2011-08-15T00:11:56 *** xiangfu has quit IRC 2011-08-15T00:20:22 *** xiangfu has joined #rtems 2011-08-15T01:05:44 *** xiangfu has quit IRC 2011-08-15T01:18:41 *** xiangfu has joined #rtems 2011-08-15T01:19:52 *** monstr-rtems has joined #rtems 2011-08-15T01:27:19 *** sebhub has joined #rtems 2011-08-15T01:31:52 good morning 2011-08-15T01:56:48 *** xiangfu has quit IRC 2011-08-15T02:09:15 sebhub: morning ! 2011-08-15T02:10:43 *** xiangfu has joined #rtems 2011-08-15T02:18:43 did somebody use the gumstix with qemu ? 2011-08-15T02:19:22 from http://www.rtems.com/ml/rtems-users/2009/september/msg00158.html it seems it worked, but i'm wondering how can i get the flash image and all other required files to boot a binary 2011-08-15T02:21:55 please have a look at the rtems-testing scripts: http://www.rtems.org/cvsweb/cvsweb.cgi/rtems-testing/sim-scripts/ 2011-08-15T02:33:39 sebhub: thanks ! 2011-08-15T02:54:52 does the printk on NDS rely on the interrupt support? 2011-08-15T02:57:02 *** xiangfu has quit IRC 2011-08-15T02:59:06 *** xiangfu has joined #rtems 2011-08-15T03:06:40 sebhub: no 2011-08-15T03:07:04 sebhub: but if you issue a printk BEFORE initializing the interrupt, it puts a mess on the interrupt context 2011-08-15T03:07:25 so that as soon as an interrupt is triggered, the execution environment for this interrupt is bad and the system crash 2011-08-15T03:08:05 dd of=flash.img bs=128k count=128 if=/dev/zero && dd of=flash.img bs=128k conv=notrunc if=u-boot-connex-400-r1604.bin && dd of=flash.img bs=1k conv=notrunc seek=4096 if=node_a.ralf && qemu-system-arm -M connex -m 289 -nographic -monitor null -pflash flash.img 2011-08-15T03:08:09 and then ... 2011-08-15T03:08:11 GUM> bootelf 400000 2011-08-15T03:08:18 Clearing .bss @ 0x02029fb0 (55788 bytes) 2011-08-15T03:08:18 ## Starting application at 0x02000000 ... 2011-08-15T03:08:18 qemu: fatal: Trying to execute code outside RAM or ROM at 0x02000000 2011-08-15T03:08:18 R00=00000001 R01=a3ee02e4 R02=40100000 R03=00000000 2011-08-15T03:08:18 R04=02000000 R05=00000001 R06=a3ee02e0 R07=00000002 2011-08-15T03:08:18 *** xiangfu has quit IRC 2011-08-15T03:08:20 R08=a3edefdc R09=a3ee0230 R10=00000000 R11=ffffffff 2011-08-15T03:08:22 sounds weird 2011-08-15T03:08:23 R12=a3edea14 R13=a3edec9c R14=a3f0a1dc R15=02000000 2011-08-15T03:08:25 PSR=600001d3 -ZC- A svc32 2011-08-15T03:08:28 [1] 9081 abort qemu-system-arm -M connex -m 289 -nographic -monitor null -pflash flash.img 2011-08-15T03:08:31 shit 2011-08-15T03:08:35 sebhub: what ? 2011-08-15T03:08:54 that the printk does such a damage 2011-08-15T03:09:17 sebhub: ok, if you want to investigate more, be my guest ! 2011-08-15T03:09:22 the fact is without this patch 2011-08-15T03:09:28 it crashed 2011-08-15T03:09:35 no problem, i check it in as it is 2011-08-15T03:09:36 and with, all my apps are working 2011-08-15T03:11:44 so now, i'm wondering why the gumstix support seems to be not working:/ 2011-08-15T03:12:22 i changed a lot in the interrupt support and the linker command files 2011-08-15T03:12:35 i only checked that it compiles and links 2011-08-15T03:13:45 for the gumstix you mean ? 2011-08-15T03:14:03 all arm bsps 2011-08-15T03:14:42 it should be in the cvs now 2011-08-15T03:14:49 ok 2011-08-15T03:14:52 thanks 2011-08-15T03:16:29 but i have a lot of warning with the nds bsp 2011-08-15T03:16:36 related to the interwork options 2011-08-15T03:16:46 I don't really know what it is and if it can cause issues 2011-08-15T03:17:04 can you please test also with the eabi tool chain? 2011-08-15T03:17:17 how can I do ? 