2014-08-11T00:08:37 good morning 2014-08-11T01:37:42 *** sebhub has quit IRC (Remote host closed the connection) 2014-08-11T01:45:33 *** kiwichris has quit IRC (Ping timeout: 272 seconds) 2014-08-11T01:57:05 *** monstr has joined #rtems 2014-08-11T02:15:52 *** sebhub has joined #rtems 2014-08-11T02:30:07 *** zoso has joined #rtems 2014-08-11T02:53:59 *** kiwichris has joined #rtems 2014-08-11T02:55:26 *** edwardk has quit IRC (Ping timeout: 250 seconds) 2014-08-11T02:58:07 *** sebhub has quit IRC (Remote host closed the connection) 2014-08-11T03:00:47 *** Hesham has joined #rtems 2014-08-11T03:03:39 *** edwardk has joined #rtems 2014-08-11T03:23:27 *** sebhub has joined #rtems 2014-08-11T04:14:22 *** mkhoory has quit IRC (Ping timeout: 245 seconds) 2014-08-11T04:37:47 *** sebhub has quit IRC (Remote host closed the connection) 2014-08-11T04:48:55 *** Hesham has quit IRC (Quit: Leaving.) 2014-08-11T05:41:59 *** Hesham has joined #rtems 2014-08-11T07:50:06 *** peerst has quit IRC (Ping timeout: 250 seconds) 2014-08-11T08:03:28 *** peerst has joined #rtems 2014-08-11T09:03:54 *** Hesham has quit IRC (Quit: Leaving.) 2014-08-11T10:19:23 *** monstr has quit IRC (Remote host closed the connection) 2014-08-11T10:27:00 *** gigetoo has quit IRC (Ping timeout: 260 seconds) 2014-08-11T10:42:11 *** Hesham has joined #rtems 2014-08-11T10:56:01 *** edwardk has quit IRC (Quit: Computer has gone to sleep.) 2014-08-11T10:58:17 *** zoso has quit IRC (Remote host closed the connection) 2014-08-11T11:01:46 *** edwardk has joined #rtems 2014-08-11T11:01:52 *** edwardk has quit IRC (Client Quit) 2014-08-11T11:18:00 *** edwardk has joined #rtems 2014-08-11T11:25:28 *** edwardk has quit IRC (Ping timeout: 250 seconds) 2014-08-11T11:25:45 *** edwardk has joined #rtems 2014-08-11T11:27:21 *** edwardk has quit IRC (Read error: Connection reset by peer) 2014-08-11T11:27:25 *** edwardk_ has joined #rtems 2014-08-11T11:31:58 *** edwardk_ has quit IRC (Ping timeout: 250 seconds) 2014-08-11T11:53:33 *** warp_ has joined #rtems 2014-08-11T11:54:02 Hi there! 2014-08-11T11:56:06 *** warp_ has quit IRC (Client Quit) 2014-08-11T12:11:50 *** edwardk has joined #rtems 2014-08-11T13:03:13 *** Hesham has quit IRC (Quit: Leaving.) 2014-08-11T13:31:57 *** edwardk has quit IRC (Ping timeout: 246 seconds) 2014-08-11T14:07:59 *** edwardk has joined #rtems 2014-08-11T14:47:14 *** gigetoo has joined #rtems 2014-08-11T15:41:34 *** edwardk has quit IRC (Quit: Computer has gone to sleep.) 2014-08-11T15:51:56 *** edwardk has joined #rtems 2014-08-11T16:45:42 *** edwardk has quit IRC (Ping timeout: 260 seconds) 2014-08-11T16:49:39 *** kiwichris has quit IRC (Ping timeout: 252 seconds) 2014-08-11T17:06:12 *** kiwichris has joined #rtems 2014-08-11T17:54:52 *** edwardk has joined #rtems 2014-08-11T19:12:37 *** mkhoory has joined #rtems 2014-08-11T23:51:02 *** Hesham has joined #rtems 2014-08-12T00:04:40 *** zoso has joined #rtems 2014-08-12T00:15:28 *** monstr has joined #rtems 2014-08-12T01:24:18 *** mkhoory has left #rtems ("Leaving") 2014-08-12T01:24:24 *** mkhoory has joined #rtems 2014-08-12T01:37:10 *** kiwichris has quit IRC () 2014-08-12T01:55:52 *** kiwichris has joined #rtems 2014-08-12T02:46:20 *** zoso has quit IRC (Ping timeout: 240 seconds) 2014-08-12T03:56:06 *** Hesham has quit IRC (Ping timeout: 246 seconds) 2014-08-12T04:08:22 *** mkhoory has quit IRC (Ping timeout: 240 seconds) 2014-08-12T04:11:21 *** Hesham has joined #rtems 2014-08-12T05:12:50 *** zoso has joined #rtems 2014-08-12T06:43:16 *** Hesham has quit IRC (Quit: Leaving.) 