you using that? 2011-04-05T15:58:24 *** hexene has joined #rtems 2011-04-05T15:58:33 i iterated somehing similar to 5.8 2011-04-05T15:58:39 i could try that 2011-04-05T15:59:13 hmm.. actually its almost identical 2011-04-05T16:00:47 lets see what changes i made.. 2011-04-05T16:05:46 *** methril_work has quit IRC 2011-04-05T16:06:26 *** peerst has quit IRC 2011-04-05T16:07:04 *** methril_work has joined #rtems 2011-04-05T16:07:49 i added following lines to ncurses-5.8/ncurses/include/term.h 2011-04-05T16:07:55 #if __rtems__ 2011-04-05T16:07:55 # include 2011-04-05T16:07:55 # define TTY struct termios 2011-04-05T16:07:55 #define TERMIOS 2011-04-05T16:11:08 5.8 is a lot newer than 5.3.. would expect some changes. :-D 2011-04-05T16:11:44 i have edited a generated file (that term.h) :l 2011-04-05T16:12:10 thats why it doesn't show in diff 2011-04-05T16:12:38 how should i make diff.. "diff -r dir1 dir2 > dir.diff" ? 2011-04-05T16:12:46 yep.. 2011-04-05T16:12:52 did you do a make install on rtems first?\ 2011-04-05T16:13:04 hmm? 2011-04-05T16:13:57 yep.. rtems was already installed .. then i build ncurses 2011-04-05T16:14:06 and install it 2011-04-05T16:14:19 RTEMS tools or RTEMS with BSP? 2011-04-05T16:14:43 both ? 2011-04-05T16:15:04 i added ncurses last.. 2011-04-05T16:15:21 then build app 2011-04-05T16:15:52 ok.. but theres no major help from ncurses-5.3 to the input problem 2011-04-05T16:16:28 5.3 could have same problem too 2011-04-05T16:17:39 you have to do a make and then make install on your RTEMS Build.. it likely needs some .h files to satisfy the autoconf. That's how the one in rtems-addon-packages is built 2011-04-05T16:18:31 are we speaking of the same thing:)? 2011-04-05T16:20:03 I don't think you are grokking me. :-D 2011-04-05T16:20:07 i added manually rtems include dir to ncurses makefiles.. :D to make it compile.. install didn't work but i manually copied files to /opt/i386-rtems4.11/.. 2011-04-05T16:20:13 when you build rtems itself, you configure, then make.. then make install 2011-04-05T16:20:28 you have to be root to copy to /opt/... 2011-04-05T16:20:40 yes.. i know.. :) 2011-04-05T16:20:43 but you can use a different prefix... I do that to test.. install in my home/bsp-install 2011-04-05T16:20:57 I dropped ncurses-5.9 in my rtems-addon-packages.. then did this 2011-04-05T16:21:11 RTEMS_MAKEFILE_PATH=/home/joel/rtems-4.11-work/bsp-install/arm-rtems4.11/edb7312/ ./bit 2011-04-05T16:21:23 ../ncurses/./tty/lib_twait.c:76:13: warning: its scope is only this definition or declaration, which is probably not what you want [enabled by default] 2011-04-05T16:21:23 ../ncurses/./tty/lib_twait.c: In function '_nc_gettime': 2011-04-05T16:21:23 ../ncurses/./tty/lib_twait.c:81:14: error: storage size of 't1' isn't known 2011-04-05T16:21:23 ../ncurses/./tty/lib_twait.c:84:2: error: dereferencing pointer to incomplete type 2011-04-05T16:21:23 ../ncurses/./tty/lib_twait.c:88:8: error: dereferencing pointer to incomplete type 2011-04-05T16:21:25 ../ncurses/./tty/lib_twait.c:92:23: error: dereferencing pointer to incomplete type 2011-04-05T16:21:27 ../ncurses/./tty/lib_twait.c:93:24: error: dereferencing pointer to incomplete type 2011-04-05T16:21:29 ../ncurses/./tty/lib_twait.c: In function '_nc_timed_wait': 2011-04-05T16:21:31 ../ncurses/./tty/lib_twait.c:168:14: error: storage size of 't0' isn't known 2011-04-05T16:21:33 make[2]: *** [../objects/lib_twait.o] Error 1 2011-04-05T16:22:31 TimeType isn't defined .. must be some ncurses private time... 2011-04-05T16:22:41 ok.. why have i tried 5.8 when there is 5.9 :) 2011-04-05T16:23:46 oh.. its new.. :D 2011-04-05T16:23:55 2011-04-04 2011-04-05T16:23:56 don't know why it is failing for that.. looks like we should have tripped it ok in autoconfs 2011-04-05T16:24:01 ncurses-5.9/config.status:#define HAVE_GETTIMEOFDAY 1 2011-04-05T16:25:19 #ifdef __rtems__ 2011-04-05T16:25:19 # include 2011-04-05T16:25:19 #endif 2011-04-05T16:25:34 added that after the MINGW32 ifdef in ncurses-5.9/ncurses/tty/lib_twait.c and it builds fine 2011-04-05T16:26:01 :-o 2011-04-05T16:26:14 I have done this before.. LOL ... 2011-04-05T16:26:30 may not work .. but builds great! 2011-04-05T16:27:02 hmm.. how do i get rid of 5.8 in /opt/i386-rtems4.11 2011-04-05T16:27:10 manually delete :D 2011-04-05T16:27:29 maybe it doesn't matter.. if 5.9 replaces all the files 2011-04-05T16:28:50 so xf ncurses-5.9.tar.gz .. what should i do next. :D 2011-04-05T16:28:57 get that makefile for 5.3? 2011-04-05T16:28:59 Drop the 5.9 tarbal into rtems-addon-packages.. then 2011-04-05T16:29:19 check out rtems-addon-packages from cvs.. put ncurses-5.9 in the top directory. 2011-04-05T16:29:28 edit bit to replace 5.3 with 5.9... 2011-04-05T16:29:34 then 2011-04-05T16:29:40 RTEMS_MAKEFILE_PATH=/home/joel/rtems-4.11-work/bsp-install/arm-rtems4.11/edb7312/ ./bit 2011-04-05T16:29:51 with RTEMS_MAKEFILE_PATH set to match yourBSP's install point 2011-04-05T16:35:41 and it fails..:D 2011-04-05T16:36:01 ok not... 2011-04-05T16:36:06 im doind it wrong.. 2011-04-05T16:38:12 how did it fail? 2011-04-05T16:39:41 wrong path.. no its compiling 2011-04-05T16:39:44 now.. 2011-04-05T16:40:22 all done.. 2011-04-05T16:40:50 :-D did you have to add those 3 lines? 2011-04-05T16:45:50 sure 2011-04-05T16:46:12 :-D 2011-04-05T16:46:14 oh noes.. stack blown with 5.9 :l 2011-04-05T16:51:22 exception 0? 2011-04-05T16:51:46 which cpu? 2011-04-05T16:51:51 Exception 0 caught at PC 1C941B by thread 167837697 2011-04-05T16:51:56 qemu i386 2011-04-05T16:52:01 ha, ralf spam 2011-04-05T16:52:13 if it doesn't say stack blown could be FPU disabled 2011-04-05T16:53:20 bye 2011-04-05T16:53:27 *** gedare has quit IRC 2011-04-05T16:53:35 what.. fpu disabled :l 2011-04-05T16:53:56 i have this #define CONFIGURE_INIT_TASK_ATTRIBUTES RTEMS_FLOATING_POINT 2011-04-05T16:54:52 ok.. then no quick guesses 2011-04-05T16:55:00 what country is PH? The spam is from there. 2011-04-05T16:55:09 descr: Bayan Telecommunications Incorporated 2011-04-05T16:55:09 descr: Roosevelt Operations Center 2011-04-05T16:55:09 descr: Bayan Building, Roosevelt Avenue, SFDM, Quezon City 2011-04-05T16:55:09 country: PH 2011-04-05T16:55:34 grrrr Philippines ... 2011-04-05T16:55:53 I blocked them from the elvis costello fans site I run... too much damned spam from one island nation 2011-04-05T17:01:56 DrJoel: can you test your ncurses with simple app 2011-04-05T17:07:09 int i=0; 2011-04-05T17:07:09 setenv("TERM", "linux", 1); 2011-04-05T17:07:09 signal(SIGINT, ncurses_finish); 2011-04-05T17:07:09 initscr(); 2011-04-05T17:07:09 while(1) 2011-04-05T17:07:11 { mvprintw(4,0,"text %d",i); i+=1; refresh(); 2011-04-05T17:07:13 } 2011-04-05T17:07:16 crash :l 2011-04-05T17:07:30 with qemu i386 2011-04-05T17:07:57 skip the signal row.. 2011-04-05T17:09:28 Email me a full test case.. time for me to head home.. will run it tomorrow 2011-04-05T17:09:31 *** DrJoel has quit IRC 2011-04-05T17:09:53 *** hexene has quit IRC 2011-04-05T17:33:20 *** projectgus has joined #rtems 2011-04-05T19:23:44 *** kiwichris has joined #rtems 2011-04-05T19:34:10 *** madrazr has quit IRC 2011-04-05T19:44:16 *** lj8175 has joined #rtems 2011-04-05T20:02:41 *** xiangfu has joined #rtems 2011-04-05T20:28:04 *** mw1 has joined #rtems 2011-04-05T20:31:44 *** methril_ has joined #rtems 2011-04-05T20:33:57 *** mwalle has quit IRC 2011-04-05T20:33:57 *** Fallenou has quit IRC 2011-04-05T20:33:58 *** methril has quit IRC 2011-04-05T20:38:16 *** Fallenou has joined #rtems 2011-04-05T20:53:51 *** dr__house has quit IRC 2011-04-05T20:53:51 *** xiangfu has quit IRC 2011-04-05T20:54:58 *** dr__hous1 has joined #rtems 2011-04-05T20:55:00 *** lj8175 has quit IRC 2011-04-05T20:55:01 *** lj8175 has joined #rtems 2011-04-05T20:55:12 *** xiangfu has joined #rtems 2011-04-05T21:10:38 *** dr__hous1 has quit IRC 2011-04-05T21:10:42 *** dr__house has joined #rtems 2011-04-05T21:10:42 *** dr__house has joined #rtems 2011-04-05T21:37:21 *** arvind_khadri has joined #rtems 2011-04-05T21:37:22 *** arvind_khadri has joined #rtems 2011-04-05T22:35:43 *** dimonov has quit IRC 2011-04-05T22:37:15 *** dimonov has joined #rtems 2011-04-05T22:53:36 *** dimonov has quit IRC 2011-04-05T22:54:06 *** dimonov has joined #rtems 2011-04-05T22:54:13 *** dimonov has joined #rtems 2011-04-05T23:52:37 *** arvind_khadri has quit IRC 2011-04-06T01:06:39 *** lj8175 has quit IRC 2011-04-06T01:50:17 *** sebhub has joined #rtems 2011-04-06T01:51:16 good morning 2011-04-06T02:09:31 *** projectgus has quit IRC 2011-04-06T02:20:36 *** lj8175 has joined #rtems 2011-04-06T03:38:16 *** cdcs has joined #rtems 2011-04-06T04:42:55 *** xiangfu has quit IRC 2011-04-06T04:49:04 *** monstr-microblaz has joined #rtems 2011-04-06T05:05:26 *** peerst has joined #rtems 2011-04-06T05:05:38 hi 2011-04-06T05:08:26 hi 2011-04-06T05:12:18 *** arvind_khadri has joined #rtems 2011-04-06T05:12:18 *** arvind_khadri has joined #rtems 2011-04-06T05:47:35 *** lj8175 has quit IRC 2011-04-06T05:53:44 *** kiwichris has quit IRC 2011-04-06T06:03:31 *** sqm has joined #rtems 2011-04-06T06:09:11 *** arvind_khadri has quit IRC 2011-04-06T06:31:01 *** lj8175 has joined #rtems 2011-04-06T06:31:53 *** cdcs has quit IRC 2011-04-06T07:21:12 *** cdcs has joined #rtems 2011-04-06T07:46:25 *** zwj has joined #rtems 2011-04-06T08:09:46 cdcs:hi 2011-04-06T08:09:59 hi 2011-04-06T08:10:06 *** xiangfu has joined #rtems 2011-04-06T08:12:45 i send a mail to Tobias to ask how to merge the AIR to RTEMS tree, but no replay . so he must be very busy 2011-04-06T08:13:17 could you help me remind him 2011-04-06T08:22:03 ok. i told him that you wanted to do that 2011-04-06T08:24:03 thank you. and about my proposal i think i should do some prepare work, have you any advice to me ? 