2010-05-24T00:40:55 *** n3oo3n has joined #rtems 2010-05-24T01:53:20 *** madrazr has joined #rtems 2010-05-24T02:01:52 morning 2010-05-24T02:28:53 *** lekernel has quit IRC 2010-05-24T02:42:10 *** lekernel has joined #rtems 2010-05-24T02:42:34 hi ! 2010-05-24T02:42:40 happy birthday DrJoel ! 2010-05-24T02:43:00 Is it Joel's bday ? 2010-05-24T02:43:07 i thinkso 2010-05-24T02:43:37 It is 2:45am where he. 2010-05-24T02:43:53 yes 2010-05-24T02:45:35 he may be asleep :p 2010-05-24T02:45:50 I think so. 2010-05-24T02:46:45 *** lekernel has quit IRC 2010-05-24T02:58:16 *** madrazr has quit IRC 2010-05-24T03:00:00 *** lekernel has joined #rtems 2010-05-24T03:35:44 *** exception13 has joined #rtems 2010-05-24T03:58:34 *** madrazr has joined #rtems 2010-05-24T04:19:19 *** lekernel has quit IRC 2010-05-24T04:30:32 *** tuxmaniac has joined #rtems 2010-05-24T04:32:26 *** lekernel has joined #rtems 2010-05-24T05:03:39 happy bday DrJoel !!! :P 2010-05-24T05:19:34 *** aniceberg has joined #rtems 2010-05-24T05:44:34 *** lekernel has quit IRC 2010-05-24T05:57:03 *** lekernel has joined #rtems 2010-05-24T06:03:17 *** zwj has joined #rtems 2010-05-24T06:31:25 *** aniceberg has quit IRC 2010-05-24T06:33:06 thanks for the birthdya wishes everyone. 2010-05-24T06:33:48 We went to a funeral about 120 miles (200 km) from home over the weekend and I just crashed when last night 2010-05-24T06:34:54 *** madrazr has quit IRC 2010-05-24T06:37:09 Happy birthday JOel 2010-05-24T06:41:13 thanks chris.. Michele had made crab cakes and after that it was just bed time 2010-05-24T07:07:53 *** lekernel has quit IRC 2010-05-24T07:16:14 *** exception13 has quit IRC 2010-05-24T07:16:15 *** mwalle has quit IRC 2010-05-24T07:16:16 *** tuxmaniac has quit IRC 2010-05-24T07:16:16 *** DrJoel has quit IRC 2010-05-24T07:16:16 *** _Lucretia_ has quit IRC 2010-05-24T07:16:17 *** tfinn has quit IRC 2010-05-24T07:16:17 *** Fallenou_ has quit IRC 2010-05-24T07:16:21 *** tiemen has quit IRC 2010-05-24T07:16:22 *** n3oo3n has quit IRC 2010-05-24T07:16:24 *** zwj has quit IRC 2010-05-24T07:16:25 *** dukeleto has quit IRC 2010-05-24T07:16:26 *** kiwichris has quit IRC 2010-05-24T07:16:28 *** lcpfnvc has quit IRC 2010-05-24T07:16:28 *** ChanServ has quit IRC 2010-05-24T07:16:29 *** haxomophone has quit IRC 2010-05-24T07:16:29 *** kuzew has quit IRC 2010-05-24T07:18:14 *** madrazr has joined #rtems 2010-05-24T07:18:14 *** zwj has joined #rtems 2010-05-24T07:18:14 *** tuxmaniac has joined #rtems 2010-05-24T07:18:14 *** exception13 has joined #rtems 2010-05-24T07:18:14 *** n3oo3n has joined #rtems 2010-05-24T07:18:14 *** kiwichris has joined #rtems 2010-05-24T07:18:14 *** DrJoel has joined #rtems 2010-05-24T07:18:14 *** mwalle has joined #rtems 2010-05-24T07:18:14 *** _Lucretia_ has joined #rtems 2010-05-24T07:18:14 *** tfinn has joined #rtems 2010-05-24T07:18:14 *** Fallenou_ has joined #rtems 2010-05-24T07:18:14 *** haxomophone has joined #rtems 2010-05-24T07:18:14 *** dukeleto has joined #rtems 2010-05-24T07:18:14 *** lcpfnvc has joined #rtems 2010-05-24T07:18:14 *** kuzew has joined #rtems 2010-05-24T07:18:14 *** tiemen has joined #rtems 2010-05-24T07:18:14 *** ChanServ has joined #rtems 2010-05-24T07:18:14 *** barjavel.freenode.net sets mode: +oo DrJoel ChanServ 2010-05-24T07:22:30 *** lekernel has joined #rtems 2010-05-24T07:32:04 *** DrJoel has quit IRC 2010-05-24T07:44:56 *** dr__house has joined #rtems 2010-05-24T07:48:54 *** aniceberg has joined #rtems 2010-05-24T07:58:54 *** lekernel has quit IRC 2010-05-24T08:02:11 *** Raj_ has joined #rtems 2010-05-24T08:03:24 *** exception13 has left #rtems 2010-05-24T08:10:19 *** lekernel has joined #rtems 2010-05-24T08:12:15 *** DrJoel has joined #rtems 2010-05-24T08:12:15 *** ChanServ sets mode: +o DrJoel 2010-05-24T08:14:51 DrJoel: Happy birthday!! 2010-05-24T08:16:35 DrJoel, Happy birthday! 2010-05-24T08:22:53 *** dr__house has quit IRC 2010-05-24T08:26:05 *** aniceberg has quit IRC 2010-05-24T08:26:25 *** aniceberg has joined #rtems 2010-05-24T08:27:43 thanks all! did you see the pictures of the balloon pit? Bazinga! 2010-05-24T08:27:58 * aniceberg waves at DrJoel 2010-05-24T08:28:29 Happy birthday DrJoel 2010-05-24T08:28:36 aniceberg: :) 2010-05-24T08:29:01 aniceberg: Jiri Gaisler emailed me a new test version of tsim with branch taken/not taken info for erc32 and leon3!! Running it for the first time now 2010-05-24T08:30:48 DrJoel, neat! But, is qemu accurate with the taken / not taken information? I felt the results were a bit weird, while with _rename_r 2010-05-24T08:35:31 *** lekernel has quit IRC 2010-05-24T08:43:23 aniceberg: I haven't compared the tests for _rename_r to the results but the branches info for the cpus we looked at made sense given the block of uncovered code. 2010-05-24T08:44:14 aniceberg: there may be issues. We haven't examined every case in detail. If you can provide a link to something you doubt in the runs from this weekend, please pass it along 2010-05-24T08:45:12 DrJoel, sure. I will have a look at the reports and let you know. a 28 byte uncovered code in _rename_r is what I am not clear about 2010-05-24T08:51:19 that's the one we used to see if the branch reports were right. It was "not taken/not taken/taken" followed by that range. That makes sense by itself but we need to look at the tests 2010-05-24T08:51:37 *** dr__house has joined #rtems 2010-05-24T08:56:39 *** exception13 has joined #rtems 2010-05-24T08:56:58 DrJoel, yes. true, it did make sense by itself, but what I feel after looking at the get_start_loc and the test... 2010-05-24T08:57:53 there is a case where "10aaaf: 3c 2f cmp $0x2f,%al" needs to be taken, covering those 14 / 28 bytes that are uncovered. 2010-05-24T08:58:18 can you provide the link to the report here? 2010-05-24T08:59:06 I am right now looking at http://www.rtems.org/ftp/pub/rtems/people/joel/coverage/pc386-OsPD-20100521-2144/annotated.html#range293 2010-05-24T08:59:41 ok.. scroll up a few instructions 2010-05-24T08:59:44 the second case, the comparison with 2F(/) should have been taken during the runs, is what i feel 2010-05-24T08:59:49 yes 2010-05-24T08:59:49 10aaad: 74 08 je 10aab7 <_rename_r+0x2b> <== NEVER TAKEN 2010-05-24T08:59:49 10aaaf: 3c 2f cmp $0x2f,%al 2010-05-24T08:59:50 10aab1: 74 04 je 10aab7 <_rename_r+0x2b> <== NEVER TAKEN 2010-05-24T08:59:50 10aab3: 84 c0 test %al,%al 2010-05-24T08:59:50 10aab5: 75 0e jne 10aac5 <_rename_r+0x39> <== ALWAYS TAKEN 2010-05-24T08:59:50 10aab7: 8d 7d bc lea -0x44(%ebp),%edi <== NOT EXECUTED 2010-05-24T09:01:40 did you pass in a relative path or do you only have an absolute path test test 2010-05-24T09:02:30 there is an absolute path in the test: /tmp.. 2010-05-24T09:02:52 in fact, there are several tests, that pass the absolute path 2010-05-24T09:04:36 Be 0x2f is / so it is testing for absolute/relative path 2010-05-24T09:04:41 what is 0x5f a test for 2010-05-24T09:05:22 sorry 0x5c is "\" 2010-05-24T09:05:54 yes the '\' 2010-05-24T09:05:56 so that is a separator test. It is also supporting DOS style path separators. 2010-05-24T09:06:36 Do you know what would help all this? Not inlining these methods. Then they would only need to be tested once in one place 2010-05-24T09:06:41 *** Raj_ has quit IRC 2010-05-24T09:07:06 yes, so, in this case, the case with 2f had to match 2010-05-24T09:08:02 yep. the leading character was never "\". 2010-05-24T09:08:02 true... not inlining could have helped here 2010-05-24T09:08:20 not inlining will also help everywhere they are used! 2010-05-24T09:08:44 oh yes, everywhere. :) 2010-05-24T09:09:37 turn rtems_filesystem_is_separator rtems_filesystem_get_start_loc and rtems_filesystem_get_sym_start_loc into methods into their own C files. One method per file. Then ensure those three routines are fully tested. I think you will be surprised how many cases this eliminates and how much the code shrinks 2010-05-24T09:12:04 DrJoel, two general questions 1. why would one want to inline if the code is larger than the usual inline functions, a case like this one? any specific reason? 2010-05-24T09:13:19 2. one method per file? and not all the 3 in one file? 2010-05-24T09:22:26 Sorry .. phone 2010-05-24T09:22:49 thats ok DrJoel 2010-05-24T09:23:21 1. inlining is usually faster than making a subroutine call. But inlining code which is large can lead to code size explosion. And inlining code with multiple branch paths leads to path explosion. You have to test all those branches in every case. That is the evil in this case. 2010-05-24T09:24:06 rtems_filesystem_is_separator is called in who knows how many places. That's 3 paths every time 2010-05-24T09:24:47 2. one method per file is a pain but it lets the linker automatically drop code out for us when it isn't referenced 2010-05-24T09:25:56 1. clear. 2. clear, but where do i need to place those source files? 2010-05-24T09:26:46 I think in libcsupport/ 2010-05-24T09:27:08 ok... 2010-05-24T09:27:29 I think you will be VERY surprised at the change in numbers after this. Size, branches and coverage 2010-05-24T09:28:52 i wish to be :) 2010-05-24T09:32:52 rtems_filesystem_get_start_loc is called 8 times itself. That is 3 branches times 8 ==> 24. After this, it will be 3 + 8 ==> 11 2010-05-24T09:33:13 that doesn't count where _is_separator is used itself 2010-05-24T09:34:29 Oh yes, thats a lot of covered code and reduced size! 2010-05-24T09:35:37 and you can focus on getting that one uncovered range in _rename_r and the supporting routines and they are gone. :-D 2010-05-24T09:37:03 yes. no change in the test code, but there will be completely covered code. :) 2010-05-24T09:37:32 Oh. another question, how is \0 a separator DrJoel ? 2010-05-24T09:39:08 it is just the end of the filename so needs to terminate evaluation of the name. May not really be needed but I don't know the impact of changing it 2010-05-24T09:40:54 the question arises since i need to pass a "" to _rename_r to hit this case, if thats necessary 2010-05-24T09:41:14 *** bubaflub has joined #rtems 2010-05-24T09:45:40 *** Raj_ has joined #rtems 2010-05-24T09:46:06 DrJoel, hi 2010-05-24T09:53:04 DrJoel, I am presently working on the fifo and the open() call returns with a EINTR. On going down the stack, I see that the call to rtems_barrier_create from fifo_open() throws a RTEMS_TOO_MANY. Is there any configuration to be done before using the the barriers? 2010-05-24T09:54:02 #define CONFIGURE_MAXIMUM_BARRIERS XX 2010-05-24T09:54:33 EINTR sounds like the wrong error for that .. should probably be ENOMEM or similar. check opengroup 2010-05-24T09:57:06 DrJoel, I have sent you the patches for lchown and utimes 2010-05-24T09:58:47 Raj_: I saw while out of town this weekend. Need to review them. But about to leave the office for a while. They are in the queue. :) 2010-05-24T09:59:50 Sure, just wanted to check that you got them :) 2010-05-24T10:00:27 *** madrazr has quit IRC 2010-05-24T10:00:28 DrJoel, what is the next thing to do regarding the work ? 2010-05-24T10:01:15 compilation and running is fine now, I guess, just need some fine tuning maybe ! 2010-05-24T10:01:57 Shall we move onto testing using the gccgo testsuite ? 2010-05-24T10:03:20 Raj_: yes. And working to get whatever patches you have merged 2010-05-24T10:03:25 have you started the FSF paperwork? 2010-05-24T10:08:01 DrJoel, yes, I sent the email out. They will send a hard-copy now I think ? 2010-05-24T10:08:16 I got their response. 2010-05-24T10:08:18 *** alexp0205 has joined #rtems 2010-05-24T10:08:27 right. It isn't a quick process. :D 2010-05-24T10:08:37 I haven't heard anything about the Google CLA though ! 2010-05-24T10:10:10 So, we can start with the testsuite, great :) ! 2010-05-24T10:11:09 DrJoel, it is throwing out a EINTR, opengroup says, it should be ENOMEM if all resources required were not allocated 2010-05-24T10:12:28 aniceberg: I suspect that. :) 2010-05-24T10:12:47 Raj_: if you did it via their website, you won't. I did it months ago and haven't gotten anything 2010-05-24T10:13:05 Yea, I did it through the website 2010-05-24T10:13:40 Hmm, they have their lawyers go through everything I guess ?! 2010-05-24T10:14:40 DrJoel, do we have to do anything about the strerror_r thing 2010-05-24T10:15:06 newlib doesn't set the errno if the length of the buffer given isn't sufficient 2010-05-24T10:16:53 Submit a patch to newlib. Good catch 2010-05-24T10:17:09 *** alexp0205 has left #rtems 2010-05-24T10:18:30 Ok 2010-05-24T10:19:29 I didn't get why they used the GNU extension version of strerror_r instead of the POSIX one 2010-05-24T10:22:07 no idea myself. Probably just a Linux centric developmetn 2010-05-24T10:22:50 Yea, maybe 2010-05-24T10:26:59 heya folks. looks like GSoC is all in full swing. 2010-05-24T10:27:04 DrJoel: best wishes 2010-05-24T10:28:38 yep. gsoc is full speed now. 2010-05-24T10:28:52 thanks for the wishes 2010-05-24T10:30:24 DrJoel: had a brief chat with gedare wrt my MMU implementation for one of his projects last week 2010-05-24T10:30:33 looks liek he *might* use my code 2010-05-24T10:31:07 * tuxmaniac is yet to get his code merged :( 2010-05-24T10:31:17 DrJoel: any news on Munich class? 