2010-05-31T00:08:52 *** dr__house has joined #rtems 2010-05-31T01:50:07 *** sebhub has joined #rtems 2010-05-31T01:50:15 good morning 2010-05-31T02:08:20 *** exception13 has joined #rtems 2010-05-31T02:13:09 *** peerst has left #rtems 2010-05-31T02:54:29 *** lekernel has joined #rtems 2010-05-31T04:54:41 *** Raj_ has joined #rtems 2010-05-31T04:56:41 sebhub, hi 2010-05-31T05:00:03 Hi DrJoel 2010-05-31T05:00:18 hi 2010-05-31T05:00:44 chris: how do we proceed with the rtems sequenced init gsoc project? 2010-05-31T05:01:46 we have to define the api and the means we want to use 2010-05-31T05:28:57 *** lekernel has quit IRC 2010-05-31T05:41:56 *** Raj_ has quit IRC 2010-05-31T06:06:44 *** Raj_ has joined #rtems 2010-05-31T06:22:43 DrJoel, hi 2010-05-31T06:41:40 sebhub, I think you should arrange an IRC meeting and we can discuss it. 2010-05-31T06:47:04 ok 2010-05-31T06:51:15 Do you guys use the umon file system code in bsp/shared ? 2010-05-31T06:56:27 no 2010-05-31T06:56:37 Ok. 2010-05-31T07:27:50 *** dr__house has quit IRC 2010-05-31T07:41:47 *** lcpfnvc has quit IRC 2010-05-31T07:46:29 *** zwj has joined #rtems 2010-05-31T07:52:41 Hi, sebhub and chris. 2010-05-31T07:54:36 hi 2010-05-31T08:01:26 *** exception13 has left #rtems 2010-05-31T08:02:07 *** zwj has quit IRC 2010-05-31T08:02:21 *** wenjie has joined #rtems 2010-05-31T08:03:20 wenjie is zwj 2010-05-31T08:05:32 *** lcpfnvc has joined #rtems 2010-05-31T08:06:38 hi 2010-05-31T08:14:06 now all we three is on the line. can we talk about how to implement the sequenced initialization api? 2010-05-31T08:15:06 We need to book a time and also have Joel about. It is late here and I would to get a commit in then finish up. Sorry about this. 2010-05-31T08:16:09 ok, i send the mail also to joel 2010-05-31T08:16:40 Great. We have a chat for the Parrot work and it is simple and quick. 2010-05-31T08:17:38 this week i am not at work on thursday and friday 2010-05-31T08:19:13 Ok. It is 1:20pm UTC time. How does 10pm UTC look ? 2010-05-31T08:24:30 not that good for me, it is 24:00 local time 2010-05-31T08:24:52 What is you local time > 2010-05-31T08:24:54 What is you local time ? 2010-05-31T08:25:01 UTC+2 2010-05-31T08:25:05 my time zone is UTC+800, my available time is 11:00 to 16:00 UTC, and saturday and sunday is all day available 2010-05-31T08:25:44 wenjie, you are close to me. It must be 9pm ? 2010-05-31T08:25:54 because now my school is not on holiday 2010-05-31T08:26:23 sebhub, in the morning your time might be a good time. 2010-05-31T08:26:31 yeah, now it is 9:26 pm in my local time 2010-05-31T08:26:51 chris you have utc+9? 2010-05-31T08:27:07 utc+10 2010-05-31T08:27:15 and joel utc-6? 2010-05-31T08:27:58 ok, joel has utc-5 in the summer 2010-05-31T08:28:40 Maybe tomorrow 12pm UTC ? 2010-05-31T08:28:48 * DrJoel waves hi.. 2010-05-31T08:29:06 Hi Joel. Just figuring out a time an IRC meeting. 2010-05-31T08:29:23 it is a national holiday here (Memorial Day) 2010-05-31T08:30:12 I'm pretty flexible 2010-05-31T08:30:30 Hi DrJoel 2010-05-31T08:30:43 Hi Raj_ any progress? 2010-05-31T08:30:59 Well, I wasn't working much today :) 2010-05-31T08:31:19 I just got out of bed so just having coffee here 2010-05-31T08:31:20 am/pm is so confusing ;-) 2010-05-31T08:31:34 but I can confirm that installing with multilib doesn't put librtemscpu in $bsp/lib 2010-05-31T08:32:01 0:00 utc+0 is ok for me 2010-05-31T08:32:07 Raj_: right.. that puts the BSP independent library in $prefix/$target/$lib 2010-05-31T08:32:49 you mean i386-rtems4.10/lib , right ? 2010-05-31T08:32:50 DrJoel, have the mount changes finished. 2010-05-31T08:33:04 Raj_: yes 2010-05-31T08:33:07 kiwichris: great 2010-05-31T08:33:13 * DrJoel is looking at some of the odd warnings in the BSPs 2010-05-31T08:33:42 DrJoel: so doesn't that mean the Makefile.inc needs to be changed inside $bsp/lib 2010-05-31T08:33:52 I have redone the tftpfs file system so it is much simpler to use 2010-05-31T08:33:58 is it really necessary to fix all these warnings before the 4.10 release? 2010-05-31T08:34:38 The TFS file system in umon needs some work. 2010-05-31T08:35:17 some are indicative of integer sizing issues and a couple of times it has turned up something serious. But I am just fixing what I can while more serious changes get made 2010-05-31T08:35:28 kiwichris: is cleaning up mount, you and Thomas fixed the ARM BSP issues 2010-05-31T08:35:33 but we need to get branched 2010-05-31T08:36:14 DrJoel, check the build results. The box looks like it was having problems. Lots of seg faults. 2010-05-31T08:36:28 It is warm ? 2010-05-31T08:37:15 no .. only 68F this morning.. but we have had storms.. it is possible that something happened over the long weekend.. I haven't been in the office since Thursday 2010-05-31T08:37:32 Ah ok. 2010-05-31T08:37:57 The build results look bad when they are not. 2010-05-31T08:38:34 I see a core dump on csb360 which did not happen on england 2010-05-31T08:38:42 Yeah. 2010-05-31T08:38:56 I have built the entire set of BSPs at least every day for the past week 2010-05-31T08:38:59 sebhub, I have a new mount command that uses a string for the type of file system. 2010-05-31T08:39:19 This means bpart can have a table of type numbers are file systems. 2010-05-31T08:40:28 oh, I did not look at the mount PR yet, I am very busy with a boot loader at the moment 2010-05-31T08:41:12 Sure. It also has a void* data value so we can pass file system specific options to the file system. 2010-05-31T08:41:34 We do not need to export ops tables. 2010-05-31T08:48:48 So DrJoel, considering that librtemscpu is in a different location with multilib enabled, do we change the test scripts, or decide to put a particular lib inside $bsp ? 2010-05-31T08:50:20 Raj_: the point of building multilib, then BSP when testing GCC is that the .h files and main RTEMS library end up in a standard location GCC looks. The BSP library is only needed when you link tests. This is how I build and run the Ada tests so I know it works. The problem was that your mulitilib install was out of sync with your BSP install 2010-05-31T08:54:46 DrJoel: Oh k, so you mean I just do the install twice, once with multilib, and then without ? 2010-05-31T08:55:43 yes.. the first is the "OS",,, the second is the BSP.. they are cooperative and need to be in sync. 2010-05-31T08:56:09 kiwichris: I don't even know that Ralf knows I do all the gcc testing this way.. cpukit + BSPs works great for building and testing tools 2010-05-31T08:56:56 He may not. I do not know. 2010-05-31T08:57:21 DrJoel: ok. Got it ! 2010-05-31T09:01:03 *** Raj_ has quit IRC 2010-05-31T09:05:36 when is the irc meeting planned for? 2010-05-31T09:05:52 http://www.timeanddate.com/worldclock/meetingtime.html?month=5&day=31&year=2010&p1=0&p2=240&p3=64&p4=-1 2010-05-31T09:05:53 It will start early your time. 2010-05-31T09:06:00 tomorrow? 2010-05-31T09:06:05 Yes 2010-05-31T09:06:31 sebhub send a email, tomorrow 12:00 UTC? 2010-05-31T09:07:06 include a link to timeanddate.com .. they have a meeting countdown page that the gsoc mentor meetings always reference 2010-05-31T09:07:09 let me find it .. hold on 2010-05-31T09:08:12 cya 2010-05-31T09:08:13 *** kiwichris has quit IRC 2010-05-31T09:08:40 http://www.timeanddate.com/counters/customcounter.html?month=06&day=1&year=2010&hour=12&min=0&sec=0&p0=0 2010-05-31T09:19:40 *** exception13 has joined #rtems 2010-05-31T09:40:29 *** lekernel has joined #rtems 2010-05-31T10:27:18 *** wenjie has quit IRC 2010-05-31T10:44:02 *** sebhub has quit IRC 2010-05-31T10:46:03 *** lekernel has quit IRC 2010-05-31T11:30:20 *** dr__house has joined #rtems 2010-05-31T12:18:29 *** krzabr has joined #rtems 2010-05-31T12:38:20 Do you know articles which decribes rtems in practical using ? 2010-05-31T12:55:09 *** krzabr has quit IRC 2010-05-31T13:24:04 *** krzabr has joined #rtems 2010-05-31T14:09:19 DrJoel: good afternoon 2010-05-31T14:25:25 *** aniceberg has joined #rtems 2010-05-31T14:31:50 dr__house: hello .. I see you got the owl picture 2010-05-31T14:32:14 DrJoel: yes :) 2010-05-31T14:36:36 How is school going? 2010-05-31T14:37:24 DrJoel: me? 5 rejects, no admits yet. Awaiting 3 :( 2010-05-31T14:37:47 I hope I have turned in all the recommendations you asked for 2010-05-31T14:38:18 DrJoel: yes, thank you for them :) 2010-05-31T14:38:36 thanks but I wasn't fishing.. just making sure I hadn't dropped the ball on anything 2010-05-31T14:39:28 DrJoel: didn't get you 2010-05-31T14:39:57 I wasn't looking for a pat on the back.. just making sure I hadn't let you down 2010-05-31T14:43:03 DrJoel: Oh, I understand :) 2010-05-31T14:47:48 *** aniceberg has quit IRC 2010-05-31T15:04:43 DrJoel: leaving for the day. See you tomorrow. Have a good day :) 2010-05-31T15:04:52 *** dr__house has quit IRC 2010-05-31T15:07:20 *** lekernel has joined #rtems 2010-05-31T15:24:28 *** exception13 has left #rtems 2010-05-31T15:29:12 *** krzabr has quit IRC 2010-05-31T15:35:13 *** krzabr has joined #rtems 2010-05-31T16:19:44 *** lekernel has quit IRC 2010-05-31T17:26:26 *** kiwichris has joined #rtems 2010-05-31T17:26:45 *** krzabr has quit IRC 2010-05-31T17:34:01 DrJoel, hi 2010-05-31T18:01:46 kiwichris: hello.. 2010-05-31T18:18:01 Just looking at the tests now 2010-05-31T18:19:32 The C++ example is also broken 2010-05-31T18:25:51 It seems the C++ test uses CONFIGURE_USE_MINIIMFS_AS_BASE_FILESYSTEM 2010-05-31T18:31:46 :-D That should work, shouldn't it? 2010-05-31T18:32:19 Yes but someone who shall remain nameless committed confdefs.h changes yesterday :) 2010-05-31T18:32:40 I didn't spot that but it has been a nice long weekend 2010-05-31T18:32:45 what was it? 2010-05-31T18:33:17 * DrJoel heads to check on the food on the grill 2010-05-31T18:33:45 I handles devfs as a single FS and not miniIMFS. 2010-05-31T18:36:11 ahhh.. makes sense. 2010-05-31T18:36:25 well just commit what you need to. 2010-05-31T18:36:37 burgers and chicken just off the grill .. time for dinner 2010-05-31T18:36:43 I will get these going. 2010-05-31T18:38:50 * kiwichris head out for a while 2010-05-31T18:39:07 *** kiwichris has quit IRC 2010-05-31T20:37:56 *** kiwichris has joined #rtems 2010-06-01T00:45:45 morning 2010-06-01T01:11:30 *** dr__house has joined #rtems 2010-06-01T01:47:43 *** exception13 has joined #rtems 2010-06-01T01:50:13 morning 2010-06-01T02:01:08 *** sebhub has joined #rtems 2010-06-01T02:01:14 good morning 2010-06-01T02:11:43 sebhub, hi 2010-06-01T02:11:53 sebhub, does the iostream test link for you ? 2010-06-01T02:12:10 with cvs head? 2010-06-01T02:12:23 Yes but before my latest updates ? 2010-06-01T02:12:32 one moment 2010-06-01T02:16:40 *** dr__house has quit IRC 2010-06-01T02:19:45 links and runs in sis 2010-06-01T02:19:59 Hmmm. 2010-06-01T02:20:16 Can you please do me a favor ? 2010-06-01T02:20:37 yes 2010-06-01T02:20:58 Can you please nm the init.o file into a text file then again with --demangle and please send me the files ? 2010-06-01T02:21:13 the init.o in the iostream test 2010-06-01T02:21:43 I think a hack may exist in confdefs.h and I am looking into why it exists. 2010-06-01T02:24:44 ok, mail is on the way 2010-06-01T02:24:53 I thought 'extern "C" { ..... };" kept everything as C bindings. 2010-06-01T02:25:16 *** dr__house has joined #rtems 2010-06-01T02:26:37 If you take a look you can see configuration_filesystem_table is being mangled. So is Init and rtems_libio_init_helper 2010-06-01T02:26:38 . 2010-06-01T02:27:04 These are all inside extern C 2010-06-01T02:27:28 Why does m68k and i386 fail. I do not understand. 2010-06-01T02:28:44 I see why it is working. 2010-06-01T02:28:55 It is the work around. 2010-06-01T02:57:20 *** dr__house` has joined #rtems 2010-06-01T02:57:20 *** dr__house has quit IRC 2010-06-01T03:26:38 *** dr__house` has quit IRC 2010-06-01T05:20:52 *** lcpfnvc has quit IRC 2010-06-01T06:41:29 sebhub, are you about ? 2010-06-01T06:55:54 yes 2010-06-01T06:56:24 Do you know why things like Initialization_tasks are not const ? 2010-06-01T06:56:55 I guess these parts are very old and during this time const was not so common 2010-06-01T06:56:56 or Device_drivers 2010-06-01T06:57:19 I think we should look at these config tables becoming const. 2010-06-01T06:57:45 yes, we should always make things const whenever possible 2010-06-01T06:58:02 I still not find why iostream does not build 2010-06-01T06:58:44 on cvs head? 2010-06-01T06:59:06 Yeah 2010-06-01T07:01:32 *** dr__house has joined #rtems 2010-06-01T07:07:07 kiwichris: sebhub is right. those are very old. const was not a keyword back when that was implemented and you started submitting. 2010-06-01T07:07:30 We could save lots of RAM with them const. 2010-06-01T07:07:59 Joel, iostream has broken with my mount change and I cannot see how. 2010-06-01T07:08:12 build or run? which target? 2010-06-01T07:08:22 i386 and m68k 2010-06-01T07:08:24 Build. 2010-06-01T07:08:35 It is clashing with dummy.o 2010-06-01T07:09:04 That means a symbol is not being generated by confdefs and having to be satisfied by dummy.o. Turn on map from linker 2010-06-01T07:09:21 It is configuration_filesystem_table 2010-06-01T07:09:33 It is being declared as C++, or mangled. 