IRC logs for #openrisc Thursday, 2014-04-24

--- Log opened Thu Apr 24 00:00:19 2014
stekernblueCmd: great, on both03:50
stekernblueCmd: as for the 'set locale' problem you mentioned, do you have the locales set up?04:48
olofk_Isn't journalctl responsible for logging now?05:13
olofk_Do you really need a separate syslogger?05:14
stekernhmmm... as far as I can understand, the systemd logger isn't running since the /dev/log socket wasn't created05:36
stekernbut I have been happily ignorant of systemd all together up until now, so I'm not to familiar with how it works05:40
stekernit's something wrong with my finger, it just refuses to do double-o's05:42
stekerncould be the brain that's faulty too...05:43
olofk_Just discovered draw.io05:47
stekern...and my /dev/log created by rsyslog doesn't seem to work properly05:54
stekernso... I try 'apt-get install systemd'05:59
stekernthat's in cap_validate_magic (called from sys_capget)06:17
stekerncoincidently, when I do 'su stefan' in qemu-user, it complains about 'qemu: Unsupported syscall: 90'. syscall 90 happens to be sys_capget06:18
stekernno idea if the problems are anyhow related06:18
stekernwonder if anyone have spent this much time to get *non*-root access to a machine...06:20
stekern_blueCmd: to answer your question if irssi is usable when running on the board, I don't see any performance issues at least06:46
stekern_it feels like running it from my normal machine06:47
stekern_...except everything is black and white...06:47
stekern_ /SET term_force_colors ON just mad matters worse...06:50
blueCmdright. I'll rebuild dbus anyway, it was a hack to get the build dependencies correct, but I guess it should work (obviously it doesn't)07:27
stekern..or then it's something else07:28
stekerndid you build that with or without the struct padding?07:28
stekernall structs passed to sys_capget seems to be of size *4 though07:29
blueCmdstekern: that was before07:35
blueCmd(pretty sure)07:35
blueCmdstekern: btw, do you think we should assume a hardware FPU? I think it kind of makes sense to require that if the performance gain is big enough07:36
blueCmdrequire for or1k Debian that is07:37
stekernthen I can't run it on mor1kx :(07:45
stekernthe FPU that or1200 have is as big as the cpu too07:45
blueCmdreally? HM07:45
blueCmdscrew floating point then :)07:46
olofk_When I get my parallella board, I can use the Epiphany as a kick-ass FPU for or1k :)08:34
stekernme too08:59
stekernaren't they in the middle of sending out boards now?09:00
olofk_Yeah. I think it should arrive any day now09:41
olofk_Hmm.. is Zynq supported in ISE, or do I need to make a Vivado for FuseSoC backend first=09:42
stekernnot sure, but obviously the artix devices are still supported by ISE09:44
stekerneven my old 13.4 version09:44
olofk_I would guess Zynq is supported too09:48
stekernis it only the kintex stuff that you need it for then?09:50
olofk_Good question09:50
stekernI would like some nice high end board to play with09:50
stekern...but I don't want to spend the $$$ they usually go for09:50
olofk_Yeah. Me too. I've been working a lot with multi gigabit interfaces at work for the past five years, but it would be fun to do some open source work in that area as well09:51
olofk_Like implementing mipi, pci express, usb2/309:51
olofk_Is Kintex the only Xilinx device with stacked wafers?09:52
stekernno idea09:52
stekernsomething like this might be fun:,400,1228&Prod=NETFPGA-1G-CML09:52
olofk_If so, that might be why it requires vivado09:52
olofk_yeah, good ol' netfpga09:52
stekern100MB is still high-end in industrial networks, I'm just waiting for that to change09:53
blueCmdwell, that's a remake of it09:53
blueCmdthe old NetFPGA had like Virtex 3 or something very very old on it09:54
blueCmdthat one has Kin09:54
blueCmdLoneTech: hola09:54
olofk_oh.. just saw that. It's a beefed-up version09:54
stekernthat could be fun too09:55
stekernnice price:
blueCmdmy should be in NYC atm09:57
blueCmdI'm looking forward to that one09:57
LoneTech indicates ISE supports most 7 family stuff, but somewhat device limited for webpack09:58
LoneTechthough iirc their newer reference designs are made only for vivado09:58
blueCmdVivado is the way to go09:58
blueCmdApparently they have stopped updating IP Cores (at least some of them, like the MAC and PHYs) for ISE09:59
stekernbah, proprietary IP-cores, who needs that? ;)10:00
LoneTechmy parallellas are on their way10:03
blueCmdyes, that's the goal10:03
blueCmdos all the way!10:03
olofk_blueCmd: Cool. I've been working with older boards from the same vendor. Are you planning on adding something to the FMC slot?10:06
olofk_And how much does that monster cost?10:06
stekernHTG-K7-PCIE-325-2 : $2,49510:13
stekernHTG-K7-PCIE-410-3 : $3,59510:13
olofk_ahh.. it said on the page10:33
stekernmmm =)10:36
blueCmdolofk_: I'm going to add a 2x QSFP module10:37
blueCmdhopefully be able to have 8x 10 Gbit/s Ethernet outputs10:38
blueCmdstekern: for locale you can just do apt-get install locales and dpkg-reconfigure locales10:44