2011-08-15T03:17:28 I have the eabi installed 2011-08-15T03:17:36 http://wiki.rtems.org/wiki/index.php/ARM-EABI 2011-08-15T03:17:41 ok, nice 2011-08-15T03:17:45 I have to go in 5 minutes 2011-08-15T03:17:46 the arm-rtemseabi4.11-gcc ? 2011-08-15T03:17:52 but I try today in the afternoon 2011-08-15T03:18:07 works only with the latest 4.11 cvs 2011-08-15T03:18:09 sebhub: ah, not this one but I can buil dit 2011-08-15T03:18:21 its available via rpm 2011-08-15T03:21:00 debian user you know ... :D 2011-08-15T03:21:28 *** xiangfu has joined #rtems 2011-08-15T03:22:15 use rpmbuild and the source rpm 2011-08-15T03:22:30 get-apt rpm 2011-08-15T04:46:24 *** xiangfu has quit IRC 2011-08-15T05:40:08 *** methril_ has quit IRC 2011-08-15T06:21:56 *** xiangfu has joined #rtems 2011-08-15T06:26:04 *** methril_work has joined #rtems 2011-08-15T07:41:25 *** lcpfnvcy has quit IRC 2011-08-15T07:42:37 *** lcpfnvcy has joined #rtems 2011-08-15T08:34:30 *** monstr-rtems has quit IRC 2011-08-15T08:43:06 *** monstr-rtems has joined #rtems 2011-08-15T09:15:13 *** antgreen has joined #rtems 2011-08-15T09:30:41 *** xiangfu has quit IRC 2011-08-15T09:41:02 *** antgreen has quit IRC 2011-08-15T09:47:50 *** xiangfu has joined #rtems 2011-08-15T09:52:35 *** xiangfu has quit IRC 2011-08-15T09:53:23 *** xiangfu has joined #rtems 2011-08-15T10:02:10 *** antgreen has joined #rtems 2011-08-15T10:41:34 *** xiangfu has quit IRC 2011-08-15T10:44:18 *** sebhub has quit IRC 2011-08-15T10:46:03 *** antgreen has quit IRC 2011-08-15T11:19:18 *** antgreen has joined #rtems 2011-08-15T14:03:46 *** monstr-rtems has quit IRC 2011-08-15T14:51:15 *** arvind_k has quit IRC 2011-08-15T16:06:08 *** antgreen has quit IRC 2011-08-15T17:45:00 *** xiangfu has joined #rtems 2011-08-15T18:06:56 *** lcpfnvcy has quit IRC 2011-08-15T18:08:47 *** lcpfnvcy has joined #rtems 2011-08-15T18:10:43 *** xiangfu has quit IRC 2011-08-15T19:53:22 *** juli1- has quit IRC 2011-08-15T19:53:41 *** juli1- has joined #rtems 2011-08-15T20:02:04 *** xiangfu has joined #rtems 2011-08-15T20:06:38 *** xiangfu has quit IRC 2011-08-15T20:06:40 *** xiangfu_ has joined #rtems 2011-08-15T20:11:20 *** xiangfu_ has quit IRC 2011-08-15T20:23:21 *** xiangfu_ has joined #rtems 2011-08-15T20:45:14 *** xiangfu_ is now known as xiangfu 2011-08-15T22:23:08 *** arvind_khadri has joined #rtems 2011-08-16T00:25:30 *** arvind_khadri has quit IRC 2011-08-16T00:39:41 *** monstr-rtems has joined #rtems 2011-08-16T00:54:04 *** sebhub has joined #rtems 2011-08-16T01:09:12 good morning 2011-08-16T01:33:10 *** arvind_khadri has joined #rtems 2011-08-16T01:35:45 *** lcpfnvcy has quit IRC 2011-08-16T01:38:24 sebhub: good morning 2011-08-16T01:38:36 sebhub: I built the crosschain for eabi yesterday 2011-08-16T01:38:45 I will compile rtems with eabi tonight :D 2011-08-16T01:46:11 ok, please use the cvs from today, otherwise c++ will not work 2011-08-16T02:40:16 *** lcpfnvcy has joined #rtems 2011-08-16T04:36:32 *** xiangfu has quit IRC 2011-08-16T05:27:30 *** arvind_khadri has quit IRC 2011-08-16T05:40:21 *** arvind_khadri has joined #rtems 2011-08-16T05:44:40 *** pzagor has joined #rtems 2011-08-16T06:44:53 *** antgreen has joined #rtems 2011-08-16T06:49:45 *** antgreen has quit IRC 2011-08-16T06:57:33 *** antgreen has joined #rtems 2011-08-16T08:15:57 *** xiangfu has joined #rtems 2011-08-16T08:18:00 *** lj8175 has joined #rtems 2011-08-16T08:30:40 *** xiangfu_ has joined #rtems 2011-08-16T08:30:46 *** xiangfu has quit IRC 2011-08-16T08:31:41 *** xiangfu_ has quit IRC 2011-08-16T09:33:34 *** pzagor has quit IRC 2011-08-16T10:00:58 *** antgreen has quit IRC 2011-08-16T10:22:21 *** arvind_khadri has quit IRC 2011-08-16T11:04:46 *** lj8175 has quit IRC 2011-08-16T11:05:03 *** sebhub has quit IRC 2011-08-16T12:25:19 *** arvind_khadri