2010-06-01T07:09:47 ahhh.. have you looked at the preprocessed output? 2010-06-01T07:09:51 The configuration_filesystem_mount_table is the same 2010-06-01T07:10:01 Yes and it looks fine 2010-06-01T07:10:06 did the extern C stay in place 2010-06-01T07:10:14 an extra } perhaps? 2010-06-01T07:10:25 There are so many extern C and they look ok. 2010-06-01T07:10:36 Maybe. 2010-06-01T07:10:44 But nothing else breaks. 2010-06-01T07:11:27 I do not get the configuration_filesystem_table as a symbol in the init.o 2010-06-01T07:11:31 brb garbage day 2010-06-01T07:16:56 The .s file has: 2010-06-01T07:17:12 .section .rodata 2010-06-01T07:17:12 .align 2 2010-06-01T07:17:12 .type configuration_filesystem_table, @object 2010-06-01T07:17:12 .size configuration_filesystem_table, 16 2010-06-01T07:17:12 configuration_filesystem_table: 2010-06-01T07:17:12 .long .LC3 2010-06-01T07:17:14 .long miniIMFS_initialize 2010-06-01T07:17:16 .long 0 2010-06-01T07:17:18 .long 0 2010-06-01T07:17:58 const is static by default in C++? 2010-06-01T07:18:29 It is in extern C 2010-06-01T07:18:55 It has appeared as a C linkage but there must be more. 2010-06-01T07:19:04 i am not sure, but const in C is something else than const in C++ 2010-06-01T07:19:25 I know static is different 2010-06-01T07:20:44 I suspect an extern "C" is missing. 2010-06-01T07:21:16 The iostream has miniIMFS as base fs but IMFS is being linked in. 2010-06-01T07:21:23 A second problem 2010-06-01T07:22:34 If you look in the top of confsdef.h a couple of headers are included outside of the extern C. I have wrapped the whole include of confsdef.h in extern C and the configuration_filesyste_table has appeared. 2010-06-01T07:25:11 heading to office .. be back on in a bit 2010-06-01T07:25:15 *** DrJoel has quit IRC 2010-06-01T07:37:45 it seems like the extern "C" is missing. adding it by hand in init.ii solves the problem 2010-06-01T07:38:48 Yes. Plus you need an "extern const rtems_filesystem_table_t ...." like at the top of confdef.h to stop the optimizer removing the code 2010-06-01T07:38:49 hm, no 2010-06-01T07:39:00 it is not the extern "C" 2010-06-01T07:39:07 http://learningcppisfun.blogspot.com/2008/02/const-in-c-and-c.html 2010-06-01T07:39:09 Oh ? 2010-06-01T07:39:41 adding an extern const is sufficient 2010-06-01T07:40:17 We have the code under extern C so this should not apply 2010-06-01T07:41:04 does extern "C" really change the semantics of keywords? 2010-06-01T07:41:11 I have others. miniIMFS does not remove things like the fifo. 2010-06-01T07:41:25 Well I have the symbol no demangled 2010-06-01T07:41:40 yes, i fixed this miniimfs problem locally 2010-06-01T07:41:49 i want to wait until 4.10 branch is out 2010-06-01T07:41:57 because this was a big change 2010-06-01T07:42:08 $ m68k-rtems4.10-nm init.o | grep configuration_filesystem_table00000000 R configuration_filesystem_table 2010-06-01T07:42:21 Oh the fact it is not that mini > 2010-06-01T07:42:26 Oh the fact it is not that mini ? 2010-06-01T07:43:12 Index: ./cpukit/sapi/include/confdefs.h 2010-06-01T07:43:13 =================================================================== 2010-06-01T07:43:15 RCS file: /usr1/CVS/rtems/cpukit/sapi/include/confdefs.h,v 2010-06-01T07:43:16 retrieving revision 1.133 2010-06-01T07:43:18 diff -u -r1.133 confdefs.h 2010-06-01T07:43:19 --- ./cpukit/sapi/include/confdefs.h 31 May 2010 13:56:37 -0000 1.133 2010-06-01T07:43:21 +++ ./cpukit/sapi/include/confdefs.h 1 Jun 2010 12:43:03 -0000 2010-06-01T07:43:22 @@ -385,7 +385,7 @@ 2010-06-01T07:43:22 sebhub: Error: "@" is not a valid command. 2010-06-01T07:43:24 * you provide your own. 2010-06-01T07:43:25 */ 2010-06-01T07:43:27 #ifndef CONFIGURE_HAS_OWN_FILESYSTEM_TABLE 2010-06-01T07:43:28 - const rtems_filesystem_table_t configuration_filesystem_table[] = { 2010-06-01T07:43:30 + extern const rtems_filesystem_table_t configuration_filesystem_table[] = { 2010-06-01T07:43:31 #if defined(CONFIGURE_FILESYSTEM_miniIMFS) && \ 2010-06-01T07:43:33 defined(CONFIGURE_FILESYSTEM_ENTRY_miniIMFS) 2010-06-01T07:43:34 CONFIGURE_FILESYSTEM_ENTRY_miniIMFS, 2010-06-01T07:43:36 can you try this? 2010-06-01T07:44:02 What does that do ? 2010-06-01T07:44:21 one moment i send you an email 2010-06-01T07:44:32 I have: 2010-06-01T07:44:34 extern const rtems_filesystem_table_t configuration_filesystem_table[]; 2010-06-01T07:44:44 const rtems_filesystem_table_t configuration_filesystem_table[] = { 2010-06-01T07:44:54 On separate lines 2010-06-01T07:47:51 *** DrJoel has joined #rtems 2010-06-01T07:47:51 *** ChanServ sets mode: +o DrJoel 2010-06-01T07:48:08 Hi Joel@work 2010-06-01T07:48:29 :) 2010-06-01T07:48:37 any progress? 2010-06-01T07:48:50 Yeap. 2010-06-01T07:49:02 IMFS_Set_handlers and miniIMFS do not play together 2010-06-01T07:49:28 :( 2010-06-01T07:49:43 I think we have a extern C issue which I will locate later 2010-06-01T07:49:49 It pulls lots in 2010-06-01T07:50:17 i already fixed the miniimfs regarding the pipe/fifo dependency 2010-06-01T07:50:37 Honestly it may be possible to deprecate miniIMFS in favor of devfs. miniIMFS may be slightly more posix compliant though 2010-06-01T07:50:41 Any patch handy ? 2010-06-01T07:51:05 If it is used then we should keep for a release 2010-06-01T07:51:07 no, i wanted to wait after the 4.10 branch 2010-06-01T07:51:22 Big fix ? 2010-06-01T07:51:27 it is a big patch 2010-06-01T07:51:44 Ok. 2010-06-01T07:51:59 quite frankly, the pipe support has several questionable things 2010-06-01T07:52:23 Fixable or not ? 2010-06-01T07:52:36 it is all software ;-) 2010-06-01T07:52:42 sebhub: that area is being worked on now by a GSOC student. there were not nearly enough tests and it is one of the biggest uncovered areas 2010-06-01T07:53:08 hm, i have a lot of imfs changes locally 2010-06-01T07:53:18 mostly cleanup of copy&paste code 2010-06-01T07:53:20 Are we all here for the seq chat ? 2010-06-01T07:53:34 I don't see the student LOL 2010-06-01T07:53:52 Ok 2010-06-01T07:54:00 he wrote an email and sayed he is delayed 2010-06-01T07:54:34 Ok. 2nd half of the Aussie and Denmark friendly for the World Cup to come. 2010-06-01T07:56:35 when can we expect 4.10? 2010-06-01T07:58:24 Any day now from my perspective .. but Chris has this to fix and Ralf had issues when I last checked 2010-06-01T07:58:50 I would like to fix gethostbyname_r or at least have a go at it. 2010-06-01T07:59:06 I have it building. 2010-06-01T08:00:44 *** zwj has joined #rtems 2010-06-01T08:01:03 Hi, DrJoel,chris , sebhub 2010-06-01T08:01:14 i am so sorry for the late 2010-06-01T08:01:16 hi 2010-06-01T08:01:35 zwj, hi 2010-06-01T08:02:05 We can start the chat about sequenced initialisation now ? 2010-06-01T08:02:19 ok 2010-06-01T08:02:21 yes 2010-06-01T08:02:53 What is the current status of the design ? 2010-06-01T08:03:26 we have no design yet 2010-06-01T08:03:30 i have made a patch based on the "RTEMS Implement" 2010-06-01T08:03:42 only the proposals on the wiki page with the different approaches 2010-06-01T08:03:55 yes 2010-06-01T08:04:05 The wiki page has some approaches and they are all good. 2010-06-01T08:04:36 yes, now we should agree on one of them (or maybe another one) 2010-06-01T08:05:11 The sort from the linker is a neat trick. What would be needed for a runtime solution ? 2010-06-01T08:05:37 That is not to use the linker to sort. 2010-06-01T08:05:44 the runtime solution needs space to store the order information and the code to sort the items 2010-06-01T08:06:09 Is this data or code ? 2010-06-01T08:06:18 both 2010-06-01T08:06:32 I see code but the data I do not see. 2010-06-01T08:07:02 you have an unorder set of items, thus each item has to carry its order information 2010-06-01T08:07:30 in term of the runtime solution, i think the code needed by squence is not so large, so it is something a little needs space 2010-06-01T08:07:43 Sure, and performance is not a big issue. Can you loop the table for each number ? 2010-06-01T08:08:12 This way sort and execute at the same time ? 2010-06-01T08:08:33 yes, i have test the patch based on the sort approch, it is ok 2010-06-01T08:08:41 Any data used ? 2010-06-01T08:08:50 sort in link time and execute in run time 2010-06-01T08:09:08 what is the problem with sort by linker? 2010-06-01T08:09:43 The linker sort is concerns me. it is a feature we then rely on. It is not standard. 2010-06-01T08:09:56 not use data, just using code _RTEMS_API_Initilzation 2010-06-01T08:10:22 you can sort the sections by hand 2010-06-01T08:10:42 Not sure I follow. 2010-06-01T08:11:01 Linux does it that way 2010-06-01T08:11:11 acturely , linux also uses the sort link 2010-06-01T08:11:14 yes 2010-06-01T08:11:27 Hmm. 2010-06-01T08:11:49 * kiwichris ponders this 2010-06-01T08:11:55 we can define some subsystem order to store the link set 2010-06-01T08:12:04 currently all linker command files in cvs are based on the GNU ld 2010-06-01T08:12:29 Yes I know and this is a good point. 2010-06-01T08:13:09 we should try to make each approach possible from a high level perspective 2010-06-01T08:13:14 On to another point. Lets just say we have a way to call the handlers in order. 2010-06-01T08:13:25 but currently i would exploit the GNU ld sort feature 2010-06-01T08:15:13 if the code needed to init is not so large, we can also consider using hard linker like linux 2010-06-01T08:15:17 Sure. I will ponder a little more and if I see anything I will let you know. 2010-06-01T08:15:30 What is a hard linker ? 2010-06-01T08:15:59 no, the linux approach is garbadge 2010-06-01T08:16:18 define some order link section, such as init.1s, init2.s 2010-06-01T08:16:34 Oh not please lets not go there. 2010-06-01T08:16:40 and in link scripts hard code the order 2010-06-01T08:16:43 this only bloats the linker command files 2010-06-01T08:16:50 and is maintainance nightmare 2010-06-01T08:17:00 This one is a definite no from me. 2010-06-01T08:17:12 and introduces the possibility of breaking RTEMS initialization on a per BSP basis. I already hate that a BSP can break C++ constructors 2010-06-01T08:17:34 wow we were quick with the hammer. Rare in RTEMS land. :-D 2010-06-01T08:18:03 My understanding is the order is numeric with cpp tricks to create labels that are sorted. Is this correct ? 2010-06-01T08:18:06 yeah, it is really hard to maintainance 2010-06-01T08:18:23 yes 2010-06-01T08:18:27 * DrJoel has to head to a customer site.. bye 2010-06-01T08:18:36 will be logged in so i can read this when I get back 2010-06-01T08:18:37 its a lexicographical sort 2010-06-01T08:19:27 So we can define domains of ordering for want of a better term. 2010-06-01T08:20:02 For example "_seq_0preRTEMS" 2010-06-01T08:20:13 yes, we can also specifiy "before" and "after" 2010-06-01T08:20:16 or "_seq_000preRTEMS" 2010-06-01T08:20:38 yes 2010-06-01T08:20:43 This means we could then allow items to be listed in this. 2010-06-01T08:21:00 We could even have preinit and init calls placed in the list. 2010-06-01T08:21:26 _seq_010preInitRTEMS ? 2010-06-01T08:21:33 Does this make sense ? 2010-06-01T08:21:47 like this __initcall_start = .; 2010-06-01T08:21:48 *(.initcall0.init) 2010-06-01T08:21:48 *(.initcall0s.init) 2010-06-01T08:21:48 *(.initcall1.init) 2010-06-01T08:21:48 *(.initcall1s.init) 2010-06-01T08:21:48 *(.initcall2.init) 2010-06-01T08:21:48 *(.initcall2s.init) 2010-06-01T08:21:49 *(.initcall3.init) 2010-06-01T08:21:50 *(.initcall3s.init) 2010-06-01T08:21:50 *(.initcall4.init) 2010-06-01T08:21:51 *(.initcall4s.init) 2010-06-01T08:21:51 *(.initcall5.init) 2010-06-01T08:21:51 *(.initcall5s.init) 2010-06-01T08:21:52 *(.initcallrootfs.init) 2010-06-01T08:21:53 *(.initcall6.init) 2010-06-01T08:21:53 *(.initcall6s.init) 2010-06-01T08:22:37 I do not understand this. Think about it as a simpler type of namespacing. 2010-06-01T08:22:40 this is linux links, we just replace it with "sort(.initcall*)" 2010-06-01T08:23:24 Can we use something like _seq so greps of symbols will not mix with C++ and others. 2010-06-01T08:24:07 these symbols are local symbols 2010-06-01T08:24:29 ? 2010-06-01T08:25:12 you have linker sets which are input sections of the linker 2010-06-01T08:25:50 Does the linker sort sections or symbols ? 2010-06-01T08:25:52 and you have items for these linker sets, these are local to a translation unit 2010-06-01T08:25:59 it sorts sections 2010-06-01T08:26:19 this is my patch, #define RTEMS_SYSINIT_CORE(order, handler) \ 2010-06-01T08:26:19 sysinit_core rtems_sysinit_ ## handler ## _core = { \ 2010-06-01T08:26:19 order, \ 2010-06-01T08:26:19 handler, \ 2010-06-01T08:26:19 }; 2010-06-01T08:26:19 #define RTEMS_SYSINIT_REF(order, handler) \ 2010-06-01T08:26:19 extern sysinit_core rtems_sysinit_ ## handler ## _core; \ 2010-06-01T08:26:21 static const *const rtems_sysinit_ ## handler ## _ref \ 2010-06-01T08:26:21 __attribute__((section(".rtems.sysinit." #order))) \ 2010-06-01T08:26:22 __attribute__((used)) \ 2010-06-01T08:26:22 = &rtems_sysinit_ ## handler ## _core; 2010-06-01T08:26:23 in linkcmd files 2010-06-01T08:26:23 * Special RTEMS Kernel constructor sections. 