2012-05-14T01:33:36 *** arvind_khadri has joined #rtems 2012-05-14T01:52:01 *** sebhub has joined #rtems 2012-05-14T02:00:25 good morning 2012-05-14T02:01:54 *** arvind_khadri has quit IRC (Quit: Leaving) 2012-05-14T02:02:05 *** arvind_khadri has joined #rtems 2012-05-14T02:37:20 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-05-14T02:56:52 *** arvind_khadri has joined #rtems 2012-05-14T02:56:52 *** arvind_khadri has joined #rtems 2012-05-14T03:54:54 *** xiangfu has quit IRC (Remote host closed the connection) 2012-05-14T06:05:51 *** cdcs has joined #rtems 2012-05-14T06:09:03 *** weiY has joined #rtems 2012-05-14T06:40:44 *** weiY has quit IRC (Ping timeout: 256 seconds) 2012-05-14T06:53:13 *** weiY has joined #rtems 2012-05-14T07:16:46 *** xiangfu has joined #rtems 2012-05-14T09:03:40 *** gedare has joined #rtems 2012-05-14T09:16:35 *** DrJoel has joined #rtems 2012-05-14T09:16:35 *** ChanServ sets mode: +o DrJoel 2012-05-14T09:16:43 sebhub, are you really there? 2012-05-14T09:16:49 hi joel 2012-05-14T09:16:55 :) you are there 2012-05-14T09:17:02 a couple of libbsd questions 2012-05-14T09:17:26 i want to write something about the sysinit, but had no time yet 2012-05-14T09:17:32 the net01 test initializes without failure but shouldn't there be some core networking stuff turned on? 2012-05-14T09:17:59 i think this will get pulled in by the if drivers 2012-05-14T09:18:15 We were wondering.. but you had a USB core 2012-05-14T09:18:23 at least this was the case with the usb wlan if drivers 2012-05-14T09:18:26 and we wondered if we needed something similar 2012-05-14T09:18:48 maybe this usb core is redundant 2012-05-14T09:19:04 OK.. Jennifer will move on to testing this on pc386 and see if it gets to a real NIC probe and make that pass 2012-05-14T09:19:34 this will not work since you have no pci bus device that can probe your nics 2012-05-14T09:19:34 Also you mentioned something about turning on bootverbose and it requiring more to be initialized or work. Can you provide a detail or two more 2012-05-14T09:20:32 to get a nic probe you need an pci bus device driver 2012-05-14T09:21:00 Yep. Were going to try Intel FXP since we had that working before with old stack on pc and mips/malta 2012-05-14T09:21:32 getting the probe to pass is step 1. Then network configuration and see if calls like socket work 2012-05-14T09:21:39 you have to add this to the nexus child devices, see _bsd_nexus_devices 2012-05-14T09:22:00 You added SYSINIT sections for it. We are hoping those are OK 2012-05-14T09:22:33 the next issue you have to solve is setting up the pci bus device driver 2012-05-14T09:22:52 without it nothing will be probed 2012-05-14T09:23:19 the bootverbose should be easily supportable, but i disable it to save code space 2012-05-14T09:24:01 one big issue will be the linkcmds files 2012-05-14T09:24:18 we should consider to use linkcmds.base for all architectures in rtems 4.12 2012-05-14T09:24:18 on bootverbose, we figured that but thought you mentioned some infrastructure piece missing. 2012-05-14T09:24:28 yep on linkcmds. We hacked psim and will do same with pc386. 2012-05-14T09:25:08 We think enough of pci bus driver is in. May not be completely right yet but it has been pulled in. 2012-05-14T09:25:24 So time to just see if it it hits fxp probe and what happens from there? 2012-05-14T09:27:23 you need a "pcib" device 2012-05-14T09:28:13 so add that to the nexus? 2012-05-14T09:28:16 the device tree should look like this "nexus" -> "pcib" -> "pci" -> "if_fxp" 2012-05-14T09:28:44 How do you find that out? You clearly are mapping something from the BSD source 2012-05-14T09:28:48 the missing link is "pcib" 2012-05-14T09:29:52 look at the DRIVER_MODULE() and MODULE_DEPEND() defines 2012-05-14T09:30:42 OK.. passing this along to Jennifer. If you think of anything else, just email us. Hopefully that is enough to go on 2012-05-14T09:32:56 an example "pcib" is this "freebsd-8.2/powerpc/mpc85xx/pci_ocp.c" 2012-05-14T09:33:20 there are more in the freebsd tree, but all are board/device/architecuture specific 2012-05-14T09:34:08 the swi01 and timeout01 test cases run on SIS 2012-05-14T09:34:22 please test your changes with them 2012-05-14T09:34:51 any simulator will do 2012-05-14T09:37:07 you should also look in the freebsd handbook for pci 2012-05-14T09:37:32 I will get jennifer to check those tests on psim and pc386 before she moves to pcib 2012-05-14T09:38:01 but when it comes time to probe pci, we really need a pci to probe and that moves us to pc386 2012-05-14T09:38:11 you have to deal also with the resource API, this might cumbersome 2012-05-14T09:39:17 maybe we end up with freebsd and a real time scheduler ;-) 2012-05-14T09:40:11 I actually wondered years ago if RTEMS could morph into a the BSD version of uCLinux .. host a lot of the BSD code but lighter, deterministic, etc. 2012-05-14T09:44:58 The FreeBSD Handbook appears to be user space oriented. Am I missing something 2012-05-14T09:45:33 hm, it should also explain the responsibilities of the pcib and the pci devices 2012-05-14T09:45:48 hi 2012-05-14T09:45:58 hi 2012-05-14T09:46:17 hey.. trying to get details on BSD configuration enough to see if we can get an Intel FXP NIC driver probe to work :) 2012-05-14T09:46:19 sebhub: would you be interested in a splay tree implementation mirroring the rbtree one? 2012-05-14T09:47:10 i have no urgent need for that 2012-05-14T09:47:12 * DrJoel regrets walking into the missing prototype warning quagmire. Turning on warnings should be discussed broadly, have plan to address, and commitment by someone who cares about that warning. 2012-05-14T09:47:29 ok. i have a working splay tree but it is hackish and lacks some features yet 2012-05-14T09:47:51 joel: yes 2012-05-14T09:47:51 not sure if i will just add them in ad-hoc or try to make a solution that shares interfaces with rbtree. 2012-05-14T09:48:49 or if i should convert everything to the freebsd tree.h ;) 2012-05-14T09:48:59 but i don't enjoy reading those macros. 2012-05-14T09:49:13 http://svnweb.freebsd.org/base/head/sys/sys/tree.h?view=log 2012-05-14T09:49:23 hard to decipher. probably easy enough to use 2012-05-14T09:49:28 yes, i wondered why we didn't use this in the first place 2012-05-14T09:50:00 yes, hard to decipher with nasty warnings and errors by the compiler 2012-05-14T09:50:13 a real c-preprocessor nightmare 2012-05-14T09:50:24 i was unaware of it. but the format of using the cpp is not desirable to me anyways 2012-05-14T09:50:27 but thats the way it is in C 2012-05-14T09:50:31 yeah. 2012-05-14T09:50:39 its a nice way to generate templated code 2012-05-14T09:50:44 rather, a working way 2012-05-14T09:50:49 nothing nice about it. 2012-05-14T09:51:12 there are reasons for C++ 2012-05-14T09:58:04 * DrJoel notes that RMS is working today after his incident in Spain 2012-05-14T09:58:11 http://ostatic.com/blog/richard-stallman-cancels-event-rushed-to-hospital 2012-05-14T09:59:01 sebhub: I also investigated briefly a threaded red-black tree, which will eliminate parent pointers. Not needed in my work, but for small targets / large trees it might make a difference. 2012-05-14T09:59:18 an interesting side-effect is that you could overlay the rbtree_node with a chain_node. 2012-05-14T10:00:10 good to hear he is doing ok. 2012-05-14T10:12:36 *** xiangfu has quit IRC (Ping timeout: 272 seconds) 2012-05-14T10:29:42 *** sebhub has quit IRC (Ping timeout: 272 seconds) 2012-05-14T10:31:03 *** weiY has quit IRC (Ping timeout: 255 seconds) 2012-05-14T10:50:38 *** QingPei has joined #rtems 2012-05-14T12:26:52 *** arvind_khadri has quit IRC (Ping timeout: 272 seconds) 2012-05-14T13:00:26 *** cdcs has quit IRC (Ping timeout: 245 seconds) 2012-05-14T13:02:29 yeah teh macros in tree.h suck 2012-05-14T13:02:42 not much you can do about it though this is the only way to realistically do it in C 2012-05-14T13:04:01 the one nice thing about it is it gets a lot of testing within freebsd including performance 2012-05-14T13:04:46 DrJoel: there are some developer docs for freebsd mostly you're expected to read the source which is very well documented.. are you looking for anything in particular? 2012-05-14T13:04:51 the handbook is directly squarly at users. 2012-05-14T13:16:02 how to build up the nexus which is the structure it steps through at initialization. Trying to ensure we initialize the right set of pieces 2012-05-14T12:24:26 *** rtemsLogger has joined #rtems 2012-05-14T12:24:29 you'll probably have to read source for that 2012-05-14T12:25:10 http://www.watson.org/~robert/freebsd/reading/ 2012-05-14T13:24:05 this is one page i've had recorded for a while 2012-05-14T13:24:27 DrJoel: check it out and see if it answers what you need -- i'm pretty sure it does 2012-05-14T13:24:48 *** arvind_khadri has joined #rtems 2012-05-14T13:35:42 *** arvind_khadri has quit IRC (Ping timeout: 252 seconds) 2012-05-14T13:57:08 *** A1Sheds has quit IRC (Ping timeout: 245 seconds) 2012-05-14T14:42:20 *** DrJoel has quit IRC (Quit: Leaving) 2012-05-14T14:52:32 *** antgreen has joined #rtems 2012-05-14T15:44:58 *** A1Sheds has joined #rtems 2012-05-14T15:44:58 *** A1Sheds has joined #rtems 2012-05-14T16:21:22 *** A1Sheds has quit IRC (Quit: puff of smoke) 2012-05-14T16:21:44 *** L84Supper has joined #rtems 2012-05-14T16:31:55 *** antgreen has quit IRC (Read error: Connection reset by peer) 2012-05-14T17:46:26 *** gedare has quit IRC (Quit: Leaving) 2012-05-14T20:14:04 *** xiangfu has joined #rtems 2012-05-14T21:25:38 *** xiangfu has quit IRC (Ping timeout: 240 seconds) 2012-05-14T21:44:16 *** xiangfu has joined #rtems 2012-05-14T22:02:38 *** xiangfu has quit IRC (Ping timeout: 244 seconds) 2012-05-14T22:05:44 *** arvind_khadri has joined #rtems 2012-05-14T22:16:36 *** xiangfu has joined #rtems 2012-05-14T22:28:57 *** arvind_khadri has quit IRC (Ping timeout: 252 seconds) 2012-05-15T00:18:16 *** arvind_khadri has joined #rtems 2012-05-15T00:18:24 *** arvind_khadri has joined #rtems 2012-05-15T00:33:36 *** xiangfu has quit IRC (Quit: Leaving) 2012-05-15T01:33:45 *** sebhub has joined #rtems 2012-05-15T01:34:14 good morning 2012-05-15T04:16:55 *** xiangfu has joined #rtems 2012-05-15T05:12:36 *** xiangfu has quit IRC (Ping timeout: 272 seconds) 2012-05-15T05:41:15 *** xiangfu has joined #rtems 2012-05-15T06:00:21 *** cdcs has joined #rtems 2012-05-15T06:03:36 *** xiangfu has quit IRC (Remote host closed the connection) 2012-05-15T09:00:51 *** weiY has joined #rtems 2012-05-15T09:17:40 *** DrJoel has joined #rtems 2012-05-15T09:17:40 *** DrJoel has joined #rtems 2012-05-15T09:17:40 *** ChanServ sets mode: +o DrJoel 2012-05-15T09:18:14 *** gedare has joined #rtems 2012-05-15T10:14:21 Hi gedare, good night 2012-05-15T10:14:27 hey weiY 2012-05-15T10:14:39 how about the github for rtems 2012-05-15T10:15:26 *** cdcs has quit IRC (Ping timeout: 245 seconds) 2012-05-15T10:15:30 the gsoc student will use github to clone and pull source code ? 2012-05-15T10:15:52 we have the account set-up, verm__ was working on some scripts for automatic updating from the master rtems.git -- we haven't added any students/mentors yet to the project, I think we will do that soon 2012-05-15T10:15:58 weiY: that is the idea yes 2012-05-15T10:16:10 weiY: i am a github newbie myself so learning while doing. :) 2012-05-15T10:16:52 yeah, github is a good idea for student to use. i think i will hear good new soon 2012-05-15T10:18:25 recently i make a report for atomic api design and send it to chris, but it seems chris is busy recently. so could i send to you and could you give me some advice about this? 2012-05-15T10:19:00 yes. 2012-05-15T10:19:36 you can email it to me 2012-05-15T10:20:00 your mail is gedare@rtems.org, right? 2012-05-15T10:20:04 yes 2012-05-15T10:20:14 i have sent to you 2012-05-15T10:20:19 okay 2012-05-15T10:20:51 thank you gedare, haha 2012-05-15T10:21:35 this could use a little more introduction 2012-05-15T10:22:01 about what? 2012-05-15T10:22:46 well, i think you should send this to rtems-devel after getting a little feedback 2012-05-15T10:23:13 so you should include a little more background just a couple sentences about C1X atomics and freebsd atomics, what is _explicit, and what is memory_order_seq_cst. 2012-05-15T10:23:30 ok, i will make a little modify and sent it to rtems devel 2012-05-15T10:23:34 it looks like you are matching the C1X atomics with the freebsd ones? 2012-05-15T10:24:37 yeah, in the past email sent to chris i have add some reference about the background, but i can send to you . so i will add this part and send it to devel 2012-05-15T10:25:14 if you have a blog then i think matching the C1X with freebsd atomics might be a nice post to write. 2012-05-15T10:25:37 you should also explain the differences between memory ordering requirements. 2012-05-15T10:27:00 i think you should refer to both netbsd and freebsd for your implementations, and design an API that covers most of the C1X capabilities 2012-05-15T10:27:57 weiY: note the distinction between an interface and an implementation 2012-05-15T10:28:15 yeah, the blog is a good idea, but i am a newbie of blog. i think i will write a blog. 2012-05-15T10:28:34 implementation will be a mix of freebsd- and netbsd-based code, and eventually C1X atomics 2012-05-15T10:28:42 interface will be something you must design 2012-05-15T10:28:49 yes, now i am foucs on the interface not implementation 2012-05-15T10:30:44 i think the goal of atomic interface design is compatible with C1X atomic ASAP, but if the implementation is diffcult for this compatible i can make a comprimise 2012-05-15T10:31:04 weiY: I suggest you work on an RTEMS internal interface for atomic operations. the important things to consider are the names of functions and what arguments/return values should be 2012-05-15T10:32:04 the interface should provide the useful primitives needed for synchronization algorithms, and is likely to mimic some *bsd or C1X atomic interface 2012-05-15T10:32:08 but will be "rtems" in flavor 2012-05-15T10:33:27 Since we ultimately hope to use language-based atomics I think the interface can be in the cpukit, but some of the implementations may need to be in libcpu since you may need to use cpu-specific features and inline assembly 2012-05-15T10:33:52 ok i understand your mean. i will consider the names of functions and arguments/return values "rtems" flavor 2012-05-15T10:35:09 yeah, this part will be the next step about the implementations 2012-05-15T10:36:00 ok. for internal interfaces the naming convention generally follows a style 2012-05-15T10:36:18 for functions that style is _API_Package_name_Method_name(...) 2012-05-15T10:36:32 where _API often is omitted and we just use _Package_name_Method_name(...) 2012-05-15T10:37:13 ok, where can i find the API name style documents? 2012-05-15T10:37:15 for structures we drop the _ so it is just typedef struct {...