RTEMS IRC Log, 10 Aug 2009

*** kaluzamorph has joined #rtems00:31
*** dr__house has joined #rtems00:44
*** sebhub has joined #rtems01:59
* dr__house is away: afk02:42
*** BBHoss has joined #rtems03:01
*** dr__house has quit IRC03:33
*** BBHoss has quit IRC03:40
*** BBHoss has joined #rtems03:40
*** cocan has joined #rtems04:00
*** dr__house has joined #rtems04:06
*** BBHoss has quit IRC04:16
*** madrazr has joined #rtems04:35
*** cocan has quit IRC05:44
*** madrazr has quit IRC05:56
*** cocan has joined #rtems06:00
*** cocan_ has joined #rtems06:26
*** cocan has quit IRC06:33
*** roxanal has joined #rtems07:23
*** dr__house has quit IRC07:24
*** _Lucretia_ has quit IRC07:25
*** _Lucretia_ has joined #rtems07:26
roxanaltuxmaniac: hi07:26
tuxmaniacroxanal: hi07:26
roxanaltuxmaniac: I just wanted to let you know I am working on the patch for the arm BSP right now07:27
roxanalI didn't had internet access over the weekend and I just saw the message07:28
tuxmaniacroxanal: I was very confused. I checked with DrJoel and he said the email was for Aaron ( agrier ) and it seems Xi is taking care of it now07:29
tuxmaniacanyay thanks for the info07:29
roxanalI see07:29
*** DrJoel has joined #rtems07:48
*** ChanServ sets mode: +o DrJoel07:48
roxanalDrJoel: good morning07:48
DrJoelroxanal: good morning.  How is the move going?07:48
roxanalDrJoel: we have everything ,but the internet :D I am at school now so I can work07:49
roxanalDrJoel: is anyone else taking care of the arm frame buffer error? or should I do it?07:50
tuxmaniacDrJoel: looks like things are working. atleast psim now hits exception when an illegal memory is accesed, executes my exception handler and decides to continue or fail07:50
tuxmaniacnow to add mature features. I still have my doubts. Will mail thomas and ask him for a short meetup today if possible07:50
tuxmaniacDrJoel: and oh. -O0 helped :)07:51
roxanalI can't test it though -- I know what is the problem is and cvs up doesn't show any changes where is should be fixed07:51
DrJoelSo far no one else has taken care of it.  It would be appreciated.07:51
DrJoeltuxmaniac: when speed isn't an issue that often improves debugging. :)07:52
roxanalDrJoel: I would like to know who changed the fb.h file but there is no entry in the ChangeLog07:52
DrJoel_Lucretia_: gcc 4.4.0 died with an internal error building arm gnat.  I am trying 4.4.1.  If that fails, I will report a bug and then will back to 4.3.x07:53
roxanalDrJoel: I will fix it but I need somebody to test it .... otherwise it will take me a while to install the tools and everything07:53
DrJoelroxanal: you can do a cvs log to verify all changes but this file have never been modified since it was initially merged -- still at revision 1.1.07:54
DrJoelI can verify it compiles .. no problem.07:54
roxanalDrJoel: I will submit some patches soon07:55
* agrier is up early07:55
agrierjoel: the arm framebuffer is for skyeye, correct?07:55
roxanalagrier: Good morning ... you are up really early -- the arm frame buffer is for skyeye I think07:56
DrJoelagrier: nds is the one actually broken ... Xi Yang is working on the one for skyeye and it is in the gumstix BSP07:57
DrJoelso there are two arm fb drivers07:57
roxanalDrJoel: how I can make a cvs up and to force to discard my changes?07:59
agriergenerate a diff and patch -R it?  :)07:59
DrJoelor remove the file you want reverted and do a cvs up08:00
roxanalthank you08:00
DrJoelroxanal: no problem08:04
_Lucretia_DrJoel: you need to check that script of yours, it decided to start deleting stuff in $HOME08:13
_Lucretia_and I had to kill the computer to stop it08:13
DrJoelWhat variables/rm command was it?  Usually that type of thing happens when base variables aren't set as expected08:15
_Lucretia_I set the variables, but i didn't call the script from within gcc-testing, is that the problem?08:21
DrJoelI usally invoke it with a full path to the script but I doubt that's the issue08:22
_Lucretia_either way, i'm not risking that again08:22
DrJoelclean_up()08:23
DrJoel{08:23
DrJoel  rm -rf ${INSTALL}/* &08:23
DrJoel  rm -rf ${BSP_INSTALL}/* &08:23
DrJoel  wait08:23
DrJoel}08:23
_Lucretia_so, i'm building from scratch using your instructions, but using the cvs versions of everything, no patches atm08:23
DrJoelWhat were those set to?08:23
_Lucretia_http://pastebin.com/m7a58c70208:23
DrJoelI don't see anything weird.  The only potential issue I see is that you the install point has subdirectories.. but that wouldn't impact the rm only when make install tries to install08:25
DrJoelIf the variables didn't propagate it would have started rm -rf at / which would be different08:26
tuxmaniacDrJoel: whenever I change something in the init.c test suite file the mmutest.exe gets relinked and re-generated. But a change in libcpu doesnt do it. Is there something that I am missing08:31
DrJoelno.. just a missing dependency .. the rtems libraries are treated as system libraries and ignored (right or wrong) as dependencies.  I usually rm -f the .exe and then build08:32
roxanalDrJoel: it looks like this frame buffer was never used, structures are missing -- I don't think it ever compiled :D08:32
tuxmaniacDrJoel: thats what I do now. thought there is something wrng with my tree. expected this anyway. thanks a lot08:34
DrJoelthat bsp compiled a long time.. fb.h is included in a lot of its source code .. I wonder it is just an NDS specific framebuffer since it looks to be copied from their toolkit08:35
DrJoeli see the problem.. when we renamed mw_fb.h -> fb.h, it conflicted with this and we are using the generic one.. not the NDS specific one from the NDS kit.  I need to rename this to "nds_fb.h" ..08:37
DrJoelmove on to something else.  I will fix this. :)08:37
roxanalDrJoel: I don't think it will fix the problem08:38
roxanaleven if you rename it, there will still be structures that are not defined08:39
DrJoelouch...08:40
DrJoelok.. let me think..08:40
roxanalfb_screeninfo on this framebuffer is not defined - i changed it in mine and it is not finding it anymore08:41
_Lucretia_DrJoel: well, it might've started at / I don't know, but it wouldn't have had permission to do anything there08:41
roxanalDrJoel: that is one example08:41
DrJoel_Lucretia_: I know :(  I just don't see where you set anything wrong and it would have confused the script08:42
roxanalI can leave the header file alone but I have to change the fb.c file to be sure it is still working08:42
_Lucretia_the only thing I can think of is the use of ${var} rather than $var08:42
_Lucretia_Iuse bash and use $var08:42
DrJoel_Lucretia_: that shouldn't make any difference.. it just ensures the var name is parsed without fear of running it into something else .. just weird .. did a symlink get followed?08:43
_Lucretia_dunno08:43
_Lucretia_i dont have any in that dir except for the newlib, libgloss ones08:44
tuxmaniacroxanal: i see fb_screeninfo struct in my tree08:44
_Lucretia_ok, svn-gcc failed08:44
tuxmaniacin the path ../rtems/cpukit/libmisc/mw-fb/mw_fb.h08:44
roxanaltuxmaniac: you don't have the latest version then mw_fb.h doesn't exit anymore and that structure was changed by me :D08:45
roxanaltuxmaniac: so I know for sure that if you get rtems now and try to compile the arm/nds it will not work08:46
tuxmaniacroxanal: ok good. thanks for the info. I will remember not to do a cvs update until I am damn sure my implementation works and there can be nothing wrong before I cvs up.08:47
tuxmaniac:)08:47
roxanal:)08:47
DrJoelroxanal: bsp's fb.h was also being installed as rtems/fb.h .. overwriting the real one.  I switched it to nds/fb.h .. maybe it will be ok now.. rebuildin08:48
roxanalDrJoel: what do you mean by the real one :) ... This file was not making any reference to a fb.h ... it was pointing to mw_fb.h which was renamed and modified by me  to08:52
roxanalDrJoel: unless you have in nds/fb.h the original mw_fb.h it will not work08:53
DrJoelThe nds bsp has a file named fb.h which it installed as rtems/fb.h.  This conflicted with the one you are working on.  So it ended up including the nds version not the renamed mw_fb.h08:54
roxanalI see ... but still we are going to need to make some changes also :)08:55
DrJoelyep09:03
DrJoelroxanal: update and see email for the errors I get now09:06
roxanalDrJoel: is there anyway I can see the mw_fb.h file?09:06
roxanalthe old one?09:06
DrJoelyou could get it from cvs, 4.9, or just read your email. I keep all old branches checked out all the time09:08
roxanalthank you ... I will fix this soon09:09
tuxmaniacsebhub: is Thomas not feeling well or something. Its been 4 days since I heard from him. May be he is a bit too busy? Its OK. Inform him to leave an email for a meeting time for tomorrow. Thanks in advance09:11
_Lucretia_DrJoel: http://pastebin.com/m5d77af32 <- svn-gcc at b-gcc2 with: ../../gcc-svn/configure --enable-threads=rtems  --with-gnu-as --disable-multilib --with-gnu-ld --with-newlib  --verbose --with-system-zlib --disable-nls --enable-version-specific-runtime-libs -enable-languages=c,ada --target=arm-rtems4.10 --prefix=$HOME/opt/rtems/install 2>&1|tee log.config.txt09:12
DrJoellook at do_one where it builds the Ada.  RTEMS has already been built by the C compiler and installed.  There is an argument added so the ada build finds RTEMS and can build the OS constants file09:14
DrJoelbuild order is VERY VERY picky with gnat09:14
tuxmaniacDrJoel: can I pm you a bit?09:16
_Lucretia_what does --enable-newlib-mb do?09:18
DrJoelmulti-byte (wide char support)09:18
_Lucretia_ok, before building b-gcc2, do I need to compile rtems using the b-gcc1 compiler?09:20
DrJoelyes.  The "stage 1" compiler is C or C/C++ so you have a c library and can build RTEMS09:21
DrJoelAda needs a lot to build09:21
_Lucretia_thought so09:23
_Lucretia_aye, i know09:23
_Lucretia_i take it the config option for rtems "--enable-ada" enables the building of the Ada runtime stuff?09:34
_Lucretia_ok, that didn't get far: ../../../../../../../../../rtems/c/src/../../cpukit/libfs/src/nfsclient/src/sock_mbuf.c:228:13: error: storage size of 'auio' isn't known09:38
DrJoelRTEMS head, newlib head, and gcc head?09:39