2010-06-01T08:26:24 */ 2010-06-01T08:26:25 . = ALIGN (16); 2010-06-01T08:26:25 rtems_sysinit_begin = .; 2010-06-01T08:26:26 http://www.rtems.com/wiki/index.php/RTEMSSequencedInitialization#RTEMS_Approach 2010-06-01T08:28:12 Yes so all I am wondering about is "RTEMS_SYSINIT_CORE(group, order, handler)" 2010-06-01T08:28:12 but there is some problem about the reference of object 2010-06-01T08:28:42 is add a group so we can manage the list a little better. 2010-06-01T08:29:13 Does "2" come before "20" ? 2010-06-01T08:29:41 ie do we need some rules ? 2010-06-01T08:29:49 yeah, it is a really problem. we should not all use num to sort 2010-06-01T08:30:05 *** exception13 has left #rtems 2010-06-01T08:30:11 for the groups we should use preprocessor defines 2010-06-01T08:30:27 Yes and they are controlled in the core. 2010-06-01T08:30:27 and we can use "0002', "0020", etc 2010-06-01T08:30:32 Great. 2010-06-01T08:31:01 This resolves the "before", "after" etc issue I had. 2010-06-01T08:31:43 yes, this not a problem, but things like "before A and after B" is not possible 2010-06-01T08:32:17 Not to the level I defined, but groups allows us to know drivers before tasks etc 2010-06-01T08:32:29 And this I think will be fine. 2010-06-01T08:32:41 freebsd define some subsystem ,we can also use this 2010-06-01T08:32:55 ok 2010-06-01T08:33:05 we are not limited in the number of domains 2010-06-01T08:33:43 Yes and if we space them and define them users can fit in around these groups within a BSP. 2010-06-01T08:34:06 For example in a group we define 0->20 for users and 100-?? for them 2010-06-01T08:34:58 yes 2010-06-01T08:35:05 It is almost full time in the soccer so I might head off. 2010-06-01T08:35:29 ok, we should define now the api on the wiki page 2010-06-01T08:35:38 we can also use the discussion page on the wiki 2010-06-01T08:35:56 yeah, but i also have a problem about how to reference the object 2010-06-01T08:35:58 I think so. I would also drop all the other ideas to the bottom or a different page. 2010-06-01T08:37:11 *** tiemen has quit IRC 2010-06-01T08:37:42 for example. about event there is no event_creat like function, so whether we should reference it in two files(eventsend.c and eventrevice.c) 2010-06-01T08:38:16 No need for this one. If the user does not call it we do not need to worry about it. 2010-06-01T08:38:32 There is not entry needed for this one. 2010-06-01T08:38:48 \/not/no/ 2010-06-01T08:39:18 but if the user want to use event, how we ensure link the event_init function into the bin 2010-06-01T08:39:53 What is event_init ? 2010-06-01T08:40:19 _Event_Manager_initialization 2010-06-01T08:40:38 you only call rtems_event_{send, receive} 2010-06-01T08:40:53 It is empty 2010-06-01T08:41:07 It should be removed 2010-06-01T08:41:38 _Event_Manager_initialization is not empty 2010-06-01T08:41:58 Hmmm got the no-event from tags 2010-06-01T08:41:59 but if we only call rtems_event_send, how we ensure link _Event_Manager_initialization into bin? 2010-06-01T08:42:19 Add to both. 2010-06-01T08:42:26 can we make this with a weak symbol reference? 2010-06-01T08:42:42 This was handled in the original patch. 2010-06-01T08:43:05 You have the node in the file you want to pull in and the references in the other files. 2010-06-01T08:43:36 if we add to both, there is a variable is not used, so it is waste space 2010-06-01T08:44:07 Only add a reference. 2010-06-01T08:44:31 i suppose this variable will vanish because it is unused? 2010-06-01T08:44:39 on event and signal is need to consider 2010-06-01T08:45:24 others there is only one to add to xxxcreat.c 2010-06-01T08:45:26 *** exception13 has joined #rtems 2010-06-01T08:46:09 ok, but before we start to change this, the new sysinit system must be implemented and tested 2010-06-01T08:46:15 it will not vanish , i test it 2010-06-01T08:46:39 Yes. I will also look at the original patch and figure out this issue and send an email. 2010-06-01T08:48:47 yeah, we can add the reference to the function. like we just add some thing like "if reference_variable == NULL ... else ..." 2010-06-01T08:49:56 that if we just call rtems_event_send or rtems_event_recevie it will ensure link into the bin 2010-06-01T09:14:00 *** bubaflub has joined #rtems 2010-06-01T09:15:44 *** kiwichris has quit IRC 2010-06-01T09:17:59 *** exception13 has left #rtems 2010-06-01T10:32:21 *** bubaflub has quit IRC 2010-06-01T10:38:36 *** bubaflub has joined #rtems 2010-06-01T10:42:41 *** sebhub has quit IRC 2010-06-01T10:53:17 bubaflub: how are things going? 2010-06-01T10:53:25 DrJoel: pretty great! 2010-06-01T10:53:30 i've got some good progress for tonight's meeting 2010-06-01T10:53:42 i found out i won't have to patch some parts of parrot 2010-06-01T10:54:11 so i've got a makefile that loads in all of the RTEMS makefile / variables 2010-06-01T10:54:44 and spits out two files - one telling the configure script what the variables are (like where CC is etc) and what steps to run/skip 2010-06-01T10:55:12 the other, a platform hints file that contains information like how big a long is etc. 2010-06-01T10:55:35 so i just need to now pull the pertinent information and get parrot to compile 2010-06-01T10:55:52 i hoped to have it compiling by today but got sidetracked a bit this weekend 2010-06-01T10:57:58 cool! that should go a long way to compiling it for every bsp 2010-06-01T10:58:10 yeah 2010-06-01T10:58:18 so now i just need to figure out which variables need to be included in the makefile 2010-06-01T10:58:29 and kiwichris has an old manually edited makefile with all the variables 2010-06-01T10:58:44 that should help a lot. 2010-06-01T10:58:46 so right now i'm just setting the variables, and trying to configure Parrot and seeing where things fiail 2010-06-01T10:58:48 yeah, definitely 2010-06-01T11:08:23 *** lekernel has joined #rtems 2010-06-01T11:26:50 *** zwj has quit IRC 2010-06-01T12:36:29 *** gedare has joined #rtems 2010-06-01T12:37:51 *** lekernel has quit IRC 2010-06-01T13:14:07 *** gedare has quit IRC 2010-06-01T13:21:04 *** tuxmaniac has joined #rtems 2010-06-01T13:21:24 \0/ 2010-06-01T13:21:53 DrJoel: has rtems for OpenRISC platform been tried before? I see a page in opencores http://opencores.org/openrisc,rtems 2010-06-01T13:22:01 any idea of the status? 2010-06-01T13:25:14 It was included in old releases but the tools never got merged into GCC back then and it withered. According to http://wiki.rtems.org/wiki/index.php/SupportedCPUs, it was in 4.6 but not in 4.7 2010-06-01T13:25:20 Is it in gcc, binutils, etc now? 2010-06-01T13:25:39 It isn't hard to revive if the tools are OK now and there is a way to test it 2010-06-01T13:27:24 by tools you mean -gcc -gdb etc right? 2010-06-01T13:27:40 because I have a project request for porting a tiny OS into OpenRISC 2010-06-01T13:27:59 DrJoel: and guess the real-time OS that came to my mind first? 2010-06-01T13:28:38 :) 2010-06-01T13:28:56 Shouldn't be too bad if the tools exist 2010-06-01T13:29:30 ok i will check on that. But definitely I have or32-gcc etc.. Not sure if its from main tree 2010-06-01T13:29:44 and the gcc version 2010-06-01T13:30:15 gcc 3.4.4 2010-06-01T13:30:18 just checked gcc-svn and it isn't there. 2010-06-01T13:30:44 yep.. now you see the problem for us from the tool view 2010-06-01T13:31:15 but you shuld be able to use that to build an or32-rtems-gcc, etc with the current newlib and their target code. Then update and fix the old port 2010-06-01T13:48:28 *** Raj_ has joined #rtems 2010-06-01T13:50:20 Hi DrJoel 2010-06-01T14:05:31 *** Raj_ has quit IRC 2010-06-01T14:08:05 *** dr__house has quit IRC 2010-06-01T14:33:24 *** tuxmaniac has quit IRC 2010-06-01T14:36:48 bubaflub: are you there? 2010-06-01T14:36:55 DrJoel: sure am 2010-06-01T14:37:10 what's up? 2010-06-01T14:37:22 I want to verify the web setup is ok for the class next week.. give me a minute and then goto https://my.dimdim.com/Joel.Sherrill 2010-06-01T14:37:36 *** krzabr has joined #rtems 2010-06-01T14:38:10 hold on .. phone call 2010-06-01T14:38:38 roger that. 2010-06-01T14:39:41 ok.. try now 2010-06-01T14:41:49 DrJoel: i'm in 2010-06-01T14:42:23 what do you see? 2010-06-01T14:42:28 can you hear me? 2010-06-01T14:42:35 i see a webcam 2010-06-01T14:42:37 you were in it for a bit 2010-06-01T14:42:39 i see the power point 2010-06-01T14:42:44 i don't hear anything, lemme check my settings 2010-06-01T14:42:47 pointing at my office wall and ceiling 2010-06-01T14:42:57 yep 2010-06-01T14:43:09 lemme switch from Safari to FireFox, i can try it in there 2010-06-01T14:43:22 ok.. that is what we will do for the class 2010-06-01T14:43:37 I watched last year using FireFox on Fedora so should be ok 2010-06-01T14:45:14 DrJoel: looks good on Firefox on Mac 2010-06-01T14:45:21 my safari has a flash blocker so i'll just stick with FF 2010-06-01T14:45:29 ok.. so I am shutting this down. 2010-06-01T14:45:35 ok 2010-06-01T14:45:54 bigger problem is that our double-sided printer/copier is in the shop so I am going to have to take my class material to a copy shop :( 2010-06-01T14:46:31 not bad for free. :) 2010-06-01T14:46:42 bummer. i know kinkos / fedex has a nice document submission thingy online 2010-06-01T14:47:11 my roommate had to print out some course materials packet (about 200 - 300 pages) and he could just upload the .doc and then lay it out through the browser 2010-06-01T14:49:06 that's what I am going to do.. just have gotten spoiled about doing it in house. 2010-06-01T14:49:45 *** krzabr has left #rtems 2010-06-01T14:50:38 ah, yeah. our printer crapped out a few months ago 2010-06-01T14:50:44 right around when some of my papers were due 2010-06-01T14:51:23 printers always die, run out of toner, or jam at the worst times 2010-06-01T15:37:43 *** exception13 has joined #rtems 2010-06-01T16:38:48 *** lekernel has joined #rtems 2010-06-01T16:43:40 lekernel: any news from your student? 2010-06-01T16:44:03 I guess that applies to mwalle as well 2010-06-01T16:45:54 yeah 2010-06-01T16:45:56 don't worry 2010-06-01T16:45:57 he had exams lately 2010-06-01T16:46:09 he was busy with exams etc. recently 2010-06-01T16:46:41 a lot of the students are having them. Just haven't heard from him and was curious 2010-06-01T16:46:48 http://sionneau.net/index.php http://github.com/fallen/rtems-milkymist 2010-06-01T16:51:04 looks good. :) 2010-06-01T16:52:21 he's on this channel btw (Fallenou) 2010-06-01T16:53:03 gn8 all 2010-06-01T16:53:13 *** lekernel has quit IRC 2010-06-01T16:54:19 Fallenou: how did I miss you?!? Blog looks like you are making progress. :) 2010-06-01T17:10:46 DrJoel: I'm decommuting in a bit, I'll catch up with you, dukeleto, and kiwichris in an hour or two 2010-06-01T17:10:53 *** bubaflub has quit IRC 2010-06-01T17:16:38 *** gedare has joined #rtems 2010-06-01T17:26:54 *** kiwichris has joined #rtems 2010-06-01T17:34:51 i'm having trouble building off head, testsuites/samples/fileio/init.c:48: error: 'msdos_ops' undeclared here (not in a function) 2010-06-01T17:36:05 gedare, I am fixing these. My mount change broke the tests. 2010-06-01T17:36:22 oh, ok. thanks, I'll try again later. :) 2010-06-01T17:36:25 I am committing the fixes now but CVS is very very slow 2010-06-01T17:36:43 hmmm.. wonder why 2010-06-01T17:36:58 Hi Joel. Cause it is CVS. 2010-06-01T17:37:03 hah 2010-06-01T17:37:20 bazinga 2010-06-01T17:37:37 kiwichris: after a couple of rounds of building and a coverage run, I think I will branch 2010-06-01T17:38:19 gedare: welcome back.. I noticed qemu has sun4v support. wonder if it works.. but we have pressing scheduler matters at hand 2010-06-01T17:39:08 DrJoel: sun4v is a surprise to me. I was only aware of some older sparc64 support, but yes, I'm planning to re-focus my efforts, might get to even code a little tomorrow 2010-06-01T17:39:13 working on infrastructure right now 2010-06-01T17:40:08 I am trying to work through mods to have the APs go to their idle thread, then make them go to a user task .. proves the mechanics of the IPI requests work 2010-06-01T17:40:09 Joel, had an email fro Claudio and he has finished his school work and back. I will arrange an IRC meeting. 2010-06-01T17:40:17 great! 2010-06-01T17:40:27 * DrJoel is ready to start merging post branching :-D 2010-06-01T17:40:34 DrJoel, yes I agree. 2010-06-01T17:40:50 about the branching. 2010-06-01T17:40:57 gedare: you should be near the top of the list. A new port is relatively straight forward to merge and not likely to be controversial 2010-06-01T17:41:10 Then we have my SMP changes, USB, and LWIP waiting in the wings 2010-06-01T17:41:28 How will the USB changes be merged in ? 2010-06-01T17:41:45 Sounds good. Now that I have sun4u support and M5 simulator "working" for sun4v, I feel pretty good about the port. 2010-06-01T17:42:12 kiwichris: I don't know.. we will have to talk with Thomas and Sebastian about them. 2010-06-01T17:42:24 gedare: part of the merge is getting me to run it. :) 2010-06-01T17:42:36 Great. The post commit method is not suitable IMO 2010-06-01T17:42:46 kiwichris: OK. 2010-06-01T17:46:03 *** bubaflub has joined #rtems 2010-06-01T17:49:43 sounds good, time for me to go though... will catch up soon 2010-06-01T17:49:56 *** gedare has quit IRC 2010-06-01T17:51:04 DrJoel, I will commit the work around for condefs.h until I find a solution. 