Wednesday, 2018-01-17

*** tpb has joined #timvideos00:00
xfxfwe can do later. sorry phone reception shite. Omw back00:02
xfxfand we have a super helpful AV person who already unlocked a bunch of stuff00:02
tumbleweedyay. We maybe we can make him unlock more things00:03
xfxfher :P00:09
*** medicalwei has quit IRC00:51
*** medicalwei has joined #timvideos00:52
*** CarlFK has quit IRC01:08
*** CarlFK has joined #timvideos01:09
*** ChanServ sets mode: +v CarlFK01:09
CarlFKgrumpermon: look, it just needs a roof. A) That shit's way too complicated.01:11
*** ewen has quit IRC01:22
*** ewen has joined #timvideos01:22
skayhere is the recipe for vegeite truffles
tpbTitle: vegemite truffles (at
*** rohitksingh-demo has joined #timvideos01:43
*** rohitksingh-demo has quit IRC01:48
*** rohitksingh-demo has joined #timvideos01:48
tpbTitle: More boards by mithro · Pull Request #57 · mithro/HDMI2USB-litex-firmware · GitHub (at
rohitksingh-demomithro: great! I'll carry it forward01:54
mithrorohitksingh-demo: Where did you get up to making the nexys_video work?01:57
rohitksingh-demomithro: have to test the generated bitstream, will that right now01:59
mithrorohitksingh-demo: Okay cool -- I was thinking that maybe we should try and get Yosys as a frontend for 7-series devices in litex-buildenv working and tested?02:00
rohitksingh-demomithro: mmm....yosys can generate the netlists right?02:02
mithrorohitksingh-demo: Yeah - vivado is still needed for the pnr+bitstream generation02:03
mithrorohitksingh-demo: Apparently I already added yosys to conda!?
rohitksingh-demomithro: we should do that02:03
rohitksingh-demomithro: awesome, much easier yosys installation \o/02:04
mithroYeah - we will still need vivado for the VHDL support...02:04
xfxfmithro: recently i keep having compatibility issues with HDMI2USB and newer crestron systems in venues; just had it in Guthrie @ UTS.  i can keep putting my scaler in, or i can poke you to figure out how we fix this properly.  i took a EDID dump.  please advise what you want to do.02:08
xfxfcarl and stefanor also are seeing this increasingly02:08
*** rohitksingh-demo has quit IRC02:28
*** rohitksingh-demo has joined #timvideos02:28
skaysomeone kick CarlFK so that he gives us a statement for his election02:34
* mithro kicks CarlFK[m] 02:44
mithroDone! :-P02:44
CarlFK[m]xfxf (IRC): [email protected]:~$ cat .minirc.dfl03:08
CarlFK[m]pu port             /dev/ttyACM003:08
*** rohitksingh-demo has quit IRC03:13
*** rohitksingh-demo has joined #timvideos03:15
*** rohitksingh-demo has quit IRC03:21
*** rohitksingh-demo has joined #timvideos03:22
*** sb0 has quit IRC03:23
*** rohitksingh-demo has quit IRC04:12
xfxflatest firmware made no difference. Inline scaler worked. Black magic USB capture works. Got info and pics mithro wanted.04:15
xfxfthis happens on all new crestron gear04:15
*** rohitksingh-demo has joined #timvideos04:15
rohitksingh-demomithro: what does it mean if memtest is okay, but H2C console's "debug ddr" is showing no output at all04:16
mithrorohitksingh-demo: The "debug ddr" only shows output when you have the bandwidth calculation stuff on I think....04:16
mithrorohitksingh-demo: It could also be broken on Opsis04:16
rohitksingh-demomithro: there is no pattern output or something like that in nexys_video?04:20
mithrorohitksingh-demo: I think there should be?04:20
rohitksingh-demoI cannot get any output from nexys_video04:21
rohitksingh-demomithro: ^04:22
mithrorohitksingh-demo: Any video output I assume you mean?04:25
mithrorohitksingh-demo: It is probably broken....04:26
rohitksingh-demomithro: yup, can't get any output...also "output0: [email protected] from pattern (underflows: 25213295)" I think there is issue with ddr3?04:26
rohitksingh-demosoo many underflows04:27