2011-04-06T08:25:38 i think you should get used to the hypervisor concepts, spatial and temporal isolation 2011-04-06T08:26:11 and you should read the AIR resports that are available on the internet 2011-04-06T08:26:16 reports* 2011-04-06T08:26:29 to get used to the internal structure of the software 2011-04-06T08:27:31 yeah, recently i download all the reports and read them carefully. now i have a basic knowlege about the AIR architecture 2011-04-06T08:28:34 and what the most diffcult about turn native RTEMS development to AIR development? 2011-04-06T08:29:39 and i think if Tobias have some good idea to merge AIR to RTEMS , this is most important work to do 2011-04-06T08:31:42 PMK is a set of patches to RTEMS source 2011-04-06T08:32:29 so i think the merge will not be difficult 2011-04-06T08:33:04 *** monstr-microblaz has quit IRC 2011-04-06T08:34:51 ok, and now the PAL is also became an independent component ? 2011-04-06T08:35:32 what do you mean by independent component? 2011-04-06T08:38:23 from the report about the AIR. before AIR-ii project it seems that PAL is included in PMK instead of independent component. and if adopt a new OS to AIR PMK must be modified. 2011-04-06T08:38:46 *** xiangfu has quit IRC 2011-04-06T08:39:09 if PAL is an independent component PMK will not be modified when adopt a new OS to AIR 2011-04-06T08:40:51 There have been efforts to abstract the PMK from the guest, but i think they are not complete 2011-04-06T08:41:37 this means that probably there are some parts of the PMK code that need to be abstracted 2011-04-06T08:42:12 ao, this parts should be abstracted to PAL 2011-04-06T08:42:35 and make it independent 2011-04-06T08:42:54 instead of calling directly RTEMS code, there must be an abstraction where other OS's hook up 2011-04-06T08:43:08 if i port Linux to AIR i will also to modify PMK 2011-04-06T08:43:24 probably yes. 2011-04-06T08:44:46 ok, now linux has an layer like PAL name paravirt_ops. so i think it can be a reference to abstract the PMK 2011-04-06T08:46:07 my knowledge of the linux kernel tends to NULL 2011-04-06T08:46:09 :) 2011-04-06T09:03:33 *** xiangfu has joined #rtems 2011-04-06T09:03:50 *** DrJoel has joined #rtems 2011-04-06T09:03:50 *** ChanServ sets mode: +o DrJoel 2011-04-06T09:04:06 DrJoel: hello :) 2011-04-06T09:06:54 hey rokka .. your test case is in my queue.. need to put a contract deliverable together and answer some gsoc email first 2011-04-06T09:07:42 sure .. :) 2011-04-06T09:07:46 ok 2011-04-06T09:08:25 btw. does anybody know where i can download latest leon2 sources? gaisler doesn't seem to provide them anymore? 2011-04-06T09:09:09 Ask Jiri? 2011-04-06T09:09:20 hmm which email? 2011-04-06T09:10:01 jiri AT gaisler DOT com .. or just email their general contact info 2011-04-06T09:11:21 *** gedare has joined #rtems 2011-04-06T09:11:46 i would be cool if they had their own support irc-channel :D 2011-04-06T09:12:16 maybe am too small customer.. only ordered the low-cost fpga evkit : 2011-04-06T09:12:16 :) 2011-04-06T09:12:34 i try jiri 2011-04-06T09:12:47 *** antgreen has joined #rtems 2011-04-06T09:16:38 My impression is that they are generally responsive. 2011-04-06T09:17:17 Maybe it depends who asks :-) 2011-04-06T09:17:50 If i email at morning they answer late in the evening. And its about same timezone. :) 2011-04-06T09:18:40 i think that is not bad... 2011-04-06T09:18:47 :D 2011-04-06T09:19:09 xc3s-1500? 2011-04-06T09:19:14 yep 2011-04-06T09:19:23 i'm also playing with it 2011-04-06T09:19:27 :-) 2011-04-06T09:19:50 i haven't got mine yet.. they are ordering xiling usb cable II from US i think.. 2011-04-06T09:20:06 * gedare has yet to run rtems on real h/w 2011-04-06T09:20:53 gedare, run it in your personal computer :) 2011-04-06T09:21:09 although i did recently get someone to boot hello world on leon3/virtex5 2011-04-06T09:21:18 mmm, wont' support 64-bit! :) 2011-04-06T09:21:19 i have RTEMS as last GRUB boot option on my laptop :) 2011-04-06T09:21:44 though i guess compat would probably work 2011-04-06T09:21:51 i think i buy some virtex5 evkit next 2011-04-06T09:22:04 gedare, simple.... add support to 64 bits :) 2011-04-06T09:22:17 it RTEMS on my boots to io-language console.. :D 2011-04-06T09:22:26 laptop.. 2011-04-06T09:22:27 i've done that once, it is someone elses turn 2011-04-06T09:24:32 how many leon3 can you put on same virtex5? what is the limiting factor? 2011-04-06T09:25:02 i think sc3s-1500 can fit only one 2011-04-06T09:26:41 i don't know. Haven't tried to generate a config (yet) 2011-04-06T09:33:48 *** zwj has quit IRC 2011-04-06T09:50:01 jiri answered that they don't distribute leon2 anymore 2011-04-06T09:50:06 hmm..:D 2011-04-06T09:50:18 who has it then.. :) 2011-04-06T09:52:43 *** sebhub has quit IRC 2011-04-06T09:54:06 *** zwj has joined #rtems 2011-04-06T09:54:23 http://www.1-core.com/resources/ 2011-04-06T09:54:25 found it 2011-04-06T10:02:04 *** xiangfu has quit IRC 2011-04-06T10:02:33 *** antgreen has quit IRC 2011-04-06T10:02:47 *** xiangfu has joined #rtems 2011-04-06T10:04:12 *** xiangfu_ has joined #rtems 2011-04-06T10:08:01 *** xiangfu has quit IRC 2011-04-06T10:19:58 *** xiangfu_ has quit IRC 2011-04-06T10:21:36 *** antgreen has joined #rtems 2011-04-06T10:37:10 *** lj8175 has quit IRC 2011-04-06T11:06:51 *** sqm has quit IRC 2011-04-06T11:30:17 *** zwj has quit IRC 2011-04-06T12:08:30 *** arvind_khadri has joined #rtems 2011-04-06T12:24:41 *** dr__house has quit IRC 2011-04-06T12:29:33 late dinner time 2011-04-06T12:59:46 *** cdcs has quit IRC 2011-04-06T15:37:39 *** mw1 is now known as mwalle 2011-04-06T16:03:47 *** kiwichris has joined #rtems 2011-04-06T16:03:59 * kiwichris waves to Joel 2011-04-06T16:14:33 *** gedare has quit IRC 2011-04-06T16:18:14 *** DrJoel has quit IRC 2011-04-06T16:19:11 *** dr__house has joined #rtems 2011-04-06T16:19:13 *** dr__house has joined #rtems 2011-04-06T16:19:26 *** kiwichris has quit IRC 2011-04-06T17:04:29 *** easwar has joined #rtems 2011-04-06T17:04:29 *** easwar has joined #rtems 2011-04-06T17:09:15 *** cdcs has joined #rtems 2011-04-06T17:10:36 *** cdcs has quit IRC 2011-04-06T17:34:04 *** projectgus has joined #rtems 2011-04-06T17:42:58 *** gedare has joined #rtems 2011-04-06T18:20:17 *** methril_ has quit IRC 2011-04-06T18:23:02 *** methril has joined #rtems 2011-04-06T18:38:42 *** antgreen` has joined #rtems 2011-04-06T18:40:35 *** antgreen has quit IRC 2011-04-06T20:39:44 *** gedare has quit IRC 2011-04-06T20:46:22 *** xiangfu has joined #rtems 2011-04-06T20:50:50 *** xiangfu has quit IRC 2011-04-06T21:02:23 *** xiangfu has joined #rtems 2011-04-06T21:02:39 *** rokka has quit IRC 2011-04-06T21:03:14 *** rokka has joined #rtems 2011-04-06T21:07:42 *** peerst has quit IRC 2011-04-06T21:37:18 *** lj8175 has joined #rtems 2011-04-06T21:55:07 *** kuzew has quit IRC 2011-04-06T22:06:10 *** kuzew has joined #rtems 2011-04-06T22:07:16 *** easwar has left #rtems 2011-04-06T22:30:04 *** kuzew has left #rtems 2011-04-06T22:42:29 *** xiangfu has quit IRC 2011-04-06T22:42:54 *** xiangfu has joined #rtems 2011-04-06T23:19:04 *** lj8175 has quit IRC 2011-04-06T23:29:26 *** peerst has joined #rtems 2011-04-06T23:54:16 *** arvind_khadri has quit IRC 2011-04-07T00:06:00 *** arvind_khadri has joined #rtems 2011-04-07T00:06:00 *** arvind_khadri has joined #rtems 2011-04-07T00:28:41 *** arvind_khadri has quit IRC 2011-04-07T00:48:11 *** monstr-microblaz has joined #rtems 2011-04-07T01:02:36 *** arvind_khadri has joined #rtems 2011-04-07T01:02:36 *** arvind_khadri has joined #rtems 2011-04-07T01:17:18 *** projectgus has quit IRC 2011-04-07T02:13:13 *** dimonov has quit IRC 2011-04-07T02:14:48 *** xiangfu has quit IRC 2011-04-07T02:16:00 *** xiangfu has joined #rtems 2011-04-07T02:18:12 *** lj8175 has joined #rtems 2011-04-07T02:26:08 *** dimonov has joined #rtems 2011-04-07T02:29:40 *** Cfen has joined #rtems 2011-04-07T03:07:11 *** dimonov has quit IRC 2011-04-07T03:31:50 *** core-ix has joined #rtems 2011-04-07T04:02:07 *** cdcs has joined #rtems 2011-04-07T04:31:34 *** xiangfu has quit IRC 2011-04-07T05:10:57 *** sqm has joined #rtems 2011-04-07T05:21:49 *** peerst has quit IRC 2011-04-07T06:29:50 *** arvind_khadri has quit IRC 2011-04-07T06:33:59 *** cdcs has quit IRC 2011-04-07T06:45:21 *** cdcs has joined #rtems 2011-04-07T07:28:44 *** DrJoel has joined #rtems 2011-04-07T07:28:44 *** ChanServ sets mode: +o DrJoel 2011-04-07T08:00:17 *** methril has quit IRC 2011-04-07T08:13:00 *** methril has joined #rtems 2011-04-07T08:28:26 *** xiangfu has joined #rtems 2011-04-07T08:35:52 *** monstr-microblaz has quit IRC 2011-04-07T08:39:45 *** cdcs has quit IRC 2011-04-07T08:39:49 *** zwj has joined #rtems 2011-04-07T08:44:24 *** sqm has quit IRC 2011-04-07T08:51:11 *** sqm has joined #rtems 2011-04-07T09:20:21 *** sqm has quit IRC 2011-04-07T09:27:31 *** sqm has joined #rtems 2011-04-07T09:52:40 *** zwj has quit IRC 2011-04-07T09:57:54 *** monstr-microblaz has joined #rtems 2011-04-07T10:04:24 *** Cfen has quit IRC 2011-04-07T10:57:41 *** sqm has quit IRC 2011-04-07T11:05:15 *** lj8175 has quit IRC 2011-04-07T12:34:14 *** dimonov has joined #rtems 2011-04-07T12:34:16 *** dimonov has joined #rtems 2011-04-07T12:57:49 *** core-ix has quit IRC 2011-04-07T13:12:01 *** monstr-microblaz has quit IRC 2011-04-07T13:21:21 very quiet in here today... 