**** BEGIN LOGGING AT Tue May 27 02:59:58 2014
May 27 04:53:04 * karki thanks Abhishek_ for the link. his 
May 27 04:53:04 <karki> b
May 27 04:53:27 * karki hates when he prematurely presses enter 
May 27 04:53:48 <karki> Any way the BBB is working as expected :)
May 27 05:07:34 <Abhishek_> Morning
May 27 06:05:59 <_av500_> gm from china
May 27 06:06:07 <_av500_> hope you are all busy coding
May 27 06:16:25 <Abhishek_> ni hao _av500_
May 27 06:16:41 <_av500_> :
May 27 06:16:42 <_av500_> )
May 27 07:00:42 <vvu> mornin _av500_
May 27 07:43:23 <Abhishek_> mranostay: what is hindering the new C compiler code generation now? Is it the same cregister issue or there are many more issues?
May 27 08:39:28 <panto> there appears to be some bloat
May 27 09:29:50 <karki> panto : what is the use of "MODULE_DEVICE_TABLE(of, pruss_dt_ids)" ?
May 27 09:30:11 <panto> boilerplace for device tree
May 27 09:31:52 <karki> boilerplace?
May 27 09:32:03 <Abhishek_> boilerplate
May 27 09:33:05 <Abhishek_> av500: Is Mans on IRC?
May 27 09:34:44 <panto> *plate
May 27 09:36:10 <karki> I kinda got confused, cause googling it told me that it was for some hot plug handling.....
May 27 09:57:30 <Abhishek_> panto: If virtio and vrings handle high speed LAN drivers in KVM/Qemu, why are they slower than mmap in the context of the BeagleBone?
May 27 09:58:06 <panto> err, it depends on what you do
May 27 09:58:17 <panto> mmap is a very crude interface
May 27 09:58:54 <panto> sure it's faster, but you have to do more stuff to have a full featured interface
May 27 10:19:40 <_av500_> Abhishek_: sometimes
May 27 10:19:52 <_av500_> but no channel that I know atm
May 27 10:20:22 <Abhishek_> k, I wanted to ask about the segfault with the asm memcpy
May 27 10:21:00 <_av500_> where does ti segfault?
May 27 10:21:04 <_av500_> run it in gdb and see
May 27 10:21:14 <_av500_> if its on preload, this needs to be enabled for user space
May 27 10:21:22 <_av500_> then try the one without preload
May 27 10:22:38 <Abhishek_> ah yes, the "pld" instruction, the UIO memory region actually is noncached.
May 27 10:23:15 <Abhishek_> IIRC
May 27 10:30:27 <Abhishek_> av500: This:  *** stack smashing detected ***: /mnt/BeagleLogic/PRULATest/Debug/PRULATest terminated. Program received signal SIGABRT, Aborted.
May 27 11:03:17 <_av500_> hmm
May 27 11:03:21 <_av500_> is this from gdb?
May 27 11:03:30 <Abhishek_> yep
May 27 11:04:31 <_av500_> what output tdoes bt give?`
May 27 11:04:34 <_av500_> for backtrace
May 27 11:09:13 <Abhishek_> av500: this: Program received signal SIGSEGV, Segmentation fault. 0x0000914a in main ()
May 27 11:20:53 <_av500_> ok
May 27 11:20:58 <_av500_> now "bt"
May 27 11:21:03 <_av500_> should give you a backtrace
May 27 11:21:47 <Abhishek_> #0  0x0000914a in main ()
May 27 11:21:53 <Abhishek_> that's all
May 27 11:22:13 <_av500_> hmm
May 27 11:22:25 <_av500_> can you put the binary somewhere?
May 27 11:22:57 <Abhishek_> should I link it with -static and give you a copy?
