2010-06-14T02:01:05 *** sebhub has joined #rtems 2010-06-14T02:01:19 good morning 2010-06-14T02:20:01 morning 2010-06-14T02:25:47 *** exception13 has joined #rtems 2010-06-14T04:48:05 *** tuxmaniac has joined #rtems 2010-06-14T05:02:20 *** lcpfnvc_ has quit IRC 2010-06-14T06:09:05 *** lcpfnvc has joined #rtems 2010-06-14T07:40:59 * DrJoel waves hello.. 2010-06-14T07:45:13 *** DrJoel has quit IRC 2010-06-14T09:05:39 *** methril_work has quit IRC 2010-06-14T09:12:42 *** tonylijo has joined #rtems 2010-06-14T09:18:00 *** dr__house has joined #rtems 2010-06-14T09:52:17 *** bubaflub has joined #rtems 2010-06-14T09:55:50 *** lcpfnvc has quit IRC 2010-06-14T10:05:19 *** tonylijo has quit IRC 2010-06-14T10:07:16 *** Fallenou has quit IRC 2010-06-14T10:17:11 *** lekernel has quit IRC 2010-06-14T10:18:54 *** DrJoel has joined #rtems 2010-06-14T10:19:02 *** ChanServ sets mode: +o DrJoel 2010-06-14T10:44:10 *** Fallenou has joined #rtems 2010-06-14T10:56:07 *** gedare_ has joined #rtems 2010-06-14T10:56:36 *** gedare_ is now known as gedare 2010-06-14T11:14:43 *** gedare has quit IRC 2010-06-14T11:43:40 *** sebhub has quit IRC 2010-06-14T11:47:48 *** dr__house has quit IRC 2010-06-14T11:48:03 *** dr__house has joined #rtems 2010-06-14T12:06:34 *** alinrus has left #rtems 2010-06-14T12:15:14 *** tonylijo has joined #rtems 2010-06-14T12:33:32 *** jacob123 has joined #rtems 2010-06-14T12:42:41 *** gedare has joined #rtems 2010-06-14T13:22:41 i have installed the tool chain for building rtems in fedora 2010-06-14T13:23:04 and took rtems from cvs 2010-06-14T13:23:46 dr__house: can you help me in building rtems 2010-06-14T13:24:31 tonylijo: cool 2010-06-14T13:25:04 tonylijo: so you did yum install of rtems packages? 2010-06-14T13:25:32 ya 2010-06-14T13:25:38 tonylijo: cool 2010-06-14T13:25:45 much easier that way, isn't it? :-D 2010-06-14T13:26:24 tonylijo: so now building rtems. First run bootstrap from the RTEMS source directory. 2010-06-14T13:26:35 yes it was pretty easier compared to building form source 2010-06-14T13:34:02 *** tuxmaniac has quit IRC 2010-06-14T13:34:55 *** jacob1231 has joined #rtems 2010-06-14T13:35:37 tonylijo: so did you run bootstrap? 2010-06-14T13:35:43 *** tonylijo1 has joined #rtems 2010-06-14T13:35:53 dr__house: it is showing an error like "configure.ac:5: error: Autoconf version 2.65 or higher is required" but bootstrap is still working 2010-06-14T13:36:27 *** jacob123 has quit IRC 2010-06-14T13:36:37 tonylijo: echo $PATH 2010-06-14T13:36:43 did you install the rtems autotools? 2010-06-14T13:36:50 yum install "*rtems*auto*" 2010-06-14T13:36:59 *** tonylijo has quit IRC 2010-06-14T13:37:42 DrJoel: /usr/local/bin:/usr/bin:/bin:/usr/local/sbin:/usr/sbin:/sbin:/home/tonylijo/bin 2010-06-14T13:38:52 tonylijo1: ah ok. You forgot to export /opt/rtems-4.xx/bin to $PATH 2010-06-14T13:39:11 tonylijo1: export PATH=/opt/rtems-4.11/bin:$PATH 2010-06-14T13:39:47 or rtems-4.10 depending on what the path is 2010-06-14T13:42:33 dr__house: now it is going well 2010-06-14T13:42:58 tonylijo1: cool 2010-06-14T13:48:51 dr__house: bootstrap finished 2010-06-14T13:49:45 tonylijo1: cool, now create a new directory outside of the source tree, navigate into it and run configure with the right options 2010-06-14T13:50:05 tonylijo1: you may refer to this http://www.rtems.com/wiki/index.php/Quick_Start 2010-06-14T13:50:43 * DrJoel frowns.. AC units are acting up at OAR. 96F/35C outside, 80F/26C and rising inside 2010-06-14T13:50:49 * dr__house is going for a stroll in the rain, will be back in a while. 2010-06-14T13:51:01 DrJoel: ouch! Power failure? 2010-06-14T13:59:24 *** jacob1231 has left #rtems 2010-06-14T14:07:03 not power.. HVAC 2010-06-14T14:20:58 dr__house: when i do sparc-rtems-4.11-gdb `find . -name ticker.exe` i got like http://pastebin.com/MiXthnJ9 2010-06-14T14:21:29 dr__house: now what is the next step 2010-06-14T14:33:02 tonylijo1: it looks like everything is working! 2010-06-14T14:35:54 bubaflub: thanks 2010-06-14T14:36:03 that shows that the ticker example program 2010-06-14T14:36:11 run correctly on sparc instruction simulator 2010-06-14T14:36:19 so it looks like you've got everything working 2010-06-14T14:36:28 you can run some more examples if you'd like, or start getting your code on 2010-06-14T14:36:44 how can i begin 2010-06-14T14:36:58 depends on what you want to do next 2010-06-14T14:38:30 i am interested to do some thing on iso filesystem . 2010-06-14T14:39:21 i have to do hello world first 2010-06-14T14:39:28 i can i start 2010-06-14T14:39:32 yeah, go ahead and compile the hello world example 2010-06-14T14:39:46 where can i get it 2010-06-14T14:40:35 i have done some simulation and printed hello world , as i have seen in some tutorials 2010-06-14T14:40:41 did you check out examples-v2? 2010-06-14T14:40:51 i imagine you did because you've got the ticker example running 2010-06-14T14:41:23 i get only .exe files only 2010-06-14T14:41:35 right 2010-06-14T14:41:38 you can do the same thing 2010-06-14T14:41:38 i have runed the hello.exe 2010-06-14T14:41:46 sparc-rtems4.10-gdb hello.exe 2010-06-14T14:41:49 ah, then you've done that 2010-06-14T14:42:03 that all just verifies that you have a correct installation of RTEMS and the tools and your system are setup correctly 2010-06-14T14:42:26 where can i get the code 2010-06-14T14:42:41 i mean the code of hello.c 2010-06-14T14:43:58 hmmmm, it should be in the checkout as well 2010-06-14T14:44:16 ok , i will find it out 2010-06-14T14:47:11 bubaflub: i got it 2010-06-14T14:47:16 awesome. 2010-06-14T14:47:21 go ahead and start looking at the source code 2010-06-14T14:47:28 if you can find the ticker example source code, look at that as well 2010-06-14T14:47:35 that'll show you some of the other features 2010-06-14T14:48:50 bubaflub: well , are there any c manuals for rtems avilable 2010-06-14T14:48:59 yeah, you can check out the RTEMS wiki and the user docs online 2010-06-14T14:49:20 http://wiki.rtems.org/wiki/index.php/Main_Page 2010-06-14T14:49:25 http://www.rtems.org/onlinedocs/releases/rtemsdocs-4.9.4/share/rtems/html/ 2010-06-14T14:49:27 respectively 2010-06-14T14:50:08 bubaflub: ok 2010-06-14T14:50:34 bubaflub: thanks for the great help you have done 2010-06-14T14:50:40 no problem 2010-06-14T14:52:52 *** tonylijo1 has left #rtems 2010-06-14T15:40:03 bubaflub: if you see tony around, for the ISO9660, he will probably need to be on qemu/pc386. 2010-06-14T15:40:18 okey dokey. i'll let him know. 2010-06-14T15:41:50 DrJoel: i'm having some linking errors with my project... do we have ptheads somewhere? 2010-06-14T15:42:04 * dr__house heads off for the day. Catch you all later. 2010-06-14T15:42:10 good day DrJoel, bubaflub 2010-06-14T15:42:14 *** dr__house has quit IRC 2010-06-14T15:46:23 *** cdcs has joined #rtems 2010-06-14T15:58:55 *** tuxmaniac has joined #rtems 2010-06-14T15:59:04 *** tuxmaniac has joined #rtems 2010-06-14T16:03:11 bubaflub: they are in the main rtems library. 2010-06-14T16:03:24 ah, lemme see if i can add that to the search path 2010-06-14T16:03:37 they are in librtemscpu.a if posix was enabled 2010-06-14T16:04:45 DrJoel: gah! that's my problem - i disabled posix when i configured and build my BSP 2010-06-14T16:06:15 :) 2010-06-14T16:06:49 ok, i'm going to try and rebuild it with --enable-posix 2010-06-14T16:06:52 and then try building parrot again 2010-06-14T16:06:54 should be able to find it 2010-06-14T16:11:19 DrJoel: another, random thing - is there an easy way to get what compiler warnings RTEMS was compiled with? 2010-06-14T16:13:26 *** tuxmaniac has quit IRC 2010-06-14T16:13:31 warning flags? or just compilation flags? 2010-06-14T16:13:51 errr, the warnings that get passed during compilation 2010-06-14T16:13:56 like -Wall, -Wimplicit blah blah 2010-06-14T16:16:08 hmmm.. for just those I don't know. Best to ask on rtems-users 2010-06-14T16:16:37 okey dokey, it's not critical now. i'll shoot an email to the list after class tonight 2010-06-14T16:18:02 *** kiwichris has joined #rtems 2010-06-14T16:24:04 *** exception13 has left #rtems 2010-06-14T16:25:49 DrJoel, hi 2010-06-14T16:26:44 *** bubaflub has quit IRC 2010-06-14T16:29:35 hey chris 2010-06-14T16:29:50 The specs parse should be ok to do. 2010-06-14T16:30:12 so it can be used to update setup.def periodically? 2010-06-14T16:30:41 Not sure about updating, but it could send you the file. 2010-06-14T16:31:09 I suppose it could do a commit if you wanted it to. 2010-06-14T16:33:31 the setup.def or the script? I suppose it could be an rtems-testing/cvs-helper or something. There are scripts in the tree to help work on rtems 2010-06-14T16:34:18 The setup.def 2010-06-14T16:35:04 Claudio ? 2010-06-14T16:35:23 ahh.. just generate it and commit the right version to head and 4.10 branch.. both are long out of date 2010-06-14T16:35:55 Hi. 2010-06-14T16:36:02 Ok. I will take a look later today. I have to check the state of the parse. 2010-06-14T16:36:33 cdcs, has the libbsdports headers helped ? 2010-06-14T16:37:16 To some level, but the headars 2010-06-14T16:37:37 that i needed the most were just empty dummies 2010-06-14T16:38:20 That is easy :) 2010-06-14T16:39:07 Question: i have one file that includes:#include 2010-06-14T16:39:45 since i have libbsdport installed, this file is from libbsdport(i think so) 2010-06-14T16:40:16 but i get..."expected specifier-qualifier-list before 'callout_t' " 2010-06-14T16:40:50 This horrible gcc error message means it cannot find the callout_t type 2010-06-14T16:42:15 Just looking in libbsdport 2010-06-14T16:43:05 yup that was my thought, struct callout is there 2010-06-14T16:43:34 but i don't know if that is a uncompleted port, or the version ported(freebsd) is different from mine 2010-06-14T16:44:48 I cannot tell. I have the attachment but it rather large and I am on a wireless 3G connection at the moment. 2010-06-14T16:45:16 cdcs, do you have an ftp site you can upload to ? 2010-06-14T16:46:06 can it be dropbox? 2010-06-14T16:46:15 What is that ? 2010-06-14T16:47:20 My problem is my build and test hardware has a cable connection and I am remote over 3G. If on an ftp type site I can download and play rather than being it over the 3G via email then copy it back over 3G. 2010-06-14T16:47:43 hi drjoel 2010-06-14T16:47:54 it's not ftp, but http 2010-06-14T16:48:07 hey gedare I noticed you filed a set of PRs for the sparc64 port. 2010-06-14T16:48:22 yep... there's not a lot in terms of how to actually get any of it to work 2010-06-14T16:48:25 but all the code is now there 2010-06-14T16:48:28 kiwichris: should gedare post to the mailing list saying the sparc64 port is available for review in the PRs XXXX-YYYY 2010-06-14T16:48:38 Yes 2010-06-14T16:48:41 first step is to build. 2010-06-14T16:48:46 i'll put together some quick start 2010-06-14T16:48:56 OK gedare you heard chris .. send an email openign the port for review 2010-06-14T16:48:58 building should be easy enough, well, after the tools 2010-06-14T16:49:00 OK 2010-06-14T16:49:08 kiwichris, http://dl.dropbox.com/u/2751858/testbed.tar 2010-06-14T16:49:13 I also think a news item on the web site. Having 64bit is big news. 2010-06-14T16:49:28 fwiw, one of the patches is 64-bit but not sparc-specific 2010-06-14T16:49:41 the first one, which addresses a sizing / type issue in some cpukit/score code. 2010-06-14T16:49:53 kiwichris: I spoke with David Maxwell for 1.5 hours today. RTEMS will be one of the featured projects in the Coverity "state of the world" annual report. We will also soon be upgraded to the latest version of Coverity for another range of issues to pop up. 2010-06-14T16:50:03 PR 1560 2010-06-14T16:50:20 DrJoel, that is great news. 2010-06-14T16:50:28 that's cool 2010-06-14T16:51:39 kiwichris: I think we are down to 2 coverity issues and I have pending fixes for both. Just need run results to close them out 2010-06-14T16:51:45 cdcs, I tend to date any file I put out and next time can it please be compressed ? 2010-06-14T16:52:05 Does this include everything or cpukit ? 2010-06-14T16:53:04 ok will date it and compress it(as soon i find the tar option for that :) 2010-06-14T16:53:44 that is my testbed, you should configure it and build-it...obviously it will not build. 2010-06-14T16:54:01 I have called it ata-testbed-20100614.tar. It is Tuesday here :) 2010-06-14T16:54:41 struct callout from is different from netbsd callout :( 2010-06-14T16:54:57 How different ? 2010-06-14T16:56:23 one has four members, the other has only one. 2010-06-14T16:56:39 like totally different 2010-06-14T16:57:23 I would hack the libbsdport one to add or remove (not which it is) with an ifdef RTEMS_NETBSD_PORT 2010-06-14T16:57:50 Once you get somewhere with this then we can look at each of these and decide what is the best solution. 2010-06-14T16:59:18 i think i will include it locally, "sys/callout.h" 2010-06-14T16:59:32 Ok. 2010-06-14T17:01:31 i will include it all locally, that way i'm sure the files are the right ones 2010-06-14T17:02:03 gedare: are pointers 32 or 64 bits? 2010-06-14T17:02:07 on sparc64 2010-06-14T17:04:10 *** gedare has quit IRC 2010-06-14T17:05:47 * cdcs offtopic... so Australia lost 4-0 in the world cup and USA got a draw on England... well done :) 2010-06-14T17:07:39 * DrJoel has had a long day.. AC failed here at OAR ... earlier was 96F/35C outside, 80F/26C and rising inside. temperature swings are exhausting 2010-06-14T17:07:52 luckily AC is now repaired 2010-06-14T17:08:33 *** DrJoel has quit IRC 2010-06-14T17:17:17 *** DrJoel has joined #rtems 2010-06-14T17:17:17 *** ChanServ sets mode: +o DrJoel 2010-06-14T17:17:27 *** DrJoel has quit IRC 2010-06-14T17:47:17 * cdcs is tired, going to sleep. See you tomorow(or later...) 