| *** tpb has joined #timvideos | 00:00 | |
| cr1901_modern | mithro: I'm working on it... | 00:14 |
|---|---|---|
| cr1901_modern | And I still have to test tinyfpga B2 after this | 00:15 |
| cr1901_modern | I may just add that later if you _really_ need the PR immediately | 00:16 |
| fitzsim | mithro: https://github.com/fitzsim/meta-hdl/tree/hx8k-example | 00:19 |
| tpb | Title: GitHub - fitzsim/meta-hdl at hx8k-example (at github.com) | 00:19 |
| fitzsim | mithro: I found a recipe for a nice simple HX8K example that I'd written at work | 00:19 |
| fitzsim | mithro: try that with SDKMACHINE=x86_64 MACHINE=qemuarm bitbake uart-transmission | 00:20 |
| fitzsim | mithro: if it works for you I'll send a PR to nrossi | 00:21 |
| fitzsim | mithro: once it's built you can load it with the usual iceprog <bit file> | 00:22 |
| fitzsim | mithro: then interact with it over the HX8K eval board UART | 00:22 |
| shorne | CarlFK[m]: other than changing the drive arguments/setup I think it should be OK. but I havent really tested it | 01:12 |
| shorne | have you had success? | 01:12 |
| *** CarlFK has joined #timvideos | 01:13 | |
| *** ChanServ sets mode: +v CarlFK | 01:13 | |
| CarlFK | shorne: um... I'm not sure I would say I have tested it, I got lost in trying to get the build script to behave | 01:24 |
| CarlFK | mithro: I fixed the space/tab problem - and have just promised I would walk someone though the howto 'soon' (days he has to get an Arty...) | 01:25 |
| mithro | CarlFK: I fixed up your changes and merged them | 02:28 |
| CarlFK | mithro: thank you! | 02:29 |
| *** CarlFK has quit IRC | 02:31 | |
| *** CarlFK has joined #timvideos | 02:42 | |
| *** ChanServ sets mode: +v CarlFK | 02:42 | |
| mithro | B2 != BX? | 03:03 |
| mithro | cr1901_modern: I'm happy for you to just do the BX first | 03:03 |
| thaytan | hiyas | 03:08 |
| mithro | hi thaytan! | 03:08 |
| thaytan | mithro, how are things? :) | 03:09 |
| mithro | Busy :-P | 03:09 |
| thaytan | if you weren't, you'd add another project | 03:11 |
| mithro | thaytan: Ha | 03:14 |
| mithro | thaytan: How's thing in the gstreamer? | 03:14 |
| thaytan | mithro, Busy :) | 03:16 |
| thaytan | GStreamer conference is coming up next month in Edinburgh | 03:17 |
| mithro | I fly to Poland before the end of the month | 03:17 |
| mithro | https://orconf.org | 03:17 |
| tpb | Title: ORConf 2018 (at orconf.org) | 03:17 |
| thaytan | that is cool | 03:19 |
| thaytan | I wonder if we could branch out and find work that would justify going to that next time :) | 03:20 |
| mithro | thaytan: FPGA multimedia acceleration! :-P | 03:22 |
| thaytan | mithro, I'd love to get a gig doing that :) | 03:28 |
| CarlFK | shorne: qemu builds, but then errors when ./scripts/build-qemu.sh tries to run it: ... does not support the option 'serial' | 03:44 |
| CarlFK | but we knew that. | 03:44 |
| CarlFK | shorne: tail of script http://paste.ubuntu.com/p/2tmTWqPDkH/ | 03:47 |
| tpb | Title: Ubuntu Pastebin (at paste.ubuntu.com) | 03:47 |
| mithro | FYI - https://elinux.org/images/6/65/An_Overview_of_the_Kernel_DMAEngine_Subsystem.pdf | 03:58 |
| *** rohitksingh_work has joined #timvideos | 04:11 | |
| *** CarlFK has quit IRC | 05:16 | |
| *** CarlFK has joined #timvideos | 06:00 | |
| *** ChanServ sets mode: +v CarlFK | 06:00 | |
| cr1901_modern | GAH, THE FBI HEADER IS KILLING ME | 06:26 |
| cr1901_modern | GNU ld _really_ wants the output to start at the beginning of a memory region, but prepending the checksum and length makes this impossible | 06:27 |
| xobs | ? | 06:27 |
| xobs | What's the issue? | 06:27 |
| cr1901_modern | I need to increment the location counter in a linker script _without_ the addresses I skipped over being FILLed | 06:28 |
| xobs | Linker scripts are a black art. I always have trouble getting them right. And I frequently end up segfaulting ld. | 06:29 |
| xobs | Are you trying to set the location counter to some offset, and have nothing come beforehand? | 06:30 |
| cr1901_modern | Screw it, I'm using a hack (create a dummy output section using ".fbi" and then DISCARD it) | 06:30 |
| cr1901_modern | xobs: Yes | 06:30 |
| cr1901_modern | And the approach I just listed will work, but it will emit ELF symbols over the region I skipped. | 06:30 |
| xobs | I doubt this is helpful, but on Chibitronics, applications start at offset 0x5900, and this is the linker script we use: https://github.com/chibitronics/ltc-compiler-layer/blob/master/support/KL02Z32-app.ld | 06:34 |