2011-04-07T13:21:31 * DrJoel is reviewing GSoC proposals for the rest of the day 2011-04-07T13:26:20 *** easwar has joined #rtems 2011-04-07T13:36:36 *** madrazr has joined #rtems 2011-04-07T14:36:01 *** DrJoel has left #rtems 2011-04-07T14:36:20 *** DrJoel has joined #rtems 2011-04-07T14:36:20 *** ChanServ sets mode: +o DrJoel 2011-04-07T14:38:25 *** madrazr has quit IRC 2011-04-07T14:45:02 *** kiwichris has joined #rtems 2011-04-07T14:45:20 Morning all 2011-04-07T14:50:33 hey kiwichris .. read any good gsoc proposals lately? 2011-04-07T14:51:17 Yeah I have and it looks like I have many more to read from emails I have 2011-04-07T14:51:55 Did you manage to look over the rtems-ld email I sent yesterday that you were cc'ed on ? 2011-04-07T14:51:59 those are notifications of comments or changes.. 2011-04-07T14:52:09 I read it but not in detail.. 2011-04-07T14:52:13 Yes. I need to look over them. 2011-04-07T14:52:35 Did you see if you use runtime loading you do not need any linker script or BSP details. 2011-04-07T14:52:36 with the gsoc deadline tomorrow and a deliverable due on a contract tomorrow that's my focus 2011-04-07T14:53:09 I will get through them today. 2011-04-07T14:53:42 Wow! 2011-04-07T14:54:30 When you load the target know all this. It links the application knowing what needs to happen 2011-04-07T14:55:39 Do I comment in the GSoC app or Google Docs ? 2011-04-07T14:57:29 I would comment in the GSoC app at this point. Notifications are going out and we are close to the wire 2011-04-07T14:58:25 If an ISO9660 image is loaded into RAM (either via tar trick or on qemu reading ISO into malloc'ed RAM), can the ramdisk be easily pointed at it? 2011-04-07T14:58:53 Did you notice Ralf on the photo on the home page of GSoC ? 2011-04-07T14:59:17 QEMU has CDROM as a device support now. 2011-04-07T14:59:54 no.. where is he.. See Bill Kendrick of TuxPaint 2011-04-07T15:00:13 ahhh upper left hand 2011-04-07T15:00:15 Right hand side one row up from the 3 cornered hat 2011-04-07T15:00:24 right hand.. yes.. the other one. :-D 2011-04-07T15:00:32 :D 2011-04-07T15:00:33 Pirate hat lady was FUNNY! 2011-04-07T15:00:41 Yeah she was fun 2011-04-07T15:00:42 Does the qemu device work with RTEMS 2011-04-07T15:00:56 It is at the IDE level. 2011-04-07T15:01:04 So yes that is a good question 2011-04-07T15:01:27 I asked and tentatively the mentor summit is likely the weekend of Oct 22. Flight Software Workshop 2011 is the week before... 2011-04-07T15:02:10 2011 Workshop on Spacecraft Flight Software (FSW-11). October 19 - 21, 2011. Hosted by The Johns Hopkins University Applied Physics Laboratory 2011-04-07T15:02:23 Where ? 2011-04-07T15:02:48 Johns Hopkins University . east coast.. Maryland.. so very possible I will be doing another hard coast-to-coast 2011-04-07T15:03:10 To far for me. 2011-04-07T15:07:07 long triangle for me.. but longer for you. 2011-04-07T15:07:28 It would be too costly for me to cover. 2011-04-07T15:07:49 I have done the Sydney to New York path many times. 2011-04-07T15:08:07 I was thinking that.. would require a "sponsor".. :) 2011-04-07T15:10:31 *** lekernel has joined #rtems 2011-04-07T15:10:51 hi 2011-04-07T15:11:10 is it OK to call rtems_message_queue_receive() from an ISR (with RTEMS_NO_WAIT of course)? 2011-04-07T15:13:47 yes.. supposed to be :-D 2011-04-07T15:16:24 DrJoel, I think the cache proposal should list the architectures to be covered with a timeline. 2011-04-07T15:19:24 I suggested that via IRC. Please note that 2011-04-07T15:19:52 DrJoel, I think the idea of an image in a ram disk is a good one. On qemu we could dd an image into a ram disk. 2011-04-07T15:20:18 I suggested he do a spreadsheet checking off what was there, could be implemented and wasn't avaiable .. then use that as a project plan.. if he has done that, then he needs to document it. Can you add that as a comment? 2011-04-07T15:20:32 :-D allows multi-architecture testing and coverage testing :-D 2011-04-07T15:20:39 Will do. 2011-04-07T15:20:42 Yes 2011-04-07T15:21:22 Sort of you for test coverage because you still need to get at the image to dd and that means host file system access which the SIS simulator does not have 2011-04-07T15:21:42 qemu is fine 2011-04-07T15:22:24 qemu is great for development ...but you can link an ISO like tarfs and just point RAM disk at that 2011-04-07T15:23:30 You would need to link the image to the executable when you can just read off the host disk inside qemu with RTEMS and dd it straight into the RAM disk of the correct size. 2011-04-07T15:24:17 I ported dd for this purpose. I have used it with MSDOS where I use mtools to make an image when load it and mount it. 2011-04-07T15:25:22 right .. on qemu that works... but if we want to test this on other architectures.... 2011-04-07T15:25:29 for his development, it is awesome 2011-04-07T15:27:49 An issue is the size of a CDROM. Can small working images be made ? 2011-04-07T15:35:04 yes... if there isn't much content.. just have to pick a mix of "interesting" things 2011-04-07T15:36:04 yes 2011-04-07T15:37:26 Cache manager proposal has a link to a nice looking spreadsheet like I suggested :-D 2011-04-07T15:38:10 Yes and I look and it is full of detail. Too much detal. 2011-04-07T15:38:23 We need a list of architectures and times against them 2011-04-07T15:39:01 I need to head off for a while 2011-04-07T15:40:47 *** cdcs has joined #rtems 2011-04-07T15:41:09 cdcs: you did the SATA GSoC project, right? 2011-04-07T15:41:17 yes 2011-04-07T15:41:25 what is preventing it from being merged? 2011-04-07T15:41:41 because is not finished 2011-04-07T15:42:00 LOL.. so one once, it isn't our fault.. what's left? 2011-04-07T15:42:41 i intend to finish it.. when i have the time :| 2011-04-07T15:43:19 this is my last year in the university and i am currently doing my MsC thesis so time is not much 2011-04-07T15:43:41 time is a precious commodity.. 2011-04-07T15:44:22 It's currently missing several controller drivers 2011-04-07T15:44:53 I am just wanting to clean out the "not merged" queue.. does it work for any devices? If so, then we can document the state and leave the others as future work 2011-04-07T15:46:57 no, the freebsd driver can be divided in three layers. Currently ported are the middle one, and the beggining of the bottom one ( the SATA and ATA controller drivers) 2011-04-07T15:47:33 ok.. just making sure we were doing our part 2011-04-07T15:47:37 as far as i remember the top one is not needed 2011-04-07T15:47:49 Also we need the busspace support 2011-04-07T15:48:27 which is an issue for the usb.. also in my queue 2011-04-07T15:48:34 yes, the interfaces that i ported need to be reviewed. I remenber that a few are just stubed 2011-04-07T15:49:14 but if you want a SATA driver ASAP i think the best choice is to look at MINIX 2011-04-07T15:49:27 busspace is needed for multiple projects 2011-04-07T15:49:53 not needed asap.. just trying to pull things together.. probably looking at a release in the fall and wanting to suck in as much as possible 2011-04-07T15:50:16 http://www.freebsd.org/cgi/man.cgi?query=bus_space&sektion=9&apropos=0&manpath=FreeBSD+8.2-RELEASE 2011-04-07T15:50:31 *** core-ix has joined #rtems 2011-04-07T15:51:10 i don't know if busspace will differ. The driver sata driver is from netbsd and the USB stack is from FreeBSD 2011-04-07T15:51:49 hmmm 2011-04-07T15:52:12 ok kiwichris I now have the man page up. will read it soon 2011-04-07T15:52:17 back to gsoc proposals 2011-04-07T15:52:24 bus space is from netbsd 2011-04-07T15:52:27 i remember that when i looked at the libbsdport and at the usb stack there were some interfaces different 2011-04-07T15:52:54 We will just have to handle this. 2011-04-07T15:53:27 DrJoel, I reviewed zwj's proposal and added some comments 2011-04-07T15:53:43 thanks.. that's the java one right? 2011-04-07T15:53:54 nope, the hypervisor one 2011-04-07T15:56:05 aaa... I like that one but still am not sure if at the end RTEMS will have a hypervisor mode.. can you tell? 2011-04-07T15:56:59 i think the merge is possible, but i don't know how much work will be needed for that 2011-04-07T15:57:49 me either.. but it is a good goal 2011-04-07T15:58:05 first problem is that the PMK is made based on RTEMS 4.8.1 2011-04-07T15:58:23 so the first goal is port it to rtems 4.10 2011-04-07T15:58:42 that's why I want to get it merged.. it is rotting already. 2011-04-07T15:59:17 from the point of view of the PMK creators rtems 4.8.1 is a requirement 2011-04-07T16:00:32 then you need to merge 2011-04-07T16:00:50 do something like "--enable-hypervisor" ? 2011-04-07T16:01:03 that's what I was thinking.. --enable-hypervisor or something 2011-04-07T16:02:10 If you are not fond of the ideia of porting linux to RTEMS hypervisor the next goals can be negociated with zwj. 2011-04-07T16:02:21 the problem is that anything based on an old fork of anythign just rots. It is off the mainstream. Even if they have some flight requirement, from a project perspective, 4.8.x is unsupported. OAR offers legacy support for 4.8 and older but the main stream project won't touch those older versions 2011-04-07T16:02:49 if the pmk code is available, I would prefer that it be updated to current RTEMS and merged into tree in a compatible way. 2011-04-07T16:03:09 I guess I should make that clear. More than likely it won't be THAT hard.. I backport patches all the time 2011-04-07T16:04:15 yes i agree with you. The two first steps are port to 4.10 and merge 2011-04-07T16:04:38 then you can think of porting it to pc386 2011-04-07T16:04:43 Is the source available anywhere? Any information? 