2010-06-14T17:47:41 *** cdcs has quit IRC 2010-06-14T17:50:46 *** kiwichris has quit IRC 2010-06-14T19:07:29 *** DrJoel has joined #rtems 2010-06-14T19:07:29 *** ChanServ sets mode: +o DrJoel 2010-06-14T19:34:14 Fallenou: how are things going? 2010-06-14T19:36:16 *** kiwichris has joined #rtems 2010-06-14T19:36:29 DrJoel, hi 2010-06-14T19:36:47 hey chris.. just starting to see if the cpukit of sparc64 builds cleanly 2010-06-14T19:36:54 excellent 2010-06-14T19:37:21 on the RFS PR about type names, are type names an actual standard or is it a convention ? 2010-06-14T19:38:20 I haven't looked at the PR in detail but in general if it is public, it starts with rtems_. If not, it doesn't matter as long as it is uncommon enough to avoid conflicts 2010-06-14T19:38:38 which PR.. let me read it again 2010-06-14T19:38:42 Ok 2010-06-14T19:40:17 I am more than happy to change and make it conform how-ever I think we will need a more formal coding standard so this is uniformly handled. 2010-06-14T19:41:18 which PR.. and assigning another to you to double check. 2010-06-14T19:42:00 https://www.rtems.org/bugzilla/show_bug.cgi?id=1556 2010-06-14T19:42:00 https://www.rtems.org/bugzilla/show_bug.cgi?id=1556 2010-06-14T19:42:40 the flood of minor PRs has overwhelmed me. I was wanting to focus on catching up today, Coverity, and maybe sparc64 2010-06-14T19:43:57 ok.. there actually is a convention for this although not written probably. :( See my followup 2010-06-14T19:44:11 Ok. 2010-06-14T19:44:18 I will sort out and commit a change. 2010-06-14T19:44:31 go ahead and put it on 4.10 also. 2010-06-14T19:45:37 What is a valid perm defined to mean ? 2010-06-14T19:47:51 it is only used a couple of places and is meant to ensure that no one ever sets a flag bit for permissions that is invalid/undefined. 2010-06-14T19:47:59 grep -r it. Maybe possible to just eliminate it. 2010-06-14T19:48:09 brb 2010-06-14T20:01:53 back.. 2010-06-14T20:02:07 Just been updating head. 2010-06-14T20:02:41 still battling the first build of a new target on my home laptop which I haven't compiled anything on since before the branch was cut LOL 2010-06-14T20:04:38 The reason I coded the RFS that way was C++. In C++ everything is natually a typedef and adding _t adds nothing. It is like _f to all functions. 2010-06-14T20:05:08 yeah.. often rtems types do not have _t on them if the type name makes it clear it is a type 2010-06-14T20:05:54 I am adding _t to everything in the RFS which is everything :) 2010-06-14T20:06:40 sparc64 multilib compiles cleanly but trips new warnings for printf formats in old familiar places like the rfs 2010-06-14T20:06:50 So valid perms are any bits that are NOT RWX ? 2010-06-14T20:07:25 I think so. for user, group and world.. so 9 bits.. but that doesn't let us set a "setuid" so may be worth killing it 2010-06-14T20:08:05 Maybe we should check just those. 2010-06-14T20:08:42 Do you mean not perform the check at all ? 2010-06-14T20:09:34 That' 2010-06-14T20:10:37 I don't see a hint of EIO in the Linux man page and it is only checked in devfs and imfs. It had an assert(0) on that path in both which I am taking to mean it was paranoia code in case we screwed something up. And all of that was long before we had real file systems or shells 2010-06-14T20:13:09 I agree it should go. I am not sure the logic is correct any way. It checks for valid perms by saying any other bit is invalid which may not be correct. 2010-06-14T20:14:09 I am not sure any test like this actually add value any more. So I think agree. 2010-06-14T20:14:27 EIO isn't defined to be returned by open. So let's just kill the code. 2010-06-14T20:26:00 I have reviewed the sparc64 patch 1 and am happy. If it were solely up to me, I would commit it now. But I would like others to concur 2010-06-14T20:29:09 I have no comments on the 64bit work 2010-06-14T20:29:16 I will kill the code. 2010-06-14T20:29:31 great.. you mean the sparc64 port? 2010-06-14T20:29:36 Yes 2010-06-14T20:29:50 thanks 2010-06-14T20:30:22 The buffer code is kind of used in the shell. Do the problems appear there as well ? 2010-06-14T20:30:40 I don't know 2010-06-14T20:30:58 the mon-editor.c coverity report? 2010-06-14T20:31:03 Yes 2010-06-14T20:31:19 I took the monitor code and created the line editor in the shell from it. 2010-06-14T20:31:42 no.. that is the only issue. I propose ch=end not ch=end + 1 .. waiting on a new coverity run to finish to verify that and libi2c changes 2010-06-14T20:31:52 I also have a shot in for the objectextend one 2010-06-14T20:32:23 Oooo that is interesting 2010-06-14T20:33:28 I don't think it is possible but I can't get 100% confidence. My try is to add a "do_extend" variable and test it in that last if. It is using index and maximum in magic ways that are inconsistent between auto-extend on and off 2010-06-14T20:33:43 The code should read easier and (I hope) make coverity happy 2010-06-14T20:34:27 Yes it is tricky this bit of code. 2010-06-14T20:36:37 Indeed :-D 2010-06-14T20:37:41 In the last run (21), there are 9 issues -- 7 are "ignore" or "false" for us. 3rd party code or analysed to be correct on our side. I have outstanding fixes for the other two. 2010-06-14T20:38:02 That will complete this "rung" for us. Then we move on to a newer and more powerful scanner 2010-06-14T20:38:30 Excellent. 2010-06-14T20:48:26 * DrJoel is heading to the bedroom.. getting late here 2010-06-14T20:48:28 night 2010-06-14T20:48:35 cya 2010-06-14T22:29:28 *** zwj has joined #rtems 2010-06-15T00:21:17 *** exception13 has joined #rtems 2010-06-15T00:46:40 *** exception13 has left #rtems 2010-06-15T01:19:59 *** lcpfnvc has joined #rtems 2010-06-15T01:20:29 morning 2010-06-15T01:50:31 *** sebhub has joined #rtems 2010-06-15T01:50:39 good morning 2010-06-15T02:07:48 sebhub, hi 2010-06-15T02:10:15 hi 2010-06-15T02:10:49 sebhub, not sure we have recovered from the beating at your team's hands yet 2010-06-15T02:11:18 it was nice ;-) 2010-06-15T02:11:31 there are still two games left 2010-06-15T02:11:42 They looks sharp and fast. Our guys where out classed. 2010-06-15T02:12:10 Yes so lets hope they can get a couple of wins. They have to win 2010-06-15T02:12:43 a lot of the team is from FC Bayern, so they know each other well 2010-06-15T02:13:10 Ah ok, it shows. Our guys play all around the world. 2010-06-15T02:23:17 *** exception13 has joined #rtems 2010-06-15T02:43:24 *** cdcs has joined #rtems 2010-06-15T02:45:16 *** dr__house has joined #rtems 2010-06-15T03:09:27 *** lekernel has joined #rtems 2010-06-15T03:28:20 Hi,sebhub 2010-06-15T03:28:34 hi 2010-06-15T03:29:05 now i also add a summary of lask week irc meeting 2010-06-15T03:29:16 i have time today so we should be able to make progress today 2010-06-15T03:29:18 and now i am looking at your mail 2010-06-15T03:29:25 ok 2010-06-15T03:39:00 sebhub: about referencing, you mean that moving the sysinit struct into the creat files? but event object has no creat files 2010-06-15T03:40:27 yes, the events are a special case 2010-06-15T03:41:04 but maybe the events need not initialization function 2010-06-15T03:42:35 yeah, i have look at the event initialization function, it is just initializae a variable 2010-06-15T03:44:28 we can consider move this function into somewhere 2010-06-15T03:44:52 yes, this should be feasible 2010-06-15T03:45:58 and should we move the sysinit struct into the creat files? 2010-06-15T03:47:15 currently this is not our concern, we have to get the specification and the api right 2010-06-15T03:47:41 but eventually we may define the sysinit entry in the create modules 2010-06-15T03:51:34 ok, about the sysinit function, if we use parameter status, we also change the initialization function, and also should discuss how to deal the startup and terminate 2010-06-15T03:53:05 http://www.rtems.com/wiki/index.php/RTEMSSequencedInitialization#Initialization_and_Termination 2010-06-15T03:53:14 do you see other alternatives, what do you favor? 2010-06-15T03:56:23 :q 2010-06-15T03:57:27 i think the first is my favor 2010-06-15T03:57:57 if we use the second we have to change the initialization functions 2010-06-15T04:01:40 sorry maybe i misunderstand the second alternatives 2010-06-15T04:07:00 i think the startup and terminate parameter should be in sequence init call function not be in sysinit_handle, like startup we can use for(initcall=rtems_sysinit_begin,initcall++,initcallrtems_sysinit_handle() 2010-06-15T04:07:41 terminate, for(initcall=rtems_sysinit_stop,initcall--,initcall>rtems_sysinit_start) (*initcall)->rtems_sysinit_handle() 2010-06-15T04:09:33 if you don't use a parameter that indicates terminate or shutdown you need two functions 2010-06-15T04:09:49 and this is not very flexible 2010-06-15T04:10:38 we don't have to change the initialization functions, we only have to provide a wrapper function for this 2010-06-15T04:22:42 its snowing here in cape town 2010-06-15T04:22:47 who is your team sebhub? 2010-06-15T04:22:54 well snowing on the mountains 2010-06-15T04:23:26 kiwichris, you supporting nz or aus? 2010-06-15T04:23:33 Germany 2010-06-15T04:24:04 he supports Australia ;-) 2010-06-15T04:24:39 ah 2010-06-15T04:24:41 :) 2010-06-15T04:26:02 yeah, i see it. but the orignal initialization function parameter is void 2010-06-15T04:26:51 we not only have initialization, we have also destruction and maybe something else in the future 2010-06-15T04:27:17 with a function parameter you are much more flexible in this way 2010-06-15T04:29:41 yeah, i agree with adding parameter, but the orignal initialization function parameter is void, we should change it. do i understand right? 2010-06-15T04:30:10 there is no need to change the functions 2010-06-15T04:30:25 we can add a new one which calls the existing one 2010-06-15T04:30:42 the new one manages the parameter handling 2010-06-15T04:39:19 every initialization fuction should add a new one that manages parameter handling? i will use this update wiki 2010-06-15T04:40:16 the current manager intitializaiton functions and the new sequenced initialization api are different things 2010-06-15T04:53:17 *** lekernel has quit IRC 2010-06-15T04:56:06 *** dr__house has quit IRC 2010-06-15T05:26:51 *** dr__house has joined #rtems 2010-06-15T05:32:43 *** madrazr has joined #rtems 2010-06-15T05:35:21 *** cdcs is now known as cdcs_awy 2010-06-15T06:16:26 sebhub:hi mentor, i have update the wiki, please help me check it, thank you 2010-06-15T06:16:40 ok 2010-06-15T06:19:00 why do we need the status field in rtems_sysinit? 2010-06-15T06:23:27 i am not sure, i think it can indicate it is startup or close down 2010-06-15T06:23:54 and how will set or use this value? 2010-06-15T06:24:02 how -> who 2010-06-15T06:26:14 the sequence code which call the init call orderly 2010-06-15T06:27:05 ok, but the handler signature is: typedef void ( *rtems_sysinit_handler )( rtems_sysinit_state state ); 2010-06-15T06:27:13 that is just my old thought, now i understand what your rtems_sequeinit_state mean, i think it should be delated 2010-06-15T06:27:38 so the init procedure will use (*handler)(INIT) and the terminate (*handler)(TERM) 2010-06-15T06:27:59 yeah, i clean up it 2010-06-15T06:28:04 ok, great 2010-06-15T06:29:04 one note about name spaces, i think RTEMS_SYSINIT is not that good, we should use SYSINIT_* and _Sysinit_* because this API will become part of the super core ApI 2010-06-15T06:29:31 it is not specific to the classic RTEMS API 2010-06-15T06:30:28 yeah, i follow the name spaces of super core API, now looking at the code of super core 2010-06-15T06:31:05 I can adjust this on the wiki page, let me know then you fixed the status field 2010-06-15T06:34:27 ok, i have fixed the status field 2010-06-15T06:34:43 ok i will clean up the names 2010-06-15T06:36:34 hm, need a new password 2010-06-15T06:39:35 *** aniceberg has joined #rtems 2010-06-15T06:45:03 what password? 2010-06-15T06:45:21 for the wiki 2010-06-15T06:45:48 i donot encount this before 2010-06-15T06:46:00 now do you acess into it? 2010-06-15T06:46:39 yes, one moment i am about to edit it 2010-06-15T06:47:44 hm, now there is a world cup, i am look at it and at the same time look at code. haha 2010-06-15T07:08:42 please have a look at the changes 2010-06-15T07:10:36 ok 2010-06-15T07:13:25 the text sysinit_core still exsit, i clean up it 2010-06-15T07:13:38 ok 2010-06-15T07:14:01 i think we don't need three order domains, two should be sufficient 2010-06-15T07:14:06 group and index 2010-06-15T07:16:12 ok, i also deleted the old one 2010-06-15T07:22:33 in the maros SYSINIT_ENTRY(handler) , there is no order parameter, how ?we define its order 2010-06-15T07:22:41 one moment 2010-06-15T07:25:25 and it seems that the _Sysinit_Reference_ ## handler is not used? 2010-06-15T07:26:30 please update the page 2010-06-15T07:27:00 yes it is not used, it is only there to pull in the module containing the sysinit entry 2010-06-15T07:30:05 yeah, it is great 2010-06-15T07:31:55 the last section also should updated, example of Message_queue_manager 2010-06-15T07:32:03 you or me to update it ? 