2014-08-12T06:44:05 *** Hesham has joined #rtems 2014-08-12T07:35:34 *** AoLaD has joined #rtems 2014-08-12T07:35:39 *** krzysiekm13 has joined #rtems 2014-08-12T07:48:06 *** Hesham is now known as hesham 2014-08-12T07:49:41 *** asuolen has joined #rtems 2014-08-12T07:50:07 *** JustJanek has joined #rtems 2014-08-12T07:51:16 *** Sched has joined #rtems 2014-08-12T07:56:45 *** ppisa has joined #rtems 2014-08-12T07:57:09 *** phipse has joined #rtems 2014-08-12T07:57:53 hi 2014-08-12T07:58:00 Hi 2014-08-12T07:58:20 hey 2014-08-12T07:58:23 Hi 2014-08-12T08:03:49 hi :) 2014-08-12T08:04:31 Hi 2014-08-12T08:06:05 hello 2014-08-12T08:08:37 *** gedare has joined #rtems 2014-08-12T08:08:37 *** ChanServ sets mode: +o gedare 2014-08-12T08:08:46 hi, sorry i'm late 2014-08-12T08:09:23 gedare: not too late 2014-08-12T08:09:30 let's get started shall we.. Sched would you like to begin? hi phipse 2014-08-12T08:10:07 Ok, last week I was still trying to make the upcall interrupt mechanism works 2014-08-12T08:10:50 hi 2014-08-12T08:10:54 but the program crash, I have described it in my email, I still can't locate the problem 2014-08-12T08:12:23 I just answered your last mail. Yesterday was soft pencils down, as we require you to write documentation, I suggest you do this first, and explain in the documentation, where you assume the failure occurrs and why. 2014-08-12T08:12:33 During the debugging, I also have update some code to fixed some bug, including to add more interrupt frame to support nest interrupt. 2014-08-12T08:12:46 OK. 2014-08-12T08:13:05 and to keep the documentation in a save place, I suggest you write a wikipage in the rtems wiki. So it doesn't get lost. 2014-08-12T08:13:15 Agreed. And also start to get any unsubmitted code into a state that it could be submitted, or at least archived. 2014-08-12T08:13:19 OK. 2014-08-12T08:14:10 ++ prepare a patch for pok and send it to the pok-devel mailinglist and explain what you did do, what works, and what does not work. 2014-08-12T08:14:44 yeah. I'm keen to see what Julien thinks about the vcpu model and whether it is viable to proceed. 2014-08-12T08:14:50 understand. 2014-08-12T08:14:50 the patch should be basicly a diff between your and their github repo. A pull request should be sufficient too, I think. 2014-08-12T08:15:19 should I attend patch in email 2014-08-12T08:15:20 ? 2014-08-12T08:15:24 phipse: regarding patches, at your convenience, we should discuss the x86 paravirt patch 2014-08-12T08:15:29 or pull request in github? 2014-08-12T08:15:44 sec, I will have a look if my pull request generated a email to the list 2014-08-12T08:16:06 OK, thank you. 2014-08-12T08:16:50 I don't think GSoC is a job, but an opportunity to start to contribute to open source project. I still want to work on it after GSoC untill it work. 2014-08-12T08:16:50 nope it didn't so the best would be to make the pull request and then send an email to pok-devel further explaining the request (including a link to the request) 2014-08-12T08:17:30 As i promise in proposal. 2014-08-12T08:17:44 phipse, understand. 2014-08-12T08:18:22 the patch & mail should generate some interest on the list, so you should get valuable input on it 2014-08-12T08:18:32 OK. 2014-08-12T08:18:40 gedare: the paravirt patch is on my todo list, didn't get to it, yet. 2014-08-12T08:18:54 Sure thing 2014-08-12T08:19:14 Ok thanks.. anything else Sched ? 2014-08-12T08:19:22 no. thank you. 2014-08-12T08:19:36 Sched: be sure to have done all this by next monday! 