Sunday, 2016-08-14

*** tpb has joined #timvideos00:00
CarlFKjea: xfxf - I have had them flip out and power cycle fixed it.00:00
jeaturns out it was the frequency. this replacement had 60hz set but camera is 5000:02
jeaas for the original box, it was going bad on the encoder after about 3mins post restart. quite odd00:02
xfxfoh right.  yeah all au cameras will be 5000:05
xfxfherm, we need to flash the firmware on it which is default 5000:06
xfxfmithro: did you build two firmwares, or just one?00:06
jeamithro just flashed it with the 50hz firmware00:06
CarlFKxfxf: pretty sure "just one"00:07
CarlFKor maybe not.  or maybe he made a 50 one... weee!!!!00:07
mithroxfxf: There is a git repo which shows what I did00:09
jeaCarlFK: a new 60hz version was made on friday night that we are using for the laptop input00:10
*** cr1901 has joined #timvideos00:13
CarlFKmithro: when you have 3-5 min.. I am having trouble, hoping it is something simple:  AssertionError: b'f' (b'f') != b'\x00' (b'\x00')00:15
tpbTitle: Ubuntu Pastebin (at
CarlFKmithro: yes, I see that f<>0, but the input file is right.   It is being served over an sshfs mount from my other test box.  md5 checks out.00:17
shenkimithro: where do the headers for the uart etc live?00:38
shenkimithro: ok. i need to build the soc first?00:40
shenkimithro: ok. what do i want to run for that?00:40
mithroshenki: the makefile in opsis-soc00:44
shenkimithro: yep. do i need to source something to set up the python paths?00:44
mithroshenki: If you set the path as I mentioned above, you'll get the right stuff00:45
shenki$ make00:45
shenkirm -rf build00:45
shenki./ --with-ethernet --nocompile-gateware --cpu-type lm3200:45
shenkiTraceback (most recent call last):00:45
shenki  File "./", line 9, in <module>00:45
shenki    from litex.gen import *00:45
shenkiImportError: No module named 'litex'00:45
shenkiMakefile:4: recipe for target 'opsis_minisoc' failed00:45
mithroshenki: you didn't do the path command probably then00:46
mithroshenki: you made sure to put the conda element *before* the others?00:46
shenki09:32 <@mithro> "echo PATH=$PWD/conda/bin:$PATH"00:46
mithroshenki: "which python"00:46
shenkimithro: ok, got it00:47
mithroshenki: I pushed some extra patchs to opsis-soc to
tpbTitle: GitHub - mithro/opsis-soc: Opsis SoC based on LiteX (at
shenkimithro: ok. should i abort my build and use that instead?00:49
mithroIf you want spiflash support, yes00:49
shenkimithro: how do i forceok00:49
mithroshenki: I need to see your changes to get micropython building....00:50
shenkimithro: mather?00:50
shenkimithro: master?00:50
mithroshenki: yes00:50
shenkimithro: should i be able to build for or1k?00:51
mithroshenki: dunno, start with lm3200:51
mithroshenki: were did you end up?00:54
shenkimithro: the fire talk00:54
shenkimithro: xilinx segfaulted \o/00:55 line 8:  4651 Segmentation fault      (core dumped) map -ol high -w -o top_map.ncd top.ngd top.pcf00:55
mithroshenki: yay - check your licensing?00:56
tpbTitle: GitHub - shenki/micropython at lm32 (at
shenkimithro: the lm32 toolchain has include path issues01:05
shenkimithro: pull again01:07
mithroshenki: Basically, you want something like this at the top of the make file01:08
tpbTitle: Pastebin | IRCCloud (at
*** cr1901 has quit IRC01:10
mithroshenki: ./litex/litex/soc/software/include/base/stdint.h01:17
jeaCarlFK: could I get your help with a veyepar thing please01:17
CarlFKjea: sure  what up?01:17
mithro./mpconfigport.h:80:20: fatal error: alloca.h: No such file or directory01:18
jeaI'm using VLC to watch the videos and confirm the times that I want. I was wondering how to properly convert the video time into the time that veyepar wants01:19
shenkimithro: hrm. we don't have an alloca01:20
shenki#include <alloca.h>01:20
CarlFKjea:   if vlc says 1:23, put 1:23 in the box.01:20
CarlFKjea: the 'pattern' is h:m:s and if you leave off h or m, then it is just seconds01:21