May 27 11:23:14 <_av500_> I cant run it
May 27 11:23:18 <_av500_> I can just look at it
May 27 11:23:22 <_av500_> see where it crashes
May 27 11:23:34 <_av500_> but ok, link static and rerun
May 27 11:27:49 <Abhishek_> _av500_: static gives a more complete backtrace:
May 27 11:27:59 <Abhishek_> #0  0x0000b956 in __libc_do_syscall ()
May 27 11:27:59 <Abhishek_> #1  0x000273ba in raise ()
May 27 11:27:59 <Abhishek_> #2  0x0000edaa in abort ()
May 27 11:27:59 <Abhishek_> #3  0x0001065c in __libc_message ()
May 27 11:27:59 <Abhishek_> #4  0x0001dfe0 in __fortify_fail ()
May 27 11:28:00 <Abhishek_> #5  0x0001df9a in __stack_chk_fail ()
May 27 11:28:00 <Abhishek_> #6  0x00009050 in main ()
May 27 11:29:21 * Abhishek_ thinks it would have been better to pastebin the backtrace
May 27 11:29:52 <_av500_> still not wiser
May 27 11:29:55 <_av500_> 9050 in main
May 27 11:29:59 <_av500_> the rest is just effect
May 27 11:30:18 <_av500_> ah
May 27 11:30:20 <_av500_> do up
May 27 11:30:25 <_av500_>  until you reach #6
May 27 11:30:29 <_av500_> then disasm
May 27 11:32:52 <_av500_> "disas"
May 27 11:34:48 <Abhishek_> https://www.irccloud.com/pastebin/NBJL1pPN
May 27 11:44:17 * karki wonders why adafruit tutorials changed their layout.....
May 27 11:46:48 <_av500_> Abhishek_: still looks like code around that stack_check
May 27 11:46:58 <_av500_> doesnt tell us what triggers it
May 27 11:47:03 <_av500_> what code are you running?
May 27 11:47:09 <_av500_> can you pastebin the source?
May 27 11:48:47 <Abhishek_> Ok, i ran objdump and here's the code
May 27 11:49:31 <Abhishek_> https://www.irccloud.com/pastebin/lWM2MVt8
May 27 11:51:07 <Abhishek_> Let me expand it a bit so that you are able to see memcpy_neon and the asm code as well:
May 27 11:51:18 <Abhishek_> https://www.irccloud.com/pastebin/6skWd06m
May 27 12:07:49 <_av500_> and with std memcpy it works?
May 27 12:08:21 <_av500_> are you addresses aligned?
May 27 12:08:23 <Abhishek_> yep, even with a C SIMD memcpy it works (2x memcpy=280ms, simd=150ms)
May 27 12:08:31 <Abhishek_> yep aligned to 64 bytes
May 27 12:08:42 <_av500_> hm
May 27 12:09:34 <_av500_> and how much do you copy?
May 27 12:09:43 <_av500_> the neon copies 128 bytes at a time
May 27 12:09:51 <Abhishek_> 4 MBytes
May 27 12:10:01 <_av500_> right
May 27 12:13:02 <Abhishek_> _av500_: IIRC Mans' code was for -mfloat-abi=soft, while what I am currently compiling is hard float. Could there be missing registers that need to be preserved?
May 27 12:13:31 <_av500_> oh
May 27 12:14:10 <_av500_> yes, could be
May 27 12:14:38 <Abhishek_> but then I wonder why it works at all with only malloc'ed regions
May 27 12:15:31 <_av500_> you copy from where to where?
May 27 12:16:45 <Abhishek_> currently there's a UIO memory region shared with the PRU, I copy from there into a buffer created with memalign(64, 64 * 1024 * 1024);
May 27 12:17:31 <Abhishek_> I compiled the sample application from that thread separately, and it does not seem to segfault
May 27 12:17:57 <Abhishek_> with memcpy, C vector-based memcpy and ASM NEON memcpy
May 27 12:18:28 <Abhishek_> all 3 work there, but I don't know why it doesn't work with my code.
May 27 12:21:09 <rseethamraju> karki: did you write an example for i2c in PyBBIO?
May 27 12:21:48 <karki> I had, never committed though.
May 27 12:23:51 <karki> I had developed the example using the arduino due (the only 3.3V I2C dev I had on me). But Alexander did not want an example where the i2c peripheral costs more than the BBB. ;)
May 27 12:30:48 <rseethamraju> karki: I’m writing the documentation for i2c
May 27 13:18:54 <karki> rseethamraju : Thank you :3 It happened to be my job!
