Wednesday, 2017-01-18

*** tpb has joined #timvideos00:00
jeamicolous: I haven't noticed it, and because of it being a tutorial we don't have audio the entire time. i'll keep monitoring though00:01
micolousOK00:01
micolousI'm in the room too00:01
micolousI notice the speaker has the lapel mic quite low00:01
jeamicolous: I'm not sure what instructions have been given to them for tutorials. I'd normally get them to have mic on when they are presenting, off while they are helping people. does that sound right to you?00:02
jeamicolous: ok, he is talking now, and i think it sounds alright00:03
micolousOK, it's just the room00:03
micolousIt sounds like it's ringing in the room00:03
jeainteresting00:03
jeaI can ask the venue AV people to check00:03
xfxfjea: if the lav mic is low that'll be the problem - worth pinging the vollies in #lca-av about that00:07
jeaxfxf: seems fine now when I listen00:22
xfxfrad00:22
jeaand room is also improved00:23
micolousyeah00:28
micolousergh, wireless keeps going down01:16
micolousAnd they have some cloud-based tutorial system where you're accessing EC2 thruogh the browser01:16
xfxf#lca2016 VIDEO: I love title slides - @CarlFK http://youtu.be/lololz01:20
xfxfor01:20
xfxf#lca2016 VIDEO: I love title slides - Carl 'donut' Karsten http://youtu.be/lololz01:20
CarlFKhttps://github.com/CarlFK/veyepar/blob/master/dj/scripts/tweet.py01:22
tpbTitle: veyepar/tweet.py at master · CarlFK/veyepar · GitHub (at github.com)01:22
jeaI think twitter handle if possible01:29
jeabut the donut is a nice touch01:29
jeacan we change his twitter handle?01:30
xfxfCarlFK: http://pastebin.com/ur1y3H6m01:37
tpbTitle: def mk_tweet(self, prefix, twitter_ids, video_name, authors, video_url): - Pastebin.com (at pastebin.com)01:37
xfxfCarlFK: http://pastebin.com/QJsiw6kV01:52
tpbTitle: def mk_tweet(self, prefix, twitter_ids, video_name, authors, video_url): - Pastebin.com (at pastebin.com)01:52
mithro_florent_: https://github.com/mithro/litex/commit/1aabe4aa2c526554f120aeec46adc544904f6a8c02:00
tpbTitle: Temp: Trying to get the SPI flash driver working in 1x mode. · mithro/[email protected] · GitHub (at github.com)02:00
_florent_mithro: can you do a test with that?:02:09
_florent_mithro:http://pastebin.com/kYy96ZVJ02:09
tpbTitle: [Python] from litex.gen import * from litex.gen.genlib.misc import timeline from lite - Pastebin.com (at pastebin.com)02:09
_florent_mithro: I worked on it yesterday, but haven't tested it yet02:09
CarlFKxfxf: #lca2017 Human-Driven Development - @aurynn http://youtu.be/qcBol_s410k02:11
mithro_florent_: Afraid it doesn't work02:38
*** CarlFK has quit IRC02:43
_florent_mithro: ok, you can keep yours for now then02:48
_florent_mithro: thanks for testing02:48
mithro_florent_: I can't see why yours would fail....02:49
_florent_mithro: I'll do a test and compare with yours02:49
mithrothe old difference I can see is a Tristate on dq....02:50
_florent_you don't need tristate in single mode02:52
_florent_hmm... in fact maybe I'm wrong if the flash is not configured correctly and use bidirectionnal miso/mosi02:52
_florent_are you using the two miso/mosi lines in your code?02:53
mithro_florent_: Hrm... I'm not sure how my code can work....02:58
_florent_mithro: I think we should  do a simulation02:59
_florent_mithro: I have some a SPISlave model that we can use03:00
mithro_florent_: I'm just trying to drop my code into your changed file03:00
_florent_mithro: ok03:02
*** tvCommitBot has joined #timvideos03:17
tvCommitBot[streaming-system] joeladdison pushed 1 new commit to master: https://git.io/vMPwl03:17
tvCommitBotstreaming-system/master 0670b73 Joel Addison: Use bootstrap columns for videos on front page03:17
*** tvCommitBot has left #timvideos03:17
*** CarlFK has joined #timvideos03:25
*** ChanServ sets mode: +v CarlFK03:25
thaytansometimes in these talks, I notice a single frame flicker on the projector output03:29
thaytanI'm in plenary atm, and saw the screen do that just now03:29