2014-08-12T08:19:43 OK. 2014-08-12T08:19:46 I got it. 2014-08-12T08:20:24 Thanks. JustJanek can you go now? 2014-08-12T08:20:45 yeah, sure 2014-08-12T08:21:04 Last week I have tried debugging the errors from the remaining arinc653 calls but I can't seem to break at rtems code within pok nor can I break at the actual arinc653 call itself. I have tried multiple arinc calls but even with the arinc calls that are working I can't step through the user-space code. 2014-08-12T08:21:40 Did you ask phipse? I know he and Sched had similar problems stepping through Pok 2014-08-12T08:21:43 I also looked into the things I want to upload to google as things I've done 2014-08-12T08:21:55 yeah, I send an email regarding this 2014-08-12T08:21:58 k 2014-08-12T08:22:26 yeah, we are working on this, I recall similar problems last year 2014-08-12T08:22:35 I wanted to talk to philipp about what I tried with pok_thread_init later 2014-08-12T08:22:44 but was it stepping through rtems code? 2014-08-12T08:22:51 or was it user-space code in general? 2014-08-12T08:23:23 good question, I think it was rtems code 2014-08-12T08:23:38 rtems is build with -g isn't it? 2014-08-12T08:24:27 I think so normally. 2014-08-12T08:24:30 I haven't specifically enabled it so unless it's enabled by default it isn't on 2014-08-12T08:24:54 and should rems be build with -g or the hello.exe? 2014-08-12T08:25:01 rtems* sorry, autocorrect 2014-08-12T08:25:15 concerning the final, I think the same applies to your project: Wikipage, patch/pull request + email to pok-devel, code scrubbing 2014-08-12T08:25:57 *** DrJoel has joined #rtems 2014-08-12T08:25:57 *** DrJoel has joined #rtems 2014-08-12T08:25:57 *** ChanServ sets mode: +o DrJoel 2014-08-12T08:26:02 no problem, I also intend to continue my work after GSoC so if that isn't a problem I hope we can get more arinc calls working 2014-08-12T08:26:02 did you try to load the rtems binary (hello.exe) into gdb? 2014-08-12T08:26:05 You want to build every step with "-g" flag 2014-08-12T08:26:22 the pc386 bsp will, I assume the virtualpok bsp does also, in its .cfg file. 2014-08-12T08:26:36 JustJanek: Have you prepared some code to demonstrate the procedures for using arinc653 calls from rtems-guest? 2014-08-12T08:26:55 like example code? 2014-08-12T08:26:57 so first you load pok.elf as symbol-file, then part1.elf, and when you reach rtems code, you load hello.exe? 2014-08-12T08:27:59 I haven't loaded hello.exe in, or tried to load it in 2014-08-12T08:28:40 might be worth a shot later, though I'm not sure at what address I should give when using add-symbol-file 2014-08-12T08:29:17 but gedare, do you mean prepared some example code? 2014-08-12T08:29:33 JustJanek: yes, some code for your submission both to Google and to RTEMS / Pok 2014-08-12T08:29:47 And some instructions to help us duplicate the results. :) 2014-08-12T08:30:06 Yes, as phipse said, we want the same things from you as we asked Sched. 2014-08-12T08:30:32 *** zoso has quit IRC (Remote host closed the connection) 2014-08-12T08:30:37 it's a bit spammy as it goes through at least one of each arinc653 subset, even ones that return errors thus far 2014-08-12T08:30:53 well, it'll be no problem 2014-08-12T08:31:05 There's a difference between "spammy" and verbose. :) 2014-08-12T08:31:10 *** jenniferA has joined #rtems 2014-08-12T08:31:13 :) 2014-08-12T08:31:16 if it is testing a lot of features, that's a good thing 2014-08-12T08:32:04 if it's not crashing on broken ones, even better ;) 2014-08-12T08:32:14 anything else JustJanek ?" 