2011-04-07T16:04:51 or adding linux as a guest 2011-04-07T16:04:53 Right.. it will never get ported anywhere is not on the main stream 2011-04-07T16:06:12 No, is still not available, but will be very soon. I cannot precise when. 2011-04-07T16:06:50 will he have code access.. that is a risk? How does one get access? 2011-04-07T16:07:13 Tobias will send him the code. 2011-04-07T16:07:51 thay are in direct contact 2011-04-07T16:07:58 i think 2011-04-07T16:08:28 I think they are also. 2011-04-07T16:09:22 you can check the PMK structure in the links provided in zwj's proporsal 2011-04-07T16:19:10 open source or closed? 2011-04-07T16:19:30 or just not available because not ready to publish? 2011-04-07T16:20:17 *** DrJoel has quit IRC 2011-04-07T16:20:53 open source. Not ready. 2011-04-07T16:36:06 *** DrJoel has joined #rtems 2011-04-07T16:36:07 *** ChanServ sets mode: +o DrJoel 2011-04-07T16:38:02 cdcs: ok 2011-04-07T16:40:03 hmmm..what just happened to me? 2011-04-07T16:41:07 DrJoel [~joel@rtems/maintainer/joel] has quit [Ping timeout: 240 seconds] 2011-04-07T16:42:46 oh well... time for me to go home anyway.. I am sure the cold beer there is better than the lack of any beer here :-D 2011-04-07T16:42:49 night all 2011-04-07T16:42:51 *** DrJoel has quit IRC 2011-04-07T16:44:03 *** tuxmaniac has joined #rtems 2011-04-07T16:44:41 :-) 2011-04-07T16:47:26 *** antgreen` has quit IRC 2011-04-07T16:50:07 *** tuxmaniac has quit IRC 2011-04-07T16:50:43 *** cdcs has quit IRC 2011-04-07T17:26:40 *** projectgus has joined #rtems 2011-04-07T17:30:21 *** lekernel has quit IRC 2011-04-07T18:05:29 *** methril has quit IRC 2011-04-07T18:06:08 *** methril has joined #rtems 2011-04-07T18:35:23 *** sqm has joined #rtems 2011-04-07T19:01:43 *** sqm has quit IRC 2011-04-07T19:22:57 Aanjhan are you online ? 2011-04-07T19:41:48 *** xiangfu has joined #rtems 2011-04-07T22:14:37 hi projectgus 2011-04-07T22:48:14 *** lj8175 has joined #rtems 2011-04-07T22:58:35 *** shayden has joined #rtems 2011-04-07T23:01:48 hi easwar 2011-04-07T23:03:17 was kinda wondering 2011-04-07T23:03:54 whether I'm still a candidate student 2011-04-07T23:04:21 easwar, which is your application ? 2011-04-07T23:04:43 kiwichris, haven't applied yet, but I'm proposing gprof 2011-04-07T23:04:51 and gcov for RTEMS 2011-04-07T23:05:37 There is a project around which has been in the GSoC in the past called the capture engine. We also trace at the instruction level. How does gcov fit here ? 2011-04-07T23:06:22 kiwichris, I mean to use the output of the capture engine with gprof 2011-04-07T23:07:13 Oh. The capture is an event system and not a sample one. Can gprof handle this ? 2011-04-07T23:07:53 hmm, not too sure 2011-04-07T23:12:05 kiwichris, I'm pretty sure my abilities go as far as reading and writing code and documentation 2011-04-07T23:12:48 I am sorry I do not understand. What do you mean ? 2011-04-07T23:13:28 kiwichris, atm, I've only run the RTEMS Hello World, and I'm supposed to send a diff to Joel 2011-04-07T23:14:02 Yes this is correct. 2011-04-07T23:14:04 but I'm unsure whether he means me to do something as simple as edit a source file and diff it against the original and send him that 2011-04-07T23:16:23 is the diff and screen cap have the same deadline as the GSOC application, tomorrow night? 2011-04-07T23:16:30 *** lj8175 has quit IRC 2011-04-07T23:16:51 do the, not is the, i mean 2011-04-07T23:16:53 *** lj8175 has joined #rtems 2011-04-07T23:22:13 this is kinda what happened last time 2011-04-07T23:22:32 I asked this question, and no one replied 2011-04-07T23:22:40 on the phone 2011-04-07T23:22:44 ok 2011-04-07T23:22:57 I was wondering whether I was asking too basic a question 2011-04-07T23:26:12 *** lj8175 has quit IRC 2011-04-07T23:30:43 *** lj8175 has joined #rtems 2011-04-07T23:58:50 *** lj8175 has quit IRC 2011-04-08T00:06:43 *** lj8175 has joined #rtems 2011-04-08T00:13:45 ?? 2011-04-08T00:15:21 *** easwar has left #rtems 2011-04-08T00:23:29 *** easwar has joined #rtems 2011-04-08T00:25:14 Am I supposed to just edit a file, show that it still builds, and send the diff? 2011-04-08T00:25:50 s/diff/diff between the original source and my edited version 2011-04-08T00:43:56 *** lj8175 has quit IRC 2011-04-08T00:44:27 *** monstr-microblaz has joined #rtems 2011-04-08T00:45:23 *** lj8175 has joined #rtems 2011-04-08T02:12:00 *** projectgus has quit IRC 2011-04-08T02:27:47 *** arvind_khadri has joined #rtems 2011-04-08T03:40:41 *** cdcs has joined #rtems 2011-04-08T03:54:36 *** core-ix has quit IRC 2011-04-08T04:24:19 *** xiangfu has quit IRC 2011-04-08T04:30:31 *** lj8175 has quit IRC 2011-04-08T04:38:07 *** core-ix has joined #rtems 2011-04-08T04:50:57 *** arvind_khadri has quit IRC 2011-04-08T05:09:38 *** arvind_khadri has joined #rtems 2011-04-08T05:17:08 *** sqm has joined #rtems 2011-04-08T05:40:41 *** sebhub has joined #rtems 2011-04-08T05:40:47 good morning 2011-04-08T05:42:42 hi 2011-04-08T05:44:18 sebhub: have you noticed before that you can put LPC arm cpu in such state that it freezes totally and can't be waked even with jtag reset, but power cycling resets it to working state 2011-04-08T05:44:49 yes, this is a nice feature 2011-04-08T05:46:41 is it documented feature ? i had bug that does that 2011-04-08T05:48:55 yes, i experience that myself 2011-04-08T05:49:18 this a documented feature? is this a joke? 2011-04-08T05:49:23 this is a nxp chip 2011-04-08T05:49:50 :)? 2011-04-08T05:51:40 do you mean.. nxp is not good? 2011-04-08T05:52:30 the chip is ok, but the nxp documentation tend to be as accurate as a fortune teller 2011-04-08T05:52:42 ;D 2011-04-08T05:53:17 if you compare this with the freescale documentation there are worlds between 2011-04-08T05:53:29 ok :) 2011-04-08T06:11:14 *** peerst has joined #rtems 2011-04-08T06:11:24 hi 2011-04-08T06:40:51 *** cdcs has quit IRC 2011-04-08T06:57:22 *** cdcs has joined #rtems 2011-04-08T07:10:17 kiwichris:hi 2011-04-08T07:34:10 *** DrJoel has joined #rtems 2011-04-08T07:34:15 *** DrJoel has joined #rtems 2011-04-08T07:34:15 *** ChanServ sets mode: +o DrJoel 2011-04-08T07:34:35 * DrJoel notes that there are about 6.5 hours left to apply to GSoC this year 2011-04-08T07:35:02 anyone need help getting a proposal in? 2011-04-08T07:38:51 DrJoel, I do 2011-04-08T07:39:10 easwar: ok shoot.. wondered where your proposal was ;) 2011-04-08T07:39:46 DrJoel, I was all at sea what you meant by diff 2011-04-08T07:40:35 I looked at Vincent's mail, and what I could make out was I edit a source file, show that the sample still builds and send the diff between the original and my edit 2011-04-08T07:40:46 Am I right so far? 2011-04-08T07:45:41 * easwar wonders why traffic on the channel either dries up or ignores the query 2011-04-08T07:48:12 sorry.. someone walked into my office 2011-04-08T07:48:22 ok... 2011-04-08T07:48:23 now onto diff 2011-04-08T07:48:46 diff is a program under UNIX/Linux which can tell you what is different between two files 2011-04-08T07:49:02 DrJoel, yeah, I know that diff exists 2011-04-08T07:49:06 and how to use it 2011-04-08T07:49:13 When you check out code from a source code control system (e.g. revision control), it can help you see what has changed 2011-04-08T07:49:24 ok 2011-04-08T07:49:38 Since RTEMS uses cvs, and you shuold have source checked out from CVS... you will edit the hello world sample in the tree to print your name. 2011-04-08T07:50:00 Once that runs, you can do a "cvs diff -u FILE" and see what you edited. Send me the output of that command. 2011-04-08T07:50:12 alright 2011-04-08T07:50:25 I used the *nix diff as opposed to the cvs diff 2011-04-08T07:50:44 but I was thinking the correct thing 2011-04-08T07:50:50 The "cvs diff" is known as a patch and with the "patch" program, I can apply your changes toa file locally and commit it 2011-04-08T07:51:11 Don't worry about not understanding.. this is all part of GSoC.. learning this .. community interaction 2011-04-08T07:51:42 DrJoel, true :) 2011-04-08T07:51:47 Is your proposal submitted? I know at least the mentors here at OAR are trying to get comments in quickly this morning 2011-04-08T07:51:58 DrJoel, not yet 2011-04-08T07:52:00 So where was Vincent's mail? 2011-04-08T07:52:06 tick...tick... tick.. 2011-04-08T07:52:13 DrJoel, on the rtems-users ml 2011-04-08T07:52:24 DrJoel, true :) 2011-04-08T07:53:29 I have had a deadline this week and that isn't helping 2011-04-08T07:58:27 oh.. I see the email.. I actually reviewed his proposal at the google site 2011-04-08T07:59:10 I just realized I hadn't confirmed wiki accounts in a week... for a music wiki site I run, we send account requests to an admin mailing list. Haven't done that yet for RTEMS Wiki so I am the bottleneck 2011-04-08T08:00:38 :D 2011-04-08T08:02:18 busy week.. my daughter did the defense of her departmental honors project this week also. 2011-04-08T08:05:32 seems so 2011-04-08T08:05:42 couple of deadlines here too 2011-04-08T08:05:43 *** monstr-microblaz has quit IRC 2011-04-08T08:10:48 Gotta attend a lecture, be back later 2011-04-08T08:33:32 *** kiwichris has quit IRC 2011-04-08T08:36:23 hmm 2011-04-08T08:39:23 shayden: what deep philosophical question has you going hmmm 2011-04-08T08:40:02 well i don't think i'm going to apply to GSoC, since I can make more desperately needed money at my other job 2011-04-08T08:40:18 but RTeMS looks interesting so i think I want to get involved and start contributing 2011-04-08T08:40:50 so, i joined the channel and am looking at the ML :D 2011-04-08T08:41:40 shayden: cool.. we are happy to help you.. what things interest you? 2011-04-08T08:41:57 *** sebhub has quit IRC 2011-04-08T08:43:07 many things, i'm in EE, and want to design various hardware, but i also want to have the software skills to use that hardware 2011-04-08T08:43:19 so maybe I'll look at porting drivers? 