2012-06-18T00:09:34 *** arvind_khadri has joined #rtems 2012-06-18T01:31:05 *** sebhub has joined #rtems 2012-06-18T01:31:16 good morning 2012-06-18T02:58:43 *** arvind_khadri has quit IRC (Ping timeout: 252 seconds) 2012-06-18T03:13:12 *** arvind_k has joined #rtems 2012-06-18T04:07:32 *** arvind_k is now known as arvind_khadri 2012-06-18T04:07:46 *** arvind_khadri has joined #rtems 2012-06-18T07:29:45 hi all, 2012-06-18T07:30:10 Is there any update about the new net stack imported from freebsd ? 2012-06-18T08:27:36 *** weiY has joined #rtems 2012-06-18T08:32:47 *** antgreen has joined #rtems 2012-06-18T09:12:36 *** xiangfu has quit IRC (Ping timeout: 246 seconds) 2012-06-18T09:20:39 *** arvind_khadri has quit IRC (Ping timeout: 246 seconds) 2012-06-18T09:25:28 *** QingPei has joined #rtems 2012-06-18T10:13:38 *** antgreen has quit IRC (Ping timeout: 245 seconds) 2012-06-18T10:50:00 *** sebhub has quit IRC (Ping timeout: 272 seconds) 2012-06-18T10:51:45 *** xiangfu has joined #rtems 2012-06-18T10:54:02 *** QingPei has left #rtems 2012-06-18T11:14:05 *** weiY has quit IRC (Ping timeout: 244 seconds) 2012-06-18T11:21:56 *** juli1 has quit IRC (Remote host closed the connection) 2012-06-18T11:24:15 *** xiangfu has quit IRC (Quit: Leaving) 2012-06-18T10:45:13 *** rtemsLogger has joined #rtems 2012-06-18T12:15:50 *** juli1 has joined #rtems 2012-06-18T14:30:46 *** panzon has quit IRC (Ping timeout: 265 seconds) 2012-06-18T14:43:27 *** panzon has joined #rtems 2012-06-18T15:35:38 *** weiY has joined #rtems 2012-06-18T15:42:20 *** weiY has quit IRC (Ping timeout: 255 seconds) 2012-06-18T16:21:52 *** kristianpaul has quit IRC (Ping timeout: 272 seconds) 2012-06-18T16:21:56 *** kristian1aul has joined #rtems 2012-06-18T16:47:36 *** kristian1aul has quit IRC (Quit: Reconnecting) 2012-06-18T16:47:45 *** kristianpaul has joined #rtems 2012-06-18T16:47:45 *** kristianpaul has joined #rtems 2012-06-18T17:52:07 *** panzon has quit IRC (*.net *.split) 2012-06-18T17:52:07 *** rokka has quit IRC (*.net *.split) 2012-06-18T17:52:16 *** rokka has joined #rtems 2012-06-18T17:52:42 *** panzon has joined #rtems 2012-06-18T17:58:02 *** kiwichris has quit IRC (Quit: This computer has gone to sleep) 2012-06-18T19:33:50 *** deb has joined #rtems 2012-06-18T19:47:22 *** deb has quit IRC (Quit: Leaving) 2012-06-18T20:03:13 *** kiwichris has joined #rtems 2012-06-18T22:35:10 * kiwichris has fixed the IRC logging to HTML 2012-06-18T22:43:15 Logging check, please ignore 2012-06-18T23:47:53 *** arvind_khadri has joined #rtems 2012-06-19T00:16:53 *** kiwichris has quit IRC (Quit: This computer has gone to sleep) 2012-06-19T00:25:40 *** xiangfu has joined #rtems 2012-06-19T00:30:22 *** kiwichris has joined #rtems 2012-06-19T02:49:47 *** sebhub has joined #rtems 2012-06-19T03:11:34 good morning 2012-06-19T05:46:21 *** kristianpaul has quit IRC (Ping timeout: 248 seconds) 2012-06-19T05:48:07 *** kristianpaul has joined #rtems 2012-06-19T08:18:10 *** QingPei has joined #rtems 2012-06-19T09:05:42 *** deb has joined #rtems 2012-06-19T09:41:53 *** arvind_khadri has quit IRC (Ping timeout: 246 seconds) 2012-06-19T10:03:21 *** xiangfu has quit IRC (Remote host closed the connection) 2012-06-19T10:56:37 *** Hesham has joined #rtems 2012-06-19T11:06:35 *** QingPei has left #rtems 2012-06-19T11:07:42 *** sebhub has quit IRC (Remote host closed the connection) 2012-06-19T11:38:42 *** arvind_khadri has joined #rtems 2012-06-19T11:43:41 *** Hesham has quit IRC (Quit: Leaving.) 2012-06-19T12:01:48 *** gedare has joined #rtems 2012-06-19T12:38:38 *** arvind_khadri has quit IRC (Ping timeout: 246 seconds) 2012-06-19T15:01:36 *** gedare has quit IRC (Remote host closed the connection) 2012-06-19T15:20:15 *** gedare has joined #rtems 2012-06-19T16:04:12 *** gedare has quit IRC (Quit: Ex-Chat) 2012-06-19T16:26:30 *** panzon has quit IRC (Ping timeout: 272 seconds) 2012-06-19T17:11:41 *** kristianpaul has quit IRC (Quit: leaving) 2012-06-19T17:11:52 *** kristianpaul has joined #rtems 2012-06-19T17:16:29 *** kristianpaul has quit IRC (Client Quit) 2012-06-19T17:17:52 *** kristianpaul has joined #rtems 2012-06-19T17:24:19 *** kristianpaul has quit IRC (Quit: leaving) 2012-06-19T17:24:44 *** user has joined #rtems 2012-06-19T17:30:38 *** user has joined #rtems 2012-06-19T17:31:34 *** user has quit IRC (Client Quit) 2012-06-19T17:31:42 *** user has joined #rtems 2012-06-19T17:34:14 *** kristian1aul has joined #rtems 2012-06-19T17:34:58 *** panzon has joined #rtems 2012-06-19T17:36:53 *** kristian1aul has quit IRC (Client Quit) 2012-06-19T17:37:02 *** kristian1aul has joined #rtems 2012-06-19T17:48:53 *** kristian1aul has quit IRC (Quit: Reconnecting) 2012-06-19T17:49:01 *** kristianpaul has joined #rtems 2012-06-19T17:49:01 *** kristianpaul has joined #rtems 2012-06-19T19:55:44 *** kiwichris_ has joined #rtems 2012-06-19T19:58:15 *** kiwichris has quit IRC (Ping timeout: 251 seconds) 2012-06-19T19:58:44 *** kiwichris__ has joined #rtems 2012-06-19T20:01:02 *** kiwichris_ has quit IRC (Ping timeout: 244 seconds) 2012-06-19T20:07:01 *** kiwichris_ has joined #rtems 2012-06-19T20:09:47 *** kiwichris__ has quit IRC (Ping timeout: 246 seconds) 2012-06-19T20:11:24 *** xiangfu has joined #rtems 2012-06-19T20:26:13 *** deb has left #rtems ("Leaving") 2012-06-19T20:37:34 *** deb has joined #rtems 2012-06-19T20:43:24 *** mikeg has quit IRC (Ping timeout: 244 seconds) 2012-06-19T22:31:17 *** deb has quit IRC (Quit: Leaving) 2012-06-19T23:53:48 *** arvind_khadri has joined #rtems 2012-06-20T01:27:58 *** kiwichris_ has quit IRC (Remote host closed the connection) 2012-06-20T01:40:49 *** kiwichris has joined #rtems 2012-06-20T01:50:31 *** sebhub has joined #rtems 2012-06-20T01:51:08 good morning 2012-06-20T03:56:09 *** cdcs has joined #rtems 2012-06-20T04:19:19 *** arvind_k has joined #rtems 2012-06-20T04:21:25 *** arvind_khadri has quit IRC (Ping timeout: 276 seconds) 2012-06-20T05:51:59 *** kiwichris has quit IRC (Quit: This computer has gone to sleep) 2012-06-20T07:21:07 *** kiwichris has joined #rtems 2012-06-20T08:27:35 *** xiangfu has quit IRC (Remote host closed the connection) 2012-06-20T08:38:23 *** weiY has joined #rtems 2012-06-20T08:41:57 *** arvind_k has quit IRC (Ping timeout: 246 seconds) 2012-06-20T09:59:44 *** QingPei has joined #rtems 2012-06-20T10:05:41 *** gedare has joined #rtems 2012-06-20T10:42:13 *** DrJoel has joined #rtems 2012-06-20T10:42:13 *** DrJoel has joined #rtems 2012-06-20T10:42:13 *** ChanServ sets mode: +o DrJoel 2012-06-20T10:44:14 DrJoel: I have sent you the files :-) 2012-06-20T10:58:00 *** weiY has quit IRC (Ping timeout: 272 seconds) 2012-06-20T11:26:45 *** sebhub has quit IRC (Remote host closed the connection) 2012-06-20T11:27:37 I got them but didn't get the .zip.001. That should be about 5-6MB 2012-06-20T11:34:18 juli1, sent email.. virus scanner is killing it. :( 2012-06-20T11:39:01 DrJoel: I resent it 2012-06-20T11:39:11 please let me know if this is ok 2012-06-20T11:43:20 sorry.. the zip split into 3 parts is confusing a virus scanner which thinks the first part should be an intact zip file. 2012-06-20T11:43:33 ok 2012-06-20T11:43:35 and now ? 2012-06-20T11:43:37 this is ok ? 2012-06-20T11:44:24 the one you sent 10 minutes ago as a .doc also got caught. It looks at content not name. 2012-06-20T11:44:56 ok 2012-06-20T11:45:03 you got the .002 and .003 ? 2012-06-20T11:45:33 yes.. they don't have a .zip header.. so look like raw data to the scanner 2012-06-20T11:45:43 ok 2012-06-20T11:45:50 I can try on your gmail account ? 2012-06-20T11:46:05 Sure.. that way we will see whose virus scanner is nailing it :) 2012-06-20T11:46:25 :-) 2012-06-20T11:46:37 sent 2012-06-20T11:48:13 *** QingPei has left #rtems 2012-06-20T11:49:27 that appears to have worked. 2012-06-20T11:49:45 the md5sum passed 2012-06-20T11:50:08 computers .. who needs em? often more trouble than they are worth 2012-06-20T11:51:16 :-) 2012-06-20T11:51:18 so 2012-06-20T11:51:25 you are able to get the whole archive ? 2012-06-20T11:51:30 is that ok for you ? 2012-06-20T11:51:50 I think so. I just need to digest things now. At least the md5sums pass and the file unzip 2012-06-20T11:52:14 I will email you an ls so you can confirm I have everything 2012-06-20T11:54:09 ok 2012-06-20T11:56:43 then I am going to buy a power supply. I bought two Dells at the same time in Dec 2008. Both have had their power supplies die. One died April 24, the other last night. Obviously built with the same batch of components. :) 2012-06-20T11:57:29 rarf 2012-06-20T11:57:31 no luck 2012-06-20T11:57:33 so 2012-06-20T11:57:35 I have to go 2012-06-20T11:57:45 is there an irc meeting tomorrow about the gsoc ? 2012-06-20T11:57:45 good night.. thanks for sending the file piece 2012-06-20T11:58:00 Yes. 2012-06-20T11:59:27 you are welcome :-) 2012-06-20T11:59:30 see you 2012-06-20T12:17:15 DrJoel: can you announce the meeting on rtems-gsoc? 2012-06-20T12:17:56 sure gedare.. no problem 2012-06-20T12:34:12 *** cdcs has quit IRC (Quit: Page closed) 2012-06-20T12:48:27 *** gedare has quit IRC (Quit: Leaving) 2012-06-20T14:48:50 *** gedare has joined #rtems 2012-06-20T15:58:34 *** deb has joined #rtems 2012-06-20T16:25:30 *** gedare has quit IRC (Quit: Leaving) 2012-06-20T16:50:27 *** DrJoel has quit IRC (Quit: Leaving) 2012-06-20T20:23:39 Before I forget, please send a link to the IRC log for the GSoC meeting to GSoC list when it has finished. 