mithroIf you are getting underflows you won't get any video output04:27
rohitksingh-demomithro: yeah, that means dma is not able to fetch data, which means probably ddr3 is not working?04:27
mithrorohitksingh-demo: Yeah probbaly04:30
shenkimithro: is your dtb generation part of the build process?04:32
shenki(the litex-buildenv)04:32
mithroshenki: Not yet04:32
shenkii don't get any serial console output after04:32
shenki[    0.000000] console [tty0] enabled04:32
shenki[    0.000000] bootconsole [litex_uart_a0] disabled04:32
shenkiusing my kernel tree and the qemu in litex-buildenv04:33
mithroshenki: Well, I have no idea what your kernel tree is -- so no idea what could be wrong....04:36
shenkiit works fine on hardware04:37
mithroSo, I guess it must be something in the qemu emulation?04:41
mithroshenki: I guess something in interrupts if it works in earlycon but not after?04:44
mithroshenki: BTW Have you tried with hardware built from that repo?04:44
mithroxobs: So I was confused, Josh is coming by tomorrow04:45
mithroxobs: Did you still the Oscilliscope access?04:45
xobsmithro: No worries.  I've mostly got it figured out.04:45
shenkimithro: yes04:46
mithroshenki: Okay04:47
*** sb0 has joined #timvideos04:50
shenkimithro: it looks like litex-buildenv points at an outdated version of liteeth?04:50
mithroshenki: There is deliberate - liteeth has change the csr interface slightly which breaks libuip, qemu emulation, etc04:51
shenkimithro: okay. so we don't plan on doing any networking for the miniconf?04:51
shenki(under linux)04:51
mithroshenki: I was hoping you would get it working?04:51
shenkii was looking at linux. i haven't looked at the other parts04:52
mithroshenki: You can just target the old csr format?04:53
shenkimithro: do you mean without the level register?04:54
mithrolevel register?04:54
shenkithe level register is the addition that _florent_ made to allow us to write a linux driver for it04:55
tpbTitle: core/mac/sram: add csr for fifo level of sram reader (for the linux d… · enjoy-digital/[email protected] · GitHub (at
mithroshenki: These were the changes that broke things ->
tpbTitle: core/mac: apply changes from misoc: remove gap_checker in rx, add pre… · enjoy-digital/[email protected] · GitHub (at
mithrojea[m]: Can you look at the edid Ryan has collected from the crestron and decode it into mode timings?04:58
mithroshenki: It would be pretty easy to fix up the drivers + qemu emulation -- it's just having time for someone to do that05:00
xfxfCarlFK[m]: that hard drive detection issue.  if you go to the menu when it says it can't detect any HDD's and reselect the "detect hard drive" option, it then detects the HDD's fine and continues on its way05:03
xfxfthis def isn't a hardware issue, it looks like a bug in the installer or something05:03
xfxfalso the installer still crashes at the end with the preseed command05:06
_florent_mithro: we can revert the liteeth changes for now if it breaks things for you05:11
shenkii think the only changes required are qemu05:13
shenkiit works for me on hardware05:13
mithroshenki: okay, if you can get things working on the hardware I can fix up the qemu emulation...05:18
shenkimithro: the bios works. i haven't tested the hdmi2usb firmware05:19
shenkiping works under linux still05:19
mithroshenki: The HDMI2USB firmware is broken with the latest liteeth05:19
shenkinow to make use of the new features, and to fix the bugs05:19
shenkimithro: any reason it should be? doesn't it use the same csr.h?05:20
mithroshenki: I think it uses one of the registers which was renamed or something?05:20
mithroWorking on something else right now05:20
shenkiit looks like it should just work to me. i will try booting it05:22
shenkimithro: what didn't work?05:24
shenki$ telnet
shenkiConnected to
shenkiEscape character is '^]'.05:24
