IRC logs for #openrisc Sunday, 2016-03-13

--- Log opened Sun Mar 13 00:00:32 2016
wallentoshorne: thanks, I will have a look03:35
bandvig_juliusb: Julius, IRC log doesn't work :(05:16
wallentoshorne: fixed in https://github.com/openrisc/binutils-gdb/commit/fe709fb78f7336c1eb1f44a5a6f100e523c86b6d05:40
wallentoI have added travis builds to binutils-gdb: https://travis-ci.org/openrisc/binutils-gdb06:05
wallentoAnd made a release of gdb 7.11: https://github.com/openrisc/binutils-gdb/tree/gdb-7.11-or1k06:05
wallentoshorne: how was it with gcc 5.2?06:52
wallentoyou missed the branch of blueCmd or was yours different/better?06:52
shornewallento: I just did install, I didnt see any issues08:24
shorneI started it up and it works ok08:25
shorneI didnt see the blueCmd branch08:25
shornewallento: where did you see the blueCmd branch?08:29
wallentoshorne: https://github.com/openrisc/or1k-gcc/tree/or1k-5.2.008:35
wallentohere08:35
shorneI was not using that08:37
shorneor1k-elf-gcc -v08:37
shornegcc version 4.9.2 (GCC)08:38
shornegcc -v08:38
shornegcc version 5.3.1 20151207 (Red Hat 5.3.1-2) (GCC)08:38
shorneI think when I build gdb it uses local gcc to x86. What does the gdb build need or1k-elf- for? ( I know it needs it, but not clear what its for )08:41
wallentoIt doesn't need it08:46
wallentoI roughly remembered you also ported gcc to 5.2, or not?08:46
shorneyes, so my host compiler is 5.3 (which needed the updates)08:47
shornethe target it the or1k-elf chain which is 4.908:47
wallentoah, okay08:48
wallentoI got confused then08:48
shornethat was no problem08:48
shorneI guess I am a bit too, I should probably try to update my or1k-elf- chain too08:48
wallentoI just rebuilt a newlib toolchain release08:49
wallentogcc 4.9.2, binutils 2.26 from upstream, newlib 2.3.0 with or1k backports and gdb 7.11 with or1k patch08:50
wallentoIf gcc 5.2 is ready we probably should merge it and I will build that too08:50
shorneis the or1k-gcc github based off of upstream git?08:51
shornei.e. is it easy to merge?08:52
shorneI see the bluecmd branch08:52
shorneneed to go for a bit..08:52
wallentoI think it is, but I will clarify it08:54
olofkwallento: Can we upstream gdb before gcc? I don't think there are any technical reasons to hold it back otherwise09:02
olofkHmm...or is sim using or1ksim? I vaguely remember some circular dependency there09:03
olofkDoes anyone know if it's possible to set top-level verilog parameters in verilator?09:03
wallentoit is not09:03
wallentoverilator09:04
olofkThat sucks :/09:04
wallentoI put this up as an issue a while ago, but the compilation forbids that09:04
wallentothats why we put the core id as port for verilated models09:04
wallentowhat do you want to do?09:04
wallentomost stuff can be done as ports fortunately09:04
olofkI have made some FuseSoC improvements so that you can easily set parameters for build targets and the other sims09:04
olofkYeah, that's true, but some things you can't09:05
wallentoah, you mean at compilation time?09:05
olofktep09:05
olofkyep09:05
olofkplusargs work fine in verilator09:05
wallentothere it should be possible09:05
olofkI need it to pass initial rom contents and other stuff09:06
wallentoI think you still need a wrapper for this09:07
olofkThat's the way we do it now, but using  a parameter would make it easy to handle it the same way for all targets09:08
wallentobut you are right, technically it should not be too hard to set parameters at compile time09:08
olofkThe other sims support it, so I find it strange that verilator doesn't09:08
wallentosame as you set defines09:09
wallentoand there is your Sunday project ;)09:09
olofkwallento: :) http://www.veripool.org/boards/2/topics/276-Verilator-Efficient-Usage-of-Verilog-Parameters09:10
olofkSmall world :)09:10
wallentoyeah, that was it09:10
wallentoregarding gdb I am not sure, in contributing it does not say much09:11
wallentowith respect to new architectures09:12
olofkhaha http://www.veripool.org/boards/2/topics/918-Verilator-Parameter-in-package-used-for-signal-width09:12
olofkAre we the only users of verilator? :)09:12
wallentofind one of Jeremy and you get a Bingo09:12
wallentoI can have a look at the code to estimate the complexity of adding +parameter+name=value09:13
olofkwallento: :)09:13
olofkHighly appreciated if you have time to look at that09:14
olofkIt really shouldn't be too hard I guess09:14
wallentoI will later, but want to finish an updated tutorial now09:15
wallentoI will create a repo openrisc/tutorials, okay?09:16
olofkGreat09:16
wallentogot the ORCONF2013 one running again09:16
olofkThey need a lot of love09:16
olofkah cool09:16
olofkPeople seem to use that one quite a lot09:16
wallentoyep, I will put up a note there also09:16
wallentoand provide prebuilt-versions09:16
wallentoas a quick start09:17
olofkgreat. It's a lot of steps to get stuff running, so have some known good images helps09:18
wallentoolofk: What's the plusarg in other simulators?11:36
wallentoI think I have drafted the feature now, will push soon for testing and ask Wilson if I am going in the proper direction11:51
mafmblueCmd: do the Debian repos of or1k still exist somewhere?12:26
mafm(or anybody who knows, of course)14:28
shornewallento: you mention you got orconf2013 running again, thats the tutorial I used. but had to change a few things17:45
shornelike use fusesoc17:46
shorneI also used the openrisc/orpsoc-cores since I figured julius's 2013 branch was pretty old17:46
shorneand so the uart pin locations was different17:47
shorneIs it ok to update the mediawiki to be more modern? or you want to create a new one and link to it?17:48
--- Log closed Mon Mar 14 00:00:33 2016

Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!