IRC logs for #openrisc Thursday, 2012-02-09

kristianpaulBtw what can you actually do with the new fpga cycle iv e board?05:43
kristianpaullike run a webserver?05:43
-!- Netsplit *.net <-> *.split quits: Blok08:47
-!- Netsplit over, joins: Blok08:51
-!- slpz is now known as slpz_10:56
Lampus_I can't compile gcc with c++13:43
Lampus_Compilation problem was occured with next error message: /libstdc++-v3/configure: line 269: retu rn: please: numeric argument required13:43
Lampus_configure: error: no 8-bit type13:44
Lampus_I'm used git version of gcc from here: git://
Lampus_Nobody here14:03
jonibotry the or1k-4.5.1 branch14:51
Lampus_jonibo: Already tried it. With smae result. But without  c++ build was successful14:53
jonibonot sure what's going on there then14:54
jonibono time to look into right now, either, i'm afraid14:54
Lampus_jonibo: seems this bug bug
Lampus_And exist patch resolving this problem:
jonibofrom 2010? somebody else must have seen this already...14:56
Lampus_jonibo: But I can't understant how to regenerate configure scripts14:56
jonibowhat distro are you on?14:57
Lampus_jonibo: ArchLinux x86_6414:57
jonibomaybe that's why...14:58
jonibodon't know if anybody else is running arch14:58
jonibohaven't seen that error on ubuntu14:58
Lampus_I'm trying create PKGBUILD's for packaging of toolchain14:58
Lampus_in arch latest versions of packages, so sometimes you can see many strange things =)14:59
joniboyeah, I know... pretty bleeding edge15:00
jonibothat's good... you can shake out bugs for us15:00
juliusbLampus_: Hi, nice to hear you're doing stuff with OpenRISC15:05
juliusbwhat is PKGBUILD?15:05
Lampus_One second...15:05
juliusbhehe :)15:06
juliusbfair cop15:07
juliusbOK, Arch Linux's package format then15:07
Lampus_Yes. Do you have any suggestions/additions to configure parameters?15:08
jonibocan you use the repository at instead of openrisc.southpole.se15:08
jonibothe version is not up to date15:08
jonibo...and I should probably just delete it15:08
Lampus_I can try it, but only tomorrow15:08
joniboyou don't want --with-sysroot if you're doing --with-newlib...15:09
Lampus_jonibo: ok. Fixed15:11
joniboseems ok15:11
Lampus_$ autoreconf error: Please use exactly Autoconf 2.64 instead of 2.68. config/override.m4:34: _GCC_AUTOCONF_VERSION_CHECK is expanded from...15:12
joniboyeah, you need autoconf 2.6415:12
jonibothat's an upstream gcc dependency15:12
Lampus_I can build package with autoconf in separate directory, like /opt15:14
juliusbI found issues when playing with GCC recently - you need to have exactly the right versions, but I foudn them easy to install under Ubuntu15:16
juliusbyou may already have them Lampus_ - try autoconf2.64 instead of just autoconf15:16
Lampus_Hmm... I can use not-standart prefix for autoconf 2.6415:16
Lampus_yep, seems I already have it =)15:17
juliusbbut I recall it was annoying to do it properly15:18
juliusbyou need to run it twice sometimes15:18
juliusbit'll give you an error the first time but not the second15:18
juliusbthere's a page on the gcc wiki about it15:19
juliusbwhich is actually out of date but you can figure out out roughly what's needed :(15:19
Lampus_Do you have some OpenRISC Quick Start documentation?15:20
Lampus_Oh, I have found some typos in OpenRISC specification draft15:21
juliusbplease submit those to the mailing list15:21
Lampus_[lampus@myhost gcc]$ PATH="/opt/autoconf/2.64/bin:$PATH" autoreconf [lampus@myhost gcc]$ echo $? 015:24
Lampus_Looks good15:24
derRichardhi! i have a question on openrisc linux support. does linux run on openrisc 1200? or only on the 1000?15:24
joniboonly on the 120015:24
jonibothat's the only implementation around today...15:24
joniboat least, the only freely available one15:24
joniboopenrisc 1000 is the architecture15:25
joniboopenrisc 1200 is the implementation15:25
joniboterrible, I know15:25
derRichardkinda confusing :)15:25
joniboit's probably actually or1200, even, not openrisc 120015:25
derRichardopenrisc support is in linux mainline. what about gcc and binutils?15:25
joniboyeah, linux is the only project where openrisc support is in mainline15:26
jonibou-boot is probably next15:26
jonibobut gcc and binutils are ___FAR___ off15:26
derRichardwhy that?15:26
jonibonobody wants to pick up the ball and run with it15:26