} Package_name_Structure_name; 2012-05-15T10:37:42 or i can refer the RTEMS source code 2012-05-15T10:37:43 i don't think it clearly is documented anywhere. DrJoel was working on a style guide but it is not ready for release and I don't know if it addresses this 2012-05-15T10:38:07 refer to the source code. the internal API mostly is seen in cpukit/score/include/rtems/score/*.h 2012-05-15T10:39:06 ok, when i design the API i can stduy those functions definitions 2012-05-15T10:39:07 * DrJoel hasn't added that rule to the style guide yet. :( 2012-05-15T10:39:30 hi DrJoel 2012-05-15T10:39:49 Think C++ packaging or Ada.. Package_long_name::method_xxx in C++ would be Package_long_name_Method_xxx in RTEMS. 2012-05-15T10:40:08 The _ is a package/method separator like the :: C++ or . in Ada 2012-05-15T10:40:23 Also private to RTEMS packages start with _ to keep them out of the public name space 2012-05-15T10:40:26 weiY: Yes. the score code is probably the most well-maintained code for internal style guidelines. 2012-05-15T10:41:31 hey gedare.. working on blog post on the "build issue" thread.. I am about done. Will you be able to read it soon? I don't want to say anything questionable, untrue, or use loaded language 2012-05-15T10:41:40 drjoel, good advice 2012-05-15T10:41:50 DrJoel: send me a draft 2012-05-15T10:42:22 It is on blogger and uses their formatting. I don't know if it will survive cut and paste 2012-05-15T10:42:33 I am willing to publish it and make edits 2012-05-15T10:42:48 either way. 2012-05-15T10:42:56 you can "preview" and copy-paste also that sould work. 2012-05-15T10:43:43 ric claus had some good negative feedback about the build system if you noticed. 2012-05-15T10:46:35 DrJoel: re the style guide, since you worked on that using asciidoc, do you want to propose replacing/augmenting the manuals with asciidoc instead of texinfo? 2012-05-15T10:46:51 or should the style guide be imported to a proper rtems manual 2012-05-15T10:47:06 I don't want to replace texinfo with asciidoc. asciidoc is good for short things 2012-05-15T10:47:16 It needs to be converted to texinfo 2012-05-15T10:47:20 ok 2012-05-15T10:47:33 I wouldn't mind seeing BSP specific notes in asciidoc but that's a VERY low priority 2012-05-15T10:59:59 *** Hesham has joined #rtems 2012-05-15T11:00:34 hi gedare 2012-05-15T11:00:39 hey hesham 2012-05-15T11:00:40 how are you ? 2012-05-15T11:00:44 good. you? 2012-05-15T11:00:47 good 2012-05-15T11:01:03 would we begin now ? 2012-05-15T11:01:27 uyes 2012-05-15T11:01:32 ok 2012-05-15T11:01:46 i've shared an Arena concept doc 2012-05-15T11:01:50 did you see it ? 2012-05-15T11:01:56 yes i saw it 2012-05-15T11:02:05 any comments ? 2012-05-15T11:03:00 it is a good listing of the high-level ideas 2012-05-15T11:03:29 you should clarify what are the possible memory allocators 2012-05-15T11:04:07 rbheap , heap 2012-05-15T11:04:20 i asked about workspace_allocate ? 2012-05-15T11:04:26 what is it ? 2012-05-15T11:04:34 that uses heap 2012-05-15T11:04:54 why not using heap directly ? 2012-05-15T11:04:55 depending on the BSP it either shares memory with the normal program heap or provides a separate heap-space 2012-05-15T11:05:18 aha 2012-05-15T11:05:48 workspace is where RTEMS kernel allocates dynamically, mostly for objects 2012-05-15T11:06:10 now i am thinking of 1:1 relation between allocator and protection 2012-05-15T11:06:40 If we want to allocate a memory of specific size 2012-05-15T11:06:46 *** dr__hous1 has joined #rtems 2012-05-15T11:06:55 that would return a pointer to that memory area 2012-05-15T11:07:11 then we use this pointer and the size of that area 2012-05-15T11:07:23 and create a domain for it 2012-05-15T11:07:30 *** dr__house has quit IRC (*.net *.split) 2012-05-15T11:07:47 or say a region 2012-05-15T11:08:02 avoid saying region because of the overlap with region manager. 2012-05-15T11:08:19 that was a mistake in my refactoring and needs to be fixed 2012-05-15T11:08:31 the region struct is used in 2011 project to describe an area of memory 2012-05-15T11:08:37 aha 2012-05-15T11:08:39 it's ok 2012-05-15T11:08:42 *** dr__hous1 is now known as dr__house 2012-05-15T11:09:09 one thing i was thinking about 2012-05-15T11:09:13 *** dr__house has quit IRC (Changing host) 2012-05-15T11:09:13 *** dr__house has joined #rtems 2012-05-15T11:09:27 Which data structure should be used to keep Arenas ? 2012-05-15T11:10:17 what do you mean? 2012-05-15T11:10:56 *** sebhub has quit IRC (Ping timeout: 265 seconds) 2012-05-15T11:11:06 i mean to keep access to Arena_Control , we could put them into rbtree 2012-05-15T11:11:09 or chain 2012-05-15T11:12:02 i guess it depends on what we need to do with them.. 2012-05-15T11:12:19 or whether it is necessary at all 2012-05-15T11:12:39 to iterate / search for the arena control. 2012-05-15T11:13:27 we would create an Arena for code area and one for stack or make it general ? 2012-05-15T11:14:03 make it general. but how to "swap" arenas that are active is the key question I guess. 2012-05-15T11:15:06 why arenas needs to be swapped ? 2012-05-15T11:15:50 well, you will have to provide a function to "walk" the arenas belonging to a task and unload/load protection domains for each one. 2012-05-15T11:16:00 eventually that is a requirement 2012-05-15T11:16:14 aha 2012-05-15T11:16:33 but initially we will just provide some global uses / no memory protection 2012-05-15T11:16:40 and evolve from there 2012-05-15T11:16:45 just keep in mind how arena will be used 2012-05-15T11:16:56 so that you don't overdesign / constrain yourself for later 2012-05-15T11:17:39 *** weiY has quit IRC (Ping timeout: 245 seconds) 2012-05-15T11:17:40 part of a task space area may be located in arena 2012-05-15T11:17:42 one thought I just had is that the Arena_Control might rely on one allocator to setup an Arena, and then provide an allocator to the application in order for the app to allocate memory from the Arena. 2012-05-15T11:18:27 or rely on the user to provide the blob of memory to the Arena, and the arena exposes an allocator for that blob 2012-05-15T11:18:47 the second approach is i think what we kind of decided on and seems simpler / better.. 2012-05-15T11:18:51 i was just to ask about that 2012-05-15T11:19:09 i'm confused when creating An arena 2012-05-15T11:19:26 we would create a heap area 2012-05-15T11:19:38 or allocate a block from a given heap area 2012-05-15T11:19:55 ? 2012-05-15T11:20:14 let's assume the arena is given a blob of memory. 2012-05-15T11:20:25 during creation 2012-05-15T11:20:27 by the user 2012-05-15T11:20:33 how the user got that memory we don't care 2012-05-15T11:21:01 so part of the API for creating an arena will include a specification of a memory blob..start address, size 2012-05-15T11:21:06 so the user would just pass a pointer to heap area when creating an arena 2012-05-15T11:21:07 ? 2012-05-15T11:21:16 that is one valid way to do it. 2012-05-15T11:21:39 void *p = malloc(1024); rtems_arena_create(..., p, 1024, ... ); 2012-05-15T11:21:55 there are other ways the user might come up with the memory blob as well. 2012-05-15T11:22:15 it doesn't really matter. what matters is that the arena gets the starting address and the size 2012-05-15T11:22:23 from there the arena is "in control" of that memory 2012-05-15T11:22:41 it is an error for the user or anyone else to access that memory without going through the arena's allocator first 2012-05-15T11:22:58 i think that makes good sense 2012-05-15T11:23:03 what do you think? 2012-05-15T11:23:04 yeah me too 2012-05-15T11:23:18 ok. so the arena can implement an allocator... 2012-05-15T11:23:32 and if protection bits are requested it can check if it can enforce the protection 2012-05-15T11:24:01 based on the allocator, size of memory blob, and hw capabilities 2012-05-15T11:24:18 the arena then would be used as a memory pool that could be partially/fully protected ? 2012-05-15T11:24:21 (some hw may require memory blob to be at least a "Page" in size, and aligned to page size) 2012-05-15T11:24:27 i think so yeah 2012-05-15T11:24:47 i'm not sure yet if we need to bake in the allocator or not 2012-05-15T11:25:13 so there would be allocate_arena 2012-05-15T11:25:27 it would take a size as a parameter 2012-05-15T11:25:43 and say a protection flag 2012-05-15T11:25:44 we should determine if it is necessary / good to have the arena allocate memory from the blob for application tasks, or if the arena should just allow tasks to access the entire blob. 2012-05-15T11:26:29 both :/ 2012-05-15T11:26:51 the second approach may be used to implement a shared Arena 2012-05-15T11:27:23 and the first to create an Arena for all code tasks' code segments or stacks 2012-05-15T11:27:39 or the task need just a protected area of memory 2012-05-15T11:28:02 work through the options and make notes about both approaches. 2012-05-15T11:28:13 advantages / disadvantages, use cases 2012-05-15T11:28:54 ok 2012-05-15T11:29:44 but if we allocated an area from Arena blob 2012-05-15T11:29:59 , we need to apply memory protection to it 2012-05-15T11:30:27 i think the behavior would be the protection applies to the entire blob 2012-05-15T11:30:28 we would use region ( struct ) and attach it to a domain 2012-05-15T11:30:32 or what you think ? 2012-05-15T11:30:33 regardless of whether you use an allocator or not 2012-05-15T11:30:57 it seems hard to do it otherwise 2012-05-15T11:30:57 the entire blob 2012-05-15T11:31:04 that would be simpler 2012-05-15T11:32:08 but if a task created an arena 2012-05-15T11:32:27 we want this task only to access that arena 2012-05-15T11:32:56 how this could be done 2012-05-15T11:32:57 ? 2012-05-15T11:33:43 it could be done be activating/deactivating page tables related to a task , that what i know 2012-05-15T11:34:14 yeah.. 2012-05-15T11:34:39 in the case of the Arena , how would it be ? 2012-05-15T11:34:48 activating/deactivating Arenas ? 2012-05-15T11:35:38 we'll have to figure out how to deal with that. at any rate there should be some function to activate/deactivate an arena 2012-05-15T11:35:54 we can later figure out what entity invokes that function, and how it knows to do that 2012-05-15T11:36:18 ok 2012-05-15T11:36:56 Also another question , in 2011 project there was the domain that consists of regions. 2012-05-15T11:37:23 why not just implementing page tables ? 2012-05-15T11:37:51 in other words , what is the benefit of the domain ? 2012-05-15T11:39:38 separates memory attributes from page tables. page tables are a hw-dependency -- we cannot rely on the hw to support "pages" 2012-05-15T11:40:14 at least that is how I was thinking at the time. I did not dig very deeply 2012-05-15T11:40:55 let me see if i understand 2012-05-15T11:41:22 there might be a domain for code segments that has only --x attrib ? 2012-05-15T11:41:49 and all code segments are part of this domain 2012-05-15T11:43:28 depends on if all code segments should always be executable at all times. 2012-05-15T11:43:51 *** sevikkk has quit IRC (Quit: Leaving.) 2012-05-15T11:44:09 a domain is intended to implement the set of memory blobs and attributes that a particular task should have 2012-05-15T11:44:35 so in our case, a domain is the set of arenas that a task "belongs to" 2012-05-15T11:44:43 this set of memory blobs are the regions struct ? 2012-05-15T11:45:02 you can walk a domain activating/deactivating arenas in order to enforce attributes. 2012-05-15T11:45:10 yeah 2012-05-15T11:45:19 that is how to map the existing implementation to our current thinking. 2012-05-15T11:46:12 coding will begins at 21st may right ? 2012-05-15T11:46:18 yep 2012-05-15T11:46:42 what part should i begin with , and what should i do till this time ? 2012-05-15T11:49:06 http://rtemsramblings.blogspot.com/2012/05/rtems-build-system-ruminations.html is up. Feedback desired 2012-05-15T11:49:16 * DrJoel wants lunch 2012-05-15T11:50:09 My wife Michele shared this cool version of Rocket Man by Elton John with ESA footage to celebrate the 40th anniversary of the release 2012-05-15T11:50:26 http://youtu.be/aa5RwoFDCxc 2012-05-15T11:51:10 Hesham: we can work out a plan for coding on Friday 2012-05-15T11:51:47 that would be good 2012-05-15T11:51:52 Hesham: work through the high-level concepts and interface for arena, and write a short draft for emailing to the rtems-devel ml 2012-05-15T11:52:09 okay 2012-05-15T11:52:44 Hesham: as for coding, we'll either start directly with arenas if the design is acceptable, or you maybe can do an initial attempt at a supercore allocator 2012-05-15T11:52:58 or bring the 2011 work up-to-date 2012-05-15T11:54:29 i will try to get the draft / design done by thursday and share it with you , then we will discuss about the coding plan on Friday 2012-05-15T11:55:46 ok sounds good 2012-05-15T11:55:49 anything else? 2012-05-15T11:56:05 i think that's enough 2012-05-15T11:58:50 DrJoel: reading. 2012-05-15T12:03:22 DrJoel: you should get rid of deprecated flags from your configure line ;) like --disable-itron 2012-05-15T12:06:22 They are generated by the bit_rtems script. If rtems-testing had branches in it, it would be OK. LOL 2012-05-15T12:07:37 I removed that from the post. Reload 2012-05-15T12:09:14 DrJoel: "This assumes a fresh checkout and a single bootstrap. If you did that for each BSP, the build would take over 24 hours without executing any tests." 2012-05-15T12:09:25 *** arvind_khadri has quit IRC (Ping timeout: 265 seconds) 2012-05-15T12:09:27 I don't see why you would do that. one fresh checkout + bootstrap is sufficient. 2012-05-15T12:09:59 however the argument about different ./configure options is important. :) 2012-05-15T12:10:08 don't forget debug flags 2012-05-15T12:10:23 which have consistently been a source of build problems for us. 2012-05-15T12:11:09 DrJoel: otherwise it looks quite nice. 2012-05-15T12:11:47 * DrJoel agrees on the one fresh checkout + bootstrap. Just making a point that if you don't optimize that, the problem gets worse. 2012-05-15T12:12:44 Actually I have a build on rtbf64a that is over 24 hours now. bit_ALL script takes that long 2012-05-15T12:13:43 *** Hesham has left #rtems 2012-05-15T12:23:41 *** QingPei has left #rtems 2012-05-15T13:01:46 *** Hesham has joined #rtems 2012-05-15T13:02:39 *** gedare has quit IRC (Quit: Ex-Chat) 2012-05-15T13:05:37 dr__house, did you graduate or just post pictures? 2012-05-15T13:08:42 *** Hesham has quit IRC (Ping timeout: 255 seconds) 2012-05-15T13:20:16 DrJoel: I just graduated :) 2012-05-15T13:20:31 *** Hesham has joined #rtems 2012-05-15T13:24:53 We thought so. Congratulations! 2012-05-15T13:25:58 :) 2012-05-15T13:25:59 thanks 2012-05-15T13:33:40 *** Hesham has quit IRC (Ping timeout: 260 seconds) 2012-05-15T13:35:08 Are you staying in the States? 