2010-05-24T10:33:30 no.. hopefully soon since it is nice to have the travel arrangements settled 2010-05-24T10:34:02 * DrJoel is leaving.. see you all later 2010-05-24T10:34:05 *** DrJoel has quit IRC 2010-05-24T10:58:54 *** aniceberg has quit IRC 2010-05-24T11:02:31 *** Raj_ has quit IRC 2010-05-24T11:21:20 *** zwj has quit IRC 2010-05-24T12:05:11 *** madrazr has joined #rtems 2010-05-24T12:15:59 *** madrazr has left #rtems 2010-05-24T13:00:39 *** krzabr has joined #rtems 2010-05-24T13:01:38 *** tuxmaniac has quit IRC 2010-05-24T13:10:23 *** krzabr has quit IRC 2010-05-24T13:41:10 good afternoon kiwichris 2010-05-24T14:02:52 *** lekernel has joined #rtems 2010-05-24T14:17:27 *** lekernel has quit IRC 2010-05-24T14:42:17 *** dr__house has quit IRC 2010-05-24T15:28:52 *** lekernel has joined #rtems 2010-05-24T15:35:47 *** exception13 has left #rtems 2010-05-24T16:27:36 *** rtemsLogger has joined #rtems 2010-05-24T16:59:10 * tfinn investigates if he can sneak in an update from RTEMS 4.5.1 to the old instrument 2010-05-24T18:38:47 not trivial, unfortunately... 2010-05-24T22:26:06 *** rtemsLogger has joined #rtems 2010-05-24T22:28:48 * kiwichris tests the logging bot 2010-05-24T22:32:19 logging test, please ignore 2010-05-24T22:46:07 *** rtemsLogger has joined #rtems 2010-05-24T22:46:43 * kiwichris tests the logging bot again 2010-05-24T22:49:17 * kiwichris take the logging bot down to sort the logs out 2010-05-24T23:28:06 *** rtemsLogger has joined #rtems 2010-05-24T23:49:27 *** alexp0205 has joined #rtems 2010-05-25T00:08:32 *** alexp0205 has quit IRC 2010-05-25T00:09:07 *** dr__house has joined #rtems 2010-05-25T03:26:34 *** rtemsLogger has joined #rtems 2010-05-25T04:31:15 *** rtemsLogger has joined #rtems 2010-05-25T04:37:19 *** tuxmaniac has joined #rtems 2010-05-25T05:18:30 *** madrazr has quit IRC 2010-05-25T05:30:02 *** madrazr has joined #rtems 2010-05-25T05:36:24 *** lekernel has quit IRC 2010-05-25T05:49:32 *** lekernel has joined #rtems 2010-05-25T05:56:08 *** lekernel has quit IRC 2010-05-25T06:08:02 *** lekernel has joined #rtems 2010-05-25T06:31:52 *** madrazr has quit IRC 2010-05-25T06:40:45 *** zwj has joined #rtems 2010-05-25T06:57:42 *** DrJoel has joined #rtems 2010-05-25T06:57:42 *** ChanServ sets mode: +o DrJoel 2010-05-25T06:58:12 * DrJoel is sure everyone wonders where rtems.org is. 2010-05-25T06:59:05 We had 95F/35C yesterday and there was an electricity failure at the office late in the afternoon. I guess no one checked back on it to turn it on. 2010-05-25T06:59:33 rtems.org is up but the other machines seem to be down 2010-05-25T07:00:17 ahh.. ok.. Jeff knows the order they have to be powered up in. Another hour or so. 2010-05-25T07:00:36 Not a problem for me. I am hacking the IRC logs in rtems.org 2010-05-25T07:00:45 They are on UPSs so I wonder why they didn't reboot. :( 2010-05-25T07:01:03 What about the switches (ethernet) ? 2010-05-25T07:01:06 * DrJoel was tinkering with an Ubuntu virtual machine when the power went out 2010-05-25T07:01:38 Hmmm... probably not on UPS.. the one in the rack with rtems.org is but there may be one for the lab which is not 2010-05-25T07:01:59 They can brown out and get a little silly. 2010-05-25T07:02:16 Jeff and I will have to check the switches from rtems.org to the lab to see what they are plugged into 2010-05-25T07:02:32 Great. 2010-05-25T07:03:06 I sent an email before it went down about ALIGN not using uintptr_t. That is causing fts.c to fail to build on m32c/m32csim 2010-05-25T07:03:46 It would seem an ok change but Ralf is a better person to ask 2010-05-25T07:04:12 I asked on the user's list but without mail.oarcorp.com up, I don't see any replies. I will follow up today 2010-05-25T07:04:32 That was all I really spotted with the warning/errors turned on and networking turned off 2010-05-25T07:04:39 I did not see any and I am not on that mail server. 2010-05-25T07:05:02 Good. There is a new ARM BSP in the tree and it is failing. 2010-05-25T07:06:47 I think I filed a PR on that.. or at least sent sebhub some email 2010-05-25T07:07:03 Ok 2010-05-25T07:12:09 * DrJoel needs to get ready to take Justin to school and head to the office .. be back on from there 2010-05-25T07:12:13 *** DrJoel has quit IRC 2010-05-25T07:22:05 the new arm bsp should be fixed now, i forgot the BSP_SMALL_MEMORY define 2010-05-25T07:22:44 Thanks. We will have to wait until tonight for the build bot to build it. 2010-05-25T07:29:47 *** madrazr has joined #rtems 2010-05-25T07:30:42 *** dr__house has joined #rtems 2010-05-25T07:36:06 *** madrazr has quit IRC 2010-05-25T07:36:18 *** dr__house has quit IRC 2010-05-25T08:05:52 *** exception13 has left #rtems 2010-05-25T08:06:43 *** madrazr has joined #rtems 2010-05-25T08:12:10 *** kiwichris has quit IRC 2010-05-25T08:18:47 *** DrJoel has joined #rtems 2010-05-25T08:18:47 *** ChanServ sets mode: +o DrJoel 2010-05-25T08:31:21 sebhub: Hi ,mentor 2010-05-25T08:34:29 i have make a small patch using sort linkcmd, but i have some question about how to reference to object 2010-05-25T08:49:02 *** Raj_ has joined #rtems 2010-05-25T08:49:39 *** Raj_ has joined #rtems 2010-05-25T08:51:47 hi 2010-05-25T08:52:36 zwj: do you use windows for develoment? 2010-05-25T08:55:56 please use unix end of line 2010-05-25T09:05:32 *** Raj_ has joined #rtems 2010-05-25T09:05:50 *** Raj_ has joined #rtems 2010-05-25T09:08:12 sebhub:hi, i send a email to you 2010-05-25T09:08:26 yes, i already looked at the patch 2010-05-25T09:08:35 it contains files with dos line endings 2010-05-25T09:08:39 i use vmware and fedora 2010-05-25T09:09:10 ao, i made patch in vmware and copy it to windows 2010-05-25T09:10:44 please avoid dos line endings 2010-05-25T09:11:41 ok, i send you another one 2010-05-25T09:12:27 *** Raj_ has quit IRC 2010-05-25T09:27:57 sebhub:do you receive my email? 2010-05-25T09:34:49 yes, but for example sequence_init.h still has dos line endings 2010-05-25T09:38:26 it is strange, i make the patch again 2010-05-25T09:39:16 *** exception13 has joined #rtems 2010-05-25T09:40:35 maybe you need to modify the source file via dos2unix 2010-05-25T09:42:29 i know the reason of it , my modified code is in my windows disk 2010-05-25T09:44:26 ok 2010-05-25T09:44:53 *** Raj_ has joined #rtems 2010-05-25T09:45:45 whether it is ok i just use dos2unix to converter the patch? 2010-05-25T09:46:16 no, you should clean up your source files 2010-05-25T09:46:43 ok 2010-05-25T09:46:55 otherwise you have to take care of this every time you create a patch 2010-05-25T09:50:46 how to use dos2unix to converter the whole dictionary files? 2010-05-25T09:52:04 you need to combine this with find for example 2010-05-25T09:52:17 unfortunately dos2unix has no recursive option 2010-05-25T09:52:29 *** dr__house has joined #rtems 2010-05-25T09:53:19 find . -type f -exec dos2unix {} \; 2010-05-25T09:53:33 *** bubaflub has joined #rtems 2010-05-25T09:53:47 but be careful with the last command 2010-05-25T09:54:14 ok, i try it 2010-05-25T09:56:37 DrJoel: hi 2010-05-25T10:31:43 *** madrazr has quit IRC 2010-05-25T10:34:16 *** sebhub has quit IRC 2010-05-25T11:59:50 *** tuxmaniac has quit IRC 2010-05-25T12:03:57 *** madrazr has joined #rtems 2010-05-25T12:03:58 *** madrazr has joined #rtems 2010-05-25T12:24:31 *** zwj has quit IRC 2010-05-25T13:17:34 *** Raj_ has quit IRC 2010-05-25T15:27:02 *** dr__house has quit IRC 2010-05-25T16:04:27 *** bubaflub has quit IRC 2010-05-25T16:30:17 *** madrazr has quit IRC 2010-05-25T16:53:18 *** DrJoel has quit IRC 2010-05-25T17:08:02 hello rtems folk 2010-05-25T17:35:52 *** bubaflub has joined #rtems 2010-05-25T17:48:34 *** exception13 has left #rtems 2010-05-25T17:50:54 *** kiwichris has joined #rtems 2010-05-25T17:58:37 kiwichris: how ya doing? is your kid ok? 2010-05-25T17:58:44 *** DrJoel has joined #rtems 2010-05-25T17:58:44 *** ChanServ sets mode: +o DrJoel 2010-05-25T17:59:06 Hi and yeah. Just that first winter around of colds. She has a virus. 2010-05-25T17:59:27 Hi Joel. 2010-05-25T17:59:55 hi.. sorry to hear about the winter virus.. 31-35C here 2010-05-25T18:00:39 It is about 15C and wet. It has been wet for over a week now. 2010-05-25T18:01:10 kiwichris: sorry to hear that. lemme know if you ever need to skip a tuesday meeting to take care of family things. 2010-05-25T18:01:24 nasty.. just hot and miserable here 2010-05-25T18:01:51 Sure. It is all sorted at my end. 2010-05-25T18:04:03 DrJoel, can we discuss the mount changes ? 2010-05-25T18:04:26 sure.. I wanted to ask you to review the chown/lchown and utime/utimes mod. Commit if you like 2010-05-25T18:04:41 Sure I will. 2010-05-25T18:04:43 But first .. how's Parrot going? 2010-05-25T18:05:47 dukeleto, are you about ? 2010-05-25T18:07:45 until he gets here, kiwichris i've got a few questions 2010-05-25T18:08:30 bubaflub, go for it 2010-05-25T18:08:31 dukeleto sent me your work that builds against 1.7.0 which looks like a script file, some patches, and a Makefile 2010-05-25T18:08:44 i'd like to give it a whirl, what's the magic command-line invocation? 2010-05-25T18:08:47 (that's question one) 2010-05-25T18:09:10 Not sure. Can you paste it here ? 2010-05-25T18:09:53 sure, i can forward you what dukeleto sent me a while ago 2010-05-25T18:09:54 one sec 2010-05-25T18:10:08 Please send it. 2010-05-25T18:10:57 i believe it's the same .tar.gz that's on the ftp on the rtems site 2010-05-25T18:11:05 is here for a few mins... 2010-05-25T18:11:09 Ok. 2010-05-25T18:11:10 kiwichris: can you review ralf's change for PR1527? https://www.rtems.org/bugzilla/show_bug.cgi?id=1527 2010-05-25T18:11:18 ok .. after meeting 2010-05-25T18:12:00 bubaflub: solving the exit() issue i think is important, there should be a TT 2010-05-25T18:12:01 DrJoel, sure 2010-05-25T18:12:34 dukeleto, TT ? 2010-05-25T18:12:50 ticket on the parrot ticket tracker 2010-05-25T18:12:54 dukeleto: ok, i'll hunt it down 2010-05-25T18:12:57 Ah ok. 2010-05-25T18:13:26 We have 2 separate issues, building parrot and then the runtime which exit is. 2010-05-25T18:14:18 kiwichris: right. since 1.7.0 parrot has gotten rid of some of the recursive make nonsense 2010-05-25T18:14:25 we need to be able to build parrot out of directory, yes? 2010-05-25T18:14:56 In my view the exit one is difficult because some parts of parrot call exist without the interpreter context. 2010-05-25T18:15:24 We tend to build out of the source in RTEMS. This keeps the source clean. This is not a requirement 2010-05-25T18:15:39 Is this what you mean ? 2010-05-25T18:15:41 ok. but we also need to generate the compiler hints 2010-05-25T18:15:52 Or just provide them. 2010-05-25T18:16:00 RTEMS tends to be static. 2010-05-25T18:16:10 right. darbelo gave me some work he did on the sparc hints 2010-05-25T18:16:25 dukeleto: is it ticket #1523? 2010-05-25T18:16:29 We can configure things like networking on or off but we should not consider this for this now. 2010-05-25T18:17:23 kiwichris: ok. 2010-05-25T18:17:32 You can think of RTEMS as a single distribution of Linux that always installs the same way and gives the same results to a configure script. 2010-05-25T18:18:06 kiwichris: ok, that makes sense. 2010-05-25T18:18:15 The hope is avoiding the runtime type tests. RTEMS's tools can perform standard compile and links tests. 2010-05-25T18:18:22 so each RTEMS configuration / bsp will always provide the same configuration information 2010-05-25T18:19:33 Yes for now. A BSP is the low layer and we can say for Parrot you need RTEMS configured in this way. 2010-05-25T18:19:55 if configured the same way (--enable-posix, --enable-networking, etc) but the CFLAGS will be the same unless changed .. agree with Chris 2010-05-25T18:19:55 By "this way" I mean networking on and POSIX on. 2010-05-25T18:20:24 okey dokey 2010-05-25T18:20:50 Ah this is a difficult part. The CFLAGS are provided by RTEMS and set for the BSP. We try to keep them common but this can vary. 2010-05-25T18:21:19 An example is an ARM bug with some optimisation on the latest gcc which needs a special disable option. 2010-05-25T18:21:52 ok 2010-05-25T18:21:54 This is a weakness of RTEMS. We do not have a formal method for exporting these. 2010-05-25T18:22:12 We do have a method. I will find it. Hang on. 2010-05-25T18:23:11 Chris probably has autoconf macros or something nicer but .. in gcc-testing/gcc/testsuites/ada, there is a Makefile.rtems which extracts information about a BSP and hacks it into something you can pass to the Ada compiler 2010-05-25T18:23:31 ok 2010-05-25T18:23:34 Each BSP installs a bsp.cfg and target.cfg file. 2010-05-25T18:23:36 It is run and generates an output file you can source in a shell script 2010-05-25T18:24:58 for example for the pc586 I have $PREFIX/i386-rtems4.10/pc586/make/bsp.cfg 2010-05-25T18:25:09 after the build and install of the pc586 bsp. 2010-05-25T18:26:37 ok. 2010-05-25T18:26:48 so let me make sure i'm tracking 2010-05-25T18:27:10 step one is building the rtems compilers and tools on our host system for whatever architecture we want to target 2010-05-25T18:27:25 say i'm on Fedora Core 12 and building the qemu / x86 stuff 2010-05-25T18:27:45 step two is building parrot with the correct configure / hints / flags to match the right target 2010-05-25T18:28:02 step three would then be to compile some simple program in parrot and having it run within the rtems-gdb 2010-05-25T18:28:05 right? 