derRichardare there technical or political problems15:26
joniboit should have been submitted for 4.715:27
jonibo... in my opinion15:27
derRichardi'm about to buy this board:,item,915:27
derRichardlet's see how well it works for me :-)15:27
jonibothat'll be lots of fun for you15:27
derRichardi hope so15:28
derRichardwhat kind of userland are you using on openrisc? busybox?15:29
derRichardwith uclibc?15:29
jonibofrom as that's the only one with openrisc support15:29
jonibothat hasn't made it upstream yet, either15:29
joniboand that won't go upstream before gcc does15:30
derRichardi c15:30
jonibosorry, yes, uclibc from openrisc.net15:30
jonibobusybox from upstream works fine15:30
Lampus_And what about PIC and dynamic linking state?15:30
jonibohaha :p15:31
derRichardbtw: are you jonas bonn?15:31
derRichardnice to meet you!15:31
derRichardi saw you posts on the linux kernel mailing list15:31
derRichard(the openrisc merge)15:31
derRichardthis is how i hered the first time about openrisc and free softcores :)15:32
juliusbjonibo: I would say it's a combination of technical and political issues - it's a fair bit of work that we've not got a serious contributor effort wotard15:36
jonibothe bickering channel is #opencores15:38
juliusbwell, come on, if there weren't technical issues it would have been done already, surely, but gcc and binutils aren't the  easiest things to master15:40
juliusbanyway, that's all I'll say, I just want to point out that there's certainly an opening for anyone who wants to come and help with our toolchain15:41
Lampus_I should go, bye15:41
stekernjonibo: u-boot is already upstream16:21
pgavinjuliusb: is there a web page somewhere that lists what needs to be done?16:36
pgavinjuliusb: I'm currently working on merging giuseppi's gcc with the most recent upstream16:36
pgavinjuliusb: and making it work with your binutils version16:37
pgavin(the or1k one)16:37
jonibopgavin: awesome!17:09
jonibowill be happy to set up a git repo for you at if you want to publish there...17:10
joniboi can try to ping giuseppe and ask him to provide you with some info about what needs doing, if you'd like17:11
pgavinok. it seems like gcc's changed quite a bit since he's last merged17:15
pgavinthey replaced some functionality with other stuff and I'm having to figure out what to do lol17:15
joniboyeah, he mentioned something along those lines when we last discussed this17:16
joniboglad you figured that our for yourself!17:16
pgavinit merged cleanly at least, no conflicts17:16
pgavinbut the or1k.c file had some compile errors17:16
joniboyeah, some functions have changed, as I understand it17:16
juliusbpgavin: great to hear18:03
juliusbpgavin: i have also done some work on the GCC port, but based on the gcc-4.5.1 version in the opencores repo18:04
juliusbmainly changing the target names to or1k-*-*18:15
juliusbso if you reckon you could take a 4.5.1 port up to latest, that'd be the one to take, but at the same time I'm not sure how advance guiseppe's port is18:16
pgavinjuliusb: basically I git cloned giuseppe's, and then git pulled from the main gcc tree, then did sed 's,or32,or1k,g'18:24
pgavinnow I'm trying to make it compile lol18:25
juliusbpgavin: nice one :)18:25
juliusbwell if you can make it build with my binutils, then sweet18:25
jonibobest way forward would be to make it work with old binutils, then replace binutils with juliusb's cgen version...18:26
jonibotoo many variables otherwise18:26
jonibogiuseppe's port is way more advanced than the 4.5.1 that's floating around18:26
joniboand then there are other patches floating about that i'm aware of that haven't been integrated into any of these trees18:26
juliusboh wow cool18:27
jonibothis whole thing's a mess18:27
jonibohopefully pgavin can pull it all together18:27
juliusbyes the thing hasn't been coherently tackled by anyone yet18:27
juliusbi'm having a go from ground up18:27
juliusbbut that took a month of my spare time just do the CGEN and binutils stuff18:27
joniboyeah, your cgen works is nice18:27
juliusb(I'm not certain binutils stuff is correct either)18:27
juliusb(still getting some odd linking issues)18:28
juliusbbut we do have the gdb sim now18:28
juliusbwhich means no more having to build or1ksim to run the regression stuff18:28

Generated by 2.15.2 by Marius Gedminas - find it at!