CarlFKso 61 can either be 61 or 1:101:21
CarlFKjea: hopefully you found this:01:22
CarlFK               ~/Videos/veyepar/troy/dnf2016/dv/crystal_ballroom/2016-07-12/11_23_21.ts :start-time=051.001:22
CarlFKunder [x]Show More01:22
CarlFK:start-time=051.0      is the vlc syntax for .. start there01:23
shenkimithro: which libc are we using for the lm32?01:27
shenkimithro: can we do a newlib build? that would give us setjmp/longjump01:28
mithroshenki: the above include/base01:28
shenkimithro: or uclibc01:28
shenkimithro: ok. those appear to be copies of glibc headers?01:29
mithroshenki: I think I've fixed the or1k issue01:30
mithroshenki: "export CPU=or1k" - just making the firmware work01:32
jeaCarlFK: sorry, had to run to a room01:35
jeaThanks for that info. I'll try it out now. I didn't realise I could just do h:m:s, so that will be much simpler01:35
shenkimithro: are you suggesting i switch back to the ork1?01:36
mithroshenki: yes, I think I am01:37
mithroshenki: Hrm, going to ork1 still seems to be missing alloca and setjmp.h ...01:39
shenkimithro: ok. we need to build a toolchain with newlib01:39
shenkimithro: it implements these things01:40
shenkimithro: for the lm32 at aleast01:40
tpbTitle: conda-hdmi2usb-packages/ at master · timvideos/conda-hdmi2usb-packages · GitHub (at
mithroshenki: the ork1 toolchain should be newlib already?01:42
shenkimithro: ok01:42
mithroshenki: The opsis-soc firmware compiles under lm32 and or1k now01:45
shenkiUse make V=1 or set BUILD_VERBOSE in your environment to increase build verbosity.01:47
shenkiLINK build/firmware.elf01:47
shenkior1k-elf-ld: warning: cannot find entry symbol _start; defaulting to 000000004000000001:47
shenkibuild/py/nlrsetjmp.o: In function `nlr_setjmp_jump':01:47
shenkinlrsetjmp.c:(.text+0x28): undefined reference to `longjmp'01:47
shenkinlrsetjmp.c:(.text+0x28): relocation truncated to fit: R_OR1K_INSN_REL_26 against undefined symbol `longjmp'01:47
shenkibuild/py/gc.o: In function `gc_init':01:47
mithroshenki: pastebin the full thing?01:59
shenki$ grep -r __divsi3 conda/lib/gcc/or1k-elf/4.9.3/02:02
shenkiBinary file conda/lib/gcc/or1k-elf/4.9.3/libgcc.a matches02:02
shenkiBinary file conda/lib/gcc/or1k-elf/4.9.3/soft-float/libgcc.a matches02:02
shenkiBinary file conda/lib/gcc/or1k-elf/4.9.3/no-delay/libgcc.a matches02:02
shenkiBinary file conda/lib/gcc/or1k-elf/4.9.3/no-delay/soft-float/libgcc.a matches02:02
shenkiBinary file conda/lib/gcc/or1k-elf/4.9.3/compat-delay/libgcc.a matches02:02
shenkiBinary file conda/lib/gcc/or1k-elf/4.9.3/compat-delay/soft-float/libgcc.a matches02:02
shenkiso those __div functions exist02:02
shenkimithro: but we're not linking them in02:02
mithroshenki: Do we just need something like -lgcc ?02:02
shenkimithro: yes02:04
shenkimithro: which you normally get by linking with the compiler frontend02:04
shenkimithro: when i do that,
mithroSo, ork1 has a crt002:06
shenkiright .and we normall don't link against it becuase we pass -nostdlib to the linker02:06
shenkiand we link with ld, not with gcc02:06
tpbTitle: Pastebin | IRCCloud (at
mithroNeed to spend some time on my own lightning talk02:07
tpbTitle: Fixing #34 by mithro · Pull Request #35 · timvideos/HDMI2USB-mode-switch · GitHub (at
shenkimithro: where would i find the source for the or1k-elf/lib/crt0.o?02:13
mithroshenki: Is that binutils or gcc?02:14
mithroshenki: - or1k-4.9.3-release02:14
tpbTitle: GitHub - openrisc/or1k-gcc: GCC port for OpenRISC 1000 (at
shenkimithro: gcc i think02:14
tpbTitle: conda-hdmi2usb-packages/meta.yaml at master · timvideos/conda-hdmi2usb-packages · GitHub (at
shenkimithro: do you have the build tree on your laptop?02:19
mithroshenki: No, I don't think I do - mostly just use travis CI for doing the builds02:20
tpbTitle: or1k-elf toolchain (newlib port for OpenRISC) (at
tpbTitle: OpenRISC GNU tool chain - OR1K :: OpenCores (at
CarlFKmithro: you fixed the assert 0==0, but I am still getting  f<>002:52