2014-08-12T08:32:21 it doesn't crash on any arinc 653 calls 2014-08-12T08:32:27 and no, that's it for now :) 2014-08-12T08:32:47 ok thanks 2014-08-12T08:32:54 AoLaD: Can you go next? Hi ppisa 2014-08-12T08:33:00 Hi gedare 2014-08-12T08:33:04 Hello 2014-08-12T08:33:22 We have tested console driver. It looks fine. I just need to push a commit with some printk (debuging) changes (clean up). Since then I am working on the Ethernet driver. Ppisa has found some work of a student during SoC 2010 for FreeBSD. He has implemented similar driver for DM335, DM355, DM365, DM64xx, DM81xx and it looks like our processor is a same as well. So I should transfer it to the RTEMS. 2014-08-12T08:33:22 Link to the SoC project: http://p4web.freebsd.org/depot/projects/soc2010/jceel_dma/sys/arm/davinci/?ac=8 2014-08-12T08:34:01 OK this week I want you to focus on cleaning up the code and getting the BSP port submitted with working clock/console. 2014-08-12T08:36:05 ok, so all the code should be acording to rtems coding standart and all functions documented? 2014-08-12T08:36:28 Preferably yes. 2014-08-12T08:36:47 For BSP code we are not always so rigid, but for GSoC students I prefer to have the rtems style and good documentation 2014-08-12T08:37:09 If you prepare a single commit on github, I can review it before you submit to rtems-devel 2014-08-12T08:37:10 ok, I'll add some nice comments 2014-08-12T08:37:25 ok, thank you 2014-08-12T08:37:38 I would remind you to be warning free but Pavel probably already does that. :) 2014-08-12T08:37:49 I would prefer to have that available by tomorrow so I can review tomorrow or Thursday, and you can do fixes and then make your submission 2014-08-12T08:38:51 that could be fine. I'll try to make till tomorow evening 2014-08-12T08:39:04 OK thanks AoLaD. 2014-08-12T08:39:22 thank you 2014-08-12T08:39:25 hi krzysiekm13 can you go now? 2014-08-12T08:39:31 yes 2014-08-12T08:39:49 so, during last week I continued work on symbol generation within covoar 2014-08-12T08:41:01 I am now able to run nm from covoar using functions from rtl-host repo, but i'm not sure if i do this in right way - i sent mail regarding this to rtems-devel, Chris and Joel today 2014-08-12T08:41:24 ok. did your pc386.in fix patch get committed? 2014-08-12T08:41:45 And do you have any other patches that still need to be submitted/committed? 2014-08-12T08:41:53 These things fall thruogh the cracks 2014-08-12T08:42:06 and it is important as the patch author to follow-up and harass committers :) 2014-08-12T08:42:28 ok I saw your email. You did not really explain how you are doing it right now though... 2014-08-12T08:42:36 so, up to now I have written interface between my coverage.py module of rtems-test and covoar 2014-08-12T08:43:01 Are covoar and rtl in the same repository? 2014-08-12T08:43:04 and I can run nm and get nm output to be then filtered (as this was done previously in shell script) 2014-08-12T08:43:06 no 2014-08-12T08:43:20 Where are they? 2014-08-12T08:43:31 covoar is currently in rtems-tools repo, but rtl-host is another repo under Chris's private branch 2014-08-12T08:43:35 ok 2014-08-12T08:43:59 I would first ask Chris about whether rtl-host should be moved into rtems-tools 2014-08-12T08:44:09 otherwise, I would add rtl-host as a git-submodule in covoar 2014-08-12T08:44:17 so that you can build rtl-host 2014-08-12T08:44:30 or even just refer to the source code directly 2014-08-12T08:44:33 *** Hesham1 has joined #rtems 2014-08-12T08:44:33 from the covoar build 2014-08-12T08:44:35 about patches: I think pc386 fix is commited, but I need to check it, I pushed my code to my github and i'm waiting for review before sending patch to mailing list 2014-08-12T08:44:56 Send another email about review, but mostly only I do code review on github 2014-08-12T08:45:01 everyone else waits until it shows up on mailing list 