IRC logs for #openrisc Friday, 2014-04-25

--- Log opened Fri Apr 25 00:00:20 2014
stekernblueCmd: ok, let's start with the first question: e-mail: yes, my saunalahti address is what I want to use.03:43
stekern(libc) will try that03:43
stekernthen, it's my turn03:43
stekernso, the build for 'bash' crashed, with not much information: http://pastie.org/910979103:46
stekern150 minutes of inactivity... why are you sitting there idling, stooopid dpkg-buildpackage?03:47
stekernblueCmd: (form) do we have someone that has 'approved request'?04:18
stekern...I should probably sign up to the binutils ML too...04:18
stekernblueCmd: also, to be able to use that 'outdated' release, I needed this: http://pastie.org/911009505:14
stekernthe new libc fixed the sys_connect problem05:42
stekernso.. now I have a log: http://pastie.org/911028105:43
stekernnot of much more help05:43
stekernbut at least that sys_connect bug got slaughtered as collatoral damage in the quest for the su bug05:46
stekernthe gcc you are using now to (re)build packages, does that have your latest __tls_get_address fixes?05:47
stekern...and when I'm at it pointing out stuff in your instructions ;)06:04
stekern'cp modded-sbuild/qemu-bootstrap /usr/sbin/' should be 'cp modded-sbuild/qemu-debootstrap /usr/sbin/'06:05
stekern..and '/debootstrap/debootstrap --second-stage' will fail since the instructions you added yesterday can't be performed before the 'mk-sbuild' command, so you have to go back and manually do that in the chrooted environment06:39
* stekern has graduated from 'blind-instruction-follower' to 'not-completely-helpless-n00b'06:41
stekernrunning '/debootstrap/debootstrap --second-stage' manually doesn't really fly, there's more stuff in mk-sbuild that should be performed after that, a better fix: http://pastie.org/911121007:08
stekern...maybe I can just do pull requests with my suggested changes instead of just pointing them out here, it'll probably make your life a lot easier...07:12
stekernolofk_: (studies) sounds like a deal made in heaven ;)07:50
olofk_stekern: Yes. Unless I have misunderstood something, I think it will work out great07:51
olofk_But I still feel suspicious. There has to be a catch somewhere :)07:52
stekernremote studies though, you'll miss out on moving away from home living the wild student life ;)07:52
stekernreminds me of something Henrik Schyffert said in a stand-up gig: "The military duties are done in completely the wrong phase in life, if someone would come now and tell you that you are going out in the woods with a couple of other guys and shoot with guns, you'd just be... *hug*"07:56
olofk_haha08:31
blueCmdstekern: sorry sorry sorry! i ment to tell you about the fakeroot bug08:41
blueCmdby default fakeroot just hangs08:41
blueCmdsomething withIPC08:41
blueCmddid latest libc solve the hang? if not, you have to use https://github.com/bluecmd/or1k-debian/issues/108:42
stekernhaha, no worries, you haven't wasted more than a couple CPU cycles on my WS =)08:42
stekernthe new libc doesn't get installed by default in the schroot, right?08:43
blueCmd(form)approver is nickc@redhat.com08:43
stekernso I'll need to install them manually there too?08:43
blueCmdstekern: no, if it appears to work you can installit in the source08:43
blueCmdi haven't dared installing it in the repository yet until I can verify that it doesn't break everything08:43
stekernok, I'll test with that first08:44
blueCmdas for a pull request, yes that would be very sweet08:45
stekernpull request - already done08:46
stekernat least for the APT thinig08:46
stekernI'll do the minor README fixes later too08:47
blueCmdthanks! merged08:54
blueCmdpushing upstream to binutils was very smooth, just like github basically - I think it's great that we have that upstreamed08:54
olofk_blueCmd: So it's officially there now?09:01
stekernnew libc didn't change things09:07
stekernthe other fakeroot works though09:07
blueCmdolofk_: yep!09:25
stekernblueCmd: this question probably got lost in my morning burst - do you have the __tls_get_addr fix when (re)building packages now?09:31
stekerne.g. is the new libc built with that?09:31
olofk_pdflatex guru needed10:00
stekern...and follow up question to that, how would I go on to build the gcc debs myself?10:17
blueCmdstekern: yes I have10:46
stekernas tony tiger would have said: G-r-r-reat!11:00
stekernscummvm isn't so easy to build...11:06
blueCmdstekern: what's the build-deps on it?11:08
blueCmddoesn't look too bad11:08