2012-06-20T20:55:14 *** xiangfu has joined #rtems 2012-06-20T21:35:45 *** ChanServ has quit IRC (*.net *.split) 2012-06-20T21:50:43 *** ChanServ has joined #rtems 2012-06-20T21:50:43 *** asimov.freenode.net sets mode: +o ChanServ 2012-06-20T22:42:16 *** deb has left #rtems ("Leaving") 2012-06-20T22:58:03 *** alseh has joined #rtems 2012-06-20T23:28:07 *** peerst has quit IRC (Quit: peerst) 2012-06-21T00:27:27 *** alseh has quit IRC (Remote host closed the connection) 2012-06-21T00:36:18 *** arvind_khadri has joined #rtems 2012-06-21T00:36:18 *** arvind_khadri has joined #rtems 2012-06-21T02:00:17 *** arvind_khadri has quit IRC (Read error: Operation timed out) 2012-06-21T02:12:09 *** sebhub has joined #rtems 2012-06-21T02:12:17 good morning 2012-06-21T02:14:18 *** arvind_khadri has joined #rtems 2012-06-21T02:47:31 *** arvind_khadri has quit IRC (Ping timeout: 264 seconds) 2012-06-21T03:58:49 good morning sebhub 2012-06-21T04:03:23 *** arvind_khadri has joined #rtems 2012-06-21T06:39:45 *** alseh has joined #rtems 2012-06-21T07:29:48 *** Hesham has joined #rtems 2012-06-21T07:45:29 *** cdcs has joined #rtems 2012-06-21T07:50:46 good morning sebhub 2012-06-21T07:58:36 rarf 2012-06-21T07:58:45 bad key :-/ 2012-06-21T08:03:24 *** mikeg has joined #rtems 2012-06-21T08:06:54 *** mikeg_ has joined #rtems 2012-06-21T08:08:24 *** mikeg_ has quit IRC (Changing host) 2012-06-21T08:08:24 *** mikeg_ has joined #rtems 2012-06-21T08:08:24 *** mikeg_ has joined #rtems 2012-06-21T08:09:36 *** mikeg has quit IRC (Ping timeout: 245 seconds) 2012-06-21T08:09:48 *** mikeg_ has left #rtems 2012-06-21T08:09:53 *** mikeg_ has joined #rtems 2012-06-21T08:13:36 *** zw_yao has joined #rtems 2012-06-21T08:16:13 *** arvind_khadri has quit IRC (Ping timeout: 265 seconds) 2012-06-21T08:28:10 *** deb has joined #rtems 2012-06-21T08:36:42 *** weiY has joined #rtems 2012-06-21T08:45:21 *** ppisa has joined #rtems 2012-06-21T08:51:02 Hi all. how long time to be leaved for net meetting ? 2012-06-21T08:51:53 about an hour or so? 2012-06-21T08:57:36 yeah, thank you 2012-06-21T09:00:02 *** xiangfu has quit IRC (Ping timeout: 246 seconds) 2012-06-21T09:04:28 *** claas_ziemke has joined #rtems 2012-06-21T09:12:24 ok so 2012-06-21T09:12:31 it starts in 45 minutes ? 2012-06-21T09:16:56 *** zw_yao has quit IRC (Quit: Leaving) 2012-06-21T09:20:20 *** zw_yao has joined #rtems 2012-06-21T09:32:27 *** jennifer has joined #rtems 2012-06-21T09:32:51 *** jennifer is now known as Guest77899 2012-06-21T09:33:24 *** jenniferA has joined #rtems 2012-06-21T09:33:24 *** WikL has joined #rtems 2012-06-21T09:35:51 *** DrJoel has joined #rtems 2012-06-21T09:35:52 *** ChanServ sets mode: +o DrJoel 2012-06-21T09:36:42 http://goo.gl/ia8sn is the GSOC IRC status document to update 2012-06-21T09:37:10 * DrJoel has another meeting 1 hr after this one. If there are any issues requiring discussion, can we do that now? 2012-06-21T09:37:57 i also would be glad when i could go first... will be quick... am sitting with my family on the trrace... am in vacation... 2012-06-21T09:38:40 I'm also ready:) 2012-06-21T09:39:45 i am also ready 2012-06-21T09:39:48 claas_ziemke, NICE! Where are you? 2012-06-21T09:40:18 100km north of berlin in a marina... its an anual thing :-) 2012-06-21T09:40:24 let me walk around the corner and let Jennifer know we are starting 2012-06-21T09:40:31 btw... are u applying to SOCIS again? 2012-06-21T09:40:35 I'm here :) 2012-06-21T09:40:47 cool. We are heading to Chattanooga this weekend for a weekend with friends.. 2012-06-21T09:41:02 Thanks jenniferA Saving me the trouble of walking.. are you wanting me to get fat? 2012-06-21T09:41:11 lol 2012-06-21T09:41:33 juli1 can confirm but I think we have applied. Gedare and kiwichris reviewed our application. 2012-06-21T09:41:33 *** WikL has quit IRC (Read error: Connection reset by peer) 2012-06-21T09:42:09 ok.. just curious... can we start? i mean can i start? :-) 2012-06-21T09:43:03 what is SOCIS about? 2012-06-21T09:43:10 * DrJoel would like to see an rtems qemu focused project.. work on improving coverage trace info, testing procedures, update and push patches, ... juli may have some outstanding qemu issues from Pok as well that need attention. 2012-06-21T09:43:17 Go on Claas 2012-06-21T09:43:21 *** WikL has joined #rtems 2012-06-21T09:43:29 ok joel... sounds nice... so what have i done? 2012-06-21T09:43:45 i fixed the memory map problems... 2012-06-21T09:44:15 then i had a hard time because i had to find out how the boot command works in u-boot 2012-06-21T09:44:24 but now i got everything right.... 2012-06-21T09:44:40 rtems boots straight to the init task now 2012-06-21T09:45:04 also i implemented a uart stub so i can print out characters to the console 2012-06-21T09:45:04 YEAH!!! Sounds like it is about time for a review and first merge! 2012-06-21T09:45:13 printk also works now 2012-06-21T09:45:33 Is there a Wiki page started for the BSP? 2012-06-21T09:45:41 printf is not working because i have not implemented the "real" driver till now 2012-06-21T09:46:01 nope sorry... i think i will put that on the todo for next week 2012-06-21T09:46:29 You can implement a polling driver and verify that works. Then use clock tick or uart isr to get interrupts working 2012-06-21T09:46:46 ok, i will have a look at that... 2012-06-21T09:47:23 When hello works with printf polled and exits as well as possible, the code needs to be up for review and a first merge. That's step 1 of a basic BSP and it will reduce the code reviewed in the next steps 2012-06-21T09:47:23 anything i can read about how drivers implemetnation in rtems? 2012-06-21T09:47:51 sounds ok 2012-06-21T09:48:03 I can email you the class section on console drivers. But looking at code as sebhub suggested earlier is always good 2012-06-21T09:48:37 there is a chapter in the BSP and device driver howto on this. 2012-06-21T09:48:48 Plus ask on the list. Many people have experience on UART drivers. :) 2012-06-21T09:48:52 ok would be great... i also like reading code alot... but docs are also nice 2012-06-21T09:49:17 http://rtems.org/onlinedocs/doc-current/share/rtems/html/bsp_howto/bsp_howto_52.html#Console-Driver 2012-06-21T09:49:33 If it sucks or you have questions, we can fix the document :) 2012-06-21T09:49:42 yeah... i will write a wiki article about the BSP, shall i include how to build the qemu emulator and how to build the nand image too? 2012-06-21T09:49:47 thx 2012-06-21T09:50:25 Yep.. those are definitely items for the Wiki and the BSPs README can reference the Wiki. Someone should have no doubt on how to run it. 2012-06-21T09:51:05 Also rtems-testing/sim-scripts needs to have a script added to run this BSP when you get hello running. If you need to add anything to the rtems-testing/qemu-support directory, feel free. 2012-06-21T09:51:26 so... to wrap up: next steps: write wiki article, clean up some code, get the code reviewed, if that goes weel i will file a pull request and announce the BSP on the ml 2012-06-21T09:51:27 The goal is for a random ignorant person (e.g. me) to pick up the BSP, build it and run things without asking questions 2012-06-21T09:51:50 ok, yes that sounds helpfull 2012-06-21T09:51:59 Yep. Focus on polled hello as milestone for review/submission, then get interrupts beyond 2012-06-21T09:52:15 ok.. you are in good shape.. who wants to go next? 2012-06-21T09:52:25 yeah thats what i ment... irq are the next level of obfuscation... 2012-06-21T09:52:33 I do 2012-06-21T09:52:33 (me) 2012-06-21T09:52:50 zw_yao, beat you WikL .. WikL is # 3 2012-06-21T09:52:53 go zw_yao 2012-06-21T09:53:02 OK... 2012-06-21T09:53:53 what I have done is: the first version of one-rbtree approach is implemented now, it can past the old POSIX key test. 2012-06-21T09:54:02 *** claas_ziemke has left #rtems 2012-06-21T09:54:34 that's great! I know you have had trouble.. 2012-06-21T09:54:45 However, I think the test may also need be re-implmented. 2012-06-21T09:54:48 Is there any email / issue you need reviewed? 2012-06-21T09:55:05 lol.. that doesn't surprise me.. is it incorrect or do we need an additional one to push unlimited threads? 2012-06-21T09:56:09 I don't know now. Actually, I'm not sure what I have implemented can work well... 2012-06-21T09:56:32 hi 2012-06-21T09:56:35 I just run the current POSIX key test... 2012-06-21T09:57:34 There are 5 key tests FWIW 2012-06-21T09:57:39 testsuites/psxtests/psxkey01 2012-06-21T09:57:39 testsuites/psxtests/psxkey02 2012-06-21T09:57:39 testsuites/psxtests/psxkey03 2012-06-21T09:57:39 testsuites/psxtmtests/psxtmkey01 2012-06-21T09:57:39 testsuites/psxtmtests/psxtmkey02 2012-06-21T09:57:55 what's the issue with your implemetnation? 2012-06-21T09:58:16 Oh, I forget the psxtmtests test... I'll check those later. 2012-06-21T09:58:41 Just good to have baseline numbers and see what impact you had 2012-06-21T09:59:23 the thread delete related stuff is still not in good status. 2012-06-21T09:59:30 *** soh_cah_toa has joined #rtems 2012-06-21T10:00:07 I'm not sure whether we need delete the key node when thread deleted. 