2010-06-15T07:32:43 ok please update it 2010-06-15T07:40:01 I added two iteration functions 2010-06-15T07:43:15 I added SYSINIT_ENTRY_WITH_INDEX 2010-06-15T07:44:13 *** DrJoel has quit IRC 2010-06-15T07:44:33 the next open issue is this: http://www.rtems.com/wiki/index.php/RTEMSSequencedInitialization#Error_Handling 2010-06-15T08:14:11 *** exception13 has left #rtems 2010-06-15T08:39:27 *** DrJoel has joined #rtems 2010-06-15T08:39:27 *** ChanServ sets mode: +o DrJoel 2010-06-15T08:40:01 * DrJoel waves hello 2010-06-15T08:41:06 kiwichris: any thoughts on this one? https://www.rtems.org/bugzilla/show_bug.cgi?id=1516 2010-06-15T08:41:18 * DrJoel would like to close some of the small PRs 2010-06-15T09:05:55 *** exception13 has joined #rtems 2010-06-15T09:12:51 *** dr__house has quit IRC 2010-06-15T09:19:51 *** tonylijo has joined #rtems 2010-06-15T09:20:04 *** jacob123 has joined #rtems 2010-06-15T09:20:41 *** bubaflub has joined #rtems 2010-06-15T09:42:35 *** lcpfnvc has quit IRC 2010-06-15T09:47:36 *** gedare has joined #rtems 2010-06-15T09:47:49 hi DrJoel 2010-06-15T09:48:34 good conversation about sharing code with sparc32. it is something i thought about trying but ultimately rejected 2010-06-15T09:52:26 honestly what scares me about the idea (although possible) is polluting the sparc32 code which is used in space applications. 2010-06-15T09:52:43 I think for that reason alone it should be separate 2010-06-15T09:53:15 not much value and a lot of risk in breaking things. 2010-06-15T09:53:19 but also, it really isn't compatible code. the biggest difference is 64-bit registers. 2010-06-15T09:53:32 OK.. just looked similar and that's always tempting 2010-06-15T09:53:43 yeah, well, i used the code as a template :) 2010-06-15T09:53:55 but there are 2x as many floating point registers 2010-06-15T09:54:03 I think (hope) Ralf will commit the cpukit code today and then we will move to the next patch. 2010-06-15T09:54:07 ok 2010-06-15T09:54:17 * DrJoel needs to review your gsoc work 2010-06-15T09:54:23 i made a note but want to bring to your attention, the interrupt handling is in libcpu 2010-06-15T09:54:30 which is abnormal 2010-06-15T09:54:39 we can move it back to cpukit now ... 2010-06-15T09:54:44 maybe 2010-06-15T09:54:54 it uses an ifdef based on the cpu model 2010-06-15T09:54:56 Not unprecedented though. sh, powerpc, x86, 2010-06-15T09:55:00 k 2010-06-15T09:55:27 I did it at first to try to avoid doing privileged accesses in cpukit 2010-06-15T09:55:35 and then it helped to extend to the sun4u architecture 2010-06-15T09:55:50 anyway, my vote is not to merge sparc32 and sparc64 at all 2010-06-15T09:55:55 at least, not now 2010-06-15T09:56:42 if someone in the sparc camp wants to participate in merging and testing, then it might be worthwhile -- but i don't think it will save much (if any) code 2010-06-15T09:57:11 and will just push complexity into ifdefs and out of the autoconf/automake separation of cpu models. 2010-06-15T09:57:18 ok.. let's just let it stay separate.. 2010-06-15T09:57:44 overall, it is merging a set of lightly used code into a set of highly used code and increasing the complexity of the highly used code 2010-06-15T09:58:04 yeah. 2010-06-15T10:00:19 I have to polish up a paper, and then I'm going to work on either gsoc or on instructions to run the niagara bsp 2010-06-15T10:02:06 DrJoel: I documented the work I did for the gsoc milestone, you can read on my project wiki 2010-06-15T10:02:23 http://code.google.com/p/rtems-sched/wiki/NewPriorityHandler and http://code.google.com/p/rtems-sched/wiki/NewReadyQueue 2010-06-15T10:02:35 which might be easier than code review at this point 2010-06-15T10:03:06 ok.. adding it to the queue 2010-06-15T10:03:23 at least, that gives the high level idea of what the code is 2010-06-15T10:03:36 i wrote it before teh code, so it might not be quite reality :) 2010-06-15T10:05:31 :D 2010-06-15T10:05:45 * DrJoel has to separate out his SMP prep work into submittable/mergable chunks 2010-06-15T10:06:09 yeah... I'd like to get a branch that I can play around with --- after teh gsoc midterm 2010-06-15T10:07:18 DrJoel: it would be helpful to me to get your changes to the global variables that you make percpu 2010-06-15T10:08:06 that's what I have to separate out. I now have a fairly large part and I need to separate out the SMP parts 2010-06-15T10:08:14 so yes.. i completely agree 2010-06-15T10:24:42 *** gedare_ has joined #rtems 2010-06-15T10:26:18 hey DrJoel i figured out my problem yesterday... my RTEMS_MAKEFILE_PATH was pointing to sparc-sis 2010-06-15T10:26:54 and so i compiled a pc586 with posix and things run 2010-06-15T10:27:21 *** gedare has quit IRC 2010-06-15T10:27:32 bubaflub: great.. how is it going overall? How many tests run, fail, and are skipped for now? 2010-06-15T10:27:50 *** gedare_ is now known as gedare 2010-06-15T10:28:10 DrJoel: not quite there yet 2010-06-15T10:28:15 because much of parrot is boot strapped 2010-06-15T10:28:24 i've gotten all of the native C code to compile 2010-06-15T10:28:43 bubaflub: are you compiling parrot as an RTEMS application? 2010-06-15T10:28:48 gedare: yep 2010-06-15T10:29:06 with the eventual goal of getting High Level Languages like Rakudo (Perl 6) to run on it 2010-06-15T10:29:15 DrJoel: so what i'm going to do is 2010-06-15T10:29:23 bubaflub: so Parrot is now completely compiled for RTEMS with no Parrot features like networking ignored 2010-06-15T10:29:47 DrJoel: hmmm... i know we have some Parrot core for networking, but i'll have to test to make sure 2010-06-15T10:30:06 DrJoel: my plan is to 1) build parrot natively, grab all the bootstrapped code (because the bootstrapped stuff is platform independent) 2010-06-15T10:30:11 2) clean the build and reconfigure for RTEMS 2010-06-15T10:30:16 3) build for RTEMS 2010-06-15T10:30:20 4) supply the bootstrapped stuff 2010-06-15T10:30:23 and then we should be running 2010-06-15T10:30:51 ahhh... so use the native parrot image to get the byte code output.. and use RTEMS to interpret that 2010-06-15T10:31:08 where can i get the explanation for the helloworld program 2010-06-15T10:31:35 tonylijo: what do you mean explanation? 2010-06-15T10:33:11 DrJoel: yep. so we've got "miniparrot" built 2010-06-15T10:33:19 and then a lot of the other code is bootstrapped from there 2010-06-15T10:33:36 bubaflub: do you have the steps taken so far documented anywhere? 2010-06-15T10:33:47 yeah, my repo is on github 2010-06-15T10:33:50 a colleague may be interested in the process 2010-06-15T10:33:53 and the readme should have what i've done so far 2010-06-15T10:33:59 k 2010-06-15T10:34:14 i just got everything to build last night, i'll try what i just explained today / tonight and update the readme and get a blog post 2010-06-15T10:34:17 tonylijo: not much to explain.. console driver, initialization and shutdown has to work and it prints some messages 2010-06-15T10:34:38 gedare: http://github.com/bubaflub/gsoc-rtems-parrot-hints-maker 2010-06-15T10:34:49 thank you. 2010-06-15T10:40:12 *** aniceberg has quit IRC 2010-06-15T10:43:52 *** sebhub has quit IRC 2010-06-15T10:55:43 *** jacob1231 has joined #rtems 2010-06-15T10:55:53 *** tonylijo1 has joined #rtems 2010-06-15T10:56:02 *** jacob123 has quit IRC 2010-06-15T10:56:59 *** tonylijo has quit IRC 2010-06-15T11:02:27 *** jacob1231 has quit IRC 2010-06-15T11:02:27 *** madrazr has quit IRC 2010-06-15T11:02:36 *** gedare has quit IRC 2010-06-15T11:05:09 *** jacob123 has joined #rtems 2010-06-15T11:05:31 *** madrazr has joined #rtems 2010-06-15T11:06:10 *** gedare has joined #rtems 2010-06-15T11:08:38 *** tonylijo has joined #rtems 2010-06-15T11:12:00 *** tonylijo1 has quit IRC 2010-06-15T11:34:09 *** zwj has quit IRC 2010-06-15T11:44:50 *** jacob123 has left #rtems 2010-06-15T11:59:56 *** tonylijo1 has joined #rtems 2010-06-15T12:00:59 *** tonylijo has quit IRC 2010-06-15T12:06:56 *** dr__house has joined #rtems 2010-06-15T12:15:30 *** jacob123 has joined #rtems 2010-06-15T12:28:00 *** jacob1231 has joined #rtems 2010-06-15T12:28:23 *** tonylijo has joined #rtems 2010-06-15T12:28:49 *** tonylijo1 has quit IRC 2010-06-15T12:29:10 *** jacob123 has quit IRC 2010-06-15T12:33:45 dr__house: can you help me with the C programming in rtems 2010-06-15T12:34:43 *** tonylijo has quit IRC 2010-06-15T12:35:01 *** tonylijo has joined #rtems 2010-06-15T12:37:25 dr__house: actually we didnt understand the #define part of the hello world program 2010-06-15T12:37:52 jacob1231: give me 10 mins, please. In the middle of something really messy :P 2010-06-15T12:38:08 dr__house: ohk 2010-06-15T12:41:40 jacob1231: the ones with CONFIGURE_ in front of them 2010-06-15T12:47:45 *** jacob123 has joined #rtems 2010-06-15T12:48:14 CONFIGURE_MAXIMUM_TASKS has 1 2010-06-15T12:48:16 *** tonylijo1 has joined #rtems 2010-06-15T12:48:32 *** tonylijo has quit IRC 2010-06-15T12:48:37 *** jacob1231 has quit IRC 2010-06-15T12:53:10 *** gedare has quit IRC 2010-06-15T13:08:58 *** jacob123 has quit IRC 2010-06-15T13:09:00 *** tonylijo1 has quit IRC 2010-06-15T13:11:55 *** tonylijo has joined #rtems 2010-06-15T13:24:34 *** jacob123 has joined #rtems 2010-06-15T13:42:13 jacob123: rtems targets embedded real-time systems. Most of these are designed with hard limits. The configure options specify how many of an OS resource your application needs. Remember you are linking the OS with your applicaiton to form a single executable 2010-06-15T13:47:57 DrJoel: are those values user defined 2010-06-15T13:48:11 *** lekernel has joined #rtems 2010-06-15T13:48:31 yes. It is strictly a matter of what you need. 2010-06-15T13:48:55 http://rtems.org/onlinedocs/doc-current/share/rtems/html/c_user/c_user00414.html 2010-06-15T13:49:08 DrJoel: but some are left undefined what does that mean 2010-06-15T13:52:10 most are left undefined by an application. it means they default to benign, usually low values (e.g. 0). 2010-06-15T14:17:52 *** jacob123 has quit IRC 2010-06-15T14:19:28 *** jacob123 has joined #rtems 2010-06-15T14:22:59 *** madrazr has left #rtems 2010-06-15T14:34:32 *** jacob123 has left #rtems 2010-06-15T15:00:37 *** tonylijo has quit IRC 2010-06-15T15:06:06 *** dr__house has quit IRC 2010-06-15T15:08:06 *** tonylijo has joined #rtems 2010-06-15T15:08:24 what is the location of the include directory in rtems 2010-06-15T15:08:55 i mean where the header files are located 2010-06-15T15:12:12 they are scattered around the tree before "preinstall" but are collected into the build tree (and install tree) and at a single point. 2010-06-15T15:12:26 Until you are comfortable with the org, I recommend "find rtems-XXX -name XXX.h" 2010-06-15T15:19:51 *** tonylijo1 has joined #rtems 2010-06-15T15:20:00 *** tonylijo has quit IRC 2010-06-15T15:27:40 *** tonylijo1 has left #rtems 2010-06-15T15:40:43 *** exception13 has left #rtems 2010-06-15T16:13:33 *** bubaflub has quit IRC 2010-06-15T16:21:36 Hi, I have some includes in "pc586/lib/include/rtems/bsd/sys/" that aren't from libbsdport, why are they there? 2010-06-15T16:21:48 *** cdcs_awy is now known as cdcs 2010-06-15T16:22:30 what are they? you can do a find from the top of your source tree for the name ... "find rtems-XXX -name XXX.h" 2010-06-15T16:22:45 *** mw|mobile has joined #rtems 2010-06-15T16:23:08 hi 2010-06-15T16:23:08 they are bsd headers, but they seem "lost" 2010-06-15T16:24:44 did they come from rtems proper? 2010-06-15T16:24:44 you can always remove the pc586/lib directory and reinstall RTEMS and libbsdport to make sure where they came from. 2010-06-15T16:28:10 rtems proper?. I know they aren't from libbsdport because i haven't found them on my libbsdport build directory 2010-06-15T16:30:45 *** mw|mobile has quit IRC 2010-06-15T16:36:12 rtems--XXX/cpukit 2010-06-15T16:37:44 "rtems/cpukit/include/rtems/bsd/sys/queue.h" 2010-06-15T16:41:09 then that is where it came from 2010-06-15T16:52:22 *** DrJoel has quit IRC 2010-06-15T16:57:21 *** bubaflub has joined #rtems 2010-06-15T16:59:18 *** alexp0205 has joined #rtems 2010-06-15T17:12:22 *** cdcs_ has joined #rtems 2010-06-15T17:12:34 *** cdcs has quit IRC 2010-06-15T17:13:55 i need some help getting qemu up and running on my system 2010-06-15T17:13:59 anyone care to hold my hand for a bit? 2010-06-15T17:18:13 *** cdcs_ has quit IRC 2010-06-15T17:26:12 kiwichris: if you're around i need a bit of help with getting qemu. i've got a parrot binary built and want to test it out on qemu and need some help loading it up 2010-06-15T17:27:26 bubaflub, I am just about and about to leave for the office so should back in 30min. 2010-06-15T17:27:38 kiwichris: ok, no rush. i'm going to keep playing with this for a bit 2010-06-15T17:27:57 for qemu I use grub2 2010-06-15T17:28:04 bb soon 2010-06-15T17:28:09 *** kiwichris has quit IRC 2010-06-15T17:28:12 ok 2010-06-15T17:58:06 *** kiwichris has joined #rtems 2010-06-15T18:00:45 *** alexp0205 has left #rtems 2010-06-15T18:05:13 kiwichris: so i've got a parrot binary and want to run it on qemu 2010-06-15T18:05:57 So this is the i386 arch ? 