thaytanis that an opsis thing?03:29
thaytanor am I just being oversensitive?03:29
CarlFKthaytan: yes - we see this to03:29
mithrothaytan: _florent_ believes we have a race condition with the framebuffer handling03:31
thaytanI wonder if it relates to things going out of sync03:31
mithrothaytan: Hrm?03:33
mithrothaytan: _florent_ thinks we might not be updating a 32bit address location of the frame buffer atomically (need to do 4 x 8bit writes)03:34
mithro_florent_: DDR works again with putting the frequency back up03:34
thaytanmithro, this problem where sometimes the captured audio/video gets out of sync03:34
thaytanI'm was musing about possible causes yesterday03:35
mithrothaytan: I don't think the encoder will see it at all03:35
thaytanOK, not that then03:35
mithrothaytan: It could be the fact that the encoder is providing a pretty variable fps03:36
thaytanmithro, shouldn't matter to matroska03:37
thaytanwhere do you notice the desync?03:37
mithrothaytan: Dunno - I think it was just the GUI03:37
xfxfthaytan: normally it's only in voctogui, not in the actual recordings/streams, so i'm assuming it's a voctogui bug03:37
xfxfsuper occasionally it's in the recordings too and we have to restart everything03:37
xfxfand it's the mux of alsa + v4l into vocto which seems to be getting out of sync, unrelated to nettime03:38
xfxfthat said, we do have to adjust the alsa audio by a ~1000ms delay to keep it in sync with v4l.  mithro said that indicates a problem (but it works for us)03:38
xfxfbut that's a constant delay, doesn't seem to drift03:39
thaytanhmm03:39
xfxfvoctogui sync issue happens at least once a day, restarting it fixes it03:41
xfxfthe audio video drift issue has only happened 2 or 3 times across all 6 rooms the entire conf03:41
xfxfspoken to the vocto guys about the sync issue, it happens for them too with SDI sources (and it's happening on our PC here with a SDI source), so unrelated to the opsis03:42
xfxfthey also don't know why it's happening either03:42
xfxfhoping somebody will figure it out as it's super annoying, but not the end of the world03:42
mithrohttp://paste.debian.net/909251/03:44
tpbTitle: debian Pastezone (at paste.debian.net)03:44
mithro_florent_: I'm just testing to see if the mimasv2 can run the uart at a decent speed04:56
cr1901_modernmithro: No. It's hardcoded.05:03
mithroWe should rewrite the PIC firmware on the mimasv205:05
mithroIt uses a similar PIC to on the TOFE LowSpeedIO board05:05
mithroWhich I have firmware for at https://github.com/timvideos/HDMI2USB-TOFE-LowSpeedIO/tree/firmware/firmware05:05
tpbTitle: HDMI2USB-TOFE-LowSpeedIO/firmware at firmware · timvideos/HDMI2USB-TOFE-LowSpeedIO · GitHub (at github.com)05:05
cr1901_modernmithro: That would be nice. Is the pic firmware for Mimasv2 open-sourced?05:05
cr1901_modernI couldn't find it05:05
mithrocr1901_modern: No05:10
mithrocr1901_modern: But as I mentioned we can probably just reuse most of the TOFE-LSIO firmware05:11
cr1901_modernFair05:11
mithrocr1901_modern: https://github.com/timvideos/HDMI2USB-TOFE-LowSpeedIO/wiki/PIC-Firmware05:17
tpbTitle: PIC Firmware · timvideos/HDMI2USB-TOFE-LowSpeedIO Wiki · GitHub (at github.com)05:17
mithrocr1901_modern: and https://github.com/timvideos/HDMI2USB-TOFE-LowSpeedIO/blob/firmware/firmware/NOTES.md05:17
tpbTitle: HDMI2USB-TOFE-LowSpeedIO/NOTES.md at firmware · timvideos/HDMI2USB-TOFE-LowSpeedIO · GitHub (at github.com)05:17
cr1901_modernmithro: I'm gonna be honest. I JUST noticed the PIC on the TOFE board. XD05:18
cr1901_modernI thought it was an FTDI type chip05:18
mithrojea: How are things going, you interested in trying to hack on the DHCP stuff again?05:43
cr1901_modernmithro: Small Windows pull req incoming. I'd like to add a "developer mode" to bypass checking out the modules at some point (because Python's module path handling is pretty awful)05:44