DrJoelor rtems 4.9?09:39
_Lucretia_everything head09:40
_Lucretia_like i said earlier, all from cvs09:40
DrJoelhmmmm09:40
_Lucretia_it is also unsanswered on your ml09:41
_Lucretia_http://www.rtems.com/ml/rtems-users/2009/april/msg00128.html09:41
DrJoeli remember it but haven't seen it in ages.09:42
DrJoeli cut and pasted the entire diff I have against newlib cvs.09:44
DrJoeland i am on head with no changes for rtems.09:44
roxanalagrier: still around?09:46
*** th_d has joined #rtems09:48
DrJoelevening thomas09:59
*** switnick has joined #rtems10:02
* agrier is at work now10:03
* agrier thinks he needs to build an arm toolchain now to play along with roxana and tux10:06
roxanalagrier: is it ok to move the enter graphics/ exit graphix during the open/close of the device like in i386 for arm/nds?10:10
*** th_d has quit IRC10:11
DrJoelroxanal: i would think so.. initialize just says "device present and quiescient".  Open says I am using it.  So usually better to do visible changes during first open and last close10:13
agrierroxana: yes.  application may want to switch in/out of graphics mode10:13
roxanalDrJoel, agrier: thank you, I thought it might be a good idea -- getting close to the fix :)10:14
_Lucretia_DrJoel: applied that patch to the files inside newlib, recompiled b-gcc1, reinstalled, rebuilt rtems, fails exactly same point, same reason10:19
roxanalagrier: also the FB_SET(GET)PALETTE return 0, should I change them to get/set the palette?10:20
agrierlucretia: I recall running into the same issue...   let me look at my notes10:23
agrierroxana: if possible, yes.10:24
roxanalpossible :)10:25
*** sebhub has quit IRC10:25
agrierlucretia: first time I ran into it, I was missing patches for newlib10:25
agrierlucretia: I ended up punting and going back to newlib 1.17.0 + patches10:26
_Lucretia_right, will try an earlier version of newlib10:27
agrierlucretia: I never did figure out what was going on with newlib CVS head.10:27
_Lucretia_dunno10:34
roxanalDrJoel: sent the mail with the patch10:46
roxanalDrJoel: I also think that this API should use the framebuffer.h and that the functions names should be renamed instead of the header file that is used right now and that was causing the initial problems -- just for consistency reasons10:50
DrJoelok.. send me instructions with a patch to change the name and I will commit it.10:54
roxanalI will send you another pach with the renaming10:54
DrJoelroxanal: cutting and pasting the warnings10:59
DrJoelcommitted in just a second10:59
DrJoelcommitted10:59
agrierarm not building for me...  hash.c (from newlib) won't compile: newlib/libc/search/hash.h:46:2: error: #error BYTE_ORDER not defined by sys/param.h11:12
agrierechoes of previous builds...11:13
DrJoelthat's a new one to me.. sounds recent11:19
*** _Lucretia__ has joined #rtems11:21
agrierI might be building the wrong target..11:25
agrierarm-rtems4.10 ?11:25
*** madrazr has joined #rtems11:32
agrierendian.h is empty... that would explain it11:34
agrierI wonder how thatĀ happened...11:34
*** _Lucretia_ has quit IRC11:36
agriernuked it, and it seems to be continuing11:40
roxanalagrier: what target are you using?11:41
agrierarm-rtems4.1011:42
agrierit's generating one for every multilib11:42
roxanaland bsp?11:42
agrierhaven't gotten there yet.  :)11:45
agrierjust building toolchain11:45
agrier(you linux users have it easy)11:45
roxanal:D11:45
*** madrazr has quit IRC12:00
roxanalI will be out for lunch 30 minutes12:04
roxanalDrJoel: please remove the header file from arm/nds/fb/fb.h since I redefined the names of the functions it isn't needed anymore -- I forgot to add that in the mail12:51
roxanalI hope I didn't break anything ... it should be working12:52
DrJoeldid my correction in the email look ok to you?12:53
DrJoelif so, I will remove that .h file and move to commit12:53
roxanalDrJoel: yes everything looks fine13:02
roxanalDrJoel: the cast was needed because buffer is defined as void * in the rtems_libio_ioctl_args_t I have to change that in another place also :) thanks for pointing it out13:06
*** _Lucretia___ has joined #rtems13:20
*** kaluzamorph has quit IRC13:22
*** _Lucretia__ has quit IRC13:37
*** BBHoss has joined #rtems13:44
switnickDrJoel: hello13:52
tuxmaniacroxanal: sorry was away. no id dont work on nds/arm13:55
roxanaltuxmaniac: no problem13:55
DrJoelbrb on phone13:55
roxanalI figured it out13:56
DrJoelroxanal: ok.. should I commit the patch? and delete nds/fb/fb.h?13:57
roxanalDrJoel: YES13:57
DrJoelok.. doing a build with it removed .. then committing13:59
switnickDrJoel: Is it possible to build rtems with 128k of text and 4k of data?14:03
*** madrazr has joined #rtems14:05
DrJoelthat is too small.  Why?14:16
DrJoel128k code is ok.. 4k data is too small i think..14:16
switnickThe atmega128 only has 4k internal sram14:20
DrJoelbut can have more external RAM right?14:21
switnickit can14:21
switnickbut I dont have this not14:21
switnicknow*14:21
switnickthis is what is causing some problem when using the debugger14:21
switnickI think atleast14:22
DrJoelthat wouldn't match the simulator configuration.  The ram configuration can be low but not that low14:22
switnickok14:23
switnickalso I'm wondering how I should go about installing the raw ISR?14:24
switnickthere is very little in the documentation14:24
DrJoelthe avr-libc manual shows how to put an interrupt attribute on a method and you can see what registers it saves/restores that way.14:27
switnickok14:32
switnickI think I understand the theory of what this should do but not so sure on all the details.14:32
DrJoelI have an example (non-RTEMS) program which uses a timer.  I am sending it to you.14:33
switnickok14:33
switnickI am familiar with setting up an ISR14:33
DrJoelcool.14:34
switnicknot so sure on how to do it with RTEMS though14:34
DrJoelThis may not help much then but it is something14:34
switnickok thanks14:34
switnickare there any implementations of _CPU_ISR_install raw_handler that would be similar to the way I would do it for avr?14:35
DrJoelthis is a "simple vectored" CPU.  The usual way is to have a common _ISR_Handler and install pointer to it in the vector table.  Depending on the architecture, it may be able to tell which IRQ to process or you have to put a simple prefix in place (push one reg, load vector number into that reg, jump to common place)14:36
DrJoelI think the m68k is probably similar -- a table of addresses for IRQ entry.14:36
DrJoel _CPU_ISR_install raw_handler puts pointer to common handler there and and that does rest of work.14:37
switnickok14:40
switnickIf I was to look up simple vectored cpu ISR handler would I find information on the best ways to do this?14:41
roxanalI am getting a build error and I don't know how to get rid of it can anyone help14:52
roxanalNo rule to make target `libmisc/bspcmdline/bspcmdline.h', needed by `../cpukit/../../../pc386/lib/include/rtems/bspcmdline.h'. Stop.14:53
DrJoelswitnick: that's the way m68k does it -- no PIC -- CPU reads handler from table and jumps without any external hardware help.14:55
DrJoel680x0 is the easiest to look at for this14:55
DrJoelroxanal: sounds like a bootstrap.  It is a new file :-D14:55
roxanalDrJoel: I did the bootstrap14:55
DrJoeldo you have   cpukit/./libmisc/bspcmdline/bspcmdline.h14:56
DrJoelis it listed in cpukit/preinstall.am?14:56
roxanalDrJoel: yes14:57
DrJoelhmmm.. have you built from scratch recently?14:58
DrJoeldo you build with --enable-maintainer-mode ?14:58
roxanali don't have --enable-maintainer-mode14:59
DrJoelyou might need to build from scratch and add that to your next configure.  It helps track changes in these files and regen erate automaticxally14:59
roxanalbuild from scratch meaning.... ?15:00
DrJoelremove the build directory, configure, make15:00
roxanalI build from scratch every time :)15:00
DrJoelenglish expression to make from nothing.. don't know where is came from .. cake from scratch is homemade without Betty Crocker box :)15:00
agrierm68k IRQ is also vectored15:01
roxanalI didn't knew if you were refering to delete the rtems sources get new ones from the server and then build :D15:01
DrJoelyes .. we call it "simple vectored" as opposed to powerpc which is "PIC vectored"15:01
DrJoelno more ideas.. heading home to take kids to get passport renewed :-D15:02
*** DrJoel has quit IRC15:02
* agrier is reminded he should do that as well15:05
roxanalagrier: usually I use cvs up to update, but it seems is not copying the libmisc/bspcmdline/ folder-- it is not in my sources15:05
_Lucretia___ok, managed to rebuild with newlib-1.17.0, but I'm now getting this error: http://pastebin.com/m1b938a2b15:05
roxanalagrier: any ideeas?15:05
*** _Lucretia___ is now known as _Lucretia_15:05
_Lucretia_i've built b-gcc1, g-rtems and this is failing on b-gcc215:06
_Lucretia_including Ada15:06
_Lucretia_into the build15:06
agrierroxanal: cvs update -d15:06
roxanalagrier: tks15:07
agrierwith ada being such a pain, I'm surprised it ever caught on15:08
_Lucretia_Anyone know where this comes from and why it can stop the compile? pragma Style_Checks ("M32766");15:15
tuxmaniacearthquakle here15:15
agrierIndia?15:21
tuxmaniacyes15:21
_Lucretia_ok15:21
_Lucretia_go get Ada, I build rtems --disable-ada, then gcc2 with ada followed by another build of rtems --enable-ada, right?15:22
* agrier wishes he knew enough to help15:22
*** BBHoss has quit IRC15:24
* _Lucretia_ wishes too15:24
roxanalagrier: I will leave in 10-15 min -- I should have internet tonight until 5 PM your hour , but i will work offline -- if not I will talk to you tomorrow15:30
*** roxanal has quit IRC15:31
agriertux, you OK over there?15:36
*** kaluzamorph has joined #rtems15:36
_Lucretia_well, the file doesn't look legal to me, but then it's like it's got a problem anyway15:37
tuxmaniacagrier: yep15:44
agriertux: it was a big one, 7.616:13
*** cocan has joined #rtems16:28
*** cocan_ has quit IRC16:30
*** chrisjohns has joined #rtems16:35
*** cocan has quit IRC17:02
*** cocan has joined #rtems17:02
*** cocan has quit IRC17:06
*** cocan has joined #rtems17:23
*** dr__house has joined #rtems17:25
*** madrazr has quit IRC18:09
*** cocan_ has joined #rtems18:10
*** cocan has quit IRC18:10
*** cocan_ is now known as cocan18:10
*** switnick has quit IRC18:14
*** dr__house has quit IRC18:46
* kiwichris is back (gone 68:37:14)20:12