2010-06-15T18:09:45 *** bubaflub has quit IRC 2010-06-15T18:14:59 *** bubaflub has joined #rtems 2010-06-15T18:15:23 bubaflub, is the arch i386 or sparc ? 2010-06-15T18:15:45 i386 2010-06-15T18:15:53 Ok 2010-06-15T18:17:20 I will email my qemu script. Hang on. 2010-06-15T18:26:54 bubaflub, just sent 2010-06-15T18:27:41 nice, thanks 2010-06-15T18:30:56 so i see the rtems-boot.img 2010-06-15T18:31:00 but what is the hda.img? 2010-06-15T18:31:12 i ran a set of commands from the wiki to create a hda.img but i'm not sure exactly what goes in there 2010-06-15T18:31:21 hdb is just set to the current directory, yes? 2010-06-15T18:31:26 That was an image to have on a second disk to test file system code. 2010-06-15T18:31:30 Yes 2010-06-15T18:31:50 Qemu will make a DOSFS disk of this directory and grub can read it. 2010-06-15T18:32:26 ah, ok 2010-06-15T18:32:46 Grub is built to look in "C:" in DOS terms for rtems-grub.cfg. 2010-06-15T18:32:55 okey dokey 2010-06-15T18:33:04 so qemu boots 2010-06-15T18:33:08 and gives me a start up menu 2010-06-15T18:33:19 and then flashes and just says "welcome to grub" 2010-06-15T18:33:27 If you do not have a default set in the rtems.grub.cfg file. 2010-06-15T18:34:03 How many files in the dir with the executable and rtems-grub.cfg ? 2010-06-15T18:34:18 30+ 2010-06-15T18:34:26 but i really only want the "parrot" executable 2010-06-15T18:34:40 Hmm, there is a limit to the size of the MSDOS disk qemu makes. 2010-06-15T18:34:48 ok, it looks like it loaded ok 2010-06-15T18:35:02 What loaded ok ? 2010-06-15T18:35:10 ah, the grub boot menu 2010-06-15T18:35:28 hmmm 2010-06-15T18:35:34 it's not giving me the menu options 2010-06-15T18:35:37 one sec, lemme check my config 2010-06-15T18:37:35 hmmm, that's strange 2010-06-15T18:37:45 one is i'm getting a command line warning that it can't open /dev/kqemu/ for the QEMU acceleration layer 2010-06-15T18:37:53 (which i assume doesn't hamper what i'm trying to do) 2010-06-15T18:37:58 Do not worry about that. It is nothing. 2010-06-15T18:38:18 Can you get a grub prompt ? 2010-06-15T18:38:49 hmmm 2010-06-15T18:38:59 i just see a "Welcome to Grub!" 2010-06-15T18:39:16 Is this in the graphic screen ? 2010-06-15T18:39:18 it doesn't accept any keyboard input or anything 2010-06-15T18:39:21 yes, graphic screen 2010-06-15T18:39:43 Can you rename the rtems-grub.cfg to something else and try again 2010-06-15T18:40:15 same thing 2010-06-15T18:40:26 so it's not finding the .cfg file? 2010-06-15T18:40:37 Can you move out of that dir and try again 2010-06-15T18:40:45 It should give a prompt 2010-06-15T18:41:09 nope, still nothing 2010-06-15T18:41:22 This means either qemu or the grub image. 2010-06-15T18:41:44 Joel has one but I do not know where it is. 2010-06-15T18:41:52 The wiki may have something. 2010-06-15T18:43:59 hmmm 2010-06-15T18:44:11 i'm using the Virtual Box image that DrJoel provided, so I wonder if he'll have any insight 2010-06-15T18:46:01 He may. He will have run qemu more than me lately. 2010-06-15T18:46:34 http://www.rtems.org/ftp/pub/rtems/people/joel/qemu/ 2010-06-15T18:46:46 Look in here. The pc386_fd maybe a suitable image 2010-06-15T18:47:50 ok, so i'll work on getting that 2010-06-15T18:48:09 and i'll talk to DrJoel about getting qemu running to test this stuff 2010-06-15T18:48:24 another option is that on the wiki there is a setup to use QEMU on a separate Virtual Box instance 2010-06-15T18:48:33 All I can suggest is you get a grub boot prompt and then you can at least see the disk and then maybe load the executable. 2010-06-15T18:48:49 and then use TFTP to load the program 2010-06-15T18:48:51 I do not use VirtualBox so I do not know. 2010-06-15T18:49:00 yeah, i can continue working without qemu, it would just be nice to test 2010-06-15T18:49:05 That could work. 2010-06-15T18:49:31 The sparc simulator in gdb is also a good choice. Joel knows this one well and it will work. 2010-06-15T18:49:32 so i'll keep forging ahead for now, and hit up DrJoel later 2010-06-15T18:49:37 okey dokey 2010-06-15T18:49:39 Yeah 2010-06-15T18:49:47 i do have sparc-sis on my machine 2010-06-15T18:49:51 Well done on getting to this point. It is impressive 2010-06-15T18:50:06 thanks 2010-06-15T18:50:09 in RTEMS yes, tools you will need to check. 2010-06-15T18:50:20 Hit sparc- a few times. 2010-06-15T18:50:35 yep, got sparc-rtems4.10-* 2010-06-15T18:50:43 it'd be nice to load it up in sparc-rtems4.10-gdb 2010-06-15T18:50:49 and load, tar sim, and run 2010-06-15T18:51:03 Sparc is great to test on because it break fragile alignment issues in code :) 2010-06-15T18:51:11 can I configure sparc with --enable-posix? 2010-06-15T18:51:15 Yes 2010-06-15T18:51:25 ok, cool 2010-06-15T18:51:38 Joel has scripts in the CVS 'rtems-testing' module. 2010-06-15T18:51:41 i think there is some code that checks for threads or pthreads, so i wanted ot make sure we can get that 2010-06-15T18:51:44 ok, cool 2010-06-15T18:51:48 Check this out and then hit make at the top. 2010-06-15T18:52:16 You should get a script to help run a test in a bin dir 2010-06-15T18:52:40 you mean a test in parrot land? 2010-06-15T18:52:57 No an RTEMS executable. 2010-06-15T18:53:06 ok, phew 2010-06-15T18:53:31 That will be you soon :) 2010-06-15T18:56:30 i get this feeling that grub isn't loading rtems-grub.cfg 2010-06-15T18:57:10 It should drop to a prompt so you can then look around. This is standard grub behaviour 2010-06-15T18:57:27 Removing that file also should do the same but this is even more certain. 2010-06-15T18:58:38 hmmm... i've got qemu 0.9.1, so that should be ok 2010-06-15T18:59:20 That is what I have. 2010-06-15T19:00:26 ok 2010-06-15T19:03:48 arg, this isn't working 2010-06-15T19:03:55 well, here's the progress report: 2010-06-15T19:04:12 the configuration throws no errors or warnings, the building just throws a few warnings 2010-06-15T19:04:26 and it builds on it's own up to miniparrot (which is where things start bootstrapping) 2010-06-15T19:05:04 if i copy one file (which i'll automate here in a bit) then it'll actually compile and link the "parrot" executable 2010-06-15T19:06:11 there appear to be some problems with some of the linking flags, but i should be able to write a script that'll configure parrot and do the copying of the bootstrapping stuff 2010-06-15T19:06:21 Isn't a miniparrot or baby parrot a chick ? ;) 2010-06-15T19:06:28 i'll write a blog post tonight and shoot you and dukeleto and email explaining this in detail 2010-06-15T19:06:35 indeed. 2010-06-15T19:06:42 oh man, grub finally booted 2010-06-15T19:07:04 Sound great. Sorry I cannot help you with the qemu side of things. 2010-06-15T19:07:08 nah, that's ok 2010-06-15T19:07:09 I know Joel can. 2010-06-15T19:07:12 i think it just took forever to load it 2010-06-15T19:07:19 ok, i'll keep playing with this and shoot ya an email in a bit 2010-06-15T19:07:20 Really. 2010-06-15T19:07:48 yeah, it just took forever to load the boot screen 2010-06-15T19:08:20 i'm in the grub command prompt now 2010-06-15T19:08:29 Excellent. 2010-06-15T19:08:35 What does ls show 2010-06-15T19:08:48 ls show s that i've got fd0, hd0, hd1, hd1,1 and cd0 2010-06-15T19:09:01 Which boot image ? 2010-06-15T19:09:29 rtems-boot.img 2010-06-15T19:09:33 Great. 2010-06-15T19:09:45 What does "ls (hd0,1)/ 2010-06-15T19:09:51 What does "ls (hd0,1)/" show ? 2010-06-15T19:10:04 uunkown device 2010-06-15T19:10:13 What about "ls (hd0)/ 2010-06-15T19:10:52 unknown filesystem 2010-06-15T19:11:06 ls (hd1,1)/ 2010-06-15T19:12:03 ah i'm loaded in the wrong directory 2010-06-15T19:14:13 I suspect hd0 is wrong and needs to be the correct one. My script may be broken I sent you. I have was testing file systems with it. 2010-06-15T19:14:21 Just check the qemu options. 2010-06-15T19:14:36 We should be able to "ls (hd0,1)/" 2010-06-15T19:14:49 and see the rtems-grub.cfg file. 2010-06-15T19:18:09 when i ls (hd1,1)/ isee the rtems-grub.cfg 2010-06-15T19:21:55 Sure but grub is built to look in hd0 2010-06-15T19:22:54 My mistake. Open up qemu.sh and delete the -hda file and change -hdb to -hda 2010-06-15T19:23:08 -hda line 2010-06-15T19:27:40 brb 2010-06-15T19:29:40 be back in a bit, i have to drive my roommate to the train station 2010-06-15T19:54:25 now i get: Booting 'Parrot' error: segment doesn't fit in memory reserved for the OS (0x0 < 0x10000) 2010-06-15T19:58:24 This is excellent. 2010-06-15T19:58:33 We can now debug it. 2010-06-15T19:59:05 You need to start qemu with -s 2010-06-15T19:59:07 hah. 2010-06-15T19:59:09 okey dokey 2010-06-15T19:59:15 Hang on and also .... 2010-06-15T19:59:59 and -S 2010-06-15T20:00:09 this stops the CPU at reset. 2010-06-15T20:00:19 You then connect with gdb 2010-06-15T20:00:54 Which file are you loading ? 2010-06-15T20:01:08 i'm trying to load the main parrot executable 2010-06-15T20:01:08 *.exe or *.ralf ? 2010-06-15T20:01:25 hmm, for some reason it doesn't have either .exe or .ralf... 2010-06-15T20:01:29 (don't know why i didn't notice that until now) 2010-06-15T20:01:59 You want the actual ELF file generated by the linker I think. Hang on and I will check 2010-06-15T20:02:26 Yeap the one from the linker 2010-06-15T20:02:52 i'm not sure when / where those files are generated in the Makefile process 2010-06-15T20:03:02 Yeap the one from the linker. 2010-06-15T20:03:18 The other files are generated from the post linker BSP hook. 2010-06-15T20:03:33 Actually the .exe file is wrong and we should not be generating it. 2010-06-15T20:03:59 We should generate files that are the linker default and for RTEMS that is the same as Linux etc 2010-06-15T20:04:35 ok, for example 2010-06-15T20:04:49 i'm only seeing "parrot" in the build directory 2010-06-15T20:05:04 That is fine 2010-06-15T20:05:27 [chris@kaka gdb]$ file /local/ccj/rtems/apps/rfs/pc386-static/filesystem-test 2010-06-15T20:05:36 ok, it's going to take a little bit to fire up qemu 2010-06-15T20:05:57 when i do `file parrot` i see 2010-06-15T20:05:57 filesystem-test: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), statically linked, not stripped 2010-06-15T20:06:14 parrot: ELF 32-bit LSB executable, Intel 80386, version 1(SYSV), statically linked, not stripped 2010-06-15T20:06:16 OH MAN 2010-06-15T20:06:18 jinx? 2010-06-15T20:06:29 Great this all looks ok 2010-06-15T20:07:07 ok 2010-06-15T20:07:14 sounds like it's clsoe 2010-06-15T20:07:16 close, even 2010-06-15T20:07:24 so i think i was confused before about Grub 2010-06-15T20:07:32 it just takes Grub a really long time to load the menu / prompt 2010-06-15T20:07:39 (for whatever reason, perhaps because it's in a huge directory) 2010-06-15T20:08:21 You can debug from reset and therefore early problems in RTEMS or you can set a break point on main and run till them 2010-06-15T20:08:59 It maybe reading the whole tree of files. 2010-06-15T20:09:11 If you limit the amount it can see it may help 2010-06-15T20:09:39 yeah 2010-06-15T20:09:49 i can probably move the executable out of that directory if that'll make things easier 2010-06-15T20:10:00 ok, so how can i attach GDB? 2010-06-15T20:11:10 I have just sent the script in an email 2010-06-15T20:11:33 Then it is just "i386-rtems4.10-gdb parrot" 2010-06-15T20:12:00 If you stop qemu at reset you will see the PC at the reset location, something like 0x0000ffff 2010-06-15T20:12:19 You can then do "b main" or "b start" then "c" 2010-06-15T20:16:34 ah, it looks like the garbage collector is trying to allocate some memory 2010-06-15T20:16:45 and then it throws a PANIC_OUT_OF_MEM 2010-06-15T20:16:53 Ok. 2010-06-15T20:17:05 How much memory does it want ? 2010-06-15T20:17:13 it looks like size=35514 2010-06-15T20:17:17 i'm guessing that's bytes 2010-06-15T20:17:17 Looks like you are up and running and debugging. 2010-06-15T20:17:20 That should work. 2010-06-15T20:17:39 The PC BSP should manage the whole thing unless something is corrupt. 2010-06-15T20:17:45 ok 2010-06-15T20:17:54 How much stack is given to the init task that runs main ? 2010-06-15T20:18:38 i'm not sure 2010-06-15T20:18:46 is that something i set at compile time? 2010-06-15T20:18:56 The qemu script gives the PC 128M of RAM so that is not a problem. 2010-06-15T20:19:03 right 2010-06-15T20:19:36 Please check the init task stack size. Do you have a init.c file that parrot is linked to ? 2010-06-15T20:21:25 ah... hrm. no. that's probably why it's blowing up 2010-06-15T20:21:32 i simply compiled parrot "as is" 2010-06-15T20:21:36 Yes I suspect so 2010-06-15T20:21:44 This one is tricky. 2010-06-15T20:21:47 doh. 2010-06-15T20:22:11 How do we add a single RTEMS specific file so parrot as is, is built to run the tests. 2010-06-15T20:22:28 The init.c file is normally a user app thing. 2010-06-15T20:22:45 I see parrot for RTEMS users as a library not an executable. 2010-06-15T20:23:06 ok, right 2010-06-15T20:23:09 and i have libparrot built 2010-06-15T20:23:39 Excellent so that issue is solved. It is just the parrot exe for RTEMS testing. 