2012-05-15T13:48:05 *** Hesham has joined #rtems 2012-05-15T13:49:09 DrJoel: yes, I managed to get a job and that is here in Philly too. So as of now staying here :) 2012-05-15T13:49:43 :) 2012-05-15T14:06:46 *** Hesham has quit IRC (Ping timeout: 252 seconds) 2012-05-15T14:25:29 *** Hesham has joined #rtems 2012-05-15T14:26:04 *** Hesham has left #rtems 2012-05-15T14:55:00 *** sevikkk has joined #rtems 2012-05-15T15:37:53 *** gedare has joined #rtems 2012-05-15T15:59:14 *** sevikkk has quit IRC (Ping timeout: 272 seconds) 2012-05-15T16:01:05 *** sevikkk has joined #rtems 2012-05-15T16:30:31 *** DrJoel has quit IRC (Quit: Leaving) 2012-05-15T17:13:37 *** L84Supper has quit IRC (Quit: puff of smoke) 2012-05-15T17:14:04 *** A0Sheds has joined #rtems 2012-05-15T18:03:01 *** gedare has quit IRC (Quit: Leaving) 2012-05-15T20:47:01 *** xiangfu has joined #rtems 2012-05-15T20:54:28 *** antgreen has joined #rtems 2012-05-15T23:09:38 *** xiangfu has quit IRC (Ping timeout: 240 seconds) 2012-05-15T23:15:04 *** kristianpaul has quit IRC (Read error: Connection reset by peer) 2012-05-15T23:20:36 *** kristianpaul has joined #rtems 2012-05-16T01:12:17 *** arvind_khadri has joined #rtems 2012-05-16T01:24:17 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-05-16T01:24:57 *** arvind_khadri has joined #rtems 2012-05-16T01:24:57 *** arvind_khadri has joined #rtems 2012-05-16T01:26:08 *** arvind_khadri has joined #rtems 2012-05-16T01:32:19 *** sebhub has joined #rtems 2012-05-16T01:32:41 good monring 2012-05-16T03:09:21 *** arvind_k has joined #rtems 2012-05-16T03:09:22 *** arvind_khadri has quit IRC (Read error: Connection reset by peer) 2012-05-16T04:31:26 *** xiangfu has joined #rtems 2012-05-16T05:19:28 *** weiY has joined #rtems 2012-05-16T06:17:34 *** arvind__ has joined #rtems 2012-05-16T06:20:27 *** arvind_k has quit IRC (Read error: Connection reset by peer) 2012-05-16T06:33:55 *** antgreen has quit IRC (Remote host closed the connection) 2012-05-16T07:22:38 *** xiangfu has quit IRC (Ping timeout: 272 seconds) 2012-05-16T07:23:58 *** xiangfu has joined #rtems 2012-05-16T07:41:26 *** gedare has joined #rtems 2012-05-16T07:46:18 hi 2012-05-16T07:50:58 hi 2012-05-16T07:54:14 *** xiangfu has quit IRC (Ping timeout: 248 seconds) 2012-05-16T08:27:15 *** xiangfu has joined #rtems 2012-05-16T08:46:24 *** xiangfu has quit IRC (Ping timeout: 245 seconds) 2012-05-16T08:47:37 *** xiangfu has joined #rtems 2012-05-16T08:57:24 *** xiangfu has quit IRC (Ping timeout: 260 seconds) 2012-05-16T09:00:15 *** weiY has quit IRC (Ping timeout: 276 seconds) 2012-05-16T09:03:15 *** xiangfu has joined #rtems 2012-05-16T09:04:42 *** weiY has joined #rtems 2012-05-16T09:58:35 *** DrJoel has joined #rtems 2012-05-16T09:58:35 *** DrJoel has joined #rtems 2012-05-16T09:58:35 *** ChanServ sets mode: +o DrJoel 2012-05-16T10:03:17 hey drjoel. 2012-05-16T10:03:49 good morning gedare 2012-05-16T10:03:55 i don't think it is a good idea to put generated files under vc. it dilutes things 2012-05-16T10:04:05 poking the bear.. :) 2012-05-16T10:04:12 yeah i suppose so. 2012-05-16T10:04:38 It is pretty clear that the build system has problems, is rough with git, etc. 2012-05-16T10:05:08 what is the status with the waf experiments? 2012-05-16T10:05:08 At least moving the autotools generated files into git would handle the timestamp OK. But it massively bloats the repo 2012-05-16T10:05:12 sucks sucks sucks 2012-05-16T10:05:38 the --enable-maintainer-mode works most of the time 2012-05-16T10:05:41 Chris and Amar were working through the BSPs and still needed to address the tests last I heard 2012-05-16T10:06:22 most of the time it does. Adding directories with configure.ac or Makefile.am is the one big exception.. but most of the time, it works. Switching branches can confuse the hell out of it though 2012-05-16T10:07:08 verm__ wants to make sure everything is "set" before unleashing it on the world 2012-05-16T10:07:27 my opinion is that we should wait for 4.12 for that change 2012-05-16T10:07:50 that depends on the rtems 4.11 release date ;-) 2012-05-16T10:07:54 yeas 2012-05-16T10:08:18 oh, i have to hurry to the train, cu 2012-05-16T10:08:54 *** xiangfu has quit IRC (Quit: Leaving) 2012-05-16T10:09:58 later sebhub. 2012-05-16T10:12:44 *** sebhub has quit IRC (Ping timeout: 244 seconds) 2012-05-16T10:23:40 At Chris' request, I added a new section to http://rtemsramblings.blogspot.com/2012/05/rtems-build-system-ruminations.html 2012-05-16T10:24:20 The new section is "How much of the make time is actually configuration?" 2012-05-16T10:24:33 Just a couple of paragraphs.. can you see if it looks OK? 2012-05-16T10:33:48 *** weiY has quit IRC (Ping timeout: 245 seconds) 2012-05-16T10:44:13 reading 2012-05-16T10:46:19 DrJoel: looks good 2012-05-16T10:46:38 pretty amazing how the time breaks down! 2012-05-16T10:46:44 yeah. 2012-05-16T10:47:03 i'm brushing up on waf stuff. i have a feeling it will come and i should be ready for it ;) 2012-05-16T10:47:56 waf's success depends on reducing the 50% configuration time. Since make -j3 used my dual cores 2/3 of the time, there isn't as much to gain there 2012-05-16T10:48:21 It may be interesting to see the difference on a quad-core.. I doubt the core utilization ratio is as good 2012-05-16T10:48:52 yes. my understanding is that it sends configure time down quite low. i had built rtems with waf once but i lost the package and can't find it 2012-05-16T10:49:23 i use a quad-core 2012-05-16T11:09:11 I updated to add quad-core configure and build times. Same basic results. 2/3 CPU utilization of 4 cores during compilation phase 2012-05-16T11:09:20 configure part is very single threaded 2012-05-16T11:09:29 but stranglely much faster on the build farm machine 2012-05-16T11:11:01 fyi i am testing daniel's patch now 2012-05-16T11:11:03 k 2012-05-16T11:33:36 *** QingPei has joined #rtems 2012-05-16T11:40:20 *** arvind__ has quit IRC (Quit: Leaving) 2012-05-16T11:40:29 *** arvind_khadri has joined #rtems 2012-05-16T11:40:29 *** arvind_khadri has joined #rtems 2012-05-16T11:55:34 *** Hesham has joined #rtems 2012-05-16T12:25:42 gedare, does his patch apply for you? I just noticed it isn't applying.. after I built all the BSPs :( 2012-05-16T12:25:54 haven't checked. one sec 2012-05-16T12:29:26 DrJoel: it applied without any errors to me. 2012-05-16T12:31:27 sigh... feel free to commit it and post to the list. I just put the ok on the list as a CYA move for you. :) 2012-05-16T12:31:50 ok i did not try to build or anything. 2012-05-16T12:32:23 odd that yours failed. 2012-05-16T12:32:29 did you checkout from an updated master? 2012-05-16T12:32:36 checkout/branch. 2012-05-16T12:33:20 $ git apply "/tmp/[PATCH 35 v4] LEON: moved register definitions into grlib header file.txt" 2012-05-16T12:33:21 error: patch failed: c/src/lib/libbsp/sparc/leon2/preinstall.am:161 2012-05-16T12:33:21 error: c/src/lib/libbsp/sparc/leon2/preinstall.am: patch does not apply 2012-05-16T12:33:21 error: patch failed: c/src/lib/libbsp/sparc/leon3/leon_smc91111/leon_smc91111.c:34 2012-05-16T12:33:21 error: c/src/lib/libbsp/sparc/leon3/leon_smc91111/leon_smc91111.c: patch does not apply 2012-05-16T12:33:28 yes 2012-05-16T12:34:09 thanks for committing that 2012-05-16T12:34:11 how odd. 2012-05-16T12:34:27 starting build now 2012-05-16T12:34:42 I did notice when I did a save as in Thunderbird, it broke it 2012-05-16T12:34:45 verm__: remote: 2: update github 2012-05-16T12:34:46 but it doesn't show. i guess the hook is running at least... 2012-05-16T12:34:52 +$(PROJECT_INCLUDE)/grlib.h: ../../sparc/shared/include/grlib.h $(PROJECT_I= 2012-05-16T12:34:53 NCLUDE)/$(dirstamp) 2012-05-16T12:35:05 ahh. probably the patch was mangled. 2012-05-16T12:35:12 SO I saved as a .txt file and that looked ok but didn't apply 2012-05-16T12:35:17 weird. I haven't had this before 2012-05-16T12:35:36 https://code.google.com/p/fetch-flagged-email/ ;) 2012-05-16T12:36:04 probably something new in thunderbird. 2012-05-16T12:36:09 Your gmail fetcher? 2012-05-16T12:36:18 yep.. but I don't rememebr it upgrading recently 2012-05-16T12:36:34 April 24 is the date on the binary 2012-05-16T12:43:33 *** QingPei1 has joined #rtems 2012-05-16T12:46:54 *** QingPei has quit IRC (Ping timeout: 272 seconds) 2012-05-16T12:46:57 gedare: please consider reviewing Arena Concept doc and commenting on it , i think it may be the initial design that would be sent to rtems-devel , So you may tell me what it also needs to be finally submitted to rtems-devel. 2012-05-16T12:47:01 https://docs.google.com/document/d/124zgmDmwx38l_-0hNYoEImre_t4UZdpVH2vWz__njxs/edit 2012-05-16T13:00:14 *** QingPei1 has left #rtems 2012-05-16T13:19:02 hi hesham. taking a look. 2012-05-16T13:19:26 DrJoel: but when did you update it, and when was the last time you dl'ed files with it ;) 2012-05-16T13:20:00 hesham: you need a little more introductory / background material 2012-05-16T13:20:06 no one will know what you are talking about 2012-05-16T13:20:29 gedare, I really don't know. Probably the last batch of his patches which is likely before 24 April 2012-05-16T13:21:47 gedare: I will add the project description and other details at the top. 2012-05-16T13:22:21 ok but don't go overboard just a summary 2012-05-16T13:22:46 ok 2012-05-16T13:24:55 start with high-level requirements 2012-05-16T13:25:04 what is needed, who will use it, so on 2012-05-16T13:25:25 that will provide some context for your design 2012-05-16T13:26:24 gedare, sparc builds look OK 2012-05-16T13:26:36 DrJoel: great 2012-05-16T13:26:48 gedare: that should be in the introduction part ? 2012-05-16T13:26:59 I was a bit surprised at the sparc having a near dupe of pci.h. :( 2012-05-16T13:26:59 Hesham: yes. 2012-05-16T13:30:36 Hesham: see my comment. the "second approach" takes a "dont-care" approach 2012-05-16T13:31:04 ok 2012-05-16T13:31:27 really the key question to consider is whether an arena should have an allocator "baked in" or not. 2012-05-16T13:32:12 Hesham: you swap 1 and 2 in your gdoc 2012-05-16T13:32:58 oh i see you kind of have a few confusing ideas 2012-05-16T13:33:07 2 approachs for creating, 2 approaches for using 2012-05-16T13:33:15 gedare: i will delete disadvantages of second approach which you commented on 2012-05-16T13:33:19 yeah 2012-05-16T13:33:45 i think you should simplify this 2012-05-16T13:34:18 i think your "two approaches of creating an arena" are the same thing 2012-05-16T13:34:34 How ? 2012-05-16T13:34:45 and correspond with having an arena that implements an allocator. the question is whether the allocator applies protection or not to the allocated chunks 2012-05-16T13:35:06 unless i missed something 2012-05-16T13:35:33 we did not agree how the arena blob will be allocated 2012-05-16T13:35:44 whether be the user who calls arena_create 2012-05-16T13:36:01 or by areana_create() itself 2012-05-16T13:36:35 ah, I see. 2012-05-16T13:36:57 So i made the two approaches for creating Arena to be discussed 2012-05-16T13:37:06 ok somehow you need to make that more clear. 2012-05-16T13:38:46 i do not have the idea how to make it more clear for now , but i will think about it later. 2012-05-16T13:39:42 ok. a suggestion is to focus on how/when the arena gets its memory pool. 2012-05-16T13:40:24 i.e. where the memory comes from. 2012-05-16T13:40:53 it seems to me the two likely answers are: 1) the workspace, and 2) user-provided 2012-05-16T13:42:17 and then your other design parameter is how the arena manages its memory pool, i.e. whether it further provides allocation within the pool back to the application. 2012-05-16T13:45:38 that make sense 2012-05-16T13:46:12 then work through how your design will satisfy the requirements you list in the intro 2012-05-16T13:46:38 and we will get some feedback from there. 2012-05-16T13:49:39 gedare: i had added a simple introduction to Arena 2012-05-16T13:53:18 Hesham: ok and you still need to identify why we are introducing arenas. what is the problem to solve and what are the requirements for the solution 2012-05-16T13:54:18 i will work on it now 2012-05-16T14:43:17 *** gedare has quit IRC (Quit: Leaving) 2012-05-16T15:37:34 *** Hesham has left #rtems 2012-05-16T16:00:18 *** arvind_khadri has quit IRC (Ping timeout: 260 seconds) 2012-05-16T16:36:23 *** A0Sheds has quit IRC (Quit: puff of smoke) 2012-05-16T16:42:51 *** A0Sheds has joined #rtems 2012-05-16T16:43:53 *** L84Supper has joined #rtems 2012-05-16T17:54:14 *** DrJoel has quit IRC (Quit: Leaving) 2012-05-16T20:03:29 *** xiangfu has joined #rtems 2012-05-16T20:53:11 *** medivhc has joined #rtems 2012-05-16T21:11:02 *** xiangfu has quit IRC (Ping timeout: 272 seconds) 2012-05-16T21:17:39 *** xiangfu has joined #rtems 2012-05-16T21:23:04 *** xiangfu has quit IRC (Ping timeout: 272 seconds) 2012-05-16T21:25:20 *** xiangfu has joined #rtems 2012-05-16T21:25:42 *** xiangfu has joined #rtems 2012-05-16T21:59:31 *** medivhc_ has joined #rtems 2012-05-16T22:09:49 *** zw_yao has joined #rtems 2012-05-16T22:12:06 *** QingPei has joined #rtems 2012-05-16T22:23:19 *** medivhc has quit IRC (Remote host closed the connection) 2012-05-16T22:33:35 *** arvind_khadri has joined #rtems 2012-05-16T23:33:07 *** Deb has joined #rtems 2012-05-16T23:34:41 *** xiangfu has quit IRC (Quit: Leaving) 2012-05-16T23:39:00 *** Deb has quit IRC (Quit: This computer has gone to sleep) 2012-05-17T00:11:12 *** panzon has quit IRC (Ping timeout: 256 seconds) 2012-05-17T00:13:22 *** panzon has joined #rtems 2012-05-17T00:42:42 *** QingPei has quit IRC (Read error: Connection reset by peer) 2012-05-17T00:44:46 *** QingPei has joined #rtems 2012-05-17T01:24:19 *** zw_yao has quit IRC (Remote host closed the connection) 2012-05-17T01:37:36 *** zw_yao has joined #rtems 2012-05-17T01:43:21 *** xiangfu has joined #rtems 2012-05-17T02:44:04 *** zw_yao has quit IRC (Remote host closed the connection) 2012-05-17T02:52:22 *** zw_yao has joined #rtems 2012-05-17T03:31:52 *** xiangfu has quit IRC (Ping timeout: 252 seconds) 2012-05-17T03:41:55 *** cdcs has joined #rtems 2012-05-17T03:46:38 *** QingPei has quit IRC (Ping timeout: 244 seconds) 2012-05-17T03:50:02 *** Deb has joined #rtems 2012-05-17T04:00:44 *** QingPei has joined #rtems 2012-05-17T04:43:12 *** zw_yao has quit IRC (Quit: Leaving) 2012-05-17T04:46:10 *** zw_yao has joined #rtems 2012-05-17T04:46:42 *** zw_yao has joined #rtems 2012-05-17T05:53:45 *** alseh has joined #rtems 2012-05-17T05:57:49 *** alseh has left #rtems 2012-05-17T05:58:56 *** alseh has joined #rtems 2012-05-17T06:21:55 *** ppisa has joined #rtems 2012-05-17T06:42:03 *** Deb has quit IRC (Quit: This computer has gone to sleep) 2012-05-17T06:43:06 *** QingPei has quit IRC (Read error: Connection reset by peer) 2012-05-17T06:43:41 *** QingPei has joined #rtems 2012-05-17T06:57:05 *** medivhc has joined #rtems 2012-05-17T07:05:21 *** weiY has joined #rtems 2012-05-17T07:17:24 *** xiangfu has joined #rtems 2012-05-17T07:24:09 *** Deb has joined #rtems 2012-05-17T07:27:54 *** zw_yao has joined #rtems 2012-05-17T07:32:08 *** xiangfu has quit IRC (Ping timeout: 255 seconds) 2012-05-17T07:41:18 *** xiangfu has joined #rtems 2012-05-17T07:50:54 *** DrJoel has joined #rtems 2012-05-17T07:50:54 *** DrJoel has joined #rtems 2012-05-17T07:50:54 *** ChanServ sets mode: +o DrJoel 2012-05-17T07:50:56 *** DrJoel has quit IRC (Client Quit) 2012-05-17T07:51:05 *** DrJoel has joined #rtems 2012-05-17T07:51:05 *** DrJoel has joined #rtems 2012-05-17T07:51:05 *** ChanServ sets mode: +o DrJoel 2012-05-17T07:51:31 * DrJoel looks around.. 