2010-05-25T18:30:39 hang on just got busy here. be back in 5 2010-05-25T18:31:10 np, take your time 2010-05-25T18:31:30 bubaflub: that's basically the sequence. On F12, you can install the cross development tools from RPM and then build and install RTEMS. 2010-05-25T18:31:41 right 2010-05-25T18:31:56 right now i'm taking the path of least resistance 2010-05-25T18:32:07 i.e. Fedora Core 12 on Virtual Box with the RPMs and targeting x86 / qemu 2010-05-25T18:32:18 but eventually i'll throw together some mega-bootstraping scripts 2010-05-25T18:32:21 For step 2, you eventually want to extract the settings automatically in a nice way and write the conf hints 2010-05-25T18:32:23 to manage all of the stuff 2010-05-25T18:32:50 yep. at first i'll provide them manually but eventually i'd like to automate that 2010-05-25T18:32:52 I don't even build the tools from source for my own "normal" development machine. I build them all the time to test gcc. Not worth the trouble 2010-05-25T18:33:08 DrJoel: ah, ok. 2010-05-25T18:33:19 so the tools we can grab as a package 2010-05-25T18:33:28 but we'll still want to pull down the rtems csv, the parrot svn repo, and who knows what else 2010-05-25T18:34:01 sholdn't need more than parrot and rtems from version control. The RTEMS tools should have the rest you need 2010-05-25T18:34:27 ok. 2010-05-25T18:34:37 i was imagining phase 5 of world domination 2010-05-25T18:34:37 are there logs for this channel? 2010-05-25T18:34:51 for example, someone could use rakudo (a perl 6 implementation) to target RTEMS 2010-05-25T18:35:00 or any other high-level language that sits on top of parrot 2010-05-25T18:35:08 but for now, i'll stick to PIR 2010-05-25T18:35:15 dukeleto: i think i saw one at one point, lemme see if i can find the logs 2010-05-25T18:35:18 dukeleto: yes.. http://www.rtems.org/irclogs/ 2010-05-25T18:35:40 bubaflub: right.. consider that "bonus" work when things just work. ;-D 2010-05-25T18:35:41 http://trac.parrot.org/parrot/wiki/Languages is the list of languages on Parrot 2010-05-25T18:35:55 DrJoel: hahaha. will do. 2010-05-25T18:37:16 so are there some definite small milestones for bubaflub to accomplish in the first week? 2010-05-25T18:37:43 dukeleto: which are the most mature? I can think that Java, Ruby, Lua, Perl and Tcl make the most sense with Tcl at the bottom since we have a native Tcl port 2010-05-25T18:37:52 most sense for RTEMS 2010-05-25T18:38:33 bubaflub: since we know you can run RTEMS hello and ticker, I would say the next step is to extract the hints needed from an installed BSP 2010-05-25T18:39:00 that should be a bunch of similar probes/extracts, etc 2010-05-25T18:39:01 DrJoel: okey dokey. i'm looking at my rtems directory now, and i'm seeing config logs but not the file that kiwichris mentioned 2010-05-25T18:39:12 it is in the install point 2010-05-25T18:39:35 You want to install RTEMS and probably use a --prefix that isn't /opt/rtems-4.10 where the RPMs are 2010-05-25T18:39:46 back 2010-05-25T18:39:54 DrJoel: ah, i see it now 2010-05-25T18:40:15 On the IRC logs I am working to have them updated every 10 minutes 2010-05-25T18:40:46 DrJoel: Rakudo Perl 6, partcl-nqp (TCL), pynie (python), cardinal (ruby) 2010-05-25T18:40:57 DrJoel: those are the most active "well-known" languages 2010-05-25T18:41:11 DrJoel: many other custom languages have a lot of activity 2010-05-25T18:41:20 kiwichris: that would be awesome 2010-05-25T18:41:25 Having these languages supported is really neat 2010-05-25T18:41:44 dukeleto: what's the current state of HLL interop? 2010-05-25T18:42:02 kiwichris: HLL interop would allow us to seamlessly share classes and code between the higher level languages on parrot 2010-05-25T18:42:14 i.e. sub class a perl 6 class in python 2010-05-25T18:42:23 or any kinda combination 2010-05-25T18:42:42 bubaflub: i think tene++ is the best person to ask about that 2010-05-25T18:42:53 okey dokey, i'll hit him up on #parrot 2010-05-25T18:43:06 bubaflub, are you happy with the build type answers ? 2010-05-25T18:43:14 bubaflub: i think most languages have not caught up to the newest parrot features, but some HLL interop is already possible, not sure how much 2010-05-25T18:43:33 kiwichris: i'm looking through the files now, probably will have a few questions in a bit 2010-05-25T18:43:46 dukeleto: yeah, i think cardinal development has stalled a bit. if i had more tuits i'd definitely hit that up 2010-05-25T18:44:03 I am here all day and please email if I am offline. 2010-05-25T18:44:33 It was fun watching parrot tests run on a credit card Coldfire board 2010-05-25T18:44:42 haha nice. 2010-05-25T18:45:00 i suppose i should eventually acquire some actual hardware to run these tests on... 2010-05-25T18:46:00 bubaflub: just made this ticket for you: http://trac.parrot.org/parrot/ticket/1657 2010-05-25T18:46:01 http://www.axman.com/?q=node/247 2010-05-25T18:46:50 dukeleto++ 2010-05-25T18:48:33 kiwichris: looks nice. will i need some kind of LCD display as well? 2010-05-25T18:48:42 (just for future reference) 2010-05-25T18:48:46 No all serial ports 2010-05-25T18:48:54 kiwichris: ok, that's what it looked like 2010-05-25T18:49:09 I tend to run all targets with serial ports even bare PC hardware 2010-05-25T18:49:23 second question, will this interface with my mac? 2010-05-25T18:49:33 (macs tend to be without serial ports) 2010-05-25T18:49:48 i.e. is there a USB interface i could use? if not, i can get some commodity hardware and throw Fedora Core or Cent OS on it 2010-05-25T18:49:56 Yes and no. Yes if you have a USB to serial port converter and not for BDM 2010-05-25T18:50:03 the debugging hardware 2010-05-25T18:50:07 kiwichris: ok. 2010-05-25T18:51:12 I use a headless PC running Linux to connect to. 2010-05-25T18:52:24 http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPC8313E-RDB is a nice PowerPC board. $320 at digikey... was ~$200 from Freescale 2010-05-25T18:53:15 this might be a toy kit, but do you guys support the package from www.nerkits.com ? 2010-05-25T18:53:54 it's based around the Atmel ATmega 168 microcontroller 2010-05-25T18:55:32 Parrot in 1024 bytes of RAM :) 2010-05-25T18:55:54 hahaha, yeah 2010-05-25T18:55:55 that's not going to happen 2010-05-25T18:56:25 kiwichris: i'm poking around /opt/rtems-4.10/make, looking at all the .cfg files 2010-05-25T18:56:39 what hints do i need to extract from the files? 2010-05-25T18:56:49 i think i have an example makefile for parrot lying around here somewhere 2010-05-25T18:57:14 Did you build a BSP and did you install it ? 2010-05-25T18:57:57 kiwichris: yes. i had the sparc-bsis and a i386 qemu 2010-05-25T18:58:17 Great. Did you use a prefix on the configure command line ? 2010-05-25T18:58:33 i believe so... 2010-05-25T18:58:38 (now i'm nervous i borked the install) 2010-05-25T18:58:49 What was the prefix you used ? 2010-05-25T18:58:57 ah, i documented it in my blog post 2010-05-25T18:59:02 lemme pull that up real quick 2010-05-25T18:59:20 bubaflub: http://trac.parrot.org/parrot/ticket/1657#comment:1 2010-05-25T19:00:05 bubaflub: those are pretty easy-to-follow instructions for making progress with exit(), let me know if you run into something. that should be doable in the first week 2010-05-25T19:00:27 dukeleto: i'm not seeing the comment on that page, i'll refresh in a bit (perhaps just a cache issue) 2010-05-25T19:01:02 dukeleto: ok, now i see it 2010-05-25T19:01:08 (perhaps a weird logged in issue) 2010-05-25T19:02:25 kiwichris: looks like i didn't put a prefix on it... weird, i thought i did 2010-05-25T19:03:05 That is ok. 2010-05-25T19:03:50 So do you have a sparc-rtems4.10 dir under /opt/rtems-4.10 ? 2010-05-25T19:03:55 i do 2010-05-25T19:04:09 and an i386-rtems4.10 2010-05-25T19:04:30 Great. This is where the RTEMS BSB, includes and libs are installed to 2010-05-25T19:04:39 bubaflub: look at http://www.rtems.org/cgi-bin/cvsweb.cgi/gcc-testing/gcc/testsuite/ada/acats/ .. that is the scripts I use to test Ada with installed BSPs. Makefile.rtems extracts what that process needs which should be similar 2010-05-25T19:04:40 BSB/BSP 2010-05-25T19:04:55 Yes 2010-05-25T19:06:08 but I for about this one .. simplest example .. http://www.rtems.org/cgi-bin/cvsweb.cgi/examples-v2/misc/extract_example/ 2010-05-25T19:06:29 It just uses a Makefile to generate a "RTEMS_SETTINGS" file. You can generate whatever format you want 2010-05-25T19:07:32 DrJoel and kiwichris: ahhhhhh 2010-05-25T19:07:33 now i get it 2010-05-25T19:08:03 so i should have a makefile or script that can generate all of the settings that i need for the parrot configure 2010-05-25T19:09:50 (light bulb turns on) 2010-05-25T19:10:16 right.. 2010-05-25T19:10:21 Yeap 2010-05-25T19:10:51 there is yet another variant on this these at the top of the rtems-graphics-toolkit CVS module 2010-05-25T19:11:11 ok DrJoel i can take a peek at that as well 2010-05-25T19:11:19 Joel, we need to document these files and make a formal API for them. 2010-05-25T19:11:46 kiwichris: agreed! 2010-05-25T19:12:16 I would also like some autoconf magic to generate the initial tar filesystem .. hint.. that is an uncovered area I need to write a test for 2010-05-25T19:13:21 DrJoel, try http://www.rtems.org/ftp/pub/rtems/people/chrisj/file-system/rfs-20100218.tar.gz 2010-05-25T19:13:47 Check the Makefile.am 2010-05-25T19:14:52 would ralf scream if I added a test that used it? 2010-05-25T19:15:23 I need it to be in a psxtests for filesystems stuff 2010-05-25T19:16:31 Which bit are you talking about ? 2010-05-25T19:17:11 Haven't looked.. just need minimum autoconf/make infrastructure to generate the tarball, convert it to object, etc 2010-05-25T19:19:02 This is covered in that tarball. You need an m4 script RTEMS_OBJECT_FORMAT then the rest is normal makefile stuff 2010-05-25T19:19:12 dukeleto: would i just need a configuration hints for different rtems packages at config/init/hints 2010-05-25T19:19:18 and then a commandline flag to specify which target i want? 2010-05-25T19:19:26 ok.. will play with it 2010-05-25T19:19:30 The args to objcopy are different for some targets 2010-05-25T19:20:15 You need aclocal/obj-format.m4 2010-05-25T19:20:29 bubaflub: they vary within the same target architecture so you really want to generate them on the fly. Two arm bsps will have different settings, BSPs are added, and setting for sxisting bsps change. You don't want to track that 2010-05-25T19:20:56 right, what i'm thinking is this: 2010-05-25T19:21:07 step 0 - get rtems stuff via RPMs 2010-05-25T19:21:15 step 1 - build rtems and install bsp 2010-05-25T19:21:27 step 2 - extract that info and generate a hints file 2010-05-25T19:21:35 2.5 - insert that hints file into the parrot build area 2010-05-25T19:21:47 For step 1 I use a different prefix to the tools install. I keep the tools install area clean. 2010-05-25T19:21:48 step 3 - instruct parrot to build that target 2010-05-25T19:22:22 so rather than tracking it in the repo we'll have the makefile that generates it and then places it into a local parrot checkout 2010-05-25T19:22:35 bubaflub: yes, you will need some config hint support 2010-05-25T19:23:00 That was a hack on my part that never really worked. 2010-05-25T19:23:03 kiwichris: shell-init-tarfile.o: rule? 2010-05-25T19:23:10 DrJoel, yeap 2010-05-25T19:23:18 ok.. will have to try that. 2010-05-25T19:24:00 dukeleto: does that sound like a reasonable plan? i can throw together a Makefile script that'll extract the information about the currently installed RTEMS to generate a hints file on the fly 2010-05-25T19:24:24 dukeleto: and once that's done, we can specify at build time which OS / configuration we want to target 2010-05-25T19:24:54 bubaflub, can a parrot build script be used when it runs ? 2010-05-25T19:25:14 kiwichris: i'm not quite sure what you mean 2010-05-25T19:25:34 bubaflub: yes, i think that sounds like a good angle of attack 2010-05-25T19:25:51 dukeleto: ok, so on the RTEMS side i'll need to get my make maker 2010-05-25T19:25:56 bubaflub, I must be missing something 2010-05-25T19:26:28 kiwichris: ok, so let's say we've got the rtems stuff all setup and shiny 2010-05-25T19:26:41 yeap 2010-05-25T19:26:42 kiwichris: when we configure parrot, there is a perl script that runs to generate the Makefile 2010-05-25T19:26:48 Yes 2010-05-25T19:27:01 that perl script also checks the hints files to see if there are any defaults that it should use 2010-05-25T19:27:15 yes 2010-05-25T19:27:29 so what i want to do is 1) write a makefile makefile that'll extract the needed information into the suitable hints format 2010-05-25T19:27:37 2) place that hints file in the local parrot checkout 2010-05-25T19:27:55 3) specify which target i want to use when building parrot 2010-05-25T19:28:00 Yes this is a great solution 2010-05-25T19:28:08 nice! 2010-05-25T19:28:27 so if i've got a sparc-bsis, i can run the hints maker, it'll put it in the parrot repo, and then when compiling parrot i'll tell it to target sparc-bsis 2010-05-25T19:28:40 the configuration will pick up the hints file and compile parrot accordingly 2010-05-25T19:29:10 What do you mean by parrot repo ? Do you mean check it in > 2010-05-25T19:29:29 no, i mean a local copy 2010-05-25T19:29:35 i mean a local checkout of the parrot repository 2010-05-25T19:29:52 i don't think it would be a good idea to check these files in as they could change / there are a lot of possibilities 2010-05-25T19:29:55 Great that is fine. These BSP files are always changing 2010-05-25T19:30:14 kiwichris: awesome sauce. 