2010-06-15T20:24:05 We need a init.c type file that has the CONFIGURE_* defined that pulls in confdefs.h 2010-06-15T20:25:43 ok, are there any examples in the examples_v2 directory that has stuff like that? 2010-06-15T20:26:07 Every test in testsuite has this type of file. 2010-06-15T20:26:32 ok 2010-06-15T20:28:22 i'm looking through the gcc-testing stuff now 2010-06-15T20:28:31 Ok. 2010-06-15T20:28:41 We renamed it not long ago to rtems-testing :) 2010-06-15T20:28:57 ah, i remember that 2010-06-15T20:29:10 i must have an old checkout 2010-06-15T20:29:19 You may have 2010-06-15T20:29:40 My parrot package should have one that works 2010-06-15T20:30:24 ah, yes i see it 2010-06-15T20:30:36 Great 2010-06-15T20:31:50 This one let me run parrot programs. The network can be removed if you want. 2010-06-15T20:31:58 Just to make it simpler. 2010-06-15T20:32:01 ah, i'll remove it only if it causes problems 2010-06-15T20:32:02 yeah 2010-06-15T20:32:36 so i need to build init.c and link it to the parrot program, right? 2010-06-15T20:33:52 Yes. There are defaults in RTEMS and tricks to get it linked but they results are often not functional. It attempts to almost nothing for those targets that are small. 2010-06-15T20:41:00 ok, so here's the master plan: 2010-06-15T20:41:28 first we need a local build of parrot so it can bootstrap of the extra files 2010-06-15T20:41:34 then we need to snag those extra files 2010-06-15T20:41:45 then we need to configure parrot for RTEMS 2010-06-15T20:41:59 move the snagged files into place 2010-06-15T20:42:09 patch the Makefile to make sure we are building and linking against init.c 2010-06-15T20:42:18 and then build parrot for RTEMS 2010-06-15T20:47:58 Sorry got distracted 2010-06-15T20:48:18 so that's my plan (which i'll document a bit more in the blog post) 2010-06-15T20:48:21 Sounds ok. 2010-06-15T20:48:25 and i'm going to try and hack this init.c in tonight 2010-06-15T20:48:33 Cool. Sorry close. 2010-06-15T20:48:38 So close. 2010-06-15T20:48:38 i'll shoot both you and dukeleto an email before i head to bed 2010-06-15T20:48:40 not sorry 2010-06-15T20:48:45 yeah, i can almost taste it 2010-06-15T20:48:51 i just want to run hello_world.pir on qemu 2010-06-15T20:48:59 well, "just" means just this week 2010-06-15T20:49:00 Can you also cc Joel. His timezone is better than mine :) 2010-06-15T20:49:04 yeah, sure thing 2010-06-15T20:50:26 *** gedare has joined #rtems 2010-06-15T21:01:13 *** zwj has joined #rtems 2010-06-15T21:03:35 kiwichris: ok, i think i've built parrot with your init.c ... i'm testing with QEMU / gdb 2010-06-15T21:03:48 Nice 2010-06-15T21:06:45 hmmm... not sure that worked 2010-06-15T21:07:47 bummer 2010-06-15T21:13:31 *** gedare has quit IRC 2010-06-15T23:11:20 *** bubaflub has quit IRC 2010-06-15T23:49:53 *** madrazr has joined #rtems 2010-06-15T23:55:05 *** kiwichris has quit IRC 2010-06-16T00:33:18 *** kiwichris has joined #rtems 2010-06-16T01:00:39 *** dr__house has joined #rtems 2010-06-16T02:11:48 *** sebhub has joined #rtems 2010-06-16T02:12:23 good morning 2010-06-16T03:02:36 *** exception13 has joined #rtems 2010-06-16T03:04:29 sebhub, hi 2010-06-16T03:04:33 hi 2010-06-16T03:05:01 The RFS changes are on 4.10. I think this blocks the release. 2010-06-16T03:05:28 I suspect 64bit related aspects of the RFS may be broken. 2010-06-16T03:05:54 I cannot tell as view CVS is broken on the 4.10 branch. 2010-06-16T03:09:09 maybe rfs has problems with 16 or 64 bit targets, but is this really a blocker? 2010-06-16T03:09:35 on the other hand i think the change from ralf is not correct 2010-06-16T03:09:49 Well yes because we have broken 32bit targets if there was a problem at all. 2010-06-16T03:09:51 the file system size may exceed 4GB? 2010-06-16T03:10:07 How can a target be standard if it cannot support uint64_t ? 2010-06-16T03:10:18 Oh yes. Files can as well. 2010-06-16T03:10:33 And all this was carefully tested and checked by me. 2010-06-16T03:10:49 I am now concerned changes like this may have gone in. 2010-06-16T03:11:00 i guess uint64_t is supported by gcc on any target 2010-06-16T03:11:06 And given CVS I cannot isolate a specific check in and see what happened. 2010-06-16T03:11:27 If it is standard it must or uint64_t means nothing. 2010-06-16T03:11:44 All this is a lot of work 2010-06-16T03:12:31 is something else broken now except the patch that i respondet too? 2010-06-16T03:12:55 We need to generate an intermediate value in the maths that is 64bit so we do not round out at 32bits 2010-06-16T03:13:28 If it is then rounded down to 32bit when assigned that is a bug. The macro was correct. 2010-06-16T03:13:47 yes, that is what i thought too, thus the mail 2010-06-16T03:14:14 Well spotted. I did not see this, only the printf stuff. 2010-06-16T03:15:36 the printf stuff seems ok, and it cannot break that much 2010-06-16T03:16:01 Yes I agree. 2010-06-16T03:29:57 I have replied. 2010-06-16T03:33:48 ok 2010-06-16T03:33:49 *** zwj has quit IRC 2010-06-16T03:34:33 hi 2010-06-16T03:34:53 *** zwj has joined #rtems 2010-06-16T03:37:24 hi 2010-06-16T04:29:05 *** nileena has joined #rtems 2010-06-16T04:29:16 *** nileena has left #rtems 2010-06-16T05:04:35 *** madrazr has quit IRC 2010-06-16T05:12:41 *** Administrator has joined #rtems 2010-06-16T05:13:00 *** Administrator is now known as Guest86950 2010-06-16T05:15:38 *** backhappy has joined #rtems 2010-06-16T05:17:31 hello,I may found an error in the RTEMS C USER'S GUIDE! 2010-06-16T05:18:17 In the 23.2.2 Basic System Information --> 2010-06-16T05:18:17 CONFIGURE_MESSAGE_BUFFER_MEMORY 2010-06-16T05:19:01 #define CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE (CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE( \ 2010-06-16T05:19:01 24, sizeof(one_message_type) + \ 2010-06-16T05:19:01 CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE( \ 2010-06-16T05:19:01 500, sizeof(other_message_type) \ 2010-06-16T05:19:01 ) 2010-06-16T05:19:47 Why not is .....#define CONFIGURE_MESSAGE_BUFFER_MEMORY(.....) 2010-06-16T05:20:40 can you please write a PR to the rtems bugzilla about this 2010-06-16T05:41:14 what's the PR? 2010-06-16T05:41:31 problem report 2010-06-16T05:43:14 oo! ok! 2010-06-16T05:43:32 thanks 2010-06-16T05:45:22 sebhub, why is arg 2 of type size_t ? 2010-06-16T05:53:26 i think this will only know the GCC experts 2010-06-16T05:53:59 Is this an invalid warning ? 2010-06-16T05:54:08 sizeof(size_t)==8 and it is unsigned so the integer conversion rank == uint64_t according to the standard 2010-06-16T05:54:23 this is a false positive warning as far as I think 2010-06-16T05:55:51 Is this on sparc64 ? 2010-06-16T05:55:59 i suppose so 2010-06-16T05:56:19 Why would the AVR break ? 2010-06-16T05:57:10 here you have sizeof(size_t) == 2? but this is a different issue 2010-06-16T05:57:17 the macro would still be ok 2010-06-16T05:57:52 So would an extra cast after the maths make any difference ? 2010-06-16T05:58:16 or cast both parts to be uint64_t 2010-06-16T05:58:54 I suspect adding the size as a uint64_t to _fs would be simplest. 2010-06-16T05:59:06 the only problem I see here is that this macro disturbs the GCC warning machinery 2010-06-16T05:59:28 Yes and all for a print in a format command :( 2010-06-16T06:00:12 with respect to the "usual integer propagation rules" i don't see a problem here 2010-06-16T06:00:53 Ok. I would rather remove the issue and avoid the whole thing. 2010-06-16T06:01:24 a static inline function should make it clear 2010-06-16T06:01:58 I need to check any performance issues. If used lots I will change to a variable in the file system struct. 2010-06-16T06:02:37 If it is just prints etc I will make it a real function and move it into a C file. 2010-06-16T06:03:25 this macro is used to evaluate the super block 2010-06-16T06:03:44 I think a real C function then. 2010-06-16T06:04:00 *** cdcs has joined #rtems 2010-06-16T06:04:07 *** zwj has quit IRC 2010-06-16T06:04:17 That is for another day. I am going to head off. 2010-06-16T06:04:26 cu 2010-06-16T06:04:32 And again thanks for picking it up. 2010-06-16T06:05:02 *** zwj has joined #rtems 2010-06-16T06:08:43 sebhub:hi 2010-06-16T06:08:54 *** kiwichris has quit IRC 2010-06-16T06:09:49 *** backhappy has quit IRC 2010-06-16T06:10:23 *** backhappy has joined #rtems 2010-06-16T06:30:15 *** zwj has quit IRC 2010-06-16T06:56:13 *** backhappy has quit IRC 2010-06-16T07:13:00 *** DrJoel has joined #rtems 2010-06-16T07:13:00 *** ChanServ sets mode: +o DrJoel 2010-06-16T07:23:05 *** DrJoel has quit IRC 2010-06-16T07:24:54 *** DrJoel has joined #rtems 2010-06-16T07:24:54 *** ChanServ sets mode: +o DrJoel 2010-06-16T07:24:59 *** DrJoel has quit IRC 2010-06-16T07:49:10 *** exception13 has left #rtems 2010-06-16T07:54:57 *** zwj has joined #rtems 2010-06-16T07:55:28 *** zwj has left #rtems 2010-06-16T08:10:07 *** DrJoel has joined #rtems 2010-06-16T08:10:08 *** ChanServ sets mode: +o DrJoel 2010-06-16T08:13:41 *** gedare has joined #rtems 2010-06-16T08:29:33 *** zwj has joined #rtems 2010-06-16T08:30:02 hi,sebhub 2010-06-16T08:30:38 hi 2010-06-16T08:31:08 now we can post the status of sequence init to the maillist 2010-06-16T08:31:38 did you update something in the meantime? 2010-06-16T08:32:33 one open point is this: http://www.rtems.com/wiki/index.php/RTEMSSequencedInitialization#Error_Handling 2010-06-16T08:33:32 not yet, the errorhanding should in sequence code of the init code? 2010-06-16T08:33:59 this is the big question 2010-06-16T08:35:06 we don't have to answer it now, but we should point this out 2010-06-16T08:35:09 if the errorhanding is in sequence code we should make a common handle to deal with it, it is a really question 2010-06-16T08:35:30 we already have a fatal error handler 2010-06-16T08:36:03 in rtems ? 2010-06-16T08:36:05 but, this fatal error depends on some initialization steps 2010-06-16T08:36:07 yes 2010-06-16T08:36:46 where's the error handler in the rtems code? 2010-06-16T08:36:50 rtems_fatal_error_occured() 2010-06-16T08:37:59 if it encounts the error it will be handlered the same way using this function? 2010-06-16T08:39:34 sry i don't understand the question 2010-06-16T08:40:53 i mean that in the whatever fatal reason it will alway goto rtems_fatal_error_occured? 2010-06-16T08:41:23 currently yes 2010-06-16T08:43:00 if there is some fatal reason before the rtems_fatal_error_occured initialization, how we handler it ? 2010-06-16T08:43:22 there isn't much to the initialization of that so it is ok. Your code is well after that 2010-06-16T08:45:30 the only critical point here is the evaluation of fatal error handlers provided by the configuration 2010-06-16T08:45:51 *** madrazr has joined #rtems 2010-06-16T08:46:10 joel: what do you think about this; http://www.rtems.com/wiki/index.php/RTEMSSequencedInitialization#Error_Handling 2010-06-16T08:46:15 ok, then we can use this function to handler the fatal error, but i think the error handler of the sequence code should only response the error about the sequence code 2010-06-16T08:48:50 right now, an error during rtems initialization is fatal. It indicates something horrible is wrong with the configuration. The routines you are calling are designed to invoke the fatal error handler themselves when something goes wrong. The entire set of these errors is in an enum. 2010-06-16T08:49:27 so .. no need to check the return status (or even have it) for RTEMS purposes 2010-06-16T08:49:27 for RTEMS managers 2010-06-16T08:50:09 I suppose though we want this to grow to later initialization points and possibly? user code? 2010-06-16T08:50:30 yes 2010-06-16T08:50:55 we may initialize bdbuf for example via this mechanism or drivers 2010-06-16T08:52:02 anything that needs no tasks 2010-06-16T08:53:52 would all failures of this code be considered fatal? If so, then we could leave them void (like global ctors) and not add return codes to all the manager adn score initializers for nothing 2010-06-16T08:55:07 we can specifiy that all errors during the sysinit invokation are fatal, i think at the moment this is reasonable 2010-06-16T08:56:28 * gedare wants to configure whether an initialiation error is fatal or not 2010-06-16T08:56:36 ;) 2010-06-16T08:57:35 why? 2010-06-16T08:58:10 this makes it more complicated and is something that the current manager initialization doesn't have 2010-06-16T08:58:47 I'm not (too) serious. Although I could see it as being useful in the future. 