2012-06-21T10:00:39 I've asked this on ml, however, gedare seems also not sure about it. 2012-06-21T10:00:51 I think you have to. Otherwise, the next thread created (which may get the same id) will get a non-zero value for the key. 2012-06-21T10:01:15 Plus with unlimited threads, the set of allocated threads can shrink so the set of valid Ids can shrink 2012-06-21T10:01:40 And you can exhaust memory quickly if the thread is created periodically. 2012-06-21T10:01:50 OK, I see. 2012-06-21T10:02:17 I also haven't done any code about the critical region. 2012-06-21T10:02:31 ppisa, I would hope the next thread (with the same id) would get an existing node .. but the value would be incorrect.. since it is supposed to be NULL 2012-06-21T10:02:46 Does anyone know which Beagle board claas is working with? The BeagleBone? I'm hoping to keep up with his progress. 2012-06-21T10:02:57 as DrJoel have said, the one-rbtree have some problem with the read. 2012-06-21T10:03:12 I wouldn't worry about the critical sections yet. We need to get the algorithm right and then we can see if you have critical section issues 2012-06-21T10:03:32 OK. 2012-06-21T10:03:38 BAD TIME, have to go to replace modem => offline 2012-06-21T10:04:03 And I'm also not decide what's next to do: 1)improve the rbtree related approah 2)try to implement the hash based approach. 2012-06-21T10:05:14 zw_yao, we can discuss with gedare offline but I think you need a complete solution (e.g. key get, set, delete, and thread delete) before switching approaches 2012-06-21T10:05:23 Then we pick the better approach and review it for critical sections 2012-06-21T10:05:56 I have mentioned the hash based approach, however, there seem be little motivation about this approach...because the hash's worst run time is not appropriate to RTEMS. 2012-06-21T10:06:02 OK, I'll do that. 2012-06-21T10:06:45 :) bad Worst case is bad for rtems :) 2012-06-21T10:07:07 Then give up the hash based approach? 2012-06-21T10:07:11 I am adding an action for you, Gedare and myself to discuss it 2012-06-21T10:07:18 No just finish the rbtree one first 2012-06-21T10:07:39 Then we evaluate its performance, etc and see what to do. You don't have a complete solution to evaluate yet 2012-06-21T10:07:57 OK:) 2012-06-21T10:08:08 *** ppisa has quit IRC (Ping timeout: 240 seconds) 2012-06-21T10:08:27 Then I think I will go on improving the rbtree approach next week. 2012-06-21T10:08:54 check the action item I added and if it isn't accurate, fix it. 2012-06-21T10:09:02 Is that all, we need to move on if we can 2012-06-21T10:09:17 Yeah, go head. 2012-06-21T10:10:14 my turn then? 2012-06-21T10:10:20 Yep.. 2012-06-21T10:10:32 me next, ok> 2012-06-21T10:10:50 (I have to leave early today, a little sleepy now, I'll check the IRC log. Sorry, bye) 2012-06-21T10:11:10 well, I'm still fighting with the compilation 2012-06-21T10:11:23 and there's a problem I posted a question about on the mailing list 2012-06-21T10:11:23 by zw_yao 2012-06-21T10:12:04 Is it "GSoC RTEMS compilation issues"? 2012-06-21T10:12:07 some solutions have been proposed, I posted a reply today but haven't received an answer yet 2012-06-21T10:12:11 yes, yes it is 2012-06-21T10:12:22 I replied this morning. 2012-06-21T10:12:58 I saw it, and posted a reply later 2012-06-21T10:13:31 I don't know that you have many options other than to add a --enable-paravirtualization OR (cheap hack) .. check out the instructions for Tiny/RTEMS and see how an odd feature flag is passed 2012-06-21T10:13:34 WikL: I do not see your reply 2012-06-21T10:14:16 I see it as sent in gmail 2012-06-21T10:14:35 http://www.rtems.org/pipermail/rtems-devel/2012-June/001283.html 2012-06-21T10:14:44 but your address has somehow escaped the cc, juli1 2012-06-21T10:14:52 that's the last time I replied 2012-06-21T10:15:03 and there is nothing newer than that 2012-06-21T10:15:12 WikL: no, I mean, you have to answer on the list :D 2012-06-21T10:15:16 keep going.. 2012-06-21T10:15:23 I have 2012-06-21T10:15:27 will look into it later 2012-06-21T10:15:29 anyways 2012-06-21T10:15:45 the question was how much of the code in score is irrelevant for the hypervisor 2012-06-21T10:16:24 and the answer is that it's pretty much all of it 2012-06-21T10:16:28 your question was "Actually, all of it. The code relates to exception handling, cpu initialization and as mentioned - cli, sti. For the hypervisor, can these be moved to the c/src/lib/libcpu/i386/pok and exclude the score/cpu/i386 altogether depending on --enable-virtualized?" 2012-06-21T10:16:37 yes 2012-06-21T10:16:51 so 2012-06-21T10:17:02 we have to decide right now what to do 2012-06-21T10:17:14 because the project is delayed regarding the initial plans 2012-06-21T10:17:26 and we need a first shot for the mid term review 2012-06-21T10:17:29 I wouldn't move it all.. 2012-06-21T10:17:55 That destroys the performance and structure of the main port on bare hw 2012-06-21T10:18:10 would it be possible to have i386-paravirtualized-rtems4.11? 2012-06-21T10:18:16 DrJoel: I think the idea of a pok BSP was nice 2012-06-21T10:18:45 The mail thread name needs to change :) 2012-06-21T10:19:26 ok so 2012-06-21T10:19:28 ? 2012-06-21T10:20:00 In fact, to be honnest, I did not catch all the reason why we cannot to a pok BSP for the i386 architecture 2012-06-21T10:20:22 I think you can and I would be perfectly OK with that. 2012-06-21T10:20:27 a Pok BSP. 2012-06-21T10:21:16 The issue is score/cpu since that is compiled any time you have CPU X (e.g. 386, sparc, or powerpc) 2012-06-21T10:21:26 *** ppisa has joined #rtems 2012-06-21T10:21:29 I think the question was that you need to ignore score/cpu 2012-06-21T10:21:33 You can ignored c/src/lib/libcpu pieces that don't make sense 2012-06-21T10:22:05 well cdcs you need a different version to compile it. 2012-06-21T10:22:35 yes. but can't you provide those functions at bsp level? 2012-06-21T10:22:37 How about a score/cpu/pok which is conditionally used by each score/cpu/XXX directory when --enable-paravirtualized is enabled 2012-06-21T10:23:05 cdcs, not and inline anything or have the proper settings for the CPU 2012-06-21T10:23:18 Is RTEMS going to be scheduling the threads within the Partition? 2012-06-21T10:23:36 yes 2012-06-21T10:23:37 If so, then it needs the context switch code, bit scan support, context initialization and a bit more 2012-06-21T10:23:45 i think yes. The hypervisor has to handle context switch 2012-06-21T10:23:49 wait wait 2012-06-21T10:23:49 It is primarily the interrupt disable and vector code 2012-06-21T10:23:56 at this time 2012-06-21T10:24:07 what about having just a pok bsp for i386 2012-06-21T10:24:16 put the necessary code in the pok bsp 2012-06-21T10:24:27 it may also require to copy some code from the cpu layer 2012-06-21T10:24:38 but once we have something that starts to work 2012-06-21T10:24:49 we can think to refactor the source tree 2012-06-21T10:24:50 I mean 2012-06-21T10:24:57 that is fine with me up to work from. 2012-06-21T10:25:04 Having something that works first before making big changes 2012-06-21T10:25:06 But I think you will end up with an uncompileable CPU kit :) 2012-06-21T10:25:15 because having new target and so on, this is really overkill 2012-06-21T10:25:37 DrJoel: maybe, maybe not. In that case, we might reconsider the thing 2012-06-21T10:25:48 but we spend too much time on discussing these issues 2012-06-21T10:25:53 I think using a conditional compile from the configure line like __RTEMS_DO_NOT_INLINE_CORE_MUTEX_SEIZE__ is the way to quickly get a conditional flag to compile on 2012-06-21T10:26:04 and adding a Pok BSP 2012-06-21T10:26:25 In addition, WikL , please propose solutions instead of waiting inputs from us. You have to make a first demo in less than 20 days .... 2012-06-21T10:26:41 DrJoel: I definitely agree 2012-06-21T10:26:51 that's the question I asked on the list 2012-06-21T10:27:08 Just ifdef out the problem areas in the score/cpu/i386 and use Pok ok versions.. add a Pok BSP. Ignore libcpu. That should let you get a working solution 2012-06-21T10:27:12 if I can just move the functions into the libbsp and conditionaly ignore score 2012-06-21T10:27:20 WikL: so, yes, go ahead 2012-06-21T10:27:59 WikL, I think you will not be able to compile CPUKIT. Just add the ifdefs on what bothers you in score/cpu and turn them into methods.. defer implementation to Pok BSP 2012-06-21T10:28:06 I agree with both of you. At this point you can change libcpu to suit your needs. Then, later, we can figure out how to include those changes in the tree. 2012-06-21T10:28:29 ok then 2012-06-21T10:28:55 WikL, you are not obligated to use libcpu although things like cpuModel don't make sense to copy. :) 2012-06-21T10:28:55 I know time is short and getting shorter 2012-06-21T10:29:18 WikL: this is mostly your problem unfortunately :-) 2012-06-21T10:29:30 that's why I am suggesting to hack score/cpu with conditionals, add a "feature env variable" for virtualization, and a new BSP 2012-06-21T10:29:42 and I still have to work out fitting the rtems image into pok 2012-06-21T10:30:09 and also adapt pok :D 2012-06-21T10:30:17 juli1, will have to help on those issues 2012-06-21T10:30:22 indeed 2012-06-21T10:30:29 for the first version it turns out that no changes to Pok are needed 2012-06-21T10:30:40 apart from the console driver in a partition 2012-06-21T10:31:06 Focus on a simple hello world Pok x86 BSP .. don't fight the tree.. just conditionalize score/cpu/i386 2012-06-21T10:31:08 WikL: are you sure ? 