has joined #rtems 2011-08-16T13:13:22 *** monstr-rtems has quit IRC 2011-08-16T15:42:41 *** arvind_khadri has quit IRC 2011-08-16T15:55:55 *** arvind_khadri has joined #rtems 2011-08-16T15:55:55 *** arvind_khadri has joined #rtems 2011-08-16T17:29:14 *** antgreen has joined #rtems 2011-08-16T18:06:44 *** arvind_khadri has quit IRC 2011-08-16T20:09:40 *** xiangfu has joined #rtems 2011-08-16T22:09:40 *** xiangfu has quit IRC 2011-08-16T22:21:29 *** xiangfu has joined #rtems 2011-08-16T23:43:11 *** antgreen has quit IRC 2011-08-17T00:19:14 *** Shisui has joined #rtems 2011-08-17T00:20:13 Shisui, hi 2011-08-17T00:23:06 hi kiwichris 2011-08-17T00:23:28 Been a while since I have heard from you. How are things going ? 2011-08-17T00:27:16 Well, it's going well, I'm cleaning up the code, improving the in-line documentation, and fixing the platforms problems 2011-08-17T00:27:40 the SIS BSP helps me to fix endianness problems 2011-08-17T00:27:57 yeah is it good for that. 2011-08-17T00:28:18 pretty hard to find them, though .. :p 2011-08-17T00:28:21 Can you please raise a PR as a feature addition so we can attach patches. 2011-08-17T00:28:44 I am sure you are getting faster and faster at doing that. 2011-08-17T00:29:09 I should write this in my resume :D (joke) 2011-08-17T00:29:22 I'll file a PR :) 2011-08-17T00:30:02 A good embedded engineer can debug using the flashing of an led :) 2011-08-17T00:30:13 (my primary concern is to take part of the bi-endian features of the filesystem, most of the multi-byte data is available in little and big endian) 2011-08-17T00:30:37 ok 2011-08-17T00:31:18 I'd prefer to debug with a flashing led rather than gdb ! 2011-08-17T00:31:58 *** monstr-rtems has joined #rtems 2011-08-17T00:32:14 I do not. gdb is ok. there are far far worse debuggers around. I am using one where a memory dump of a menu or 2 then a dialog then a single column of numbers 2011-08-17T00:32:34 .. dump is a menu ... 2011-08-17T00:32:50 Makes dumping a 1K table impossible 2011-08-17T00:33:34 it looks like a funny debugger ! 2011-08-17T00:34:49 you end up just doing prints which is ok up to a point 2011-08-17T00:37:10 well, GDB is really useful to dig into bugs, but sometimes my stack and registers were messed up and I was unable to find why (e.g. : ATA thread related problems) 2011-08-17T00:37:16 which is really frustrating. 2011-08-17T00:37:51 Maybe using -O would help. 2011-08-17T00:38:19 -0 ? 2011-08-17T01:12:18 *** arvind_khadri has joined #rtems 2011-08-17T01:12:18 *** arvind_khadri has joined #rtems 2011-08-17T01:33:15 Sorry I was distracted. I mean compile optimisation 2011-08-17T01:38:36 *** arvind_khadri has quit IRC 2011-08-17T01:43:19 oh, no matter what level I used, I got the same problems anyway. 2011-08-17T01:43:57 Ok. It could related to whatever bug you are looking into 2011-08-17T01:58:34 *** arvind_khadri has joined #rtems 2011-08-17T02:31:17 *** sebhub has joined #rtems 2011-08-17T02:54:25 sebhub: hi, I see your post on the list 2011-08-17T02:54:43 but I'm not sure how can I check the heap size and so on : I'm really not familiar with linked scripts :/ 2011-08-17T03:16:12 *** xiangfu has quit IRC 2011-08-17T03:18:22 can you debug on the nds? 2011-08-17T03:18:33 *** xiangfu_ has joined #rtems 2011-08-17T03:19:36 *** Shisui has quit IRC 2011-08-17T03:23:27 juli1-: what do you mean ? 2011-08-17T03:23:33 sebhub: at this time, I emulate the system 2011-08-17T03:23:49 meaning, that, until now, the emulation worked fine and worked as on the hardware 2011-08-17T03:25:09 can you set a break point at _start and debug the start code? 2011-08-17T03:25:27 hmmm, no, I don't think so 2011-08-17T03:25:48 but also, you can try the RTEMS 4.10 and the cvs with desmume 2011-08-17T03:25:58 you'll see that the cvs is totally broken :/ 2011-08-17T03:26:56 it should be easy to fix, if you can debug the start sequence 2011-08-17T03:27:26 what do you mean with totally broken, do you reach bsp_start? 