2010-06-01T17:51:19 This is the need for extern C in the iostream test 2010-06-01T17:51:30 good $localtime, kiwichris 2010-06-01T17:51:33 ok. 2010-06-01T17:51:38 * DrJoel waves good night.. the rain has stopped .. time to go home 2010-06-01T17:51:42 *** DrJoel has quit IRC 2010-06-01T17:52:08 bubaflub, yeah it is morning. 2010-06-01T18:18:06 *** exception13 has left #rtems 2010-06-01T18:37:52 arg i missed DrJoel 2010-06-01T18:37:54 damn 2010-06-01T19:01:22 kiwichris: ready for a quick rundown of my progress this week? 2010-06-01T19:01:36 Yes 2010-06-01T19:01:40 sweet 2010-06-01T19:01:44 so, i might have told you this earlier 2010-06-01T19:01:52 but i found out we won't need to patch the Parrot configuration step 2010-06-01T19:02:03 I saw this. That is great. 2010-06-01T19:02:05 so i have a Makefile that generates two separate files 2010-06-01T19:02:41 Does this mean we run a different procedure for RTEMS when building Parrot ? 2010-06-01T19:03:18 what do you mean? 2010-06-01T19:03:24 when building parrot for RTEMS? 2010-06-01T19:03:51 Yes. Is it the normal procedure documented in Parrot or is it custom for RTEMS ? 2010-06-01T19:03:58 it will be a normal procedure 2010-06-01T19:04:13 That is excellent. 2010-06-01T19:04:18 the configure step can either take command line options 2010-06-01T19:04:21 of the form --no-gmp etc. 2010-06-01T19:04:29 or it can be in a file 2010-06-01T19:04:38 Ok. 2010-06-01T19:04:55 so now we run the configuration step with --file=/home/rtems/whatever/ 2010-06-01T19:04:56 So you run a make and that is passed to parrot. 2010-06-01T19:05:01 Ok. 2010-06-01T19:05:11 Can you makefile run parrot ? 2010-06-01T19:05:15 not yet 2010-06-01T19:05:25 Do you think that is a good idea ? 2010-06-01T19:05:34 oh, i think i miscommunicated 2010-06-01T19:05:38 the makefile generates two separate files 2010-06-01T19:05:40 one is the hints file 2010-06-01T19:05:44 the other is the configuration file 2010-06-01T19:05:48 Yes 2010-06-01T19:05:59 so the procedure would be 2010-06-01T19:06:08 1) build RTEMS 2010-06-01T19:06:20 2) run my makefile to generate those files 2010-06-01T19:06:24 3) build parrot with those two files 2010-06-01T19:06:51 In understand. Can 2) and 3) be combined ? 2010-06-01T19:06:56 hmmmm 2010-06-01T19:06:57 probably 2010-06-01T19:07:02 we'd have to know where the parrot source directory is 2010-06-01T19:07:08 then the makefile can just copy the hints file to the right place 2010-06-01T19:07:13 The makefile sees it has a new hints or config and then runs parrot. 2010-06-01T19:07:15 and invoke `perl Configure.pl --file=...` 2010-06-01T19:07:27 Yes. 2010-06-01T19:07:46 yeah, that could be added 2010-06-01T19:08:06 From a new RTEMS users point of view it is just a single make command. NO need to understand what we are doing to make it happen 2010-06-01T19:08:16 yeah, definitely 2010-06-01T19:08:31 Great. The simpler the better the result. 2010-06-01T19:08:31 right now those steps are in the README but even easier would be something completely automated 2010-06-01T19:08:41 Yes I understand. 2010-06-01T19:08:50 i think that's something definitely to add 2010-06-01T19:09:06 This is a big step. 2010-06-01T19:09:10 Well done. 2010-06-01T19:09:27 so now it configures with the right gcc and everything 2010-06-01T19:09:37 Love it. 2010-06-01T19:09:40 i.e. the bsp specific gcc 2010-06-01T19:09:47 i just need to figure out the library path 2010-06-01T19:10:07 Let me have a look. Hang on 2010-06-01T19:10:38 sure, lemme nopaste / gist my current output 2010-06-01T19:10:56 Ok 2010-06-01T19:12:14 http://gist.github.com/421722 2010-06-01T19:12:21 I cannot see the way automake does it at the moment. 2010-06-01T19:13:14 The list of libs is wrong. Where did this come from ? 2010-06-01T19:13:26 i haven't set the libs yet, i think those are auto-configed 2010-06-01T19:13:29 which i need to override 2010-06-01T19:14:16 All you need is -lrtemsbsp -lrtemscpu plus whatever else you decide to test for and add 2010-06-01T19:14:23 okey dokey 2010-06-01T19:14:29 is that already set in a RTEMS makefile variable? 2010-06-01T19:14:34 for example -lnfs, -lopenssh etc 2010-06-01T19:14:36 i can pass the CFLAGS or CCFLAGS 2010-06-01T19:14:50 What about LDFLAGS ? 2010-06-01T19:14:55 hmm, i believe so 2010-06-01T19:15:12 I think this is the one. Any one just for libs ? 2010-06-01T19:15:20 yeah 2010-06-01T19:15:33 I would use that one. 2010-06-01T19:15:43 How would you test for the extra libs ? 2010-06-01T19:16:20 I know you have a makefile but should autoconf be considered ? 2010-06-01T19:16:34 sure - i just don't have too much experience with autoconf 2010-06-01T19:16:38 whatever you think would be most portable 2010-06-01T19:17:10 Is the Makefile using make or shell or something else ? 2010-06-01T19:17:13 i would have done a perl script or a bash script 2010-06-01T19:17:16 just plain ole make 2010-06-01T19:17:33 autoconf is just a script with m4 sugar 2010-06-01T19:17:40 okey dokey 2010-06-01T19:17:50 and autoconf is required for a RTEMS build, right? 2010-06-01T19:18:13 If you have autoconf you can add tests for libs and headers and adjust what you pass. 2010-06-01T19:18:30 very cool, i'll look into that as well 2010-06-01T19:18:37 i'll start with getting this Makefile configuring / building 2010-06-01T19:19:03 Is the current stuff you have available for download ? 2010-06-01T19:19:04 and then i can move the stuff to autoconf 2010-06-01T19:19:05 yeah 2010-06-01T19:19:33 Care to remind an old leaky mind :) 2010-06-01T19:19:36 github.com/bubaflub/gsoc-rtems-parrot-hints-maker 2010-06-01T19:21:56 Ok I see. 2010-06-01T19:22:24 In 4.11 (RTEMS next release) I am looking to remove RTEMS_MAKEFILE_PATH from RTEMS. 2010-06-01T19:22:59 ah, ok 2010-06-01T19:23:15 is there another way i can get the path to that file? 2010-06-01T19:23:32 We can do this with autoconf. I will have a play and get back to you. Please ping me Friday if I have not responded. 2010-06-01T19:23:39 okey dokey 2010-06-01T19:23:55 Yes with an autoconf --prefix option which is standard. 2010-06-01T19:25:01 Environ vars to control builds is considered fragile in some cases. If the user changes something in the environment it can break things and no one knows. 2010-06-01T19:25:18 Think about very controlled builds for critical applications. 2010-06-01T19:26:07 yeah, definitely 2010-06-01T19:26:08 Please keep going on the libs etc and do not wait for me. 2010-06-01T19:26:11 sure sure 2010-06-01T19:26:15 i'll try and improve what i got here 2010-06-01T19:26:33 but i can imagine someone building rtems and then changing some variables like RTEMS_MAKEFILE_PATH 2010-06-01T19:26:43 and that might screw up the Parrot build 2010-06-01T19:27:03 Or different users with different hosts. The combinations or problems compound. 2010-06-01T19:27:08 definitely 2010-06-01T19:27:59 ok, cool 2010-06-01T19:28:03 The other thing is having all RTEMS user autoconf. RTEMS uses it for better or worse and that will not change on the mid term future. 2010-06-01T19:28:18 ok. so autoconf it is 2010-06-01T19:28:29 for next week i'm shooting to have the library situation figured out 2010-06-01T19:28:32 and have it configure cleanly 2010-06-01T19:28:37 building would be nice, but might be a bit much 2010-06-01T19:28:38 I think so. As I said let me play and get you over the start up hump 2010-06-01T19:28:44 yeah 2010-06-01T19:28:51 and autoconf is definitely in the future 2010-06-01T19:29:04 Only because this is what RTEMS uses. 2010-06-01T19:29:29 it sounds a bit more suited for what i want the Makefile to do 2010-06-01T19:29:36 All except the makefile include which Ralf and I would like to see go. Ralf hates it. 2010-06-01T19:29:47 haha, ok 2010-06-01T20:26:24 kiwichris: when you said ''-lrtemsbsp -lrtemscpu" do you *literally* mean "rtemsbsp" or do you mean something like "sparc-sis"? 2010-06-01T20:26:58 No I mean those exact names. 2010-06-01T20:27:11 They are found under BSP specific directories. 2010-06-01T20:27:53 okey dokey 2010-06-01T20:28:01 i found the place to override the libraries in the hints file 2010-06-01T20:28:11 Cool 2010-06-01T20:28:27 You must be getting close to a build. 2010-06-01T20:28:33 i can almost taste it! 2010-06-01T20:37:03 dukeleto: i've got a stupid perl question for ya 2010-06-01T20:42:29 bubaflub: yo 2010-06-01T20:42:44 i'm trying to setup a hints file during to be used during configuration 2010-06-01T20:42:51 and it's in it's own package namespace 2010-06-01T20:42:56 package init::hints::sparc-sis 2010-06-01T20:43:03 and it's complaining about not finding init::hints::sparc 2010-06-01T20:43:06 i imagine that's because the - 2010-06-01T20:43:10 right? 2010-06-01T20:43:21 can i have a package name with hyphens? 2010-06-01T20:43:28 (you can tell i don't do a whole lot of perl OO) 2010-06-01T20:44:00 bubaflub: no, hyphens are not valid perl variable name or namespace characters 2010-06-01T20:44:18 reckon'd as much 2010-06-01T20:44:21 for now i'll just stick with init::hints::sparcsis 2010-06-01T20:44:24 bubaflub: there is a way to get around that with fiddling in the symbol table 2010-06-01T20:44:29 yikes 2010-06-01T20:44:36 bubaflub: an underscore may be nicer to read :) 2010-06-01T20:44:38 haha 2010-06-01T20:44:39 ok 2010-06-01T20:44:53 bubaflub: yeah, don't mess with the symbol table unless you have a really good reason :) 2010-06-01T20:44:59 haha 2010-06-01T20:44:59 i don't 2010-06-01T20:45:03 just a poor choice in names 2010-06-01T20:45:07 i'm switching to underscores 2010-06-01T20:59:07 kiwichris and dukeleto: more progress! though the configuration still doesn't work, it gets pretty far through the script. http://gist.github.com/421722 2010-06-01T20:59:21 i've gotten the hints file to load correctly and to override the libs 2010-06-01T20:59:51 i realize we'll have to provide a *lot* of information in that hints file 2010-06-01T21:01:44 Yes. I cannot remember what I did. 2010-06-01T21:01:44 bubaflub: yes 2010-06-01T21:02:02 bubaflub: have you read perldoc Configure.pl 2010-06-01T21:02:39 bubaflub: there should be a way to only run certain configure stages 2010-06-01T21:02:51 dukeleto: yep, got it under my pillow... 2010-06-01T21:02:56 Is parrot normally built with -Wall ? 2010-06-01T21:02:56 bubaflub: :) 2010-06-01T21:02:57 dukeleto: yes, i can comment out each step i don't want to run 2010-06-01T21:03:20 kiwichris: i think we have different warning flags for different compilers 2010-06-01T21:03:22 kiwichris: no, not usually, it has different -W's for different compilers 2010-06-01T21:03:46 Maybe the -Wall should be stripped from the flags. It seems to confuse things. 2010-06-01T21:04:02 kiwichris: ok, i'll make a note to get rid of it 2010-06-01T21:04:05 bubaflub: yes, you will have to provide a decent amount of info, but you are getting pretty far right now 2010-06-01T21:04:21 bubaflub: is your rtems_parrot_config_directives under version control? 2010-06-01T21:04:42 dukeleto: yes and no. that file and the hints file is generated from the Makefile 2010-06-01T21:05:07 the Makefile (+ README and TODO) are on my github 2010-06-01T21:05:31 bubaflub: ok 2010-06-01T21:05:53 bubaflub: ./test_18161: ./test_18161: cannot execute binary file <-- this is because the $PWD is not what Configure.pl thinks it is 2010-06-01T21:06:32 bubaflub, I cheated with my build. I picked up the C files that worked and used them. I avoided the configuration. 2010-06-01T21:07:30 kiwichris: yeah, this will be a little bit more work up front but i think will track trunk much easier 2010-06-01T21:07:41 bubaflub: the branch that wants to add --builddir to Configure.pl would help with the "cannot execute binary file" issue 2010-06-01T21:07:53 bubaflub, sure will. Mine was a hack and really big :) 2010-06-01T21:08:00 bubaflub: yes, i agree, this way is more generic and will be maintainable 2010-06-01T21:08:02 dukeleto: i think it's because that file was built and linked with a sparc-gcc and run on a linux VM 2010-06-01T21:08:12 bubaflub: ah, that could be it too :) 2010-06-01T21:08:34 bubaflub: you are right, i was confusing that with a similar-looking error 2010-06-01T21:10:18 bubaflub: any other questions? i am about to go for a rainy jog :) 2010-06-01T21:11:26 dukeleto: nah, i'm about to sit down for dinner 2010-06-01T21:11:45 i'll email y'all when my blog post is up 2010-06-01T21:11:48 have a good $localtime, everyone 2010-06-01T21:11:58 bubaflub: sounds good 2010-06-01T21:12:40 dukeleto, good on you. Now I feel guilty. 2010-06-01T21:27:34 *** bubaflub has quit IRC 2010-06-02T02:02:25 *** sebhub has joined #rtems 2010-06-02T02:03:04 good morning 2010-06-02T02:11:24 sebhub, hi 2010-06-02T02:11:44 What do you mean by a file system item ? 2010-06-02T02:13:44 rtems_filesystem_location_info_t 2010-06-02T02:15:12 So the iop may reference the mount table entry and iop may be referenced by the path loc ? 