jeamithro: Things are going pretty well. I'm just going to try to setup some monitoring things, then I'll look at DHCP and other thigns05:45
cr1901_modern_florent_: Out of curiosity, why was this a merge commit? https://github.com/enjoy-digital/litex/commit/d23d7a591bf8a54067f68dc54c04d45b1a3bac7205:45
tpbTitle: Merge pull request #18 from mithro/spiflash1x · enjoy-digital/[email protected] · GitHub (at github.com)05:45
mithrojea: Okay05:45
cr1901_moderninstead of a rebase*05:45
jeatumbleweed: are you about?05:46
tumbleweedjea: yeah, sitting in plenary right now05:46
jeatumbleweed: ok. question: can you see anything odd on boardwalk from where you are?05:47
jeawhen i look at the stream, it seems to be out of sync between audio and video, but the people in the room tell me that it looks in sync05:47
jeajust wondering if there is anything odd in logs05:48
tumbleweedlooks like things happened about 10 mins ago05:50
tumbleweedhard to say what, not a login, but almost everything restarted05:50
jeai did a logout and login to try to deal with it05:51
tumbleweedurgh, there's still pulseaudios left from previous logins05:51
jeabut it didn't completely fix it by the looks of it05:51
jeaoh really?05:51
tumbleweedsome killall pulseaudios may be a good idea05:51
jeado you want to do that now, or will we leave it and deal with it later?05:52
jeait is not that bad, just a bit more obvious than some others05:52
tumbleweedI can poke at it05:52
mithrocr1901_modern: Merge is the normal default05:53
tumbleweedjea: done05:53
jeatumbleweed: you might need to check again. they logged out and logged back in after you poked things, i think05:55
tumbleweedjea: yeahI see05:56
tumbleweedwe probbaly should have warned them a little more05:56
jeayeah05:56
jeai've told them now that if it freezes, just restart the GUI05:56
tumbleweedyeah05:57
xfxfi thought the systemd spawning things on ssh'ing in was fixed?05:59
xfxfor is this when logging out / logging in from the local machine?05:59
tumbleweedit was06:00
tumbleweedyes06:00
xfxfoh06:01
xfxfis the solution to not log out / back in again, or something else?06:01
tumbleweedthe solution is a commit that I'll push when I have my gpg keys again06:02
xfxfok, np06:02
cr1901_modernmithro... idk how to tell you this, but... the mimasv2 target is booting from flash on my end06:28
cr1901_modernall the way to H2U> prompt06:28
cr1901_modernI didn't CHANGE anything06:28
cr1901_modernWas it SERIOUSLY "not enough RAM" that broke everything?06:28
cr1901_modernhttps://github.com/mithro/HDMI2USB-litex-firmware/commit/a4b7a12ac3d3cac9e3861d76e77ce6304495fb16 Okay, so the design works without any timing errors from Xilinx PAR at 83.33 MHz when booting from SPI flash. What changed?07:18
cr1901_modernmithro: ^^07:18
mithroI get timing errors on that08:13
cr1901_modernOh, but it works...08:18
cr1901_modernIn any case, I'm running a few tests by building a bunch of bitstreams that only vary in subtle ways08:19
cr1901_modernBut HDMI2USB boots on my end08:19
cr1901_modernIdk why yours is different08:19
cr1901_modernmithro: Curiously, with the bad base bitstream (with the bad SRAM size), I'm no longer able to get any output from the UART before the CPU goes to hell08:36
cr1901_modernOnly difference is the RAM size... I think I'm going to have to use my brain for this one.08:36
cr1901_modernmithro: Ignore what I was saying. https://logs.timvideos.us/%23timvideos/%23timvideos.2016-12-23.log.html#t2016-12-23T20:54:11 This is why I should keep a log of observed behavior. B/c after a while I start mixing up what I observe for a given set of changes10:00
cr1901_modernThen again, I suppose the IRC log *is* the log10:00
CarlFKtumbleweed: https://github.com/CarlFK/veyepar/blob/master/dj/scripts/enc.py#L442-L46210:10
tpbTitle: veyepar/enc.py at master · CarlFK/veyepar · GitHub (at github.com)10:10
cr1901_modernhttps://github.com/mithro/HDMI2USB-litex-firmware/commit/e7f2be6b2c9a5ef6b532541195cc556e5217b043 Was... was the SPIflash mapped *twice* before this commit?10:16