2014-08-12T08:45:08 So you are better off just sending your code to the mailing list. :) 2014-08-12T08:45:17 ok :) 2014-08-12T08:45:24 *** hesham has quit IRC (Ping timeout: 260 seconds) 2014-08-12T08:45:33 *** Hesham1 is now known as hesham 2014-08-12T08:45:39 Alright. Anything else? 2014-08-12T08:45:42 so i will prepare patch and send it to mailing list tomorrow 2014-08-12T08:45:43 When does SOCIS end? 2014-08-12T08:45:48 Awesome 2014-08-12T08:45:53 with the end of august 2014-08-12T08:45:56 Ok 2014-08-12T08:46:03 sounds good and you are making great progress, keep it up 2014-08-12T08:46:15 How much is left in bash? 2014-08-12T08:46:15 but i'm going also to continue this project after the end 2014-08-12T08:46:33 Yes, we just need to be clear about deadlines is all 2014-08-12T08:46:36 so that nothing gets missed 2014-08-12T08:47:02 next up, hesham 2014-08-12T08:47:08 well, coverage flow is migrated to python - i can run coverage analysis directly by passing --coverage option to rtems-test (actually for pc386 only :) ) 2014-08-12T08:47:13 git checkout master 2014-08-12T08:47:16 Hi gedare, DrJoel 2014-08-12T08:47:24 well that was the wrong window :) 2014-08-12T08:47:29 :) 2014-08-12T08:48:01 Last week, I was working on cleaning up the code, submit for review, and fix up any issues according to your comments. 2014-08-12T08:48:12 Hi hesham. Regarding the floating-point support, I'd leave the stubs in place to make it easier to implement 2014-08-12T08:48:44 I would recommend moving them to a dedicated file. 2014-08-12T08:48:45 Ok, I will have to retrieve the Context_initialize_fp() again 2014-08-12T08:48:55 All FP functions? 2014-08-12T08:49:12 you can submit the improved cpu_halt() function as a follow-on patch with the power mgmt stuff 2014-08-12T08:49:33 Yes. When it is implemented, it is likely save and restore will be .S and initialize can be whereever. static inline, macro, .S, etc. 2014-08-12T08:49:38 I have the patch ready to be sent now, but I have some questions 2014-08-12T08:49:46 DrJoel: It's implemented today 2014-08-12T08:49:57 Oh.. sorry.. 2014-08-12T08:50:03 I mean power management 2014-08-12T08:50:03 regarding the nesting, do add the check on whether to switch or not. I don't quite understand why the arch manual says nesting won't occur 2014-08-12T08:50:05 not FP 2014-08-12T08:50:05 Leave it with the switch code then 2014-08-12T08:50:36 It does not say "won't" occur, it says should not occur 2014-08-12T08:50:47 once you enable interrupts before calling the user handler, another interrupt can happen 2014-08-12T08:50:48 I guest because of the limited size of the task stack they allocate 2014-08-12T08:51:01 I assume the arch manual means that it doesn't have IRQ levels so a higher priority can't interrupt. And unless you explicitly reenable interrupts, they won't 2014-08-12T08:51:19 We'll have to look closely at it 2014-08-12T08:51:22 Regarding conditionally switching the RTEMS stack, would it just to check ISR_NEST_LEVEL value? 2014-08-12T08:51:24 but you should implement the framework for nesting 2014-08-12T08:51:28 Yes 2014-08-12T08:51:39 There should be examples in other ports 2014-08-12T08:51:43 gedare: I did implement that 2014-08-12T08:51:46 ok great. 2014-08-12T08:52:01 In our case, we have a dedicated stack, so should be OK to switch and nest. Whether it happens or not, the ISR_NEST_LEVEL is a part of the porting framework that other code can depend on . 2014-08-12T08:52:35 And if it's greater than 1, I should not do anything else related to the stack? 