blueCmdlibfluidsynth-dev is missing, and that needs cmake11:09
blueCmdaah11:09
blueCmdI can build cmake now! finally11:09
blueCmd(it depends on qt)11:09
blueCmdit has been a huge pain to be without it11:09
stekernyes... I ended up in qt, then I gave up ;)11:17
stekernbut there was a problem with doxygen too11:17
stekernit needs to be >= 1.8.x something11:17
stekernand we have 1.7.211:17
stekernbut I guess you have that sorted out, since I think it was cmake that went into the doxygen dep11:18
stekernI went on to try to build emacs23 instead11:19
stekern...but having qt, that's nice11:20
stekernis it only the atomics that are arch specific?11:21
stekernI briefly looked at it way back, and I remember the atomics11:21
blueCmdatomics in qt is currently using pthread mutexes11:29
blueCmdbecause we havent added or1k specifics to it11:29
blueCmdfor doxygen thats just an equiv (empty package) used to trick sbuild that it actually has doxygen11:30
blueCmdbuilding the real one now11:30
stekernyeah, I saw the issue about qt atomics11:32
stekern...but the question really was, is there no other things that are arch specific? i.e. there are generic implementations for everything else?11:32
blueCmdin debian? a few things are specific11:51
blueCmdlibatomic, libgc, libffi mainly11:51
blueCmdthere are some bugs marked as "port" in or1k-debian11:51
stekernno, I meant in qt itself11:51
blueCmdyes, only atomics11:52
blueCmdafaik11:52
stekernok11:52
stekernso, in my learning experiment to build emacs23, I've got the build-dep of flex-old covered11:53
stekernthen it needs xaw3d, but that fails when it tries to do: /usr/bin/make -C lib/Xaw3d clean11:54
stekernwhere is /«BUILDDIR»/?11:55
blueCmdasd12:03
blueCmdah, /build usually, but AFAIK it's thrown away when sbuild exits12:03
blueCmdhah, qt4 doesn't really look 100% correct: http://openrisc.debian.net/tmp/2014-04-25-135622_1024x768_scrot.png but it's "close enough" for now :)12:03
LoneTechit looks appropriately experimental :)12:07
stekernblueCmd: yes, it gets thrown away, I'll experiment with apt-get source & friends12:08
blueCmdapt-get build-dep X12:10
blueCmdapt-get source X12:10
stekerncheck and check12:10
blueCmdDEB_BUILD_OPTIONS=nocheck dpkg-buildpackage -b -us -uc12:10
blueCmdif that doesn't work, you apply some fix and want to run it again that command will clean the directory and start over12:10
blueCmdDEB_BUILD_OPTIONS=nocheck dpkg-buildpackage -b -us -uc -nc12:10
blueCmdthat will not12:11
stekernwhat does the 'nocheck' do?12:11
stekernotherwise, 'check' ;)12:11
blueCmdskips "make check" on some packages12:11
stekernah, ok12:11
blueCmdwhich takes forever, and if it uses pthreads will fail anyway12:11
stekern...and cleaning before or not doesn't matter for me right now, since the issue seems to be that the Makefile is missing all together12:13
blueCmdsometimes it gets into a 'not-quite-clean' state where it will try to do 'make clean' before running ./configure12:15
blueCmdthen I usually just run it with -nc12:15
stekernah, there's a Imakefile12:17
blueCmdAh, one of those weird make systems12:18
stekernsomething goes wrong with xmkmf I think12:20
blueCmdRebuild it!12:21
stekernthis is the Makefile it generates: http://pastie.org/911199012:22
stekernno wonder it can't find the 'clean' target in that ;)12:23
blueCmdclearly the file uses the wrong font-color12:27
blueCmdit shouldn't write the text in white with white background, everybody knows that12:28
stekernmaybe it's whitespace source?12:28
* blueCmd doesn't understand :(12:29
stekernhttp://en.wikipedia.org/wiki/Whitespace_(programming_language)12:29
blueCmdah, (without reading the article) is it the one with space and tabs?12:29
stekernand linefeeds12:30
blueCmdhah, nicce12:30
stekerntoday I once again got the honour to start up the old PC from 1995 here12:33
blueCmdhow come?12:33
stekernthere are some conformance tests that uses an ISA board12:33
olofk_:)12:33
blueCmd*whirr* *faster whirr* *whine* *BEEEP*12:34
stekernspot on imitation blueCmd12:34
olofk_blueCmd: I could feel it :)12:34
stekernit's like stepping into a time-machine12:34
olofk_Is it complete with a nice 50Hz CRT as well?12:35
stekernit's funny, the c64 doesn't do that as much, since it doesn't sound so much12:35
stekernamiga 500 floppy sounds otoh...12:35
blueCmdyou need loud harddrives, loud fans and a high-pitched beeper - otherwise it's not the same12:36
stekernolofk_: unfortunately not, it's connected to 22" LCD12:36
