| --- Log opened Mon Dec 29 00:00:14 2014 | ||
| ytreza | Hello, | 07:23 |
|---|---|---|
| ytreza | I'm on ubuntu 14.04. I would like to cross compile to or1k. I downloaded the toolchain binaries http://opencores.org/or1k/OpenRISC_GNU_tool_chain, https://github.com/juliusbaxter/mor1kx-dev-env/wiki/OpenRISC-tool-chain-installation-guid. But when i try to compile, i says "/opt/or1k-toolchain/libexec/gcc/or1k-elf/4.8.0/cc1: error while loading shared libraries: libmpc.so.2: cannot open shared object file: No such file or directory". | 07:28 |
| ytreza | I install libmpc but still doesn't compile. Can someone help me please ? Thank you. | 07:28 |
| olofk | juliusb: Could you update your mor1kx-dev-env to point to the official build instructions? It looks like several people use this as an entry point, but I guess it's a bit outdated | 11:16 |
| olofk | Nice guides btw. I haven't looked at them before | 11:23 |
| olofk | I just tried to build upstream binutils 2.25 with only --target=or1k-elf and --prefix=/opt/openrisc. It builds just fine. Does this mean that I won't need all the extra configure options in the build instructions for stage 1 or1k-src ? | 11:40 |
| olofk | Can I for example check if this was built with --enable-shared ? | 11:41 |
| rah | https://www.youtube.com/watch?v=8-OR0QppWKA&feature=youtu.be&t=28m23s | 13:19 |
| rah | anyone seen this? | 13:19 |
| -!- FreezingAlt is now known as FreezingCold | 13:31 | |
| olofk | rah: Cool. | 15:15 |
| olofk | I didn't know that the CCC talks were available already | 15:15 |
| juliusb | olofk: sure. which official build instructions, though? maybe I should just delete it and point to the orpsoc-cores guide? | 16:03 |
| juliusb | What about this: https://github.com/juliusbaxter/mor1kx-dev-env/blob/master/README | 16:13 |
| juliusb | Ah I think I should also add a link to fusesoc | 16:17 |
| olofk | juliusb: The fusesoc/orpsoc-cores repos don't contain very much info | 16:17 |
| juliusb | OK well I'll leave it then, as it will now point to some reasonably up-to-date guides | 16:18 |
| _franck_ | olofk: could you please look at my fusesoc pull requests when you have some spare time ? | 16:51 |
| olofk | _franck_: I'm looking at this commit https://github.com/fjullien/fusesoc/commit/81a87a4db8babe9f1d728c63b3769a8960bb892f | 21:09 |
| olofk | Can't remember exactly how this was supposed to work, but I get the feeling that this line is for picking up libs in the dependencies of the top-level core | 21:10 |
| _franck_ | to be honest, I didnt investigate that much. I just saw there was two times what was needed and so I just fixed it | 21:11 |
| _franck_ | not sure that's the good way to do it | 21:11 |
| olofk | Maybe line 44 should be removed instead | 21:11 |
| olofk | I created a monster. I haven't got a fucking clue how this works anymore | 21:12 |
| _franck_ | yeah line 44 seems better | 21:14 |
| _franck_ | :) | 21:14 |
| olofk | Compiling verilator now. I need to do some tests before I decide which line to remove | 21:20 |
| _franck_ | of course | 21:20 |
| olofk | _franck_: Found another issue now. FuseSoC complains that it needs SYSTEMC_LIBDIR and SYSTEMC_INCLUDE even when I'm running a regular C++ model | 21:27 |
| ams | lisp. | 21:28 |
| ams | woops .. | 21:28 |
| olofk | ams: Hey! No cursing in here | 21:28 |
| _franck_ | crap, I should have added a "if self.src_type == 'systemC':" | 21:28 |
| olofk | _franck_: Yeah, I think that should be enough. I'll fix that | 21:28 |
| olofk | I just noticed that it's not very easy to figure out where to add the --verbose flag to fusesoc command line :) | 21:34 |
| _franck_ | I know, I had to try several times before I found out | 21:37 |
| olofk | I remember you said something like that. I understand you completely now :) | 21:37 |
| olofk | I think I go for removing line 44 instead. Not really sure, but it looks more correct | 21:37 |
| olofk | Yep. I tested both now and you patch doesn't work when a dependency needs a lib (like mor1kx-generic depends on elf-loader) | 21:40 |
| olofk | There is an issue with the other patch in that pull request too. It will fail if the same file name is encountered in several directories | 21:44 |
| olofk | OTOH, I'm not too worried about that. It's better than the current situation where I guess that the log files will not be written at all | 21:45 |
| _franck_ | it's worst, it just crash | 21:46 |
| olofk | Why doesn't it always crash? | 21:46 |
| olofk | ah.. all my C files are located in the root dir of their cores | 21:47 |
| _franck_ | exact | 21:47 |
| olofk | I think the best solution would be to just dump everything in a single file. Right now we're producing a lot of empty files | 21:48 |
| _franck_ | FYI, this is my current testing core: https://github.com/fjullien/orpsoc-cores/tree/systemc/cores/test_systemc | 21:49 |
| olofk | That is so cool. It's not much code at all to set up | 21:53 |
| olofk | _franck_: Does gcc print the filename for errors and warnings? | 22:01 |
| _franck_ | olofk: it's working. Just need to clean/improve/test it. | 22:01 |
| _franck_ | olofk: don't remember | 22:01 |
| olofk | I'll do some tests | 22:01 |
| _franck_ | next time I'll open issues instead half/not working pull requests ;) | 22:02 |
| olofk | haha | 22:03 |
| _franck_ | sorry for that | 22:03 |
| olofk | No worries | 22:03 |
| olofk | I'm just happy to have users who want to improve stuff | 22:04 |
| olofk | ok, gcc prints out the filename, so I'm writing std{err,out} to verilator.{gcc,g++}.{err,out}.log instead | 22:05 |
| olofk | And just appends the prints from all compilations | 22:05 |
| _franck_ | ok, I let you work for me. good night | 22:05 |
| olofk | Sounds like a plan. Good night :) | 22:06 |
| --- Log closed Tue Dec 30 00:00:15 2014 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!