2014-08-12T08:52:35 about the 64-bit load/store, some architectures have double-word ld/st especially for floating point regs, that might overlap general-purpose regs, it is something worth looking into. 2014-08-12T08:52:58 Well... 2014-08-12T08:53:01 But load/store double registers can come after the initial code is merged. 2014-08-12T08:53:09 If you think nesting shouldn't happen, you should probably abort! 2014-08-12T08:53:14 gedare: FP is not implemented in mor1kx CPU 2014-08-12T08:53:22 ok 2014-08-12T08:53:23 And not fully implemented on or1ksim 2014-08-12T08:53:25 ok 2014-08-12T08:53:37 They use some host FP instructions for or1ksim 2014-08-12T08:53:54 Do the cleanup we discussed, prepare a separate patch for the power management 2014-08-12T08:53:56 *** krzysiekm13 has quit IRC (Quit: Page closed) 2014-08-12T08:54:01 Send the cleaned-up initial port 2014-08-12T08:54:02 Ok, I have fixed all the issues we discussed about 2014-08-12T08:54:04 So we can commit it 2014-08-12T08:54:10 I haev seen architectures that had no FPU and could save/load up to four registers in a single insrtuction. Just for speed. Saes the instruction decode cycles and basically gets you a cache line in one fetch into registers 2014-08-12T08:54:14 Then you can add features and improve from there. 2014-08-12T08:54:32 Also, we need documentation now 2014-08-12T08:54:38 You need to add a section to the CPU manual 2014-08-12T08:54:39 DrJoel: Yes, ARM is doing that (pushing and poping the whole stack in one instruction) 2014-08-12T08:54:53 And to update various wiki pages to indicate the new support exists 2014-08-12T08:55:03 gedare: Sure that's on my TODO list 2014-08-12T08:55:19 We'd like the updated CPU manual with the initial port 2014-08-12T08:55:35 (Otherwise we've found it is hard to get people to write documentation in retrospect.) 2014-08-12T08:55:35 Ok, regarding the process of submitting patches... 2014-08-12T08:55:57 doc/cpu_supplement and add a chapter for this port. A lot of that should be easy to write now that you have done it. 2014-08-12T08:56:35 For cpukit, would I submit a big bulk patch like I did (fixed one)? 2014-08-12T08:56:46 Or should I divide it up? 2014-08-12T08:57:11 IMO One big one so we can build. 2014-08-12T08:57:18 Warnign free :) 2014-08-12T08:57:39 DrJoel: Yes I forgot to tell in the e-mail that I have fixed warnings 2014-08-12T08:57:48 Awesome! 2014-08-12T08:58:10 For static inline functions you mentioned, 2014-08-12T08:58:24 other ports does not provide it static inline 2014-08-12T08:59:04 unless you mean _CPU_ISR_Get_level 2014-08-12T08:59:05 Do they do macros? 2014-08-12T08:59:51 Ok, I misread the comment, I thought it's related to the following function, sorry.. 2014-08-12T09:00:00 The disable must be a macro since the single parameter is set as a side-effect. 2014-08-12T09:00:20 He was referring to the mfspr and mtspr functions 2014-08-12T09:00:21 Would I still need to provide the _CPU_IDLE function in a separate file? 2014-08-12T09:00:21 OK. Macros vs inline doesn't matter. inlines tend to be better because of type checking, etc. 2014-08-12T09:00:27 us 2014-08-12T09:00:44 You should provide BSP_PROVIDES_IDLE 2014-08-12T09:00:46 gedare: They are already static inline 2014-08-12T09:00:48 or whatever feature macro it is 2014-08-12T09:00:53 oh, nvm then 2014-08-12T09:01:09 but I did just see _CPU_ISR_Set_level is inline, without static. 2014-08-12T09:01:23 It should always be 'static inline' not just 'inline' 2014-08-12T09:01:27 gedare: Yes I think these what DrJoel talked about 2014-08-12T09:01:30 ok 2014-08-12T09:01:40 regarding BSP_PROVIDE_IDLE 2014-08-12T09:01:41 #if (CPU_PROVIDES_IDLE_THREAD_BODY == FALSE) 2014-08-12T09:01:41 * This routine is the body of the system idle thread. 