2011-08-17T03:28:09 I don't think so 2011-08-17T03:28:16 in fact, nothing is written 2011-08-17T03:28:22 the screen remains totally blank 2011-08-17T03:29:28 and you cannot use a debugger on that target or simulator? 2011-08-17T03:48:06 hmmm, no 2011-08-17T03:48:14 the emulator itself has some debugging facilities 2011-08-17T03:48:20 but that's all 2011-08-17T03:48:29 then its pretty hard to figure out what is going wrong 2011-08-17T03:50:54 indeed :D 2011-08-17T03:51:06 that is why it was already difficult for me to find the initial bug 2011-08-17T03:56:17 *** peerst has joined #rtems 2011-08-17T03:56:23 hi 2011-08-17T04:20:51 what is this for a simulator for nds? 2011-08-17T04:22:04 sebhub: I don't know what do you think :D 2011-08-17T04:22:08 but it works quite good 2011-08-17T04:22:12 give it a try ! 2011-08-17T04:22:13 :D 2011-08-17T04:23:34 is it a special one or is it based on skyeye or qemu? 2011-08-17T04:25:53 *** Shisui has joined #rtems 2011-08-17T04:30:13 *** sebhub has quit IRC 2011-08-17T04:32:21 *** sebhub has joined #rtems 2011-08-17T04:37:47 the DeSumME seems to have GDB support: http://vgcoding.blogspot.com/2011/01/how-to-debug-your-nds-rom-using-visual.html 2011-08-17T04:47:54 *** xiangfu_ has quit IRC 2011-08-17T06:14:22 *** antgreen has joined #rtems 2011-08-17T07:08:29 sebhub: ok, but I didn't manage to get it working 2011-08-17T07:08:44 sebhub: if it works for you, just tell me how you do it ! 2011-08-17T07:08:59 because for me, it is not working :/ 2011-08-17T08:24:50 sry, i have not time to debug the nds, i will revert the changes if we have no solution before the 4.11 release 2011-08-17T08:53:43 *** antgreen has quit IRC 2011-08-17T08:54:58 sebhub: no problem, don't worry, it's just that on my side, i didn't manage 2011-08-17T08:55:09 and just in case, you might also be interested in testing this bsp 2011-08-17T09:15:11 *** arvind_khadri has quit IRC 2011-08-17T11:41:01 *** sebhub has quit IRC 2011-08-17T13:54:51 *** mwalle has quit IRC 2011-08-17T14:06:55 *** mwalle has joined #rtems 2011-08-17T14:19:51 *** monstr-rtems has quit IRC 2011-08-17T15:58:20 *** Shisui has quit IRC 2011-08-17T19:13:09 *** xiangfu has joined #rtems 2011-08-17T19:15:23 *** xiangfu has joined #rtems 2011-08-17T20:04:22 *** xiangfu has quit IRC 2011-08-17T21:36:04 *** xiangfu has joined #rtems 2011-08-17T22:10:19 *** xiangfu has quit IRC 2011-08-18T00:46:12 *** Shisui has joined #rtems 2011-08-18T00:50:24 *** monstr-rtems has joined #rtems 2011-08-18T01:31:19 *** sebhub has joined #rtems 2011-08-18T01:33:24 good morning 2011-08-18T01:47:40 *** arvind_khadri has joined #rtems 2011-08-18T02:01:18 *** arvind_khadri has quit IRC 2011-08-18T02:10:54 hi 2011-08-18T02:13:52 *** arvind_khadri has joined #rtems 2011-08-18T02:24:12 sebhub: good morning ! I manage to get the cvs version of rtems at my work (firewall issues !) 