*** cocan_ has joined #rtems21:53
*** cocan has quit IRC21:53
*** cocan_ is now known as cocan21:53
*** cocan_ has joined #rtems00:45
*** cocan has quit IRC00:46
*** cocan_ is now known as cocan00:46
tuxmaniacheya gang01:15
kiwichrishi01:15
tuxmaniachi kiwichris . good evening01:15
tuxmaniacin the linkcmds file where we define the memory sections01:34
tuxmaniacMEMORY01:34
tuxmaniac  {01:34
tuxmaniac        RAM       : ORIGIN = 0, LENGTH = 16M01:34
tuxmaniac        PSIM_REGS : ORIGIN = 0x0c000000, LENGTH = 16M01:34
tuxmaniac        VECTORS   : ORIGIN = 0xFFF00000, LENGTH = 0x2000001:34
tuxmaniac  }01:34
tuxmaniachere if i set rAM = 16 M and PSIM_REGS = 16M01:34
tuxmaniacit means i can address any memory location in the 16M of RAM right01:34
kiwichrisAll this does is provide some error checking.01:34
kiwichrisWhen linking. I suspect you will need to edit a psim config file.01:35
tuxmaniackiwichris: yeah there is a device tree01:35
tuxmaniackiwichris: but I am trying to make psim see 32 MB01:35
kiwichrisIs RAM a device ?01:35
tuxmaniacbut the bsp files just 16Mb01:35
tuxmaniacto generate a dsi exception when i access an inbetween location01:35
kiwichrisIf there some other setting in the BSP ?01:36
kiwichrisI do not know this BSP well.01:36
tuxmaniacno. i dont think so. i have a device tree which I have configured01:36
tuxmaniacye ram is a device for psim01:36
tuxmaniacat least from the configs I did01:37
tuxmaniachmm let me try01:37
kiwichrisYou will need to change the linkcmd file if you have more static (.data + .bss) data that 16M01:37
tuxmaniachmm any idea whats these lines in linkcmds01:38
tuxmaniacRamBase = DEFINED(RamBase) ? RamBase : 0x0;01:38
tuxmaniacRamSize = DEFINED(RamSize) ? RamSize : 8M;01:38
tuxmaniacwhich one does the bsp take. is itthe one in the MEMORY thing above or the above two lines whihc has Ramsize01:39
*** kaluzamorph has quit IRC01:39
kiwichrisIf you define RamSize to be something it will take that value else RAM will be 8M01:39
kiwichrisNo idea. Check the symbols with nm or grep the source.01:39
tuxmaniacok. doing it01:40
tuxmaniacthanks01:40
* kiwichris is away: I'm away but logging01:48
*** chrisjohns has quit IRC01:49
tuxmaniacsigh. ishould have rtfm before asking here. nice write up on linkcmds in there on rtems docs01:49
lee__morning02:00
*** sebhub has joined #rtems02:00
lee__sebhub, morning02:37
*** lee2 has joined #rtems03:09
lee__hi all03:44
lee__have a simple problem...03:44
lee__trying to generate ipc key in rtems...03:44
lee__any help on this?03:45
*** _Lucretia__ has joined #rtems04:05
*** _Lucretia_ has quit IRC04:18
*** _Lucretia__ is now known as _Lucretia_04:18
lee__hi... :(04:34
lee__does rtems support shm or mmap calls yet? or will it? if not what is an alternative?04:34
kiwichrislee__, you have a single address space so why do you need IPC ?04:41
lee__hi chirss04:42
lee__*chris04:42
lee__remember that app I ported ...04:43
lee__well it uses IPC.... :(04:43
lee__and I basically just commented out those bits... which isn't much04:43
lee__and was not needed ...04:43
lee__for example!!! little code snippet04:43
lee__if (!disable_shm_write) {04:44
lee__      FILE *fh;04:44
lee__      fh = fopen(file_name, "w");04:44
lee__      if (fh == NULL) {04:44
lee__         cm_msg(MERROR, "ss_shm_flush", "Cannot write to file %s, please check protection", file_name);04:44
lee__      } else {04:44
lee__         /* write shared memory to file */04:44
lee__         fwrite(adr, 1, size, fh);04:44
lee__         fclose(fh);04:44
lee__      }04:44
lee__   }04:44
lee__   return SS_SUCCESS;04:44
lee__im just curious04:45
lee__if you were wondering why im asking?04:45
kiwichrisI remember.04:46
*** dr__house has joined #rtems04:46
_Lucretia_anyone here (apart from joel) know anything about the Ada side of rtems?04:47
kiwichris_Lucretia_, sorry I do not.04:47
lee__kiwichris, nevermind... :P04:49
kiwichrislee__, does this mean all is ok ?04:49
lee__kiwichris, did u see that matt gitau didn't get cited04:50
lee__kiwichris, was ok before04:50
lee__kiwichris, I was just going through some includes and wondering why thats all04:50
kiwichrislee__, for the late tackle ?04:50
lee__kiwichris, yip...04:50
_Lucretia_ok04:51
lee__kiwichris, lol and now SARU was yesterday called before IRB for wearing an arm band ... lol...04:51
kiwichrislee__, should he be cited. he was late and sin binned. It was not that bad a hit04:51
kiwichrislee__, now that is really really bad.04:52
kiwichrislee__, you can hurt some one with one of those.04:52
lee__kiwichris, lol... you talking about the arm bands or git's tackle?04:53
kiwichrislee__, arm bands04:53
lee__kiwichris, hehe... yip they sure are deadly...04:55
lee__kiwichris,  i see carter will not be playing for nz against aus...04:55
lee__kiwichris, think he's very much missed in nz side04:55
kiwichrislee__, it cannot be good04:55
kiwichrislee__, what ever is wrong. I think the world cup is at the center of all this.04:56
kiwichrislee__, kiwi do not want to be favorites04:56
lee__kiwichris, hehe might be... dont wanna choke in nz, in front of all the fans... I really dont want that nz rugby coaching position now04:58
kiwichrislee__, yeah that is correct. Rather hostile at the moment04:59
lee__kiwichris, Henry really must have the toughest job ever in the world at the moment04:59
* kiwichris off for a while04:59
* kiwichris is away: I'm away but logging04:59
_Lucretia_gcc-svn needs a patch for Ada, I'm certain05:09
*** cocan has quit IRC05:34
*** cocan has joined #rtems05:35
lee__very sille question again!? but does anyone know a command in RTEMS similar to vxworks sysClkRateGet()?05:45
lee__*command = function05:46
lee__*sille = silly05:46
lee__soz... was a long weekend here for us ;)05:46
*** dr__house has quit IRC07:01
*** dr__house has joined #rtems07:03
*** dr__house has quit IRC07:06
*** dr__house has joined #rtems07:09
*** madrazr has joined #rtems07:12
lee__never mind found it07:13
lee__vxworks function is very cryptic :P07:13
lee__rtems much better... :P07:14
lee__rtems_clock_get_ticks_per_second(void)07:14
tuxmaniaclee__: mmap no. but it will soon support at a different level. one of the gsoc projects was intended to be that way. But none of the POSIX APIs served the pupose. Since RTEMS is a single process multi threaded system07:35
tuxmaniacthere was no need for IPC07:36
lee__tuxmaniac, thanx07:47
lee__<--- learning more and more of rtems each day :)07:48
lee__tuxmaniac, what should one use (rtems/rtems/sem.h OR posix/semaphore.h ) both seem to provide same functionality, albeit in a different format07:50
tuxmaniaclee__: i am also a new comer here doing my google project with RTEMS. But from the paths you gave me it looks like semaphore.h is used if you need the POSIX compliant one07:51
tuxmaniacmay be I am wrong and you might have to wait for the other folks to comment07:51
lee__hehe07:51
lee__yip... also thought posix/semaphore.h07:52
tuxmaniaclee__: in the configure script when you give --enable-posix semaphore.h cane be used. else I think you need to use sem.h07:52
tuxmaniacconfigure command I mean07:53
*** mwalle has quit IRC08:03
*** th_d has joined #rtems08:04
*** mwalle has joined #rtems08:05
mwallehi ;)08:06
lee__hi08:06
*** madrazr has left #rtems08:08
*** DrJoel has joined #rtems08:12
*** ChanServ sets mode: +o DrJoel08:12
DrJoeldr__house: the two coremutex cases should be reachable via classic api08:19
DrJoelthe mutexinit.c one is POSIX specific08:19
*** madrazr has joined #rtems08:21
DrJoelthe coremutex.inl:171 should be as easy as creating a priority ceiling Classic API semaphore (binary attribute) unlocked with priority ceiling equal to 1 and then doing a semaphore lock on it to acquire it08:21
DrJoelthen sleep, let another thread of priority=1 block attempting to acquire same semaphore.  Then release it.08:22
*** madrazr has quit IRC08:27
*** madrazr has joined #rtems08:27
_Lucretia_finalyl08:30
_Lucretia_DrJoel: http://pastebin.com/m1b938a2b <- lastest svn tree08:30
DrJoel_Lucretia_: finally it built? or that I showed up?08:30
_Lucretia_showed up08:30
DrJoelit is 8am here.  :)08:31
_Lucretia_it's failing on the generated files s-soccons or something08:31
_Lucretia_DrJoel: pah!08:31
_Lucretia_the file contains no package ... is ... end ...; just a pragma, dunno if that's legal08:32
DrJoelby any chance is there a message somewhere in the logs about not finding an include file?08:32
DrJoelI can't remember.. socket.h maybe?08:32
_Lucretia_lemme see08:32
*** madrazr has quit IRC08:33
_Lucretia_shit! I deleted it!08:33
*** madrazr has joined #rtems08:33
_Lucretia_hokay, won't take me long to get it back08:33
*** madrazr has quit IRC08:33
_Lucretia_don't go anywhere08:33
*** madrazr has joined #rtems08:33
_Lucretia_DrJoel: I take it that socket.h is for the rtems source not the native?08:34
DrJoelyes.  If it couldn't find it, it goes downhill from there08:34
_Lucretia_I did just see if it exists in the opt/rtems dir and it is there at least08:35
DrJoelthat's good.. now to the Ada build log.. maybe it didn't find it08:35
DrJoeldid you have something like this on your configure line?     CFLAGS_FOR_TARGET=-B${BSP_INSTALL}/${TARGET}/${BSP_BUILT}/lib/ \08:36
*** madrazr has quit IRC08:36
*** madrazr has joined #rtems08:37
lee2DrJoel, Hi, simple question does one use rtems/rtems/sem.h or posix/semaphore.h to get a similar functionality as vxworks semBLib.h ...08:37
_Lucretia_DrJoel: no I didn't08:37
*** madrazr has left #rtems08:38
DrJoellee2: if that's what I think it is, then you want rtems.h and rtems_semaphore_xxx as a "simple binary semaphore".  The .h files you mention should not be directly included by the application08:38
DrJoel_Lucretia_: then the Ada build isn't finding the bsp .h files.  Add that to the Ada configure and try that one step again08:38
lee2http://www.fys.uio.no/studier/kurs/fys4220/docs/vxworks/ref/semBLib.html#top08:38
lee2that is a description of what Im looking for ...08:38