2010-06-02T02:16:30 these locations are not only used by iop related things, also chroot() uses them for example 2010-06-02T02:18:32 The iop can be handled with the sem it contains. The mt_entry may need references. 2010-06-02T02:18:57 its not only the reference, problematic is the path to create these location items 2010-06-02T02:19:13 Go on. 2010-06-02T02:19:44 the locations will be created by the eval methods of the file system 2010-06-02T02:19:53 Yes. 2010-06-02T02:20:05 to invoke these methods you need a mount point 2010-06-02T02:20:11 Yes 2010-06-02T02:20:21 these methods usually use file system resources and data structures 2010-06-02T02:20:49 They may contain specific data but hold no resources. 2010-06-02T02:20:55 Or they may. 2010-06-02T02:21:08 someone can interrupt you between the mount point evaluation and the call to the file system eval method 2010-06-02T02:21:17 now someone can unmount the file system 2010-06-02T02:21:30 and now you use freed resources and data structures 2010-06-02T02:21:35 Ah ok. I will take a closer look at this. Thanks for the detail. 2010-06-02T02:21:59 I will reference this chat in the PR. I am about to head out. 2010-06-02T02:22:05 ok 2010-06-02T02:22:43 this situation happend to me with usb sticks and this usually renders the system to be unusable afterwards 2010-06-02T02:25:05 Should we create separate PRs for these and have this one depend on them ? 2010-06-02T02:25:36 * kiwichris is out for a while 2010-06-02T02:45:14 *** lcpfnvc has joined #rtems 2010-06-02T02:46:11 morning 2010-06-02T02:52:47 hi 2010-06-02T05:10:51 *** Raj_ has joined #rtems 2010-06-02T05:25:58 *** Raj_ has quit IRC 2010-06-02T06:28:44 *** dr__house has joined #rtems 2010-06-02T07:19:08 *** DrJoel has joined #rtems 2010-06-02T07:19:08 *** ChanServ sets mode: +o DrJoel 2010-06-02T07:22:27 *** kiwichris has quit IRC 2010-06-02T07:22:47 aanjhan: I checked a bit into or32 2010-06-02T07:39:47 *** DrJoel has quit IRC 2010-06-02T08:01:00 *** zwj has joined #rtems 2010-06-02T08:13:46 *** gedare has joined #rtems 2010-06-02T08:16:13 sebhub:hi, good night 2010-06-02T08:17:29 *** tiemen has joined #rtems 2010-06-02T08:17:33 hi all 2010-06-02T08:18:38 sebhub:about the sort order, how should we define the "order"? 2010-06-02T08:18:47 hi 2010-06-02T08:32:38 hi 2010-06-02T08:34:04 i would use a triple (domain, {before,actual,after},index) 2010-06-02T09:02:06 *** DrJoel has joined #rtems 2010-06-02T09:02:06 *** ChanServ sets mode: +o DrJoel 2010-06-02T09:12:33 *** cdcs has joined #rtems 2010-06-02T09:16:31 hi DrJoel ! 2010-06-02T09:16:37 sorry i missed you yesterday 2010-06-02T09:16:43 how are you ? 2010-06-02T09:21:35 no problem.. sorry I haven't said hi when you have been on. 2010-06-02T09:21:43 Looks like from your blog that you are doing well 2010-06-02T09:27:38 *** sebhub has quit IRC 2010-06-02T09:35:48 i'm still kindof busy with my exams and my courses but it's ending soon 2010-06-02T09:36:28 i've read almost all http://www.rtems.org/onlinedocs/releases/rtemsdocs-4.9.4/share/rtems/html/bsp_howto/index.html 2010-06-02T09:36:33 all of the students are going through that. 2010-06-02T09:36:49 oh i have a question ! 2010-06-02T09:37:00 the bsp howto is good reading. 2010-06-02T09:37:01 do i have to pay to attend to the open class over the web ? 2010-06-02T09:37:02 It also helps to step through the initialization of a working BSP and watch what happens 2010-06-02T09:37:10 no.. it is free for gsoc students 2010-06-02T09:37:17 oh super :) 2010-06-02T09:37:22 do i have to register somewhere ? 2010-06-02T09:37:33 i've seen some mails about that, but i cannot find them again, sorry 2010-06-02T09:37:39 s/mails/e-mails/ 2010-06-02T09:38:11 No. I will send out a URL to the GSOC mailing list. I thought you might have to since last year I sent out personal invites but this time I noticed you can just give out the URL of the online meeting. Much easier 2010-06-02T09:38:19 You missed the other question.. what timezone? :-D 2010-06-02T09:38:54 I am on Central Standard. The same as Chicago. Monday is a kick start which basically covers the GSOC hello world project 2010-06-02T09:40:08 I'm in Paris, it's GMT + 2 right now 2010-06-02T09:42:16 I'm in Huntsville Alabama and it is GMT -5 2010-06-02T09:43:09 at least, this time of year 2010-06-02T09:43:39 yep. :) 2010-06-02T09:44:59 DrJoel: is the starting hour of the class decided yet ? 2010-06-02T09:45:46 it is supposed to be 8am CST but depends on when the in-person (paying) folks actually get here. 2010-06-02T09:47:45 *** bubaflub has joined #rtems 2010-06-02T09:47:51 In case it wasn't clear, your first focus is to get Hello World running using polled IO. 2010-06-02T09:48:01 mornin' DrJoel 2010-06-02T09:48:19 Thursday and Friday are BSP and Driver related and focus on the order you want to bring up a BSP in. So that is the most important part for you. 2010-06-02T09:49:15 The first part is theory and patterns from a grad RT class I taught.. then SuperCore architecture (1/2 day), then Classic API with debugging and tuning mixed in, then BSP and Driver 2010-06-02T09:50:51 *** alexp0205 has joined #rtems 2010-06-02T09:52:14 *** lcpfnvc has quit IRC 2010-06-02T09:52:52 so i guess 8am CST makes 14h UCT so 16h in Paris 2010-06-02T09:52:58 it's ok for me :) 2010-06-02T09:53:47 ok good 2010-06-02T09:54:06 i will try to have the basic bsp working during the week-end 2010-06-02T09:55:32 cool! It looks like 4.10 will branch this week which makes it possible to incrementally merge if you want 2010-06-02T09:59:30 *** exception13 has joined #rtems 2010-06-02T10:07:24 *** zwj has quit IRC 2010-06-02T10:08:11 :) 2010-06-02T10:08:18 DrJoel: you checked a bit iinto or32 and ? 2010-06-02T10:10:32 they have newer tools like gcc 4.3 or 4.4 on their website which should make your life a lot easier. 2010-06-02T10:10:43 wow. on whose site? 2010-06-02T10:10:46 The rtems they have there is 4.5 which is ~8-10 years old. 2010-06-02T10:11:04 opencores.org 2010-06-02T10:11:25 http://opencores.org/openrisc,gnu_toolchain 2010-06-02T10:11:42 But we had it in 4.6 which is still ancient but a lot more recent. 2010-06-02T10:12:05 interesting they have linux 2.6.24 running 2010-06-02T10:12:23 Thanks for the pointer DrJoel :-) 2010-06-02T10:12:38 You should be able to take their toolchain, add the or32-rtems target, use the latest newlib with our patches, add libc/machine/or32 and then update the port from the last source we have. 2010-06-02T10:12:44 gedare: anything yet needed from me for the MMU thingi? 2010-06-02T10:13:19 aanjhan: no, i figured out a way around 2010-06-02T10:13:56 DrJoel: hmm. any news yet on Munich? 2010-06-02T10:15:04 not yet.. need to email Thomas.. thanks for reminding me.. only a month away 2010-06-02T11:00:53 *** cdcs_ has joined #rtems 2010-06-02T11:01:00 *** cdcs has quit IRC 2010-06-02T11:07:16 *** lekernel has joined #rtems 2010-06-02T11:09:21 *** cdcs_ is now known as cdcs 2010-06-02T11:23:39 Hi DrJoel 2010-06-02T11:23:59 gedare: hello 2010-06-02T11:24:04 Ticker on pc386 / qemu isn't working for me, from head pulled about an hour ago 2010-06-02T11:24:54 :( 2010-06-02T11:25:13 it hangs during initialization. i'll investigate, thought i'd see if it's just me 2010-06-02T11:25:20 Can you set a break point at rtems_fatal_error_occurred? I think I am seeing the same thing. Mount is failing 2010-06-02T11:25:33 I thought it was my smp mods and had just built sis to debug 2010-06-02T11:25:33 here is a backtrace: 2010-06-02T11:25:36 #0 0x0010d7fa in _Internal_error_Occurred ( 2010-06-02T11:25:36 the_source=INTERNAL_ERROR_RTEMS_API, is_internal=0, the_error=2882338818) 2010-06-02T11:25:36 at ../../../../../../rtems/c/src/../../cpukit/score/src/interr.c:61 2010-06-02T11:25:36 #1 0x0010cac2 in rtems_fatal_error_occurred (the_error=2882338818) 2010-06-02T11:25:36 at ../../../../../../rtems/c/src/../../cpukit/sapi/src/fatal.c:37 2010-06-02T11:25:36 #2 0x0010854c in rtems_filesystem_initialize () 2010-06-02T11:25:38 at ../../../../../../rtems/c/src/../../cpukit/libcsupport/src/base_fs.c:60 2010-06-02T11:25:40 #3 0x00100642 in bsp_libc_init (heap_begin=0x138558, heap_size=131877544, 2010-06-02T11:25:42 sbrk_amount=0) 2010-06-02T11:25:44 at ../../../../../../../../rtems/c/src/lib/libbsp/i386/pc386/../../shared/bsplibc.c:28 2010-06-02T11:25:48 #4 0x0010047a in boot_card (cmdline=0x125f90 "(hd1)/ticker.exe") 2010-06-02T11:25:50 at ../../../../../../../../rtems/c/src/lib/libbsp/i386/pc386/../../shared/bootcard.c:93 2010-06-02T11:25:52 #5 0x001000a5 in zero_bss () 2010-06-02T11:25:54 at ../../../../../../../../rtems/c/src/lib/libbsp/i386/pc386/start/start.S:267 2010-06-02T11:25:56 #6 0x00125f90 in _boot_multiboot_memory () 2010-06-02T11:25:58 #7 0x00000000 in ?? () 2010-06-02T11:26:00 this is with stock head. 2010-06-02T11:26:10 yep.. same thing 2010-06-02T11:26:22 we need to ping kiwichris.. he broke something 2010-06-02T11:26:31 tsk tsk. ok 2010-06-02T11:26:57 fwiw hello still works 2010-06-02T11:27:13 not that it's any consolation. :) 2010-06-02T11:27:25 Probably uses a different filesystem for root.. just guessing 2010-06-02T11:27:41 file a PR and assign it to him. 2010-06-02T11:27:49 ko 2010-06-02T11:29:30 I will build locally with a clean no-SMP tree and see if I can track it down but have meeting at customer site in a bit 2010-06-02T11:30:04 k.. I'm not familiar with the fs code, i might take a gander but unlikely to come up with anything. 2010-06-02T11:33:33 Chris was changing the mount to take strings and fighting the initialization/configuration. So he likely broke it 2010-06-02T11:33:46 This is the last change before branching 2010-06-02T11:34:11 yeah. cvs wasn't compiling yesterday afternoon from his changes :p 2010-06-02T11:34:23 cool 2010-06-02T11:36:44 IMO the sparc64 is the first addition after the branch.. some work will be on 4.10 branch for release prep .. but it should be mostly frozen 2010-06-02T11:37:27 sounds good. I should be able to spin a "stable" bsp that merges against the 4.10 release when it happens 2010-06-02T11:37:46 and most of the tool support is in place now it looks like. 2010-06-02T11:38:18 mount returns -1 and fatal_error is called. 2010-06-02T11:38:32 * gedare has exhausted his knowledge of the fs code base :) 2010-06-02T11:40:00 I was willing to go a bit deeper than that but that was about where I wanted to quit 2010-06-02T11:40:18 i also will report the parameters to the mount call 2010-06-02T11:40:22 chris can go from there :) 2010-06-02T11:40:29 good time for a long lunch! 2010-06-02T11:44:15 *** dr__house has quit IRC 2010-06-02T11:55:04 *** alexp0205 has left #rtems 2010-06-02T11:56:57 *** cdcs has quit IRC 2010-06-02T12:33:51 *** exception13 has left #rtems 2010-06-02T12:55:37 *** dr__house has joined #rtems 2010-06-02T13:50:25 *** gedare has quit IRC 2010-06-02T14:25:42 *** dr__house has quit IRC 2010-06-02T14:57:04 *** bubaflub has quit IRC 2010-06-02T16:07:03 *** lekernel has quit IRC 2010-06-02T16:21:35 *** DrJoel has quit IRC 2010-06-02T17:22:11 *** kiwichris has joined #rtems 2010-06-02T17:48:43 *** kiwichris has quit IRC 2010-06-02T18:37:01 *** kiwichris has joined #rtems 2010-06-02T20:16:53 *** DrJoel has joined #rtems 2010-06-02T20:16:53 *** ChanServ sets mode: +o DrJoel 2010-06-02T20:17:36 DrJoel, hi 2010-06-02T20:17:58 hey.. I just got back from Boy Scouts.. so what was the default fs in 4.9? 2010-06-02T20:18:01 * DrJoel goes looking 2010-06-02T20:18:32 miniIMFS was default. 2010-06-02T20:18:34 It could have been miniIMFS 2010-06-02T20:18:57 const rtems_filesystem_mount_table_t configuration_mount_table = { 2010-06-02T20:18:57 #ifdef CONFIGURE_USE_IMFS_AS_BASE_FILESYSTEM 2010-06-02T20:18:57 &IMFS_ops, 2010-06-02T20:18:57 #else /* using miniIMFS as base filesystem */ 2010-06-02T20:18:57 &miniIMFS_ops, 2010-06-02T20:18:58 #endif 2010-06-02T20:19:18 Given the size saving is so small because miniIMFS is broken I thought IMFS should be used. Maybe that is wrong. 2010-06-02T20:19:35 Seb has a patch to fix this for 4.11 2010-06-02T20:19:43 we might be able to effectively go down to devfs 2010-06-02T20:19:50 how invasive is the patch? 2010-06-02T20:20:05 Yes I agree. I think all tests should say use devfs and default to IMFS. It is more user friendly 2010-06-02T20:20:14 He said it was big. 2010-06-02T20:20:33 I suspect you need to set up extra call tables. 2010-06-02T20:20:44 for the set handlers call 2010-06-02T20:20:55 I need some help with running all tests on sis 2010-06-02T20:20:56 depends on what he trimmed to make it smaller.. it has to support directories, symlinks and device nodes 2010-06-02T20:21:22 Yes, but it uses the set handler call and that pulls in the rest of the IMFS 2010-06-02T20:21:35 Nothing actually uses it 2010-06-02T20:21:47 IMO.. devfs should replace miniIMFS and miniIMFS can be killed in 4.11 as long as devfs is a suitable replacement 2010-06-02T20:21:54 Yeap I agree 2010-06-02T20:22:03 We should make it a suitable replacement. 2010-06-02T20:22:24 devfs is the equivalent of the device name tables we had pre-filesystem support. 2010-06-02T20:22:24 I have built leon2 BSP and I want to run all tests. 