2012-06-21T10:31:31 I think that was the conclusion during out Saturday(?) meeting 2012-06-21T10:31:37 *our 2012-06-21T10:31:59 :) 2012-06-21T10:32:37 WikL, and juli1 .. anything else? Don't fight the tree WikL .. hack on it .. email me if you have trouble adding and/or using a "feature variable on configure line" 2012-06-21T10:32:52 yes 2012-06-21T10:33:02 WikL: no, there would be few change 2012-06-21T10:33:09 where? 2012-06-21T10:33:09 http://www.rtems.org/wiki/index.php/TinyRTEMS#Configuring_RTEMS_for_Size has an example of using a couple of them 2012-06-21T10:33:18 WikL: but there will be definitely some changes to do, that cannot work out of the box 2012-06-21T10:35:20 ok.. can we move on.. I only have 25 minutes before I need to be in another meeting 2012-06-21T10:35:29 me, next? 2012-06-21T10:35:37 can i be next ? 2012-06-21T10:35:37 jenniferA, can you take over to finish at that point if we aren't? 2012-06-21T10:35:43 ok 2012-06-21T10:35:48 weiY, then Hesham .. got to be quick :) 2012-06-21T10:35:53 ok 2012-06-21T10:35:54 weiY, how goes it? 2012-06-21T10:35:56 I know what to do now, will post to the list about following steps 2012-06-21T10:36:16 this week my main work is to import the atomic implementation from FreeBSD to rtems 2012-06-21T10:36:25 for i386 architecuture 2012-06-21T10:37:05 and there are some freebsd specific macor, so i have to change them to rtems 2012-06-21T10:37:20 and modify them to rtems name style 2012-06-21T10:38:14 and then i am study the concurrencyKit source code and think how to port its test case to rtems 2012-06-21T10:38:30 ok.. just follow what gedare pointed out about staying true to licenses and sources 2012-06-21T10:38:43 Cool.. then no major issues at this point 2012-06-21T10:39:01 WikL, I am adding an action item to make sure we give you help adding a flag :) 2012-06-21T10:39:23 ok, my first plan is to design the single task atomic operation test case without memory barrier 2012-06-21T10:39:29 thanks 2012-06-21T10:39:54 there are some questions 2012-06-21T10:40:18 are we continuing the conversation later or are we done for now? 2012-06-21T10:40:21 first is whether there are macro to indicate 64 bit support? 2012-06-21T10:40:42 WikL, up to Juli.. I probably can't be here but gedare or someone else may be here 2012-06-21T10:41:11 Because i think not all architecuture support 64bit atomic API 2012-06-21T10:41:49 weiY, we don't have an x86_64 port (yet) so no.. there isn't a flag. If this code is in a CPU specific place, then you can use the gcc flag which lets you know that you are in 64bit mode 2012-06-21T10:42:23 the code is in score 2012-06-21T10:42:32 atomic API 2012-06-21T10:43:31 but are the i386 bits in score/cpu/i386 or inline outside that directory? When there are 15 implementations, how will they be organzed? 2012-06-21T10:44:54 this flag is to indicate whether the 64bit atomic API should be supported 2012-06-21T10:45:35 probably more value to ignore x86-64 and see how it is on any other different architecture 2012-06-21T10:45:35 all implementations are in score/cpu/i386/rtems/score 2012-06-21T10:45:40 ok :) 2012-06-21T10:46:05 Then adding one for say PowerPC or SPARC or MIPS would show more about future path than adding unused 64 bit code 2012-06-21T10:46:36 yeah, then i can remove the 64bit atomic API support. if the architecture can support 64 bit i can implemente it in the architecture directory 2012-06-21T10:47:53 ok.. then you shuold have some code ready for review soon? 2012-06-21T10:47:55 this problem i know how to solve 2012-06-21T10:48:14 ok.. 2012-06-21T10:48:23 souinds like you don't have any mentor action items :) 2012-06-21T10:48:29 anything else? 2012-06-21T10:48:31 i think after i add the basic test case i can let all code to be reviewd 2012-06-21T10:48:39 if not, then we can move to Hesham 2012-06-21T10:48:48 i am here 2012-06-21T10:48:51 another question 2012-06-21T10:49:17 whether there are some code sample to refer about import freebsd code to rtems? 2012-06-21T10:51:22 the biggest and best is in rtems-libbsd. We try desperately to preserve their formatting and only add lines 2012-06-21T10:51:42 was just gonna suggest that 2012-06-21T10:51:52 freebsd-to-rtems.py 2012-06-21T10:52:30 where are those source code? the link 2012-06-21T10:52:38 You can do a tkdiff of the freebsd file to the one in that library and that might help you 2012-06-21T10:52:53 http://git.rtems.org/rtems-libbsd.git/ 2012-06-21T10:53:25 I think the file libbsd.txt (asciidoc format) has guidelines and instructions 2012-06-21T10:53:41 ok, thank you. i will study it 2012-06-21T10:54:48 can i start now ? 2012-06-21T10:54:58 weiY, you done? 2012-06-21T10:55:23 yeah, done 2012-06-21T10:56:01 and .. just in from a 5 week engagement in the Catskills.. Hesham 2012-06-21T10:56:11 round of applause.. getting silly in here 2012-06-21T10:56:25 i don not know it :) 2012-06-21T10:56:35 I have implemented two interfaces for Arena manager, and changed previous mmu a little to work with Arena. 2012-06-21T10:56:35 And as ric suggested in devel m-l, i tried to figure out how we could avoid translating from high-level to bsp code. 2012-06-21T10:57:21 so i have created functions like rtems_mm_set_write at high-level layer ( cpukit/libmm/* 2012-06-21T10:58:05 and mid level functions that maps to it _CPU_Memory_Protection_Set_write 2012-06-21T10:58:09 that sounds good 2012-06-21T10:58:29 previous work implemented this on powepc 2012-06-21T10:58:32 and _CPU_Memory_Pro... is usually implemented where in the tree? 2012-06-21T10:58:58 at /libcpu/powepc/mpc6xx/mmu 2012-06-21T10:59:08 pagetable_.c 2012-06-21T10:59:23 that's the previous work 2012-06-21T10:59:36 ok.. then we can discuss if the _CPU_.. name is correct.. adding an action to figure that out. Usually _CPU_ indicates it is in score/cpu 2012-06-21T11:00:03 ok 2012-06-21T11:00:05 the problem is 2012-06-21T11:00:27 the previous work implemented on PowerPC arch 2012-06-21T11:00:38 and my project will work on ARM 2012-06-21T11:01:22 gedare suggested that i may have to dig into powepc code and edit it to get these functions work at low-level 2012-06-21T11:01:51 i have tried to understand previous mmu code for setting attributed like that 2012-06-21T11:02:15 but i think it's not right and also never tested 2012-06-21T11:02:51 sorry .. feeding printer 2012-06-21T11:03:08 So, i may ignore low-level code for powerpc 2012-06-21T11:03:11 it may be able to be tested on psim or qemuppc 2012-06-21T11:03:50 i will have to write that part of code again for powepc 2012-06-21T11:03:51 Yes from a "do I have to test it now" view.. but before end of project get your ARM and it to use same API. Gedare may be able to help test. I think he is more familiar with testing that. 2012-06-21T11:03:58 not just edit it? 2012-06-21T11:04:05 refactor? move around? 2012-06-21T11:04:17 yeah 2012-06-21T11:04:33 ok.. we can discuss that on devel and figure it out 2012-06-21T11:04:52 I also created some test cases for two functions i have implemented 2012-06-21T11:05:12 i have another question 2012-06-21T11:05:26 go ahead 2012-06-21T11:05:36 ric suggested to have an opaque attribute type for Arena 2012-06-21T11:05:45 at the high-level 2012-06-21T11:06:08 but i do not figure out how it should be, and whether it's important or not 2012-06-21T11:06:34 that's after having rtems_mm_set_write and similar functions 2012-06-21T11:06:52 Personally I think it would be better to have a core set of things that are the same across all architectures.. read,write, execute.. etc. and have space for accessing specific architecture capabilities beside those 2012-06-21T11:07:31 we have managed to have most APIs work across all architectures. But BSP and CPU specific additions, helpers. Core MMU functionality should be the same across all MMUs 2012-06-21T11:07:47 basic protection, address mapping, etc 2012-06-21T11:08:10 in the previous work this attributes were kept at uint8_t bit pattern and translated at BSP , CPU layer 2012-06-21T11:08:39 that should be fine with N bits defined across all architectures and some left for extra stuff 2012-06-21T11:08:57 ric suggested to avoid any translation 2012-06-21T11:09:30 and attributes should be manipulated at bsp layer only 2012-06-21T11:09:39 it wouldn't be translated. For example, XXX_READ_ONLY would be defined all the time 2012-06-21T11:09:49 so me and gedare agreed that we delete this bit type 2012-06-21T11:09:59 manipulated yes (between libcpu and BSP code maybe) 2012-06-21T11:10:33 OK. But we need to discuss if there can be a core set of attributes that are known.. otherwise this isn't easy to document and "read only", "execute" will end up different on different targets for no particular reason 2012-06-21T11:11:00 personally I suspect there is really a portable list of capabilities and it is just a matter of if architecture X supports a particular one 2012-06-21T11:11:13 but this is gedare's research arena so he should get a big vote. 2012-06-21T11:11:14 there are initial set of attributes ( read,write,exec,cacheable) 2012-06-21T11:11:45 ok i will discuss with gedare about that issue 2012-06-21T11:12:10 I think those are very close to universal.. but there may be others.. say security levels or something 2012-06-21T11:12:40 I should not be traveling as much for a while so should be able to be more involved in the discsussion 2012-06-21T11:12:48 I am adding this as an action item for us 2012-06-21T11:13:11 i thought we could have a struct with bool type indicating whether these permissions are supported by HW or not . I do not know if it make sense or not 2012-06-21T11:13:21 type ** 2012-06-21T11:14:14 I was thinking of something simmialr earlier.. a method "what do you support?" 