lee2:)08:38
DrJoellee2: yep.  That's simple binary..08:39
_Lucretia_BSP_INSTALL is where I've installed rtems? I'm dumping everything to $HOME/opt/rtems, so I would assume that I should use CFLAGS_FOR_TARGET=-B$HOME/opt/rtems/arm-rtems4.10/edb7312/lib ?08:39
_Lucretia_DrJoel: will doo08:39
DrJoel_Lucretia_: that looks right08:39
lee2DrJoel, ta... :P08:39
*** madrazr has joined #rtems08:39
DrJoellee2: that variation was added by Eric Norum years ago to make it easier to port VxWorks applications LOL08:39
_Lucretia_cool08:40
lee2hehe08:40
lee2hint hint... its what im doing08:40
lee2:P08:40
DrJoelFWIW rtems_task_variable_* was also added for that reason08:40
*** switnick has joined #rtems08:45
*** th_d has quit IRC08:50
dr__houseDrJoel: sorry had been out to get some medication for my sore throat. So should the mutexinit.c case go on psxtests?08:56
DrJoelyes.. isn't mutexinit.c itself in posix/src?08:57
dr__houseDrJoel: yeah08:57
dr__houseDrJoel: so does that test go onto psxtests/psx05?08:58
DrJoeldr__house: if it is in score and reachable via classic api, that's the preferred way to test08:58
DrJoellooks reasonable since it isn't much more than attr init/create/destroy08:59
dr__houseDrJoel: ok, yeah.08:59
*** roxanal has joined #rtems09:00
*** cocan has quit IRC09:07
*** cocan has joined #rtems09:07
tuxmaniacHello DrJoel09:23
switnickDrJoel: Hello09:24
tuxmaniacswitnick: how are things coming along?09:25
switnickOk, some things are a little over my head.09:26
switnickhow about you?09:26
tuxmaniacswitnick: just looking over the pit I should say09:27
tuxmaniacpeeping out may be. yet to make it jump out completely :)09:28
switnickdo you think you will make it for the pencils down date?09:30
DrJoelswitnick: it is always that way for me on a new port.  the architecture specific issues are hard and strange09:30
switnickIt would be good if I can get some extra help for the next few days09:31
DrJoelremember we want you to stay active after the pencil down date.  We want you to stay in the community.  Work on something else if you want.. we will help you there too09:31
switnickI would like to get interrupts working before09:31
DrJoelswitnick: trying ..  does cvs match your code09:31
switnickI made some drastic changes to cpu init09:32
switnickI think that is why the patch was failing09:32
DrJoeldoes it work? how many tests run?09:32
switnickI can get ticker running09:32
DrJoelyou can always just send me the files09:32
DrJoelhello?09:32
DrJoelanything else?09:32
switnickhello09:32
switnickyes09:32
tuxmaniacDrJoel: we (thomas and me ) put out a next steps which seems like getting interesting. like dynamic changes to memory protection values and stuff.09:32
DrJoelavr is a very small memory target. I expect we will find that a number of tests are simply too large to fit09:33
DrJoelhave you tried any of the sptests?09:33
switnickyes dont get any output09:33
DrJoeltuxmaniac: cool..09:33
switnickI think there is still a problem09:33
DrJoelhmmm.. even from sp01?  that's the same as ticker09:33
switnicktried it yesterday and didnt haca any luck09:34
DrJoelthat's weird.09:34
DrJoelcatch cvs up and I will look at that09:34
tuxmaniacDrJoel: also we started talks on a GSoC report. like detailed description of project, contribution, status, next scope etc. we will discuss the topics to include in the report / blog post in the coming couple of days09:35
* DrJoel brought his multimeter in to OAR today. working on discrete IO and analog out drivers09:35
switnickOK09:35
DrJoeltuxmaniac: yes .. everyone needs to do something like that. And update the wiki09:35
tuxmaniacand oh I miht be meeting you this time around in october . Planning to hit Munich for the Oktoberfest sometime around oct 2-409:36
DrJoelwe turned in my 3 sons passports for renewal yesterday (OMG the paperwork nightmare!!!) .. I hope this class happens .. Munich is lovely in October and it will be nice to meet you finally09:37
switnickDrJoel: My wired internet is still down, will go home soon and use the wire there.09:38
DrJoelone sons picture was wrong.. his head wasn't the right size in the picture.  same son forgot his driver's license and in spite of what state dept said, they didn't need birth certificate... <sigh> very expensive process to get or renew passports09:38
tuxmaniacDrJoel: haha Schengen VISA photo dimesnion FTW09:39
DrJoeltwo of the boys photos from same place, different day were ok.  luckily we just hadn't liked the one pic of that son from the "good day" and it was a retake.  So we had a picture we didn't like and used that09:40
tuxmaniac:)09:41
DrJoelgovernment paperwork and pain is the same world wide I think.  Instructions stink.09:41
*** switnick has quit IRC09:42
*** cocan has quit IRC09:42
dr__houseDrJoel: surprisingly enough, I didn't face a lot of problems for my passport09:42
tuxmaniacDrJoel: and oh. there are enough bugs in the implementation now. We also discussed about the enabling/disabling mmutests for targets you rised a few days back today.09:43
tuxmaniacWe decided to merge the source only after we find a proper way to do this because we felt its not good to have tests that doesnt run on other targets09:43
DrJoelgood idea but we don't need to let it bitrot.  If this requires BSP specific support, then you could consider adding a macro to bsp.h which indicates the MMU capabilities are supported and the test can use that to turn itself on and off.09:46
DrJoelLook at pppd and "SMALL". :)09:46
tuxmaniachmm will check09:47
*** switnick has joined #rtems10:02
switnickDrJoel: sent a diff for avr dir10:05
* dr__house heads for dinner, be back in a little while10:11
*** dr__house has quit IRC10:11
*** madrazr has left #rtems10:14
tuxmaniacDrJoel: still the oarcorp site has rendering problems in FF. In Opera and IE it seems fine10:23
DrJoelok.. i will pass that along.. i am only responsible for rtems.org/com10:25
tuxmaniacDrJoel: this site is the one referenced in embedded brains - RTEMS Class ite10:25
tuxmaniacsite*10:25
DrJoelany particular FF version or all?10:26
tuxmaniacatleast the one in whuch I am using . 3.0.11 under Fedora 1010:26
*** sebhub has quit IRC10:26
*** cocan has joined #rtems10:46
_Lucretia_hokay! gsocket.h:190:24: fatal error: sys/socket.h: No such file or directory10:49
_Lucretia_used: ../../gcc-svn/configure --enable-threads=rtems  --with-gnu-as --disable-multilib --with-gnu-ld --with-newlib  --verbose --with-system-zlib --disable-nls --enable-version-specific-runtime-libs -enable-languages=c,ada --target=arm-rtems4.10 --prefix=$HOME/opt/rtems CFLAGS_FOR_TARGET=-B$HOME/opt/rtems/arm-rtems4.10/edb7312/lib 2>&1|tee log.config.txt10:49
_Lucretia_and it's in include/sys alright10:50
_Lucretia_so, unless that CFLAGS_ part needs to go before the ../*/configure part10:50
DrJoel_Lucretia_: check the configure command in the do_one script.. it is right there10:51
_Lucretia_i am10:51
_Lucretia_I don't have these: --enable-newlib-mb --enable-newlib-iconv10:53
DrJoelis sys/socket.h $HOME/opt/rtems/arm-rtems4.10/edb7312/lib/sys/socket.h?11:02
DrJoelis that -B option on the command line that can't find socket.h?11:03
*** dr__house has joined #rtems11:04
_Lucretia_yeah11:06
_Lucretia_and I think it's failing because I missed the / off the end of lib11:06
_Lucretia_so it might not be constructing the path properly11:06
DrJoelthat is needed on a -B option to gcc.. strange though it sounds11:20
_Lucretia_same error11:28
_Lucretia_the final configure line is this: ../../gcc-svn/configure --enable-threads=rtems  --with-gnu-as --disable-multilib \11:28
_Lucretia_  --enable-newlib-mb --enable-newlib-iconv \11:29
_Lucretia_  --with-gnu-ld --with-newlib  --verbose --with-system-zlib --disable-nls \11:29
_Lucretia_  CFLAGS_FOR_TARGET=-B$HOME/opt/rtems/arm-rtems4.10/edb7312/lib/ \11:29
_Lucretia_  --enable-version-specific-runtime-libs \11:29
_Lucretia_  --enable-languages=c,ada --target=arm-rtems4.10 --prefix=$HOME/opt/rtems 2>&1|tee log.config.txt11:29
_Lucretia_which is identical to yours11:29
_Lucretia_in the do_one11:29
DrJoelhopefully that works :-D11:30
_Lucretia_no, it doesnt11:31
_Lucretia_In file included from s-oscons-tmplt.c:106:0:11:31
_Lucretia_gsocket.h:190:24: fatal error: sys/socket.h: No such file or directory11:31
_Lucretia_here is the preceding log: http://pastebin.com/m351e5e5011:35
_Lucretia_can you see if there is something wrong with that?11:35
*** agrier has quit IRC11:38
DrJoelwhere is the command that tried to compile s-oscons-tmplt.c?  This is mostly configure output11:39
_Lucretia_http://pastebin.com/mfa20ac111:40
_Lucretia_line 1211:40
_Lucretia_ish11:40
*** dr__house has quit IRC11:43
DrJoelthat line does not have the edb7312 directory in it if I am reading it correctly11:51
_Lucretia_yer not wrong11:52
_Lucretia_line 7 has this: -B/home/laguest/opt/rtems/arm-rtems4.10/lib/11:52
_Lucretia_so does line 1411:53
_Lucretia_and 1511:53
_Lucretia_but of them add on the include/sys part11:54
_Lucretia_all the -isystem parts point to home/laguest/opt/rtems/arm-rtems4.10/include and home/laguest/opt/rtems/arm-rtems4.10/sys-include11:54
DrJoelbut none of them pick up the installed arm/edb7312 BSP11:54
DrJoeleach directory in that list points to something but not the bsp11:54
_Lucretia_and /home/laguest/opt/rtems/arm-rtems4.10/sys-include doiesnt exist11:55
_Lucretia_hmmm, shouldn't it be -b rather than -B?11:57
_Lucretia_ah, no. found it in the manual, finally11:58
DrJoelthe sys-include not existing is no big deal.  That's where you drop precompiled things12:01
_Lucretia_ok12:01
_Lucretia_are you sure that should be CFLAGS_FOR_TARGET=-B$HOME/opt/rtems/arm-rtems4.10 instead?12:02
DrJoelgcc assumes it is a "system directory".  libraries in that directory.  Also will add "include" to the end of that and use it as an include directory.  That's how the BSPs are installed.12:06
_Lucretia_can you select the bsp with -march or -mtune or -m or whatever flag they use now?12:08
DrJoelno.. it isn't installed like that.  i would like to see us add a -mrtemsbsp=XXX to handle the standard way of using a bsp but it has never made the top of the list and no one has ever paid for it.. so just an idea :(12:14