2012-05-17T07:59:40 I know it is early but here is what I am looking at for the format of a report from the GSOC IRC meeting. https://docs.google.com/document/d/1wuGL8Hb64XRTgqKlom6Nx8YxTbLnPiwAQUCBZr19QX0/edit 2012-05-17T08:00:13 My idea was to let folks edit it for their project, to show attendance, etc. and then lock it down and publish it. 2012-05-17T08:01:43 *** arvind_khadri has quit IRC (Ping timeout: 245 seconds) 2012-05-17T08:02:38 * tuxmaniac looks around and doesnt find DrJoel :P 2012-05-17T08:03:14 trust me.. even trying to lose weight, I am pretty easy to spot.. :) 2012-05-17T08:03:55 You have a Klout score of 44. :) 2012-05-17T08:05:20 DrJoel: is that good or bad? 2012-05-17T08:05:21 :P 2012-05-17T08:05:32 DrJoel: hahaha. that was the comment I knew was coming 2012-05-17T08:05:52 * tuxmaniac is looking at all the docs shared by Hesham. 2012-05-17T08:06:57 well I have a 52 but Michele has a 56. It is pretty good. 2012-05-17T08:07:35 http://www.wired.com/epicenter/2012/04/ff_klout/all/1 for more information. Interesting. 2012-05-17T08:09:01 I see folks in the status report. Does the format look OK? Suggestions appreciated. We will be stuck with this format all summer. :) 2012-05-17T08:11:17 If you are here early, feel free to add a a few sentences on your project status. Probably will make meeting go faster 2012-05-17T08:20:29 DrJoel: what about every report having a "DONE", "Next steps", "Red flags (if any)" 2012-05-17T08:20:34 *** weiY has quit IRC (Ping timeout: 244 seconds) 2012-05-17T08:21:07 *** jennifer has joined #rtems 2012-05-17T08:22:13 *** zw_yao has quit IRC (Remote host closed the connection) 2012-05-17T08:23:39 *** weiY has joined #rtems 2012-05-17T08:23:54 *** zw_yao has joined #rtems 2012-05-17T08:24:58 Hi joel, what content should we add to the report 2012-05-17T08:28:00 Can't open the file 2012-05-17T08:30:04 I found the reason. Network problem. Sorry. 2012-05-17T08:32:45 Deb.. you are in China, right? 2012-05-17T08:32:54 Yeah. 2012-05-17T08:33:20 It seems that I cannot establish secure connection with google doc server 2012-05-17T08:33:21 * cdcs says hello 2012-05-17T08:33:25 I recall at various times Chinese students had problems getting to certain Google sites. Those were "Great Firewall" issues. 2012-05-17T08:33:35 *** zw__yao has joined #rtems 2012-05-17T08:33:35 Yes. 2012-05-17T08:33:45 It seems to be the problem 2012-05-17T08:33:52 I am trying to find the proxy server 2012-05-17T08:33:55 Can you get there non-secure? 2012-05-17T08:34:36 It is google who requires secure connection 2012-05-17T08:34:53 So, I don't think I can use unsecure connection 2012-05-17T08:35:11 yean, you can use goagent in china to access the google doc 2012-05-17T08:36:19 *** zw_yao has quit IRC (Ping timeout: 252 seconds) 2012-05-17T08:38:09 Thank you. 2012-05-17T08:49:29 https://docs.google.com/document/d/1wuGL8Hb64XRTgqKlom6Nx8YxTbLnPiwAQUCBZr19QX0/edit is the Status report for the GSOC IRC meeting. It is currently world editable. Please add a few sentences on your project. When the meeting starts, hopefully everyone will add themselves as attendees 2012-05-17T08:57:31 *** zw__yao has quit IRC (Read error: Connection reset by peer) 2012-05-17T09:14:07 *** alseh has quit IRC (Remote host closed the connection) 2012-05-17T09:22:00 *** ppisa has quit IRC (Ping timeout: 256 seconds) 2012-05-17T09:32:39 *** alseh has joined #rtems 2012-05-17T09:33:03 *** xiangfu has quit IRC (Quit: Leaving) 2012-05-17T09:33:53 @weiY, do you have the goapent.zip file? I don't know when the download can be done. 2012-05-17T09:33:53 Deb: Error: "weiY," is not a valid command. 2012-05-17T09:34:27 weiY, do you have the goagent.zip file? I don't know when the download can be done? 2012-05-17T09:34:32 . 2012-05-17T09:42:32 yes 2012-05-17T09:42:46 i can send you, your email? 2012-05-17T09:43:38 *** QingPei has quit IRC (Read error: Connection reset by peer) 2012-05-17T09:44:03 *** QingPei has joined #rtems 2012-05-17T09:44:06 i use goagent and chrome to access all the web 2012-05-17T09:44:14 panx@seas.upenn.edu 2012-05-17T09:44:24 thank you very much 2012-05-17T09:47:40 *** gedare has joined #rtems 2012-05-17T09:48:23 https://docs.google.com/document/d/1wuGL8Hb64XRTgqKlom6Nx8YxTbLnPiwAQUCBZr19QX0/edit is the Status report for the GSOC IRC meeting. It is currently world editable. Please add a few sentences on your project. When the meeting starts, hopefully everyone will add themselves as attendees 2012-05-17T09:49:17 I can't access it 2012-05-17T09:49:19 *** weiY has quit IRC (Ping timeout: 245 seconds) 2012-05-17T09:49:51 Even it is world editable 2012-05-17T09:50:59 i didn't realize we were having the first meeting today! happenchance I logged on. 2012-05-17T09:51:22 * gedare is not good with dates and times these days 2012-05-17T09:51:56 i sent an email 2012-05-17T09:51:57 I didn't have much time on this project lately because I am on the travel all the time 2012-05-17T09:52:08 Thank you 2012-05-17T09:52:16 that's ok it is expected at this point that there would not be much time available 2012-05-17T09:55:19 How are you doing lately? 2012-05-17T09:55:48 *** weiY has joined #rtems 2012-05-17T09:55:51 I've tried to update the attendees and missing. I don't think I got it right but please check 2012-05-17T09:56:58 Deb: i'm doing well. very busy 2012-05-17T09:57:11 am here 2012-05-17T09:57:13 \o/ 2012-05-17T09:57:19 me too .o/ 2012-05-17T09:57:50 i am here 2012-05-17T09:57:57 https://docs.google.com/document/d/1wuGL8Hb64XRTgqKlom6Nx8YxTbLnPiwAQUCBZr19QX0/edit is the Status report for the GSOC IRC meeting. It is currently world editable. Please add a few sentences on your project. When the meeting starts, hopefully everyone will add themselves as attendees 2012-05-17T09:57:59 what is that \o/ and .o/? 2012-05-17T09:58:46 i thought they are a emoticons 2012-05-17T09:58:50 gedare: just emoticons to indicate raising hands 2012-05-17T09:58:50 Haha, that is smile, which is not complete typed, I guess 2012-05-17T09:58:59 If you are a student and haven't added a few sentences of status to that document, please do so. 2012-05-17T09:59:15 gedare: \o/ hand waving :D 2012-05-17T09:59:16 Joel, I am still trying to set up the proxy 2012-05-17T09:59:25 i have add some sentences about the project status 2012-05-17T09:59:31 ahh. 2012-05-17T09:59:33 hi deb 2012-05-17T09:59:41 have you got goagent 2012-05-17T09:59:48 not yet 2012-05-17T09:59:59 i can send you , what is your email 2012-05-17T10:00:07 Deb.. if you need to, just put it on pastebin or some such site and I will put it in. 2012-05-17T10:00:22 panx@seas.upenn.edu 2012-05-17T10:00:39 *** zw_yao has joined #rtems 2012-05-17T10:01:11 Deb: paste it somewhere where you can paste text :) Someone can update it. 2012-05-17T10:01:35 I think all students are listed as either attending or missing now. Best attempt at mentors 2012-05-17T10:02:25 Joel, Kevin isn't listed do you want me to add him to the missing? 2012-05-17T10:02:35 Deb: you are in UPenn?! 2012-05-17T10:02:44 please. Did I add 11 students? :( 2012-05-17T10:02:56 yes 2012-05-17T10:03:03 Deb: what school? what major? 2012-05-17T10:03:13 seas 2012-05-17T10:03:17 Deb: or are you a mentor, pardon me if so 2012-05-17T10:03:18 embs 2012-05-17T10:03:24 gedare, What's wrong with the student list? Jennifer is adding Kevin and that will make 11. :) 2012-05-17T10:03:25 Deb: wow! 2012-05-17T10:03:26 not mentor 2012-05-17T10:03:37 ? 2012-05-17T10:03:43 Deb: I just graduated from Penn 2012-05-17T10:03:46 like yesterday 2012-05-17T10:03:50 dr__house: Deb hehe 2012-05-17T10:03:55 and I majored in EMBS 2012-05-17T10:04:10 zw_yao, i think you edited the file first. How are things going? 2012-05-17T10:04:21 DrJoel: deng hengyi is listed twice 2012-05-17T10:04:36 surname / given name confusion :) 2012-05-17T10:04:42 Just fix it please. :) 2012-05-17T10:04:50 someone just did like a ninja 2012-05-17T10:05:02 haha 2012-05-17T10:05:08 Which year? 2012-05-17T10:05:10 DrJoel, I'm reading the doc of uthash this days. try to start the hash approach design. 2012-05-17T10:05:22 Is everyone's name correct? If not, then quietly fix it. 2012-05-17T10:05:25 Deb: this year. Graduated in 2012. Last monday :P 2012-05-17T10:05:33 Wow 2012-05-17T10:05:47 Deb: are you a second year student now>? 2012-05-17T10:05:52 Do I know you? May I ask what is your name 2012-05-17T10:05:53 Yes 2012-05-17T10:05:56 dr__house: Deb shhhh. meeting in progress 2012-05-17T10:05:57 You have been active asking questions. Any outstanding items we owe you? 2012-05-17T10:06:01 Deb: Santosh Vattam 2012-05-17T10:06:05 Deb: that's my name 2012-05-17T10:06:22 ********** Please yield the floor to zw_yao 2012-05-17T10:06:43 thanks, gedare. 2012-05-17T10:07:17 zw_yao: Sure: do you have any questions or comments or requests? 2012-05-17T10:07:53 (or complaints) 2012-05-17T10:08:28 I have problem of the schedule. since we have a schedule in the GSoC proposal, should I follow that exactly? 2012-05-17T10:08:51 zw_yao: scheduling on these projects is always loose. work with your mentor to revise it as you go 2012-05-17T10:09:01 Do what's technically best. 2012-05-17T10:09:09 OK. 2012-05-17T10:09:47 any other concerns? 2012-05-17T10:09:59 And about the design as Gedare sayed on GSoc-rtems ml. 2012-05-17T10:10:26 what kind of design should be completed before 5.23(code start)? 2012-05-17T10:10:49 we want to see a plan for what you will be coding before you actually start coding 2012-05-17T10:11:00 that way you don't go too far along and then realize something is wrong with your design 2012-05-17T10:11:18 *** WikL has joined #rtems 2012-05-17T10:11:22 Do you have a direction for the solution? Know where to implement it in tree, data structures touched, etc. As gedare said we want to make sure you are doing the right thing before you code 2012-05-17T10:11:32 https://docs.google.com/document/d/1wuGL8Hb64XRTgqKlom6Nx8YxTbLnPiwAQUCBZr19QX0/edit is the Status report for the GSOC IRC meeting. It is currently world editable. Please add a few sentences on your project. When the meeting starts, hopefully everyone will add themselves as attendees 2012-05-17T10:11:45 say my project, there seems two approaches now: the hash and red black tree, should I complete both approach before start coding? 2012-05-17T10:12:11 zw_yao: you should work with your mentors and the devel community to determine which approach will satisfy the requirements of your project 2012-05-17T10:12:39 and what will be the most acceptable solution for rtems 2012-05-17T10:13:36 OK, I think we couldn't talk too much about the technicall design on IRC? 2012-05-17T10:13:50 correct not at this meeting anyways. 2012-05-17T10:14:01 perhaps later, but that is better done on the mailing list or with your mentors. 2012-05-17T10:14:18 Late arrivals .. please put a few sentences in the doc I referenced 2012-05-17T10:14:20 what are your goals before the next meeting? 2012-05-17T10:14:29 I think I haven't decided which approach will be choosed. 2012-05-17T10:15:10 *** ppisa has joined #rtems 2012-05-17T10:15:12 I would suggest you sketch both approaches and send an email to the rtems-devel mailing list to solicit some feedback. 2012-05-17T10:15:19 WeiY you are up next. I don't have your name in the same order in the document in the attendee list and on your project section. Can you fix it? I don't want to insult you 2012-05-17T10:15:26 I think I'll complete the hash table approach design, and decide which is better(or both need to be implemented). 2012-05-17T10:15:29 +1 2012-05-17T10:16:07 zw_yao, probably best if not a lot of work. The "surface" where the algorithm meets the key code is quite small. 2012-05-17T10:16:10 zw_yao: make sure you get some feedback from the community before you spend too much of your time on it. if you understand the merits of both designs then you can help us to help you. 2012-05-17T10:16:34 hi joel , what i should fix ? 2012-05-17T10:16:49 the order of what i add 2012-05-17T10:16:51 weiY: the preferred ordering of your name 2012-05-17T10:16:57 weiY: surname given or given surname 2012-05-17T10:16:59 DrJoel, you mean both approach? 2012-05-17T10:17:22 yes... the modifications to *key* are very small compared to the size of the "algorithm implementation" 2012-05-17T10:17:26 gedare, I see. I should collect feedback as early as possible. 2012-05-17T10:17:33 ok it is right on the doc 2012-05-17T10:17:44 DrJoel, OK. 2012-05-17T10:17:51 zw_yao, ready to move on? 2012-05-17T10:17:58 weiY, you are up next 2012-05-17T10:18:00 zw_yao: yes. send rtems-devel an email about your project, the two approaches, and your initial ideas. 2012-05-17T10:18:09 Just another question about uthash. 2012-05-17T10:18:46 zw_yao: probably that should be handled by your mentor. :) 2012-05-17T10:19:22 OK, I see, I'll discuss the technique on ml. 2012-05-17T10:19:25 thanks zw_yao 2012-05-17T10:19:28 great 2012-05-17T10:19:29 That's all. 2012-05-17T10:19:42 weiY, highlight status and any issues or things that need to be done? 2012-05-17T10:19:47 done by mentors I mean 2012-05-17T10:19:50 ok, i have some question how to submit the source code, using github? and about the process of development, which i mean how to use git to development under rtems 2012-05-17T10:20:14 code submission will come later. there are two kinds of code submission 2012-05-17T10:20:26 1) submit to rtems: done through rtems-devel ml by mailing patches or through bugzilla 2012-05-17T10:20:33 2) submit to google: done at end of summer through melange 2012-05-17T10:21:16 using github: we have not gotten github support fully worked out. the primary rtems repo is not getting pulled properly. hopefully we get that fixed and get all students set up with accounts and instructions. it's new for me too 2012-05-17T10:21:19 you mean i should highlight in the rtems gsoc irc meeting? 2012-05-17T10:21:40 and on (1) please don't dump LOTS of code at the end if you can submit pieces as you go. If your project makes sense to do that way :) 2012-05-17T10:22:04 yes.. high points of status and your needs 2012-05-17T10:22:04 as for using git... read the Git wiki pages on RTEMS Wiki, and read the references those pages point to for learning more about git. ask questions on rtems-devel too. 2012-05-17T10:22:41 i hear openhatch has good tutorials on using git as well. so you might try that. 2012-05-17T10:23:26 weiY: I saw you have a good email thread on rtems-devel so hopefully your design question is being resolved there 2012-05-17T10:23:27 ok, i have read the wiki. but i want to know whether there are difference between the formal developmenter and student 2012-05-17T10:23:29 weiY, are the discussions on the list helping you come to a conclusion on technical direction and solution? 