2012-06-21T11:14:22 yeah 2012-06-21T11:15:01 that could be useful for importing attributes from BSP to high-level code as i think 2012-06-21T11:15:04 but for that to make sense, there has to be a core set of universal things to support 2012-06-21T11:15:42 i think r/w/x/c are universal enough 2012-06-21T11:15:59 If there is a new useful capability on a new MMU, we can add it to the core set and others not support it. :) 2012-06-21T11:16:51 every BSP has to decide whether to support this capability or not 2012-06-21T11:18:09 yes, we have to decide a set of universal permissions and allow extending this permissions 2012-06-21T11:18:13 Default is "I don't". "get capabilities" fills in all fields as false. Each MMU/BSP implementation only sets bits/fields it actually supports to true. If a new capability is added, then it is set to false by default 2012-06-21T11:18:52 I've got to go now, will post to the devel when I have something. Thanks for the help 2012-06-21T11:18:53 that's what i am thinking of 2012-06-21T11:18:58 *** WikL has quit IRC () 2012-06-21T11:19:51 i will discuss with gedare about this design 2012-06-21T11:19:58 ok.. I added some actions for us on your project. I think we just need some final decisions. If we have to group voice chat with gedare, we can arrange that 2012-06-21T11:20:31 Is anyone else here? 2012-06-21T11:20:38 yes, me 2012-06-21T11:20:41 ok i will, thanks 2012-06-21T11:21:00 alseh, how are things going 2012-06-21T11:21:22 there were going pretty bad becaues I could not run on any simulator 2012-06-21T11:21:39 ... now pisa helped and ported the cirrus driver from linux 2012-06-21T11:22:04 it offers the packed pixel format microwindows requires 2012-06-21T11:22:07 I saw.. if you can address Gedare's license comment, then we can merge that. 2012-06-21T11:22:28 Personally I would be prone to make that the default FB driver but that is for discussion 2012-06-21T11:22:30 I thought I will work first on NXLib ... to get that going 2012-06-21T11:22:59 I agree , and I would like to merge fb_vga and fb_cirrus files 2012-06-21T11:23:03 I think, that there should be option to have more fb drivers and probe them. 2012-06-21T11:23:11 and a define to select the drivers 2012-06-21T11:23:24 probing would be better ... 2012-06-21T11:24:22 however, I thought that I will work on this after I successfully run a NXlib test 2012-06-21T11:25:26 also, I think a ARM simulator should be working ... and as I heard, claas had u-boot working, and I should ask him for some advice 2012-06-21T11:25:29 Problem is on which level drivers should be switched, if we should expect that functions frame_buffer_control etc. should exist only onece, than there should be one level below them which defines actual device for each minor. 2012-06-21T11:26:07 Other option is to not keep single major and register them by name only. 2012-06-21T11:26:41 But nayway, NXlib is primary now. 2012-06-21T11:26:47 We probably should consider the NIC configuration as a model where you can pick the set you want to support in your exe but they probe to see if their hardware is present. In this case, I guess the first would be the default FB 2012-06-21T11:29:51 so, I guess that I coud work this week on NXLib, and we should think for the next meeting on how to progress with pc386 fb driver ? 2012-06-21T11:30:01 *** mikeg_ has quit IRC (Ping timeout: 245 seconds) 2012-06-21T11:30:14 If you address the license question, then I am happy to merge that code. 2012-06-21T11:30:58 nxlib and general fb configuration are issues 2012-06-21T11:31:26 ppisa, the libchip model would even work.. a table of potential drivers with probes (which can be hard-coded to return true) to see what's there 2012-06-21T11:32:02 As for the driver, I think that even VESA protected VBE is feasible. This would help on real HW. 2012-06-21T11:32:35 That would be an improvement 2012-06-21T11:32:39 On the other hand Cirrus is nice, that it can work even on PowerPC when it is inserted into system. 2012-06-21T11:33:21 And it is really simple and does not cause lags as BIOS calls. 2012-06-21T11:33:50 agreed... having anything nice on qemu and then having nxlib working leaves driver work to those who care and gives the community a nice platform to show the capability on 2012-06-21T11:34:22 actually I think it depends on how much RTEMS is used on real x86 HW 2012-06-21T11:35:59 alseh, anyone can use qemu to experiment and see what nxlib, etc can do. Is it right for their app, etc. If they like it, they can right a driver for their hardware. We can never support everything but we can support something everyone can see the feature running on. 2012-06-21T11:36:22 I used to refer to this as a "virtual reference platform". RTEMS needs those as marketing and test engines 2012-06-21T11:37:06 I think that alseh refers to VBE solution. But it is good for Virtual Box and other as well. 2012-06-21T11:37:46 yes, indeed, I was refering that since we can use cirrus driver for QEMU, VBE loses a bit from it's importance 2012-06-21T11:37:58 aaahhh... 2012-06-21T11:38:06 sorry.. 2012-06-21T11:38:19 how about nxlib then vbe.. add capability, then driver? 2012-06-21T11:38:48 that was what I was thinking 2012-06-21T11:39:05 and maybe try to find some enviroment for other simulated platforms as well to prove it's value 2012-06-21T11:39:09 I have looked into MW and there are needed header files. But it wort to implement protected mode entry. 2012-06-21T11:39:21 grow the RTEMS feature set, then spread it across more hardware.. 2012-06-21T11:39:26 ok.. I trust you ppisa :) 2012-06-21T11:40:06 hi, sorry for not being present earlier, but was in a meeting 2012-06-21T11:40:24 its ok sebhub .. I've been bouncing back in and out.. 2012-06-21T11:40:30 did xiang show up today? 2012-06-21T11:40:52 DrJoel: yeah, didn't you have an important meeting? 2012-06-21T11:40:54 not according to the report 2012-06-21T11:41:05 he seems to be pretty inactive 2012-06-21T11:41:20 it was another telecon that I lead for an Open Group standard.. I have a sub-committee with a handful of people and none showed :) 2012-06-21T11:41:50 lol wow 2012-06-21T11:42:08 sebhub, that's not good.. please send him a wakeup email. 2012-06-21T11:44:50 bye, i have to reach the train 2012-06-21T11:45:37 later 2012-06-21T11:45:46 alseh, are you ok? 2012-06-21T11:45:57 need anything specific or are you on track? 2012-06-21T11:46:29 any other students hanging around? 2012-06-21T11:46:35 Hello 2012-06-21T11:46:45 well, I am a bit behind how I planned, but I believe I will recover 2012-06-21T11:47:11 same for me ;) 2012-06-21T11:47:24 ok alseh just ask if you need help. you seem to be on track now 2012-06-21T11:47:35 deb .. how are thinks? 2012-06-21T11:47:41 not so good 2012-06-21T11:47:47 I am behind the schedule 2012-06-21T11:47:57 ok.. what can we do to put you on track? 2012-06-21T11:48:46 Here is what I am trying to find solutions for 2012-06-21T11:49:00 I installed the CDT plugin for eclipse 2012-06-21T11:49:06 *** sebhub has quit IRC (Ping timeout: 246 seconds) 2012-06-21T11:49:39 But it has automatically generated the Makefile for me 2012-06-21T11:49:45 I don't want it 2012-06-21T11:50:12 I am buffled whether or not I can use a single line command to do the cross-compilation 2012-06-21T11:50:23 there is an Eclipse plugin for RTEMS on top of the CDT which subhub wrote and could have said more about 2012-06-21T11:50:58 OK. 2012-06-21T11:51:02 What is that? 2012-06-21T11:51:12 Can you just use Eclipse as an editor and create the Makefiles by hand? I assume you are building outside the main RTEMS tree. The Makefiles in examples-v2 are quite simple 2012-06-21T11:51:30 OK 2012-06-21T11:51:46 Where can I get examples-v2 2012-06-21T11:51:49 *** weiY has quit IRC (Ping timeout: 245 seconds) 2012-06-21T11:51:50 It really would make you productive quicker 2012-06-21T11:51:53 git.rtems.org 2012-06-21T11:52:05 http://git.rtems.org/examples-v2/ 2012-06-21T11:52:05 OK 2012-06-21T11:52:16 I use eclipse sometimes with RTEMS and can give you a guide on how to manage external makefile 2012-06-21T11:52:50 Install a BSP, export RTEMS_MAKEFILE_PATH=${prefix}/${target}/${bsp} where the variables reflect what you built and installed 2012-06-21T11:52:53 then type make 2012-06-21T11:53:36 If using sis, then very simple to build and run. If you have any trouble, I will email you a log of my doing it 2012-06-21T11:54:12 OK 2012-06-21T11:54:15 Thank you. 2012-06-21T11:54:24 *** Hesham has quit IRC (Quit: Leaving.) 2012-06-21T11:54:28 I will write a makefile 2012-06-21T11:54:42 Another question 2012-06-21T11:54:58 Not now. 2012-06-21T11:55:00 shoot 2012-06-21T11:55:17 I may be able to do it. I will try it first 2012-06-21T11:55:25 If problem occurs 2012-06-21T11:55:31 I will then ask 2012-06-21T11:55:41 ok.. don't fight things too long.. :) 2012-06-21T11:56:16 OK 2012-06-21T11:58:02 any other issues? 