| tpb | Title: ltc-compiler-layer/KL02Z32-app.ld at master · chibitronics/ltc-compiler-layer · GitHub (at github.com) | 06:34 |
| cr1901_modern | xobs: https://sourceware.org/binutils/docs/ld/Output-Section-Discarding.html This appears to be the correct method | 06:43 |
| tpb | Title: LD: Output Section Discarding (at sourceware.org) | 06:43 |
| cr1901_modern | http://ix.io/1mrW This being my linker script | 06:44 |
| *** rohitksingh_work has quit IRC | 06:47 | |
| xobs | Oh, I see. You can't define user_flash as being 8 bytes later? | 06:49 |
| cr1901_modern | No, once I have a non-emtpy section, the 8 bytes will get emitted. So I create an empty section that does nothing except increment the location counter. | 06:51 |
| xobs | I think the approach I usually take is the same as yours, though -- emit symbols to fill in the offending section. | 06:51 |
| cr1901_modern | From what I can gather, not all assignments to "." create space in the output file. But an assignment to "." will emit an ELF section with NOBITS type | 06:52 |
| cr1901_modern | xobs: I see, looks like you had the same idea https://github.com/chibitronics/ltc-compiler-layer/blob/master/support/KL02Z32-app.ld#L123-L131 | 07:03 |
| tpb | Title: ltc-compiler-layer/KL02Z32-app.ld at master · chibitronics/ltc-compiler-layer · GitHub (at github.com) | 07:03 |
| xobs | Those were added because I kept segfaulting ld, and discovered the workaround hack was to add those symbols. They didn't seem to affect the output file at all, and I never found anyone else who was having the same problem. | 07:04 |
| xobs | But yeah, the discard section was useful to get the compile size down. | 07:04 |
| xobs | For Chibitronics, the "blink an LED" is mostly 48 bytes of program header, which go at the front of the resulting image, followed by just a few bytes of syscalls. | 07:05 |
| cr1901_modern | God, getting tinyfpga to work initially is a bunch of hacks T_T. Already 2 FIXMEs | 07:06 |
| *** rohitksingh_work has joined #timvideos | 07:07 | |
| cr1901_modern | Gonna have to add a third one to force the use of nextpnr until litex-buildenv has a way to override arguments to plat.build() | 07:08 |
| xobs | The program header is a C struct that gets populated with values that come from the ld script, such as _data, _edata, and _text. So your program could have a "struct loader_header { uint32_t crc; uint32_t addr; } attribute((section(".progheader")));" and define one of those in a support library somewhere. | 07:08 |
| xobs | Ah, or the way we do it in Tomu is https://github.com/im-tomu/tomu-quickstart/blob/master/include/toboot.h#L109 | 07:10 |
| tpb | Title: tomu-quickstart/toboot.h at master · im-tomu/tomu-quickstart · GitHub (at github.com) | 07:10 |
| * cr1901_modern still needs to get a Tomu this century | 07:11 | |
| cr1901_modern | Aaaand arachne-pnr isn't going to route this design, is it?! | 07:11 |
| * cr1901_modern is grumpy tonight | 07:11 | |
| xobs | And then we define the .text section (which comes first) as having a SORT, and make sure the toboot header is always first: https://github.com/im-tomu/tomu-quickstart/blob/master/tomu-efm32hg309.ld#L29 | 07:12 |
| tpb | Title: tomu-quickstart/tomu-efm32hg309.ld at master · im-tomu/tomu-quickstart · GitHub (at github.com) | 07:12 |
| * cr1901_modern strongly suggests adding LITEX_EXTRA_CMDLINE="-Ob use_nextpnr True" to your make command line when using tinyfpga | 07:20 | |
| cr1901_modern | xobs: I'll check in a bit when I'm done throwing out all electronic devices from my home | 07:21 |
| cr1901_modern | tinyfpga: If you're awake, could I ask for a HUGE favor? | 07:33 |
| cr1901_modern | Actually I'll prob just make a PR | 07:34 |
| cr1901_modern | mithro: https://imgur.com/yx0693L Proof that it works. Cleaning up and sending a PR. | 07:49 |
| tpb | Title: Imgur: The magic of the Internet (at imgur.com) | 07:49 |
| *** paddatrapper has quit IRC | 08:00 | |
| *** paddatrapper has joined #timvideos | 08:10 | |
| *** paddatrapper has joined #timvideos | 08:14 | |
| cr1901_modern | mithro: https://github.com/timvideos/litex-buildenv/pull/55 | 08:50 |
| tpb | Title: Add TinyFPGA BX Platform by cr1901 · Pull Request #55 · timvideos/litex-buildenv · GitHub (at github.com) | 08:50 |