2011-08-18T02:24:39 sebhub: so, in fact, there are already some errors in the link script, I fixed that, but even, it appears that it continues to crash :/ 2011-08-18T02:24:39 ok, great 2011-08-18T02:25:07 however, thsi the eabi, I no longer have any compilation warning 2011-08-18T02:50:30 *** xiangfu has joined #rtems 2011-08-18T03:30:14 *** xiangfu` has joined #rtems 2011-08-18T03:31:22 *** xiangfu has quit IRC 2011-08-18T03:31:58 *** xiangfu` has quit IRC 2011-08-18T03:32:15 *** xiangfu has joined #rtems 2011-08-18T03:37:14 *** xiangfu has quit IRC 2011-08-18T03:42:48 *** xiangfu has joined #rtems 2011-08-18T04:47:45 *** xiangfu has quit IRC 2011-08-18T05:37:45 *** pzagor has joined #rtems 2011-08-18T05:41:34 *** sebhub has quit IRC 2011-08-18T05:46:20 *** pzagor has quit IRC 2011-08-18T05:48:13 *** sebhub has joined #rtems 2011-08-18T07:20:41 *** Shisui has quit IRC 2011-08-18T07:32:44 *** Shisui has joined #rtems 2011-08-18T07:41:30 *** zwj has joined #rtems 2011-08-18T07:41:54 *** xiangfu has joined #rtems 2011-08-18T07:57:25 *** zwj has quit IRC 2011-08-18T08:00:24 *** zwj has joined #rtems 2011-08-18T08:15:17 *** antgreen has joined #rtems 2011-08-18T08:24:24 *** xiangfu has quit IRC 2011-08-18T08:51:46 *** Shisui has quit IRC 2011-08-18T08:56:11 *** zwj has quit IRC 2011-08-18T08:56:24 *** xiangfu has joined #rtems 2011-08-18T09:01:45 *** xiangfu has quit IRC 2011-08-18T09:09:33 *** xiangfu has joined #rtems 2011-08-18T09:25:52 *** sebhub has quit IRC 2011-08-18T09:44:42 *** xiangfu has quit IRC 2011-08-18T11:00:31 *** pzagor has joined #rtems 2011-08-18T11:09:43 *** arvind_khadri has quit IRC 2011-08-18T11:15:48 *** Shisui has joined #rtems 2011-08-18T11:40:14 *** pzagor has quit IRC 2011-08-18T11:43:58 *** monstr-rtems has quit IRC 2011-08-18T13:00:29 *** pzagor has joined #rtems 2011-08-18T13:05:09 *** pzagor has quit IRC 2011-08-18T14:55:22 *** antgreen has quit IRC 2011-08-18T16:20:03 *** Shisui has quit IRC 2011-08-18T18:47:24 *** lcpfnvcy has quit IRC 2011-08-18T18:49:03 *** lcpfnvcy has joined #rtems 2011-08-18T19:47:35 *** xiangfu has joined #rtems 2011-08-18T21:52:05 *** lcpfnvcy has quit IRC 2011-08-18T22:01:03 *** lcpfnvcy has joined #rtems 2011-08-18T22:41:14 *** xiangfu has quit IRC 2011-08-18T22:52:53 *** xiangfu has joined #rtems 2011-08-19T00:13:43 *** monstr-rtems has joined #rtems 2011-08-19T00:38:52 *** xiangfu has quit IRC 2011-08-19T00:51:53 *** xiangfu has joined #rtems 2011-08-19T01:31:49 *** arvind_khadri has joined #rtems 2011-08-19T01:31:49 *** arvind_khadri has joined #rtems 2011-08-19T01:33:13 *** dimonov has quit IRC 2011-08-19T03:25:34 *** Shisui has joined #rtems 2011-08-19T03:26:55 *** arvind_khadri has quit IRC 2011-08-19T03:29:35 hi :) 2011-08-19T03:39:51 *** arvind_khadri has joined #rtems 2011-08-19T04:12:44 *** arvind_khadri has quit IRC 2011-08-19T04:25:38 *** arvind_khadri has joined #rtems 2011-08-19T04:36:11 *** arvind_khadri has quit IRC 2011-08-19T04:49:16 *** arvind_khadri has joined #rtems 2011-08-19T05:26:03 *** Shisui has quit IRC 2011-08-19T05:29:33 *** Shisui has joined #rtems 2011-08-19T07:09:32 *** dr__house has quit IRC 2011-08-19T07:09:37 *** dr__house has joined #rtems 2011-08-19T07:09:50 *** dr__house has joined #rtems 2011-08-19T07:16:44 re (or hello everybody :D) 2011-08-19T08:40:51 *** monstr-rtems has quit IRC 2011-08-19T09:00:04 *** arvind_khadri has quit IRC 2011-08-19T09:12:08 *** arvind_khadri has joined #rtems 2011-08-19T09:28:38 *** xiangfu has quit IRC 2011-08-19T10:02:05 *** arvind_khadri has quit IRC 2011-08-19T11:47:34 *** Shisui has quit IRC 2011-08-19T12:11:46 *** Shisui has joined #rtems 2011-08-19T12:25:39 *** Shisui has quit IRC 2011-08-19T12:43:40 *** Shisui has joined #rtems 2011-08-19T13:03:18 *** Shisui has quit IRC 2011-08-19T13:33:12 *** mw2 has joined #rtems 2011-08-19T13:34:09 *** lcpfnvcy_ has joined #rtems 2011-08-19T13:38:30 *** lcpfnvcy has quit IRC 2011-08-19T13:38:30 *** mwalle has quit IRC 2011-08-19T13:54:40 *** Shisui has joined #rtems 2011-08-19T14:35:53 *** Shisui has quit IRC 2011-08-19T14:58:54 *** Shisui has joined #rtems 2011-08-19T15:04:00 *** Shisui has quit IRC 2011-08-19T16:32:00 *** methril_work has quit