_Lucretia_well, i'm searching through the log and the bsp is used in the CFLAGS in various places12:14
DrJoelyep.. but this is just getting the tools to build12:15
DrJoelyou are on the svn head right?12:15
DrJoelsometimes passing CFLAGS_FOR_TARGET breaks in the Ada.12:15
_Lucretia_yeah, everything is head12:17
_Lucretia_http://pastebin.com/m47a0484f <- line 2 includes the bsp, might be a bit more context for you to look at12:18
DrJoelthe word edb7312 is simply not there12:29
_Lucretia_I've just searched for it, and it's there12:30
_Lucretia_on that pastebin12:30
_Lucretia_line 112:30
_Lucretia_lin1212:30
_Lucretia_balls, 212:30
_Lucretia_"CFLAGS=-B/home/laguest/opt/rtems/arm-rtems4.10/edb7312/lib/ -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes" ada/s-oscons.ads12:30
_Lucretia_at the end12:30
_Lucretia_well, ner the end of the line12:30
DrJoeldid you install rtems multilib or just for edb7312?  there are two rtems builds in the script12:33
DrJoelthe multilib one is where the sys/socket.h is getting found.12:33
DrJoelIt doesn't look like CFLAGS_FOR_TARGET is propagating down.. email myself and laurent@guerby.net with your configure command, bzipped configure and build log, and version info and explanation. My build procedure is covering up the issue that CFLAGS_FOR_TARGET is not propagating down12:34
_Lucretia_for b-gcc1, b-rtems and b-gcc2, they all have --disable-multilib set on the configure cmdline12:40
DrJoelnot in my my script :) and that's why it works for me.12:41
_Lucretia_ok, you want my configure script?12:41
DrJoelSomething has broken the ada build for passing CFLAGS_FOR_TARGET down ..12:41
_Lucretia_the gcc/configure?12:41
DrJoelno.. email laurent and myself with the logs, explanation, etc.. he will know how to fix it .. he has fixed this multiple times in the past as I recall12:42
DrJoelyes.. gcc condfigure12:42
_Lucretia_the oarcorp one?12:43
*** cocan has quit IRC12:43
_Lucretia_ok, sent12:51
_Lucretia_not sure that's all the info yo need12:52
_Lucretia_bbiab12:54
*** cocan has joined #rtems13:00
tuxmaniacok heading to bed early tonight. dont want to have another earful :) later folks13:34
DrJoel_Lucretia_: i hope Laurent can help.  FYI he is in France but often answers email late in the evening13:37
DrJoel_Lucretia_: 4.4.1 failed with the same bug I have reported on 4.5 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=4077513:58
DrJoeltrying a 4.3.4 build now :(13:58
_Lucretia_hmmm, that link is a different error, no?14:28
DrJoeli'm getting further than you because my procedure includes multilib rtems and the .h files are in places gcc finds them by defualty14:32
DrJoeldefault14:32
DrJoel_Lucretia_: native built for 4.3.4, leaving arm going14:54
_Lucretia_k15:01
*** cocan has quit IRC15:02
*** cocan has joined #rtems15:16
*** switnick has quit IRC15:51
*** switnick has joined #rtems15:52
*** cocan_ has joined #rtems15:54
*** cocan has quit IRC15:56
*** cocan_ is now known as cocan15:56
*** cocan has quit IRC16:06
*** DrJoel has quit IRC16:09
*** roxana has joined #rtems16:48
*** roxanal has quit IRC16:50
*** roxana is now known as roxanal17:15
*** switnick_ has joined #rtems18:19
*** switnick_ has quit IRC18:34
*** switnick has left #rtems18:54
*** chrisjohns has joined #rtems20:35
*** roxanal has quit IRC20:46
*** chrisjohns__ has joined #rtems21:06
*** chrisjohns has quit IRC21:16
*** chrisjohns__ has quit IRC21:30
*** agrier has joined #rtems00:44
*** dr__house has joined #rtems01:09
*** sebhub has joined #rtems01:52
*** cocan has joined #rtems02:27
*** dr__house has quit IRC02:41
*** cocan has quit IRC02:51
*** cocan has joined #rtems03:09
*** dr__house has joined #rtems03:13
lee2morning all03:14
lee2is there a rtems task example?03:14
kiwichrislee2, testsuites/samples !03:26
lee2ah03:30
lee2nevermind found it03:30
lee2hehe03:30
lee2ta for the help03:30
kiwichrisno probs03:30
lee2ah since you are here03:31
lee21 more question03:31
kiwichrissoot03:31
kiwichrisshoot03:31
lee2in vxworks one has this structure: VX_TASK_SPAWN03:31
lee2but I cant seem to find a similar one for rtems..03:32
lee2not a problem, but just wondering... would keep my could rather simple03:32
kiwichrisWhat does it do ?03:32
lee2 ts = (VX_TASK_SPAWN *) param;03:32
lee2   status =03:32
lee2       taskSpawn(ts->name, ts->priority, ts->options, ts->stackSize,03:32
lee2                 (FUNCPTR) thread_func, ts->arg1, ts->arg2, ts->arg3,03:32
lee2                 ts->arg4, ts->arg5, ts->arg6, ts->arg7, ts->arg8, ts->arg9, ts->arg10);03:32
lee2basically consist of all the 'variables' for creating a task...03:33
kiwichrisOh. I see why RTEMS does not have this.03:33
kiwichrisDo you mean arguments ?03:34
kiwichrisLiek argc/argv ?03:34
lee2not sure I follow what you mean?03:34
lee2but param is a void*03:35
lee2structure03:35
lee2like I said its not an issue, but it would just make the code easier to maintain thats all03:35
lee2:)03:35
kiwichrisI wonder why vxWorks decided on only 10 arguments ?03:36
kiwichrisThe need for more than one is silly.03:36
kiwichrisYou either have many many like argc/argv or you have 1. The number 10 is strange.03:36
lee2i agree... hence the question03:37
lee2:P03:37
kiwichrisWith 1 you just pass the "param" and then cast it in the task.03:37
kiwichrisand save all that stack space you cannot recover.03:37
lee2brb... bathroom call :(03:38
*** dr__house has quit IRC03:44
*** lainwir3d has joined #rtems03:45
*** madrazr has joined #rtems04:14
*** cocan has quit IRC04:31
*** cocan_ has joined #rtems04:32
*** cocan_ is now known as cocan04:32
*** cocan has quit IRC04:59
*** cocan has joined #rtems04:59
*** dr__house has joined #rtems05:03
*** madrazr1 has joined #rtems05:08
*** madrazr has quit IRC05:09
*** cocan has quit IRC05:20
*** madrazr1 is now known as madrazr05:21
*** cocan has joined #rtems05:21
*** dr__house has quit IRC05:25
*** chrisjohns has joined #rtems05:50
chrisjohnsHi Cocan05:55
*** dr__house has joined #rtems06:05
*** dr__house has quit IRC06:37
*** roxanal has joined #rtems06:43
*** dr__house has joined #rtems06:44
*** madrazr has quit IRC06:44
*** madrazr has joined #rtems06:44
*** cocan has quit IRC07:24
*** cocan has joined #rtems07:25
*** cocan has quit IRC07:29
*** cocan has joined #rtems07:43
*** chrisjohns has quit IRC08:30
*** DrJoel has joined #rtems08:34
*** ChanServ sets mode: +o DrJoel08:34
DrJoelkiwichris: are you still awake?08:34
_Lucretia_you get anything from laurent? I got nothing so far08:40
DrJoeli have nothing from him08:40
_Lucretia_ay08:40
_Lucretia_how did your build go?08:40
_Lucretia_on the earlier version?08:41
DrJoelit finished and there are about 50 ACATS failures.. they look like legitimate issues :(  I will be reporting them08:41
_Lucretia_o, not good08:47
*** madrazr has quit IRC08:58
*** madrazr has joined #rtems08:59
*** cocan has quit IRC09:06
dr__houseDrJoel: good morning :)09:09
DrJoeldr__house: morning09:09
DrJoelor evening09:09
dr__houseDrJoel: grr, I forgot a semicolon! :(09:09
DrJoelthat was the weirdest error message I have ever seen from a semi-colon!!!09:10
dr__houseDrJoel: heh, guess I should take a screenshot of it :P09:11
DrJoeland shame yourself forever LOL09:14
dr__houseDrJoel: yeah :'(09:15
DrJoelso from here it should be easy to add the missing test code.  Did you update so all that is needed now is a diff?09:17
dr__houseDrJoel: yeah updating now09:17
DrJoelthese shouldn't be too hard to finish out from here..09:18
dr__houseDrJoel: yeah09:18
*** lainwir3d has quit IRC09:20
*** cocan has joined #rtems09:22
*** madrazr has left #rtems09:29
*** switnick has joined #rtems09:49
*** madrazr has joined #rtems10:01
*** madrazr has quit IRC10:12
dr__houseDrJoel: updated the tree and the make gives me this error, http://paste.ubuntu.com/251985/10:24
DrJoelI redid all the coverage stuff in the Makefiles.  It is now driven by --enable-coverage.  I am hoping to get it merged soon.10:28
DrJoelYou will need to back off all of the coverage patch10:28
DrJoelThe simplest way to do that is "cvs up -Pd 2>&1 | grep -v ^cvs".  Then remove any files that are modified and not by you.  Then update again10:29
dr__houseDrJoel: ah ok, but those files didn't give me any conflicts during update10:29
dr__houseDrJoel: cool, will do10:29
DrJoelthis version touches all different places in the sources.  Adds cpukit/wrapup-coverage which is easier to add to. :)10:29
dr__houseDrJoel: cool. And coverage can be enabled with a --enable-coverage! Thats awesome news :D10:30
roxanalDrJoel: where in the tree should my framebuffer test be? I have it in my tree in testsuites/samples/fbtest10:32
DrJoeldr__house: I hope that's how it lands. :)10:37
DrJoelroxanal: samples seems good. How will it build on a BSP without a framebuffer?10:37
dr__houseDrJoel: doing bootstrap now10:37
DrJoelroxanal: or should this be an external example like network-demos?10:38
DrJoelroxanal: if you can know bsp has framebuffer, then just not build.. see "SMALL" in pppd sample10:38
dr__houseDrJoel: should I include --enable-coverage when running configure now?10:40
DrJoelno.. it isn't in the tree at all.  it is still an add-on patch in rtems-coverage.10:41
dr__houseDrJoel: oh ok10:42
dr__houseDrJoel: will the coverage work now that I have updated the tree?10:42
DrJoeleasier to change set of things tested.. rtems_coverage.o is built in wrapup-coverage using a Makefile.am similar to the one in wrapup for librtemscpu.a.  So uncomment one line and add another component.  I tried /dev/null and another libmisc directory overnight for grins10:42
DrJoelare you running on arm for local testing?10:43
dr__houseDrJoel: no, sparc10:43
dr__houseDrJoel: ouch, I think I realized my mistake. That shouldn't matter right now, I am not running coverage tests for sparc, so.10:44
dr__houseDrJoel: dinner time. Be back in a while10:44
roxanalDrJoel: Ok I will take into account the bsp problem10:45
DrJoelwith sis, there isn't any coverage support so just keep your tree clean10:46
dr__houseDrJoel: back, yeah got that :)11:03
dr__houseDrJoel: build went fine :)11:03
DrJoeldr__house: great!11:03