2010-06-16T08:59:20 But the complexity would probably be too limiting 2010-06-16T09:00:40 errors during initialization are software or configuration bugs 2010-06-16T09:00:56 such applications are not worth to live long 2010-06-16T09:01:40 initialization fatal errors are reserved for things that are truly horrible and reflect an inconsistent system configuration that you shouldn't be allowed to run the first task with 2010-06-16T09:02:06 Lab failures. This is not a desktop OS. If you screwed up, catch it early before you blow up a rocket or kill someone 2010-06-16T09:02:17 * DrJoel is heading home to take care of a sick wife. may be back on later 2010-06-16T09:02:21 *** DrJoel has quit IRC 2010-06-16T09:03:42 zwj: can you please update the wiki page with the results of this discussion, please don't delete anything, instead explain if something was refused 2010-06-16T09:04:06 *** gedare_ has joined #rtems 2010-06-16T09:04:37 grrrr... my fedora (well, X) keeps lockign up :/ 2010-06-16T09:05:45 ok, i will update in a moment 2010-06-16T09:06:01 now have a dinner 2010-06-16T09:06:38 *** gedare has quit IRC 2010-06-16T09:06:59 *** gedare_ is now known as gedare 2010-06-16T09:07:22 have a nice meal 2010-06-16T09:29:57 *** bubaflub has joined #rtems 2010-06-16T09:39:42 ls 2010-06-16T09:39:47 oops, wrong window 2010-06-16T09:48:33 *** tonylijo has joined #rtems 2010-06-16T10:15:12 *** cdcs has quit IRC 2010-06-16T10:23:42 *** cdcs has joined #rtems 2010-06-16T10:47:11 *** sebhub has quit IRC 2010-06-16T10:56:38 *** madrazr has quit IRC 2010-06-16T11:05:08 *** tonylijo1 has joined #rtems 2010-06-16T11:06:53 *** tonylijo has quit IRC 2010-06-16T11:29:31 *** zwj has quit IRC 2010-06-16T11:33:19 *** tonylijo has joined #rtems 2010-06-16T11:33:24 *** tonylijo has left #rtems 2010-06-16T11:35:25 *** tonylijo1 has quit IRC 2010-06-16T11:54:56 *** exception13 has joined #rtems 2010-06-16T12:32:08 *** madrazr has joined #rtems 2010-06-16T12:47:09 *** n3oo3n has quit IRC 2010-06-16T12:47:09 *** aanjhan has quit IRC 2010-06-16T12:50:14 *** madrazr has quit IRC 2010-06-16T12:50:42 *** DrJoel has joined #rtems 2010-06-16T12:50:42 *** ChanServ sets mode: +o DrJoel 2010-06-16T13:09:38 *** madrazr has joined #rtems 2010-06-16T13:35:54 *** DrJoel has quit IRC 2010-06-16T13:36:08 *** DrJoel has joined #rtems 2010-06-16T13:36:08 *** ChanServ sets mode: +o DrJoel 2010-06-16T13:36:37 *** madrazr has quit IRC 2010-06-16T14:06:41 bubaflub: does your utime PR have the correct patch? 2010-06-16T14:06:58 DrJoel: ? 2010-06-16T14:14:05 *** gedare has quit IRC 2010-06-16T14:14:25 bubaflub: https://www.rtems.org/bugzilla/show_bug.cgi?id=1529 2010-06-16T14:14:29 sorry in bug commit mode 2010-06-16T14:15:35 no, i don't have this yet 2010-06-16T14:16:03 is it on the CVS head? 2010-06-16T14:17:10 crap.. confused parrot and go again. This was a patch to help out the Go port. LOL 2010-06-16T14:17:21 haha, that's ok 2010-06-16T14:17:39 i was going to say i didn't think that was my issue, but then again you know plenty of things i don't... 2010-06-16T14:19:00 you and vinu are (in my muddled mind) both tripping over issues which require minor tweaks of RTEMS to make things work. But I don't think you have identified any of these yet 2010-06-16T14:19:20 DrJoel: i had to patch a bit of the parrot source to get it compile clean on sparc 2010-06-16T14:19:31 something about AF_LOCAL 2010-06-16T14:19:41 grrr.. networking constant 2010-06-16T14:19:49 yeah, but it's ok 2010-06-16T14:19:53 it was only in one place 2010-06-16T14:20:07 and i realized that targeting sparc first would be a big pain 2010-06-16T14:20:23 so i switched to i386 and now everything is configuring and compiling clean 2010-06-16T14:20:34 and qemu is booting, it's just throwing an error 2010-06-16T14:20:43 i think i have to compile / link correctly with the init.c 2010-06-16T14:21:06 *** gedare has joined #rtems 2010-06-16T14:23:38 This is where looking at how the ada main is run might help.. see the ada-examples CVS module. It has a "magic" init task which knows what to do. Then runs gnat_main 2010-06-16T14:23:57 thanks for the pointer, i'll definitely check it out 2010-06-16T14:25:21 I also just committed a change to cpukit/score/src/objectallocate.c which is useful for debugging in your situation. If you would ever have an OS object (thread, mutex, etc) fail to allocate, it prints a message to tell you about the type so you can know what to configure more of 2010-06-16T14:25:34 It has to be conditionally compiled to turn on 2010-06-16T14:25:38 ah, that sounds good 2010-06-16T15:03:36 *** bubaflub has quit IRC 2010-06-16T15:15:38 *** gedare has quit IRC 2010-06-16T15:16:07 *** gedare has joined #rtems 2010-06-16T15:31:32 drjoel: your hvac working today? :) 2010-06-16T15:32:37 yep.. but yesterday during the rain, the roof leak appeared to start again 2010-06-16T15:32:45 did you get the patch I sent out for review? 2010-06-16T15:32:59 I am in the process of trying to build the sparc64 bsps 2010-06-16T15:33:46 i sent back a review already :p 2010-06-16T15:34:12 i have a similar problem today -- getting windows and patio replaced, i've had a gaping hole in my kitchen for the past 2.5 hours 2010-06-16T15:34:20 oh my.. just arrived 2010-06-16T15:34:37 I think kitchen remodel is the worst.. that sucked. 2010-06-16T15:34:53 wife wants new cabinets, so it'll happen again 2010-06-16T15:34:55 we need some bathroom remodeling.. out to the studs kind of work 2010-06-16T15:35:02 roxana is putting new cabinets in their new place 2010-06-16T15:35:07 yikes 2010-06-16T15:35:29 it can take a LONG LONG time but we were lucky.. only 3 weeks from start of tear down to almost functional again. range took 6 weeks though. 2010-06-16T15:35:48 roxana is making? hanging? or paying someone? 2010-06-16T15:36:00 ikea, doing it themselves i think 2010-06-16T15:36:06 i'll probably give a hand if i have time 2010-06-16T15:36:17 ohhh... 2010-06-16T15:37:09 we found a guy about 30 miles outside the city limits with a cabinet shop.. all custom made with slate counter top for less than Home Depot's cheapest ones you could order 2010-06-16T15:37:19 very cool 2010-06-16T15:37:30 we'll probably do home depot, although i'd like to shop around personally 2010-06-16T15:37:49 but time is $ :) 2010-06-16T15:38:04 we probably saved $10K and got better stuff 2010-06-16T15:38:28 well, we're in a starter home, so i don't think we'll go too big 2010-06-16T15:38:33 I have no idea in your area if it is possible to even find the small town cabinet maker 2010-06-16T15:38:49 yeah me either, not without driving an hour i would guess. 2010-06-16T15:39:17 we are in a 5 bedroom house but the kitchen is only 12x12.. we did floor to ceiling cabinets as much as possible. 2010-06-16T15:39:41 There are probably pictures in Michele's facebook.. nothing too fancy.. 2010-06-16T15:39:54 cool. I'd like to do rollout cabinets :) 2010-06-16T15:40:12 the extravagant things were the farm sink and a piece of stained glass as the panel in the door above the fridge 2010-06-16T15:40:21 ah, neat 2010-06-16T15:40:53 and the stained glass was ~$400 custom made .. so not terrible 2010-06-16T15:41:42 it looks awesome.. 2010-06-16T15:51:03 well, i finished the paper i was working on, and got those sparc64 BSPs refitted, reviewed your patch... not a bad day I suppose :) 2010-06-16T15:57:52 I just finished your email. Not much to address. have to recompile before generatign another patch. 2010-06-16T15:58:29 FWIW I left some SMP code in there in the new files.. it was harder to remove it than to keep it. I suppose I could eliminate all the RTEMS_SMP ifdef's but it seemed like unnecessary work 2010-06-16T16:02:15 I think so too, and it helped me to see it for comparison 2010-06-16T16:03:24 I think this is a good baseline. The next patch will have the BSP interface. I want the work to be open 2010-06-16T16:10:35 yeah. its a pretty big patch but a lot of it is repetitive in the cpu interrupt handling code. 2010-06-16T16:10:54 anyway, im off for today. bye 2010-06-16T16:10:59 I tried to minimize the damage to ports not supporting SMP :D 2010-06-16T16:11:03 night 2010-06-16T16:11:04 *** gedare has quit IRC 2010-06-16T16:11:06 *** DrJoel has quit IRC 2010-06-16T16:39:19 *** exception13 has left #rtems 2010-06-16T16:47:24 *** dr__house has quit IRC 2010-06-16T17:39:31 *** cdcs has quit IRC 2010-06-16T17:44:52 *** DrJoel has joined #rtems 2010-06-16T17:44:52 *** ChanServ sets mode: +o DrJoel 2010-06-16T17:45:10 *** DrJoel has quit IRC 2010-06-16T19:25:05 *** kiwichris has joined #rtems 2010-06-16T23:02:58 *** n3oo3n has joined #rtems 2010-06-16T23:42:20 *** kiwichris has quit IRC 2010-06-17T00:36:04 *** kiwichris has joined #rtems 2010-06-17T00:48:27 *** tiemen has quit IRC 2010-06-17T01:39:01 *** dr__house has joined #rtems 2010-06-17T01:39:03 *** dr__house has joined #rtems 2010-06-17T01:51:03 *** sebhub has joined #rtems 2010-06-17T01:51:15 good morning 2010-06-17T02:39:03 *** arvind_khadri has joined #rtems 2010-06-17T03:29:13 *** arvind_khadri has quit IRC 2010-06-17T03:42:32 *** arvind_khadri has joined #rtems 2010-06-17T03:46:40 sebhub, hi 2010-06-17T04:07:02 *** kiwichris has quit IRC 2010-06-17T04:14:30 *** exception13 has joined #rtems 2010-06-17T04:26:16 *** madrazr has joined #rtems 2010-06-17T04:28:15 *** arvind_khadri has quit IRC 2010-06-17T04:41:02 *** arvind_khadri has joined #rtems 2010-06-17T04:41:43 *** madrazr has quit IRC 2010-06-17T05:08:32 *** arvind_khadri has quit IRC 2010-06-17T05:15:13 *** exception13 has quit IRC 2010-06-17T05:21:00 *** arvind_khadri has joined #rtems 2010-06-17T05:54:56 *** madrazr has joined #rtems 2010-06-17T05:56:39 *** backhappy has joined #rtems 2010-06-17T05:57:26 CONFIGURE_MP_MAXIMUM_PROXIES is the maximum number of concurrently active 2010-06-17T05:57:26 thread/task proxies in a multiprocessor system.What's the thread/task? 2010-06-17T07:18:59 what's the thread/task proxies? 2010-06-17T07:37:02 *** arvind_khadri has quit IRC 2010-06-17T07:54:40 *** tuxmaniac has joined #rtems 2010-06-17T08:20:45 *** tuxmaniac has quit IRC 2010-06-17T08:57:37 *** bubaflub has joined #rtems 2010-06-17T08:59:45 *** backhappy has quit IRC 2010-06-17T09:36:23 *** exception13 has joined #rtems 2010-06-17T09:53:08 *** DrJoel has joined #rtems 2010-06-17T09:53:08 *** ChanServ sets mode: +o DrJoel 2010-06-17T10:18:06 *** madrazr has quit IRC 2010-06-17T10:29:32 *** gedare has joined #rtems 2010-06-17T10:29:47 morning drjoel 2010-06-17T10:31:10 morning.. I think you and Ralf crossed PR comments. :) 2010-06-17T10:31:41 FWIW chris is right on the offset issue.. varies from 2 -> 4 -> 8 depending on the target 2010-06-17T10:34:01 yeah, that's what i meant by sizeof(void*) -- and I realized it wasn't do-able in the code, I just wasn't being clear 2010-06-17T10:34:14 it does seem like a good autoconf issue 2010-06-17T10:38:38 *** bubaflub has quit IRC 2010-06-17T10:38:41 I hope it turns out to be non-controversial.. I don't feel like drama 2010-06-17T10:41:03 *** sebhub has quit IRC 2010-06-17T10:41:34 as i understand, anything touching autoconf = drama :) 2010-06-17T10:43:21 mostly just scary code to me. :) 2010-06-17T10:43:43 heh. fwiw, i'd love to see the helenos code removed, but as i said, i'm not doing it 2010-06-17T10:45:04 it isn't worth the trouble. really 2010-06-17T10:45:47 right. and it isn't very much code, but it isn't in "RTEMS" style, so harder to review I suppose. 2010-06-17T10:46:36 In this case (like the Nintendo DS), the "criticality" of the target in RTEMS isn't worth the effort 2010-06-17T10:46:46 *nod* 2010-06-17T10:46:52 there is some ugly code copied from the nds development world 2010-06-17T10:51:11 *** bubaflub has joined #rtems 2010-06-17T11:14:58 *** gedare_ has joined #rtems 2010-06-17T11:16:18 *** gedare has quit IRC 2010-06-17T11:16:56 *** gedare_ has quit IRC 2010-06-17T11:17:12 *** gedare_ has joined #rtems 2010-06-17T11:17:20 *** gedare_ is now known as gedare 2010-06-17T11:17:53 drjoel: fyi, I haven't run usiii on a free simulator, only on simics 2010-06-17T11:19:00 ok.. sorry for overstating.. it SHOULD eventually run on qemu is our believe, right? 2010-06-17T11:21:10 all code merged now. you should do a fresh check out at this point and verify we didn't break anything :D 2010-06-17T11:51:34 eventually I think so.. and I will do that. 2010-06-17T11:52:34 I don't fret too much over properly licensed, functional code from a 3rd party in a BSP. Just as long as it is possible to update it if needed 2010-06-17T11:52:49 ok, that's kind of what i thought 2010-06-17T11:58:25 checked out cvs, going to build niagara and try it out. 2010-06-17T12:06:14 *** dr__house has quit IRC 2010-06-17T12:06:57 crap, alignment error... 2010-06-17T12:07:15 oh 2010-06-17T12:07:24 of course, since the object extend hasn't been applied 2010-06-17T12:07:32 i'll locally make that change and see 2010-06-17T12:11:37 niagara hello and ticker run on M5 (after making that alignment fix) 2010-06-17T12:15:01 great! 2010-06-17T13:04:06 *** madrazr has joined #rtems 2010-06-17T13:42:17 DrJoel: sent you a short blurb for a news item... going to go watch soccer for a bit ;) 2010-06-17T13:42:28 if you want more details or other angles let me know and I can write up a little more 2010-06-17T13:42:30 * gedare sick of writing 2010-06-17T13:44:01 enjoy the soccer.. 2010-06-17T15:00:52 drjoel: nice spot on the gcc define 2010-06-17T15:01:44 but now i have to go fix a toilet. if it isn't one thing... 2010-06-17T15:01:48 I was in the process of adding a long list of "if def this cpu" when I happened upon it in the output of "gcc -E -dM - heh, funny how that happens 2010-06-17T15:02:07 I still don't have the one I worked on this weekend completely right . Flaps falls too quickly 2010-06-17T15:02:33 too much tension on the pull chain 2010-06-17T15:02:57 maybe.. anyway, i have to replace the wax ring, so i get to pull the whole thing. fun 2010-06-17T15:03:06 don't think i'll be back, so i'll turn off now. Later 2010-06-17T15:03:37 are you doing a "real wax ring" or the fluidmaster rubber o-ring alternative? 