2010-06-02T20:22:32 Yes and I think it is fine. 2010-06-02T20:22:35 so how do we fix this immediate problem ... 2010-06-02T20:22:44 I have a patch. We just use the IMFS. 2010-06-02T20:22:53 leon2 requires tsim and the only licensed copy is on britain. 2010-06-02T20:23:05 What do I build for sis ? 2010-06-02T20:23:15 --enable-rtemsbsp=sis 2010-06-02T20:23:19 Ok. 2010-06-02T20:23:27 How do I run all the tests ? 2010-06-02T20:23:34 I have gcc-testing 2010-06-02T20:23:54 do a make in gcc-testing.. it creates a bin directory.. in there is a script named sis and sis-gdb. 2010-06-02T20:24:05 Just make ? 2010-06-02T20:24:08 sis *.exe will run them all with timeouts and place the logs in a "log/" subdirectory 2010-06-02T20:24:14 yes.. not autoconf'ed 2010-06-02T20:24:24 Running 2010-06-02T20:25:23 all of the scripts in gcc-testing/bin that look like the name of a BSP in fact are wrapper scripts to run that BSP. The argument list is common 2010-06-02T20:25:52 if the sim is in gdb BSP-gdb does the gdb setup for you. If it is external like qemu, skyeye or tsim, it tries to do the remote for you 2010-06-02T20:26:10 Do I need all that the README says I need if I have tools and RTEMS already built ? 2010-06-02T20:26:41 hmm.. probably not.. I bet the readme is focused on testing gcc and need to be in the gcc subdirectory 2010-06-02T20:27:14 Yes and I think that confused me. May be we should have a couple of sections with a testing just RTEMS section. 2010-06-02T20:27:39 or README.rtems etc 2010-06-02T20:27:54 I will visit that.. This started focused on testing gcc and grew wildly :-D 2010-06-02T20:28:24 It is great but my eye kinda said more coffee when I got to GNATS :) 2010-06-02T20:28:50 GNATS or GNAT? I hope GNATS is dead.. :-D 2010-06-02T20:29:08 For Ada, you probably felt a disturbance in the force from Germany 2010-06-02T20:29:10 GNAT sorry 2010-06-02T20:29:36 Yes the dark side started to tremble 2010-06-02T20:30:25 On the bright side the failure showed the filesystem control worked. You did not have the miniIMFS :) 2010-06-02T20:30:28 the sad thing is that GNAT/RTEMS works great and Ada is really an ideal solution for safety-critical embedded systems. With SPARK notation in the comments, you can do formal prove analysis on the code 2010-06-02T20:30:36 :-D 2010-06-02T20:31:00 brb 2010-06-02T20:31:10 And have you written a paper or article to publish that says this ? 2010-06-02T20:31:14 ;) 2010-06-02T22:40:22 *** kiwichris has quit IRC 2010-06-02T22:47:38 *** kiwichris has joined #rtems 2010-06-02T23:03:21 *** kiwichris has quit IRC 2010-06-03T00:22:25 *** dr__house has joined #rtems 2010-06-03T00:38:55 *** kiwichris has joined #rtems 2010-06-03T00:44:56 *** kiwichris has quit IRC 2010-06-03T01:24:09 *** kiwichris has joined #rtems 2010-06-03T01:28:11 *** kiwichris has quit IRC 2010-06-03T01:33:11 *** kiwichris has joined #rtems 2010-06-03T01:45:20 *** lcpfnvc has joined #rtems 2010-06-03T01:46:46 morning :D 2010-06-03T01:47:23 *** kiwichris has quit IRC 2010-06-03T01:50:24 *** dr__house has quit IRC 2010-06-03T01:58:51 *** kiwichris has joined #rtems 2010-06-03T02:12:24 *** exception13 has joined #rtems 2010-06-03T02:23:57 *** lekernel has joined #rtems 2010-06-03T02:32:35 *** dr__house has joined #rtems 2010-06-03T02:37:17 *** kiwichris has quit IRC 2010-06-03T03:20:28 *** kiwichris has joined #rtems 2010-06-03T03:26:00 *** cdcs has joined #rtems 2010-06-03T03:26:59 cdcs, hi 2010-06-03T03:27:07 hi: 2010-06-03T03:27:27 cdcs, welcome back. I hope the exams and study went well. 2010-06-03T03:27:52 regarding the chat time: 11am GMT is also fine for me 2010-06-03T03:28:08 yes they went :) 2010-06-03T03:28:35 Great. At the moment it is 6pm here and normally that is food etc time and a bit busy for a couple of hours 2010-06-03T03:28:57 i just woke up 2010-06-03T03:29:09 I am happy to settle on Thursday at 9pm or 11am UTC 2010-06-03T03:29:35 It is a little early for Joel but he can see the logs and chip in after that if need be. 2010-06-03T03:30:00 We need to get the sample application working for you. 2010-06-03T03:30:37 between that two i prefer 9pm if you don't mind... 11 am is lunch time ;) 2010-06-03T03:31:04 Is that Thursday your time ? 2010-06-03T03:31:27 thursday GMT time 2010-06-03T03:31:43 Sure. That is 7am Friday my time. 2010-06-03T03:32:04 i am GMT+1 2010-06-03T03:32:15 Thanks 2010-06-03T03:33:06 yes we need to get the sample application working 2010-06-03T03:33:24 I am just uploading my latest RFS example application. 2010-06-03T03:34:11 ok 2010-06-03T03:34:34 regarding the NetBSD driver: 2010-06-03T03:35:07 Yeap 2010-06-03T03:35:12 the NetBSD system calls that need to be ported, should be ported to libbsdport? 2010-06-03T03:35:26 * kiwichris has uploaded a new RFS tester app 2010-06-03T03:35:41 I do not know. 2010-06-03T03:36:05 Do we repeat what maybe in libbsdport or do we add the driver into there and get it to work. 2010-06-03T03:36:45 Maybe it is easier to try and see what pops out. 2010-06-03T03:37:08 ok 2010-06-03T03:37:30 NetBSD driver interface is a little different from RTEMS one 2010-06-03T03:37:37 Maybe add a new dir into libbsdport for this 2010-06-03T03:37:56 we don't have a initialize call 2010-06-03T03:37:58 Which RTEMS interface ? The libbsdport or RTEMS ? 2010-06-03T03:38:07 RTEMS 2010-06-03T03:38:25 The processes for libbsdport is ... 2010-06-03T03:38:37 Add BSD code to libbsdport 2010-06-03T03:38:46 Add glue between the BSD code and RTEMS. 2010-06-03T03:39:15 The glue will transform the RTEMS call outs to the BSD ones. 2010-06-03T03:39:59 The idea behind libbsdport is not to touch the BSD if that can be helped. 2010-06-03T03:40:44 ok 2010-06-03T03:43:25 i was asking that because in netBSD there is a piece of the kernel that calls all disk drivers probe function 2010-06-03T03:43:29 So we add a new dir to libbsdport called ahic and copy and paste some configure.ac logic for another dir and then copy a makefile. 2010-06-03T03:44:37 then if the probe returns 0 it calls that driver attach function and inits the drive 2010-06-03T03:44:41 ok 2010-06-03T03:44:55 Great. That all seems normal. 2010-06-03T03:45:39 Look in init.c in the rfs tester application 2010-06-03T03:45:46 yup this functionality needs to be included in libbsdport 2010-06-03T03:46:10 Which functionality ? 2010-06-03T03:46:23 init.c... i'm there 2010-06-03T03:46:29 circle thru the driver 2010-06-03T03:46:35 Yes. 2010-06-03T03:47:00 Look at the area that is "#if LIBBSDPORT" 2010-06-03T03:48:11 A dummy interface is created called libbsdport_config 2010-06-03T03:48:36 The normal RTEMS TCP/IP calls it as if it is a single fully ported RTEMS driver 2010-06-03T03:49:06 It then makes the needed calls to libbsdport to set up the various detected interfaces 2010-06-03T03:50:02 Notice the table "libbsdport_netdriver_table". This is the table of drivers probed. 2010-06-03T03:50:22 We need a similar process for AHIC drivers. 2010-06-03T03:50:50 yup 2010-06-03T03:50:52 For now we can have main all the various bits and pieces we need. 2010-06-03T03:51:14 Once we have something we can refine these packaging things. 2010-06-03T03:51:38 For now we can have main CALL the various bits and pieces we need. 2010-06-03T03:53:11 So do you want to have a go at making a new dir in libbsdport to add this code ? 2010-06-03T03:54:41 the configure.ac can be the one on libbsdport root? 2010-06-03T03:54:51 Yeap. 2010-06-03T03:55:32 and the makefile can be the one on bsd_eth 2010-06-03T03:55:37 _ 2010-06-03T03:57:08 ? 2010-06-03T04:04:40 Yeap 2010-06-03T04:04:45 Sorry was distracted. 2010-06-03T04:06:29 a minute...i am trying to compile the rfs test application 2010-06-03T04:10:28 i386-rtems4.11-objcopy -O elf32-i386 --remove-section=.comment --remove-section=.note --strip-unneeded filesystem-test.exe filesystem-test.nxe 2010-06-03T04:10:29 i386-rtems4.11-objcopy: 'filesystem-test.exe': No such file 2010-06-03T04:10:31 make: *** [filesystem-test.ralf] Error 1 2010-06-03T04:13:06 also should i copy Makefile.in, Makefile.am and Makefile?( i am a bit on the dark on this... 2010-06-03T04:13:07 make: *** [filesystem-test.ralf] Error 1 2010-06-03T04:13:57 sorry i have a "oversensitive" enter 2010-06-03T04:15:58 No. Only the *.ac and *.am. 2010-06-03T04:16:09 The *.in etc are generated. 2010-06-03T04:16:40 I am off. 2010-06-03T04:17:09 I need to sort homey things at my end :) 2010-06-03T04:20:13 so configure.ac and makefile.am? 2010-06-03T04:54:24 *** lekernel has quit IRC 2010-06-03T05:24:32 *** cdcs has quit IRC 2010-06-03T05:24:36 *** cdcs has joined #rtems 2010-06-03T05:25:36 *** dr__house has quit IRC 2010-06-03T05:58:16 *** dr__house has joined #rtems 2010-06-03T06:09:52 *** tuxmaniac has joined #rtems 2010-06-03T06:42:43 *** dr__house has quit IRC 2010-06-03T07:07:19 *** zwj has joined #rtems 2010-06-03T07:26:53 *** DrJoel has quit IRC 2010-06-03T08:00:10 *** kiwichris has quit IRC 2010-06-03T08:10:21 *** DrJoel has joined #rtems 2010-06-03T08:10:21 *** ChanServ sets mode: +o DrJoel 2010-06-03T08:17:11 *** exception13 has left #rtems 2010-06-03T08:23:07 *** Raj_ has joined #rtems 2010-06-03T08:32:55 DrJoel: hi 2010-06-03T08:34:09 Raj_: hello 2010-06-03T08:34:43 I was testing gccgo, have some doubts 2010-06-03T08:35:22 One test, checks for the command line arguments, but I see that entering cmdline arguments in rtems 2010-06-03T08:35:46 can only be through a prompt or as macros during compilation, right ? 2010-06-03T08:37:28 Right. 2010-06-03T08:37:42 We probably need to just expect that one to fail until we get some magic set up. 2010-06-03T08:38:03 hmm .. 2010-06-03T08:38:20 And do you know why go_tg.o is used, during testing ? 2010-06-03T08:39:02 it is test glue for dejagnu. probably like the one for C/C++ testing, it provides wrappers for exit, etc 2010-06-03T08:39:08 I has to do with the dejagnu framework, many tests are failing because of not finding that file 2010-06-03T08:39:43 if you look at the C/C++ rundeja script, it explicitly compiles a test glue file. You probably have to do the same thing 2010-06-03T08:39:58 Yea, I thought so, because gcc tests also have a similar gcc_tg.o too 2010-06-03T08:40:17 Well I can find go_tg.o in testsuite/go 2010-06-03T08:41:02 Some tests find it, while some can't, around 1000 are expected passes, and 150 are unexpected fails 2010-06-03T08:44:32 that is tedious but should be easy to fix once you see the trick (or Ian helps) 2010-06-03T08:44:42 In C/C++, some take a different name for the object file 2010-06-03T08:57:25 DrJoel: These are the main 2 problems I saw, rest of the tests are passing on tweaking some config params 2010-06-03T08:58:02 will keep you updated, working on the go patches at the moment. 2010-06-03T08:59:13 That's amazing! The Ada testing kind of went that way. Nothing ran, then a big chunk ran, fixed one thing, then a lot more ran. :) 2010-06-03T09:07:13 DrJoel: But, this are just the bare compilers tests, the library tests are different :). Most of the problems will be in that i guess ! 2010-06-03T09:07:26 *compiler 2010-06-03T09:15:27 *** exception13 has joined #rtems 2010-06-03T09:27:45 *** lcpfnvc has quit IRC 2010-06-03T09:33:05 *** madrazr has joined #rtems 2010-06-03T09:43:20 *** Raj_ has quit IRC 2010-06-03T10:01:54 *** bubaflub has joined #rtems 2010-06-03T10:12:29 *** gedare has joined #rtems 2010-06-03T10:19:50 mornin' DrJoel 2010-06-03T10:20:06 hey bubaflub .. how is your day? 2010-06-03T10:20:18 DrJoel: pretty good... just started an hour or two ago 2010-06-03T10:20:28 but my extra memory showed up at work today which is pretty exciting 2010-06-03T10:20:35 i can now run Virtual Box without closing everything else on my computer 2010-06-03T10:21:08 :) I had OAR order some extra memory for my desktop for the same reason 2010-06-03T10:21:58 maybe i'll give vbox a try with my new laptop...fc12 hangs occasionally while i use firefox: p 2010-06-03T10:22:01 also, good morning 2010-06-03T10:22:19 gedare: i'm running FC12 inside virtual box, it's not too shabby 2010-06-03T10:22:42 and maybe after GSoC i'll work on getting Mac binaries so i don't have to run VirtualBox at all 2010-06-03T10:22:53 heh. on the other hand, not being able to use the web browser might be a boon to productivity. 2010-06-03T10:22:58 hahaha 2010-06-03T10:23:30 i don't see why you couldn't build from sources. 