IRC 2011-08-19T20:20:28 *** kiwichris has quit IRC 2011-08-20T02:18:15 *** kiwichris has joined #rtems 2011-08-20T07:22:01 *** arvind_khadri has joined #rtems 2011-08-20T08:41:52 *** Shisui has joined #rtems 2011-08-20T08:49:50 *** kristianpaul has quit IRC 2011-08-20T09:08:36 *** arvind_khadri has quit IRC 2011-08-20T09:14:48 *** kristianpaul has joined #rtems 2011-08-20T09:20:57 *** arvind_khadri has joined #rtems 2011-08-20T09:41:17 *** zwj has joined #rtems 2011-08-20T09:55:16 *** arvind_khadri has quit IRC 2011-08-20T11:14:50 *** kristianpaul has quit IRC 2011-08-20T12:33:24 *** kristianpaul has joined #rtems 2011-08-20T12:33:24 *** kristianpaul has joined #rtems 2011-08-20T14:03:38 *** arvind_khadri has joined #rtems 2011-08-20T14:15:50 *** Shisui has quit IRC 2011-08-20T15:19:10 *** methril has joined #rtems 2011-08-20T15:22:25 *** Shisui has joined #rtems 2011-08-20T16:56:27 *** Shisui has quit IRC 2011-08-20T19:57:49 *** kristianpaul has quit IRC 2011-08-20T20:04:53 *** kristianpaul has joined #rtems 2011-08-20T22:18:11 *** arvind_khadri has quit IRC 2011-08-20T22:19:36 *** methril_ has joined #rtems 2011-08-20T22:20:14 *** methril has quit IRC 2011-08-21T00:00:44 *** arvind_khadri has joined #rtems 2011-08-21T00:39:29 *** verm__ has quit IRC 2011-08-21T00:40:48 *** verm__ has joined #rtems 2011-08-21T01:52:09 *** methril_ has quit IRC 2011-08-21T02:04:48 *** methril_ has joined #rtems 2011-08-21T02:09:45 *** methril_ has quit IRC 2011-08-21T02:21:52 *** methril_ has joined #rtems 2011-08-21T02:54:14 *** xiangfu has joined #rtems 2011-08-21T02:58:56 *** xiangfu has quit IRC 2011-08-21T03:19:35 *** Shisui has joined #rtems 2011-08-21T03:21:20 *** methril_ has quit IRC 2011-08-21T03:33:54 *** methril_ has joined #rtems 2011-08-21T05:30:55 *** zwj has quit IRC 2011-08-21T09:01:49 *** panzon has joined #rtems 2011-08-21T12:37:55 *** arvind_khadri has quit IRC 2011-08-21T13:01:28 *** Shisui has quit IRC 2011-08-21T14:01:07 *** kristianpaul has quit IRC 2011-08-21T14:01:42 *** kristianpaul has joined #rtems 2011-08-21T14:31:40 *** kristianpaul has quit IRC 2011-08-21T15:04:27 *** DrJoel has joined #rtems 2011-08-21T15:04:27 *** DrJoel has joined #rtems 2011-08-21T15:04:27 *** ChanServ sets mode: +o DrJoel 2011-08-21T15:05:15 * DrJoel has spent the afternoon trying to close PRs .. many with patches from GSoC students :-D 2011-08-21T15:11:09 *** marta has joined #rtems 2011-08-21T15:13:25 hello joel and all, I'd like to discuss SMP if you have a moment (yes, that's this group from France again) 2011-08-21T15:18:04 I recognized you. Sure on SMP.. Jennifer should be back to tidy things up again tomorrow. We got interrupted by another project. 2011-08-21T15:19:19 In fact, it is more about our patches 2011-08-21T15:19:42 ok. I admit they went to the back of the SMP queue.. but I want to address them 2011-08-21T15:19:51 We looked what Jennifer is doing and we've passed that stage >3 months ago 2011-08-21T15:20:25 your use case was very different from what the main stream goal was. Desirable but not primary 2011-08-21T15:20:34 I wanted to ask what we can do to help: we can make a big code drop with our approach 2011-08-21T15:20:41 ok. ask ahead on the patches. I haven't looked at it in multiple weeks though 2011-08-21T15:21:13 I think that we have more or less the same goal 2011-08-21T15:21:40 I need to check with Jennifer on what is left to merge first. I do not know if it is all in or there are a couple of PRs still open. If we have no SMP PRs open, then we have no more code. So merging from you will cause us absolutely zero grief on that front 2011-08-21T15:21:45 Currently we have mostly fixes of deadlocks/not enough locks everywhere 2011-08-21T15:22:04 :) 2011-08-21T15:22:07 that's good because we suspected we missed some. 2011-08-21T15:22:54 I hate a bunch of PRs but can you break into deadlock fixes and one for each additional lock? 2011-08-21T15:22:59 At some point we stopped sending stuff because of not much reply. We're silently merging things, but it is diverging more and more 2011-08-21T15:23:02 a patch per deadlock fix would overwhelm us 2011-08-21T15:23:13 let's converge. 