2012-06-21T11:58:26 not yet 2012-06-21T11:59:04 ok.. let me know when you get hello from examples-v2 working :) 2012-06-21T11:59:07 anyone else here? 2012-06-21T11:59:17 kevin is here 2012-06-21T11:59:20 me yes 2012-06-21T11:59:42 go ahead 2012-06-21T11:59:43 I can get normal ones working 2012-06-21T12:00:06 so it sounds like the best solution regarding the in_cksum_arm.S issue to use the ARM c version right? 2012-06-21T12:00:24 https://docs.google.com/document/d/1hksizaEMVB_xFqRnsrni_L5KeBwZapLLkSkzu-4qNWY/edit needs to be updated everyone :) 2012-06-21T12:00:36 Just when I need some headers added, the compilation options buffles me sometimes. 2012-06-21T12:00:44 I think so. It avoids the .S issue and gets us code from FreeBSD without an advertising clause 2012-06-21T12:01:10 With the examples-v2 Makefiles, that is easy. 2012-06-21T12:01:13 should i delete the file though? 2012-06-21T12:01:27 yes and remove it from the .py file 2012-06-21T12:01:52 ok 2012-06-21T12:03:18 i'll add that to my status and have it done later today 2012-06-21T12:04:31 i'm still a little unclear as to why freebsd maintains our changes though 2012-06-21T12:04:35 good :) ... Keep in mind you will need to regenerate the Makefiles and do a clean build after its removal 2012-06-21T12:05:10 we are too. :) We are lettign out changes to the FreeBSD source pile up, then review it, then work with them to see what can be accepted. Our goal now is to make minimal changes 2012-06-21T12:06:00 soh_cah_toa, With a single ARM BSP linking, you should be able to check the rest of them, right? 2012-06-21T12:07:01 yes, granted that using in_cksum.c instead get the build to work and tests passing 2012-06-21T12:07:50 :) 2012-06-21T12:08:00 Then move through other architectures and BSPs on in_cksum 2012-06-21T12:08:18 right 2012-06-21T12:08:27 and i guess that's it. the other stuff i'm struggling w/ is just "general" things regarding rtems-libbsd as a whole; not necessarily specific to my project 2012-06-21T12:09:14 blinders are a good thing with this project even from our perspective. Many many details and you only need to solve one problem at a time. As you do that your understanding will grow 2012-06-21T12:09:33 Jennifer and I have spent a lot of time talking to Chris Johns and Sebastian about FreeBSD 2012-06-21T12:10:08 you're right. one of my weaknesses is trying to grok everything at once 2012-06-21T12:10:15 ;) 2012-06-21T12:12:23 jenniferA: real quick. you said i need to regenerate the Makefile but rtems-libbsd uses a hard-coded Makefile 2012-06-21T12:12:45 no Makefile.in 2012-06-21T12:12:56 Run the .py script 2012-06-21T12:13:17 right ok 2012-06-21T12:13:48 and probably with a -R script to get the code back to the BSD tree before you do that 2012-06-21T12:14:14 Hopefully we will see a patch for the shared arm linkcmds and the .py file soon. :) 2012-06-21T12:15:10 yeah, that in_cksum_arm.S was holding me back for a bit 2012-06-21T12:15:24 sounds good then. i'm done if there are any other students left 2012-06-21T12:15:55 ok.. anyone else here 2012-06-21T12:17:06 going once.. 2012-06-21T12:18:23 going twice.. 2012-06-21T12:18:26 and we are out of here... 2012-06-21T12:18:38 thanks.. 2012-06-21T12:18:44 yup 2012-06-21T12:18:50 medivhc and Xi Yang were missing today 2012-06-21T12:19:32 everyone please update the status file. We want them to be accurate. I still see a lot of XXX 2012-06-21T12:19:38 https://docs.google.com/document/d/1hksizaEMVB_xFqRnsrni_L5KeBwZapLLkSkzu-4qNWY/edit 2012-06-21T12:19:48 k 2012-06-21T12:20:52 * DrJoel has an errand to run.. dropping off 2012-06-21T12:20:56 *** DrJoel has quit IRC (Quit: Leaving) 2012-06-21T12:22:29 *** Hesham has joined #rtems 2012-06-21T12:28:09 *** ppisa has left #rtems ("Kopete 0.12.7 : http://kopete.kde.org") 2012-06-21T12:29:23 *** Hesham has quit IRC (Quit: Leaving.) 2012-06-21T13:09:10 *** cdcs has quit IRC (Quit: Page closed) 2012-06-21T14:04:12 *** soh_cah_toa has quit IRC (Ping timeout: 246 seconds) 2012-06-21T14:13:52 *** jenniferA has quit IRC (Quit: Leaving) 2012-06-21T14:48:03 *** soh_cah_toa has joined #rtems 2012-06-21T15:03:47 *** Hesham has joined #rtems 2012-06-21T15:41:58 *** alseh has quit IRC (Ping timeout: 245 seconds) 2012-06-21T15:48:47 *** Hesham has quit IRC (Quit: Leaving.) 2012-06-21T16:36:22 *** zw_yao has quit IRC (Read error: Connection reset by peer) 2012-06-21T16:43:24 *** alseh has joined #rtems 2012-06-21T16:54:29 *** soh_cah_toa has left #rtems ("leaving") 2012-06-21T16:56:46 *** alseh_ has joined #rtems 2012-06-21T16:58:29 *** alseh has quit IRC (Ping timeout: 246 seconds) 2012-06-21T17:18:11 *** alseh_ has quit IRC (Ping timeout: 244 seconds) 2012-06-21T17:27:09 *** DrJoel has joined #rtems 2012-06-21T17:27:09 *** ChanServ sets mode: +o DrJoel 2012-06-21T17:28:33 *** DrJoel has quit IRC (Client Quit) 2012-06-21T17:30:32 *** alseh_ has joined #rtems 2012-06-21T19:05:45 *** DrJoel has joined #rtems 2012-06-21T19:05:45 *** ChanServ sets mode: +o DrJoel 2012-06-21T19:05:54 deb.. really here? 2012-06-21T19:24:10 *** DrJoel has quit IRC (Quit: Leaving) 2012-06-21T19:38:04 *** alseh_ has quit IRC (Ping timeout: 246 seconds) 2012-06-21T19:38:08 *** alseh__ has joined #rtems 2012-06-21T20:31:10 *** xiangfu has joined #rtems 2012-06-21T20:37:57 *** alseh_ has joined #rtems 2012-06-21T20:39:36 *** alseh__ has quit IRC (Ping timeout: 272 seconds) 2012-06-21T20:55:58 *** deb has quit IRC (Quit: This computer has gone to sleep) 2012-06-21T21:15:01 *** deb has joined #rtems 2012-06-21T21:21:24 *** alseh_ has quit IRC (Ping timeout: 245 seconds) 2012-06-21T22:33:13 *** deb has quit IRC (Quit: Leaving) 2012-06-21T22:49:40 *** dr__house` has joined #rtems 2012-06-21T22:49:40 *** dr__house` has joined #rtems 2012-06-21T22:49:47 *** dr__house` has left #rtems 2012-06-22T00:14:47 *** alseh has joined #rtems 2012-06-22T01:01:04 *** arvind_khadri has joined #rtems 2012-06-22T01:12:15 *** sebhub has joined #rtems 2012-06-22T01:42:18 good morning 2012-06-22T02:07:23 *** alseh has quit IRC (Read error: Operation timed out) 2012-06-22T03:01:18 good morning sebhub 2012-06-22T03:01:30 hi 2012-06-22T03:24:30 *** arvind_khadri has quit IRC (Quit: Leaving) 2012-06-22T03:24:44 *** arvind_khadri has joined #rtems 2012-06-22T03:25:41 *** arvind_khadri has joined #rtems 2012-06-22T03:25:41 *** arvind_khadri has joined #rtems 2012-06-22T03:44:25 *** arvind_khadri has quit IRC (Ping timeout: 260 seconds) 2012-06-22T04:26:29 *** arvind_khadri has joined #rtems 2012-06-22T08:03:18 *** deb has joined #rtems 2012-06-22T08:05:58 *** xiangfu has quit IRC (Quit: Leaving) 2012-06-22T08:31:30 *** weiY has joined #rtems 2012-06-22T08:39:06 *** weiY has quit IRC () 2012-06-22T08:44:49 *** arvind_khadri has quit IRC (Ping timeout: 276 seconds) 2012-06-22T09:10:18 *** mikeg has joined #rtems 2012-06-22T10:11:38 *** sebhub has quit IRC (Ping timeout: 240 seconds) 2012-06-22T10:15:11 *** Hesham has joined #rtems 2012-06-22T11:43:16 *** gedare has joined #rtems 2012-06-22T11:43:20 hi Hesham 2012-06-22T11:43:33 hi gedare, how are you ? 2012-06-22T11:43:42 pretty good. busy 2012-06-22T11:43:44 these days 2012-06-22T11:43:51 i noticed that 2012-06-22T11:44:07 :) how is your work coming along? 2012-06-22T11:44:36 i wanna have some discussions with you to make final decisions of the design 2012-06-22T11:44:56 Dr joel added some actions in the IRC log for our project 2012-06-22T11:45:18 so, when you are not busy we can talk about these issues 2012-06-22T11:45:42 we can chat now. i'm catching up on email too... will check the irc status log now 2012-06-22T11:45:52 good 2012-06-22T11:46:09 there are major issues you can pick one to start with :) 2012-06-22T11:46:21 ah i see 2012-06-22T11:46:47 permissions type , what should it looks like? ( i discussed that with dr joel ) 2012-06-22T11:46:52 leave off the naming issue for now... that should be easy to fix with search/replace 2012-06-22T11:47:00 okay 2012-06-22T11:47:20 permissions is tricky as evidenced by the powerpc-specific wimg bits...which do matter 2012-06-22T11:47:21 mmu implementation for powerpc 2012-06-22T11:47:49 i think set_attributes function is never tested 2012-06-22T11:47:57 given that the powerpc has the most "mature" implementation it would be best to have it working, but i understand if you are hesitant to dive in to learning the low-level features... 2012-06-22T11:48:06 hmmm 2012-06-22T11:48:10 so i can dig with and discuss about the code to solve that problem 2012-06-22T11:48:29 yeah. that's something that you can harass aanjhan and quanming about 2012-06-22T11:48:52 i can chat with them 2012-06-22T11:49:10 ok so we will let low-level issues for now ? 2012-06-22T11:49:25 until i chat with aanjhan or quanming 2012-06-22T11:49:47 ok. also make sure you cc peter dufault if you have his email.. he has done mmu work with other powerpc targets i think 2012-06-22T11:49:57 so he might have some other feedback perhaps 2012-06-22T11:50:09 ok i will 2012-06-22T11:50:18 i have created two new branches for test cases and powepc work 2012-06-22T11:50:42 i tried to create my own test case but i have a problem with dummy.o 2012-06-22T11:50:57 what branch? 2012-06-22T11:50:58 so i added my test case to mmutest 2012-06-22T11:51:06 to git 2012-06-22T11:51:51 mid/low-level_Work 2012-06-22T11:52:03 brb 2012-06-22T11:52:07 which contain changes at libcpu/* 2012-06-22T11:52:08 tyt 2012-06-22T11:56:13 back..looking 2012-06-22T11:56:48 in libtests? 