May 27 13:19:47 <karki> tell me if you need any help. Though I believe the source code itself is sufficiently documented.
May 27 13:21:08 <rseethamraju> ya I figured.
May 27 13:23:29 <rseethamraju> So I’m just copy pasting your stuff
May 27 13:39:58 <karki> Does anyone have a "build" dir under "/lib/modules/<kernel-version>" ?
May 27 13:44:58 <_av500_> Abhishek_: can you reduce the per copy size frown from 4MB until the crahs goes away?
May 27 13:45:04 <_av500_> does it happen for a 128 byte copy?
May 27 13:46:19 <Abhishek_> _av500_: segfault for a 256 byte copy
May 27 13:47:28 <_av500_> hmm
May 27 13:47:57 <_av500_> can you make a test case with no PRU code, just the memory source an dest?
May 27 13:48:08 <_av500_> allocate both mems like you dpo and copy
May 27 13:50:00 <Abhishek_> _av500_: Still a segfault, I removed all PRU code, it's now just two memalign and a memcpy and it still crashes, strange
May 27 13:50:19 <karki> vvu Abhishek_ : how do you go about getting the "build" directory if it is not already present at  "/lib/modules/<kernel-version>/build"
May 27 14:27:22 <Abhishek_> _av500_: Very strange. If I call the function "indirectly" i.e. through a function pointer then no segfault. Any ideas?
May 27 14:30:49 <_av500_> Abhishek_: pastebin the test codee
May 27 14:33:34 <joel_> s/codee/codes/
May 27 14:36:35 <Abhishek_> _av500_: https://gist.github.com/abhishek-kakkar/e2a4a0aefd7d1e7fc15d
May 27 14:37:02 <Abhishek_> This does not give any segfault
May 27 14:51:33 <karki> panto : how do you go about getting the "build" directory if it is not already present at  "/lib/modules/<kernel-version>/build"?
May 27 15:29:47 <karki> praveendath92 : ping
May 27 16:05:38 <praveendath92> karki: Pong
May 27 16:05:57 <karki> how did you get your makefile to work?
May 27 16:06:14 <karki> praveendath92 : how do you go about getting the "build" directory if it is not already present at  "/lib/modules/<kernel-version>/build"
May 27 16:06:37 <praveendath92> You mean headers?
May 27 16:06:51 <karki> yes.
May 27 16:07:43 <praveendath92> I'm currently doing all my tests on my Ubuntu
May 27 16:08:00 <praveendath92> You are looking for bone headers?
May 27 16:08:26 <karki> yes.
May 27 16:08:42 <praveendath92> Just a min.
May 27 16:08:44 <karki> Ah, so you don't really need bone specific stuff, do you?
May 27 16:08:47 <praveendath92> I'm searching/
May 27 16:08:52 <karki> for your drivers.
May 27 16:09:03 <praveendath92> Not really actually.
May 27 16:09:15 <karki> I'm trying out a few stuff. I'll check if it works first.
May 27 16:09:42 * karki is trying all the relevant apt-get 's he can find!
May 27 16:11:05 <praveendath92> I don't think apt-get worked for me.
May 27 16:11:58 <karki> praveendath92 : yeah, apt-get didn't work for me either.
May 27 16:12:12 * karki withstands another round of "FAIL"
May 27 16:12:28 <praveendath92> Haha.
May 27 16:13:47 <karki> I really can't do anything without kernel headers! :(
May 27 16:14:00 <praveendath92> karki: This worked for me. https://gist.github.com/praveendath92/9640684
May 27 16:16:29 <karki> you are running ubuntu on the BBB?
May 27 16:16:39 <karki> I'm running debian :/
May 27 16:17:44 <praveendath92> Nope. Debian.
May 27 16:17:50 <praveendath92> That worked for debian
May 27 16:18:14 <karki> Yes. I see it is working for me.
May 27 16:18:18 * Abhishek_ is running the 15 May version of Debian
May 27 16:18:28 * karki keeps his fingers crossed.