2014-08-12T09:01:41 * 2014-08-12T09:01:41 * NOTE: This routine is actually instantiated by confdefs.h when needed. 2014-08-12T09:01:41 */ 2014-08-12T09:01:41 void *_Thread_Idle_body( 2014-08-12T09:01:42 uintptr_t ignored 2014-08-12T09:01:42 ); 2014-08-12T09:01:43 #endif 2014-08-12T09:02:04 and, inline functions that are part of an interface must be in a header file. 2014-08-12T09:02:08 So I should set CPU_PROVIDES_IDLE_THREAD_BODY to true ? 2014-08-12T09:02:10 That's from score/include/rtems/score/thread.h. Delete the CPU idle body and define CPU_PROVIDES_IDLE_THREAD_BODY to FALSE 2014-08-12T09:02:36 Your idle body is the same as the generic one. So you don't need to provide it. 2014-08-12T09:02:47 DrJoel: I update it 2014-08-12T09:02:49 I think there is a standard way to have the BSP provide it though. 2014-08-12T09:02:55 If or1k has halt or lower power mode instruction, then you put that in a loop. 2014-08-12T09:02:58 With sleep instruction 2014-08-12T09:03:18 There are 4 power modes (added them today) 2014-08-12T09:03:39 Yeah, set it to FALSE, and then you can provide it in the BSP 2014-08-12T09:03:44 Look at erc32/startup/bspidle.c for an example. But that's BSP specific. If you have an arch version, then OK. 2014-08-12T09:03:58 yeah true, if it is at arch-level do it in the CPU port 2014-08-12T09:04:00 gedare: But it's part of the CPU 2014-08-12T09:04:05 alright, lets move on... handle the rest of this by email 2014-08-12T09:04:11 ok, then provide it in your cpu port 2014-08-12T09:04:15 The sleep mode must allow for quick response to interrupts. You probably can't go into a deep sleep. 2014-08-12T09:04:16 and define the macro to TRUE 2014-08-12T09:04:17 ;-) 2014-08-12T09:04:24 make it part of your follow-up patch, OK? 2014-08-12T09:04:34 and for now just leave the stub function you included 2014-08-12T09:04:37 gedare: Ok 2014-08-12T09:04:48 Alright. Moving on to asuolen 2014-08-12T09:04:59 ok 2014-08-12T09:05:12 This week I have added interrupt-driven access to the I2C bus and corrected some pending issues with SPI and the I2C buses. During this process I have created some config options on the raspberry Pi BSP configure.ac script which allows to enable any of the buses and to set its IO mode (polled or interrupt driven). 2014-08-12T09:06:19 great. this week, as with everyone else, please focus on code cleanup, documentation, and submission. 2014-08-12T09:06:20 The bus and devices registration are now made in bsp_predriver_hook function, regarding that the bus is enabled in the configure.ac script 2014-08-12T09:06:31 perfect 2014-08-12T09:07:24 as for documentation should I just fill in the wiki and document the code? 2014-08-12T09:08:23 Yeap 2014-08-12T09:08:48 Maybe write some howto for raspberrypi users, blog posts are great for that though. 2014-08-12T09:09:07 Enough so that your code is maintainable and usable. 2014-08-12T09:09:29 Folks tend to rely on examples so try to have one or more that are decent. 2014-08-12T09:09:37 yes will do that 2014-08-12T09:09:58 Yes I have some examples already, just need to explain how to use them 2014-08-12T09:11:25 I assumed you did. :) 2014-08-12T09:11:59 Good. Also, for the RTEMS project, it might be nice to provide a "shopping list" of affordable hardware in case someone wants to setup a test environment with your setup 2014-08-12T09:12:33 There is some driver code that I also created to test the buses. The driver I have done for the mcp23008 I2C device is working and can be submitted also. It allows to expand the number o gpio pins on the Pi using the I2C bus 2014-08-12T09:12:46 *** Sched has quit IRC () 2014-08-12T09:13:24 ok.. well, focus on getting the code in shape this week 2014-08-12T09:13:39 That really does go for all the students, since the deadline is looming. 