2012-06-22T11:57:34 try not to add temporary files to git... files ending in ~ 2012-06-22T11:57:53 aha 2012-06-22T11:58:01 what is dummy.o? 2012-06-22T11:58:03 ok i deleted them anyway 2012-06-22T11:58:35 i dunno that object file must be created when i was compiling mmutest1 2012-06-22T11:59:08 so i added my test cases to previous mmutest1 2012-06-22T11:59:12 and it worked 2012-06-22T11:59:19 sounds like something broken 2012-06-22T12:00:03 is it ok to add test cases to mmutest1 ? 2012-06-22T12:00:04 so...what am i looking for? 2012-06-22T12:00:05 for arena 2012-06-22T12:00:17 sure for now...if the mmutest1 worked in the first place 2012-06-22T12:00:23 but really we need to get the libmm code merged 2012-06-22T12:00:27 that is primary importance 2012-06-22T12:00:38 ok let's define goals for libmm 2012-06-22T12:01:08 i think there is a problem with psim when the code invoked isr routines 2012-06-22T12:02:00 the code for libmm is entirely compiled successfully 2012-06-22T12:02:18 after i have fixed some issues i told you 2012-06-22T12:02:46 and made a stub for uninstall_mpe interface because it's not implemented/used 2012-06-22T12:03:26 ok that's good... then test cases? 2012-06-22T12:03:47 i don't want you to get stuck on the arena stuff before libmm is finished 2012-06-22T12:03:56 ok 2012-06-22T12:04:03 you mean previous mmu test cases ? 2012-06-22T12:04:30 yes. some kind of test code for libmm 2012-06-22T12:04:31 i have already implemented two Arena interfaces 2012-06-22T12:04:43 ok 2012-06-22T12:05:02 but we still need to submit a clean "arena free" libmm + tests 2012-06-22T12:05:16 you need clean separation between libmm and arena 2012-06-22T12:05:27 because arena depends on libmm 2012-06-22T12:05:35 for now i included libarena.* at libmm 2012-06-22T12:05:35 so we can't have any code in libmm depending on arena 2012-06-22T12:06:18 i see.. .libarena.c? 2012-06-22T12:06:30 we can move arena code any where, 2012-06-22T12:06:34 yeah 2012-06-22T12:06:36 ok 2012-06-22T12:06:46 libarena.c and .h at libmm now 2012-06-22T12:06:53 that's fine for now as long as the arena code is separable from the rest of libmm 2012-06-22T12:06:54 you can check them at libmm branch 2012-06-22T12:07:08 ok 2012-06-22T12:07:37 arena code just include libmm.h 2012-06-22T12:07:39 when you have git commits you should try to separate them into smaller units 2012-06-22T12:08:06 ok 2012-06-22T12:08:13 i see the commit that adds arena code to libmm... 2012-06-22T12:08:37 i created one for libmm work ( including libarena and fixes at libmmu ) 2012-06-22T12:08:38 i'd like you to create a fresh branch of rtems that adds only the code needed for libmm...nothing about arenas 2012-06-22T12:08:48 and none of these ~ files or .swp files... 2012-06-22T12:09:02 you can customize your .gitignore so they won't get added 2012-06-22T12:09:03 ok 2012-06-22T12:09:38 ok i will 2012-06-22T12:09:42 what's else ? 2012-06-22T12:10:03 where do I see how bsp provides attributes? 2012-06-22T12:10:48 or is it just the stubs now... 2012-06-22T12:10:51 rtems_mm_set_* 2012-06-22T12:10:53 i just created set_* functions, i wanted to discuss with you about attributes type 2012-06-22T12:10:56 ok 2012-06-22T12:11:16 rtmes_mm_set_* are at limmu.h 2012-06-22T12:11:20 i see those. 2012-06-22T12:12:04 dr joel suggested we should have universal attributes that works for all archs 2012-06-22T12:12:20 i suggested r/w/x/c permissions 2012-06-22T12:12:27 yes in the abstract this is good 2012-06-22T12:12:36 and he told me to discuss with you about that 2012-06-22T12:13:10 Also i have suggested that attributes type could be a struct that have bool field for possible permissions 2012-06-22T12:13:40 BSP code can set them to true if it HW support it 2012-06-22T12:13:41 one approach would be to typedef a struct in libmm.h and then have the BSP provide the actual definition of the struct 2012-06-22T12:14:04 because the big issue is ... how efficiently can the BSP load mmu entries with the given attributes 2012-06-22T12:14:28 if the BSP can store exactly the bit pattern necessary to load the mmu control register, and that gets passed to it from libmm, then no translation occurs during mmu enforcement 2012-06-22T12:14:36 and what about the opaque type ? 2012-06-22T12:15:27 the opaque type is exactly what is typedef'd...libmm doesn't know what is in the struct just that the struct exists and can be passed as a pointer 2012-06-22T12:16:17 how should the definition look like at BSP code ? 2012-06-22T12:16:34 basically in libmm.h you have... 2012-06-22T12:16:47 typedef struct rtems_mm_attributes_struct; 2012-06-22T12:16:55 and in each bsp you will have somwhere.... 2012-06-22T12:17:02 struct rtems_mm_attributes_struct { ... } 2012-06-22T12:17:14 sorry, typedef struct rtems_mm_attributes_struct rtems_mm_attributes; probably 2012-06-22T12:17:24 woul be in libmm.h 2012-06-22T12:17:27 aha what should be at { ... } 2012-06-22T12:17:32 the bit pattern ? 2012-06-22T12:17:38 for mmu control register ? 2012-06-22T12:17:41 whatever makes the most sense for the bsp 2012-06-22T12:18:24 that's ok for defining them 2012-06-22T12:18:39 what about reading them at high-level layer ? 2012-06-22T12:19:48 well i think that can be done through an interface to the bsp 2012-06-22T12:20:03 i don't think the time-critical part is in the getting/setting of these attributes.... 2012-06-22T12:20:09 the time-critical part is in loading the mmu 2012-06-22T12:20:25 that's why we let the BSP define the most convenient definition of the tyep 2012-06-22T12:20:27 type* 2012-06-22T12:20:37 aha 2012-06-22T12:20:48 so you'll add some new functions just like the other ones that go down to the bsp... 2012-06-22T12:21:02 and pass around rtems_mm_attribute* 2012-06-22T12:21:27 that for getting them ? 2012-06-22T12:22:03 So now i will create a typedef at high-level code and a stub at BSP code for attributes 2012-06-22T12:22:15 yeah... in fact if the rtems_mm_attribute is part of the mpe then you just pass around an mpe 2012-06-22T12:22:21 great 2012-06-22T12:22:28 good 2012-06-22T12:22:37 what else should i do 2012-06-22T12:22:50 try to get powerpc working ;-) 2012-06-22T12:22:59 make usable test cases for libmm... 2012-06-22T12:23:04 maybe start with something really simple 2012-06-22T12:23:14 i will chat with aanjhan or quanming 2012-06-22T12:23:14 instead of trying to reuse the mmutest1 code 2012-06-22T12:23:30 k 2012-06-22T12:23:54 What is the due date and what code should be submitted before mid-term eval ? 2012-06-22T12:24:19 You tell me ;) 2012-06-22T12:24:40 I will tell you and you add to them :) 2012-06-22T12:24:40 Get a clean version of libmm branch pushed to your github... 2012-06-22T12:24:52 without arena code 2012-06-22T12:24:54 yes 2012-06-22T12:25:07 this will be a branch that will be considered for submitting to rtems 2012-06-22T12:25:16 so it needs to be working... 2012-06-22T12:25:20 have test cases... 2012-06-22T12:25:31 a working architecture e.g. powerpc/psim 2012-06-22T12:25:32 Should i ignore Makefile changes / additions ? 2012-06-22T12:25:40 ? 2012-06-22T12:25:48 that related to arena 2012-06-22T12:25:55 yes for libmm... 2012-06-22T12:26:00 aha ok 2012-06-22T12:26:13 it should be only the changes necessary to have libmm working. 2012-06-22T12:26:25 ok that's an easy task 2012-06-22T12:26:33 about the test cases .... 2012-06-22T12:26:36 yep 2012-06-22T12:26:51 the first test case only works and psim halts after that 2012-06-22T12:27:11 that's the previous libmm work 2012-06-22T12:27:26 I think that's because of isr 2012-06-22T12:27:41 *** MMU ALUT TEST 1 BEGINS *** 2012-06-22T12:27:41 initialize the memory protect manager 2012-06-22T12:27:41 cpu 1, cia 0x3840: double interrupt - MSR[RI] bit clear when attempting to deliver interrupt, cia=0x3840, msr=0x40; srr0=0x8500(cia), srr1=0x82072(msr); trap-vector=0x700, trap-msr=0x40 2012-06-22T12:27:41 /home/hesham/ppc-test/powerpc-rtems4.11/c/psim/testsuites/libtests/mmutest1/mmutest1.exe: Caught signal -1 at address 0x3840 2012-06-22T12:28:00 try to start over with a new mmu test 2012-06-22T12:28:49 you mean i create test to previous libmm code ? 2012-06-22T12:29:06 just create a test for the libmm... 2012-06-22T12:29:11 make it simple 2012-06-22T12:29:19 no timer 2012-06-22T12:29:26 that should avoid interrupts i hope 2012-06-22T12:29:47 anyway i don't know enough to help debug the powerpc/psim right now :( 2012-06-22T12:30:03 i will eliminate isr routines, is it ok ? 2012-06-22T12:30:09 yea 2012-06-22T12:30:56 i don not know about powerpc or psim too :) but i will chat with aanjhan/quanming to get them work 2012-06-22T12:31:21 that's the second goal 2012-06-22T12:31:42 ok... copy hello world and add libmm to it 2012-06-22T12:31:57 that should give you a simple stripped-down code to work from 2012-06-22T12:32:25 actually i have tried that solution with heapwalk test case :) 2012-06-22T12:32:41 i will try it with hello world 2012-06-22T12:33:04 what's the due date ? 2012-06-22T12:35:45 rolling due dates... hard to set due dates. you can try to break down each task you have to do and determine how long they should take 2012-06-22T12:36:06 ideally you can have a minimally working libmm posted to github next week 2012-06-22T12:36:13 with small test case that works on powerpc 2012-06-22T12:36:31 then you can add/fix attributes 2012-06-22T12:36:58 i can submit libmm tomorrow 2012-06-22T12:37:11 heapwalk is problematic... adding mmu protection bits to heap can cause trouble during allocation/free since the heap metadata lie just outside the heap buffers. 