2010-06-03T10:23:55 gedare: no doubt, but i'd like to get some MacPorts stuff setup 2010-06-03T10:24:01 makes it easier to upgrade 2010-06-03T10:24:24 i see 2010-06-03T10:25:39 also might lower the barrier of entry for other people on Mac OS X 2010-06-03T10:26:20 yeah 2010-06-03T10:27:02 i have some scripts to build from sources, but it isn't fool-proof and it takes time. 2010-06-03T10:33:11 there is a simple build script in the gcc-testing cvs module which shouuld help 2010-06-03T10:35:18 somewhat, it doesn't apply patches though. :p 2010-06-03T10:42:33 don't trust scripts to do that and stay up to date on the patches. :-D 2010-06-03T10:43:09 yeah. that's the lack of fool-proofiness in my approach. :) 2010-06-03T10:43:44 which would be easier if the most recent patches were named consistently, instead of having to look by date 2010-06-03T10:47:21 Adding a script to contrib/crossrpms which let the patch and tool versions get inserted like they are in the RPMs is better long term 2010-06-03T10:55:02 *** madrazr1 has joined #rtems 2010-06-03T10:56:55 *** madrazr has quit IRC 2010-06-03T11:09:37 *** madrazr1 has quit IRC 2010-06-03T11:10:49 *** cdcs has quit IRC 2010-06-03T11:12:40 *** arvind_khadri has joined #rtems 2010-06-03T11:14:02 DrJoel, hiii :) 2010-06-03T11:14:15 hey arvind_khadri how is life? 2010-06-03T11:14:38 DrJoel, am fine, how are things at your end ? I just finished my exams :) 2010-06-03T11:15:36 getting ready for an RTEMS class next week. And my wife just found a renaissance faire near atlanta she would like to go to this weekend :) 2010-06-03T11:16:21 :) oh cool, atlanta is a big place around that side right, my friend who stays in SC also keeps talking about it :) 2010-06-03T11:18:23 DrJoel, ^ 2010-06-03T11:18:42 btw, when can I start working on the project :) 2010-06-03T11:18:53 http://www.garenfest.com 2010-06-03T11:19:00 we have costumes.. :-D 2010-06-03T11:19:06 * gedare giggles 2010-06-03T11:24:24 today's goal: dummy ready queue interface 2010-06-03T11:24:58 well, not dummy, just wrap the current ready chain accesses. 2010-06-03T11:25:10 *** madrazr has joined #rtems 2010-06-03T11:25:42 gedare: that is a very good goal. There is someone I need to introduce you to who is doing PhD work on a hardware implemented scheduler for RTEMS. So this interface would help them I think 2010-06-03T11:25:48 arvind_khadri: whenever you want :-D 2010-06-03T11:26:15 yes please do. my phd work also involves hardware-based scheduling. 2010-06-03T11:26:35 i'd rather not scoop / be scooped. 2010-06-03T11:27:00 fwiw, that is why i proposed this gsoc project -- to make it easier to identify scheduling points. 2010-06-03T11:27:22 (and now I am interested in SMP stuff too, curses!) 2010-06-03T11:30:21 :D I want to write up the scheduler simulator. I believe there is a paper in there. Are you under pressure to publish along the way? 2010-06-03T11:30:38 yup. I'd be glad to work with you on that. 2010-06-03T11:31:13 *** zwj has quit IRC 2010-06-03T11:50:54 * DrJoel hates avery labels 2010-06-03T11:52:34 ha 2010-06-03T11:55:51 what are task proxies? or where can I read about them 2010-06-03T11:56:55 They are the structures enqueued to represent remote tasks blocking. When a proxy wakes up we send a message to another node to wake it up 2010-06-03T11:58:24 oh i see. i'll ignore it for now, but that model might help with remote dispatching as well. 2010-06-03T11:58:36 I'm copying the threadq code as a basis for the refactored readyq 2010-06-03T11:58:42 rayther, the interface. 2010-06-03T12:30:12 *** madrazr has quit IRC 2010-06-03T12:44:13 *** arvind_khadri has quit IRC 2010-06-03T12:57:54 *** arvind_khadri has joined #rtems 2010-06-03T13:13:47 *** arvind_khadri has quit IRC 2010-06-03T13:17:12 *** Fallenou has quit IRC 2010-06-03T13:20:07 *** Fallenou has joined #rtems 2010-06-03T13:26:36 *** arvind_khadri has joined #rtems 2010-06-03T14:03:28 *** cdcs has joined #rtems 2010-06-03T14:06:34 *** arvind_khadri has quit IRC 2010-06-03T14:19:48 *** arvind_khadri has joined #rtems 2010-06-03T14:23:37 *** cdcs has quit IRC 2010-06-03T14:31:28 *** tuxmaniac has quit IRC 2010-06-03T15:08:21 *** cdcs has joined #rtems 2010-06-03T15:10:30 *** arvind_khadri has quit IRC 2010-06-03T15:45:57 *** cdcs has quit IRC 2010-06-03T15:46:40 uh oh thunder. 2010-06-03T15:51:41 :) 2010-06-03T15:52:02 making good progress though, i like the new readyq interface that is shaping up. 2010-06-03T15:53:09 a little bit bloated compared to what i had in mind, but decent flexibilty. 2010-06-03T15:53:23 a good first cut *nods* 2010-06-03T15:53:37 I can't wait to start reviewing it. 2010-06-03T15:54:18 FWIW with the class next week and wanting to branch 4.10 ASAP. (it is ready), I think that I will want to merge the sparc64 port week after next 2010-06-03T15:54:31 OK I will keep that in mind. 2010-06-03T15:55:06 I don't see it as a hard or risky merge. :-D 2010-06-03T15:55:07 I'll get the sources up-to-date with the CVS and prepare a clean branch, then get you some instructions on how to build and test. 2010-06-03T15:55:19 nah, new architectures are pretty well separated 2010-06-03T15:55:44 Did you add a chapter to the cpu supplement? 2010-06-03T15:56:00 Do you want me to include the ultrasparc 3 BSP as well? I don't know of any open source simulator to test it with 2010-06-03T15:56:11 I haven't done much to document. :) 2010-06-03T15:56:27 Well, I have a lot of notes, but nothing cohesive or in RTEMS documentation formats. 2010-06-03T15:57:36 might as well use the us3 bsp as well, who knows maybe someone in the community will try it on real hardware *hides under the table* 2010-06-03T15:57:56 Yes if the US3BSP works somewhere. Just need to know what it does run on and document that. 2010-06-03T15:58:02 K 2010-06-03T15:58:10 Will also want to have a kit of some sort for the pieces required to run m5 2010-06-03T15:59:01 *nod* I should be able to come up with something. We have the tools and scripts to build and run it, so it isn't too much work to package it 2010-06-03T15:59:55 It looks like I'll have the basic ready queue refactored today or tomorrow 2010-06-03T16:00:24 Made a lot more progress than my timeline expected, of course having hacked through this code before helps. 2010-06-03T16:00:41 So I'll block off some time to set up the sparc64 code for merging. 2010-06-03T16:00:58 Thanks.. it will give me a chance to try Ralf's new RPMs 2010-06-03T16:04:35 * gedare should have proposed porting RTEMS to Sparc64 as a GSOC project 2010-06-03T16:04:36 ;) 2010-06-03T16:04:53 might have had to tweak my moral compass a little bit. 2010-06-03T16:04:55 * DrJoel doesn't think you are at risk of not passing :D 2010-06-03T16:04:58 hehe 2010-06-03T16:05:19 * DrJoel proposed being a student and a mentor and giving myself a passing grade .. but that seemed wrong 2010-06-03T16:05:52 i think i saw you pondering that, a couple months back 2010-06-03T16:07:14 grrr, stupid line breaks breaking sed. 2010-06-03T16:10:26 is paranoia missing it's .scn file on purpose? 2010-06-03T16:11:42 no.. 2010-06-03T16:11:52 i don't see it ;) 2010-06-03T16:12:32 must not be there... 2010-06-03T16:26:57 *** kiwichris has joined #rtems 2010-06-03T16:27:13 alright! well the ready queue has been mostly captured now i hope. 2010-06-03T16:27:17 Morning 2010-06-03T16:27:19 hi chris 2010-06-03T16:27:25 gedare, hi 2010-06-03T16:27:42 just about to call it an evening. :) 2010-06-03T16:28:00 I hope it is a pleasant one. 2010-06-03T16:28:15 thanks, and you have a nice day. 2010-06-03T16:28:48 night drjoel 2010-06-03T16:29:32 night 2010-06-03T16:29:41 DrJoel, bye 2010-06-03T16:30:52 DrJoel, can gcc-testing be easily changed to rtems-testing ? 2010-06-03T16:30:58 *** bubaflub has quit IRC 2010-06-03T16:31:18 yes... just a move in the repository 2010-06-03T16:31:35 What needs to change locally for users with a checkout ? 2010-06-03T16:31:59 Change the Root or Repo file in each CVS directory to change from gcc-testing to rtems-testing 2010-06-03T16:32:11 Ok. 2010-06-03T16:32:36 I will hack a script today and send to you or the list if you make a post about the change. 2010-06-03T16:34:11 *** gedare has quit IRC 2010-06-03T16:38:49 For test failures I think we need to mod the screens to define a number of variable fields 2010-06-03T16:39:05 that makes sense 2010-06-03T16:39:14 then have something to parse the screen 2010-06-03T16:39:35 This means the tests should copy the screen file to the location of the executable. 2010-06-03T16:39:57 This includes the install phase. 2010-06-03T16:44:01 I think they may have done that eons ago 2010-06-03T16:44:41 What is that saying about destined to repeat ourselves :) 2010-06-03T16:45:06 email sent and I wrote the script. Not too hard 2010-06-03T16:45:17 Many thanks. 2010-06-03T16:45:46 I have to replace a ground fault outlet which has died on our outdoor pond tonight but think I will start cutting the branch late tonight or in the morning 2010-06-03T16:46:17 there is a class next week so I want to cut it and do some builds. I know Ralf likes to delete targets which don't have tools at this point. So that has to happen after branching 2010-06-03T16:46:22 I am ready. Maybe a heads up on the main list ? 2010-06-03T16:46:31 yeah 2010-06-03T16:46:36 Great. 2010-06-03T16:46:56 Should the miniIMFS RFC be posted to the main list ? 2010-06-03T16:47:07 :) 2010-06-03T16:53:31 DrJoel, we can now hide the ops handlers for file systems and remove the includes all over the place for them. 2010-06-03T16:54:01 sure .. if they are private to a fs, no reason to publicize them 2010-06-03T16:54:31 The interface is now a string label. 2010-06-03T17:00:03 DrJoel, is the "tests" component in bugzilla ok for "rtems-testing" ? 2010-06-03T17:01:34 I don't know.. I tend to think not.. what do you think? 2010-06-03T17:02:08 I am not sure. Should "tests" be renamed "testing" and this can include the tests ? 2010-06-03T17:02:20 I think "tests" and "testing" is confusing. 2010-06-03T17:02:43 agreed 2010-06-03T17:04:00 *** exception13 has left #rtems 2010-06-03T17:04:28 So "tests" -> "testing" and the description be "RTEMS Testing support scripts, tests, and samples. This include the 'rtems-testing' module." 2010-06-03T17:08:10 Sure.. if you explain it that clearly. :) 2010-06-03T17:08:53 Is the description I put clear ? Take a look in bugzilla. I have made the change :) 2010-06-03T17:10:27 looks good but my name shows up a lot for responsible LOL but I have learned to reassign quickly 2010-06-03T17:11:28 Yes. I took the filesystem. I should take the shell, oh I have that already. 2010-06-03T17:12:08 they are pretty spread out.. we just need more core maintainers. 2010-06-03T17:12:17 with write permission. 2010-06-03T17:17:45 Yes. With commits and bug on lists this has made visible all the work that goes on. 2010-06-03T17:18:24 *** DrJoel has quit IRC 2010-06-03T17:37:57 *** cdcs has joined #rtems 2010-06-03T17:45:06 *** cdcs has quit IRC 2010-06-03T18:51:16 *** kiwichris has quit IRC 2010-06-03T19:57:22 *** kiwichris has joined #rtems 2010-06-03T20:04:52 *** DrJoel has joined #rtems 2010-06-03T20:04:52 *** ChanServ sets mode: +o DrJoel 2010-06-03T20:05:58 DrJoel, should I add 4.11 (CVS) as a version to bugzilla ? 2010-06-03T20:09:06 yes, please 2010-06-03T20:09:14 WIll do. 2010-06-03T20:09:53 Michele and I replaced the ground fault outlet on the pond.. tough when the outlet is 3 inches off the ground in the stones around the pond .. and by stones I mean ROCKS! 2010-06-03T20:10:04 running coverage on erc32 overnight.. will branch in AM 2010-06-03T20:10:17 I have been doing "4.11 (CVS)" rather than CVS or HEAD. Should I keep this ? The downside is anything that was entered at 4.10 or "4.10 (CVS)" will be left at that version. 2010-06-03T20:10:37 PRs? 2010-06-03T20:10:45 Well done. It is sooooo wet here no one would outside and not near power. 2010-06-03T20:11:12 Yes. If you look at the PR entered during 4.10 development, they will stay pointing to that version. 2010-06-03T20:11:38 We now move to 4.11. The problem is the issues may still be in CVS. 2010-06-03T20:12:07 I think we should default to "HEAD". 2010-06-03T20:12:20 That would be a nice solution.. 2010-06-03T20:12:28 We have 4.11 as a milestone. 2010-06-03T20:12:40 when we spot the PRs pointing as 4.10 (CVS), we will just have to change them 2010-06-03T20:13:00 Yes. I will rename 4.10 CVS to just 4.10 now. 2010-06-03T20:13:19 so are we just days away from starting to discuss testing goals for 4.11? :-D 2010-06-03T20:13:35 I would like to know what is required testing to get the cpukit done 2010-06-03T20:13:39 An integration issues for big changes. 2010-06-03T20:14:04 yes.. that needs to be discussed. USB is one big one coming 2010-06-03T20:14:08 Yeap. 2010-06-03T20:14:27 I would like to progress the testing items we have discussed. 2010-06-03T20:14:28 I need to do a write up on LWIP so everyone knows it isn't a toy.. used very heavily 2010-06-03T20:14:44 For 4.10 release I need to update the filesystem doco. 2010-06-03T20:14:58 I noticed it has structures etc that need updating. 2010-06-03T20:15:07 I think having a test plan would go a long way to helping. we can start with what we do now, then add "could do" items and move them up as they get done 2010-06-03T20:15:43 Could the test plan be a testing model for RTEMS which is applied to all releases ? 2010-06-03T20:16:01 with the class next week, I just want to get the branch cut and some builds started on 4.10 and 4.11. I won't do much at all next week 2010-06-03T20:16:29 Sure. Could the testing plan be a textinfo doc and something that lives for ever ? 2010-06-03T20:17:20 I think it should be. It is a Project Procedures document. We want it to be something we are proud of and can help us decide when a release should occur 2010-06-03T20:18:09 Yes that is an important section. It can also over public access to information, eg commit logs, bug lists. 