2014-08-12T09:13:51 And it is good to get code submitted ASAP to avoid bitrot 2014-08-12T09:14:07 Ok will do that 2014-08-12T09:14:22 I think that wraps it up... no meeting next week.. Thanks everyone for a great GSoC! 2014-08-12T09:14:26 asuolen: Did you get a Geert Board? 2014-08-12T09:14:52 DrJoel: no I have not 2014-08-12T09:15:09 OK.. just curious .. looked like a nice IO board for the Pi 2014-08-12T09:18:10 All, I will be here the next few weeks to answer questions, etc 2014-08-12T09:18:22 DrJoel: After using the new sleep IDLE instruction, I got "Fatal Error 5 Halted" from a call to _CPU_Fatal_halt, though, ticker continues working as expected 2014-08-12T09:18:26 So even though this is the last official meeting, feel free to drop by Tuesday mornings to chat 2014-08-12T09:19:16 gedare: Thanks for helping us during GSoC 2014-08-12T09:19:53 And the same for all community and mentors :) 2014-08-12T09:21:22 yes, you all have our thanks 2014-08-12T09:22:37 It is mostly our pleasure 2014-08-12T09:23:00 We just want to see you all succeed and continue to contribute to open-source as you can 2014-08-12T09:23:45 hesham is the instructin in a loop? 2014-08-12T09:24:19 Nope, should it? other ports does just emit asm function with sleep instruction 2014-08-12T09:24:39 I second gedare and really encourage you to help us get all your code merged. We now have hard work to help get it all reviewed and merged. :) 2014-08-12T09:25:23 The idle thread needs to be in a loop 2014-08-12T09:25:31 Well... 2014-08-12T09:25:34 I take that back. 2014-08-12T09:25:58 It probably needs to be in a loop. 2014-08-12T09:25:59 ;-) 2014-08-12T09:26:09 *** asuolen has quit IRC (Ping timeout: 246 seconds) 2014-08-12T09:26:10 No.. the idle thread body is a loop. This is the erc32 2014-08-12T09:26:13 void *bsp_idle_thread( uintptr_t ignored ) 2014-08-12T09:26:13 { 2014-08-12T09:26:13 while (1) { 2014-08-12T09:26:13 ERC32_MEC.Power_Down = 0; /* value is irrelevant */ 2014-08-12T09:26:13 } 2014-08-12T09:26:13 return NULL; 2014-08-12T09:26:13 } 2014-08-12T09:26:33 If he fell off the bottom, it would have been a fatal error 2014-08-12T09:27:20 yea 2014-08-12T09:27:55 The CPU should awake up upon any pending interrupts 2014-08-12T09:27:56 hesham: which one wasn't in a loop? 2014-08-12T09:28:16 void _CPU_Thread_Idle_body( void ) 2014-08-12T09:28:16 { 2014-08-12T09:28:16 do { 2014-08-12T09:28:16 _OR1K_CPU_Sleep(); 2014-08-12T09:28:16 } while (0); 2014-08-12T09:28:16 } 2014-08-12T09:28:27 it should be while(1); 2014-08-12T09:28:51 Sorry... 2014-08-12T09:29:04 The do {..} while(0) is syntactic sugar to allow macros to always expand properly on if then else statements 2014-08-12T09:29:20 No big deal. It is late there :) 2014-08-12T09:30:24 DrJoel: Yes, I used to do it, I got it wrong here ;) It's now working fine, thanks :) 2014-08-12T09:31:33 Awesome! 2014-08-12T09:32:10 gedare: Do you know of any gcc patches I haven't merged? I am working through a few of mine in the background. 2014-08-12T09:32:25 sebhub posted something recently 2014-08-12T09:32:29 let me look 2014-08-12T09:32:54 [PATCH] RTEMS: Add multilibs for ARM 2014-08-12T09:32:57 search for that 2014-08-12T09:33:38 [PATCH v2] RTEMS: Add multilibs for ARM 2014-08-12T09:33:41 looks like he updated it 2014-08-12T09:33:44 so look for the newer one 2014-08-12T09:34:16 also, this one: [PATCH] RTEMS: Add Nios 2 support 2014-08-12T09:34:35