2011-04-08T08:44:58 I'm reading about the RosettaOS apis project on the wiki 2011-04-08T08:46:16 *** hexene has joined #rtems 2011-04-08T08:47:07 DrJoel: regarding the timeline, will I have to add cache support to simulators too? 2011-04-08T08:53:59 hexene: you mean actually changing the code in the simulators? 2011-04-08T08:54:14 DrJoel: yes? 2011-04-08T08:55:10 I would say if it is a valid instruction for the target architecture the simulator is faking out, it should at least silently accept the instruction if it is valid. So ideally, yes. 2011-04-08T08:55:51 otherwise, we have to say the BSP is for a specific CPU model that only exists in the simulator. 2011-04-08T08:56:21 did the comments make sense? pick the order of the architectures.. block time for each.. submit patches during each.. that's your schedule 2011-04-08T08:58:11 so given that I might need to tinker with the simulators too, I don't really know how long each arch would take. 2011-04-08T08:59:36 DrJoel: how would you suggest I draw up the timeline? 2011-04-08T08:59:48 And you don't know which ones are missing what either... 2011-04-08T09:00:13 I have yet to test each sim 2011-04-08T09:01:11 Could do it in a couple of days though (hopefully) 2011-04-08T09:01:17 how about you make a list of deficiencies in simulators as you go.. file PRs as appropriate.. and then address fixing them AFTER the cache code is finished for all architectures. This makes it more of what we call "bonus" work.. if all goes well, you do this. And maybe with good PRs filed, others will fix them. Then it is just a matter of testing patches and providing feedback 2011-04-08T09:01:50 As you move from architecture to architecture, you will have to bring up another BSP/toolset. Should get quicker and quicker. :-D Same task over and over 2011-04-08T09:02:12 ok, makes sense :) 2011-04-08T09:03:49 if the simulator is in gdb or another FSF project, you will have to have paperwork filed to submit code. 2011-04-08T09:04:26 oh 2011-04-08T09:05:40 not a big deal to do it. We can start the process Monday.. just ping me.. One student submitting a proposal has already gone through this 2011-04-08T09:06:14 ok 2011-04-08T09:09:38 @DrJoel: I think I should take more discuss about the API structure. Is that OK if the final implementation is not as same as it described in proposal? 2011-04-08T09:09:38 sqm: Error: "DrJoel:" is not a valid command. 2011-04-08T09:10:22 DrJoel: I think I should take more discuss about the API structure. Is that OK if the final implementation is not as same as it described in proposal? 2011-04-08T09:12:51 DrJoel: timeline is coming.I think updating now is not a good idea if not necessary 2011-04-08T09:18:09 sqm: of course it is OK. the final version of something is very seldom exactly like it was conceived. There are always insights you get as you go along and optimizations 2011-04-08T09:19:40 *** zwj has joined #rtems 2011-04-08T09:20:34 thanks. I will build the old project on psim next week 2011-04-08T09:23:31 cool! And hopefully we can work with aanjhan and see what we can do to get that code merged. I am really wanting to clean the queue of unmerged code 2011-04-08T09:34:56 *** lj8175 has joined #rtems 2011-04-08T09:38:15 *** antgreen has joined #rtems 2011-04-08T09:54:09 *** arvind_khadri has quit IRC 2011-04-08T10:00:02 DrJoel: hi joel 2011-04-08T10:01:45 hey.. just replied to an email for you 2011-04-08T10:02:10 yeah, hehe . just recevice your mail 2011-04-08T10:02:32 this week has been busy.. work, GSoC, etc 2011-04-08T10:03:25 yeah, i seems that this year the students to rtems GSOC is less that last year 2011-04-08T10:03:27 DrJoel: which distro and dev environment do most RTEMS devs use? 2011-04-08T10:04:10 Fedora 14 ... that's what all the build farm machines are running .. 2011-04-08T10:04:28 ok, thanks 2011-04-08T10:04:29 Ralf Corsepius is on Steering Committee and is also a Fedora Packaging person 2011-04-08T10:04:34 very easy on Fedora 2011-04-08T10:04:35 and i also send a mail to Tobias to ask for his idea about how to merge AIR code to RTEMS. but it seems that he is also busy 2011-04-08T10:04:39 hehe 2011-04-08T10:05:56 cdcs, Hi. Tobias is so busy that i have no replay until now . could you help me ping him . thank you 2011-04-08T10:06:46 zwj: I think as long as they are willing.. we can find a way.. needs to be licensed properly.. that is a discussion point on both sides.. without seeing how code integrates into rtems, hard to say if LGPL is OK. pure hypervisor code is probably OK as LGPL .. since there is separation of address spaces and no static linking with application code.. but mods to RTEMS need to be GPL+exception 2011-04-08T10:07:05 *** arvind_khadri has joined #rtems 2011-04-08T10:10:58 yeah, we should first contract with Tobias to discuss about these question. and Tobias has said hyervisor is open and ARINC653 APEX is closed. we just merge hypervisor is ok 2011-04-08T10:11:19 *** hexene has quit IRC 2011-04-08T10:12:51 and if there is a way to make RTEMS configurable that both Native RTEMS and Hypervisor RTEMS, it will be perfect. but before look the hypervisor code it is hard to say 2011-04-08T10:21:17 agreed.. a configure option is what I am hoping for 2011-04-08T10:22:27 Do you think this will win me a contest for two tickets to Avenue Q which asked for 3 Q words? 2011-04-08T10:22:29 The quirky queen quivered and quaked as the quintessential quintet quarrelled that quadrophonic sound qualified as quackery. With qualms, the quandary was quickly quenched when they quickly played a quarter of a quarto as a quartet. 2011-04-08T10:24:19 what is this? Tongue twister? hehe 2011-04-08T10:25:03 A random sentence I made up to use as many Q words as possible.. hopefully it will win since it is WAY over the top of just list 3 Q words 2011-04-08T10:27:39 yeah, there are 18 Q words. 2011-04-08T10:29:09 found a scrabble site that listed words that start with q. :-D 2011-04-08T10:32:47 rokka: your ncurses program works on sis... just sent you a png 2011-04-08T10:34:39 what is sis :) 2011-04-08T10:35:24 SPARC Instruction Simulator 2011-04-08T10:35:31 will give it a whirl on pc386 2011-04-08T10:35:38 okay :) 2011-04-08T10:35:55 i tested only with qemu/pc386 2011-04-08T10:37:13 on vga or com1 as console? 2011-04-08T10:37:40 both 2011-04-08T10:37:51 ok.. 2011-04-08T10:44:55 works on pc386 using COM1 as console 2011-04-08T10:45:24 :l 2011-04-08T10:45:47 my tools or something is broken 2011-04-08T10:45:52 looks great on vgacons 2011-04-08T10:46:29 so i have to compile whole toolchain again 2011-04-08T10:46:43 or somethinvg 2011-04-08T10:46:54 I have added ncurses 5.9 to rtems-addon-packages .. why don't you update and use its build magic 2011-04-08T10:47:12 maybe i can try that too 2011-04-08T10:47:58 DrJoel, just added myself to the gprof project page as a volunteer 2011-04-08T10:48:04 DrJoel: thanks for help 2011-04-08T10:58:02 DrJoel, Do I need to implement only mcount and profil? 2011-04-08T10:58:22 or are there other routines too that are required to enable profiling of RTEMS 2011-04-08T10:58:24 ? 2011-04-08T10:58:41 *** arvind_khadri has quit IRC 2011-04-08T10:59:41 mcleanup too 2011-04-08T10:59:44 k 2011-04-08T11:02:51 *** cdcs has quit IRC 2011-04-08T11:02:58 easwar: I don't know.. did you include the gcov support as well? 2011-04-08T11:04:15 DrJoel, nope 2011-04-08T11:04:24 I'm planning that as future work 2011-04-08T11:06:01 ok.. then if you finish profiling and time is left you can start that.. mention it as potential "bonus" work 2011-04-08T11:06:54 *** sqm has quit IRC 2011-04-08T11:12:19 *** lj8175 has quit IRC 2011-04-08T11:42:32 *** core-ix has quit IRC 2011-04-08T11:47:24 i have to go to bed . byb 2011-04-08T11:47:43 DrJoel: what disc requirements does a build machine usually have? I'm building a VM for RTEMS... 2011-04-08T11:48:58 hmmm... let me check the ones I created for students 2011-04-08T11:49:25 thanks! 2011-04-08T11:50:06 I gave those 16GB .. that's enough to build tools if you need to... 1.5 GB of RAM .. my Windows machine has plenty of disk space and 4GB RAM 2011-04-08T11:50:17 the HD is expandable 2011-04-08T11:50:43 does everyone have their proposal in? ... tick.. tick.. 1h10m 2011-04-08T11:51:21 k 2011-04-08T11:52:14 i have a fulltime job lined up for the summer, and joined a cubesat project before i heard about GSoC 2011-04-08T11:52:28 i don't want to apply and set myself up for failure at everything 2011-04-08T11:55:48 DrJoel, done 2011-04-08T11:55:52 :) I have heard talk of RTEMS on CubeSat2 2011-04-08T11:56:57 *** easwar has left #rtems 2011-04-08T12:01:42 nice! 2011-04-08T12:03:06 RTEMS has a lot of space applications.. 2011-04-08T12:07:48 *** tuxmaniac has joined #rtems 2011-04-08T12:28:14 excellent 2011-04-08T12:28:18 i'll keep that in mind 2011-04-08T12:28:40 the computer/software eng group for the project will be notified 2011-04-08T12:29:36 http://flightsoftware.jhuapl.edu/fsw10.html has some very nice presentations. plus video of some (all?) of the presentations. Many mention RTEMS. I gave a presentation and was on a panel 2011-04-08T12:29:48 *** zwj has quit IRC 2011-04-08T12:32:40 *** cdcs has joined #rtems 2011-04-08T12:40:44 1m20m until GSoC Student Application deadline... tick .. tock... tick.. 2011-04-08T12:41:26 How many applicants? 