2010-05-25T19:30:24 so my goals are the following: 2010-05-25T19:30:51 Not only that there are 116 current in RTEMS 2010-05-25T19:31:19 1) make parrot not call the system exit() when running on RTEMS via a compile time macro, tracked at http://trac.parrot.org/parrot/ticket/1657 2010-05-25T19:31:46 2) write a Makefile that will extract the current information of an installed BSP and write the appropriate parrot hints file 2010-05-25T19:32:07 3) patch the parrot Configuration script so we can override what OS / platform we are targeting 2010-05-25T19:32:19 4) write an introductory blog post 2010-05-25T19:32:25 sound good? 2010-05-25T19:32:33 Sounds great. 2010-05-25T19:32:43 With the exit, what will you call ? 2010-05-25T19:33:31 hmmm.. haven't thought about that. dukeleto? 2010-05-25T19:33:46 This is where I got stuck. 2010-05-25T19:35:52 we've got some C functions where we can hook on exit 2010-05-25T19:35:59 i think as long as we don't actually call the system exit 2010-05-25T19:36:01 we can do whatever we want 2010-05-25T19:36:29 Sort of. You need to have a context which is the instance of Parrot the call relates to. 2010-05-25T19:37:33 I see Parrot_exit (this is ok), exit, and exit_fatal. 2010-05-25T19:37:39 what would happen if we just don't call exit() ? 2010-05-25T19:37:49 i just want bubaflub to be able to run the parrot test suite 2010-05-25T19:37:49 yeah, Parrot_exit takes the current interpreter context 2010-05-25T19:38:05 dukeleto: ah, then i can check the test suite output 2010-05-25T19:38:06 do we need to call a different function, instead of exit() ? 2010-05-25T19:38:22 dukeleto, if this can happen then Parrot could remove the exit !! 2010-05-25T19:38:57 Yes but which instance of parrot do we exit if more than one is running ? 2010-05-25T19:39:47 kiwichris: what do you mean that parrot could remove the exit? 2010-05-25T19:40:21 Oh I miss understood your "not call exit" sentence. Sorry. 2010-05-25T19:41:30 If an exit call is a check that should never happen once ported and tested then maybe that is ok. 2010-05-25T19:42:06 I tripped over some because I had bad compiler settings and default hints. 2010-05-25T19:42:48 By ok I mean ok to leave as is in the Parrot source. 2010-05-25T19:44:31 kiwichris: so i should leave in the exit() for now? 2010-05-25T19:45:15 It depends. I think some do need to be fixed. For example the one a bad command line parameter would be nice if fixed. 2010-05-25T19:45:35 They all need to be checked. 2010-05-25T19:46:17 If they are too difficult to be fixed and we can say this is not normal for a working set up then we could argue it is ok to leave. 2010-05-25T19:47:12 Is it a suitable way to progress ? I do not want us to become stuck on something that may not happen once we have a working system. 2010-05-25T19:47:28 kiwichris: this is a run time problem, not a compile time problem, yes? 2010-05-25T19:47:41 Yes. 2010-05-25T19:47:57 ok, for the mean time then i'll work on getting the compiling good to go 2010-05-25T19:48:04 and i'll hit the dev list asking for some suggestions 2010-05-25T19:48:16 and hopefully by them time i've got a clean compile going, we've got an answer 2010-05-25T19:48:36 because this primarily affects the test suite rebooting the whole system 2010-05-25T19:48:42 (erroneously rebooting) 2010-05-25T19:49:04 Yes that is correct. For example you try a bad argument on the shell command line and you reboot. 2010-05-25T19:49:16 Makes you get to know the commands well :) 2010-05-25T19:50:19 hahaha 2010-05-25T19:50:23 kaboom. 2010-05-25T19:50:50 and a possible work around would be to have an error handler on the shell to trap that before we explode 2010-05-25T19:51:24 This works for a single thread only. 2010-05-25T19:52:03 dukeleto: should i also take care of TT #1176? 2010-05-25T19:55:38 * kiwichris inspects the Parrot faces page 2010-05-25T19:56:47 dukeleto: or better yet, take care of TT #1048? 2010-05-25T19:57:58 bubaflub: maybe. what does it buy for parrot on rtems? 2010-05-25T19:58:15 bubaflub: if parrot on rtems would use that heavily, than yes 2010-05-25T19:58:18 the ticket for pow() and powl() is still open 2010-05-25T19:58:31 and it seemed like the way to take care of that was for a configure-time check 2010-05-25T19:58:49 and then 1048 suggested having a configuration function that checked for the existence of other functions 2010-05-25T19:58:55 and i figured we might as well do it right 2010-05-25T19:58:58 bubaflub: yes, currently Parrot just uses pow() now, which is sub-optimal 2010-05-25T19:59:10 ok, i'll put it on my list of things to investigate 2010-05-25T19:59:18 will that block the building of parrot on rtems? 2010-05-25T19:59:24 i.e. suboptimal = breaks the build? 2010-05-25T19:59:25 bubaflub: yes, what kind of time frame do you think 1048 would take? 2010-05-25T19:59:39 bubaflub: no, i changed it so it doesn't block the build 2010-05-25T19:59:43 dukeleto: ok, great. 2010-05-25T19:59:55 so i'll focus on getting the build stuff done for now 2010-05-25T19:59:56 b 2010-05-25T20:00:00 ut i'll keep that ticket in view 2010-05-25T20:00:19 i don't think it would be too hard - try calling a function with a certain name, and if it errors then you don't have it 2010-05-25T21:12:04 kiwichris DrJoel dukeleto: blog post: http://blog.bobkuo.com/2010/05/beginning-parrot-on-rtems/ - please check for any factual errors etc. 2010-05-25T21:12:42 also, if you want filtered access to the Google Summer of Code stuff, use http://blog.bobkuo.com/category/google-summer-of-code/ 2010-05-25T21:32:41 ok, time for a homebrew and pizza with my fiance. see you guys tomorrow 2010-05-25T21:32:55 s/see/talk with on IRC/g 2010-05-25T21:32:58 *** bubaflub has quit IRC 2010-05-25T23:13:45 *** Raj_ has joined #rtems 2010-05-25T23:15:38 *** Raj_ has joined #rtems 2010-05-25T23:16:08 *** Raj_ has joined #rtems 2010-05-25T23:40:12 *** Raj_ has joined #rtems 2010-05-25T23:40:26 DrJoel: hello 2010-05-26T00:15:44 *** Raj_ has quit IRC 2010-05-26T01:01:46 *** madrazr has joined #rtems 2010-05-26T01:05:20 *** madrazr1 has joined #rtems 2010-05-26T01:07:45 *** madrazr has quit IRC 2010-05-26T01:12:10 *** madrazr1 has quit IRC 2010-05-26T01:12:34 *** madrazr has joined #rtems 2010-05-26T01:12:55 *** madrazr has left #rtems 2010-05-26T02:09:53 *** sebhub has joined #rtems 2010-05-26T02:10:55 good morning 2010-05-26T03:06:05 sebhub, are you about ? 2010-05-26T03:22:00 yes 2010-05-26T03:22:16 Just playing with the IRC logs index page. 2010-05-26T03:22:49 Can you have a look ? 2010-05-26T03:23:04 yes, can you give me the link? 2010-05-26T03:23:14 Should I have the weeks go forward for each year or back ? 2010-05-26T03:23:25 http://www.rtems.org/irclogs/index.html 2010-05-26T03:23:44 This view is backwards and I tend to think it is wrong. 2010-05-26T03:26:50 i would use a table like the mailing list archives 2010-05-26T03:27:32 Huh ? 2010-05-26T03:28:04 http://www.rtems.org/pipermail/rtems-users/ 2010-05-26T03:28:27 Then what ? 2010-05-26T03:33:01 Sorry I do not follow how a month table can be created with nicks etc. 2010-05-26T03:42:26 you mean the order within a particular log? 2010-05-26T03:42:33 i think that is ok 2010-05-26T03:43:23 Yes. I think anything else is too hard. I have auto updating of the logs every 10minutes 2010-05-26T03:43:49 *** alexp0205 has joined #rtems 2010-05-26T04:24:05 *** dr__house has joined #rtems 2010-05-26T04:27:21 *** alexp02 has joined #rtems 2010-05-26T04:27:21 *** alexp0205 has quit IRC 2010-05-26T04:30:12 *** Raj_ has joined #rtems 2010-05-26T04:30:38 DrJoel: hello 2010-05-26T04:36:32 *** madrazr has joined #rtems 2010-05-26T04:37:47 *** dr__house` has joined #rtems 2010-05-26T04:38:00 *** dr__house has quit IRC 2010-05-26T04:38:07 *** dr__house` is now known as dr__house 2010-05-26T04:43:34 *** alexp0205 has joined #rtems 2010-05-26T04:45:29 *** alexp02 has quit IRC 2010-05-26T05:00:50 *** Raj_ has quit IRC 2010-05-26T05:27:11 *** zwj has joined #rtems 2010-05-26T05:29:58 *** exception13 has joined #rtems 2010-05-26T05:33:48 *** alexp0205 has left #rtems 2010-05-26T06:20:33 *** tuxmaniac has joined #rtems 2010-05-26T06:38:35 *** madrazr has quit IRC 2010-05-26T06:38:36 *** madrazr1 has joined #rtems 2010-05-26T06:38:53 *** madrazr1 is now known as madrazr 2010-05-26T06:39:06 *** madrazr has joined #rtems 2010-05-26T06:48:16 *** madrazr1 has joined #rtems 2010-05-26T06:48:16 *** madrazr has quit IRC 2010-05-26T07:14:32 apparently missed Raj again 2010-05-26T07:17:46 *** madrazr1 has quit IRC 2010-05-26T07:19:31 *** DrJoel has quit IRC 2010-05-26T07:52:01 *** kiwichris has quit IRC 2010-05-26T08:07:56 *** DrJoel has joined #rtems 2010-05-26T08:07:56 *** ChanServ sets mode: +o DrJoel 2010-05-26T08:14:10 *** dr__house has quit IRC 2010-05-26T08:18:49 *** Raj_ has joined #rtems 2010-05-26T08:19:00 DrJoel: hi :) 2010-05-26T08:27:48 *** Raj_ has quit IRC 2010-05-26T08:28:13 *** madrazr has joined #rtems 2010-05-26T08:28:24 *** madrazr has left #rtems 2010-05-26T08:29:40 *** dr__house has joined #rtems 2010-05-26T08:37:44 well there he went again :( 2010-05-26T08:37:51 hey dr__house 2010-05-26T08:38:58 *** cdcs has joined #rtems 2010-05-26T08:50:37 *** exception13 has left #rtems 2010-05-26T08:51:10 DrJoel: good morning :) 2010-05-26T08:56:16 :) 2010-05-26T08:56:57 Waiting to hear from Raj.. he was on the verge of a nice coverage change. Not inlining 3 routines in the filesystem/system call infrastructure which had 1-3 branches in each!! 2010-05-26T09:03:18 DrJoel: Raj and coverage? Or did you mess up the names? 2010-05-26T09:03:27 DrJoel: This was the change that you were talking about around 2 days back right? 2010-05-26T09:03:51 probably messed up names 2010-05-26T09:03:53 dr__house: right 2010-05-26T09:03:58 on phone brb 2010-05-26T09:09:22 DrJoel: heh, I guess you meant Bharat. In fact you have mailed Raj too regarding this. I guess that mail should've been sent to Bharat. :P 2010-05-26T09:17:38 :( stupid me 2010-05-26T09:42:20 *** madrazr has joined #rtems 2010-05-26T09:57:25 *** dr__house has quit IRC 2010-05-26T10:17:42 *** exception13 has joined #rtems 2010-05-26T10:24:24 *** sebhub has quit IRC 2010-05-26T10:34:50 *** arvind_khadri has joined #rtems 2010-05-26T10:44:45 *** madrazr has quit IRC 2010-05-26T10:47:29 *** arvind_khadri has quit IRC 2010-05-26T10:47:47 *** zwj has quit IRC 2010-05-26T11:25:17 *** Raj_ has joined #rtems 2010-05-26T11:25:46 *** dr__house has joined #rtems 2010-05-26T11:26:25 *** krzabr has joined #rtems 2010-05-26T11:28:38 DrJoel: hello 2010-05-26T11:29:21 hey Raj_ how are things going? 2010-05-26T11:30:39 DrJoel: starting off on testing 2010-05-26T11:30:49 stuck on that basically :) 2010-05-26T11:31:53 the first test is hard. Is it dejagnu and running them in general? 2010-05-26T11:32:55 Well, as you said I will try to run the gcc C testsuite first, then go for Go 2010-05-26T11:33:11 :D that should work easily enough 2010-05-26T11:34:38 :D Well I am stuck on that ! 2010-05-26T11:35:57 I haven't yet figured out how to run the tests .. in a step-by-step manner 2010-05-26T11:37:03 You have to have a gcc build tree with C built. Then the rundeja script can set up the magic variables to run tests. It has an example at the end to just run a subset of tests. 2010-05-26T11:37:57 I do that from inside gcc-testing ? 2010-05-26T11:38:40 those scripts assume that but you can do it by hand with your own build tree if you set the variables right 2010-05-26T11:39:12 But I dont see a rundeja script inside gcc-testing ?! 2010-05-26T11:39:24 gcc directory 2010-05-26T11:40:08 rundeja or runtest ? 2010-05-26T11:41:10 have you updated recently? runtest is not in my tree. 2010-05-26T11:41:41 no runtest is for running dejagnu 2010-05-26T11:42:00 I dont have rundeja 2010-05-26T11:43:16 let's start from the beginning.. are you in the rtems gcc-testing cvs module? 2010-05-26T11:43:24 Yes I am 2010-05-26T11:43:40 $ ls gcc 2010-05-26T11:43:40 ChangeLog do_one gcc_status rtems_gcc_main.c test_driver 2010-05-26T11:43:40 CVS do_tests Makefile.rtems_gccmain rundeja testsuite 2010-05-26T11:43:59 Yes I see rundeja 2010-05-26T11:44:07 funny I did a find and it didn't show it !! 2010-05-26T11:44:10 that's the script that runs the C/C++ test suite 2010-05-26T11:44:24 do_tests is supposed to work on an installed toolset but I haven't used that in a long time 2010-05-26T11:45:05 ok cool. :) 2010-05-26T11:45:28 so I just need to run rundeja.. but how will it know about where my build directory and stuff are ? 