DrJoeli just managed to add libcsupport to the set of code being covered.  Running tests now.  It will be interesting to see how much of that is covered.11:04
dr__houseDrJoel: cool11:05
DrJoelthis phase establishes a base set of code and targets analyzed.  Over time that can expand.11:05
dr__houseDrJoel: hmm..11:05
DrJoelnot before the end of August.  You and I are focused on erc32.  We should review the others to ensure they didn't turn up a new case that is needed.  But that's it. :)11:06
*** madrazr has joined #rtems11:07
*** sebhub has quit IRC11:07
dr__houseDrJoel: hmm..11:09
DrJoeldr__house: when you add libcsupport, there are some BIG chunks11:36
DrJoel193 uncovered ranges found11:36
DrJoelBytes Analyzed          : 10674811:36
DrJoelBytes Not Executed      : 2052811:36
DrJoelPercentage Executed     : 0.807711:36
DrJoelPercentage Not Executed : 0.192311:36
dr__houseDrJoel: wow! Ok, that was overwhelming :P11:37
DrJoelThe 15 largest range from 336 to 2 over 2K.11:37
DrJoelA few new tests would knock that down quick.  There are a few areas that are used but don't have tests.  The code for /etc/passwd and /etc/group parsing are used by the shell and telnetd but none of it is covered.11:38
dr__houseok11:38
DrJoeltermios has a lot.  getcwd and a few of the stat variants are not hit11:38
DrJoelLays out the next steps11:39
dr__houseDrJoel: hmm.. ok11:39
switnickDrJoel: how do you get the $Id at the beginning of the files you create?12:08
DrJoelwhen it is committed to cvs and checked out the next time, the revision number is changed.  The string $Id$ is what the source code control system looks for12:09
DrJoelYou just put them in comments12:10
switnickok12:10
switnickI have included support for simulavr in gcc testing12:10
switnickI dont think it is complete12:11
switnickbut it will run for avr12:11
DrJoelswitnick: cool.  send it and I will merge it.12:15
switnickjust sent it12:16
DrJoelyou forgot to attach the new file avrtest.in :)12:20
DrJoelI need to work on the gdb script support for attaching to external simulators .. skyeye and simulavr and qemu all use the gdb server model12:20
switnicksorry will send it along12:22
DrJoelno problem.  Merged12:32
DrJoelthe bspcheck faults routine is for grepping the log for simularot or BSP exception messages that indicate the test has failed badly12:33
DrJoelThat's an FYI.. so if you see any message as you go along that debugging that fits that description, we add it12:33
switnickok12:34
switnickDrJoel: have you implemented the time limit for the simulator yet?12:41
DrJoel-m  <nanoseconds>     maximum run time of <nanoseconds>12:45
DrJoelsimulated nanoseconds12:45
*** cocan_ has joined #rtems13:18
*** cocan has quit IRC13:20
*** cocan_ is now known as cocan13:20
tuxmaniacdr__house: happy birthday. enjoy the day :)13:31
*** roxanal has quit IRC13:33
*** roxanal has joined #rtems13:33
DrJoeldr__house: happy birthday!13:35
dr__househey! thank you guys! :D13:55
dr__houseDrJoel: tuxmaniac:^^13:55
DrJoeldr__house: so what is the traditional birthday treat in your part of the world?14:01
dr__houseDrJoel: traditionally people get cakes. But I stopped getting cake since high school. Its just any other day for me, except that people wish me :)14:03
tuxmaniacDrJoel: we go to temple and pray. then its time for some lifting and kicking butts14:05
DrJoel:) cakes are the norm here also.  I have a thing for ice cream cakes.  One of my sons birthday is in late October so he usually gets a "graveyard".  It is a flat dish of chocolate pudding with cookies in it as headstones, pumpkins, etc .. halloween :)14:05
dr__houseDrJoel: yeah and then a grand lunch/dinner at home, with lots of sweets :)14:05
DrJoeltuxmaniac: so not much different than many other days? LOL14:05
tuxmaniachehe14:06
*** switnick has left #rtems15:00
*** switnick has joined #rtems15:35
*** cocan has quit IRC15:46
*** switnick has quit IRC16:41
kiwichrisDrJoel, are you around ?17:34
* kiwichris off to the office17:44
* kiwichris is away: I'm away but logging17:44
*** kiwichris has quit IRC17:48
*** madrazr has quit IRC17:57
dr__houseok all, leaving for the day. Have a good day everyone. Good day DrJoel :)18:20
*** dr__house has quit IRC18:20
*** kiwichris has joined #rtems19:50
*** kiwichris has quit IRC23:34
lee2morning all.... what a lovely day :P01:47
lee2snow fell last night on our mountains and now its bloody crispy, very sunny morning... :)01:47
*** madrazr has joined #rtems02:05
*** sebhub has joined #rtems02:09
lee2sebhub, morning02:46
*** tuxmaniac has quit IRC02:55
*** tuxmaniac has joined #rtems02:56
*** madrazr has quit IRC03:01
*** tuxmaniac is now known as adultling03:13
*** agrier has quit IRC03:14
*** mwalle has quit IRC03:14
*** _Lucretia_ has quit IRC03:14
*** adultling is now known as tuxmaniac03:14
*** rokka has quit IRC03:15
*** rokka has joined #rtems03:15
*** agrier has joined #rtems03:16
*** mwalle has joined #rtems03:16
*** _Lucretia_ has joined #rtems03:16
*** cocan has joined #rtems03:27
*** cocan_ has joined #rtems03:46
*** cocan has quit IRC03:46
*** cocan_ is now known as cocan03:46
*** dr__house has joined #rtems04:03
*** cocan has quit IRC04:06
*** cocan has joined #rtems04:07
*** madrazr has joined #rtems05:02
*** dr__house has quit IRC05:16
*** kiwichris has joined #rtems05:20
*** madrazr has quit IRC05:52
*** kiwichris has quit IRC06:31
*** kiwichris has joined #rtems06:31
*** cocan_ has joined #rtems06:42
*** cocan has quit IRC06:42
*** cocan_ is now known as cocan06:42
*** kiwichris has quit IRC07:32
* DrJoel apparently didn't kill the client overnight07:57
*** madrazr has joined #rtems07:59
*** cocan has quit IRC08:10
sebhubhi08:39
*** cocan has joined #rtems09:03
*** madrazr has left #rtems09:09
*** th_d has joined #rtems09:10
*** roxana has joined #rtems09:16
*** roxanal has quit IRC09:18
*** switnick has joined #rtems09:34
*** th_d has quit IRC09:39
*** roxana is now known as roxanal10:21
*** madrazr has joined #rtems10:51
*** madrazr has quit IRC11:27
*** _Lucretia__ has joined #rtems11:40
*** sebhub has quit IRC11:48
*** dr__house has joined #rtems11:51
dr__houseDrJoel: good morning :)11:51
DrJoelevening.. so you survived the birthday..11:51
dr__houseDrJoel: lol yeah :D11:52
DrJoelany luck on the patch?11:53
dr__houseDrJoel: not yet, had a few things to ask, I think I can send it over tonight. :)11:54
*** _Lucretia_ has quit IRC11:55
dr__houseDrJoel: what routine do I call to create a thread? I mean is there a classic API for that? or should pthread_create be called?11:58
DrJoelrtems_task_create.. see sp01/init.c for an example11:59
DrJoelor as I sometimes do.. cd sptests ; grep -r "XXX" . when I want to see which tests use something11:59
dr__houseDrJoel: ah ok, I did try that, but I searched for rtems_thread_create :P12:00
DrJoelthread==task.. api chooses name12:00
dr__houseDrJoel: oh ok, cool12:01
*** _Lucretia__ is now known as _Lucretia_12:45
*** cocan_ has joined #rtems13:10
*** cocan has quit IRC13:10
*** cocan_ is now known as cocan13:10
*** roxanal has quit IRC13:12
*** madrazr has joined #rtems13:13
*** cocan has quit IRC13:14
*** cocan has joined #rtems13:32
*** cocan has quit IRC14:26
*** cocan has joined #rtems14:27
*** switnick has quit IRC15:19
*** cocan has quit IRC15:41
DrJoeldr__house: any luck?15:41
*** switnick has joined #rtems15:42
*** cocan has joined #rtems15:43
*** switnick has quit IRC15:49
*** switnick has joined #rtems15:53
dr__houseDrJoel: sorry had been to have some tea. Things are improving :)15:57
*** switnick has quit IRC15:57
DrJoeldr__house: :)15:59
* DrJoel has had luck .. yesterday I saw interrupts from discrete inputs on an embedded PC I am doing some drivers for. Today I saw the analog outputs work. :)16:00
DrJoelfyi i won't be in tomorrow.. helping daughter move to dorm for junior year :D16:01
dr__houseDrJoel: Oh, ok. Cool :)16:03
DrJoelwith any luck, you will have a patch before I leave today and I can give you a new report.. hint16:04
dr__houseDrJoel: hopefully, yeah16:10
DrJoel:)16:10
dr__houseDrJoel: ok, time to pester you again. What should be the initial modes and attributes arguments for rtems_task_create? I have given it as RTEMS_DEFAULT_MODES, and RTEMS_DEFAULT_ATTRIBUTES. But the routine is failing with error code 5.16:13
DrJoelif you look in the manual (or cpukit/rtems/include/rtems/rtems/status.h) you will see it is16:19
DrJoel  RTEMS_TOO_MANY                 =  5,16:19
DrJoelwhich is a hint that you need to configure another task .. 2 instead of 1.16:20
DrJoel:)16:20
dr__houseDrJoel: ah! ok.16:25
dr__houseDrJoel: the description on status.h says that I have tried to attempted to create too many instances of the same object class, which means that I have tried to created one too many tasks right?17:03
dr__housewhen you say that I need to configure another task, it means I have to create another semaphore?17:04
dr__houseI am sorry if I am behaving extra ordinarily dumb today, but I desperately want to get the patch out today :(17:05
DrJoel#define CONFIGURE_MAXIMUM_TASKS             217:06
DrJoelthe tasks need to share the semaphore..17:06
dr__houseDrJoel: cool, got the point. LOL, I frame a 3 line question and the answers one line pasted from the code, irony :P17:07
DrJoel:) it is an advantage of living the code for years... :)17:08
DrJoelwhat's the legal drinking age in India?17:08
dr__houseDrJoel: I'll get there someday :)17:08
dr__houseDrJoel: Its 18 I guess17:08
DrJoelIt's 21 in the U.S. and if you all have noticed the output of ticker or sp01, check out the date17:08
DrJoelTA1  - rtems_clock_get_tod - 09:00:00   12/31/198817:09
DrJoelI am treating 12/31/2009 as RTEMS 21st birthday.. we don't know the official day development started but that is counting :)17:09
dr__houseDrJoel: yeah I have thats the date.17:09
dr__houseDrJoel: awesome! :D17:09
DrJoelI know some devel was done in 1988 (or design).  But that date was just New Year's Eve in the first year .. so a popular date for us to use in tests17:10
dr__houseDrJoel: lets just think that its today CDT, that way it gets to be 13th Aug! :D17:10