2010-06-17T15:03:49 if you aren't good at it, the o-ring alternative is harder to screw up 2010-06-17T15:03:54 * DrJoel is not good at it 2010-06-17T15:05:06 yeah, with the wax ring you've gotta drop the toilet straight down without too much moving around 2010-06-17T15:06:17 kind of sensitive and no real way to know it is OK until it leaks. :( 2010-06-17T15:06:36 yep, the kind of thing i leave to a landlord and/or professional 2010-06-17T15:07:23 *** madrazr has quit IRC 2010-06-17T15:08:27 *** gedare has quit IRC 2010-06-17T15:58:33 *** DrJoel has quit IRC 2010-06-17T15:58:47 *** bubaflub has quit IRC 2010-06-17T16:28:51 *** exception13 has left #rtems 2010-06-17T17:34:37 *** kiwichris has joined #rtems 2010-06-17T17:50:01 *** bubaflub has joined #rtems 2010-06-17T17:50:18 hello kiwichris 2010-06-17T18:37:15 *** bubaflub has quit IRC 2010-06-17T20:13:39 *** DrJoel has joined #rtems 2010-06-17T20:13:39 *** ChanServ sets mode: +o DrJoel 2010-06-17T20:13:47 * DrJoel waves hello 2010-06-17T20:20:30 kiwichris: are you there? 2010-06-17T20:56:47 ping chris 2010-06-17T21:00:41 chris.. /home/joel/rtems-4.11-work/build/gcc-testing/rtems/network-demos/tftpTest/init.c:39: undefined reference to `rtems_bsdnet_initialize_tftp_filesystem' 2010-06-17T21:00:41 collect2: ld returned 1 exit status 2010-06-17T21:00:47 network-demos on head and 4.10 2010-06-17T21:00:54 *** DrJoel has quit IRC 2010-06-17T21:35:07 *** DrJoel has joined #rtems 2010-06-17T21:35:07 *** ChanServ sets mode: +o DrJoel 2010-06-17T21:40:30 *** DrJoel has quit IRC 2010-06-17T23:59:54 *** madrazr has joined #rtems 2010-06-18T01:12:48 *** sebhub has joined #rtems 2010-06-18T01:12:57 good morning 2010-06-18T02:27:37 *** madrazr has left #rtems 2010-06-18T02:27:49 hi seb 2010-06-18T02:28:11 hi 2010-06-18T02:28:17 today is my last day with Frye 2010-06-18T02:28:26 I start with intel on monday 2010-06-18T02:28:44 nice 2010-06-18T02:28:54 there's got to be some way to sneak RTEMS in there 2010-06-18T03:25:36 *** lee_ has joined #rtems 2010-06-18T03:37:35 *** lee_ has quit IRC 2010-06-18T03:41:07 *** lee__ has joined #rtems 2010-06-18T04:02:15 *** kiwichris has quit IRC 2010-06-18T04:35:13 *** exception13 has joined #rtems 2010-06-18T05:00:20 *** dr__house has joined #rtems 2010-06-18T05:04:41 *** sebhub has quit IRC 2010-06-18T05:47:12 *** lekernel has quit IRC 2010-06-18T05:51:04 *** lekernel has joined #rtems 2010-06-18T07:02:59 *** madrazr has joined #rtems 2010-06-18T07:38:38 *** exception13 has left #rtems 2010-06-18T07:50:32 *** dr__house has quit IRC 2010-06-18T08:40:23 *** DrJoel has joined #rtems 2010-06-18T08:40:23 *** ChanServ sets mode: +o DrJoel 2010-06-18T10:37:08 *** rtemsLogger has joined #rtems 2010-06-18T10:54:44 *** gedare has joined #rtems 2010-06-18T11:09:53 * DrJoel reports that rtems.org is now downloading about 1.5G in OS updates after rebooting with Fedora 12 on it. Still work to be done after that. 2010-06-18T11:10:02 hah 2010-06-18T11:10:21 that's a lot of stuff for a server ! 2010-06-18T11:10:36 Yep! It does a lot of stuff for RTEMS. :-D 2010-06-18T11:11:26 The tedious part is making sure all the services still work. CVS appears to be back up OK though 2010-06-18T11:11:41 httpd is still down 2010-06-18T11:20:59 *** kuzew has quit IRC 2010-06-18T11:22:39 *** kuzew has joined #rtems 2010-06-18T12:36:26 http://gedare-csphd.blogspot.com/2010/06/building-and-booting-rtems-sparc64.html 2010-06-18T12:37:19 when the wiki comes back i'll add a link to the download I made to help build a bootable image 2010-06-18T12:43:11 Great. The download of 1.5 Gb is done. Now yum says it has ~2750 steps to complete the installation. 2010-06-18T12:45:50 haha 2010-06-18T12:46:47 well yum was polite enough to download all of the updates before complaining about some conflicts. 2010-06-18T12:47:40 normal upgrade on this machine.. grub had to be manually reinstalled from emergency disk, yum wouldn't run after upgrade DVD done until some symlinks to hack point to certain dynamic libraries were in place,... 2010-06-18T12:58:59 *** exception13 has joined #rtems 2010-06-18T13:02:35 *** dukeleto has quit IRC 2010-06-18T13:07:58 fun for you 2010-06-18T13:16:14 about 900 of ~2800 yum steps done 2010-06-18T13:20:21 *** gedare has quit IRC 2010-06-18T13:22:15 *** gedare has joined #rtems 2010-06-18T14:57:59 *** madrazr has quit IRC 2010-06-18T15:02:07 *** rtemsLogger has joined #rtems 2010-06-18T15:21:50 *** bubaflub has quit IRC 2010-06-18T15:21:58 *** bubaflub has joined #rtems 2010-06-18T15:27:54 *** gedare has quit IRC 2010-06-18T15:38:09 *** exception13 has left #rtems 2010-06-18T15:50:10 *** lekernel has quit IRC 2010-06-18T16:52:16 how goes the update, DrJoel ? 2010-06-18T16:53:23 slowly .. the server is back up .. Bugzilla is being a bitch about the upgrade. We think most things are OK 2010-06-18T16:53:35 It was 1.5GB of updates to download!!! 2010-06-18T16:56:45 haha. nice. 2010-06-18T16:56:53 yeah, Bugzilla has a bunch of Perl dependencies and sometimes those can be fragile 2010-06-18T16:59:42 Yep. Combine that with a number of database inconsistencies, it is just now deciding not to like. :( 2010-06-18T17:22:59 *** DrJoel has quit IRC 2010-06-18T17:23:49 *** DrJoel has joined #rtems 2010-06-18T17:23:50 *** ChanServ sets mode: +o DrJoel 2010-06-18T17:23:51 *** DrJoel has quit IRC 2010-06-18T17:36:35 *** bubaflub has quit IRC 2010-06-18T17:54:02 *** aniceberg has joined #rtems 2010-06-18T18:36:29 *** aniceberg has quit IRC 2010-06-19T00:08:12 *** madrazr has joined #rtems 2010-06-19T01:49:54 *** kv has joined #rtems 2010-06-19T02:20:50 *** madrazr has quit IRC 2010-06-19T02:27:45 *** kv has quit IRC 2010-06-19T03:21:01 *** exception13 has joined #rtems 2010-06-19T03:33:52 *** madrazr has joined #rtems 2010-06-19T03:54:36 *** madrazr has quit IRC 2010-06-19T06:32:55 *** madrazr has joined #rtems 2010-06-19T07:38:38 *** _Lucretia__ has quit IRC 2010-06-19T09:12:18 *** krzabr has joined #rtems 2010-06-19T09:34:29 *** DrJoel has joined #rtems 2010-06-19T09:34:29 *** ChanServ sets mode: +o DrJoel 2010-06-19T09:38:01 *** madrazr has quit IRC 2010-06-19T09:39:47 *** madrazr has joined #rtems 2010-06-19T09:46:19 *** madrazr has quit IRC 2010-06-19T09:46:29 *** madrazr has joined #rtems 2010-06-19T09:57:59 *** mpanetta has joined #rtems 2010-06-19T10:11:04 Does anyone know if RTEMS handles Cortex-M3 core uC's yet? 2010-06-19T10:12:05 rtems is microkernel or monolithic ? 2010-06-19T10:12:40 mpanetta: I don't know There is a simulator BSP I did for the simulator in gdb to explore what arm variations work. I know the a9 has issues. see arm/gdbarmsim and add a make/custom for the CPU flags you are concerned about. 2010-06-19T10:18:03 Ahh ok, if A9 does not work then M3 won't. I was just looking through the CVS source and it seems that it is not supported. At least not in cpukit as far as I can tell. 2010-06-19T10:18:28 M3 is thumb2 only. 2010-06-19T10:18:59 there are so many variations these days.. Thumb should work but not thumb2 if I understand correctly. But not that much code? right? 2010-06-19T10:21:49 The thumb2 devices are much more simplified WRT interrupt handling and so forth because the hardware stacks the interrupt frame for you. It is almost completely different then standard ARM interrupt structure from what I know. 2010-06-19T10:22:23 And there is no switching between arm and thumb mode (arm mode does not exist) 2010-06-19T10:22:29 ahhh.. I recall someone telling me it was different. Simplified is good though. 2010-06-19T10:23:03 Shouldn't be terribly difficult since you don't have to figure out the architecture completely.. just transform the existing code into it. Tedious yes. :) 2010-06-19T10:23:27 * DrJoel notes that he has plenty of uncovered consulting time coming up and was about to send the NPR-ish fund drive email begging soon 2010-06-19T10:25:25 It might be possible to squeeze it in to the existing infrastructure, but there would be a few ifdefs to disable the mode switching code at a minimum. 2010-06-19T10:26:14 It is also possible to easily switch an alternate file and just avoid the ifdef. Probably better .. is there a mixed mode with Thumb2 available? 2010-06-19T10:26:20 DrJoel: This is only a hobby for me else I would devote more time to it :( 2010-06-19T10:26:48 DrJoel: Not with the M0 and M3, I don't know about the A8,9 2010-06-19T10:26:52 it is a hobby as much as a paying thing to me also. Here it is Saturday and I am adding a test 2010-06-19T10:27:01 Cool :) 2010-06-19T10:27:28 If you want to tackle it as a slow project, there are people interested so you could get help. No one wants to tackle it on their own so far 2010-06-19T10:27:33 *** krzabr has left #rtems 2010-06-19T10:28:20 Yeah I think I asked on the list about a year ago, it seemed that way then too ;) 2010-06-19T10:29:01 I would rather switch files. Ifdefs are useful but ugly at times :) 2010-06-19T10:29:11 Esp for something that is so different... 2010-06-19T10:29:17 I am trying on big projects that are at best a mix of hobby, students and paid work (if lucky) to make a project plan which is a bunch of tiny steps that can be done by whoever is interested 2010-06-19T10:29:35 that is how I am approaching SMP. First patch for review put into Bugzilla yesterday 2010-06-19T10:29:42 If I started the port as a second CPU core type next to ARM, would that make it easier or harder to merge it back when it is done? 2010-06-19T10:30:37 I would ask on the list so Ralf can comment. We are having this discussion about the sparc64. Similar enough where you are tempted to merge it but different enough to scare me about screwing up the sparc which is used in space apps 2010-06-19T10:30:42 RTEMS is getting SMP? 2010-06-19T10:30:50 working on it... 2010-06-19T10:30:54 Neat 2010-06-19T10:31:10 first step is moving some score data into a "per cpu" structure. That is the first step 2010-06-19T10:31:13 Ok, I will make a list posting. 2010-06-19T10:31:15 the patch does that 2010-06-19T10:31:57 I have multi-core systems booting on qemu (1-32 cores) and leon3 (4 cores). I can initialize, do intercore interrupts, and am working on the forcing another core to dispatch part now. 2010-06-19T10:32:21 Gedare's GSOC project is refactoring a scheduler out of the current core implementation so we can add an smp scheduler. 2010-06-19T10:32:34 there is the issue of OS locking and interrupts left to mentally solve 2010-06-19T10:35:31 Will it be doing interrupt scheduling between CPU's or will the interrupts just be pinned at first? 2010-06-19T10:36:15 pinned to first cpu at first for simplicity .. this is complicated enough.. I have a project plan that has a "simple baseline" 2010-06-19T10:37:07 I also put together an RTEMS Scheduler Simulator which builds about 200 files in the tree with a command line under Linux so you can see how a scheduling algorithm behaves. The goal is to have RTEMS support SMP and be useful for easy experimentation with scheduling algorithms 2010-06-19T10:37:24 Gedare did an EDF for RTEMS so we are hoping it is there out of the box and easy to configure.. 2010-06-19T10:37:39 I am by no means an RTOS expert, I was mainly just asking out of curiosity :) 2010-06-19T10:37:59 Learn on the job. :-D Computer science is all about learning 2010-06-19T10:38:09 Yes it definitly is 2010-06-19T10:38:49 And that is what makes it fun IMO 2010-06-19T10:39:21 if you don't learn, you become obsolete. I had 3 reasons in my 20s for getting a PhD. (1) expected to burn out or be obsolete in 30s . needed management option (2) may want to teach in university (3) good card for writing books and looking impressive to publishers 2010-06-19T10:39:51 so far (1) hasn't happened... (2) is unlikely to happen since I really hated teaching the 1 grad class I taught .. and (3) is not going to happen until the kids are gone 2010-06-19T10:40:03 haha 2010-06-19T10:41:51 I have learned I like to do more research type projects.. RTEMS is still fun after 20 years on and off.. I have mostly done consulting on other things and done this part time or hooby 2010-06-19T10:42:53 I am mostly self taught... Makes it a bit harder to get a job in the field of my choice... However I do know what I am doing most of the time and the people that have hired me like me and my work. 2010-06-19T10:45:26 one of OAR's best people has a music degree and worked 1 year as a high school band director.. not his cup of tea after all 2010-06-19T10:45:29 The last thing I did with RTEMS was fix the support for the MRM32, before it died... I wonder if they even still sell that thing. 2010-06-19T10:45:54 We still have one in the lab. But I have no idea if we could buy another 2010-06-19T10:46:30 I don't think Freescale makes the 68332 anymore. 2010-06-19T10:46:42 that would be a major obstacle .. :) 2010-06-19T10:46:56 heh yeah it would be 2010-06-19T11:23:31 Sent the email. Hopefully it makes it through. 2010-06-19T12:22:15 hi DrJoel :) 2010-06-19T12:22:22 i don't know if you got my previous message 2010-06-19T12:22:24 but it works now ! 2010-06-19T12:23:24 /win/j #milkymist 2010-06-19T12:23:31 sorry 2010-06-19T12:38:55 i just pushed 2010-06-19T12:39:03 i didn't have internet where we were working 2010-06-19T13:12:37 mpanetta: yes it did. After the upgrade yesterday, Mediawiki needed hand tweaking to get running again and Bugzilla was a two hour nightmare to get back again. I had to handwrite SQL and correct the database so the upgrade script would run. But we forgot to put the sendmail configuration file back in place so that was broken overnight. 2010-06-19T13:13:02 Fallenou: yes .. saw that!! I just send you an email. 