CarlFKAssertionError: 'b'f'' (b'f') != 'b'\x00'' (b'\x00')02:52
tpbTitle: Ubuntu Pastebin (at
*** Bertl is now known as Bertl_zZ04:16
shenkimithro: what were you mentioning at lunch about toolchain 'hello world' programs?04:16
mithroshenki: ?04:17
tpbTitle: or1k-elf toolchain (newlib port for OpenRISC) (at
shenkimithro: oh cool. they have prebuilt toolchians04:23
shenkimithro: bug is in our linker script04:37
shenkimithro: it links if i don't use it04:37
mithroshenki: compare the linker scripts?04:38
shenkithis is the default one
mithrobtw are you using CROSS=1 ?04:46
shenkithat doesn't exist in my makefile04:47
shenkimithro: do you think we could use the linker script as-is, but swap out 0x15000000 for main_ram?04:48
mithroshenki: that is in the micropython Makefile?04:48
jeaCarlFK: you around still?05:21
CarlFKjea: yep05:21
jeawe just tried running and hit some syntax issues05:21
jeado you know of anything that has changed in that area?05:22
shenkimithro: can you load this on to a board?05:22
jeaCarlFK: the syntax error is in the googleapiclient module, which seems unusual05:23
CarlFKjea:  I bet it is a pthon3 thing.. just a sec...05:23
jeayeah, we were thinking that. just surprised that it was an issue in the module05:24
CarlFKif you can find a py3 version, great.  but here is my fix (kinda scary, but really nothing is at risk;05:25
jeaah, it looks like an old version of that package. maybe i should just upgrade it05:25
CarlFKcd $(python3 -c "from distutils.sysconfig import get_python_lib; print( get_python_lib())")05:26
CarlFK2to3 -w steve vidscraper05:26
CarlFK(thats from my install script, I thought there were the goog stuff too05:26
CarlFKcd.. gets you into the site_packages dir (or dist_packages)05:27
CarlFK2to3 -w ... ports the code!05:27
jeaok, lets give that a go05:27
jeaCarlFK: is there a reason you are using git+
tpbTitle: GitHub - CarlFK/google-api-python-client (at
CarlFKjea: I think because I tweeked the open().read() to take a file handle instead of a file name05:41
CarlFKbut that isn't needed now05:41
CarlFKso that is the reason, but it is no longer a valid reason.05:42
jeaI'll switch to the standard one so i have the py3 support05:42
jeanearly there. now to work out why google thinks the file is invalid05:57
jeaah, this might be the issue you were mentioning about open05:59
CarlFKoh right05:59
shenkimithro: ive finished it06:11
shenkimithro: i need to test it06:11
jeaok, i understand what is happening now. I'll patch it later06:11
CarlFKjea: it used to do a little ascii progress bar, but that broke06:12
*** CarlFK has left #timvideos06:12
*** CarlFK has joined #timvideos06:14
*** ChanServ sets mode: +v CarlFK06:14
CarlFKjea:  when you are rooting around, if you see how to do a call back, let me know.06:14
*** cr1901 has joined #timvideos06:41
*** cr1901 has quit IRC06:41
*** rohitksingh has joined #timvideos09:12
*** CarlFK has quit IRC09:18
*** rohitksingh has quit IRC10:14
*** rohitksingh has joined #timvideos10:17
*** rohitksingh has quit IRC10:49
*** CarlFK has joined #timvideos11:04
*** ChanServ sets mode: +v CarlFK11:04
*** Bertl_zZ is now known as Bertl13:32
*** rohitksingh has joined #timvideos15:31
*** springermac_ has joined #timvideos16:27
*** springermac is now known as Guest5725216:27
*** Guest57252 has quit IRC16:30
*** springermac__ has joined #timvideos16:30
*** springermac__ is now known as springermac16:31
*** springermac_ has quit IRC16:32
*** springermac_ has joined #timvideos17:11
*** springermac is now known as Guest5720317:11
*** Guest57203 has quit IRC17:13
*** springermac__ has joined #timvideos17:13
*** springermac__ is now known as springermac17:13
*** springermac_ has quit IRC17:15
*** Bertl is now known as Bertl_oO17:59
*** rohitksingh has quit IRC18:08
*** rohitksingh has joined #timvideos18:23
*** rohitksingh has quit IRC20:25

Generated by 2.13.1 by Marius Gedminas - find it at!