2010-05-26T11:46:01 Notice that rundeja is getting a lot of settings from "VERSIONS" which is where it gets some info frmo 2010-05-26T11:46:32 when run automatically .. it is run by do_one (a single target) and it cd's to the build directory before invoking it 2010-05-26T11:46:38 echo "Running GCC Tests..." 2010-05-26T11:46:38 cd ${BASEDIR}/b-gcc1-${cpu} || exit 1 2010-05-26T11:46:38 time sh -x ${SCRIPTDIR}/gcc/rundeja ${bsp} \ 2010-05-26T11:46:38 >${LOGDIR}/${cpu}-gcctests-${bsp}.log 2>&1 2010-05-26T11:47:08 Shouldn't be too difficult to set the variables it needs hacky and use it yourself 2010-05-26T11:47:49 Ok so do_one is just for one target .. hmm 2010-05-26T11:48:49 test_driver can update, build native, install automake/conf, then loops over all the targets you want to build and test. It can do binutils, gdb, and gcc for C, C++ and Ada. I am hoping it grows to include Go. 2010-05-26T11:50:13 Thanks, this has saved me a lot of time. Now I have a big picture idea of gcc-testing ! :) 2010-05-26T11:50:59 Isn't this information available on the wiki, made a superficial google search, didn't get any help ! 2010-05-26T11:52:34 Hmm... I don't know. 2010-05-26T11:56:02 *** madrazr has joined #rtems 2010-05-26T11:56:03 Will get back to you with new doubts in a day's time :D ! 2010-05-26T11:56:40 Doubts are common any time you are in new and uncharted territory 2010-05-26T12:06:26 DrJoel: the VERSIONS file has options to specify the automake source directory 2010-05-26T12:06:39 what if I haven't made automake from source ? 2010-05-26T12:08:59 same for binutils and gdb 2010-05-26T12:15:06 *** Raj_ has quit IRC 2010-05-26T12:19:36 *** Raj_ has joined #rtems 2010-05-26T12:31:09 *** Raj_ has quit IRC 2010-05-26T14:16:38 *** krzabr has quit IRC 2010-05-26T14:44:53 *** _Lucretia__ has joined #rtems 2010-05-26T14:45:15 *** _Lucretia_ has quit IRC 2010-05-26T14:58:42 btw could Qt Embedded work on RTEMS? 2010-05-26T15:00:54 it's LGPL and looks much nicer and feature-rich than Nano-X 2010-05-26T15:01:28 I don't know how heavy that would be though... 2010-05-26T15:01:50 statically linked binaries under linux are 15+ MB 2010-05-26T15:03:19 Statically linked and LGPL generally is not cool for "real" embedded applications. To comply you have to ship your closed source objects and source for LGPL components so they can be relinked 2010-05-26T15:03:23 BUT .... 2010-05-26T15:04:25 when we have dynamic loading, we may be able to consider that shared libraries and it might be more ok then. 2010-05-26T15:04:34 But it would be cool and useful for students/experimenters 2010-05-26T15:25:27 *** lekernel has quit IRC 2010-05-26T15:27:45 hi 2010-05-26T15:28:11 mwalle: hello 2010-05-26T15:28:25 DrJoel, how are you? 2010-05-26T15:29:03 very busy this week. Monday was my birthday, my daughter started her summer internship and two of my sons graduate from high school Friday! 2010-05-26T15:29:38 have you seen the latest coverage reports? nice html. 2010-05-26T15:30:02 Happy belated birthday :) 2010-05-26T15:30:15 Thanks 2010-05-26T15:31:51 not yet 2010-05-26T15:33:27 nice :) 2010-05-26T15:35:27 do you use any scripting language for parsing and generating the html file or is it a part of covoar 2010-05-26T15:36:33 mhhh 2010-05-26T15:36:36 http://www.rtems.org/ftp/pub/rtems/people/joel/coverage/lm32_evr-OsPd-20100522-0703/annotated.html#range98 2010-05-26T15:36:49 looks strange 2010-05-26T15:37:04 why is this one instruction not executed? 2010-05-26T15:37:52 *** lekernel has joined #rtems 2010-05-26T15:38:38 one can think, that context_restore doesnt return, but then, why are the last five instructions executed.. 2010-05-26T15:43:11 *** tuxmaniac has quit IRC 2010-05-26T15:44:49 mwalle: The contents of each run directory are out of covoar so C/C++. The per BSP index files are in shell. 2010-05-26T15:45:02 That range is because that does not return 2010-05-26T15:45:48 I think you can arrange _CPU_Context_Restart_self to not be a macro and have it be a real label in the .S file. Then mark that prototype as noreturn 2010-05-26T15:45:49 ah there is a local jump :) 2010-05-26T15:46:37 which jumps to 8005d68 2010-05-26T15:46:39 from where 2010-05-26T15:46:45 * mwalle is very tired 2010-05-26T15:47:28 i wondered why 8005d64 isnt executed, while 8005d68 is. but theres a jump which skips the call to the context_restore 2010-05-26T15:48:32 btw do you arithmetically round up and down, when you calculate the overall coverage? or always round down? 2010-05-26T15:49:47 would have to check on rounding.. it is just C++ math and printf 2010-05-26T15:50:28 In that path, _CPU_Context_Restart_self (e.g. _CPU_Context_restore on lm32) does not return since it is a restart self 2010-05-26T15:50:40 See how it is done on the arm.. 2010-05-26T15:50:45 DrJoel, yes i got it :) 2010-05-26T15:50:46 void _CPU_Context_restore( Context_Control *new_context ) 2010-05-26T15:50:46 RTEMS_COMPILER_NO_RETURN_ATTRIBUTE; 2010-05-26T15:51:12 minor nit but will eliminate that 2010-05-26T15:51:43 *** cdcs has quit IRC 2010-05-26T16:05:37 *** cdcs has joined #rtems 2010-05-26T16:18:43 *** dr__house has quit IRC 2010-05-26T16:59:21 *** exception13 has left #rtems 2010-05-26T17:12:35 *** madrazr has quit IRC 2010-05-26T17:34:39 *** kiwichris has joined #rtems 2010-05-26T17:39:56 Hello 2010-05-26T17:41:10 DrJoel, hi 2010-05-26T17:47:44 hey .. hold on.. jessica is at the office and I had to check her oil 2010-05-26T17:49:22 Hope she had some to get to the office :) 2010-05-26T17:50:05 it was down between 1.5L and 2L.. :( making noise.. 2010-05-26T17:50:23 Ouch 2010-05-26T17:50:40 I think the valve cover gasket is leaking .. but there is oil collected in the center of the one cover and a bolt in the center which I think is really the problem 2010-05-26T17:51:02 damn lucky she didn't burn it up... got the typical girl "i forgot" 2010-05-26T17:51:36 we always check the fluids VERY frequently in vehicles after they join our fleet. you have to know what is happening.. then you can back off 2010-05-26T17:51:37 "Oh is that what that light means !!" 2010-05-26T17:52:41 Is the psx test all working ? 2010-05-26T17:53:23 all of them or just the one I was adding? 2010-05-26T17:53:39 The loadfs one ? 2010-05-26T17:54:25 I have a funny problem with the mount function changes I have. 2010-05-26T17:54:57 I added psxload01 to test untar .. probably misnamed. 2010-05-26T17:55:10 all of the psxtests seem to be passing 2010-05-26T17:55:31 I include a shell init script I run to set up the test. Much nicer that lots of C code. 2010-05-26T17:55:51 Is this in the tree or outside the tree ? 2010-05-26T17:56:09 It will be in the tree. I just wrote it this afternoon. :) 2010-05-26T17:56:30 I agree on the shell init but for the tests we need to keep them light so they can run on as many target boards as possible 2010-05-26T17:56:32 Great. 2010-05-26T17:56:38 Yeah. 2010-05-26T17:56:50 The mount problem is strange. 2010-05-26T17:57:31 The shell script runs and sets up an environment with a correct root and current directory 2010-05-26T17:57:45 I then start the CLI shell and its root file system is all NULL 2010-05-26T17:58:17 The handling of the root fs is different but it is set as before so something else is wrong. 2010-05-26T17:58:46 Does psxfile01 run ok? 2010-05-26T17:59:05 No idea. I am not doing this in the tree. 2010-05-26T17:59:33 It is built with -O2 and that can be a pain. 2010-05-26T17:59:46 By it I mean RTEMS 2010-05-26T18:00:46 I just need to debug it but with -O2 and a gdb stack trace window I get silly fetches from memory and that effects the BDM driver. It does not like it. 2010-05-26T18:01:21 So the root IMFS is setup OK? That is what I was concerned about. 2010-05-26T18:01:27 Yeap 2010-05-26T18:01:47 It is set up and I can make dirs mount NFS plus other things. 2010-05-26T18:02:08 I something is corrupting something 2010-05-26T18:02:48 I do not understand why we have the eval calls to in the env set up code. 2010-05-26T18:03:13 Also should a new env take the current dir from the parent or reset to the root ? 2010-05-26T18:03:41 The calls are also in the base_fs.c initialise call. 2010-05-26T18:04:18 by new env, you mean new shell or the "process env"? 2010-05-26T18:07:40 In rtems_libio_set_private_env 2010-05-26T18:10:08 I think it should take current director from the parent. 2010-05-26T18:10:22 FWIW if you are touching this code, fix Till's spelling of implicitly. :) 2010-05-26T18:10:32 I do to and will do. 2010-05-26T18:10:48 Also can we just copy the parent without the need to make the calls ? 2010-05-26T18:11:10 They are the same before and after. 2010-05-26T18:11:10 yes... just need to keep the comment that enforces that the root node will never change. :) 2010-05-26T18:11:28 Can it ? 2010-05-26T18:11:45 The root has to be the first entry of the mount table chain control 2010-05-26T18:12:03 This is accessed as that in the code somewhere. Not good but it is present. 2010-05-26T18:13:12 no I don't think we could change the root fs on the fly. Right now, we only allow IMFS as root but I suppose some day that could change 2010-05-26T18:13:32 Not great to assume it it first .. 2010-05-26T18:13:36 I plan to go over this code in 4.11 2010-05-26T18:13:41 fyi.. erc32 branch coverage 2010-05-26T18:13:49 Bytes Analyzed : 122276 2010-05-26T18:13:49 Bytes Not Executed : 17644 2010-05-26T18:13:49 Percentage Executed : 85.57 2010-05-26T18:13:49 Percentage Not Executed : 14.43 2010-05-26T18:13:49 Uncovered ranges found : 399 2010-05-26T18:13:50 Total branches found : 2246 2010-05-26T18:13:52 Uncovered branches found : 1135 2010-05-26T18:13:56 693 branches always taken 2010-05-26T18:13:58 442 branches never taken 2010-05-26T18:14:14 Is this with the file system etc ? 2010-05-26T18:15:15 yes.. you can check out gcc-testing/rtems-coverage/do_coverage where it generates the list of symbols tested. Not everything in cpukit is there... no libblock or monitor or shell I think 2010-05-26T18:15:26 Bytes Analyzed : 70136 2010-05-26T18:15:26 Bytes Not Executed : 52 2010-05-26T18:15:26 Percentage Executed : 99.93 2010-05-26T18:15:26 Percentage Not Executed : 0.07414 2010-05-26T18:15:26 Uncovered ranges found : 6 2010-05-26T18:15:27 Total branches found : 1204 2010-05-26T18:15:29 Uncovered branches found : 514 2010-05-26T18:15:31 247 branches always taken 2010-05-26T18:15:33 267 branches never taken 2010-05-26T18:15:40 for core only (score, sapi, rtems, and posix 2010-05-26T18:15:51 Nice 2010-05-26T18:16:43 So out of 2408 branch decisions, we take ~80% of them 2010-05-26T18:17:00 sorry.. exercise them with our current tests 2010-05-26T18:17:23 I understand. Is it hard to get branches taken ? 2010-05-26T18:18:01 you mean to create tests for them? 2010-05-26T18:18:09 Yes 2010-05-26T18:18:54 sometimes but a lot are just cases where the "return EINVAL" was optimized into a single place. So two error checks look executed but we didn't actually generate both errors in the tests 2010-05-26T18:19:18 Some I expect will be devilish. :) 2010-05-26T18:19:21 Ok. 2010-05-26T18:25:47 time to head home.. night 2010-05-26T18:25:50 *** DrJoel has quit IRC 2010-05-26T20:00:56 *** kiwichris has quit IRC 2010-05-26T21:28:00 *** bubaflub has joined #rtems 2010-05-26T21:40:12 any makefile experts here? i need an actual $ in a string to not interpolate. i've tried $$ and \$. any ideas? 2010-05-26T23:12:22 *** kiwichris has joined #rtems 2010-05-26T23:45:15 kiwichris: i sent ya an email with a link to my git repo with preliminary work. i'm heading to bed for now. catch ya later. 2010-05-26T23:45:21 *** bubaflub has quit IRC 2010-05-27T00:26:29 *** dr__house has joined #rtems 2010-05-27T02:03:47 *** tfinn has quit IRC 2010-05-27T02:20:21 *** madrazr has joined #rtems 2010-05-27T02:40:49 *** exception13 has joined #rtems 2010-05-27T03:00:56 *** haxomophone has quit IRC 2010-05-27T03:11:37 morning 2010-05-27T03:18:15 *** madrazr has quit IRC 2010-05-27T03:54:13 *** madrazr has joined #rtems 2010-05-27T06:33:46 *** madrazr has quit IRC 2010-05-27T07:02:57 *** DrJoel has joined #rtems 2010-05-27T07:02:57 *** ChanServ sets mode: +o DrJoel 2010-05-27T07:03:38 kiwichris: good morning from my perspective.. how are the winter viri there? 2010-05-27T07:10:57 *** DrJoel has quit IRC 2010-05-27T07:25:08 *** madrazr has joined #rtems 2010-05-27T07:38:37 *** kiwichris has quit IRC 2010-05-27T07:42:02 *** Raj_ has joined #rtems 2010-05-27T08:35:17 *** DrJoel has joined #rtems 2010-05-27T08:35:17 *** ChanServ sets mode: +o DrJoel 2010-05-27T08:51:20 *** exception13 has left #rtems 2010-05-27T09:06:15 *** Raj_ has quit IRC 2010-05-27T09:09:27 *** bubaflub has joined #rtems 2010-05-27T09:11:38 bubaflub: any luck? 2010-05-27T09:11:46 DrJoel: ? 2010-05-27T09:17:15 what's happening on gsoc today? just checking 2010-05-27T09:17:57 DrJoel: ah, yes that. 2010-05-27T09:18:13 i emailed dukeleteo and kiwichris, but forgot to include you 2010-05-27T09:18:19 i've got my preliminary work on a git repo 2010-05-27T09:18:25 http://github.com/bubaflub/gsoc-rtems-parrot-hints-maker 2010-05-27T09:18:37 cool .. 