blueCmdhttps://www.youtube.com/watch?v=hSJDIGiepgU12:41
blueCmdstekern: doxygen built and added to repo12:52
stekern\o/12:52
blueCmdyes, now everything gcc build-deps on is available! :D12:53
stekernso... I'm trying to follow your advice to rebuild xutils-dev (which contains xmkmf), but it depends on cpp-4.7 (which is not available)12:54
stekernso, did you hack-build that?12:55
blueCmdyes, probably13:11
blueCmdI filed a bug for something like that that they shouldn't build-depend on a specific version like that13:11
blueCmdAH!13:11
blueCmdHah13:11
blueCmdDo you know why they didn't want to build with 4.8?13:12
blueCmdbecause the output produced by the binary was just an empty file!13:12
blueCmdI recall that now13:12
blueCmdI have a fix for that somewhere13:12
stekernoutput by xmkmf?13:13
blueCmdhttps://bugs.debian.org/cgi-bin/bugreport.cgi?bug=74243613:13
blueCmdyes13:13
stekernyeah, it's not empty, there's some garbage in there ;)13:13
blueCmdit's not garbage, it's comments!13:13
blueCmdstekern: cmake added to repo13:16
stekernright you are13:16
stekernbut it looks like garbage without the context13:16
blueCmdyep13:16
olofk_blueCmd: Extremely comprehensive and polite answers you got in that bug report13:16
blueCmdstekern: rebuild with the fix in that, verify and please send me the debs! :)13:17
blueCmdstekern: if you want I can set you up with access to upload debs13:17
blueCmdolofk_: yes, I complained about that13:17
stekernwill do13:17
blueCmdolofk_: actually I dropped of the bug to ask why is that13:17
stekerncool thanks13:18
blueCmdthe answer I got was 'Don't drop off the bug'13:18
blueCmdso I asked the same question in the bug13:18
blueCmdand the answer then was 'See the changelog?'13:18
blueCmdthe changelog says something like 'It doesn't work'13:18
stekernyeah, isn't that a reason?13:18
olofk_:)13:18
blueCmdSure, but it's also shorter to write than 'Don't drop off the bug' + 'See the changelog' :)13:19
olofk_Next time someone reports an issue I will respond "Go fuck yourself"13:19
blueCmdolofk_: that's the Linus-spriti!13:19
olofk_Ah right. Forgot that there is already someone like that :)13:20
blueCmdchecking size of __int64... 013:22
blueCmdhah13:22
stekernisn't that more the systemd spirit?13:22
blueCmdyes - Linus is only that when something is wrong13:23
blueCmdI don't think he's arogant as such13:23
stekernif you would have responded in the bug-report "Go fuck yourself", that'd be more like it I think13:24
blueCmdso PCH is broken in gcc apparently, haven't looked too much on it15:25
blueCmdboth boost and qt has problems with it15:25
stekernsounds like fun16:45
stekernI bet it hasn't got much tested before16:45
blueCmdstekern: olofk: posted http://i.imgur.com/dSowPtR.jpg at work today17:30
stekernheh, wut?17:31
stekernthat's like our response to those shirtless putin photos ;)17:35
stekernit's fun to watch configure sometimes: http://pastie.org/911268117:51
* blueCmd pats his fake_atomics.o17:52
blueCmdlinker complains about missing __sync* or __atomics*? no problem!17:53
blueCmdis it really atomic? well, who needs atomicity anyway?17:53
stekernblueCmd: is it a know issue that pkg-config can't find any modules?18:04
stekernhmm.. it works in my 'real' install18:06
blueCmdnope, that should work18:06
stekernhttp://pastie.org/911273718:11
blueCmdI never used it like that, I just never got an error from ./configure so I assumed it worked :)18:13
stekernah, well, I never used it like that neither. until now, when I get errors from ./configure18:14
stekern=)18:14
stekernbut it's obviously scanning the right path, but doesn't seem to find anything there...18:15
stekernit works in my or1ksim/atlys setup too18:16
stekern...but not with qemu-user in the same rootfs18:17
stekernwtf...18:18
blueCmdyou broke something!18:28
stekernmmm, but it has to be qemu that I broke in that case18:29
blueCmdhas the bootcode support for orpsoc been added?18:53
blueCmdarbitary boot code*18:54
olofkDoes anyone know of an online markdown parser I can use to test my syntax?20:09
olofkBefore I push it to github20:09
olofkFound it!20:09
olofkAh.. but tables is a github extension anyway :/20:11
olofkblueCmd: arbitary boot code?20:33
blueCmdlike uboot20:37
olofkStill not sure what you mean. We have OpenRISC support in uboot and barebox21:30
blueCmdolofk: yes, but I want to be able to embedd uboot in the FPGA image22:00
--- Log closed Sat Apr 26 00:00:21 2014

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