2012-06-22T12:37:19 sure but it is not working right? ;) 2012-06-22T12:37:34 libmm work only, not test cases ;) 2012-06-22T12:38:03 should i create new branch for test cases ? 2012-06-22T12:40:24 test cases need to be submitted with libmm 2012-06-22T12:40:43 along with any other changes to powerpc/psim that are required to make it testable/usable 2012-06-22T12:41:01 aha 2012-06-22T12:41:02 ok 2012-06-22T12:41:04 we can't just submit cpukit/libmm....it's useless without supporting cpu/bsp code and test case to show how to use it and that it works 2012-06-22T12:42:00 i see 2012-06-22T12:42:34 i will not submit it until all of them fixed 2012-06-22T12:43:15 yea 2012-06-22T12:43:18 can't submit broken code. 2012-06-22T12:43:54 ok 2012-06-22T13:11:50 *** Hesham1 has joined #rtems 2012-06-22T13:14:34 *** Hesham has quit IRC (Ping timeout: 255 seconds) 2012-06-22T13:29:47 *** gedare has quit IRC (Quit: Leaving) 2012-06-22T14:44:12 *** deb_ has joined #rtems 2012-06-22T14:46:14 *** deb has quit IRC (Ping timeout: 244 seconds) 2012-06-22T14:00:12 *** rtemsLogger has joined #rtems 2012-06-22T15:28:50 *** Hesham1 has quit IRC (Quit: Leaving.) 2012-06-22T19:41:09 *** deb_ has quit IRC (Read error: Connection reset by peer) 2012-06-22T20:57:08 *** deb has joined #rtems 2012-06-22T22:45:56 *** deb has left #rtems ("Leaving") 2012-06-22T23:33:13 *** weiY has joined #rtems 2012-06-22T23:33:24 Hi chris, are you around 2012-06-23T02:28:39 *** panzon has quit IRC (Ping timeout: 255 seconds) 2012-06-23T02:43:12 *** kiwichris_ has joined #rtems 2012-06-23T02:43:13 *** kiwichris has quit IRC (Read error: Connection reset by peer) 2012-06-23T02:49:40 hi, anyone are around 2012-06-23T02:53:35 yep 2012-06-23T02:53:55 hi juli1 2012-06-23T02:53:59 hi 2012-06-23T02:54:16 if i want add a .h file to rtems/rtems what i should do 2012-06-23T02:54:24 i have add .h to makefile.am 2012-06-23T02:54:30 yes 2012-06-23T02:54:37 and enable the maintainer mode in rtems 2012-06-23T02:54:37 hi 2012-06-23T02:54:38 but it did not appeare 2012-06-23T02:54:50 hi chris 2012-06-23T02:54:57 did you enabled the maintainer mode or run bootstrap ? 2012-06-23T02:55:06 preinstall.am did not have the .h file 2012-06-23T02:55:06 Finally get together on IRC. Nice 2012-06-23T02:55:18 no enabled maintainer mode 2012-06-23T02:55:24 There is a perl script that makes those files. 2012-06-23T02:55:45 yeah, chris. today all the time i will on IRC 2012-06-23T02:56:14 Oh. I have the flue so have not been around. 2012-06-23T02:56:21 juli1, you mean i enabled maintainer mode the .h will exsit in preinstall.am? 2012-06-23T02:57:38 Have a look at the preinstall mode in bootstrap in the root dir. It runs ampolish3 2012-06-23T02:57:56 chris, i have complete a general atomic load test case without any memory barrier 2012-06-23T02:57:57 And that script maintains the preinstall files. 2012-06-23T02:58:02 now i am building it 2012-06-23T02:58:05 Nice. 2012-06-23T02:58:25 weiY: no, the enable-maintainer mode avoid you to run bootstrap all the time 2012-06-23T02:58:30 Looking forward to see it running 2012-06-23T02:58:45 how to turn to preinstall mode in bootstrap? 2012-06-23T02:59:35 Look at the '-p' install mode 2012-06-23T02:59:52 ok, i have seen the ampolish3. i try ampolish3 Makefile.am >preinstall.am 2012-06-23T03:00:40 or './bootstrap -p' or '../../../../bootstrap -p' 2012-06-23T03:01:17 juli1, but if i do not run bootstrap the configure script will not be generated 2012-06-23T03:01:30 chris, ok, i will try 2012-06-23T03:02:57 chris, if i build rtems to support smp, whether enable-smp option is enough? 2012-06-23T03:03:10 weiY: yes but AFTER ... 2012-06-23T03:03:25 and how i test the smp rtems on qumu for i386? are there any reference? 2012-06-23T03:03:47 You would need to ask Joel. He is not about this weekend 2012-06-23T03:04:02 yeah, the ./bootstrap -p is ok for me. thank you really 2012-06-23T03:04:28 ok, if joel is around i will ask 2012-06-23T03:04:50 Just run the tests in single CPU mode and see how it goes. 2012-06-23T03:05:04 Is there an --enable-smp option for configure ? 2012-06-23T03:05:19 If there is then try qemu with a couple of processors. 2012-06-23T03:05:33 chris, now i just write the general atomic test case without any memory barrier and only one task. so later the mult-task test case with memory barrier should be designed carefully 2012-06-23T03:06:00 yeah, there is a --enable-smp option 2012-06-23T03:06:04 I would write for more than 1 task. It will still run on a single cpu 2012-06-23T03:06:21 We know this is all for SMP but they need to run on a single cpu 2012-06-23T03:08:50 but memory barrier will only impact mult-cpu 2012-06-23T03:09:11 if all run on a single cpu the impact of memory barrier will not be test 2012-06-23T03:12:01 there are lots of scenes to be tested. 1. one task + atomic operation + single cpu 2. mult-tasks +atomic operation + single cpu 3. mult-tasks +atomic operation + mult-cpu 2012-06-23T03:24:07 *** weiY has quit IRC (Ping timeout: 264 seconds) 2012-06-23T04:37:06 *** weiY has joined #rtems 2012-06-23T08:01:31 *** zw_yao has joined #rtems 2012-06-23T10:37:17 *** weiY has quit IRC (Ping timeout: 252 seconds) 2012-06-23T11:46:32 *** panzon has joined #rtems 2012-06-23T11:46:36 *** panzon_ has joined #rtems 2012-06-23T11:49:25 *** panzon_ has quit IRC (Client Quit) 2012-06-23T15:29:22 *** zw_yao has quit IRC (Read error: Connection reset by peer) 2012-06-23T16:34:44 *** deb has joined #rtems 2012-06-23T21:33:28 *** deb has quit IRC (Quit: This computer has gone to sleep) 2012-06-23T22:38:55 *** deb has joined #rtems 2012-06-23T23:49:02 *** deb has quit IRC (Quit: This computer has gone to sleep) 2012-06-24T00:16:00 *** arvind_khadri has joined #rtems 2012-06-24T00:52:52 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T01:11:41 *** arvind_khadri has joined #rtems 2012-06-24T01:11:41 *** arvind_khadri has joined #rtems 2012-06-24T01:21:52 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T01:23:47 *** arvind_khadri has joined #rtems 2012-06-24T01:23:47 *** arvind_khadri has joined #rtems 2012-06-24T01:29:24 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T01:30:02 *** arvind_khadri has joined #rtems 2012-06-24T01:39:24 *** alseh has joined #rtems 2012-06-24T01:41:00 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T01:43:07 *** arvind_khadri has joined #rtems 2012-06-24T02:02:57 *** arvind_khadri has quit IRC (Quit: Leaving) 2012-06-24T02:04:11 *** arvind_khadri has joined #rtems 2012-06-24T02:30:16 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T02:30:40 *** arvind_khadri has joined #rtems 2012-06-24T02:30:40 *** arvind_khadri has joined #rtems 2012-06-24T02:49:53 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T02:52:56 *** weiY has joined #rtems 2012-06-24T03:00:41 *** arvind_khadri has joined #rtems 2012-06-24T03:00:41 *** arvind_khadri has joined #rtems 2012-06-24T03:35:45 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T03:36:12 *** arvind_khadri has joined #rtems 2012-06-24T03:36:12 *** arvind_khadri has joined #rtems 2012-06-24T03:41:48 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T03:42:07 *** arvind_khadri has joined #rtems 2012-06-24T03:53:01 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T03:53:24 *** arvind_khadri has joined #rtems 2012-06-24T03:53:24 *** arvind_khadri has joined #rtems 2012-06-24T04:06:12 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T04:06:35 *** arvind_khadri has joined #rtems 2012-06-24T04:06:35 *** arvind_khadri has joined #rtems 2012-06-24T04:09:27 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T04:09:51 *** arvind_khadri has joined #rtems 2012-06-24T04:09:51 *** arvind_khadri has joined #rtems 2012-06-24T04:13:21 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-06-24T04:13:45 *** arvind_khadri has joined #rtems 2012-06-24T05:37:10 *** weiY has quit IRC (Ping timeout: 248 seconds) 2012-06-24T06:30:25 *** alseh has quit IRC (Remote host closed the connection) 2012-06-24T07:53:50 *** QingPei has joined #rtems 2012-06-24T08:15:17 *** arvind_khadri has quit IRC (Ping timeout: 265 seconds) 2012-06-24T08:31:43 *** arvind_khadri has joined #rtems 2012-06-24T08:31:43 *** arvind_khadri has joined #rtems 2012-06-24T08:36:08 *** arvind_khadri has quit IRC (Ping timeout: 245 seconds) 2012-06-24T08:38:31 *** deb has joined #rtems 2012-06-24T08:53:41 *** arvind_khadri has joined #rtems 2012-06-24T09:23:24 *** arvind_khadri has quit IRC (Ping timeout: 246 seconds) 2012-06-24T09:40:28 *** arvind_khadri has joined #rtems 2012-06-24T09:42:34 *** QingPei has left #rtems 2012-06-24T09:59:23 *** arvind_khadri has quit IRC (Ping timeout: 244 seconds) 2012-06-24T10:19:05 *** weiY has joined #rtems 2012-06-24T11:04:19 *** weiY has quit IRC (Ping timeout: 245 seconds) 2012-06-24T12:49:28 *** deb has quit IRC (Quit: This computer has gone to sleep) 2012-06-24T14:15:58 *** deb has joined #rtems 2012-06-24T21:17:49 *** soh_cah_toa has joined #rtems 2012-06-24T21:19:52 *** soh_cah_toa has quit IRC (Client Quit) 2012-06-24T22:55:49 *** deb has quit IRC (Quit: This computer has gone to sleep)