2010-05-27T09:18:40 and got my preliminary blog post describing 2010-05-27T09:18:42 what i want to do 2010-05-27T09:18:43 http://blog.bobkuo.com/2010/05/beginning-parrot-on-rtems/ 2010-05-27T09:19:47 so right now i need to figure out what values Parrot needs to be configured with 2010-05-27T09:19:53 and where I need to store those values 2010-05-27T09:20:26 as well as patch the Configuration.PL script to allow me to override the auto-detected architecture 2010-05-27T09:20:53 i have a pretty good idea where / how to store those configuration values, so i'll just need to make sure i set *everything* 2010-05-27T09:21:21 for example, we detect if you have GMP installed. of course with the auto-detect it will detect if the host system has GMP 2010-05-27T09:21:36 most likely an embedded system doesn't have GMP, so that needs to be unconditionally overridden 2010-05-27T09:22:13 Update the rtems wiki page on Parrot to link to your repo and blog 2010-05-27T09:22:21 DrJoel: okey dokey. 2010-05-27T09:22:59 hmmm.. it needs GMP on the target system.. interesting. I am pretty sure we don't build that to the target now 2010-05-27T09:23:12 That may be a side project to get that built and installed 2010-05-27T09:23:28 DrJoel: it doesn't need it, but we auto detect if it's available 2010-05-27T09:23:42 that's just one example of a library we detect if you have it 2010-05-27T09:24:17 ok. Then detecting it not there is good but that would leave that as a TODO. Just keep a list of libraries that Parrot will use if there. If you autodetect them missing we can port them over time and they will just get used. 2010-05-27T09:24:33 yep. 2010-05-27T09:24:55 those optional libraries more likely to cause run-time issues, not compile-time issues 2010-05-27T09:25:06 so my main focus will be getting compilation error-free 2010-05-27T09:25:19 and then tweaking those settings to make sure we accurately reflect what is available 2010-05-27T09:25:22 We have a similar issue with Go. It uses a few things we don't have. Now we are letting them get disabled and then adding the missing pieces on our side 2010-05-27T09:25:52 some higher level languages (like Perl 6, Python, etc.) on top of Parrot might require some other libraries 2010-05-27T09:25:59 so i'll make notes of what we may need 2010-05-27T09:28:07 That's all we can do. FWIW when Roxana finished up Microwindows last year, I got a number of optional support libraries to compile to RTEMS in just a couple of days 2010-05-27T09:28:45 DrJoel: awesome. 2010-05-27T09:28:56 i updated the wiki page with some links under the "Resources" section 2010-05-27T09:29:48 Great! 2010-05-27T09:30:20 When you update your blog, post a message to rtems-users and let the community know you have made some progress. 2010-05-27T09:30:37 yeah, i'll hit both the RTEMS and the Parrot lists today with my initial blog posts 2010-05-27T09:32:57 *** alexp0205 has joined #rtems 2010-05-27T09:43:08 *** alexp0205 has quit IRC 2010-05-27T09:55:27 *** Raj_ has joined #rtems 2010-05-27T09:55:37 DrJoel: hello 2010-05-27T09:55:52 Raj_: hi! 2010-05-27T09:56:14 I tried testing gccgo today and yesterday 2010-05-27T09:56:29 and... 2010-05-27T09:56:41 Some doubts :) 2010-05-27T09:57:02 do I need to build rtems in a special way to do the testing using qemu ? 2010-05-27T09:57:36 specifically do I need to set USE_COM1_AS_CONSOLE=1 and BSP_PRESS_KEY_FOR_RESET=0 ? 2010-05-27T09:57:51 during rtems build configure 2010-05-27T09:58:31 *** alexp0205 has joined #rtems 2010-05-27T09:58:58 The do_one script does that. I don't think you have to have USE_COM1_AS_CONSOLE since pc386_fda handles that dynamically for you but it is safer 2010-05-27T09:59:48 do_one script builds rtems you mean ? 2010-05-27T10:00:04 yep. it can. 2010-05-27T10:00:23 do_one [options] CPU BSP 2010-05-27T10:00:24 -A - binutils, stage 1, C++, C/C++ multilib, Ada, RTEMS 2010-05-27T10:00:24 -T - run GCC and Ada tests 2010-05-27T10:00:24 -b - install binutils (from source or copy installed RPM contents) 2010-05-27T10:00:24 -D - install gdb (from source or copy installed RPM contents) 2010-05-27T10:00:24 -n - native compiler 2010-05-27T10:00:25 Well I tried without building rtems using do_one 2010-05-27T10:00:27 -1 - stage 1 C/C++ compiler 2010-05-27T10:00:28 -c - do not include C++ 2010-05-27T10:00:30 -m - do not build multilib 2010-05-27T10:00:32 -a - stage 2 Ada compiler 2010-05-27T10:00:34 -r - build RTEMS 2010-05-27T10:00:38 -g - run GCC C/C++ tests 2010-05-27T10:00:40 -C - run ACATS 2010-05-27T10:00:42 -d - do not clean up (delete build directories) 2010-05-27T10:00:44 -v - verbose 2010-05-27T10:00:46 -M - do NOT email test results to gcc-testresults 2010-05-27T10:01:22 What happened when you ran? 2010-05-27T10:02:22 Well one peculiar error was that it was saying undefined references to lchown and utimes 2010-05-27T10:02:27 and just those two :) 2010-05-27T10:03:30 These were the only errors though ! 2010-05-27T10:03:55 Then I made the Go wrappers of these functions point to chown and utime instead and tried again 2010-05-27T10:04:03 those are build errors. Did it run anything? 2010-05-27T10:04:45 Many tests were just compilation and error tests, which went ok 2010-05-27T10:05:13 but execution tests, none of them worked, all of them had these errors ! 2010-05-27T10:06:55 like what? give me some output 2010-05-27T10:06:58 That's when I remembered another problem - I built rtems for pc486, because of the __sync* builtins issue in GCC 2010-05-27T10:07:47 Sorry, I don't have that specific output now, I will mail them to you in some time ! 2010-05-27T10:07:58 that shouldn't be much trouble... just check the -cpu option on qemu 2010-05-27T10:08:14 *** exception13 has joined #rtems 2010-05-27T10:08:52 But the boards for dejagnu ? 2010-05-27T10:09:17 I mean only pc386 is supported, right ? 2010-05-27T10:10:31 Because the tests, they say now that "Simulator pc386 is missing" 2010-05-27T10:10:59 ohhh... you need to do a "make" in gcc-testing and add ..../gcc-testing/bin to your path. :) 2010-05-27T10:11:24 Yea that was my next question :) 2010-05-27T10:12:48 *** exception13 has left #rtems 2010-05-27T10:12:55 *** exception13 has joined #rtems 2010-05-27T10:14:21 And the cpu option on qemu you were talking about ? 2010-05-27T10:19:51 it would go in sim-scripts/pc386.in I think where it sets the arguments on how to invoke qemu 2010-05-27T10:21:11 *** exception13 has left #rtems 2010-05-27T10:21:16 *** exception13 has joined #rtems 2010-05-27T10:22:34 Ok. Let me try that 2010-05-27T10:26:24 *** exception13 has left #rtems 2010-05-27T10:27:03 *** exception13 has joined #rtems 2010-05-27T11:12:05 *** exception13 has left #rtems 2010-05-27T11:12:17 *** exception13 has joined #rtems 2010-05-27T11:55:56 *** tuxmaniac has joined #rtems 2010-05-27T12:07:06 *** DrJoel has quit IRC 2010-05-27T12:44:23 *** Raj_ has quit IRC 2010-05-27T13:21:20 *** tuxmaniac has quit IRC 2010-05-27T14:25:45 *** alexp0205 has left #rtems 2010-05-27T16:03:11 *** bubaflub has quit IRC 2010-05-27T16:46:35 *** dr__house has quit IRC 2010-05-27T16:51:22 *** exception13 has left #rtems 2010-05-27T16:53:34 *** kiwichris has joined #rtems 2010-05-27T17:27:11 *** madrazr has quit IRC 2010-05-27T17:46:13 *** kiwichris has quit IRC 2010-05-27T18:04:13 *** dukeleto has quit IRC 2010-05-27T18:10:51 *** dukeleto has joined #rtems 2010-05-27T19:03:41 *** kiwichris has joined #rtems 2010-05-27T19:23:26 *** DrJoel has joined #rtems 2010-05-27T19:23:26 *** ChanServ sets mode: +o DrJoel 2010-05-27T19:23:51 kiwichris: how's the bug tracking going? 2010-05-27T19:30:36 Better, I just got some coffee !! 2010-05-27T19:30:54 I have confdefs.h sorted and documented. 2010-05-27T19:31:06 I have mounts working, well NFS is working 2010-05-27T19:31:34 All you need to do for NFS is call mount. No need to initialise in any special way 2010-05-27T19:31:55 Just looking at the TFTP FS. 2010-05-27T19:32:22 DrJoel, do you know of a way to get a path given a pathloc ? 2010-05-27T19:36:17 *** alexp0205 has joined #rtems 2010-05-27T20:08:02 kiwichris: not off the top of my head. Email Jennifer.. she will be in the office tomorrow 2010-05-27T20:08:30 I suspect we cannot looking at getcwd. It works from the top down. 2010-05-27T20:08:48 I have decided to change the mount_me handler to take the mount_point as an argument 2010-05-27T20:09:02 This way a file system will know the mount point 2010-05-27T20:09:05 whatever makes sense and makes everything clearer 2010-05-27T20:09:18 the getcwd implementation is from newlib I think.. 2010-05-27T20:09:24 It means we can mount the TFTP FS anywhere and not just /TFTP 2010-05-27T20:09:42 I thought we could now so that is definitely an improvement. 2010-05-27T20:09:42 It will also help a df command which we need badly 2010-05-27T20:09:58 Did you see the new version of the patch? I think it is properly formed now 2010-05-27T20:10:13 Not yet. 2010-05-27T20:11:11 I mainly posted it because of the autotools issues involved. It is pushing a new area and we want it dead right. The test is pretty mundane 2010-05-27T20:11:24 The only addition comment relates to the other versions of these file. Should we make sure all aclocal dirs are at the same level for files shared ? 2010-05-27T20:12:00 I checked and canonicalize-tools is already different. Whether this is right or not I don't know. 2010-05-27T20:12:04 I have been using that test for years in my test setup which is automake based. 2010-05-27T20:12:29 Different else where in the tree or just this file ? 2010-05-27T20:13:00 elsewhere in the tree. There are 3 versions in the tree 2010-05-27T20:13:09 And they are different ? 2010-05-27T20:13:15 yep 2010-05-27T20:13:34 I will raise this in the PR and point it to this discussion. 2010-05-27T20:14:22 The NFS mount is nice now. All you need is a shell and then you can: 2010-05-27T20:14:41 $ mount -t nfs joel:/work /work 2010-05-27T20:14:46 ok thanks. please review generate_symbols in gcc-testing/rtems-coverage. It has the list of sub-libraries in the development analysis. If you want something added, it is obvious how to do it. 2010-05-27T20:14:56 Marvelous!!! 2010-05-27T20:15:10 Yeah no extra calls need to be made. 2010-05-27T20:15:27 The confdefs.h is now a little complex with disable, set base etc option. 2010-05-27T20:15:53 well confdefs.h is supposed to be as complicated as needed to make life easy for applications 2010-05-27T20:16:17 I have documented the options in confdefs.h and will update the doc soon 2010-05-27T20:16:32 We need a file systems section for the manual. 2010-05-27T20:16:44 OK. Whatever makes sense... 2010-05-27T20:16:54 well back to real life.. family needs me 2010-05-27T20:17:10 I think all we need is to document the options for confdefs.h and the mount options. 2010-05-27T20:54:40 that sounds good 2010-05-27T21:08:00 DrJoel, do you anything about the pathconf_limis_and_options struct in the mount table ? 2010-05-27T21:08:03 *** dr__house has joined #rtems 2010-05-27T21:24:58 I think I implemented it but it has been years 2010-05-27T21:25:21 I suspect it is not initialised and that is not great. 2010-05-27T21:26:30 I will add a default to the mount command. It would be good to know what the fields mean. 2010-05-27T21:26:56 man fpathconf 2010-05-27T21:27:18 eg IMFS has posix_vdisable = 6 ! 2010-05-27T21:28:35 and did you see the comment in ./libfs/src/imfs/imfs_config.c /* XXX this structure should use real constants */ 2010-05-27T21:28:50 sounds like Jennifer and I never went back and corrected it. :( 2010-05-27T21:28:56 Ah no. I look in tftpfs and it is a copy of IMFS :) 2010-05-27T21:29:11 they are just unique numbers for debugging. 2010-05-27T21:29:23 ok. I will clean this up. 2010-05-27T21:29:41 We need to add doc for this. I also need to think the RFS using this data. 2010-05-27T21:29:45 * DrJoel wonders if 4.10 will ever branch 2010-05-27T21:30:36 I know I am digging deeper into this stuff, but it is almost done. 2010-05-27T21:30:56 I wasn't really picking at you.. just in general 2010-05-27T21:31:29 I know. Given the API change I am happy to get this done now for 4.10. 2010-05-27T21:31:47 yep.. as much as I want a release, I want things right 2010-05-27T21:32:34 fyi the Go student is now starting to figure out how to run the gcc Go test suite. Luckily it is DejaGNU also so the infrastructure I have for C/C++ is useful 2010-05-27T21:33:05 That is great. Did he get a resolution about the utime API ? 