dr__houseDrJoel: RTEMS is almost as old as me! Man, I am so overwhelmed with that information. I am working on a project thats almost as old as me. I am so previleged to be doing my contribution to it!17:13
dr__houseDrJoel: ok, now getting an RTEMS_UNSATISFIED error17:18
DrJoeldr__house: how much stack space did you ask for on the create?17:28
DrJoelreduce it to minimum or ...17:29
DrJoel#define CONFIGURE_EXTRA_TASK_STACKS         (4 * RTEMS_MINIMUM_STACK_SIZE)17:29
DrJoelsomething like that from sp0117:29
dr__houseDrJoel: I had asked for twice the minimum, should I reduce it?17:29
DrJoeljust use minimum.. confdefs.h assumes stacks are minimum unless you ask for more memory17:29
dr__houseDrJoel: cool17:29
DrJoeldoes that make sense?  it assumes each task has a minimum but doesn't know you tried to create with more17:30
dr__houseDrJoel: ah ok, and when I want to create with more I do it by specifying  #define CONFIGURE_EXTRA_TASK_STACKS         (4 * RTEMS_MINIMUM_STACK_SIZE) . Right?17:31
DrJoelclose .. it is kind of weird.. you created 1 task and asked for 2 * minimum.17:32
DrJoelconfdefs.h assumed 1 * minimum.  So you are one short.17:32
DrJoel #define CONFIGURE_EXTRA_TASK_STACKS         (RTEMS_MINIMUM_STACK_SIZE)17:32
DrJoelsay you asked for 40K .. then it would be..17:33
DrJoel #define CONFIGURE_EXTRA_TASK_STACKS         (40K - RTEMS_MINIMUM_STACK_SIZE)17:33
dr__houseDrJoel: ah ok, got it. When I want to ask for more stack space, I do  #define CONFIGURE_EXTRA_TASK_STACKS         (XX- RTEMS_MINIMUM_STACK_SIZE) and then confdefs.h understands that I need that much more space allocated for the stack17:34
DrJoelright.. it is above minimum * number of tasks. :)17:35
DrJoelsimilarly confdefs.h has to be told about how much memory you will need for message buffers.  The size and max is up to you completely17:35
dr__houseDrJoel: ah ok. got it. Thats for message queues. :)17:36
DrJoelyes..17:36
DrJoelheading home.. i will probably log on later from home and will check email in the morning before I leave to take daughter to school.  If you have something you want run, just shoot it my way.17:37
DrJoelI can easily add one test and rerun.. probably < 1 minute to do it now when I have a built tree17:37
* DrJoel waves bye17:37
*** DrJoel has quit IRC17:37
dr__houseDrJoel: fine, bye17:37
*** _Lucretia_ has quit IRC17:59
*** _Lucretia_ has joined #rtems18:02
*** _Lucretia_ has quit IRC18:09
*** DrJoel has joined #rtems18:12
*** ChanServ sets mode: +o DrJoel18:12
* DrJoel is at home and will be walking by 18:12
DrJoeldr__house: if you need help, just ask18:12
*** cocan has quit IRC18:15
*** cocan has joined #rtems18:17
*** _Lucretia_ has joined #rtems18:42
dr__houseDrJoel: sure, was outside for some fresh air :)18:48
*** cocan has quit IRC18:51
*** madrazr has left #rtems18:52
*** cocan has joined #rtems18:52
dr__houseDrJoel: when I try to obtain the lock from the thread, it fails with RTEMS_INVALID_ID error.19:07
DrJoelDid you pass &Mutex_id to rtems_semaphore_create?  Did you create the semaphore before the task ran?19:12
dr__houseDrJoel: yeah, in fact the control passes on to the task routine19:13
DrJoelit is better to create objects threads use before threads to avoid race conditions on creation.. if that is the problem.19:14
DrJoelCould be a simple programming error also. :)19:14
dr__houseDrJoel: well, I am thinking the latter, considering my extraordinary programming skills :P19:15
DrJoelemail my gmail the program and I will look before i go back to the family19:15
DrJoelprogramming errors are easy to create and take experience to spot.  programs with threads present their own challenges :)19:16
dr__houseDrJoel: hmm..19:17
dr__houseDrJoel: Sent the mail. Will go to sleep now, its 0545hrs here. I so wanted to finish it today. Hopefully will get it done tomorrow.19:19
DrJoelgo to bed. :)  the email is here19:20
*** roxanal has joined #rtems19:20
dr__houseDrJoel: good day, catch you later :)19:21
DrJoeli just built it.. there is a warning19:21
dr__houseDrJoel: oh, what does it say?19:21
DrJoel../../../../../../../rtems/c/src/../../testsuites/sptests/sp65/init.c:57: warning: passing argument 2 of 'rtems_task_start' from incompatible pointer type19:21
DrJoel../../../../../sis/lib/include/rtems/rtems/tasks.h:387: note: expected 'rtems_task_entry' but argument is of type 'void * (*)(rtems_id *)'19:21
DrJoel../../../../../../../rtems/c/src/../../testsuites/sptests/sp65/init.c:57: warning: passing argument 3 of 'rtems_task_start' makes integer from pointer without a cast19:21
DrJoel../../../../../sis/lib/include/rtems/rtems/tasks.h:387: note: expected 'rtems_task_argument' but argument is of type 'rtems_id *'19:21
DrJoel../../../../../../../rtems/c/src/../../testsuites/sptests/sp65/init.c: In function 'Task_1':19:21
DrJoel../../../../../../../rtems/c/src/../../testsuites/sptests/sp65/init.c:78: warning: passing argument 1 of 'rtems_semaphore_obtain' makes integer from pointer without a cast19:21
DrJoelline 78 put a "*" in front of Mutex_id19:23
dr__houseDrJoel: ok19:23
dr__houseDrJoel: building now19:23
*** cocan has quit IRC19:24
DrJoeli missed something hold on19:24
dr__houseDrJoel: ok, its not showing the error. But the routine doesn't seem to wake up after I say sleep(1)19:24
*** cocan has joined #rtems19:24
DrJoel#define CONFIGURE_APPLICATION_DOES_NOT_NEED_CLOCK_DRIVER19:25
DrJoelCONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER19:25
dr__houseDrJoel: ok done, building19:26
dr__houseDrJoel: the test completed. Now to put a break point and check if I hit it.19:26
dr__houseDrJoel: looks like I haven't hit it19:30
DrJoelhold on... looking myself at case again19:30
dr__houseDrJoel: ok19:30
DrJoeldo you even get to an instruction or line above it?19:36
DrJoeloh.. go on to bed.. it is too late for you... i will email when i figure out something .. I am getting confused19:37
dr__houseDrJoel: heh, ok fine. Catch you later, bye. Have a good day :)19:38
dr__houseDrJoel: actually that applies to me too :P19:39
DrJoelI need to rebuild on my home laptop in "coverage mode" .. my executable wasn't build the same way19:39
dr__houseDrJoel: ah ok19:39
dr__housesee you then, bye19:39
*** dr__house has quit IRC19:39
*** cocan has quit IRC20:02
*** cocan has joined #rtems20:17
*** kiwichris has joined #rtems20:24
DrJoelfor those following along from home dr__house hit the case he was trying.  We are now down to ...20:30
DrJoel4 uncovered ranges found20:30
DrJoelBytes Analyzed          : 6988820:30
DrJoelBytes Not Executed      : 2820:30
DrJoelPercentage Executed     : 0.999620:30
DrJoelPercentage Not Executed : 0.000400620:30
DrJoel====== 15 Largest Range Sizes ======20:30
DrJoel      2  420:30
DrJoel      1  820:30
DrJoel      1  1220:30
DrJoel=====================================20:30
DrJoelon the sparc/erc32.  Really 3 cases.20:30
DrJoelkiwichris: hello?20:30
kiwichrisHi JOel20:30
DrJoeldid you see either of my emails today?20:31
DrJoeli would appreciate the feedback on the hw watchdog/reset stuff. I really want to get it together for some odd reason. :)20:31
kiwichrisYeap. Look at one and that all seems fine. I will sort out the detail and get back to you. Not at the wdog one yet. Looking at BDM issue. Also dog was going mad so had to go to the park :)20:32
DrJoelfyi i also can now do coverage reports on more code.. libcsupport, libblock, filesystems, libmisc, .. anything but shell and networking in cpukit20:32
kiwichrisWhy not the shell ?20:32
DrJoeljust haven't included it yet.  Figured testing it would be a different type of testing.  We can turn it on when we want.20:33
DrJoelSo far it doubles the size of the tested binary, and drops the coverage to ~80%20:33
kiwichrisThe filesystem one will be interesting. Creating test cases will be a challenge20:34
DrJoelyes.. filesystem will be..20:34
DrJoelwe are takign Jessica to college tomorrow and I am going to start on a termios test. :<20:34
kiwichrisThe one will be easier that the file system.20:34
DrJoelbut there are no tests for the /etc/passwd, /etc/groups support.  termios and that are the largest in my set right now20:34
kiwichrisHere is a comment from the RFS:20:35
kiwichris        /*20:35
kiwichris         * This should never happen. Here so Joel can remove once his coverage20:35
kiwichris         * testing gets to the file systems.20:35
kiwichris         */20:35
kiwichris20:35
kiwichris:D20:35
DrJoeli agree... I wonder if it makes sense to add a "fstest" category .. and have standard tests that can operate on a mounted filesystem of any type.. just link the test with a different configuration/init to run the same test on a different filesystem/root test point.20:36
DrJoelI like that comment.20:36
DrJoelI thought the coverage might help to keep you honest at this point and let you know what was untested20:36
kiwichrisThe issue with a file system test is testing the actual file systems limits. For example a fat12 file system only has 64 root dir entries and code exists to handle this20:36
DrJoeli learned my bspcmdline code has two uncovered paths and I didn't add any tests for stringto. :(20:37
kiwichrisHe he. Big brother of the code world :)20:37
DrJoelsounds like fatal errors (/dev/null has one :( )... but that fat12 is possible to hit.. a larger limit might be impossible in practive20:37
DrJoelpractive20:37
DrJoelpractice ... grrrr20:37
kiwichrisIf you a ramdisk it should be ok. The cache has a unit test which is in my ftp area20:38
kiwichrisThe ATA, best of luck !20:38
kiwichrisThe ATA driver, best of luck !20:38
DrJoelthat's a hardware test. :-D20:38
DrJoeli am only sticking to abstract layers20:39
kiwichrisSure.20:39
DrJoelTCP/IP is off limits for a LONG TIME20:39
kiwichrisThe RAM disk will allow file system to driver path20:39
kiwichrisYeap I agree.20:39
kiwichrisThe file system is one area I can help wtih advice.20:39
kiwichrisThe bdbuf code would be great20:40
DrJoelI just want to get to the point where new code must pass the "coverage" metric before we let the submitter slip away :)20:40
DrJoelcatching up on existing code is a challenge.  It has been almost two years since I started this quest to get here <sigh>20:41
kiwichrisThis is a great way to make sure we have a complete test suite.20:41