2011-08-21T15:23:17 Some deadlock fixes require bigger changes 2011-08-21T15:23:46 that's ok.. can that be broken out. How large is entire change set? 2011-08-21T15:24:32 Very large, can get the figure for you (we're on git, so it's easier) 2011-08-21T15:24:53 There's a big issue in all the mutex/spinlock/semaphore things 2011-08-21T15:25:09 git movement for rtems is planned.. looking for best time window.. wanted GSOC to close 2011-08-21T15:25:30 coool....currently the cvs update is taking several hours :) 2011-08-21T15:26:24 so...what to do... can we agree on a big code drop at the last merge point we did? 2011-08-21T15:26:25 huh? are you on a slow network? It isn't a server here. I updated on hotel wifi last week and that wifi sucked. 2011-08-21T15:27:14 We're runing gitcvs, that may be part of the reasons 2011-08-21T15:27:16 but 2011-08-21T15:27:38 gitcvs is SLOW SLOW SLOW.. I ran it directly on server and it took > 30 minutes 2011-08-21T15:28:00 I think it may still be running at least every hour.. not sure offhand 2011-08-21T15:28:19 similar to our experience... 2011-08-21T15:28:34 not ideal, but imports everything 2011-08-21T15:29:00 I think we can commit to reviewing a mega-patch.. I think there is only one SMP patch left for jennifer to merge looking at Bugzilla (all were in there) 2011-08-21T15:29:16 the isr one? 2011-08-21T15:29:27 yes.. there might be a follow up past that. 2011-08-21T15:29:34 for us it's ok, we have a similar things (with different names of course) 2011-08-21T15:30:12 the intention was to add a wrapper for "my core" versus "all cores" .. then follow up with SMP ISR code. I honestly don't remember is that is all of it or one patch following 2011-08-21T15:30:35 good, that's the same that we have 2011-08-21T15:30:41 But we can start reviewing yours. If we try this, does it supercede your other patches? 2011-08-21T15:31:13 Some of these changes are mechanical and large .. just to get a layer inserted. Hard to separate that from meat for review purposes 2011-08-21T15:31:15 we need to merge first, it will be somewhat painful 2011-08-21T15:31:42 We've already had this when we submitted the first set 2011-08-21T15:31:54 We had to dig into the commit layers 2011-08-21T15:31:57 ok.. I will get jennifer to finish merging on our side. then we can focus on yours 2011-08-21T15:32:38 OK... so we'll make a big code drop. I'll check if we can easily merge now. If not, we'll point exactly at the moment when it diverges 2011-08-21T15:32:44 we have a master list of PRs for SMP we have been keeping internally and we tried VERY hard to keep commits small and reviewable. It was a lot of extra work to keep it that way but got reviews 2011-08-21T15:33:50 when thinking of it now it could have been a good idea on our side too, but now it's too late 2011-08-21T15:34:00 If you can submit "functional areas", then the set will shrink as we commit. We had that trouble. Review changed the master code and our SMP on old baseline diverged. :( 2011-08-21T15:34:14 yep.. horses have left the barn. 2011-08-21T15:34:36 I think we can try do split, but I want to first see the total diff 2011-08-21T15:34:44 BTW 2011-08-21T15:34:53 Email myself and Jennifer. She can give a summary of what we have left so you can size up what is coming. Shouldn't be much 2011-08-21T15:35:10 have you added the mutex for isr_disable already? 