2010-05-27T21:33:26 * kiwichris intended the pun 2010-05-27T21:34:17 LOL.. I think Go will just use what's available and doesn't care .. utimes is preferred .. even if it "may be removed", it is still there in the current standard 2010-05-27T21:34:30 need to start heading to bed .. catch you later 2010-05-27T21:34:36 cya 2010-05-27T21:59:54 *** alexp0205 has quit IRC 2010-05-27T23:20:19 *** dr__house has quit IRC 2010-05-27T23:45:01 *** kiwichris has quit IRC 2010-05-28T00:38:55 *** kiwichris has joined #rtems 2010-05-28T01:00:50 *** Raj_ has joined #rtems 2010-05-28T01:39:39 *** exception13 has joined #rtems 2010-05-28T01:42:48 *** Raj_ has quit IRC 2010-05-28T02:20:19 *** Raj_ has joined #rtems 2010-05-28T02:51:00 *** Raj_ has quit IRC 2010-05-28T04:07:01 *** dr__house has joined #rtems 2010-05-28T04:17:54 *** aniceberg has joined #rtems 2010-05-28T04:20:03 *** aniceberg has quit IRC 2010-05-28T04:20:30 *** aniceberg has joined #rtems 2010-05-28T05:09:23 *** madrazr has joined #rtems 2010-05-28T07:10:12 *** madrazr has quit IRC 2010-05-28T07:53:21 *** lekernel has quit IRC 2010-05-28T07:53:43 *** lekernel has joined #rtems 2010-05-28T07:59:07 *** tuxmaniac has joined #rtems 2010-05-28T07:59:18 *** exception13 has left #rtems 2010-05-28T09:13:14 *** bubaflub has joined #rtems 2010-05-28T09:14:35 *** madrazr has joined #rtems 2010-05-28T09:17:36 *** zwj has joined #rtems 2010-05-28T09:19:25 *** madrazr has quit IRC 2010-05-28T09:19:31 *** madrazr1 has joined #rtems 2010-05-28T09:20:41 *** madrazr has joined #rtems 2010-05-28T09:23:58 *** madrazr1 has quit IRC 2010-05-28T09:24:12 *** dr__house` has joined #rtems 2010-05-28T09:24:19 *** dr__house has quit IRC 2010-05-28T09:24:28 *** dr__house` is now known as dr__house 2010-05-28T09:27:53 *** tuxmaniac has quit IRC 2010-05-28T09:41:55 *** exception13 has joined #rtems 2010-05-28T09:55:03 *** dr__house has quit IRC 2010-05-28T10:01:48 *** madrazr has quit IRC 2010-05-28T10:03:59 *** zwj has quit IRC 2010-05-28T10:11:25 *** zwj has joined #rtems 2010-05-28T10:13:42 *** alexp0205 has joined #rtems 2010-05-28T10:37:28 *** zwj has quit IRC 2010-05-28T10:38:39 *** aniceberg has quit IRC 2010-05-28T10:43:25 *** aniceberg has joined #rtems 2010-05-28T10:46:57 *** zwj has joined #rtems 2010-05-28T10:58:04 *** alexp0205 has quit IRC 2010-05-28T11:18:06 *** alexp0205 has joined #rtems 2010-05-28T11:43:08 *** mwalle has quit IRC 2010-05-28T12:02:45 *** zwj has quit IRC 2010-05-28T12:14:52 *** alexp0205 has left #rtems 2010-05-28T12:21:31 *** dr__house has joined #rtems 2010-05-28T13:00:38 * DrJoel waves hi 2010-05-28T13:12:43 * dr__house waves back :) 2010-05-28T13:23:48 well relatives are arriving .. see you all later 2010-05-28T13:23:51 *** DrJoel has quit IRC 2010-05-28T13:34:33 *** aniceberg has quit IRC 2010-05-28T13:50:37 *** krzabr has joined #rtems 2010-05-28T13:52:10 which scrypt language is widest using in real time programming ? 2010-05-28T13:52:15 anybody knows ? 2010-05-28T13:52:48 krzabr: what exactly do you mean? 2010-05-28T13:56:41 i'm interested which languages are used under realtime systems 2010-05-28T13:58:29 like eel 2010-05-28T13:58:32 http://eel.olofson.net/ 2010-05-28T13:59:28 *** krzabr has left #rtems 2010-05-28T13:59:38 *** krzabr has joined #rtems 2010-05-28T14:08:02 *** lekernel has quit IRC 2010-05-28T14:14:00 *** krzabr has quit IRC 2010-05-28T14:22:03 *** lekernel has joined #rtems 2010-05-28T15:11:02 *** madrazr has joined #rtems 2010-05-28T15:36:12 *** madrazr has quit IRC 2010-05-28T15:38:23 *** exception13 has left #rtems 2010-05-28T15:51:11 *** madrazr has joined #rtems 2010-05-28T16:04:56 *** bubaflub has quit IRC 2010-05-28T16:26:34 *** madrazr has left #rtems 2010-05-28T17:15:23 *** dr__house has quit IRC 2010-05-28T17:47:00 *** kiwichris has quit IRC 2010-05-28T17:58:18 *** bubaflub has joined #rtems 2010-05-28T18:10:17 *** kiwichris has joined #rtems 2010-05-28T18:24:09 *** bubaflub has quit IRC 2010-05-28T18:24:30 *** bubaflub has joined #rtems 2010-05-28T18:43:01 *** lekernel has quit IRC 2010-05-28T19:30:20 *** bubaflub_ has joined #rtems 2010-05-28T19:33:05 *** bubaflub has quit IRC 2010-05-28T19:33:05 *** bubaflub_ is now known as bubaflub 2010-05-28T19:53:58 kiwichris: hey, i was poking around the parrot configuration script and it looks like what i would have needed to patch is already there. i should have some goodies by next Tuesday 2010-05-28T20:16:14 *** alexp0205 has joined #rtems 2010-05-28T20:54:22 *** alexp0205 has left #rtems 2010-05-28T20:57:54 bubaflub, hi 2010-05-28T20:58:07 That is great news 2010-05-28T20:58:50 kiwichris: yep. instead of specifying --cc and all those options on the commandline, i can actually pass it a file with all of the configuration options 2010-05-28T20:58:57 and there is a command-line switch to specify which hints file to use 2010-05-28T20:59:04 so now i've just got to generate those two files 2010-05-28T20:59:08 oh, and by the way, how is your daughter? 2010-05-28T20:59:29 She is much better now. Just a stupid virus. 2010-05-28T20:59:49 That is great news on build system. The less we do the better. 2010-05-28T20:59:55 *** zwj has joined #rtems 2010-05-28T21:00:09 good to hear. 2010-05-28T21:00:21 and yeah, i agree. darbelo had already put in that option a while ago. so the cross-building should be a bit easier 2010-05-28T21:00:28 just need to figure out *all* of the options that we need on and off 2010-05-28T21:00:34 there is even the option to skip configuration steps 2010-05-28T21:00:36 brb 2010-05-28T21:00:39 so i can have it not probe if we are on a mac, etc. 2010-05-28T21:00:41 k 2010-05-28T21:25:32 kiwichris: i'm going to drive around aimlessly with my fiancee, i'll catch ya later 2010-05-28T21:25:34 *** bubaflub has quit IRC 2010-05-28T21:25:59 cya. sorry still busy. Have fun. 2010-05-28T21:33:22 *** kiwichris has quit IRC 2010-05-28T22:32:47 *** kiwichris has joined #rtems 2010-05-28T23:28:36 *** madrazr has joined #rtems 2010-05-29T00:12:11 *** lcpfnvc has quit IRC 2010-05-29T00:12:15 *** kiwichris has quit IRC 2010-05-29T00:47:46 *** dr__house has joined #rtems 2010-05-29T01:23:46 *** madrazr has quit IRC 2010-05-29T02:06:40 *** kiwichris has joined #rtems 2010-05-29T02:16:52 *** kiwichris has quit IRC 2010-05-29T02:42:22 *** kiwichris has joined #rtems 2010-05-29T04:30:42 *** lekernel has joined #rtems 2010-05-29T05:09:12 *** peerst has joined #rtems 2010-05-29T07:55:28 *** tuxmaniac has joined #rtems 2010-05-29T08:42:06 *** tuxmaniac has quit IRC 2010-05-29T09:07:47 *** kiwichris has quit IRC 2010-05-29T09:28:30 *** zwj has quit IRC 2010-05-29T09:35:01 *** zwj has joined #rtems 2010-05-29T09:47:14 *** dreamkeep has joined #rtems 2010-05-29T09:50:36 hello 2010-05-29T09:51:11 *** dreamkeep has quit IRC 2010-05-29T10:05:07 *** dr__house has quit IRC 2010-05-29T10:40:38 *** mwalle has joined #rtems 2010-05-29T11:10:11 *** exception13 has joined #rtems 2010-05-29T11:11:22 *** zwj has quit IRC 2010-05-29T11:28:35 *** Raj_ has joined #rtems 2010-05-29T11:35:14 *** Raj_ has quit IRC 2010-05-29T12:46:53 *** dr__house has joined #rtems 2010-05-29T12:51:17 *** lcpfnvc has joined #rtems 2010-05-29T13:02:55 *** aniceberg has joined #rtems 2010-05-29T13:24:13 *** madrazr has joined #rtems 2010-05-29T13:30:30 *** aniceberg has quit IRC 2010-05-29T13:31:16 *** tuxmaniac has joined #rtems 2010-05-29T13:35:42 *** tuxmaniac has quit IRC 2010-05-29T13:36:50 *** tuxmaniac has joined #rtems 2010-05-29T14:12:55 *** dr__house has quit IRC 2010-05-29T15:59:01 *** lcpfnvc has quit IRC 2010-05-29T16:48:49 *** exception13 has left #rtems 2010-05-29T16:51:15 *** bubaflub has joined #rtems 2010-05-29T16:57:52 *** bubaflub has quit IRC 2010-05-29T16:59:41 *** madrazr has quit IRC 2010-05-29T17:14:07 *** madrazr has joined #rtems 2010-05-29T17:17:05 *** lekernel has quit IRC 2010-05-29T17:17:31 *** madrazr has left #rtems 2010-05-29T17:35:30 *** Fallenou_ is now known as Fallenou 2010-05-29T17:38:00 *** kiwichris has joined #rtems 2010-05-29T17:49:32 *** aniceberg has joined #rtems 2010-05-29T18:01:22 *** DrJoel has joined #rtems 2010-05-29T18:01:22 *** ChanServ sets mode: +o DrJoel 2010-05-29T18:03:26 Joel ? 2010-05-29T18:03:48 yes? Just woke up from a nap.. long weekend so far 2010-05-29T18:04:09 Nice. Just woke up :) 2010-05-29T18:04:26 Well it is 6pm here so ... 2010-05-29T18:04:39 Sunday morning sleep in. Just watching the tennis replays from Paris 2010-05-29T18:04:58 Yeah I have Foxclocks running and it is great for checking times. 2010-05-29T18:05:11 Just fixing the TFTP file system. 2010-05-29T18:05:34 i saw you posted a number of PRs .. had no idea how many had fixes in hand or close 2010-05-29T18:06:12 I assigned the filesystem to me. You have most of the cats in bugzilla 2010-05-29T18:06:37 that's because I am the default for most of them. :( 2010-05-29T18:06:39 I will go through them all as I fix the things I am on. 2010-05-29T18:06:51 If you see one you think I should address ASAP ping me 2010-05-29T18:06:51 Yeah and that is a bit tough on you. 2010-05-29T18:06:58 I will. 2010-05-29T18:08:22 So Jordan and Rob graduated high school last night. We had ~25 relatives in. This morning, I visited with them and then went to a local nature preserve to help with a 10th anniversary of geocaching "bltiz". They had GPS receivers to loan to people and a course for them to run on the trail 2010-05-29T18:09:14 Congrats on the graduations and well done to both of them. 2010-05-29T18:09:27 :) 2010-05-29T18:10:46 It was a wild west show of relatives.. neices, nephews, parent, current in-laws, former spouses and our ex in-laws, and every grandparent 2010-05-29T18:34:42 *** tuxmaniac has quit IRC 2010-05-29T18:47:17 Hello DrJoel 2010-05-29T18:48:41 broke my fedora and now fixing @ 2am :(. Running do_coverage once to make sure all was fine... 2010-05-29T18:49:19 *** alinrus has joined #rtems 2010-05-29T18:49:43 Also, now facing a problem i didnt have earlier. the .exe.cov each take 11MB and quickly I am out of disk space. not sure why it popped up now. 2010-05-29T18:57:45 * aniceberg needs to head home now. 2010-05-29T18:58:30 later folks. have a nice weekend. 2010-05-29T18:58:34 *** aniceberg has quit IRC 2010-05-29T21:10:30 *** zwj has joined #rtems 2010-05-29T23:48:34 *** dr__house has joined #rtems 2010-05-30T01:08:27 *** alinrus has quit IRC 2010-05-30T01:25:38 *** kiwichris has quit IRC 2010-05-30T02:04:10 *** lcpfnvc has joined #rtems 2010-05-30T02:33:33 *** alinrus has joined #rtems 2010-05-30T02:34:34 DrJoel: hi ,are you around? 2010-05-30T02:37:03 when i install rtems-4.11-arm-rtems4.11-gdb.i386, it say libpython2.4.so.1.0 is needed ,but in my fedroa10 libpython2.5.so exists 2010-05-30T02:37:42 so it fails to install libpython2.4.so.1.0 2010-05-30T02:37:57 anyone knows how to resolve it? 2010-05-30T03:06:11 ok, i have resolved it . i rpm --force the libpython2.4.so.1.0 2010-05-30T04:24:45 *** lekernel has joined #rtems 2010-05-30T05:06:03 *** lcpfnvc has quit IRC 2010-05-30T05:22:31 *** exception13 has joined #rtems 2010-05-30T05:25:12 *** lcpfnvc has joined #rtems 2010-05-30T06:13:01 *** tuxmaniac has joined #rtems 2010-05-30T06:46:14 *** Raj_ has joined #rtems 2010-05-30T06:46:23 DrJoel: hi 2010-05-30T07:29:23 *** _Lucretia__ is now known as _Lucretia_ 2010-05-30T07:29:30 *** _Lucretia_ has joined #rtems 2010-05-30T08:05:13 *** exception13 has left #rtems 2010-05-30T08:05:19 *** exception13 has joined #rtems 2010-05-30T08:17:07 hi,everyone 2010-05-30T08:29:37 today when i used rtems-sparc-gdb to debug the sample_test there is nothing to print. do you know the reason? thanks 2010-05-30T08:57:16 * DrJoel waves good morning 2010-05-30T08:57:38 zwj.. how is the test printing? 2010-05-30T08:57:50 Raj_: hi 2010-05-30T08:58:03 Hi 2010-05-30T08:58:28 As I said the other day, those errors are persisting 2010-05-30T08:58:36 DrJoel: starting program: hello.exe program exited normally 2010-05-30T08:59:08 it seems the program is run ok but no console information 2010-05-30T08:59:13 zwj: Is this with the standard hello and fedora 12 RPM for sparc-rtems gdb? 2010-05-30T08:59:43 Raj_: which error? Sorry it ihas been a long weekend with two of my sons graduating high school, many relatives, etc 2010-05-30T09:00:16 Ahh. Congrats on the good news :) 2010-05-30T09:00:21 the hello.exe is modified by me with sequence_init code, and use fedora 10 rpm for sparc-rtems gdb 2010-05-30T09:00:43 Well, I tried running the Go tests inside the gcc testsuite using gcc-testing module 2010-05-30T09:00:57 my development envoriment is fedora 10 2010-05-30T09:01:02 zwj: are you initializing the device drivers? printf requires the console driver to work 2010-05-30T09:01:08 the compilation ends in 2 errors for nearly all cases 2010-05-30T09:01:50 Raj_: ok what are the errors? 2010-05-30T09:02:07 i doesnot modify the device initializing code, i just modify the RTEMS_API_INITIALIZION 2010-05-30T09:03:25 b-i486-gccgo/i386-rtems4.10/./libgo/.libs/libgo.a(libsyscall.a.o): In function `libgo__.syscall.Lchown': b-i486-gccgo/i386-rtems4.10/libgo/../../../gccgo/libgo/syscalls/sysfile_rtems.go:272: undefined reference to `lchown' 2010-05-30T09:03:49 zwj ... hmmm.. when you run does it exit or lock up? 2010-05-30T09:04:02 DrJoel: this is for lchown, and the next is for utimes, exactly similar error 2010-05-30T09:04:29 Raj_: does your rtems have those in the tree? Did you do a make install on that? Or did the script do it? 2010-05-30T09:04:51 Raj_: Or did you install multilib before you added these and haven't done it since? 2010-05-30T09:05:08 Yes, my rtems has those in the tree 2010-05-30T09:05:35 DrJoel: i use sparc-rtems-gdb hello.exe and tar sim, load, r and then it print start" hello.exe and program exited normally" 2010-05-30T09:05:37 And I have compiled multilib too 2010-05-30T09:05:42 Is the installed librtemscpu.a the one with it. 2010-05-30T09:05:50 Raj_: and installed? 