2011-04-08T12:43:01 10-ish 2011-04-08T12:43:56 only four in the wiki :( 2011-04-08T12:48:51 *** hexene has joined #rtems 2011-04-08T12:49:35 DrJoel: I've updated the proposal to include the timeline 2011-04-08T12:51:02 *** arvind_khadri has joined #rtems 2011-04-08T12:52:15 ok.. will review 2011-04-08T13:20:33 DrJoel: Shall I club the cpus together? 2011-04-08T13:24:35 what do you mean? I think you should look at architectures and then variations within that architecture... 2011-04-08T13:31:37 *** madrazr1 has joined #rtems 2011-04-08T13:35:54 *** madrazr1 is now known as madrazr 2011-04-08T13:35:59 *** madrazr has joined #rtems 2011-04-08T13:44:22 15m until GSoC Student Application deadline... tick .. tock... tick.. 2011-04-08T13:44:32 a proposal showed up about 15 minutes ago. :D 2011-04-08T13:49:18 DrJoel: not enough time to group cpus according to arch similarity? What do I do? :( 2011-04-08T13:50:45 Just add a note above the timeline that you will do that.. focus on architecture and variations within each architecture. Even though timeline has specific models 2011-04-08T13:51:00 You comment indicated you would do that anyway.. but just add a note 2011-04-08T13:51:30 ok 2011-04-08T13:51:41 \o/ 2011-04-08T13:55:01 *** tuxmaniac has quit IRC 2011-04-08T13:55:39 ok done 2011-04-08T13:56:06 :D 2011-04-08T13:56:56 11 proposals in so far 2011-04-08T14:01:10 pencils down everyone 2011-04-08T14:04:21 Melange sent 164,000 emails sent in the past 12 hours 2011-04-08T14:05:12 5474 student proposals overall 2011-04-08T14:05:32 the last proposal was created on 2011-04-08 18:59:56 2011-04-08T14:05:52 :) 2011-04-08T14:06:17 lol. that was close! 2011-04-08T14:06:40 ' doubled the numbers in the last 24 hours ... 2011-04-08T14:08:04 SRabbelier: "numapps" is 417 mentoring orgs applied and 175 were accepted this year. In 2010, 367 mentoring orgs applied and 151 were accepted; 5539 proposals were submitted by 3464 students, and 1026 were accepted 2011-04-08T14:08:13 random stats from gsoc channel 2011-04-08T14:11:16 What are the rtems proposals? Besides the ones on the wiki? 2011-04-08T14:12:59 haven't checked the lists against each other.. but from memory iso9660, hypervisor, mmu, cache, gcj, posix timing tests, posix compliance tests, port of resource reservation framework, and lua 2011-04-08T14:13:15 that order does not reflect my scoring.. only random nature of failing mind 2011-04-08T14:14:00 Resource Reservation Framework? 2011-04-08T14:14:23 http://www.frescor.org/ 2011-04-08T14:16:16 cool 2011-04-08T14:42:46 time for me to head home.. see you all soon 2011-04-08T14:42:50 *** DrJoel has quit IRC 2011-04-08T15:10:03 *** hexene has quit IRC 2011-04-08T15:26:44 *** panzon has joined #rtems 2011-04-08T15:27:01 hi 2011-04-08T15:27:10 is someone there? 2011-04-08T15:28:28 I have a problem when I follow the "Quick Start" instructions 2011-04-08T15:31:22 when I execute: sparc-rtems4.11-gdb `find . -name ticker.exe` 2011-04-08T15:31:39 I get sparc-rtems4.11-gdb: error while loading shared libraries: libtinfo.so.5: cannot open shared object file: No such file or directory 2011-04-08T15:39:12 Hi. 2011-04-08T15:39:22 Did you build the tools? or are they prebuilt? 2011-04-08T16:24:24 *** dufault has joined #rtems 2011-04-08T16:25:02 OK, I'm showing my age. I haven't used IRC before. Am I now in the RTEMS IRC? 2011-04-08T16:43:37 yes you are~ 2011-04-08T16:45:17 Ah, good. I guess I thought it was very active. 2011-04-08T16:46:00 sometimes. It depends on the time of day 2011-04-08T16:47:53 I'm hooking up to help out with GSOC if I can help. 2011-04-08T17:12:53 I think there isn't any student in here.. 2011-04-08T17:22:00 *** methril_work has quit IRC 2011-04-08T17:26:27 *** core-ix has joined #rtems 2011-04-08T17:33:42 *** cdcs has quit IRC 2011-04-08T17:45:30 *** madrazr has quit IRC 2011-04-08T18:46:22 *** panzon has quit IRC 2011-04-08T20:44:38 *** xiangfu has joined #rtems 2011-04-08T20:55:18 *** shayden has quit IRC 2011-04-08T21:00:35 *** shayden has joined #rtems 2011-04-08T21:17:14 *** easwar has joined #rtems 2011-04-08T21:21:55 *** xiangfu has quit IRC 2011-04-08T21:43:07 *** shayden has quit IRC 2011-04-08T21:45:37 *** shayden has joined #rtems 2011-04-08T22:12:08 *** zwj has joined #rtems 2011-04-08T22:25:49 *** zwj has quit IRC 2011-04-08T22:26:13 *** zwj has joined #rtems 2011-04-08T23:32:08 *** zwj has quit IRC 2011-04-09T00:08:29 *** arvind_khadri has quit IRC 2011-04-09T00:11:47 *** sqm has joined #rtems 2011-04-09T00:16:01 *** sqm has quit IRC 2011-04-09T00:20:24 *** sqm has joined #rtems 2011-04-09T03:10:15 *** panzon has joined #rtems 2011-04-09T03:27:17 ...Hi 2011-04-09T03:28:27 I'm looking for some help 2011-04-09T03:29:48 yesterday I was followint the "Quick Start" 2011-04-09T05:37:00 *** panzon_ has joined #rtems 2011-04-09T05:40:35 *** panzon has quit IRC 2011-04-09T06:27:01 *** dufault has quit IRC 2011-04-09T06:28:21 *** dufault has joined #rtems 2011-04-09T06:30:20 *** dufault has left #rtems 2011-04-09T06:38:59 *** panzon_ has quit IRC 2011-04-09T07:44:08 *** panzon_ has joined #rtems 2011-04-09T08:00:17 *** methril has quit IRC 2011-04-09T08:08:02 *** panzon_ has quit IRC 2011-04-09T08:13:22 *** methril has joined #rtems 2011-04-09T09:14:59 *** cdcs has joined #rtems 2011-04-09T09:16:20 *** dufault has joined #rtems 2011-04-09T09:17:23 *** dufault has quit IRC 2011-04-09T09:47:03 *** gedare has joined #rtems 2011-04-09T09:53:45 *** DrJoel has joined #rtems 2011-04-09T09:53:45 *** ChanServ sets mode: +o DrJoel 2011-04-09T10:12:49 *** lekernel has joined #rtems 2011-04-09T10:12:55 hi 2011-04-09T10:13:40 what is the reason for including the RTEMS version in the toolchain executable names? can't they be simply named e.g. lm32-rtems-gcc ? 2011-04-09T10:17:01 becasue the toolchain is different fro each rtems release 2011-04-09T10:17:11 for* 2011-04-09T10:19:18 so when you are building a RTEMS-4.xx version you should use a rtems4.xx- toolchain 2011-04-09T10:22:29 *** gedare has quit IRC 2011-04-09T10:25:31 ok, but why make that so obvious that every application or lib, even those that are portable across rtems releases, have to know about it? 2011-04-09T10:26:04 the linux kernel has a similar problem for modules, still we can simply use the 'gcc' command to compile regular stuff on Linux machines 2011-04-09T10:27:06 so you see any problem (especially autoconf related) if I name my toolchain lm32-rtems-* ? 2011-04-09T10:27:43 the target name is assumed to be the prefix or the tools for matching purposes.. 2011-04-09T10:27:52 many people have multiple rtems branches installed at the same time 2011-04-09T10:28:45 but i think you can name your tools whatever you want if you change the autoconf options. 2011-04-09T10:29:10 yes and it is supposed to work.. but it isn't the way most people do it so might break sometimes 2011-04-09T10:30:13 my biggest concern is some autoconf script might break if run for a "lm32-rtems" prefix without version number. those are such a pain to fix... 2011-04-09T10:30:47 right.. that's what my warnign 2011-04-09T10:32:38 you only will know if you try :) 2011-04-09T10:34:35 btw, what's the difference between, say, a ppc-rtems4.9 toolchain and a ppc-rtems4.10 one? 2011-04-09T10:35:08 why can't we use the 4.10 toolchain to compile things for 4.9? 2011-04-09T11:11:06 *** sqm has quit IRC 2011-04-09T11:11:28 different libc for starters.. the .h files for posix functionality have to be in sync 2011-04-09T11:12:31 worse problems are possible because all testing was with a particular gcc. We saw trouble a few years ago when gcc starting hoisting code outside of interrupt disable critical sections.. had to tinker with the isr disable code to tell gcc it was a memory barrier.. gcc can improve and break things.. 2011-04-09T11:52:27 *** DrJoel has quit IRC 2011-04-09T13:30:55 *** madrazr1 has joined #rtems 2011-04-09T13:39:28 *** easwar has left #rtems 2011-04-09T16:35:26 *** cdcs has quit IRC 2011-04-09T17:45:45 *** madrazr1 is now known as madrazr 2011-04-09T17:45:53 *** madrazr has joined #rtems 2011-04-09T18:00:05 *** lekernel has quit IRC 2011-04-09T19:22:17 *** madrazr has quit IRC 2011-04-09T21:40:45 *** lj8175 has joined #rtems 2011-04-09T21:48:03 *** easwar has joined #rtems 2011-04-09T21:48:03 *** easwar has joined #rtems 2011-04-09T22:04:40 *** xiangfu has joined #rtems 2011-04-09T23:04:45 *** lj8175 has quit IRC 2011-04-09T23:27:07 *** methril has quit IRC 2011-04-09T23:27:26 *** methril has joined #rtems 2011-04-09T23:39:37 *** sqm has joined #rtems 2011-04-09T23:42:31 *** xiangfu has quit IRC 2011-04-10T00:49:48 *** sqm has quit IRC 2011-04-10T00:50:12 *** sqm has joined #rtems 2011-04-10T01:10:45 *** lj8175 has joined #rtems 2011-04-10T01:38:17 hi, does somebody have used gcc-trunk code to compile rtems-gcc ? 2011-04-10T03:23:11 *** xiangfu has joined #rtems 2011-04-10T03:27:34 *** xiangfu has quit IRC 2011-04-10T03:39:39 *** xiangfu has joined #rtems 2011-04-10T03:40:12 *** xiangfu_ has joined #rtems 2011-04-10T03:51:54 *** sqm has quit IRC 2011-04-10T04:37:04 *** sqm has joined #rtems 2011-04-10T05:49:53 *** lj8175 has quit IRC 2011-04-10T06:36:10 *** sqm has quit IRC 2011-04-10T06:40:21 *** sqm has joined #rtems 2011-04-10T07:32:24 *** lj8175 has joined #rtems 2011-04-10T07:39:11 *** lj8175 has quit IRC 2011-04-10T07:39:34 *** lj8175 has joined #rtems 2011-04-10T08:30:17 *** cdcs has joined #rtems 2011-04-10T08:56:57 Hi, does RTEMS support getcontext ? 