stekern | hmm, the target specific code, especially in the toolchain support is really, really messy in clang | 06:48 |
---|---|---|
stekern | everything is just hardcoded with if-elses for different archs and other corner cases | 06:49 |
stekern | I'm a bit reluctant smacking on even more or1k specific stuff there to get 'or1k-linux-clang hello_world.c -o hello_world' work out of the box | 06:52 |
stekern | (using gcc as a shell for as and ld of course works, but it would be nice to not have the gcc dependency) | 06:59 |
TarekEldeeb | hello all, I've compiled the or32-elf-sim and toolchain, tried a simple test program .. the simulator cannot decode most of the instructions | 15:36 |
TarekEldeeb | 00002070: e4832000 <invalid> (executed) [cycle 149, #130] 00002074: 13fffffe l.bf -2 (next insn) | 15:36 |
TarekEldeeb | What did I miss that caused this behavior ? | 15:37 |
--- Log closed Thu Jun 14 16:48:19 2012 | ||
--- Log opened Thu Jun 14 16:53:52 2012 | ||
-!- Irssi: #openrisc: Total of 14 nicks [0 ops, 0 halfops, 0 voices, 14 normal] | 16:53 | |
-!- Irssi: Join to #openrisc was synced in 24 secs | 16:54 | |
juliusb | mmm, odd stuff from Tarek there | 18:08 |
juliusb | stekern: so currently how does LLVM work, though? It relies on GCC's frontend? | 18:10 |
juliusb | or rather, the work you're doing? | 18:10 |
stekern | juliusb: no, it "relies" on using gcc as a shell for binutils (as and ld) | 20:33 |
stekern | i.e. you pass gcc the assembly and let it pass it on to gas and ld with the right commandline options | 20:34 |
stekern | you could of course just do that manually (and it's done manually in many projects) | 20:34 |
stekern | or you could define a toolchain in clang (that's what I'm looking at now) | 20:35 |
stekern | or actually, there is a "Linux" toolchain already defined, but that's not really suitable for us, so I think I'll just create my own "or1klinux" class and set it up like it's set up in gcc | 20:36 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!