2010-05-30T09:05:58 find $prefix -name librtemscpu.a 2010-05-30T09:05:59 Yes installed too :) 2010-05-30T09:06:34 DrJoel: and i just modified the code _RTEMS_API_Initialize function 2010-05-30T09:06:43 zwj .. can you break at Init and exit see if it gets to them? 2010-05-30T09:06:57 termios uses RTEMS Classic API Semaphores and it could have issues 2010-05-30T09:08:29 brb .. getting coffee 2010-05-30T09:08:46 DrJoel: i set break at Init and exit, but i doesnot break. and also print program exited normally 2010-05-30T09:09:23 DrJoel: Yes, librtemscpu.a is present in the directory 2010-05-30T09:12:00 zwj: bzip -9 the executable and send it to me 2010-05-30T09:12:47 Raj_: do an nm on it and make sure the symbols lchown, etc are in it .. if not, you need to install an rtems with them 2010-05-30T09:13:11 DrJoel: Already did that, they are present :) 2010-05-30T09:13:21 OK, i send the hello.exe to you 2010-05-30T09:13:30 so you did an nm on the library? 2010-05-30T09:13:40 Yes 2010-05-30T09:14:02 if yes, then what is a command line for a failing build? You may not have tinkered with the "VERSIONS" file to make it point to the right place 2010-05-30T09:15:03 DrJoel: I think the versions is ok, the command line is good 2010-05-30T09:15:50 Executing on host: /home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/gcc/testsuite/go/../../gccgo -B/home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/gcc/testsuite/go/../../ /home/vinu/Desktop/GSoC/clean/rtems/tools/gccgo/gcc/testsuite/go.go-torture/execute/array-1.go go_tg.o -I/home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/i386-rtems4.10/./libgo -w -O2 -fomit-frame-pointer -finline-functions -isystem /home/vinu 2010-05-30T09:16:10 DrJoel: What your email? your name dot gmail or oarcorp.com? 2010-05-30T09:16:28 zwj: either of those work 2010-05-30T09:16:33 I am at home and can get both 2010-05-30T09:16:44 Raj_: checkign a long here ... hold on 2010-05-30T09:17:47 and I don't have a log :( so what is the path to the rtemscpu.a 2010-05-30T09:17:54 DrJoel: the copy was not complete, will do a line by line 2010-05-30T09:18:05 Executing on host: /home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/gcc/testsuite/go/../../gccgo -B/home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486 2010-05-30T09:18:31 gccgo/gcc/testsuite/go/../../ /home/vinu/Desktop/GSoC/clean/rtems/tools/gccgo/gcc/testsuite/go.go-torture/execute/array-1.go go_tg.o -I/home/vinu/Desktop 2010-05-30T09:20:05 DrJoel: i have sent it to your gmail 2010-05-30T09:20:08 GSoC/clean/rtems/tools/b-i486-gccgo/i386-rtems4.10/./libgo -w -O2 -fomit-frame-pointer -finline-functions -isystem /home/vinu/Desktop/GSoC/clean/rtems 2010-05-30T09:20:41 tools/b-i486-gccgo/i386-rtems4.10/./newlib/targ-include -isystem /home/vinu/Desktop/GSoC/clean/rtems/tools/gccgo/newlib/libc/include -B/home/vinu/Desktop/ 2010-05-30T09:21:07 GSoC/clean/rtems/opt/rtems-4.10/i386-rtems4.10/pc386/lib/ -specs bsp_specs -qrtems -B/home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/i386-rtems4.10/./ 2010-05-30T09:21:44 newlib/ -L/home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/i386-rtems4.10/./newlib /home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/gcc/ 2010-05-30T09:22:05 goRtems_init.o -Wl,-Ttext,0x00100000 -L/home/vinu/Desktop/GSoC/clean/rtems/tools/b-i486-gccgo/i386-rtems4.10/./libgo -L/home/vinu/Desktop/GSoC/clean/rtems/ 2010-05-30T09:22:26 tools/b-i486-gccgo/i386-rtems4.10/./libgo/.libs -Wl,-wrap,exit -Wl,-wrap,_exit -Wl,-wrap,main -Wl,-wrap,abort -lm -o /home/vinu/Desktop/GSoC/clean/rtems/ 2010-05-30T09:22:41 tools/b-i486-gccgo/gcc/testsuite/go/array-1.x (timeout = 300) 2010-05-30T09:22:53 zwj: It is blowing up in _RTEMS_API_Initialize() 2010-05-30T09:23:15 (gdb) r 2010-05-30T09:23:16 Starting program: /tmp/hello.exe 2010-05-30T09:23:16 Breakpoint 3, _RTEMS_API_Initialize () 2010-05-30T09:23:16 at /mnt/hgfs/rtems/c/src/../../cpukit/sapi/src/rtemsapi.c:68 2010-05-30T09:23:16 68 /mnt/hgfs/rtems/c/src/../../cpukit/sapi/src/rtemsapi.c: No such file or directory. 2010-05-30T09:23:16 in /mnt/hgfs/rtems/c/src/../../cpukit/sapi/src/rtemsapi.c 2010-05-30T09:23:18 (gdb) s 2010-05-30T09:23:20 70 in /mnt/hgfs/rtems/c/src/../../cpukit/sapi/src/rtemsapi.c 2010-05-30T09:23:22 (gdb) 2010-05-30T09:23:26 71 in /mnt/hgfs/rtems/c/src/../../cpukit/sapi/src/rtemsapi.c 2010-05-30T09:23:28 (gdb) 2010-05-30T09:23:30 Program exited normally. 2010-05-30T09:23:36 Since you have the right source, you should be able to see which variable(s) aren't right 2010-05-30T09:24:13 DrJoel: should I send you a mail of the cmd line ? 2010-05-30T09:25:16 Raj_: that would help.. Include the path of your librtemscpu.a 2010-05-30T09:25:30 *** titurel has joined #rtems 2010-05-30T09:25:52 by the way path to rtemscpu.a is - /home/vinu/Desktop/GSoC/clean/rtems/opt/rtems-4.10/i386-rtems4.10/pc386/lib/ 2010-05-30T09:28:21 DrJoel: I have sent you the mail 2010-05-30T09:28:30 got it.. reading phdcomics.com LOL 2010-05-30T09:28:55 :-D 2010-05-30T09:29:49 -B/home/vinu/Desktop/GSoC/clean/rtems/opt/rtems-4.10/i386-rtems4.10/pc386/lib/ 2010-05-30T09:29:49 -specs bsp_specs -qrtems 2010-05-30T09:30:02 is the fragment that should make it pick up the librtemscpu.a 2010-05-30T09:30:29 i386-rtems4.10-nm -g /home/vinu/Desktop/GSoC/clean/rtems/opt/rtems-4.10/i386-rtems4.10/pc386/lib/librtemscpu.a | grep lchown 2010-05-30T09:30:40 What's the output 2010-05-30T09:31:17 Yes, let me check 2010-05-30T09:32:19 DrJoel: no output ! 2010-05-30T09:32:34 then lchown isn't in there. :-D 2010-05-30T09:32:46 if i386-rtems4.10/lib/librtemscpu.a I get output 2010-05-30T09:33:08 I never did nm for the bsp specific lib 2010-05-30T09:33:15 but that sounds like it is in the build tree 2010-05-30T09:33:43 no I mean /home/vinu/Desktop/GSoC/clean/rtems/opt/rtems-4.10/i386-rtems4.10/lib/librtemscpu.a 2010-05-30T09:33:53 that's the one from the multilib build.. and aparently you haven't done an install on it since adding the new methods 2010-05-30T09:34:32 how do i check the config options again ? 2010-05-30T09:34:35 for the build ? 2010-05-30T09:35:12 DrJoel: ok, i know the reason, thanks 2010-05-30T09:35:24 When I build to test Ada, I build and install RTEMS twice, multilib and then for a BSP. See the do_one script 2010-05-30T09:35:55 You just have missed that.. A simple case will work but this build is finding that multilib library without the symbols 2010-05-30T09:36:26 as an experiment just remove the non-bsp specific library and run again. It should find the one for the BSP (does it have the symbols) 2010-05-30T09:36:56 let me see 2010-05-30T09:38:41 DrJoel: my option for running single case examples is -B/home/vinu/Desktop/GSoC/clean/rtems/opt/rtems-4.10/i386-rtems4.10/pc486/lib/ 2010-05-30T09:39:38 and pc386 is a symlink to that.. the args look right so that if you remove the one in i386-rtems4.10/lib, it will find that one. If it has the symbols, the tests should link 2010-05-30T09:40:13 So you mean it links against the other librtemscpu, but why ? 2010-05-30T09:40:20 when I have given this option ? 2010-05-30T09:41:47 before it in the library search path. That is the default system library to find libc.a, etc. It also lets i386-rtems4.10/include be picked up automatically without being aware of the BSP. That's why Ada is built with a multilib RTEMS first and then a BSP. It makes it easier to find .h files with no magic. Especially the TCP/IP ones 2010-05-30T09:43:45 DrJoel: The single case is still building after removing the other librtemscpu.a 2010-05-30T09:44:51 try the tests now 2010-05-30T09:45:12 ok 2010-05-30T09:46:13 DrJoel: But why the tests now, I just removed the librtemscpu.a 2010-05-30T09:46:21 the tests should still fail 2010-05-30T09:47:20 There should be one in the bsp specific directory with lchown and it should now be finding that one instead of the one left over from an older pre-lchown multilib build. 2010-05-30T09:47:44 Eventually you will have to clean up the install point and install the tools and rtems multilib + BSP but for now, this should resolve things 2010-05-30T09:48:59 I had to delete the librtemscpu inside pc486/ ? 2010-05-30T09:49:13 the bsp-specific one is this right ? 2010-05-30T09:49:39 no.. the one which was NOT BSP specific ... i386-rtems4.10/lib/librtemscpu.a .. it did not have lchown and is being picked up. 2010-05-30T09:50:12 DrJoel: No, this one had lchown 2010-05-30T09:50:19 when you are modifying things and adding/deleting/changing things in installed libraries you have to be VERY careful you are getting the library you mean to. 2010-05-30T09:50:20 the bsp-specific didn't have it 2010-05-30T09:50:45 How about we not try to hack it. :) 2010-05-30T09:50:53 build and install multilib.. 2010-05-30T09:50:57 build and install BSP 2010-05-30T09:51:00 then run tests. 2010-05-30T09:51:03 then we should be sure 2010-05-30T09:51:16 Well I can copy the lib/librtemscpu into pc486/ ? 2010-05-30T09:51:22 wouldn't that work ? 2010-05-30T09:51:22 sure 2010-05-30T09:51:27 should 2010-05-30T09:51:51 let me try that, but I haven't understood yet why my single cases are working :) 2010-05-30T09:52:14 different combinations of arguments are picking up the libraries in a different order 2010-05-30T09:53:51 DrJoel: Yes, but the options in my single case examples are to pick up the bsp-specific librtemscpu 2010-05-30T09:54:50 I cannot understand why it picks up the other one because there are no other options and LD_LIBRARY_FLAGS is not set too ? 2010-05-30T09:55:38 I am confused at this point.. we just need to make sure the installed RTEMS .h files and libraries are consistent with youy current source and have all the symbols.. it will work then 2010-05-30T09:57:53 DrJoel: Yes, well, that's why I was confused too, because I thought that was the case :) 2010-05-30T09:58:29 When doing C, C++, and Ada tests for reporting, I usually remove everything at the test install point and build EVERYTHING from scratch 2010-05-30T10:00:19 Hmm, well now it's saying simulator pc386 missing :) 2010-05-30T10:00:21 *** lcpfnvc has quit IRC 2010-05-30T10:00:29 PATH? 2010-05-30T10:01:27 yes ofcourse, I switched the computer off from yesterday :D 2010-05-30T10:05:16 DrJoel: Yes, this error is gone :), but some new errors :) . Have to go for dinner now :) . Bye 2010-05-30T10:05:25 bye 2010-05-30T10:05:31 *** Raj_ has quit IRC 2010-05-30T10:16:20 *** lcpfnvc has joined #rtems 2010-05-30T11:19:02 *** zwj has quit IRC 2010-05-30T12:06:46 *** alexp0205 has joined #rtems 2010-05-30T12:07:27 *** Raj_ has joined #rtems 2010-05-30T12:11:01 DrJoel: hello 2010-05-30T12:13:24 Raj_: so what's the new error 2010-05-30T12:13:54 The new error is related to the config options I guess 2010-05-30T12:14:11 it says something like work space greater than work area 2010-05-30T12:14:57 that's where the tedious part comes in. Hmm.. that sounds like the message in bootcard.c which means you don't have enough RAM on the board. Look at the pc386.in file in sim-scripts and increase the memory (the -m option to qemu) 2010-05-30T12:15:34 Well, I actually set the maximum threads, mutexes and conditional variables quiet high 2010-05-30T12:15:54 will reduce them first, then try to increase memory of qemu 2010-05-30T12:16:44 Well, I was wondering why the librtemscpu inside pc486 folder didn't have lchown and utimes ? 2010-05-30T12:17:10 isn't that installed during rtems installation ? 2010-05-30T12:28:19 DrJoel: why isn't the correct librtemscpu installed in pc486/lib after rtems compilation and installation ? 2010-05-30T12:34:02 Raj_: it is supposed to be.. watch the log while installing and see the path... maybe the configure is wrong or something. Symlink may be confusing things also 2010-05-30T12:34:08 * DrJoel is heading out to run errands.. bbl 2010-05-30T13:01:48 Raj_: any luck? Waiting for kids to be ready.. 2010-05-30T13:02:12 librtemcpu.a is copied only for multilib 2010-05-30T13:03:01 I don't know how librtemscpu is installed inside pc486/lib :) 2010-05-30T13:03:16 hmm.. ok.. must be detecting it is already there as a multilib. When you build and install just as a single BSP without it previously been installed as multilib, it goes into the BSP directory 2010-05-30T13:03:27 there may be debug code in score/src/objectallocate.c which might help 2010-05-30T13:04:40 you mean if I install without multilib, then it should be installed inside pc486/lib right ? 2010-05-30T13:05:18 DrJoel: will try that and checkout the code too. Thanks 2010-05-30T13:05:43 * Raj_ calls it a day 2010-05-30T13:05:59 *** Raj_ has quit IRC 2010-05-30T13:13:46 *** alexp0205 has left #rtems 2010-05-30T13:39:00 *** tuxmaniac has quit IRC 2010-05-30T13:59:03 *** madrazr has joined #rtems 2010-05-30T14:09:52 *** lekernel has quit IRC 2010-05-30T14:17:14 *** titurel has left #rtems 2010-05-30T14:57:01 *** madrazr has quit IRC 2010-05-30T15:12:22 *** madrazr has joined #rtems 2010-05-30T15:14:33 *** alinrus has quit IRC 2010-05-30T16:00:50 *** exception13 has left #rtems 2010-05-30T16:28:53 *** dr__house has quit IRC 2010-05-30T17:08:07 *** madrazr has left #rtems 2010-05-30T19:10:10 *** kiwichris has joined #rtems