2012-05-17T10:23:45 weiY, no. You are a developer without write access 2012-05-17T10:24:36 weiY: primarily the formal developer has write access. I think we will have to update instructions with how to use github with rtems but we have to get it working before we can do even that ;) 2012-05-17T10:24:40 hi joel. until now the discussions on the list are not enough for me to do the decition 2012-05-17T10:26:03 so i think we should be more actively to discuss on the list 2012-05-17T10:26:29 weiY: I think you'll have to push a little harder on the list then, and also solidify a few solid options for how to move forward 2012-05-17T10:27:06 Ok gedare, i will work hard and more actively to push this 2012-05-17T10:27:27 you should try to have some idea of how your interface will appear, where the implementation will reside, and so on. 2012-05-17T10:27:36 weiY, sometimes direct questions followed by "ping you didn't answer me" or "I think I should do X" help. People often need prodding 2012-05-17T10:28:20 weiY, anything else? 2012-05-17T10:28:49 yeah, it seems no 2012-05-17T10:29:11 :) 2012-05-17T10:29:23 and now to bdbufs? 2012-05-17T10:29:34 ok 2012-05-17T10:30:07 weiY, thanks.. keep up the good work. Getting answers from a mailing list is often hard. Check out how effective even I am at getting replies. Don't be discouraged 2012-05-17T10:31:14 medivhc, how are things going? 2012-05-17T10:31:49 yeah, joel. i got it. i will learn more about how to work effective under the open source environment. until now i am very happy that i have got what i needed from the rtems people. they are very nice 2012-05-17T10:32:31 I need to gather statistics from bdbuf. But don't know where to start. 2012-05-17T10:32:35 weiY, did I pass along my hope/wish for a plugin model for algorithms along the way? Like Gedare did for the CPU scheduler 2012-05-17T10:32:50 what type of stats? 2012-05-17T10:33:25 kiwichris may be able to help if you can do it on qemu. He has some RTEMS shell scripts to do NFS copies of massive amounts of data 2012-05-17T10:33:40 joel, about my project atomic? 2012-05-17T10:34:01 hit rate I think 2012-05-17T10:34:06 DrJoel: it does not make much sense for weiY's project. 2012-05-17T10:34:14 DrJoel: weiY: we can discuss it later though. 2012-05-17T10:34:18 sorry that was for medivhc 2012-05-17T10:34:19 in my memory i hear about this plugin model but i am not sure 2012-05-17T10:34:27 doesn]atomic 2012-05-17T10:34:30 ahh. 2012-05-17T10:34:33 doesn't apply to atomic 2012-05-17T10:34:48 weiY, we can discuss that outside the meeting until it is clear 2012-05-17T10:34:50 ok, we can discuss after the meeting 2012-05-17T10:35:01 medivhc: have you sent your question to the rtems-devel mailing list? 2012-05-17T10:35:10 i thought i saw something about bdbuf stats but it wasn't very clear 2012-05-17T10:35:29 There is a mail, but not reply. 2012-05-17T10:35:32 ah it was buried in your project intro 2012-05-17T10:35:35 for stats, ask on the devel list for some test cases. I had to invent some for my PhD work. kiwichris has some scripts he used for RFS development which might be good enough to baseline with 2012-05-17T10:35:39 you should highlight your question as a single clear email 2012-05-17T10:35:47 +1 2012-05-17T10:37:01 I found the bdbuf show command. do we still use it ? 2012-05-17T10:37:03 gathering stats / writing a test case for that will help evaluate solutions. have you learned enough about the bdbuf? 2012-05-17T10:37:16 medivhc: that will have to be answered offline we cannot answer here 2012-05-17T10:37:33 medivhc, I am not even sure the right technical people are here to answer that 2012-05-17T10:38:04 medivhc, I added those as action items 2012-05-17T10:38:49 anything else before we move on.. (I have another meeting in 22 minutes) 2012-05-17T10:39:27 Deb, You are up next :) 2012-05-17T10:39:38 Hello 2012-05-17T10:39:44 I don't have much question 2012-05-17T10:39:53 medivhc: send your question to rtems-devel as just a question, and talk directly to your mentor if you haven't already 2012-05-17T10:40:12 I am still trying to figure out the test suites and the tests splition 2012-05-17T10:40:15 medivhc: by the next meeting you should have some idea bout the shape of bdbuf code and where you will be making changes. 2012-05-17T10:40:21 Deb, if you have questions on particular tests or are looking for a particular style of test, let me know. 2012-05-17T10:40:33 OK. 2012-05-17T10:40:35 I can tell you which are "error tests", "functional tests", etc. 2012-05-17T10:40:36 Ok I will work harder on this . 2012-05-17T10:40:58 That helps. 2012-05-17T10:41:03 Thank you. 2012-05-17T10:41:19 Deb: Do you have some ideas yet about how your project will "unfold"? 2012-05-17T10:41:24 For example, sp09 is a HUGE error test which needs to be broken into multiple tests. Too large for many targets. 2012-05-17T10:41:51 Deb: I know your mentor verm__ had some fairly specific tools he was interested in having you use 2012-05-17T10:42:38 Yeah, but I need to start with things that doesn't need with tools so specific yet 2012-05-17T10:42:52 Deb: ok so you have some idea where you will start and what code you will write/change? 2012-05-17T10:42:58 Deb, don't be shy about asking questions on the devel list. :) 2012-05-17T10:43:41 OK. I think I should stick with the proposal first 2012-05-17T10:43:56 And make adjustments if I need to. 2012-05-17T10:43:58 ok.. anythign we need to do for you Deb? 2012-05-17T10:44:45 alseh, you are up next 2012-05-17T10:44:53 Your knowledge on the layers of OS and testing suits helps 2012-05-17T10:44:53 ok Deb please send to rtems-devel the first steps you'll be taking so you can get some early feedback. thanks 2012-05-17T10:45:02 Thank you. 2012-05-17T10:45:18 Hello all 2012-05-17T10:45:34 As I wrote in the document, I was very busy lately 2012-05-17T10:45:55 up till now I built the toolkit to be able to start working on my main machine 2012-05-17T10:46:15 And just got a few pointers from my mentor on where I should look 2012-05-17T10:46:36 this weekend I will make a better view on the project 2012-05-17T10:47:01 although, I have a question on what version of NXlib I should get to work, git or release ? 2012-05-17T10:47:11 ok. that's fine gsoc isn't officially starting yet :) 2012-05-17T10:47:42 alseh, that's fine. Just let the ideas form. I know Kevin on TCP/IP is finishing finals so in similar boat 2012-05-17T10:47:42 ask on rtems-devel, but I would lean toward using releases and writing scripts to make it easy'ish to integrate future releases 2012-05-17T10:47:57 I know, but I was hoping to have a little more time 2012-05-17T10:48:20 if I use python for scripts, do you think it is ok ? 2012-05-17T10:48:25 alseh: yes. 2012-05-17T10:48:35 Hello alesh and all others I consider the starting versions is important and repository as well. 2012-05-17T10:48:59 If you need a new script. The framework for building is there now. If you replace that, you should move to waf (which is Python based) 2012-05-17T10:49:15 But the baseline work is to update the individual libraries and submit any patches upstream 2012-05-17T10:49:27 alseh: it is hard to track another projects development head when doing integration work like this. ppisa is your mentor and he sounds interested in both releases and development head of nxlib 2012-05-17T10:49:43 so you may want to do the integration with a release and then work on getting scripts that can pull the nxlib git head as well as an option 2012-05-17T10:50:03 this level of detail should be figured out on rtems-devel mailing list or with your mentor :) 2012-05-17T10:50:12 FWIW nxlib is a SLOW moving project. it won't be hard to track :) 2012-05-17T10:50:19 I expect that there should be created light fork for NXLIB and Nano-X on GIThub 2012-05-17T10:50:45 I would also 2012-05-17T10:50:58 i'd like to see an initial design that includes these ideas sent to rtems-devel 2012-05-17T10:51:13 +1 2012-05-17T10:51:32 And yes, Nano-X has activity peak in 2010 and more activity in 2011 but it is quite quiet now. 2012-05-17T10:51:48 and of course ppisa and alseh can discuss this after the meeting or in private :) 2012-05-17T10:51:51 * gedare drops subtle hints. 2012-05-17T10:51:55 of course :) 2012-05-17T10:52:00 Ok 2012-05-17T10:52:04 and I have a organizational question, on google melange it says that we shoul start a google-code project ? 2012-05-17T10:52:15 i think that is old information 2012-05-17T10:52:25 afaik students do not need a google-code project anymore 2012-05-17T10:52:31 since code submission is through melange 2012-05-17T10:52:36 but we will keep an eye on that for you. 2012-05-17T10:52:48 I was not sure of that, thank you :) 2012-05-17T10:52:50 did you get your student paperwork submitted to google? -- THAT GOES FOR ALL STUDENTS. 2012-05-17T10:52:53 that would be at the end to get paid and confirm you did something. 2012-05-17T10:53:24 PLEASE PLEASE PLEASE .. be able to put together a packet of work to submit to Google to get paid for at end ... however they want it 2012-05-17T10:53:37 alseh: we should have the rtems github set-up soon I hope, and then we will give instructions for development there... 2012-05-17T10:53:48 alseh: I think we will have a clone of the graphics lib there too for you to work from. 2012-05-17T10:53:58 so I will make the forks on rtems-github, right ? 2012-05-17T10:54:02 ok.. WikL is up next 2012-05-17T10:54:02 yeah 2012-05-17T10:54:10 and I have another meeting in 5 minutes.. 2012-05-17T10:54:12 those instructions will come..when we know we have it working. 2012-05-17T10:54:14 ok, thank you 2012-05-17T10:54:28 I hope the mentors / students will put some status and action items in the status report. 2012-05-17T10:54:32 hello viewers 2012-05-17T10:54:37 I want to make it public soon 2012-05-17T10:54:41 WikL: hi. any news, questions, etc? 2012-05-17T10:54:44 Hello 2012-05-17T10:54:46 right now I'm dealing with Pok more than I am with RTEMS 2012-05-17T10:55:29 so I guess I'll just wait until Julien comes back and check directly with him 2012-05-17T10:55:33 ok. hopefully you are asking questions / getting answers from your mentor then ;) 2012-05-17T10:55:33 WikL, I expect that 2012-05-17T10:55:42 Can you run anything on Pok yet? 2012-05-17T10:55:48 I can 2012-05-17T10:56:00 Cool! 2012-05-17T10:56:01 but I'm not sure if I'm getting the expected results 2012-05-17T10:56:10 LOL .. that will require Julien 2012-05-17T10:56:17 WikL: do you have some idea of what you will be doing first / your design goals? 2012-05-17T10:56:36 also the fact that I'm using a trial version of the simulator makes it a bit harder 2012-05-17T10:57:06 not yet, I'm not yet familiar enough with what I will be working with exactly 2012-05-17T10:57:34 trial version of tsim? 2012-05-17T10:57:35 WikL: ok. i encourage you then to write a report about what you learned so far, some idea of what you need to learn, and a rough outline of where you think things are headed 2012-05-17T10:58:34 tsim-eval is limited to 2^32 processor ticks, but that is more than enought to delevop. 2012-05-17T10:58:51 that's what I thought 2012-05-17T10:59:05 WikL, we can discuss this on devel, but I HOPE that the virtualized BSP(s) for Pok can share a LOT of code across the architectures. And I am curious how the clock tick gets into the virtualized OS inside a partition 2012-05-17T10:59:18 but one of the example applications which doesn't do much more than print to the screen and allocate some memory 2012-05-17T10:59:24 hits the limit 2012-05-17T10:59:49 WikL: you'll have to dig at this with your mentor or on the mailing list. 2012-05-17T10:59:52 what is the ARINC schedule of the application? It may be too big. 2012-05-17T11:00:15 WikL: any other issues before we close this meeting? 2012-05-17T11:00:18 email me and Julien privately about this? We may be able to get Gaisler to give you a time-limited copy with lower limits 2012-05-17T11:00:29 ok.. gedare.. you are in charge.. I have an open group meeting now 2012-05-17T11:00:37 DrJoel: ok have fun 2012-05-17T11:00:45 nope, I'm fine apart from that 2012-05-17T11:00:45 bye joel 2012-05-17T11:01:01 cdcs: ARINC schedule? 2012-05-17T11:01:19 WikL: OK good. your project is large so a good design is going to be necessary. so keep plugging at it. 2012-05-17T11:01:26 ********88*** end of the meeting 2012-05-17T11:01:41 ok people feel free to mingle :) 2012-05-17T11:01:45 Hi gedare, about the talk of a plugin model for algorithms along the way? do you konw what joel mean? 2012-05-17T11:02:09 weiY: DrJoel was referring to the bdbuf project by medivhc as using some plugins I think. I'm not quite sure. 2012-05-17T11:02:14 Wikl, The partition schedule. 2012-05-17T11:02:20 weiY: but I don't believe it applies to your project 2012-05-17T11:02:28 Woo, meeting is very effective! 2012-05-17T11:02:43 zw_yao: good i'm glad you thought so 2012-05-17T11:02:44 ok, i will check this with joel. thank you 2012-05-17T11:02:49 welcome. 2012-05-17T11:02:57 mentors please update google doc 2012-05-17T11:03:20 i closed it already >.< 2012-05-17T11:03:27 gedare, and Joel, many thanks for your work. 2012-05-17T11:03:37 next week we will schdule meeting again? 2012-05-17T11:03:46 weiY: should be same bat time, same bat channel. 2012-05-17T11:03:55 hope i'm not dating myself with that reference....hmm 2012-05-17T11:04:10 well, bat man is forever. 2012-05-17T11:04:15 ;-) 2012-05-17T11:04:21 I tried different configurations for the partitions' time slots, but it doesn't help 2012-05-17T11:04:27 (__) 2012-05-17T11:05:10 good night . i have to go to bed 2012-05-17T11:05:30 weiY: good night 2012-05-17T11:05:38 I'm also going to bed, Good night! 2012-05-17T11:05:40 Any application that runs forever will hit the simulation limit. 2012-05-17T11:05:46 bye 2012-05-17T11:05:55 i have to check that application 2012-05-17T11:06:12 but Julien will know that better than me 2012-05-17T11:06:25 I started with the paritions-threads example 2012-05-17T11:06:34 zw_yao: good night too.. 2012-05-17T11:06:50 late meeting over in china :) 2012-05-17T11:07:17 nope, still here 2012-05-17T11:08:14 bye everyone~ 2012-05-17T11:08:22 *** medivhc has quit IRC (Quit: Leaving) 2012-05-17T11:08:23 And the firewall is so "great" 2012-05-17T11:08:34 * gedare grins. careful Deb :P 2012-05-17T11:08:40 it's watching you!! 2012-05-17T11:08:59 Yeah, I am praising it 2012-05-17T11:09:01 \o/ waves to it 2012-05-17T11:09:10 :) 2012-05-17T11:09:26 that's not an easy emote to make. 2012-05-17T11:09:26 I can't even trying to google proxy server 2012-05-17T11:09:46 i think that example does not exit 2012-05-17T11:09:48 grad students practice hard :P 2012-05-17T11:09:52 so if you are seeing the printfs 2012-05-17T11:09:55 i think you are fine :P 2012-05-17T11:09:57 not all of them 2012-05-17T11:10:25 for example, I never get the message about freed memory 2012-05-17T11:10:29 *** weiY has quit IRC (Ping timeout: 255 seconds) 2012-05-17T11:11:47 I'll post the output I'm getting on my project's list 2012-05-17T11:12:12 ok 2012-05-17T11:12:28 that is awkward, the windows are big enough 2012-05-17T11:12:54 have to go now. 2012-05-17T11:12:55 bye 2012-05-17T11:13:10 thx, bye 2012-05-17T11:17:11 to alesh: I would have to go in short time. Have you somethink to discuss now? My suggestion is, that we both check actual versions of RtemsGraphicToolkit and compare it with current Nano-X head and NXlib till next meeting. According to found differences we consider on which to start. 