2011-04-10T09:01:35 seems not , because cann't find it in source code 2011-04-10T09:10:23 i have used only setjmp longjmp 2011-04-10T09:10:53 so.. gues not :) 2011-04-10T09:19:13 rokka :) 2011-04-10T10:02:12 *** xiangfu_ has quit IRC 2011-04-10T10:25:04 Hi, how big is RTEMS/i386's stack ? 2011-04-10T10:27:55 you can configure it with 2011-04-10T10:28:03 #define CONFIGURE_INIT_TASK_STACK_SIZE 2011-04-10T10:28:03 #define CONFIGURE_MINIMUM_TASK_STACK_SIZE 2011-04-10T10:28:14 or with parameter when you create task 2011-04-10T10:28:40 put stack size as number after those defines 2011-04-10T10:29:00 like #define CONFIGURE_INIT_TASK_STACK_SIZE (32*1024) 2011-04-10T10:29:18 those have to be before #include 2011-04-10T10:29:19 rokka: you are so nice, thank you! 2011-04-10T10:29:49 :) 2011-04-10T10:30:43 rokka, and what is the default ? 2011-04-10T10:33:34 maybe 4096 bytes 2011-04-10T10:34:12 it's cpu dependent 2011-04-10T10:34:34 don't know.. :) 2011-04-10T10:34:46 *** dufault has joined #rtems 2011-04-10T10:34:52 i386 is 4K 2011-04-10T10:35:39 mips is 8k 2011-04-10T10:35:40 many are 4k ... 2011-04-10T10:37:34 :) 2011-04-10T10:42:18 rokka: thank you, my case is working after change to 64k :) 2011-04-10T10:42:23 :D 2011-04-10T10:42:24 good 2011-04-10T10:53:07 *** sqm has quit IRC 2011-04-10T10:55:29 *** cdcs has quit IRC 2011-04-10T11:02:31 *** lj8175 has quit IRC 2011-04-10T11:08:54 *** shayden has quit IRC 2011-04-10T11:10:56 *** shayden has joined #rtems 2011-04-10T11:21:54 *** methril_ has joined #rtems 2011-04-10T11:26:54 *** methril has quit IRC 2011-04-10T11:27:00 *** easwar has quit IRC 2011-04-10T11:36:54 *** easwar has joined #rtems 2011-04-10T11:37:16 *** antgreen has quit IRC 2011-04-10T11:38:43 *** antgreen has joined #rtems 2011-04-10T14:23:34 *** dufault has left #rtems 2011-04-10T15:49:39 anyone home? 2011-04-10T15:51:30 :D 2011-04-10T15:51:43 you mean available for helping? 2011-04-10T15:53:44 rokka, more like to find a mentor for the gprof project 2011-04-10T15:54:15 or some suggestions or so 2011-04-10T16:00:37 oh.. im too busy to mentor :l 2011-04-10T16:00:44 and not so good with rtems yet 2011-04-10T16:00:53 list of mentors: http://wiki.rtems.org/wiki/index.php/RTEMSSummerOfCode 2011-04-10T16:01:27 i think they are better available at business time :) 2011-04-10T16:05:36 rokka, business time? what's the reference timezone? 2011-04-10T16:09:49 it was just a guess, and i think they are around the world.. 2011-04-10T16:10:05 i meant monday-friday 2011-04-10T16:10:37 oh 2011-04-10T16:10:39 :) 2011-04-10T16:10:56 where are you from 2011-04-10T16:11:07 rokka, currently in Charlotte, NC 2011-04-10T16:11:09 USA 2011-04-10T16:11:14 originally from India 2011-04-10T16:11:18 :D 2011-04-10T16:11:55 i think DrJoel will be here in about 16 hours 2011-04-10T16:12:33 8:00AM CDT-8 ? 2011-04-10T16:13:33 rokka, ok 2011-04-10T16:15:44 rokka, do you mean 1600 hours CDT? 2011-04-10T16:16:02 coz if it was 8AM CDT, he would be here already :) 2011-04-10T16:16:15 i mean ;): wait_hours(16); 2011-04-10T16:16:49 whoa 2011-04-10T16:16:53 alright 2011-04-10T16:16:59 :D 2011-04-10T16:17:11 so monday 8:00 AM CDT-8 2011-04-10T16:20:12 oh 2011-04-10T16:20:14 right 2011-04-10T16:20:25 business hours 2011-04-10T16:20:31 got it 2011-04-10T16:20:51 im having trouble with compiling i386-rtems4.11 :DD 2011-04-10T16:20:57 toolchain 2011-04-10T16:21:12 rokka, hmm 2011-04-10T16:21:15 i did it before.. but currently.. gcc hates me :D 2011-04-10T16:21:34 i posted rtems-users mailing list 2011-04-10T16:21:38 have you joined it? 2011-04-10T16:21:44 rokka, yes 2011-04-10T16:22:29 ok.. my mail should be latest one 2011-04-10T16:22:37 mysterious problem 2011-04-10T16:22:44 i think my ubuntu is broken :D 2011-04-10T16:24:39 rokka, no idea, I'm completely new to RTEMS and have only built HelloWorld so far 2011-04-10T16:24:51 :D 2011-04-10T16:24:52 with precompiled toolchains at that 2011-04-10T16:25:03 yep. thats good way to start 2011-04-10T16:25:58 so you study at some university? 2011-04-10T16:26:28 rokka, looks like configure is failing, doubt there is use in doing make without a successful configure 2011-04-10T16:26:50 rokka, yes, university of north carolina at charlotte 2011-04-10T16:27:05 configures are not perfect.. thats too bad :l 2011-04-10T16:27:31 okey 2011-04-10T16:28:00 do you have good courses about embbeded programming 2011-04-10T16:28:50 rokka, quite a bit 2011-04-10T16:30:14 ok 2011-04-10T16:31:30 what do they use as embedded os? 2011-04-10T16:37:56 rokka, we do bare-metal mostly 2011-04-10T16:38:04 but MicroC/OS-II 2011-04-10T16:38:25 is what we're taught as an embedded OS 2011-04-10T16:42:19 ok.. rtems is only i have tried 2011-04-10T16:42:24 only one 2011-04-10T16:43:13 application was a sounding rocket :) 2011-04-10T17:08:39 where did you get idea for gprof? 2011-04-10T17:16:06 rokka, it was an open project on the wiki 2011-04-10T17:16:44 *** projectgus has joined #rtems 2011-04-10T17:31:45 okay:) 2011-04-10T18:06:53 *** gedare has joined #rtems 2011-04-10T18:49:08 *** kiwichris has joined #rtems 2011-04-10T18:49:21 gedare, hi 2011-04-10T18:49:27 kiwichris: hello 2011-04-10T18:49:38 I'm skimming some of the project proposals. 2011-04-10T18:49:39 I agree with the comments on the cache project 2011-04-10T18:49:49 yeah.. it is a bit of a big-bite I think 2011-04-10T18:50:00 I would like to see an order of importance to RTEMS. 2011-04-10T18:50:10 I think that is something missing on the open projects page 2011-04-10T18:50:19 I meant to mention it last year 2011-04-10T18:50:22 On testing, I also raised this. I am wondering if something in flagged as not tested is better than not. 2011-04-10T18:50:50 I don't quite follow you 2011-04-10T18:50:52 I am meaning in terms of which cache processors to do 2011-04-10T18:51:00 Oh, yes 2011-04-10T18:51:21 I feel for some processors unless we have hardware and some code to test with they will be difficult to validate 2011-04-10T18:51:33 Yeah, and I wouldn't want to commit code like that. 2011-04-10T18:51:38 This means we need a user with real hardware and a real driver to sign off the code 2011-04-10T18:51:42 Right. 2011-04-10T18:51:49 It is a catch 22 2011-04-10T18:52:03 If the code is present it stops people adding to BSBs or applications. 2011-04-10T18:52:15 Actually, it means the work shouldn't be done unless someone is willing to step forward and test it or wants it. 2011-04-10T18:52:35 If it is present but built with a compile warning that may be a suitable work around 2011-04-10T18:52:55 But if there are a couple sims that support it, or easily tested HW, then it can be good to have the capabilities implemented to create cache manager tests 2011-04-10T18:52:59 yeah 2011-04-10T18:53:01 Some users come and go over the years or sit on a release for years 2011-04-10T18:53:06 Yeah 2011-04-10T18:53:23 Oh yes we should order the support done based on actual ability to test 2011-04-10T18:53:44 Importance and testing all first. 2011-04-10T18:53:58 You'd have to talk to Joel about whether doing something with compiler warnings about untested code is sufficient... 2011-04-10T18:54:06 If the project defines 100 devices and we get 10 done then that is 10 less we need to do 2011-04-10T18:54:21 Joel and I have been talking about this topic a bit lately. 2011-04-10T18:54:25 Ok 2011-04-10T18:54:36 I'd rather see 2 devices proposed and 10 devices finished. 2011-04-10T18:54:40 ;) 2011-04-10T18:54:44 The situation we were chating about is the USB stack. 2011-04-10T18:54:57 Oh, yeah... that is a tough one 2011-04-10T18:55:08 He wanted me to think about doing it this summer, but I'm busy with "real" work. 2011-04-10T18:55:16 We cannot get this code merged because I will not allow it in until a busspace API is added to the score 2011-04-10T18:55:42 We need an score namespace RTEMS specific bus space type API. 2011-04-10T18:55:49 That makes sense 2011-04-10T18:55:51 On to this we map the BSD API. 2011-04-10T18:56:03 We can then port BSD code and map it to RTEMS. 2011-04-10T18:56:14 And the various other bus devices that are lying around 2011-04-10T18:56:15 How-ever as we wait for the bus space API the usb stack rots. 2011-04-10T18:56:20 *nod* 2011-04-10T18:56:21 Yes 2011-04-10T18:56:39 So we are wondering about a --enable-experimental configure option 2011-04-10T18:56:56 so we can release RTEMS yet have long term projects in main. 2011-04-10T18:57:11 It avoids using branches with version control 2011-04-10T18:57:29 Yeah, it makes sense. 2011-04-10T18:57:35 Just have to sell the SC on it 2011-04-10T18:57:46 (ralf) 2011-04-10T18:57:49 So I see the cache support in the source but not tested as maybe experimental 2011-04-10T18:58:20 Yeah. I like the idea in principal, in practice it has to be moderated to make sure there isn't too much "experimental" bitrot 2011-04-10T18:58:24 I would love to be able to add the bus space API but I have other tasks. The current one is dynamic loading and this is a muct to finish. 2011-04-10T18:58:49 yeah 2011-04-10T18:58:53 I would look at an experimental build each week with the build bot 2011-04-10T18:59:21 I am to add email reports to the build bot so we could see how things go. 2011-04-10T18:59:22 ok, that seems reasonable. I only mention it because, e.g., I know Joel hardly ever tests with debugging on :) 2011-04-10T18:59:40 :) 2011-04-10T18:59:45 I always do. 2011-04-10T18:59:52 And so does Seb 2011-04-10T19:00:10 me too... although I drift from the CVS head sometimes. 