2011-08-21T15:36:10 added code but disabled since it made initial debug even harder. Too many broken things. 2011-08-21T15:36:27 yes 2011-08-21T15:36:41 that was the hardest thing, get around those deadlocks 2011-08-21T15:36:42 hmmm... smpisr.c or isrsmp.c (forget name) is not committed. I think it is in a PR right now 2011-08-21T15:36:56 yes, it is in the latest one 2011-08-21T15:37:36 ok.. that should be the end of the code on our side. But Jennifer needs to confirm that. 2011-08-21T15:37:46 ok 2011-08-21T15:38:16 the good news is.. if we don't screw up non-xMP configurations, no one cares right now. LOL 2011-08-21T15:38:41 no users == no complaints :-D 2011-08-21T15:38:55 we're checking with non-smp builds too :) 2011-08-21T15:38:58 but we need to get it all working before we branch 2011-08-21T15:39:19 branch for? 2011-08-21T15:39:30 my primary goal is not to make non-SMP larger or slower while adding this 2011-08-21T15:39:34 4.11 series 2011-08-21T15:40:40 I agree...patching two branches later on will be a mess 2011-08-21T15:41:11 I hate to not have annual branches of new release series but getting SMP working isn't going to make an early fall target. December maybe.. more like Q1 calendar 12 2011-08-21T15:41:36 you can't schedule free software easily if you don't like to accept having incomplete work 2011-08-21T15:42:17 for me it's better to delay the release than have something beta-quality 2011-08-21T15:43:06 we have always followed that. push to make them but don't let them happen unless code is ready. I try to compensate by making more .x releases on 4.9 and 4.10 while 4.11 is cooking 2011-08-21T15:43:09 from the good news: for us smp works quite reliably 2011-08-21T15:44:11 great! It was working pretty well for us but there were a couple of lock ups which we assumed were a deadlock. Decided to merge and open it to more eyes. 2011-08-21T15:44:47 qemu doesn't simulate that accurately .. blocks of instructions before switching cores.. tsim is better -- cycle accurate .. but we couldn't use it well enough to get deep insight. Hoped someone with better tools could help 2011-08-21T15:45:09 we're on cycle accurate 2011-08-21T15:45:21 and understand the simulator. :-D 2011-08-21T15:45:29 and processor intimately :D 2011-08-21T15:45:46 and we're running in 16 cores, not 2-4 as you :) 2011-08-21T15:47:09 oh.. one set of patches pending that are not SMP might impact you.. a new scheduler is coming (EDF) and another entry point in scheduler plug-in framework is being added. I was going to review it and try to merge it today but it was missing a new file. So probably this week 2011-08-21T15:47:50 i haven't seen that pr 2011-08-21T15:47:57 which one is it? 2011-08-21T15:48:14 1897 is the tests.. one before (I think) is scheduler. 2011-08-21T15:55:21 I think I'll post a review of the scheduler tomorrow :) 2011-08-21T15:55:31 which scheduler? 2011-08-21T15:56:16 edf 2011-08-21T15:56:33 ok.. go for it.. 2011-08-21T15:56:36 i am heading off 2011-08-21T15:56:38 night 2011-08-21T15:56:41 *** DrJoel has quit IRC 2011-08-21T15:56:53 *** marta has quit IRC 2011-08-21T16:33:08 *** lcpfnvcy_ has quit IRC 2011-08-21T16:33:34 *** lcpfnvcy has joined #rtems 2011-08-21T19:09:17 *** DrJoel has joined #rtems 2011-08-21T19:09:17 *** ChanServ sets mode: +o DrJoel 2011-08-21T19:10:01 *** DrJoel has quit IRC 2011-08-21T20:11:30 *** xiangfu has joined #rtems 2011-08-21T20:38:06 *** kristianpaul has joined #rtems 2011-08-21T20:48:11 *** xiangfu has quit IRC