tpbTitle: Remove spiflash memmap, as it map to ROM location. · mithro/[email protected] · GitHub (at github.com)10:16
tumbleweedCarlFK: vb=50k ab=50k preset=ultrafast progress=1 movflags=+faststart10:30
cr1901_modernLooks like that commit didn't have any effect (don't think it was optimized out tho)10:42
CarlFKhttp://veyepar.ndv.cdn.nextdayvideo.com/veyepar/lca/lca2017/dv/plenary/2017-01-17/08_27_44.ts.mp411:19
CarlFKhttp://veyepar.nextdayvideo.com/main/C/lca/S/lca2017/L/plenary/D/18/E/12018/#ep-sm11:23
tpbTitle: veyepar: Slot (at veyepar.nextdayvideo.com)11:23
xfxfSLOT!11:26
*** CarlFK has quit IRC11:59
*** sorlob has joined #timvideos12:32
*** sorlob has quit IRC12:36
*** CarlFK has joined #timvideos13:18
*** ChanServ sets mode: +v CarlFK13:18
*** rohitksingh has joined #timvideos14:08
*** Joelw_ has joined #timvideos15:26
*** tumbleweed has quit IRC15:26
*** ivodd_ has joined #timvideos15:26
*** tumbleweed_ has joined #timvideos15:26
*** tumbleweed_ is now known as tumbleweed15:26
*** tumbleweed has joined #timvideos15:26
*** Joelw has quit IRC15:26
*** ivodd has quit IRC15:26
*** olasd has quit IRC15:26
*** olasd has joined #timvideos15:27
*** fumblehool has joined #timvideos16:49
*** vystrela has quit IRC19:01
*** vystrela has joined #timvideos19:01
*** fumblehool has quit IRC19:10
*** froztbyte has quit IRC19:22
*** froztbyte has joined #timvideos19:33
CarlFK6:30... time for me to wake up.  of course.19:33
*** sb0 has quit IRC19:58
*** rohitksingh has quit IRC20:07
*** sb0 has joined #timvideos20:42
CarlFKheading to resi brefkast20:46
tumbleweedCarlFK: all rax are synced20:52
tumbleweedsome of the rats are probably in sync too21:12
CarlFKneat - to the bus!!21:21
*** CarlFK has quit IRC21:21
*** CarlFK has joined #timvideos21:48
*** ChanServ sets mode: +v CarlFK21:48
*** CarlFK has quit IRC21:53
mithrojea: we got the turbot running "make gateware" at the hackfest, right?22:08
mithrojea: what did we end up using for a disk drive?22:08
tumbleweedI thought I saw carlfk's M2 USB drive hanging off it22:09
jeaI think we got it building stuff. it was running off my microsd card, which is probably still in it22:17
thaytanmorning22:29
*** CarlFK has joined #timvideos22:52
*** ChanServ sets mode: +v CarlFK22:52
tumbleweedOK, I bought the domain. http://carlrants.com/23:26
tpbTitle: Carl Rants - Live from the linux.conf.au TOC (at carlrants.com)23:26
*** tumbleweed has quit IRC23:32
*** tumbleweed has joined #timvideos23:32
thaytanis the rust tutorial not being streamed?23:45
tumbleweedjea: ^^23:51
jeathaytan: it should be23:51
jealet me check23:52
thaytanjea, sorry - it is23:52
thaytanit just was marked as 'not started' in the web ui23:52
jeathaytan: missing audio?23:52
thaytanconfused me23:52
jeaah, might not have refreshed yet23:52
thaytan<tim> using headphones to watch show on projector, can't check :)23:53
thaytan<tim> a/v sync seems way off23:53
thaytan<tim> using html5 player in firefox23:53
thaytan<tim> not that it matters much23:53
jeayes, a/v sync seems off to me too, but i'm waiting to see camera again23:54
jeaboth wellingtons seem a bit out of sync23:55
thaytanis it possible to archive the direct matroska feeds before voctomix?23:56
thaytanI'm really curious if it's capture or mixing that introduces the desync23:56
tumbleweednot impossible, but I don't think we want to risk the extra I/O23:57
CarlFKthaytan: probably, but trying to .. yeah, all that.23:57
tumbleweedthe desync problems are on the most loaded machine23:57
*** hozer has quit IRC23:57
thaytantumbleweed, really? is it saturating the network link?23:57
thaytanif it's just a disk io problem, could feed it back out to a separate machine23:58
tumbleweedyeah, we have space on the network (2x 1gbps cards because we can't fit on one)23:58
tumbleweedwhich makes we wonder why we didn't just do link aggregation, rather than the two IP thing23:58

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!