2011-04-10T19:00:29 anyway, perhaps send an e-mail on the users list 2011-04-10T19:00:35 That is normal. I sometimes have many head checked out 2011-04-10T19:00:48 see if any users are interested in seeing experimental features in releases 2011-04-10T19:01:16 There are some. I am currently planning to clean up the file system layer to add locking etc 2011-04-10T19:01:36 does that relate to till's threaded newlib patches at all? 2011-04-10T19:01:48 he was complaining about that being unresolved 2011-04-10T19:01:56 I am not sure. I may take a look at them. 2011-04-10T19:02:07 bugzillarot ;) 2011-04-10T19:02:14 It is the fact pathlocs etc are not reference counted and locked 2011-04-10T19:02:29 There is a PR for this issue. 2011-04-10T19:02:44 *nod* I think I skimmed it... tbh I avoid most FS issues though. 2011-04-10T19:03:04 If you look at open you can see calls to a file system with out locks and the file system could be unmounted while in open, as an example. 2011-04-10T19:03:15 yeah 2011-04-10T19:03:21 The other is the offset in iop is not thread safe. 2011-04-10T19:03:27 Look at read 2011-04-10T19:03:42 A read and seek at the same time breaks the code. 2011-04-10T19:04:22 There are important applications using files and the RFS with many hundreds of files. 2011-04-10T19:05:12 maybe there is a better way to track open projects than the wiki, that can allow scoring or something 2011-04-10T19:06:01 My time is taken with support work and this rules the priority for me 2011-04-10T19:06:17 it has to :) 2011-04-10T19:06:43 I have an ELF loader loading object files from archives or actual files on SPARC (not 64bit) 2011-04-10T19:06:46 anyway, it isn't terribly important, except when gsoc comes 2011-04-10T19:06:56 nice 2011-04-10T19:07:05 PowerPC i386 m68k and ARM 2011-04-10T19:07:23 I am now working on an RTEMS linker to handling the object file dependences 2011-04-10T19:07:24 this ties in well with the lua project 2011-04-10T19:07:35 Great. It is a dlopen API. 2011-04-10T19:08:14 I'm not sure what lua uses, it is mostly layered on posix, so it might have to be hacked on a little if it uses fork/exec 2011-04-10T19:08:19 I have the host RTEMS linker reading all the symbols for all libraries. I now have to implement the resolver 2011-04-10T19:08:32 Python uses dlopen 2011-04-10T19:08:37 ok 2011-04-10T19:09:03 I suspect it is dlopen because it is usually just one process 2011-04-10T19:09:44 The really nice thing about dynamic loading which only occurred to me the other day is no need for linkcmd files to build an applicatoin 2011-04-10T19:10:17 Yeah, although they'll never go away. :) 2011-04-10T19:10:31 The real win is that GPL code can be dynamically linked against 2011-04-10T19:10:54 Can it ? Is dlopen ok ? I would not be so sure. 2011-04-10T19:11:06 My understanding is that it can, maybe I'm mistaken. 2011-04-10T19:11:11 It makes LGPL code easier to manage 2011-04-10T19:11:13 IANAL :) 2011-04-10T19:11:18 that definitely can be. 2011-04-10T19:11:33 I thought the OS trap type interface was ok. 2011-04-10T19:11:58 The Linux kernel modules is not proven yet 2011-04-10T19:12:16 oh 2011-04-10T19:13:17 And that is the other way around to our problem. I suspect it will never get tested until you have an 'iPad" type product that someone wants to go after 2011-04-10T19:17:53 you use python with rtems, right? 2011-04-10T19:18:05 * gedare pokes kiwichris 2011-04-10T19:18:08 yeah I have in the past 2011-04-10T19:18:19 you just link the required modules in a C header file? 2011-04-10T19:18:31 and the linker only pulls in those python modules? 2011-04-10T19:18:47 I required the application to create a C file that pulled in the modules 2011-04-10T19:19:05 *nod* 2011-04-10T19:19:25 just making sure I understood the recommended flow... I know the student who is proposing the lua project, did you read the proposal? 2011-04-10T19:19:38 It worked really well. Within a minute of getting it going I was read my email on an IMAP server on a credit card size Coldfire board 2011-04-10T19:19:57 nice 2011-04-10T19:20:04 Yes. I like it. I think any language on RTEMS is a win. 2011-04-10T19:20:20 i was thinking you or peter would be good mentors for that project.. 2011-04-10T19:20:31 if it got in :) 2011-04-10T19:20:42 You can have a C/C++ realtime part to your application and a control management part in other language 2011-04-10T19:20:50 Great. 2011-04-10T19:20:53 yeah 2011-04-10T19:21:05 OK good to know you are interested. 2011-04-10T19:21:17 I told him to send in a proposal, since he was looking at this already 2011-04-10T19:21:22 In one project we had boxes all around the place and they read configurations from POP email accounts. 2011-04-10T19:21:26 wasn't sure if anyone would care. :) 2011-04-10T19:21:52 I do. I have talked with a Python dev about getting RTEMS support into the Python main code. 2011-04-10T19:21:54 nice 2011-04-10T19:22:25 I have also been involved with Parrot on RTEMS. My lack of time has lapsed that work which is a shame 2011-04-10T19:23:03 We also have Go from last year and now Java this year. 2011-04-10T19:23:28 yeah 2011-04-10T19:23:48 What I would like is a way to package these 3rd party bits in RPMs 2011-04-10T19:24:06 Then a repo install and you can add it to your application 2011-04-10T19:25:20 that's the problem with some of these language projects... they're loosely coupled to the rtems distribution 2011-04-10T19:26:06 Yeah. Once RTL (runtime loading) is sorted I hope we will have a clearer picture on how to proceeed 2011-04-10T19:26:12 lol i just read the ada proposal 2011-04-10T19:26:24 it made me laugh 2011-04-10T19:26:36 good humour 2011-04-10T19:26:42 I agree... 2011-04-10T19:27:40 there are a lot of mmu-related projects floating around too 2011-04-10T19:27:49 I think these need to be unified and serialized somehow 2011-04-10T19:28:53 Yes. I think the top level API is important because it sets the direction for the actual low level parts. 2011-04-10T19:29:22 yeah. defining clearly the need and interface for RTEMS MMU services is top priority here 2011-04-10T19:29:40 gprof is to use the capture engine how-ever that is an event system and grof is sampling based. 2011-04-10T19:30:06 someone hacked gprof to use the clock isr 2011-04-10T19:30:08 I still like adding an attribute call to the region API in the classic API 2011-04-10T19:31:14 I like the ISO9660 project. A CDROM for RTEMS for a PC is a nice idea. 2011-04-10T19:32:45 yeah, and it can be tested with an iso 2011-04-10T19:32:57 and the ramdisk driver 2011-04-10T19:33:08 yeah 2011-04-10T19:33:24 This removes the ATAPI mess and anyone hacking the current ATA driver which I do not want to see happen 2011-04-10T19:33:33 :) 2011-04-10T19:33:49 Once we have the bus space API we can get last years GSoC SATA code into RTEMS 2011-04-10T19:34:11 bus space API bus space API grrrrr 2011-04-10T19:40:03 *** xiangfu has joined #rtems 2011-04-10T19:42:29 is there a write-up anywhere of the bus space API project? :) 2011-04-10T19:47:34 kiwichris: do you know how the gcj project will be tested? is there a jvm available? 2011-04-10T19:48:05 or does the gcj do java-to-c translation of everything 2011-04-10T19:48:11 Does gjc generate native code ? 2011-04-10T19:48:17 I think it can. 2011-04-10T19:48:26 begs the question why to use java :) 2011-04-10T19:48:30 I think it is straight to machine code 2011-04-10T19:48:35 ok that makes sense 2011-04-10T19:48:44 Because you have an existing code base. 2011-04-10T19:48:47 yeah 2011-04-10T19:49:08 i'd like to see a nicely abstracted GC layer 2011-04-10T19:49:20 so that real-time garbage collectors can be implemented 2011-04-10T19:49:23 RTEMS as a project should enable users rather than dictate. 2011-04-10T19:49:25 and plugin 2011-04-10T19:49:28 *nod* 2011-04-10T19:49:55 Yeah this is a big area of research and part of why I want Parrot on RTEMS, 2011-04-10T19:50:10 The Parrot VM need a real-time GC 2011-04-10T19:50:21 There are at least 2 real-time JVMs I've heard of that run RTEMS 2011-04-10T19:50:24 It is a weakness with most VMs 2011-04-10T19:50:35 JamaicaVM and FijiVM 2011-04-10T19:50:50 Do they have a real-time GC ? 2011-04-10T19:51:11 FijiVM does, i met the guy who wrote it.. it is his research phd thesis work 2011-04-10T19:51:16 real time garbage collection 2011-04-10T19:51:23 Nice. 2011-04-10T19:51:36 they had an open source version too...although I lost track of that one 2011-04-10T19:51:43 but the FijiVM is commercial 2011-04-10T19:51:57 http://www.ovmj.net/oscj/ 2011-04-10T19:52:36 Nice. Is this in the WIki ? 2011-04-10T19:53:41 i don't know. 2011-04-10T19:53:54 It should be. It looks great 2011-04-10T20:54:22 *** gedare has quit IRC 2011-04-10T21:03:03 Is gedare likely to be back? 2011-04-10T21:03:28 no idea. Can I help ? 2011-04-10T21:05:52 kiwichris, I was thinking of who the mentor would be for the gprof project 2011-04-10T21:05:57 and from the Wiki history 2011-04-10T21:06:07 gedare wrote the initial page 2011-04-10T21:06:24 assuming of course, the project gets accepted 2011-04-10T21:07:09 The mentors put their hands up then we may have a chat and handle any resolutions then Joel oks this. 2011-04-10T21:07:54 kiwichris, ok 2011-04-10T21:16:17 *** sqm has joined #rtems 2011-04-10T21:53:05 *** lj8175 has joined #rtems 2011-04-10T22:20:36 *** antgreen has quit IRC 2011-04-10T22:21:00 *** antgreen has joined #rtems 2011-04-10T23:22:28 *** lj8175 has quit IRC