shenkiH2U 00:00:32>05:24
shenkiH2U 00:00:32>05:24
mithrotftp booting05:24
shenkithe hdmi2usb doesn't tftp boot, does it? that's the bios' job05:25
shenkihdmi2usb firmware, that is05:25
shenkibios tftp booting is working for me05:25
mithroI could just be confusing qemu and real hardware05:25
shenkiperhaps you had a mismatch between old software and new gateware?05:26
shenkii've got a Super Legit© method to keeping our device tree in sync with the generated addresses of the gateware05:29
shenkigrep -h "^#" ./build/arty_net_or1k.linux/software/include/generated/csr.h ./build/arty_net_or1k.linux/software/include/generated/mem.h | grep -v include > ~/dev/kernels/litex/include/dt-bindings/litex.h05:29
shenkiand then something like this in the device tree:
shenkiit's the worst. but at least it's simple to update05:31
*** shorne has quit IRC05:33
*** shorne has joined #timvideos05:35
rohitksingh-demomithro: you have pull request from Joe :p05:41
mithroshenki: Joe?05:41
shenkinot me05:45
tpbTitle: FuPy (FPGA MicroPython) on Mimas v2 and Arty (at
ewenIn summary, seems to work (with one minor tweak) for lm32 soft CPU on Mimas v2 and Arty A7 (but or1k soft CPU didn't compile -- system.h not found)05:53
*** futarisIRCcloud has quit IRC05:55
rohitksingh-demoewen: Awesome work and the blog post! \o/06:00
ewenrohitksingh-demo: Thanks!06:05
ewenmithro: litex readthedocs appears not to be building because has only a logo PNG in it.  Is there documentation somewehre else that should be rendered?06:12
mithroewen: Oh....06:13
mithroI guess _florent_ didn't import the migen/misoc docs?06:13
ewenPerhaps not :-(06:14
ewenmithro: exists, but seems last updated 201506:15
tpbTitle: Migen manual Migen 0.5.dev0 documentation (at
ewenmithro: And exists, but seems to be failing to build06:16
xfxfCarlFK[m], stefanor: do we have deployable rooms yet?  (i wanna test vocto capture)06:48
*** mcarden has quit IRC06:50
jeaxfxf, CarlFK, tumbleweed - are you all still at google?07:16
CarlFKjea; yes07:16
felix_xobs, mithro:
tpbTitle: USB Sniffer | Ultra-Embedded (at
jeaCarlFK: cool, I'll come back then07:17
felix_haven't tried that, but might be interesting07:17
xobsfelix_: That does look like it could be interesting.07:17
mithroxobs: I have all parts for that here btw07:20
*** futarisIRCcloud has joined #timvideos07:25
cr1901_modernmithro: Is it a known issue that the mimasv2 constrains for the PMOD ports are dead wrong (in Migen/Litex _and_ Numato's provided file)?07:34
jea[m]CarlFK (IRC): can you let me in please07:56
CarlFKjea: sending doorman07:56
jea[m]Thanks :)07:56
*** ewen has quit IRC08:47
*** paddatrapper has quit IRC08:48
*** paddatrapper has joined #timvideos09:01
*** paddatrapper has quit IRC09:08
*** paddatrapper has joined #timvideos09:12
*** paddatrapper has quit IRC09:17
tpbTitle: Script to download prebuilt firmware · GitHub (at
*** futarisIRCcloud has quit IRC09:55
*** paddatrapper has joined #timvideos10:12
*** futarisIRCcloud has joined #timvideos10:46
rohitksingh-demomithro: do we have ATX to DC Barrel Jack connector?10:57
futarisIRCcloudewen: Cheers. FuPy works fine on my Arty here. Haven't tested it on Mimas v2 yet.11:06
*** sb0 has quit IRC12:03
*** rails_ has joined #timvideos13:16
*** rails_ is now known as rails[I]13:17
*** futarisIRCcloud has quit IRC13:25
*** thaytan has quit IRC13:34
*** thaytan has joined #timvideos13:36
*** ChanServ sets mode: +v thaytan13:36
*** CarlFK has quit IRC13:42
*** rohitksingh-demo has quit IRC13:43
*** sb0 has joined #timvideos13:59
*** CarlFK has joined #timvideos14:02
*** ChanServ sets mode: +v CarlFK14:02
cr1901_modern_florent_: I've been reading the SD spec this morning to debug some issues I'm having on my end. Why do you delegate the length of the card response (either 48 or 132 bits) to the firmware >>14:19