| *** twoolie has joined #timvideos | 09:48 | |
| *** Shari2 has joined #timvideos | 10:09 | |
| *** twoolie has quit IRC | 10:14 | |
| *** Shari2 has quit IRC | 11:11 | |
| *** rohitksingh_work has quit IRC | 12:35 | |
| *** twoolie has joined #timvideos | 13:32 | |
| *** rohitksingh has joined #timvideos | 13:42 | |
| *** rohitksingh has quit IRC | 13:46 | |
| *** rohitksingh has joined #timvideos | 14:06 | |
| *** CuriousLearner has quit IRC | 14:15 | |
| *** CuriousLearner has joined #timvideos | 14:18 | |
| *** rohitksingh has quit IRC | 14:30 | |
| *** waldo323 has quit IRC | 14:31 | |
| *** rohitksingh has joined #timvideos | 14:59 | |
| *** waldo323 has joined #timvideos | 14:59 | |
| mithro | cr1901_modern: Why is it yellow? | 15:02 |
| *** waldo323 has quit IRC | 15:09 | |
| *** twoolie has quit IRC | 15:10 | |
| *** CuriousLearner has quit IRC | 15:15 | |
| mithro | cr1901_modern: Comments added | 15:17 |
| *** CuriousLearner has joined #timvideos | 15:20 | |
| *** waldo323 has joined #timvideos | 15:20 | |
| *** rohitksingh has quit IRC | 15:37 | |
| *** CuriousLearner has quit IRC | 15:40 | |
| *** CuriousLearner has joined #timvideos | 15:43 | |
| *** swalladge has quit IRC | 15:52 | |
| *** swalladge has joined #timvideos | 16:27 | |
| CarlFK | mithro: do you know about this, qemu-system-or1k -M litex ... if=mtd,format=qcow2,file=build/arty_net_or1k//qemu.qcow2,serial=n25q128a13 ... Block format 'qcow2' does not support the option 'serial' | 17:01 |
| CarlFK | mithro: more context: http://paste.ubuntu.com/p/2PbXpgwknG/ | 17:01 |
| tpb | Title: Ubuntu Pastebin (at paste.ubuntu.com) | 17:01 |
| CarlFK | mithro: thats from new qemu from shornes's repo: export QEMU_REMOTE=https://github.com/stffrdhrn/qemu.git | 17:05 |
| tpb | Title: GitHub - stffrdhrn/qemu: Official QEMU mirror. Please see http://wiki.qemu.org/Contribute/SubmitAPatch for how to submit changes to QEMU. Pull Requests are ignored. (at github.com) | 17:05 |
| mithro | CarlFK: dunno - that serial was used to provide the flash chip to emulate -- so maybe the option has just changed names? | 17:10 |
| CarlFK | mithro: something like that - shorne was aware of "it changed" and took a guess at what might work, and id didn't | 17:12 |
| CarlFK | I'll open an issue | 17:12 |
| *** tac-tics has quit IRC | 17:29 | |
| *** tac-tics has joined #timvideos | 17:43 | |
| cr1901_modern | mithro: Idk why it's yellow... prob my terminal application | 19:26 |
| mithro | fitzsim: "NOTE: Tasks Summary: Attempted 500 tasks of which 500 didn't need to be rerun and all succeeded." | 19:54 |
| *** Kripton has quit IRC | 20:59 | |
| *** Kripton has joined #timvideos | 21:03 | |
| cr1901_modern | mithro: https://github.com/timvideos/litex-buildenv/pull/55#discussion_r216706869 What do you mean by this? | 21:14 |
| tpb | Title: Add TinyFPGA BX Platform by cr1901 · Pull Request #55 · timvideos/litex-buildenv · GitHub (at github.com) | 21:14 |
| *** tac-tics has quit IRC | 22:41 | |
| mithro | cr1901_modern: I mean there should just be switches which modify the template in some way rather than having to override whole template | 23:04 |
| *** twoolie has joined #timvideos | 23:10 | |
| *** twoolie has quit IRC | 23:15 | |
| *** twoolie has joined #timvideos | 23:20 | |
| *** mauz555 has joined #timvideos | 23:24 | |
| fitzsim | mithro: ok, I guess uart-transmission succeeded, then? | 23:39 |
| fitzsim | mithro: it looks like you re-ran it to make sure, but nothing needed to be re-run | 23:39 |
| fitzsim | mithro: that's a great thing about Yocto is that its dependency tracking is probably the best I've seen; it only rebuilds stuff when needed, and it seems to always rebuild everything that needs to be rebuilt | 23:40 |
| fitzsim | in comparison, buildroot seems simpler, but less sophisticated (Makefiles and stamp files) although I've only used buildroot a little bit | 23:40 |
| cr1901_modern | mithro: I don't know how we would inject such options. Additionally, I figured that the whole point of the template is that you can manually override it if needed | 23:54 |
| cr1901_modern | having the backend fill in some replacement vars | 23:54 |
| cr1901_modern | mithro: I added #ifdef MAIN_RAM_BASE b/c I got a compile failure w/o it. Will check | 23:55 |
| *** twoolie has quit IRC | 23:58 | |
| *** mauz555 has quit IRC | 23:58 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!