IRC logs for #openrisc Wednesday, 2016-10-12

--- Log opened Wed Oct 12 00:00:53 2016
olofkDoes anyone have experience with msys2 on windows?03:11
olofkI could never have imagined things would be so complicated03:12
simoncookolofk: Yes, much better than msys104:12
olofksimoncook: I can't really tell if you're serious :)04:12
simoncookOh trust me, I am. Maybe I just did things wrong with msys1 and went with "install everything" and watch everything melt with the wrong versions. These days I use msys2 for an environment for building things on Windows. Well... a version I stuck in a git repo because I'm not setting that up twice :)04:13
olofksimoncook: Then you could perhaps tell me how to build elfutils?04:14
olofkWhich seems to be the new replacement for libelf04:14
simoncookIt looks like this hasn't bundled its configure script, so it looks like you'd need to install autoconf/automake to build the configure script, and then do the autoconf dance of "autoreonf" followed by "automake --add-missing"04:19
simoncookthough depending on whether you're building for msys/mingw (the real complexity), it seems the mingw gcc can't find __thread support, but I'm not sure if I'm missing something there04:20
olofkYes! We're stuck on the __thread thing right nwo04:21
olofkAnd I'm not sure I understand the distinction between msys2 and ming6404:21
olofkmsys2 is an environment for installing mingw?04:21
olofkIs msys2 shell only for installing new msys2 stuff, and everything else should be run from mingw64?04:22
simoncookMy understanding (though this could just be wrong), is that msys2 is the environment and has its own libraries for most things, possibly somewhat closer to cygwin but not that far, whereas mingw* is the toolchain you should use for building things to run natively outside of the mingw environment, so it links against more Windows libraries, etc.04:24
olofkSo where do I want to install elfutils (if I at some point get it working)?04:25
simoncookThat doesn't really matter that much, as you should be able to move most things around and they still work04:29
olofkGood to know. Still haven't got a clue where or why I would want to move anything04:31
olofkOr how to get around the __thread issue04:31
simoncookI've looked at the config.log, and it seems the error is actually the linker complaining it doesn't know what '-z' is, but how that's not in this version of the linker or why that flag is trying to be used I'm not sure, I have older binutils that lists all the -z options in its --help04:31
olofkI found an msys2 package for elfutils04:31
simoncookaah, or just install that, avoids the building issue completely04:32
olofkNo, because I still need to build it04:33
olofkIt's just a PKGBUILD file04:34
simoncookI'm afraid I'm not having much luck narrowing this down. I removed the problematic flags from the and libelf/ to see if I got any further and now I'm hitting errors in "WARNING libc does not have argp"04:43
olofkThanks for trying04:43
mor1kx[mor1kx] bandvig pushed 1 new commit to marocchino_devel:
mor1kxmor1kx/marocchino_devel c7f9e7e Andrey Bacherov: Tomasulo algorithm has been implemented.13:33
bandvigolofk: as I remember, you want to see Tomasulo implementation. It is time.13:34
SMDnotebandvig: great news!14:50
bandvigSMDnote: thanks15:06
SMDnotebandvig: have you measured the perf gain compared to cappuccino?15:10
bandvigBy coremark with LTO: CAPPUCCINO: 123.915737 [i/sec] (2.48 i/mhz) MAROCCHINO: 129.533679 [i/sec] (2.59 i/mhz)15:14
bandvigmuch more improvement is for FPU operation15:14
bandvigbut I haven't got exactly digits right now15:14
-!- Netsplit *.net <-> *.split quits: imphil, Hoolootwo, SMDnote, LoneTech17:19
-!- Netsplit *.net <-> *.split quits: rokka, Amadiro17:23
-!- Netsplit *.net <-> *.split quits: bentley`17:24
-!- Netsplit *.net <-> *.split quits: bentley`19:00
-!- Netsplit over, joins: bentley`19:15
-!- Hooloo42 is now known as Hoolootwo21:59
--- Log closed Thu Oct 13 00:00:54 2016

Generated by 2.15.2 by Marius Gedminas - find it at!