2012-05-17T11:17:24 ppisa: it's alseh not alesh 2012-05-17T11:18:07 Exuse, I memonized it according to one Czech name. 2012-05-17T11:19:57 dose.t matter, i know who you are speaking with 2012-05-17T11:20:42 I agree with what you say, I will have an exam tommorow, and then I will start on working 2012-05-17T11:21:09 OK 2012-05-17T11:23:42 Thanks Joel and Gedare for moderation. We have students presentations next week at IRC meeting time, but I expect to be online. 2012-05-17T11:24:00 So bye to everybody and see you next week. 2012-05-17T11:24:21 ok ppisa take care 2012-05-17T11:24:35 *** ppisa has left #rtems ("Kopete 0.12.7 : http://kopete.kde.org") 2012-05-17T11:26:55 good bye. see you tomorrow. 2012-05-17T11:28:07 I need to go bed too. Good night. 2012-05-17T11:28:45 *** Deb has quit IRC (Quit: Leaving) 2012-05-17T11:49:12 *** arvind_khadri has joined #rtems 2012-05-17T11:50:20 *** Hesham has joined #rtems 2012-05-17T11:56:30 did i miss the meeting ? 2012-05-17T11:56:34 yeah 2012-05-17T11:56:40 it started two hours ago 2012-05-17T11:56:57 so sorry i thought it was 7 pm by my time 2012-05-17T11:57:29 it's ok. should be the same time next week. 2012-05-17T11:57:43 i will be there 2012-05-17T11:57:57 DrJoel will post meeting minutes and action items I think. 2012-05-17T11:58:38 that's good 2012-05-17T11:59:32 yep.. just missing status from 3 projects. I just posted to gsoc mailing list 2012-05-17T12:03:31 DrJoel: i am one of the missing students , could i update the doc ? 2012-05-17T12:03:53 yes.. please go ahead .. 2012-05-17T12:03:59 https://docs.google.com/document/d/1wuGL8Hb64XRTgqKlom6Nx8YxTbLnPiwAQUCBZr19QX0/edit 2012-05-17T12:04:09 thanks 2012-05-17T12:04:16 no problem. 2012-05-17T12:08:51 *** alseh has quit IRC (Quit: Leaving) 2012-05-17T12:31:30 *** arvind_khadri has quit IRC (Ping timeout: 255 seconds) 2012-05-17T12:38:06 I'm out, thanks for the meeting. 2012-05-17T12:38:11 later, dudes 2012-05-17T12:38:25 *** WikL has left #rtems 2012-05-17T12:43:57 *** QingPei1 has joined #rtems 2012-05-17T12:47:06 *** QingPei has quit IRC (Ping timeout: 272 seconds) 2012-05-17T13:01:52 *** cdcs has quit IRC (Quit: Page closed) 2012-05-17T13:10:05 *** gedare has quit IRC (*.net *.split) 2012-05-17T13:10:40 *** gedare has joined #rtems 2012-05-17T13:50:58 *** gedare has quit IRC (*.net *.split) 2012-05-17T13:51:03 *** gedare_ has joined #rtems 2012-05-17T13:52:08 *** gedare_ is now known as gedare 2012-05-17T14:11:20 *** jennifer has quit IRC (Quit: Leaving) 2012-05-17T14:30:12 *** Hesham has quit IRC (Quit: Leaving.) 2012-05-17T15:34:57 Hi all. I hope the meeting went well. 2012-05-17T15:36:15 be back soon 2012-05-17T15:42:52 *** QingPei1 has quit IRC (Read error: Connection reset by peer) 2012-05-17T15:43:47 *** QingPei has joined #rtems 2012-05-17T15:57:59 quiet in here now 2012-05-17T16:54:48 *** DrJoel has quit IRC (Quit: Leaving) 2012-05-17T17:47:47 *** Deb has joined #rtems 2012-05-17T18:43:22 *** QingPei has quit IRC (Read error: Connection reset by peer) 2012-05-17T18:44:00 *** QingPei has joined #rtems 2012-05-17T20:35:08 *** QingPei has quit IRC (Ping timeout: 272 seconds) 2012-05-17T20:50:36 *** QingPei has joined #rtems 2012-05-17T21:03:07 *** xiangfu has joined #rtems 2012-05-17T22:14:34 kiwichris: meeting was nice 2012-05-17T22:14:56 Excellent. Wish I could have attended. I had a tennis match at 7am 2012-05-17T22:14:58 most of the students showed. we gave some feedback and mostly encouragement to post questions on the ml 2012-05-17T22:15:00 Needed to sleep 2012-05-17T22:15:11 nice early morning exercise. 2012-05-17T22:15:34 Yeah, 2hr yesterday and 1.5hr today. Feeling a bit sore now 2012-05-17T22:15:45 i bet! 2012-05-17T22:16:26 I think following the FreeBSD and NetBSD API is the best path. 2012-05-17T22:16:29 kiwichris: the student working on bdbufs had some questions, hopefully he'll send them to the mailing list. he wanted to know about gathering stats. 2012-05-17T22:16:37 *** zw_yao has quit IRC (Remote host closed the connection) 2012-05-17T22:16:48 yes. they are apparently slightly divergent though 2012-05-17T22:16:50 What sort of stats ? 2012-05-17T22:16:59 i'm not certain the only one he quoted was hit rate 2012-05-17T22:17:23 he asked in his introductory email too but it slipped by anyone 2012-05-17T22:17:30 Ok. You need to be careful getting ones that matter 2012-05-17T22:17:55 neither joel nor i knwo enough about it to give guidance so we said to ask mentor or ml 2012-05-17T22:17:56 The issue with states and hit rates is a 100G disk has many blocks. How do you count stats for them 2012-05-17T22:18:32 The idea behind a cache is the map a part of a large external space. 2012-05-17T22:18:42 rtems-devel @april 26: [GSoC] Bdbuf Improvements 2012-05-17T22:18:55 Are the states just misses and hits as globals ? 2012-05-17T22:19:00 ok. 2012-05-17T22:19:03 Just looking 2012-05-17T22:20:21 Ok. 2012-05-17T22:21:01 Lets get Xiang to post a specific question about the bdbuf stats and go from there. 2012-05-17T22:21:20 There is not only hit and miss, there is also the resizing that happens and the effects this can have. 2012-05-17T22:22:41 I also think "hottest" maybe worth considering. Here performance is lost to maintain a table of the hottests spots on disk. 2012-05-17T22:23:13 yeah it would be best for him to ask :) 2012-05-17T22:23:32 Thinking about this it may pay to add hooks for stats and then the code can be standard and special stat features added at runtime. 2012-05-17T22:23:50 The issue as pointed out in the email is how a user profiles the cache usage 2012-05-17T22:24:20 Rebuilding is not what a user would want to do. 2012-05-17T22:24:39 I think Seb is the mentor ? 2012-05-17T22:25:06 correct. the hesitant mentor :) 2012-05-17T22:25:33 it would be good to have plugin and turn on/off stats I would guess. 2012-05-17T22:25:34 I will nudge him to get the email sent and also talk to him about the stat collection 2012-05-17T22:25:39 k 2012-05-17T22:25:41 I agree 2012-05-17T22:25:56 my student is slowly converging to a solid plan for arenas 2012-05-17T22:26:07 after meandering through a jungle 2012-05-17T22:26:23 that should be coming to the ml i would hope early next week 2012-05-17T22:26:23 Great. 2012-05-17T22:26:51 Duke over at parrot posted this link http://pcwalton.github.com/blog/2012/04/23/why-lifetimes/ 2012-05-17T22:26:57 looked interesting 2012-05-17T22:27:09 *** zw_yao has joined #rtems 2012-05-17T22:28:24 The address space is being carved up at a range of different levels. 2012-05-17T22:28:39 took a long time to get to his thesis. :) 2012-05-17T22:28:57 :) 2012-05-17T22:29:08 and left us hanging 2012-05-17T22:29:24 yeah, would like to see the research 2012-05-17T22:30:02 sending an email to respond to the atomics 2012-05-17T22:30:12 Great 2012-05-17T22:30:15 just to push the student to picking an interface and start mocking up a design for us 2012-05-17T22:30:28 Ok. 2012-05-17T22:31:05 I am still pondering atomics in C++ and what it gives us in the embedded world. 2012-05-17T22:31:15 it's not c++ i think it is C 2012-05-17T22:31:19 C11 2012-05-17T22:31:22 Oh 2012-05-17T22:31:27 My mistake 2012-05-17T22:31:43 but.. it is broken right now in compilers and who knows what targets will be supported 2012-05-17T22:31:53 Is this all part of the parallel processing world 2012-05-17T22:32:13 yeah and making it easier to write thread-safe code in multicore systems 2012-05-17T22:32:37 for some reason the C-langauge folks thought it is a good idea for the C language to support. i'm not so sure but eh 2012-05-17T22:33:02 For a kernel we need control not a compiler 2012-05-17T22:34:07 yeah. i'm thinking of this atomics business as another part of the score/cpu like interrupts and ctxt switch. 2012-05-17T22:34:26 though i abhor putting more code udner score/cpu :) 2012-05-17T22:34:36 i'd like to abolish that directory. i think it all belongs in libcpu 2012-05-17T22:34:51 Interesting. 2012-05-17T22:34:54 haven't quite worked through the ramifications for that yet. 2012-05-17T22:35:41 The structure is arch -> cpu -> bsp -> target 2012-05-17T22:36:16 in a "kind of" relationship 2012-05-17T22:36:18 yeah. an arch subdir would be fine too... "libarch" 2012-05-17T22:36:46 I think step 1 is to "unbend" the includes then move from there 2012-05-17T22:37:05 Having 150K headers installed in a build is just plain stupid 2012-05-17T22:37:11 yep 2012-05-17T22:37:20 Seen Joel's blog ? 2012-05-17T22:39:40 yeah some good posts recently. good place for him to get some of his venting done. 2012-05-17T22:39:49 productively ;) 2012-05-17T22:40:03 hehe 2012-05-17T22:45:59 heading to bed goodnight 2012-05-17T22:50:03 cya 2012-05-17T23:20:30 *** zw_yao has quit IRC (Ping timeout: 244 seconds) 2012-05-17T23:42:33 *** Deb has quit IRC (Quit: This computer has gone to sleep) 2012-05-18T00:32:43 *** sevikkk has quit IRC (Read error: Operation timed out) 2012-05-18T00:33:22 *** zw_yao has joined #rtems 2012-05-18T00:44:38 *** sevikkk has joined #rtems 2012-05-18T00:44:50 *** xiangfu has quit IRC (Read error: Connection reset by peer) 2012-05-18T00:51:53 *** xiangfu has joined #rtems 2012-05-18T01:41:06 *** zw_yao has quit IRC (Remote host closed the connection) 2012-05-18T01:49:36 *** sebhub has joined #rtems 2012-05-18T01:49:57 *** arvind_khadri has joined #rtems 2012-05-18T01:50:17 good morning 2012-05-18T02:42:39 *** arvind_khadri has quit IRC (Remote host closed the connection) 2012-05-18T02:45:11 *** arvind_khadri has joined #rtems 2012-05-18T02:45:11 *** arvind_khadri has joined #rtems 2012-05-18T03:17:18 *** xiangfu has quit IRC (Ping timeout: 272 seconds) 2012-05-18T03:20:24 *** arvind_khadri has quit IRC (Ping timeout: 252 seconds) 2012-05-18T03:34:55 *** xiangfu has joined #rtems 2012-05-18T03:41:18 *** xiangfu has quit IRC (Read error: Connection reset by peer) 2012-05-18T03:55:33 *** Deb has joined #rtems 2012-05-18T04:24:14 *** cdcs has joined #rtems 2012-05-18T04:53:27 *** xiangfu has joined #rtems 2012-05-18T07:04:53 *** arvind_khadri has joined #rtems 2012-05-18T07:05:18 *** arvind_k has joined #rtems 2012-05-18T07:21:57 *** gedare is now known as gedare-is-not-he 2012-05-18T07:22:13 *** gedare-is-not-he is now known as gedare-not-here 2012-05-18T07:44:34 *** arvind_k has quit IRC (Ping timeout: 272 seconds) 2012-05-18T08:24:42 *** lcpfnvcy has quit IRC (Ping timeout: 265 seconds) 2012-05-18T08:43:41 *** Deb has left #rtems ("Leaving") 2012-05-18T08:50:19 *** DrJoel has joined #rtems 2012-05-18T08:50:19 *** DrJoel has joined #rtems 2012-05-18T08:50:19 *** ChanServ sets mode: +o DrJoel 2012-05-18T09:06:04 *** lcpfnvcy has joined #rtems 2012-05-18T09:12:33 *** gedare has joined #rtems 2012-05-18T09:30:17 good morning gedare 2012-05-18T09:31:10 *** DrJoel changes topic to "Welcome to RTEMS (http://www.rtems.org). Please feel free to ask questions and, if you can, answer those from others." 2012-05-18T09:37:23 hey DrJoel 2012-05-18T09:37:54 having a good week? 2012-05-18T09:38:29 I don't know .. LOL 2012-05-18T09:38:42 yesterday was the last day of high school for my youngest 2012-05-18T09:38:45 heh then probably not :0 2012-05-18T09:38:47 awww\ 2012-05-18T09:38:50 I have lost a few pounds :) 2012-05-18T09:39:00 hopefully in a healthy/good way 2012-05-18T09:39:12 i have too this winter. lot of bicycling 2012-05-18T09:40:23 Mine is from trying to eat very low carb. No beer, bread, or potatoes in over a week. 2012-05-18T09:40:34 Custom M&M's for Justin's graduation: http://sphotos.xx.fbcdn.net/hphotos-ash3/538142_10150837736020914_540805913_9638550_1384936743_n.jpg 2012-05-18T09:40:57 Hopefully you can see that. It is his "Durrr" face in cap and gown.. 2012-05-18T09:43:44 well i've cut beer out too which helps. 2012-05-18T09:43:59 heh 2012-05-18T09:45:20 turns out liquor is low carbs.. learning to appreciate tequila more .. LOL 2012-05-18T09:45:51 Mentioned you in new blog post: http://rtemsramblings.blogspot.com/2012/05/technical-debt-and-rtems.html 2012-05-18T09:46:55 you've been rocking the posts. reading it now, nice idea i was recently reading about technical debt and found it quite appropriate to our current situation 2012-05-18T09:47:41 it was defined a little differently though, being that sometimes you make decisions in sw development "quick and dirty" that builds up technical debt 2012-05-18T09:47:52 do it too much and your project is technically bankrupt ;) 2012-05-18T09:47:59 :) Yep. 2012-05-18T09:48:17 But evolving, living code has to be modified across the board. 2012-05-18T09:48:25 My recent pci.h work was technical debt 2012-05-18T09:48:39 yeah. well, paying it off 2012-05-18T09:48:39 :) 2012-05-18T09:49:30 he also mentioned something like coding with blinders on. Sometimes you have to work on a small piece of the solution that you can get done without worrying that you don't have the entire, perfect solution yet. 2012-05-18T09:49:50 I think it is that "doing" often helps you see different solutions 2012-05-18T09:50:01 yeah 2012-05-18T09:50:15 you have to get in the weeds to appreciate the view 2012-05-18T09:51:18 nice post. 2012-05-18T09:51:40 although the end i think should be more like "avoid incurring technical debt" rather than paying debt that is owed :) 2012-05-18T09:51:49 but still a good point and well made I think. 2012-05-18T09:54:06 You are right. We make two kinds of patch comments -- avoid debt and don't push it on us. :) 2012-05-18T09:55:13 back in a bit. 2012-05-18T10:04:16 *** sebhub has quit IRC (Remote host closed the connection) 2012-05-18T10:15:59 *** lcpfnvcy has quit IRC (*.net *.split) 2012-05-18T10:16:11 *** lcpfnvcy has joined #rtems 2012-05-18T10:20:55 *** xiangfu has quit IRC (Remote host closed the connection) 2012-05-18T10:25:15 back 2012-05-18T10:26:05 *** arvind_k has joined #rtems 2012-05-18T10:27:56 DrJoel: did you see daniel's email yet 2012-05-18T10:29:03 the interrupt one? 2012-05-18T10:29:07 yeah 2012-05-18T10:29:20 i don't quite follow him 2012-05-18T10:30:27 me either .. I think it is simple but lost in the language 2012-05-18T10:30:35 ok, yeah. 2012-05-18T10:31:37 glad i'm not the only one ;) 2012-05-18T10:32:56 *** cdcs has quit IRC (Ping timeout: 245 seconds) 2012-05-18T10:34:14 Replying :) 2012-05-18T10:34:41 Beagleboard project still has no status. 2012-05-18T10:34:56 I may update it to "no status" after lunch and post it. 2012-05-18T10:35:02 do it 2012-05-18T10:36:24 *** arvind_k has quit IRC (Read error: Operation timed out) 2012-05-18T10:41:42 dr__house, do you have any status on Beagleboard project? 2012-05-18T10:55:20 *** Hesham has joined #rtems 2012-05-18T10:57:30 hi hesham 2012-05-18T10:57:39 hi gedare 2012-05-18T10:57:51 any progress since yesterday? 2012-05-18T10:58:16 i edited the introduction according to our discussion 2012-05-18T10:58:39 looks like you're still including allocation as part of your motivation? 