cr1901_modernas opposed to autodetecting it?14:19
*** mcarden has joined #timvideos14:27
_florent_cr1901_modern: i don't remember, probably for flexibility, can you point to the code that does it?14:40
cr1901_modern_florent_: Stuff like this
tpbTitle: litesdcard/sdcard.c at master · enjoy-digital/litesdcard · GitHub (at
cr1901_modernAFAICT, the direction of the transfer and the length of the control response are implicit in the command14:49
cr1901_modernIn any case, I suspect that the reason I'm having SD card woes is that my SD card doesn't actually support SET_BLOCK_COUNT14:50
* cr1901_modern doesn't know for sure right now14:51
_florent_cr1901_modern: we are doing that since it's easier to do in software, this way the gateware does not need to know which command is what and just transfers values from the software to the sdcard16:19
_florent_cr1901_modern: set_block_count should indeed not report an error16:24
cr1901_modern_florent_: Does that mean old SD cards are not supported?16:30
_florent_cr1901_modern: your sdcard need to support multiblock write/read16:39
_florent_cr1901_modern: which is probably not the case of old sdcards16:39
_florent_cr1901_modern: but we could try to support it16:39
cr1901_modern I don't think I have any spares atm :(16:39
_florent_cr1901_modern: just that for our usecase, only supporting new sdcards was fine16:40
cr1901_modern_florent_: Atm, not sure how invasive the changes are to support it. I need to do more testing16:40
cr1901_modernI don't _know_ for sure that this SD card doesn't support SET_BLOCK_COUNT.16:40
cr1901_modernJust that SET_BLOCK_COUNT errors out, and I need to query the actual response code16:41
cr1901_modern_not_ the cmdevt that is returned in CSR16:41
_florent_cr1901_modern: if you only get an error for this command, your card is probably not supporting multiple block16:43
_florent_cr1901_modern: you can also use sdcard_write_single_block / sdcard_read_single_block16:43
_florent_cr1901_modern: what is your usecase for litesdcard btw?16:44
cr1901_modernSee privmsg16:44
cr1901_modern_florent_: According to simplified spec, multiple read/multiple write are mandatory commands to support. And those work w/o returning an error.16:48
cr1901_modernIt's SET_BLOCK_COUNT that causes problems16:49
cr1901_modern(page 83 on the document itself- page 101 in Adobe reader)16:49
cr1901_modern>this way the gateware does not need to know which command is what and just transfers values from the software to the sdcard16:51
cr1901_modernThat's fine/fair. Users prob shouldn't need to know the spec anyway to use the firmware. It just confused me when debugging.16:51
*** rails[I] has quit IRC18:26
CarlFK[m]im waking up...20:33
*** CarlFK has quit IRC21:01
jea[m]CarlFK: I’m awake and leaving now for the coffee shop21:01
micolous[m]Morning campers21:35
jea[m]Hello micolous22:41
micolous[m]Heading to the room22:46
*** CarlFK has joined #timvideos22:46
*** ChanServ sets mode: +v CarlFK22:46
Kamilionhrm, ran across a new HDMI capture box I wasn't familiar with22:49
Kamilionanybody seen/experienced this lil' guy before?22:50
CarlFK[m]no - im intrigued by the reset button22:55
CarlFK[m]"You can buy other more capable device (e.g. Raspberry PI) which can do the same job for less money or better. "23:03
CarlFK[m]how can a pi get hdmi signal?23:03
*** rails[I] has joined #timvideos23:09
*** rails[I] has quit IRC23:09
*** rails[I] has joined #timvideos23:09
*** futarisIRCcloud has joined #timvideos23:34
futarisIRCcloudKamilion: The LKV373A has been mentioned on Hackaday a few times before.  From what I recall, 1080p gets scaled down to a weird resolution.  720p is ok. Everything is h.264 across multicast packets over ethernet.23:38

Generated by 2.13.1 by Marius Gedminas - find it at!