May 27 16:18:44 <Abhishek_> that should be roughly a 500MB download
May 27 16:20:41 <karki> praveendath92 : how long did it take you?
May 27 16:22:18 <praveendath92> I did it when I was in college
May 27 16:22:27 <praveendath92> So, around 10-15 mins max.
May 27 16:22:53 * Abhishek_ knows :)
May 27 16:23:52 <praveendath92> Haha
May 27 16:24:37 <karki> what net speed do you have at college :o
May 27 16:26:17 <Abhishek_> The net experience depends on many factors [can be insanely fast sometimes] but is marred by the amount of proxies and port blockers.
May 27 16:26:47 <Abhishek_> Using git and IRC is a pain
May 27 16:26:52 <praveendath92> No more proxy in IITB
May 27 16:27:11 <praveendath92> git works but yeah irc is still a pain.
May 27 16:36:20 * vvu is done with his exams! OS was like majestic
May 27 16:36:40 <praveendath92> Yo vvu \m/
May 27 16:36:44 <Abhishek_> cool, party time now :)
May 27 16:36:55 <vvu> first sleep :))
May 27 16:37:09 <praveendath92> Yes. yes. 1st thing after exams.
May 27 16:52:25 <karki> praveendath92 : I get " Cannot mkdir: No space left on device" :P
May 27 16:52:38 <praveendath92> Haha.
May 27 16:52:46 <praveendath92> How much is your sdcard?
May 27 16:53:02 <karki> 16GB
May 27 16:53:10 <karki> but my fs is much lesser.
May 27 16:53:18 <Abhishek_> did you grow the filesystem?
May 27 16:53:21 <karki> have to expand I guess
May 27 16:53:27 <praveendath92> Oh. I remember getting something like that.
May 27 16:53:32 <Abhishek_> look in /opt/scripts
May 27 16:53:33 <karki> Abhishek_ : not yet.
May 27 16:53:34 <praveendath92> Yeah. You have to karki
May 27 16:53:48 <karki> why /opt/scripts
May 27 16:54:29 <karki> how to expand the FS via cmd line on the BBB?
May 27 16:54:40 <Abhishek_> karki: /opt/scripts/tools/grow_partition.sh
May 27 16:55:18 <praveendath92> karki: Once everything is up and running I recommend you backup the image on BB.
May 27 16:55:57 <karki> praveendath92 : yeah. I'll do that.
May 27 16:56:26 <praveendath92> I made backups to cloud. Lol.
May 27 17:05:13 <Abhishek_> I once lost my entire BeagleLogic code folder in the days prior to coding period to a single rm -rf [there wasn't really any code in that though, just cloned repositories]. I'm now careful with backups.
May 27 17:08:35 * Abhishek_ finds it a little hard to imagine that the system images and stuff we care about is all held in trapped gate charges. One flip and it's gone...
May 27 17:13:30 <praveendath92> Haha. Abhishek_ You could restore that I think.
May 27 17:14:10 <karki> alias your rm
May 27 17:15:48 <Abhishek_> I dunno really. it was a python script gone wrong, and I couldn't restore it. I had to recreate the tree.
May 27 17:16:40 <Abhishek_> It has been > 3 weeks since then
May 27 18:09:04 * karki is happy that everything is working fine :) Thanks Abhishek_ and praveendath921
May 27 18:09:20 <karki> G'nite :)
May 27 18:09:24 <praveendath921> Yo karki.
May 27 18:09:29 <praveendath921> Back it up!
May 27 18:09:36 <praveendath921> :D Night :)
May 27 18:09:37 <karki> oh, yeah!
May 27 18:09:48 <karki> Thanks for reminding :P
May 27 18:09:54 <Abhishek_> G'night karki.
May 27 18:21:15 * vvu hates the internet here...goes so slow to ssh
May 27 19:23:38 <Abhishek_> ds2: does a 128-bit OCP for the DDR interface (EMIF) means it is most efficient to write 128 bits at a time?
May 27 21:00:51 <tridge> good morning
**** ENDING LOGGING AT Wed May 28 02:59:58 2014