2012-05-18T10:58:48 i deleted it 2012-05-18T10:59:00 the first two sentences? 2012-05-18T10:59:12 3 sentences even. 2012-05-18T10:59:37 i went back and re-read the conversation I had with kiwichris about arenas. I think it would be good for you to re-read it also now 2012-05-18T10:59:37 That's mean he wanna apply protection to pre-allocated area 2012-05-18T10:59:40 do you still have that? 2012-05-18T10:59:57 with chris ? 2012-05-18T11:00:17 an email i sent you labelled GSoC inquiry on mar 22 2012-05-18T11:00:29 has an attachment 'gedare-kiwichris-arenas.txt' 2012-05-18T11:00:58 I think i did not get it 2012-05-18T11:01:07 i will check it 2012-05-18T11:01:11 see if you have it in your mail box otherwise I can resend . 2012-05-18T11:01:15 give me a moment 2012-05-18T11:01:17 ok 2012-05-18T11:02:53 i got it 2012-05-18T11:03:31 okay. great. read through it later 2012-05-18T11:03:56 it should help you to re-establish a good direction. i think combined with the experiences you had the last 2 weeks 2012-05-18T11:03:58 i wanna agree with accepted design today as i have a lot of exams for the next few days 2012-05-18T11:04:02 ah, Ok 2012-05-18T11:04:22 read that .txt file quickly 2012-05-18T11:04:34 or slowly. however long it takes to understand :0 2012-05-18T11:04:43 i will read it now 2012-05-18T11:11:13 gedare:Simply what i understand is , attached tasks are the only ones have access to this Arena 2012-05-18T11:11:35 Allocation is not our concern now 2012-05-18T11:13:01 correct. but we may want to add allocation at some level---probably within an "arena" public API 2012-05-18T11:13:13 so it should be included in the conversation somewhere, just not front and center 2012-05-18T11:14:10 the main problem we want to solve is how to make it easier for users to use/ apply memory attributes in a flexible way while still maintaining efficiency and having some hope of giving real-time guarantees 2012-05-18T11:15:40 "I think the name supports this. You alloc some memory, set some attributes and then share this memory with selected tasks. Other tasks cannot access the memory " 2012-05-18T11:15:45 That's what you mean ? 2012-05-18T11:16:31 Quoted 2012-05-18T11:17:10 yeah that would be some high-level mechanism we would eventuallly like to support 2012-05-18T11:17:31 or even further.... 2012-05-18T11:17:44 allocate memory from within a shared arena, and you automatically have a way to share with other tasks 2012-05-18T11:17:56 as long as they can get the same pointer that you allocated 2012-05-18T11:18:10 lots of possibilities... 2012-05-18T11:18:20 we want to avoid making too many restrictions on what can be done with the memory 2012-05-18T11:18:43 Could we concentrate on memory protection for now and let the part of allocation later ? 2012-05-18T11:18:49 yep 2012-05-18T11:19:09 i think it is safe to assume that the user provides the arena with a chunk of memory and the requested attributes on that chunk, and then adds tasks to the arena 2012-05-18T11:19:10 just to make determined goal for today discussion to work on 2012-05-18T11:19:29 we should not care about the allocation of the initial chunk of memory for the arena 2012-05-18T11:19:31 yeah i am with you 2012-05-18T11:19:53 allocation should only come in when we consider embedding an allocator within an arena, but that will be a high-level mechanism 2012-05-18T11:19:57 we need only to provide protection for that chunk 2012-05-18T11:20:02 in the low level the hw still just sees the original single chunk of memory 2012-05-18T11:20:12 and even in the kernel it sees the one chunk of memory and the set of tasks that can access it 2012-05-18T11:20:32 embedding an allocator is just a way to carve up the chunk of memory into smaller pieces that the application can use 2012-05-18T11:20:38 exactly 2012-05-18T11:20:48 The question is how we make a link between Arena and attached tasks 2012-05-18T11:20:49 allocator can easily come at the end :0 2012-05-18T11:20:52 yes 2012-05-18T11:21:11 i was thinking of something like even control block 2012-05-18T11:21:20 event* 2012-05-18T11:21:31 how do you mean 2012-05-18T11:22:33 In microC OS || there is an event control block that controls the behaviour of semaphores , mailbox , queue 2012-05-18T11:23:02 i think that's similar to priority table in the case of scheduler 2012-05-18T11:23:35 So we setup a table with a maximum entries of tasks can be attached to one Arena 2012-05-18T11:24:02 why limit it? 2012-05-18T11:24:04 and use some "fixed time algorithms" to determine the attached tasks 2012-05-18T11:24:16 that's an option anyway 2012-05-18T11:24:38 well, when do we need to know which task is in which arena? 2012-05-18T11:24:41 in that case tasks can still be attached to Arena , but not activated 2012-05-18T11:24:46 or when do we need to know which arenas a given task is in? 2012-05-18T11:25:16 the Arena has a pointer to that block 2012-05-18T11:25:26 step back for a second 2012-05-18T11:25:32 let's think about the problem you're trying to solve 2012-05-18T11:25:58 What do we need to know, and when do we need to know it? 2012-05-18T11:26:05 or rather, who (in rtems) needs to know it? 2012-05-18T11:26:22 the problem is that we wanna make a link between Arena and tasks attached to it . 2012-05-18T11:27:20 by that link , we can determine which tasks attached to a specific Arena 2012-05-18T11:28:03 why/who needs to know that? 2012-05-18T11:28:04 that must be time fixed to be compatible with real-time issues 2012-05-18T11:28:56 How could a task know that it has access to that Arena ? 2012-05-18T11:29:06 i mean it's attached to that Arena 2012-05-18T11:29:33 does it matter if the task knows? 2012-05-18T11:29:44 What if a non-attached task want to access this Arena ? 2012-05-18T11:30:06 ok that is another question to resolve 2012-05-18T11:30:18 how a task can attach to an arena 2012-05-18T11:30:56 That's why i was thinking of that block that control attaching , activating/deactivating 2012-05-18T11:31:57 let's assume for now that the set of arenas is somehow known 2012-05-18T11:32:09 yeah 2012-05-18T11:32:18 so a task attaches to an arena by invoking a function directly "arena_attach" or similar. 2012-05-18T11:32:34 yeah 2012-05-18T11:32:49 my question is still: why/who must know the relationship between tasks and arenas? 2012-05-18T11:32:55 it is a many-to-many relationship 2012-05-18T11:33:48 why , is to know the set of tasks attached to that Arena 2012-05-18T11:34:01 what do you need to know that for? 2012-05-18T11:34:10 what is the requirement that means you have to know it 2012-05-18T11:34:42 For example if you want to check if a given task is attached to that Arena 2012-05-18T11:35:09 the one reason I can think is when you delete / destroy an arena. 2012-05-18T11:35:14 or if you want to have this task attached to Arena but still "offline" 2012-05-18T11:35:43 let's think about this a different way 2012-05-18T11:35:45 me ? i do not delete it 2012-05-18T11:35:58 so, a task is executing 2012-05-18T11:35:59 I just give that option to the developer 2012-05-18T11:36:15 the task can create an arena... 2012-05-18T11:36:21 and become resident in that arena 2012-05-18T11:36:29 it could join some other existing arenas too 2012-05-18T11:36:36 somehow knowing what those arenas are, right? 2012-05-18T11:36:58 yeah 2012-05-18T11:37:19 let's consider two different "forks" in our story... 2012-05-18T11:37:20 I think that will be embedded to TCB 2012-05-18T11:37:29 in one direction, an arena is deleted 2012-05-18T11:37:34 by this task 2012-05-18T11:37:46 in the other direction, the task has a context switch to a different task 2012-05-18T11:37:58 what has to happen in each of these two cases? 2012-05-18T11:38:52 We could make some restriction to deleting an Arena 2012-05-18T11:39:14 for example , only the task which created an Arena can delete it 2012-05-18T11:39:21 ok that doesn't matter too much 2012-05-18T11:39:26 what happens when an arena is deleted? 2012-05-18T11:39:56 It returns a status code to all attached tasks that it's deleted 2012-05-18T11:40:08 how would it do that when only one task runs at a time? 2012-05-18T11:40:16 and only one task calls "arena_delete" 2012-05-18T11:40:59 let's back up again and ask a single question: does a task need to know what arenas it is using? 2012-05-18T11:41:04 I think that would be through the table of attached tasks 2012-05-18T11:42:10 so when an arena is deleted you would iterate through the table of all tasks attached to the arena, doing...something? 2012-05-18T11:42:41 returning status code that this tasks could not use this arena anymore 2012-05-18T11:43:07 i don't understand where you are returning this status code, so maybe we will come back to this later. 2012-05-18T11:43:09 and cut the connection between that area of memory and that task 2012-05-18T11:44:00 i will return it to the task attached to an Arena 2012-05-18T11:44:14 well, you really want the task to just see the area of memory as if it were part of the global space. so whatever attributes are on the global space should apply. 2012-05-18T11:44:55 yeah, but the task could be doing anything else at the time, so there is no "return" to use. I can see why you are thinking of an "event"-like table now 2012-05-18T11:45:12 you would have to implement some kind of asynchronous mechanism to notify the tasks that they were removed from the arena 2012-05-18T11:45:24 that seems a little bit unnecessary to me 2012-05-18T11:45:52 yeah that's what i mean. 2012-05-18T11:45:57 ok, lets go down the other fork in the road instead. 2012-05-18T11:46:08 what happens during a task context switch? 2012-05-18T11:46:31 happens to what ? 2012-05-18T11:46:56 lets say there are tasks 1 and 2, and arenas a,b,c,d,e 2012-05-18T11:47:01 task 1 is resident in a,b,c 2012-05-18T11:47:06 task 2 is resident in c,d,e 2012-05-18T11:47:19 and there is a context switch from 1->2 2012-05-18T11:47:31 so now task 2 is running instead of task 1 2012-05-18T11:47:41 what has to happen from the point-of-view of arena? 2012-05-18T11:48:45 maybe deactivating a,b Arenas 2012-05-18T11:49:20 Do we agree that there is a "event"- like control block ? 2012-05-18T11:49:45 suppose there is 2012-05-18T11:49:56 but yes, you have to deactivate a,b and activate d,e 2012-05-18T11:50:38 you have to activate c for the use of task 2 2012-05-18T11:51:02 if we use event block 2012-05-18T11:51:05 sure if you deactivated c. 2012-05-18T11:52:35 we may mark tasks belongs to Arena C as deactivated , and Reactivate the new task (2) of the Arena C in that block 2012-05-18T11:52:35 so you have for each arena an array of the tasks that are resident 2012-05-18T11:52:49 yeah 2012-05-18T11:53:03 that's the event- block i'am taking about 2012-05-18T11:53:25 so you will go through every arena and deactivate it, and search its array for task 2 and activate the arena if found? 2012-05-18T11:54:31 that could be fast if we used something like hash table 2012-05-18T11:54:55 so every Arena has an entry in a bit pattern for example 2012-05-18T11:56:04 though it limits the number of arenas. and also the number of tasks 2012-05-18T11:56:39 yeah , but still make the operation fast and fixed time at the end 2012-05-18T11:56:51 either way you have to touch every arena once so the timing depends on how many arenas there are 2012-05-18T11:57:04 No 2012-05-18T11:57:17 how do you deactivate and activate the correct arenas? 2012-05-18T11:58:21 I am thinking of activating/deactivating the tasks part of the table 2012-05-18T11:58:37 aha so you have a 2d array? 2012-05-18T11:58:57 one dimension is arenas one dimension is tasks? 2012-05-18T11:59:09 no 2012-05-18T11:59:25 it's something like bit pattern table 2012-05-18T11:59:31 i don't see how you are finding the arenas that a task is resident in 2012-05-18T11:59:51 in order to deactivate those arenas 2012-05-18T11:59:56 and then activate the arenas for the next task 2012-05-18T12:00:12 I do not deactivate Arena 2012-05-18T12:00:31 i deactivate the bit maps to this task in that Arena table 2012-05-18T12:00:48 how will you know to reactivate those bits later? 2012-05-18T12:01:14 In a context switch 2012-05-18T12:01:36 that's where it gets done, yes, but how does the context switch know that task 2 should have c,d,e activated? 2012-05-18T12:01:44 Or maybe the developer wants to deactivate this task from using that arena for some reason 2012-05-18T12:01:48 where does the context switch look for that? 2012-05-18T12:02:13 for the embedded parts of Arena in TCB 2012-05-18T12:02:15 developer deactivation would be a separate concern. 2012-05-18T12:02:34 so what does the context switch find in the TCB? 2012-05-18T12:03:09 Not sure how it's implemented right now 2012-05-18T12:03:28 how is the context switch going to figure out what arenas a task belongs to? 2012-05-18T12:03:29 But in uCOS || there was a way to do that in fixed time 2012-05-18T12:04:16 maybe a table of bit pattern 2012-05-18T12:04:34 i can review the implementation part of that concern and tell you 2012-05-18T12:05:00 i'm struggling with understanding your choice of data structures to represent the mapping between tasks and arenas. 2012-05-18T12:05:32 I don't know why it is important for an arena to know which tasks are resident, and I don't know how a task (or ctxt switch) can figure out which arenas it belongs to 2012-05-18T12:05:57 these are important design elements 2012-05-18T12:06:24 The most important part of making a map , is to meet real-time constrains 2012-05-18T12:06:41 the real-time constraint is that the context switch must be in bounded time 2012-05-18T12:06:48 bounded, constant time 2012-05-18T12:06:53 yeah 2012-05-18T12:07:15 That's why a mapping exist 2012-05-18T12:07:19 during the context switch the important thing is to find out which arenas task 1 belongs to, and which arenas task 2 belongs to 2012-05-18T12:07:45 because the cs code has to deactivate the arenas containing 1 and activate the arenas containing 2 2012-05-18T12:07:53 in order to enforce the memory attributes 2012-05-18T12:08:52 could you review the part of event-control block of a book , maybe it illustrate that part better than me. 2012-05-18T12:10:26 The idea if that table make it clear how we could attach/deattach/activate/deactivate tasks belongs to that Arena 2012-05-18T12:11:27 That's what came into my mind when you speak about multitasking. 2012-05-18T12:11:31 it seems like having an event system just to notify tasks about their arena state is heavy 2012-05-18T12:11:43 it also does not seem like it will solve the problem of determining which arenas a task belongs to 2012-05-18T12:12:49 i have two ideas about this 2012-05-18T12:13:27 1) TCB points to a small array of pointers to ACBs. then cs just walks that array deactivating for task 1 / activating for task 2 2012-05-18T12:13:46 2) TCB points to a linked list of ACBs. Then cs walks the linked list. 2012-05-18T12:15:00 how to deactivate/activate task 1 ? 