2010-06-19T13:13:31 Yuk, glad to hear it is fixed! 2010-06-19T13:13:32 If you can see how to add this as a script to gcc-testing/simscripts like the qemu-lm32_evr that would be great. Submit it all as a PR and we can start reviewing 2010-06-19T13:14:44 this was a very painful upgrade that took a full day for some reason. grub had to be hand-installed via an emergency disk. 1.5GB of updates to download and sort through after the update. Then yum broke due to a library problem and had to be fixed by hand. Then the normal web services that need attention 2010-06-19T13:15:01 We moved from Fedora 11 -> 12 on the server 2010-06-19T13:15:29 * DrJoel goes back to house chores.. toilet supply line is leaking I think 2010-06-19T13:25:12 *** madrazr has joined #rtems 2010-06-19T14:19:44 *** madrazr has quit IRC 2010-06-19T14:23:16 *** cdcs has joined #rtems 2010-06-19T14:58:29 *** cdcs has quit IRC 2010-06-19T15:14:46 *** lekernel has joined #rtems 2010-06-19T15:23:48 *** cdcs has joined #rtems 2010-06-19T16:14:14 *** lekernel has quit IRC 2010-06-19T16:14:46 *** cannedfish has joined #rtems 2010-06-19T16:38:39 *** cdcs has quit IRC 2010-06-19T17:34:33 *** exception13 has left #rtems 2010-06-19T17:37:02 *** cannedfish has quit IRC 2010-06-19T18:33:25 *** bubaflub has joined #rtems 2010-06-19T18:52:59 hello DrJoel 2010-06-19T19:27:03 *** bubaflub has quit IRC 2010-06-19T19:54:40 *** _Lucretia_ has joined #rtems 2010-06-19T22:26:30 *** kv has joined #rtems 2010-06-20T01:29:48 *** kv has quit IRC 2010-06-20T02:43:43 *** tfinn has quit IRC 2010-06-20T02:48:01 *** mpanetta has quit IRC 2010-06-20T02:49:50 *** mpanetta has joined #rtems 2010-06-20T03:12:07 *** lekernel has joined #rtems 2010-06-20T03:30:34 *** lekernel has quit IRC 2010-06-20T05:32:37 *** exception13 has joined #rtems 2010-06-20T07:30:01 *** cdcs has joined #rtems 2010-06-20T08:31:31 *** cdcs has quit IRC 2010-06-20T08:43:23 *** cdcs has joined #rtems 2010-06-20T09:41:30 *** backhappy has joined #rtems 2010-06-20T09:44:37 IN the RTEMS API Configuration Table ,what's the macro correspond to the User initialization tasks table? 2010-06-20T09:45:32 hold on.. will point to link once Drupal upgrade is finished.. need to concentrate :) 2010-06-20T09:49:35 http://www.rtems.org/onlinedocs/releases/rtemsdocs-4.9.4/share/rtems/html/c_user/c_user00427.html 2010-06-20T09:50:07 If you define CONFIGURE_RTEMS_INIT_TASKS_TABLE, it assumes every other attribute (method name Init) 2010-06-20T09:50:22 ok!I think you should check the 4.99 C USER's guide, the RTEMS API Configuration Table----User initialization tasks table 2010-06-20T09:50:51 maybe a little error 2010-06-20T09:51:10 what specifically? I don't think that text has changed in years 2010-06-20T09:51:45 User initialization tasks table 2010-06-20T09:51:45 is the address of the Initialization Task Table. This table contains 2010-06-20T09:51:45 the information needed to create and start each of the initialization 2010-06-20T09:51:45 tasks. The format of this table will be discussed below. When using 2010-06-20T09:51:45 the rtems/confdefs.h mechanism for configuring an RTEMS application, 2010-06-20T09:51:46 the value for this field corresponds to the setting of the macro 2010-06-20T09:51:48 CONFIGURE_EXECUTIVE_RAM_WORK_AREA. 2010-06-20T09:52:03 the last line ? 2010-06-20T09:55:17 Also the Basic System Information -----CONFIGURE_MESSAGE_BUFFER_MEMORY also exist a little question, #define CONFIGURE_MESSAGE_BUFFERS_FOR_QUEUE(...) should change into the #define CONFIGURE_MESSAGE_BUFFER_MEMORY(...) 2010-06-20T09:55:48 Can you file a PR for each of these? Please. 2010-06-20T09:56:45 I think the issue with the init task table one is that not only is it wrong but that field doesn't exist in the main configuration table anymore. It was split into its own table with a known name. I suspect I need to review that entire chapter :( 2010-06-20T09:57:24 The other I just want to see and track. it is Sunday morning here. Plus this will likely have to be address in 4.9 and 4.10 branche 2010-06-20T09:58:18 sorry,I am a newer,can't file a PR well,maybe you can do that? 2010-06-20T09:58:35 oh you can do it. :) Don't sell yourself short 2010-06-20T09:59:08 https://www.rtems.org/bugzilla/ it is pretty straight forward from there 2010-06-20T09:59:20 ok!let me try! 2010-06-20T10:01:04 Hope you find more error in this charpter,there 's many little error that I can't find,you know.. 2010-06-20T10:01:19 *** cdcs_ has joined #rtems 2010-06-20T10:01:22 oh don't wish finding doc errors on me.. :-D 2010-06-20T10:01:23 *** cdcs has quit IRC 2010-06-20T10:02:07 haha you bet? I can give you more immediately~ 2010-06-20T10:02:32 *** cdcs_ is now known as cdcs 2010-06-20T10:04:03 there 's no macro about barriers in document~,you should add 2010-06-20T10:04:39 just file a number of small PRs.. docs on active release branches have to be fixed also for some of these. 2010-06-20T10:06:20 ok! try my best! 2010-06-20T10:06:49 I'll be here a few more minutes and anyone on IRC can help if you run into trouble filing a PR 2010-06-20T10:07:03 It is even better if you can provide a diff as an attachment but let's focus on filing first 2010-06-20T10:07:57 ok!I think I can do both~ 2010-06-20T10:08:47 thanks. It makes it a lot easier. and smaller patches are easier to backport .. that's why keeping the issues distinct is important.. plus the error may not apply to an older branch 2010-06-20T10:12:30 which type of the attachment should I use? doc? txt? 2010-06-20T10:13:35 *** cdcs_ has joined #rtems 2010-06-20T10:13:51 when you make the single change, do a "cvs diff -u XXX.t >/tmp/prXXX.diff 2010-06-20T10:14:03 it should autodetect if you click patch 2010-06-20T10:14:10 *** cdcs has quit IRC 2010-06-20T10:16:43 thanks! 2010-06-20T10:17:06 thank you 2010-06-20T10:17:18 it is hard to get doc issues reported :) 2010-06-20T10:18:25 Maybe I wil find more to help you improve! 2010-06-20T10:18:36 that would be appreciated. 2010-06-20T10:19:49 Share world! I should do that! 2010-06-20T10:50:20 *** madrazr has joined #rtems 2010-06-20T11:05:45 hi, where should my rtems errno.h be? 2010-06-20T11:06:24 errno.h is a posix .h file ... 2010-06-20T11:08:33 strange... i have #include but i get error: 'errno' undeclared 2010-06-20T11:10:10 maybe not in the file that is using it? 2010-06-20T11:10:16 or before a macro that uses it? 2010-06-20T11:11:33 errno.h is my first include and is in the same file where i use errno... 2010-06-20T11:11:38 I just submit a PR, but some description write wrong after I submit.. 2010-06-20T11:11:41 i can't find errno.f 2010-06-20T11:12:13 i can't find errno.h in rtems folder... 2010-06-20T11:23:15 *** backhappy has quit IRC 2010-06-20T11:27:39 it is under /opt/rtems-4.11/..... 2010-06-20T11:27:43 * DrJoel is going awk 2010-06-20T11:37:23 *** cdcs_ has quit IRC 2010-06-20T11:39:37 *** cdcs has joined #rtems 2010-06-20T14:13:49 When was the last time someone did a test compile on the nds BSP for HEAD? 2010-06-20T14:13:59 Anyone now? 2010-06-20T14:14:02 er know 2010-06-20T14:14:38 Yesterday.. http://www.rtems.org/ftp/pub/rtems/people/chrisj/pq/html/arm.html 2010-06-20T14:19:59 That is for the 1136js tho. I tried compiling the nds BSP last night and some of the tests are failing to compile. 2010-06-20T14:20:11 scroll down.. all BSPs are there 2010-06-20T14:20:23 Since I have an NDS with a card I figured I could use it to get myself back up to speed with RTEMS :) 2010-06-20T14:20:26 ohhh tests.. I committed some changes.. update 2010-06-20T14:20:26 Sorry 2010-06-20T14:20:29 Ahh! 2010-06-20T14:20:32 Ahh ok 2010-06-20T14:20:52 one of the summer of code students is working on new tests and I was trying to clean some stuff up 2010-06-20T14:20:57 brb 2010-06-20T14:21:01 ok 2010-06-20T14:26:11 Still getting the error. Maybe something has changed? 2010-06-20T14:26:12 ../../../../../nds/lib/include/rtems/confdefs.h:1891:5: error: 'CONFIGURE_MAXIMUM_FIFOS' undeclared here (not in a function) 2010-06-20T14:26:26 Its the spfifo02 test 2010-06-20T14:28:12 Hmm maybe that is my fault... Did I miss enabling somthing at config time? 2010-06-20T14:40:29 No.. it is transient. i was/am working on confdefs.h for fifos. hold on.. let me commit something 2010-06-20T14:41:38 update now and try again 2010-06-20T14:43:47 *** madrazr has left #rtems 2010-06-20T14:46:05 *** kuzew has quit IRC 2010-06-20T14:47:08 *** kuzew has joined #rtems 2010-06-20T14:47:51 Ok, sec 2010-06-20T14:50:14 hopefully it will be ok. 2010-06-20T14:50:17 * DrJoel is going awk 2010-06-20T14:56:55 now I get: ../../../../../nds/lib/include/rtems/confdefs.h:316:7: error: #error "No FIFOs or PIPES configured" 2010-06-20T14:57:25 On spfifo01 2010-06-20T14:58:13 Maybe NDS just does not support them. 2010-06-20T14:58:36 I can compile with --enable-tests=samples for now 2010-06-20T14:59:13 err 2010-06-20T14:59:22 that was for spfifo02 not spfifo01 2010-06-20T15:25:50 *** cdcs_ has joined #rtems 2010-06-20T15:26:10 *** cdcs has quit IRC 2010-06-20T15:28:11 *** cdcs_ is now known as cdcs 2010-06-20T16:34:02 *** lekernel has joined #rtems 2010-06-20T16:51:26 DrJoel: i'm gonna write a mail to gsoc-rtems + milkymist-dev with explanations about my progression, with a link to a blog post (that i also have to write) 2010-06-20T16:51:42 great! 2010-06-20T16:51:48 i'm really glad it runs :) 2010-06-20T16:52:07 things tend to be less frustrating after hello and ticker run 2010-06-20T16:52:09 not sure it's 100% clean though 2010-06-20T16:52:15 yes 2010-06-20T16:52:19 both of them are running :) 2010-06-20T16:52:30 i had the chance to test on actual hardware thanks to lekernel saturday 2010-06-20T16:52:41 but i still have qemu to test more 2010-06-20T16:53:06 try to write the helper for the script to run it in "rtems-testing/sim-script". Then you can easily run every test and see which pass/fail. 2010-06-20T16:53:20 oh ok i will read doc about that 2010-06-20T16:54:42 Fallenou, it's clean enough (afaik) 2010-06-20T16:55:09 about the places for the #defines and stuff like that 2010-06-20T16:55:11 except that the benchmark timer isn't properly calibrated but it's a detail 2010-06-20T16:55:12 and the indentation 2010-06-20T16:55:42 you think we got it wrong in the benchmarking function ? 2010-06-20T16:56:20 it's ok, but there are those #define'd time constants which may need some review 2010-06-20T16:56:25 I think it would be best if you clean up what you have, write the rtems-testing/sim-script helper, and submit it. Then we can review it and merge what you have. At this point, it should be bugs and additional drivers. 2010-06-20T16:56:28 ah, yes, check the indentation 2010-06-20T16:56:41 Plus we can give feedback on less code at a time. :) 2010-06-20T16:56:49 DrJoel: you can have a look in my github if you want to :) 2010-06-20T16:56:51 indentation == 2 spaces, no tabs 2010-06-20T16:56:56 aouch 2010-06-20T16:56:59 i hate spaces =( 2010-06-20T16:57:21 i will have to configure my vim to use spaces when i type tab 2010-06-20T16:57:23 sorry.. also nothing past column 79 2010-06-20T16:57:31 ok 2010-06-20T16:58:02 Also double check that you haven't copied something that should be shared. That is hard to spot for anyone but the author. 2010-06-20T16:58:33 i think we put something in shared, but it could be put in the BSP 2010-06-20T16:59:07 ? 2010-06-20T16:59:09 but it's not stupid since someone could use lekernel's uart and timer IP-core inside his project with lm32 2010-06-20T16:59:20 so it could be shared too ... 2010-06-20T16:59:25 ah 2010-06-20T16:59:26 well, yes 2010-06-20T16:59:28 lekernel: yes the milkymist_console and timer 2010-06-20T16:59:32 i think it belongs in shared 2010-06-20T16:59:43 yes i guess 2010-06-20T17:00:06 when i will port it to M1 i will have to use it :) 2010-06-20T17:00:23 there's nothing to port to M1... it'll work the same as on the ML401 2010-06-20T17:00:31 ooooh :( not funny ! =( 2010-06-20T17:00:32 :) 2010-06-20T17:02:17 that may be a case for a "BSP variant" if it is very very similar.. I don't know the differences so can't say. This is a way to change CPU_CFLAGS, a bit of conditional compilation, and enable/disable parts of the BSP if they are very very close 2010-06-20T17:04:41 yes we could do that :) 2010-06-20T17:07:45 well, the M1 port will hopefully be 100% compatible with the ML401, and the ML401 port will become obsolete soon 2010-06-20T17:07:51 lekernel: i could use the screenshot for my blog post ;) 2010-06-20T17:07:55 yes 2010-06-20T17:25:29 DrJoel, i have a doubt about includes... In my application i include in main.c file, then i have another unrelated .c file that includes "../sys/errno.h". When i try to compile i get a errno undefined error in main.c, if i remove errno.h from sys/errno.h it compiles correctly 2010-06-20T17:28:00 *** exception13 has left #rtems 2010-06-20T17:30:21 chris and I would have to see the code. I suspect you have something wrong that we won't catch by discussion. 2010-06-20T17:33:29 ok, i will send a mail to Chris :). I have worked around it. 2010-06-20T18:08:17 *** cdcs has quit IRC 2010-06-20T18:48:02 *** kiwichris has joined #rtems 2010-06-20T18:48:33 Hi all 2010-06-20T21:58:22 *** backhappy has joined #rtems 2010-06-20T23:24:48 *** kiwichris has quit IRC