2010-06-03T20:18:27 Plus the daily builder process. 2010-06-03T20:18:53 These are large changes in the work that underpins the 4.10 release. 2010-06-03T20:19:46 Indeed a lot has changed and improved. The coverage testing, your building, the commit ml, the tools testing, 2010-06-03T20:20:08 The simulator testing 2010-06-03T20:20:46 yes.. and we have to get the daily builder running the tests on all the simulators.. that would be a major milestone forward 2010-06-03T20:21:02 Yes, This is a 4.11 PR under testing. 2010-06-03T20:21:29 IRC and IRC logs is another item. 2010-06-03T20:22:34 bugzilla updated. 2010-06-03T20:23:28 PR 1255 is a big item for 4.11 2010-06-03T20:24:02 yes. It needs attention 2010-06-03T20:28:04 I will post an email about soon'ish for discussion. 2010-06-03T20:29:42 Thanks.. I think we need to set some concrete goals for 4.11 soon and work to them. When they are met, release. Test plan goals, features, etc 2010-06-03T20:29:54 since there are a number of features in the queue, that part is not the delay.. 2010-06-03T20:30:02 lwip, usb, sparc64, 2010-06-03T20:30:09 I would a PR for each one. 2010-06-03T20:30:26 dynamic loading, SMP progress, etc 2010-06-03T20:30:49 We should then be able to look at bugzilla and ask for milestone 4.11 bugs and know what needs to be done. 2010-06-03T20:30:53 I will ask gedare to file a PR for sparc64. I intend to merge that after the class. it is pretty non-controversial 2010-06-03T20:31:01 that would be a first. :) 2010-06-03T20:31:10 :) 2010-06-03T20:31:12 I hate to run but I should be helping Michele with dinner 2010-06-03T20:31:17 I may be back on .. 2010-06-03T20:31:21 He can then file the patch and it can be reviewed. 2010-06-03T20:31:43 I need to pop out for a while 2010-06-03T20:41:44 *** kiwichris has quit IRC 2010-06-03T22:31:05 *** kiwichris has joined #rtems 2010-06-03T23:16:36 *** kiwichris has quit IRC 2010-06-04T01:24:24 *** lcpfnvc has joined #rtems 2010-06-04T01:28:06 morning 2010-06-04T01:31:37 morning 2010-06-04T02:44:20 *** exception13 has joined #rtems 2010-06-04T02:47:25 *** cdcs has joined #rtems 2010-06-04T02:56:32 *** lekernel has joined #rtems 2010-06-04T04:21:06 *** lekernel has quit IRC 2010-06-04T04:25:43 *** aniceberg has joined #rtems 2010-06-04T04:33:41 *** tuxmaniac has joined #rtems 2010-06-04T06:09:29 *** peerst has joined #rtems 2010-06-04T06:09:34 hi 2010-06-04T06:30:52 *** exception13 has left #rtems 2010-06-04T07:29:40 * DrJoel waves hi before he logs off and moves to the office... will be back in about 45 minutes 2010-06-04T07:29:46 *** DrJoel has quit IRC 2010-06-04T07:30:19 *** exception13 has joined #rtems 2010-06-04T08:45:24 *** lekernel has joined #rtems 2010-06-04T08:53:39 *** DrJoel has joined #rtems 2010-06-04T08:53:39 *** ChanServ sets mode: +o DrJoel 2010-06-04T09:08:42 * DrJoel begins tagging the 4.10 branch.. 2010-06-04T09:16:26 Hello DrJoel 2010-06-04T09:17:55 hello aniceberg .. how are things going for you? 2010-06-04T09:18:10 is the fifo/pipe filling in? 2010-06-04T09:18:26 DrJoel, done with the exams... more focus on coding now... :) 2010-06-04T09:18:30 oh yes... 2010-06-04T09:18:48 its been going well... was documenting certain things now... whilst the coverage is running 2010-06-04T09:19:00 DrJoel, the .cov now is as large as 2.2GB 2010-06-04T09:19:27 which was not the case earlier... and the coverage analysis takes a huge time now. 2010-06-04T09:20:09 for pc386? Are you using the qemu from the couverture git repo? They moved their code to a git repo. Bugs in qemu can do that. 2010-06-04T09:21:26 the couverture i have is not from the git repo 2010-06-04T09:22:05 they just fairly recently did this. But I did see large repos when using a buggy version 2010-06-04T09:23:40 oh. i think i am using the older version... 2010-06-04T09:23:44 i will check this. 2010-06-04T09:32:07 a question from jennifer and I. She is going to go through the ranges and put together a list of what order to do them in. She will give some explanations. Would you like the "work list" to go on the wiki? 2010-06-04T09:33:16 How about here: http://wiki.rtems.org/wiki/index.php/RTEMS_Coverage_Analysis#Current_Status 2010-06-04T09:40:13 *** lekernel has quit IRC 2010-06-04T09:50:49 *** alexp0205 has joined #rtems 2010-06-04T09:54:38 *** lcpfnvc has quit IRC 2010-06-04T09:57:25 *** bubaflub has joined #rtems 2010-06-04T10:01:45 DrJoel, was away... 2010-06-04T10:02:05 DrJoel, sure... that would also help me in creating a concreate schedule too 2010-06-04T10:08:28 ok.. Sending Jennifer the link 2010-06-04T10:09:59 DrJoel, I am now checking the changes and sending them. should be done in some time, if the qemu is quick now. 2010-06-04T10:10:49 Just attach it to the PR. 2010-06-04T10:11:00 I look forward to seeing a big jump. How much do you think you are getting? 2010-06-04T10:11:49 1% is already there... 2010-06-04T10:12:34 That's awesome! I think you will cross 90% this summer 2010-06-04T10:12:49 fifo_open, with one task should be doing that much. a little more should not be a problem with 2 tasks on 2010-06-04T10:15:25 sebhub thinks there is some questionable code in pipe/fifo so keep your eyes open.. you may want to check with him 2010-06-04T10:16:33 *** alexp0205 has quit IRC 2010-06-04T10:16:37 sure I will... i did come across this in the irclogs 2010-06-04T10:19:55 *** alexp0205 has joined #rtems 2010-06-04T10:26:17 *** aniceberg has quit IRC 2010-06-04T10:43:11 *** alexp0205 has quit IRC 2010-06-04T11:01:58 *** alexp0205 has joined #rtems 2010-06-04T12:12:03 *** alexp0205 has left #rtems 2010-06-04T12:48:24 *** madrazr has joined #rtems 2010-06-04T13:21:50 *** lekernel has joined #rtems 2010-06-04T14:33:58 * DrJoel waves 2010-06-04T14:35:03 * cdcs waves back 2010-06-04T14:35:08 *** tuxmaniac has quit IRC 2010-06-04T14:35:52 * DrJoel thinks the 4.10 branch is cut... yeah! 2010-06-04T14:36:11 my development tree is switched to 4.11 and I have a 4.10 working area set up. 2010-06-04T14:36:29 *** DrJoel has quit IRC 2010-06-04T14:59:33 *** madrazr has quit IRC 2010-06-04T15:19:32 *** cdcs has quit IRC 2010-06-04T15:31:00 *** cdcs has joined #rtems 2010-06-04T15:59:31 *** cdcs has quit IRC 2010-06-04T16:14:19 *** cdcs has joined #rtems 2010-06-04T16:30:04 *** lekernel has quit IRC 2010-06-04T16:31:29 *** cdcs_ has joined #rtems 2010-06-04T16:31:29 *** cdcs has quit IRC 2010-06-04T16:33:51 *** cdcs_ is now known as cdcs 2010-06-04T16:50:05 *** bubaflub has quit IRC 2010-06-04T16:50:19 *** cdcs has quit IRC 2010-06-04T17:12:42 *** gedare has joined #rtems 2010-06-04T17:35:39 *** exception13 has left #rtems 2010-06-04T18:13:48 *** bubaflub has joined #rtems 2010-06-04T18:21:22 *** gedare has quit IRC 2010-06-04T21:13:48 *** bubaflub has left #rtems 2010-06-04T22:09:31 *** zwj has joined #rtems 2010-06-04T23:32:28 *** zwj has quit IRC 2010-06-05T01:45:10 *** zwj has joined #rtems 2010-06-05T02:08:35 *** madrazr has joined #rtems 2010-06-05T03:16:19 *** cdcs has joined #rtems 2010-06-05T03:18:20 *** cdcs is now known as cdcs_a 2010-06-05T03:20:21 *** madrazr has quit IRC 2010-06-05T04:13:16 *** zwj has quit IRC 2010-06-05T04:17:20 *** zwj has joined #rtems 2010-06-05T04:33:33 *** lekernel has joined #rtems 2010-06-05T04:56:09 *** lekernel has quit IRC 2010-06-05T05:24:05 *** zwj has quit IRC 2010-06-05T05:29:10 *** zwj has joined #rtems 2010-06-05T05:45:05 *** exception13 has joined #rtems 2010-06-05T06:15:09 *** zwj has quit IRC 2010-06-05T06:19:43 *** zwj has joined #rtems 2010-06-05T06:54:08 *** DrJoel has joined #rtems 2010-06-05T07:06:17 *** aniceberg has joined #rtems 2010-06-05T07:14:15 hi aniceberg how are you? 2010-06-05T07:14:50 DrJoel, I am fine. How are you? 2010-06-05T07:15:02 I thought you were not here. :DS 2010-06-05T07:16:08 I am in Georgia in the hotel getting ready to go to the Faire. My wife is doing some repair on her dress and my daughter's dress. 2010-06-05T07:16:23 Pics will be on facebook tonight or tomorrow 2010-06-05T07:16:40 DrJoel, the couverture from the git repository does not take the -trace option... and the svn does not have the qemu-r5688 directory LOL 2010-06-05T07:17:16 Did you switch the branch? I don't remember how. 2010-06-05T07:17:45 join their mailing list and ask. But verify you used the right qemu binary. I know I end up with the wrong one a lot 2010-06-05T07:18:17 * DrJoel is grabbing a shower 2010-06-05T07:18:53 was using from the svn till some time and this one, spits out 2.2G of trace :D. moved to git, and this one does not take the -trace option. I got the one from the rtems site now. 2010-06-05T07:19:04 that should work :) 2010-06-05T07:46:06 we have a qemu on rtems? mwalle has a nice qemu at github. That's where I get the source for the lm32 version with trace 2010-06-05T07:46:44 you had given me the link to qemu @ private/coverage :) 2010-06-05T08:20:25 *** DrJoel has quit IRC 2010-06-05T08:27:11 *** DrJoel has joined #rtems 2010-06-05T08:31:51 *** DrJoel has quit IRC 2010-06-05T08:38:33 *** DrJoel has joined #rtems 2010-06-05T08:44:25 *** DrJoel has quit IRC 2010-06-05T09:16:45 *** dr__house has joined #rtems 2010-06-05T09:19:10 *** zwj has quit IRC 2010-06-05T09:19:47 *** zwj has joined #rtems 2010-06-05T09:59:27 *** zwj has quit IRC 2010-06-05T10:01:46 *** zwj has joined #rtems 2010-06-05T10:22:21 *** gedare has joined #rtems 2010-06-05T10:22:51 * dr__house finally got a clean build of RTEMS from source. 2010-06-05T10:22:54 * dr__house is happy 2010-06-05T10:23:37 :) 2010-06-05T10:27:02 *** zwj has quit IRC 2010-06-05T10:29:03 some long pending work beckons 2010-06-05T10:31:46 *** zwj has joined #rtems 2010-06-05T10:54:41 anybody online who groks the rtems make system (for the application not for building rtems I mean) 2010-06-05T10:58:28 *** aniceberg has quit IRC 2010-06-05T11:52:16 *** dr__house has quit IRC 2010-06-05T12:30:57 *** gedare has quit IRC 2010-06-05T12:46:03 *** Fallenou has quit IRC 2010-06-05T12:56:32 *** dr__house has joined #rtems 2010-06-05T13:07:15 *** madrazr has joined #rtems 2010-06-05T13:07:36 *** madrazr has left #rtems 2010-06-05T13:16:06 *** zwj has quit IRC 2010-06-05T13:32:11 *** Fallenou has joined #rtems 2010-06-05T14:08:33 *** gedare has joined #rtems 2010-06-05T14:14:23 *** gedare has quit IRC 2010-06-05T16:28:30 *** exception13 has left #rtems 2010-06-05T17:18:37 *** dr__house has quit IRC 2010-06-05T17:44:07 *** gedare has joined #rtems 2010-06-05T17:44:34 *** aniceberg has joined #rtems 2010-06-05T17:52:25 *** gedare has quit IRC 2010-06-05T18:07:09 *** cdcs_a has quit IRC 2010-06-05T18:07:10 *** cdcs has joined #rtems 2010-06-05T18:38:41 *** aniceberg has quit IRC 2010-06-05T18:49:55 *** cdcs has quit IRC 2010-06-05T20:23:34 *** zwj has joined #rtems 2010-06-05T22:01:23 *** zwj has quit IRC 2010-06-05T22:40:13 *** zwj has joined #rtems 2010-06-06T00:18:34 *** dr__house has joined #rtems 2010-06-06T00:28:03 *** zwj has quit IRC 2010-06-06T00:31:03 *** zwj has joined #rtems 2010-06-06T01:16:49 *** zwj has quit IRC 2010-06-06T01:17:36 *** zwj has joined #rtems 2010-06-06T01:19:41 *** zwj has quit IRC 2010-06-06T01:19:59 *** zwj has joined #rtems 2010-06-06T02:10:10 *** zwj has quit IRC 2010-06-06T02:48:18 *** exception13 has joined #rtems 2010-06-06T03:02:43 *** dr__house has quit IRC 2010-06-06T03:37:46 *** dr__house has joined #rtems 2010-06-06T04:45:05 *** lekernel has joined #rtems 2010-06-06T05:25:46 *** dr__house has quit IRC 2010-06-06T05:44:32 *** tuxmaniac has joined #rtems 2010-06-06T07:41:54 hi 2010-06-06T07:42:23 is it standard time (UTC - 6) or daylight saving time (UTC - 5) right now in Alabama ? 2010-06-06T08:03:56 *** aniceberg has joined #rtems 2010-06-06T08:06:03 *** cdcs has joined #rtems 2010-06-06T08:38:19 *** dr__house has joined #rtems 2010-06-06T08:44:38 *** exception13 has left #rtems 2010-06-06T09:08:54 *** exception13 has joined #rtems 2010-06-06T09:39:01 *** aniceberg has quit IRC 2010-06-06T11:01:39 *** gedare has joined #rtems 2010-06-06T11:10:08 *** cdcs has quit IRC 2010-06-06T11:11:46 *** dr__house has quit IRC 2010-06-06T11:23:18 *** dr__house has joined #rtems 2010-06-06T11:37:41 *** Raj_ has joined #rtems 2010-06-06T11:44:38 *** cdcs has joined #rtems 2010-06-06T11:48:19 *** Raj_ has quit IRC 2010-06-06T12:30:08 *** gedare has quit IRC 2010-06-06T12:55:50 *** madrazr has joined #rtems 2010-06-06T14:22:45 *** krzabr has joined #rtems 2010-06-06T14:32:03 *** krzabr has quit IRC 2010-06-06T14:33:52 *** cdcs has quit IRC 2010-06-06T14:38:58 *** madrazr has quit IRC 2010-06-06T14:40:13 *** tuxmaniac has quit IRC 2010-06-06T15:41:51 *** dr__house has quit IRC 2010-06-06T16:24:57 *** cdcs has joined #rtems 2010-06-06T16:31:55 *** exception13 has left #rtems 2010-06-06T17:40:33 *** cdcs has quit IRC 2010-06-06T18:55:52 *** gedare has joined #rtems 2010-06-06T19:40:11 *** gedare has quit IRC 2010-06-06T19:59:29 *** DrJoel has joined #rtems 2010-06-06T19:59:29 *** ChanServ sets mode: +o DrJoel 2010-06-06T19:59:47 *** DrJoel has quit IRC