2012-05-18T12:15:29 take task 1's TCB, follow the pointer to either the array of linked list of ACBs and iterate through deactivating the arenas 2012-05-18T12:15:49 or if the hw supports it just deactivate all open arenas 2012-05-18T12:16:11 then take task 2's TCB and follow the pointer to iterate through activating its arenas 2012-05-18T12:16:47 the time of the operation depends solely on the number of arenas each task is resident in, which can be limited (without limiting the total number of arenas or tasks) 2012-05-18T12:18:29 and if arenas are turned off then the cost is a single pointer in the tcb 2012-05-18T12:18:30 what's the way to deactivate an Arena ? 2012-05-18T12:18:58 i assume this depends on the hw, and that it can be done if you have a pointer to that arena's control block. 2012-05-18T12:19:19 probably by removing entries from the TLB. for some hw this may mean dumping the entire tlb. 2012-05-18T12:19:38 and the Arena does not have event-like control block ? 2012-05-18T12:19:46 i don't see why it would need one. 2012-05-18T12:20:11 what if we need to list the attached tasks of the arena 2012-05-18T12:20:19 now, going back to the other fork: deleting an arena (or listing the attached tasks of an arena) 2012-05-18T12:20:38 or notify the the tasks attached to arena 2012-05-18T12:20:48 yeah 2012-05-18T12:20:53 we don't want to limit the number of tasks that can use an arena, and neither of these operations need to be real-time constrained to constant time operations 2012-05-18T12:21:06 so I would say a linked list of tasks for each arena would work fine 2012-05-18T12:21:29 you just walk the linked list of the tasks performing whatever operation...removing this arena from their TCB or doing something else. 2012-05-18T12:21:42 what's the maximum number of tasks that RTEMS support ? 2012-05-18T12:21:54 no hard maximum. 2012-05-18T12:22:30 which is why solutions in ucosii won't necessarily work for tems 2012-05-18T12:22:32 rtems* 2012-05-18T12:22:32 If we walk through the linked list that may slow down the operation 2012-05-18T12:22:41 does that matter for these operations? 2012-05-18T12:22:54 anyway the operation still will take the same amount of time as there are tasks attached to the arena 2012-05-18T12:23:14 since it iterates through all tasks. the linked list is only a problem if it is used to try to lookup a task given an arena. 2012-05-18T12:24:02 right? 2012-05-18T12:24:26 what if it's found at the front of the linked list ? 2012-05-18T12:24:43 does it matter? 2012-05-18T12:24:57 anyway i don't know of an operation that will look for a task given an arena. 2012-05-18T12:25:17 just operations that will do something for all tasks of one arena 2012-05-18T12:25:19 or for all arenas of one task 2012-05-18T12:25:48 these kinds of operations make perfect sense to use either a linked list if the number of "all" is not known, or an array if the number of "all" is fixed and small. 2012-05-18T12:25:54 the time of the operations will depends on the number of linked list entries, right ? 2012-05-18T12:26:08 the time of these operations depend on the number of tasks resident in an arena 2012-05-18T12:26:29 in the linked list approach it is the same amount of time / operations 2012-05-18T12:26:46 if you had to scan through an array of all arenas or all tasks it would be larger 2012-05-18T12:27:19 but does this solution meet real-time issues ? 2012-05-18T12:27:36 the real-time issue for these operations is less important. 2012-05-18T12:27:55 the primary concern to have is how the context switch time is affected by using arena 2012-05-18T12:27:58 but you said we had to meet them 2012-05-18T12:28:23 it's sufficient to have a bound on the time 2012-05-18T12:28:37 for deleting an arena, the bound on the time is the number of tasks that can be part of an arena 2012-05-18T12:29:04 that means we have to limit the those numbers to get the time bound right ? 2012-05-18T12:29:06 for a context switch, the bound is the number of arenas that a task can be part of 2012-05-18T12:29:30 we care mostly about the bound on the context switch time, so the number of arenas a task can be part of should be limited 2012-05-18T12:30:05 but we care quite a bit less on the bounds for creating, joining, or deleting an arena I think. 2012-05-18T12:30:06 yeah i see 2012-05-18T12:30:15 we can then use my idea of a table 2012-05-18T12:30:26 to use some more memory 2012-05-18T12:30:35 but to have a faster context swich 2012-05-18T12:30:43 switch* 2012-05-18T12:30:49 and fixed time 2012-05-18T12:31:06 no table for ACB 2012-05-18T12:31:15 right. 2012-05-18T12:31:22 the TCB points to a table of ACB pointers 2012-05-18T12:31:45 yeah or a bit pattern or anything like that 2012-05-18T12:32:00 i don't think a bit pattern is the right thing to use here. 2012-05-18T12:32:16 with a bit pattern you still have to scan it to see which bits are set 2012-05-18T12:32:28 with an array of pointers you just iterate through them accessing the ACBs directly 2012-05-18T12:32:52 again... the primary concern is the cost of iterating these structures, not the cost of looking for something in them 2012-05-18T12:33:01 as opposed to scheduling, where the concern is the cost of searching for the next task to run. 2012-05-18T12:33:03 It's ok , i am just thinking of bit patterns to make this "walk" faster 2012-05-18T12:33:18 bit pattern is a way to make a search faster 2012-05-18T12:33:24 yeah 2012-05-18T12:33:34 using a bit pattern to guide the walk is much slower than traversing either an array or a linked list. 2012-05-18T12:34:20 ok 2012-05-18T12:34:53 so you think you can revise your "introduction to arenas" document? 2012-05-18T12:35:10 and start to fill out the design goals/requirements and initial design ideas 2012-05-18T12:35:40 and to tell how the link between tasks/arenas 2012-05-18T12:36:13 yes, and what the relationship is, what operations require knowledge about the relationship, and what are the time-critical ones 2012-05-18T12:36:33 we've identified at least that iterating through all of the arenas for a given task is time-critical 2012-05-18T12:37:04 yeah 2012-05-18T12:37:11 and that some operations (like arena_delete) will need to iterate through all of the tasks for a given arena 2012-05-18T12:37:27 (and furthermore for those tasks it may need to search for that arena within the TCB->ACB) 2012-05-18T12:37:28 the attached tasks to agiven Arena will be kept in a linked list right ? 2012-05-18T12:37:53 do you think so? 2012-05-18T12:38:02 you said so 2012-05-18T12:38:07 but do you think so :) 2012-05-18T12:38:34 yeah because we do not wanna limit the number of attached tasks to a given Arena 2012-05-18T12:38:48 and we do not know the number of them 2012-05-18T12:39:05 will we ever need to search for a specific task attached to an arena? 2012-05-18T12:39:37 when it's deleted 2012-05-18T12:39:55 not searching 2012-05-18T12:40:10 iterating through attached tasks . 2012-05-18T12:40:52 right ? 2012-05-18T12:41:48 sounds right. i don't know the answer ;) 2012-05-18T12:42:59 if Arena is deleted , that would iterate to that list , and delete This ACB from the linked list of pointers , so every entry would be something like , TCB->ACB.delete 2012-05-18T12:43:47 delete is an inline function to delete a given entry from the linked list 2012-05-18T12:44:30 by then , the ACB is deleted from the TCB linked list which include Arenas a task resident on 2012-05-18T12:44:35 does it make since ? 2012-05-18T12:44:37 sense 2012-05-18T12:44:43 more like... for ( node in acb->list ) node->tcb->arenas->acb[x] = NULL; list_remove(node->tcb->arenas) 2012-05-18T12:45:13 yeah yeah , i am concerned with the idea for now ;) 2012-05-18T12:45:19 you'll get it figured 2012-05-18T12:45:26 ok work from there... 2012-05-18T12:45:41 try to think through / sketch what each of the arena operations will do in this design 2012-05-18T12:45:58 is not that a port of the implementation ? 2012-05-18T12:46:05 a part 2012-05-18T12:46:16 yes but you need to have some clue about what you are going to implement so that your design doesn't suck 2012-05-18T12:46:22 i wanna finish the design doc 2012-05-18T12:46:35 you have to know how the various arena functions will use the data structures 2012-05-18T12:46:41 in order to be certain that your data structures are good 2012-05-18T12:47:10 ok 2012-05-18T12:47:17 anything else? 2012-05-18T12:47:38 i will edit the design doc and make some figures 2012-05-18T12:47:44 ok great 2012-05-18T12:47:51 and study/pass your exams :) 2012-05-18T12:47:56 when we should show it to the community ? 2012-05-18T12:48:08 let's iterate one more time and see how it looks 2012-05-18T12:48:12 i have to study hard for the next days 2012-05-18T12:48:14 I Think it's getting close 2012-05-18T12:48:26 when are your exams? 2012-05-18T12:48:39 will you be able to meet on Tuesday? 2012-05-18T12:49:15 I had a plenty of time in the final exams , i could work though the project withing final exams 2012-05-18T12:49:51 but i have 4 exams/interviews on monday/tuesday 2012-05-18T12:50:11 after then i am can continue on the project 2012-05-18T12:50:36 i will be able to meet on Tuesday , but i think i'll not do much progess 2012-05-18T12:50:42 progress ** 2012-05-18T12:50:47 ok work on the doc when you have time. i'll check in and make edits and we'll plan to see each other thursday for the weekly update meeting and if we need we can meet some friday. i'd like to have you send out your design either next friday or or the following tuesday 2012-05-18T12:51:29 that weekend is a holiday weekend in the US, and I will not be around 2012-05-18T12:51:49 It's Okay 2012-05-18T12:52:03 i will work through the doc withing this week 2012-05-18T12:52:15 within 2012-05-18T12:52:26 ok 2012-05-18T12:52:29 anything else ? 2012-05-18T12:52:38 not that i can think of 2012-05-18T12:53:14 good meeting. long meeting :) 2012-05-18T12:53:30 you might want to go re-read that irc chat again i had with kiwichris 2012-05-18T12:53:34 yeah but getting close to the final design :) 2012-05-18T12:53:42 and think about it in terms of the design 2012-05-18T12:54:07 more like the initial design 2012-05-18T12:54:09 ;-) 2012-05-18T12:54:09 I will read our log along with kiwichris log 2012-05-18T12:54:13 ok. 2012-05-18T12:54:20 good luck! 2012-05-18T12:54:39 I discovered that the hardest part is the design not the implementation :) 2012-05-18T12:54:45 thanks :) 2012-05-18T12:59:19 then you've learned a useful lesson already 2012-05-18T13:00:38 That's for sure 2012-05-18T13:18:33 *** Hesham has quit IRC (Quit: Leaving.) 2012-05-18T13:40:04 DrJoel: What do you think about hosting an LXR for RTEMS? 2012-05-18T13:41:06 i'm not sure if it makes sense or not :) 2012-05-18T13:42:38 *** Hesham has joined #rtems 2012-05-18T13:44:03 *** antgreen has joined #rtems 2012-05-18T13:44:26 *** gedare has quit IRC (Quit: Ex-Chat) 2012-05-18T13:51:52 *** Hesham has quit IRC (Quit: Leaving.) 2012-05-18T14:18:34 *** arvind_k has joined #rtems 2012-05-18T14:23:45 *** DrJoel has quit IRC (Quit: Leaving) 2012-05-18T14:27:28 *** Hesham has joined #rtems 2012-05-18T15:20:39 *** ppisa has joined #rtems 2012-05-18T15:33:14 *** gedare-not-here is now known as gedare 2012-05-18T15:48:05 *** antgreen has quit IRC (Read error: Connection reset by peer) 2012-05-18T15:54:21 *** arvind_k has quit IRC (Ping timeout: 246 seconds) 2012-05-18T16:49:35 *** ppisa has left #rtems ("Kopete 0.12.7 : http://kopete.kde.org") 2012-05-18T23:48:17 *** L84Supper has quit IRC (Quit: puff of smoke) 2012-05-18T23:48:28 *** A2Sheds has joined #rtems 2012-05-19T01:37:33 *** sevikkk has quit IRC (Quit: Leaving.) 2012-05-19T01:47:06 *** sevikkk has joined #rtems 2012-05-19T03:39:05 *** Deb has joined #rtems 2012-05-19T04:32:55 *** arvind_khadri has joined #rtems 2012-05-19T04:56:54 *** Deb has left #rtems ("Leaving") 2012-05-19T05:17:03 *** arvind_khadri has quit IRC (Ping timeout: 256 seconds) 2012-05-19T05:29:24 *** arvind_khadri has joined #rtems 2012-05-19T06:03:44 *** sevikkk has quit IRC (Read error: Connection reset by peer) 2012-05-19T06:12:36 *** arvind_khadri has quit IRC (Read error: Connection reset by peer) 2012-05-19T06:12:52 *** sevikkk has joined #rtems 2012-05-19T06:17:52 *** arvind_khadri has joined #rtems 2012-05-19T06:24:26 *** sevikkk has quit IRC (Quit: Leaving.) 2012-05-19T06:24:58 *** sevikkk has joined #rtems 2012-05-19T06:48:36 *** arvind_khadri has quit IRC (Ping timeout: 246 seconds) 2012-05-19T07:02:17 *** arvind_khadri has joined #rtems 2012-05-19T07:52:14 *** arvind_khadri has quit IRC (Ping timeout: 245 seconds) 2012-05-19T08:15:49 *** arvind_khadri has joined #rtems 2012-05-19T08:25:54 *** arvind_khadri has quit IRC (Read error: Operation timed out) 2012-05-19T08:57:50 *** arvind_khadri has joined #rtems 2012-05-19T08:58:31 *** arvind_khadri has joined #rtems 2012-05-19T10:17:50 *** lcpfnvcy has quit IRC (Read error: Operation timed out) 2012-05-19T10:39:19 *** arvind_khadri has quit IRC (Ping timeout: 245 seconds) 2012-05-19T10:40:19 *** gedare has quit IRC (Read error: Operation timed out) 2012-05-19T10:55:47 *** gedare has joined #rtems 2012-05-19T11:48:04 *** alseh has joined #rtems 2012-05-19T13:57:21 *** gedare has quit IRC (Ping timeout: 246 seconds) 2012-05-19T14:18:49 *** alseh has quit IRC (Remote host closed the connection) 2012-05-19T16:35:08 *** kiwichris has quit IRC (Quit: This computer has gone to sleep) 2012-05-19T17:20:57 *** gedare has joined #rtems 2012-05-20T03:48:30 *** xiangfu has joined #rtems 2012-05-20T05:57:05 *** arvind_khadri has joined #rtems 2012-05-20T06:14:28 *** arvind_khadri has quit IRC (Ping timeout: 276 seconds) 2012-05-20T06:31:00 *** xiangfu has quit IRC (Ping timeout: 246 seconds) 2012-05-20T06:49:05 *** xiangfu has joined #rtems 2012-05-20T08:37:42 *** xiangfu has quit IRC (Ping timeout: 246 seconds) 2012-05-20T08:49:50 *** xiangfu has joined #rtems 2012-05-20T08:52:03 *** xiangfu has quit IRC (Read error: Connection reset by peer) 2012-05-20T08:52:53 *** xiangfu has joined #rtems 2012-05-20T08:54:54 *** xiangfu has quit IRC (Read error: Connection reset by peer) 2012-05-20T09:11:33 *** xiangfu has joined #rtems 2012-05-20T09:14:40 *** xiangfu has quit IRC (Client Quit) 2012-05-20T09:15:16 *** xiangfu has joined #rtems 2012-05-20T10:09:24 *** xiangfu has quit IRC (Ping timeout: 246 seconds) 2012-05-20T10:09:49 *** arvind_khadri has joined #rtems 2012-05-20T11:05:53 *** Hesham has joined #rtems 2012-05-20T11:08:05 *** Hesham1 has joined #rtems 2012-05-20T11:08:06 *** Hesham has quit IRC (Read error: Connection reset by peer) 2012-05-20T11:17:21 *** Hesham1 has quit IRC (Ping timeout: 252 seconds) 2012-05-20T11:18:33 *** Hesham has joined #rtems 2012-05-20T13:17:20 *** A2Sheds has quit IRC (Quit: puff of smoke) 2012-05-20T13:17:30 *** A0Sheds has joined #rtems 2012-05-20T13:17:31 *** A0Sheds has joined #rtems 2012-05-20T17:15:40 *** kiwichris has joined #rtems 2012-05-20T19:13:09 *** arvind_khadri has quit IRC (Ping timeout: 248 seconds) 2012-05-20T21:10:47 *** kiwichris has quit IRC (Ping timeout: 250 seconds) 2012-05-20T21:15:50 *** kiwichris has joined #rtems 2012-05-20T23:49:12 *** xiangfu has joined #rtems