DrJoelI agree.. and being able to do it on sparc and arm with similar results is amazing..20:41
DrJoeli can't wait until we work with skyeye to get bfin and coldfire :)20:42
DrJoeli think qemu might be able to be augmented to do this also20:42
DrJoelits traces are pretty thorough so may be a baseline20:42
DrJoelhttp://libre.adacore.com/libre/tools/coverage/  hmmmm.... i wonder who is involved20:45
DrJoelENST Paris is listed as a partner.  I probably know one of the principals.  I will email them. :)20:46
kiwichrisInteresting project20:48
DrJoelI am emailing my contact.  I hope it pans out.  It would be nice to add x8620:48
kiwichrisYeah. Alos standard host rpm installs is important.20:49
* kiwichris look over wd code20:53
kiwichrisJoel, I think the whole package should be "reset" of which a wdog is one type.20:54
DrJoelok.  so it is a reset driver?20:54
DrJoelthat would add "immediate" as an option, right?20:54
kiwichrisYes it is a driver for aspects of reset hardware on targets.20:54
DrJoelok.. i will rename it all and shoot it back early next week20:55
kiwichrisIt may be ony able to cause a reset and have no watchdog eg a PC20:55
kiwichrisThere may also be more than wdog on a board. I have seem this.20:55
kiwichrisIn this case there is no detault.20:55
DrJoelok.. and i have now seen two different wd implementations on embedded PCs so far <sigh>20:55
DrJoelall will be registered but the first configured/found is /dev/reset (or whatever)20:56
kiwichrisI think you have a reset driver, then an ioctl to manage them.20:56
DrJoeljust like console, rtc, etc20:56
DrJoelok..20:56
kiwichrisYeap.20:56
DrJoelthen i will wrap helper on top of that when you and thomas are happy20:57
kiwichrisRTEMS_RESET_WD_HIT_ALL20:57
kiwichrisYeap.20:57
DrJoelWhat's that?20:57
kiwichrisAn ioctl to hit all wd's at once.20:57
kiwichrisThere should also be a means for a user to ask what was the cause of the last reset.20:58
DrJoeloh... makes sense.. come in through any minor and hit all of them20:58
* DrJoel has to run.. bed time around here.20:58
DrJoelnight20:58
kiwichrisYeap.20:58
kiwichrisDo we need minors for every wd ? Hmmm this is a good question. I am not sure. I will think about it.20:58
*** cocan has quit IRC21:05
DrJoelok.. please think..it won't impact the initial design in the least.  I was thinking of a primary and a set of secondaries.. could be on different periods ..21:58
DrJoelfyi i think i just figured out the difference in results between erc32 and leon2/3.  Next run may have them all equal.  :-D22:03
*** BBHoss has joined #rtems23:14
*** kiwichris has quit IRC23:29
*** BBHoss has quit IRC01:04
*** BBHoss has joined #rtems01:04
*** sebhub has joined #rtems01:52
sebhubgood morning01:54
*** _Lucretia__ has joined #rtems01:58
*** _Lucretia_ has quit IRC02:15
*** BBHoss has quit IRC02:41
*** madrazr has joined #rtems03:02
*** BBHoss has joined #rtems03:14
*** BBHoss has quit IRC03:28
*** _Lucretia___ has joined #rtems03:40
*** madrazr has left #rtems03:50
*** _Lucretia__ has quit IRC03:56
*** _Lucretia___ is now known as _Lucretia_03:59
lee2 /join #ubuntu05:30
*** lee2 has quit IRC05:32
*** lee2 has joined #rtems05:32
lee__anyone around05:33
*** madrazr has joined #rtems06:13
*** dr__house has joined #rtems06:15
DrJoellee2: don't run to ubuntu .. it  won't meet your real time needs ;)06:45
DrJoeldr__house: did you see the email? congratulations06:45
DrJoeldr__house: I didn't notice until everything was committed that you hit the other coremutex case.  Please add that one to sp65.doc and send me a patch.06:45
dr__houseDrJoel: yeah saw it. Thank you :)06:45
dr__houseDrJoel: you mean the coremutexsurrender.c too?06:46
DrJoelYep.  Look at the report.  It wasn't there.06:46
dr__houseDrJoel: awesome :D06:46
DrJoelI need to look at the gumstix report.  I think there may be some odd paths that are hit on sparc that aren't on it.  That would indicate a need for a few more easy tests.06:47
dr__houseDrJoel: great, cool :D06:48
DrJoelI think so.  leon2 is 3 ranges and 24 bytes. same for leon3.  erc32 is 4 for 28 bytes.  leon2/3 should hit 100%06:51
dr__houseDrJoel: cool.06:52
DrJoelsmdk2410 is 13 and 132 bytes.  So worth some analysis.  (probably needs nanoseconds since last tick added to BSP like gumstix and edb7312 to move up)06:53
DrJoelsomehow edb7312 is reporting 100%.06:53
DrJoelgumstix is 31 uncovered ranges found06:53
DrJoelBytes Analyzed          : 6489206:53
DrJoelBytes Not Executed      : 31606:53
DrJoelso sparc is looking good.. arm is close behind06:54
DrJoeland look what i found last night... http://libre.adacore.com/libre/tools/coverage/06:54
DrJoelI emailed a contact at the university involved.06:54
dr__houseDrJoel: are they competition ;)06:57
lee2DrJoel, hehe... not running to ubuntu... guy in our lab has ubuntu and asked me to help with network issues...06:57
DrJoelno.. just adding coverage to another simulator .. so great .. and it looks like they want to include branch taken/not taken which is great06:58
DrJoellee2: :)06:58
* DrJoel needs to log off.. moving daughter to school today.. things are hectic.06:58
lee2DrJoel, boss away, now I become 'mrfixit now'06:58
lee2:(06:58
DrJoeldr__house: work on the next cases.. I will be online tomorrow probably (gmail for sure)06:58
DrJoeldr__house: I will have phone and can obviously do gmail or gchat on a G1. ;)06:59
*** dr__house has quit IRC06:59
* DrJoel waves bye06:59
*** DrJoel has quit IRC06:59
lee2tata06:59
*** dr__house has joined #rtems06:59
tuxmaniaclee2: is it tata for bye in your place as well?07:10
lee2tuxmaniac, ta-ta as in bye-bye07:11
tuxmaniaclee2: yeah I know. its the same thing in India as well :)07:15
lee2hehe07:25
*** roxanal has quit IRC07:29
tuxmaniacsebhub: is Thomas busy? or may be he forgot the change in meeting time? Can you please verify. Sorry for troubling you07:43
*** madrazr has quit IRC07:44
*** madrazr has joined #rtems07:44
*** th_d has joined #rtems08:04
*** th_d has quit IRC08:10
*** roxanal has joined #rtems08:35
*** switnick has joined #rtems08:39
*** dr__house has quit IRC08:46
*** madrazr has quit IRC08:50
*** sebhub has quit IRC09:09
*** Cathrak has quit IRC09:21
*** cocan has joined #rtems09:37
*** Cathrak has joined #rtems09:48
*** cocan_ has joined #rtems10:58
*** cocan has quit IRC10:58
*** cocan_ is now known as cocan10:58
*** switnick has quit IRC12:22
*** madrazr has joined #rtems12:51
*** dr__house has joined #rtems12:54
*** switnick has joined #rtems13:09
*** _Lucretia__ has joined #rtems13:21
*** _Lucretia_ has quit IRC13:37
*** cocan has quit IRC13:39
*** cocan has joined #rtems13:40
*** cocan has quit IRC13:42
*** _Lucretia__ is now known as _Lucretia_13:46
* tuxmaniac wonders why everyone in he world is learning git now13:52
agriersourceforge is shifting to it?14:15
tuxmaniacagrier: hehe i dont think so14:15
*** kiwichris has joined #rtems17:31
*** madrazr has left #rtems17:39
*** dr__house has quit IRC17:45
*** madrazr has joined #rtems17:54
*** switnick has quit IRC17:55
*** madrazr has quit IRC18:34
*** _Lucretia__ has joined #rtems21:13
*** _Lucretia_ has quit IRC21:29
*** chrisjohns has joined #rtems21:55
*** kiwichris has quit IRC22:08
*** _Lucretia__ is now known as _Lucretia_02:38
*** madrazr has joined #rtems03:06
*** chrisjohns has quit IRC03:35
*** lainwir3d has joined #rtems04:27
*** madrazr has quit IRC06:40
*** Cathrak has quit IRC07:01
*** Cathrak has joined #rtems07:10
*** madrazr has joined #rtems08:51
*** DrJoel has joined #rtems10:07
*** ChanServ sets mode: +o DrJoel10:07
tuxmaniacDrJoel: received the email yesterday. the silly bug is fixed. Now more securing of code with checks and cleanups before starting off the next phase10:34
tuxmaniacand also the final GSoC documentation outline10:34
DrJoeli saw that while on the road yesterday.  Spent the day moving my daughter in to the dorm, buying cleaning and school supplies and running cords for her10:35
tuxmaniacDrJoel: lucky children. I had no one when I moved into my dorm last year. and no body visiting me during the weekends since I am far away from parents. SOmetimes makes me miss them. But thats about it :)10:36
DrJoelwe didn't help last year but I did the year before that.  Her brothers were not old enough to drive and we had to be here for them on a school day.  Luckily she is in a town only 2 hours drive away and there are other relatives there.  I can understand how you miss your family.  Being a LONG way away from your family is hard10:37
tuxmaniacyeah and seeing fellow friends apartment mates parents visiting their kids is the time when I miss them. rest of the time work and trave andgeneral stuff keeps me occupied10:39
tuxmaniac:)10:39
DrJoelyep.  Jessica only comes home a few times during the semester.10:48
*** madrazr has quit IRC10:55
* DrJoel wrote a test on the drive yesterday to exercise all termios tcsetattr.. all bauds, character sizes, parity and stop bit options. going to see how the coverage for termios improves in a bit10:56
*** lainwir3d has quit IRC11:58
*** lainwir3d has joined #rtems12:12
*** madrazr has joined #rtems12:25
*** madrazr1 has joined #rtems12:31
*** madrazr has quit IRC12:31
*** madrazr1 has quit IRC13:03
*** switnick has joined #rtems13:12
*** switnick has quit IRC13:19
*** lainwir3d has quit IRC17:33
*** Cathrak has quit IRC01:20
*** madrazr has joined #rtems03:06
*** madrazr has quit IRC03:37
*** lee21 has joined #rtems03:54
lee21afternoon everyone :)04:07
*** cocan has joined #rtems08:19
*** _Lucretia_ has quit IRC08:35
*** cocan has quit IRC08:48
*** roxanal has quit IRC08:53
* DrJoel waves hello09:25
*** lee23 has joined #rtems09:56
*** _Lucretia_ has joined #rtems09:59
*** lee21 has quit IRC10:05
*** madrazr has joined #rtems10:48
*** DrJoel has quit IRC11:01
*** lee23 has left #rtems12:19
*** lainwir3d has joined #rtems15:30
*** madrazr has left #rtems16:12
*** lainwir3d has quit IRC17:06
*** DrJoel has joined #rtems17:30
*** ChanServ sets mode: +o DrJoel17:30
*** DrJoel has quit IRC21:01
*** kiwichris has joined #rtems21:08
*** kiwichris has quit IRC23:55

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!