olofk_Just realized that I'm entitled to the digilent academic price since I'm going to study this autumn11:28
olofk_Maybe I'll get one of these beauties,400,819&Prod=GENESYS11:29
olofk_I mean it's half price if you're a student11:30
blueCmdI want to get something where I can run a lot of or1k-debians and have it as build-cluster!12:07
blueCmdI'm thinking if I can make a ddr3 controller that splits an 8GB DDR3 module into 64x 128MB 'partitions' somehow12:08
stekernwhy does it need to split it?12:55
blueCmdstekern: do you think a multiplex/arbiter would be enough?13:30
stekernwell, it'll boil down to that in the end anyway, won't it?13:32
stekernI mean, at some level you have to arbiter the accesses to the memory13:33
stekernolofk_: where what are you going to study?13:48
olofk_stekern: I'm going to study remote half time while I'm on parental leave13:50
olofk_Think it was Umeå13:50
olofk_Going to read about Linux Device Drivers and wireless sensor networks (=lots of spare time) :)13:50
olofk_And by doing that I'm allowed to put my daughter in day care a few hours a week and still be at home13:51
stekernmore struct padding problems, now the other way around18:47
stekernthe reason why the /dev/log socket connect fails is that sys_connect is called with a len of 112, but sizeof(*sunaddr) = 11018:49
FindetonI am back, just like terminator18:52
blueCmdstekern: is that constructed in libc maybe?19:18
stekernblueCmd: not sure, perhaps19:23
stekerncould be libpam too19:24
stekernif I want to build packages myself, then I should follow the notes after: #### Create sbuild schroot for building packages19:25
stekernis everything there done on the 'host'?19:26
stekernto question 1 and 2?19:27
stekernok, good, then I had understood things right19:27
blueCmdYes, it _might_ be a bit out of date - so let me know if anything doesn't work19:28
blueCmdI'll take a readthrough19:28
blueCmdthe sources, use the snapshot listed in instead19:29
stekernI think one problem I'm having is that my user "stefan" isn't allowed to read /var/lib/dpkg/lock and such19:30
stekernsources.list, instead of...?19:31
blueCmdinstead of the sources listed in the READMe19:31
stekernthe one in ~/.mk-sbuild.sources19:31
blueCmdstekern: ah, NFS related?19:31
stekernno, I mean... that's my host19:32
stekernbare with me, I'm still confused by this stuff...19:32
blueCmdI don't blame you19:32
blueCmdhost = fast x64 computer19:32
stekernI'm at the stage where I'm just blindly follow your instructions and hoping I'll start to understand something myself soon ;)19:33
stekernyes, and in a not chrooted environment19:33
blueCmdsbuild works by setting up a chroot and using qemu-or1k-static as a 'shell'19:33
blueCmdso when running 'sudo schroot -c unstable-or1k -d /tmp' I get a new 'or1k' Debian install every time I run that19:34
blueCmdsbuild will then use that to build packages (or you can just do apt-get source / apt-get build-dep / dpkg-buildpackage from inside the chroot)19:34
stekernok, let's start from the 'sudo schroot -c unstable-or1k -d /tmp', that gives me: No chroots are defined in ‘/etc/schroot/schroot.conf’ or ‘/etc/schroot/chroot.d’19:35
stekern(and the /var/lib/dpkg/lock was when I tried to run 'builder/ bash')19:36
blueCmdstekern: doesn't the first line say something about not carying about the lock error? ;)19:37
blueCmdstekern: for the schroot you first need to do the "mk-sbuild" step19:37
stekernyes, I've done that19:38
stekernfirst line where?19:38
blueCmdfrom build.sh19:38
blueCmdecho "Please ignore any lock related errors below, it's irrelevant"19:39
blueCmdapt-get update $APT_OPTS19:39
stekernheh, ok, you're right19:39
stekernbut to my defense, it was all the way at the top, before a lot of lines with 'Get: bla bla'19:40
stekernso I didn't see it19:41
blueCmdhaha, yes19:43
blueCmdit's more so I don't feel bad to not have it documented :P19:43
stekernso, this is the full output from builder/
stekernso, that InRelease says: Valid-Until: Sun, 06 Apr 2014 20:56:12 UTC19:55
stekerncan I just changed the snapshot archive to something newer?19:57
stekernfor instance?19:57
stekernwhile waiting for an answer, I'll try, what's the worst that can happen? (famous last words)19:59
stekernah, ok, now I understand the comment before the mk-sbuild20:12
stekernit should be read: "run this twice, logout and login after the first time"20:13
stekernnot "run this twice, then login logout" as I did :/20:14
stekernlogout login20:14
_franck__stekern: as you worked on the ps2 core you must be a PS2 expert, don't you ?20:18
_franck__is the ps2 working with all ps2 keyboard you tested ?20:18
_franck__ah the is a module parameter to not reset the keyboard during init....20:24
stekern_franck__: yes, it has worked with *all*20:26
stekernall=1 in this case ;)20:26
_franck__did you use the reset parameter ?20:27
stekernnot sure if I know what you mean20:28
stekernmodule as in Linux module or module as in verilog module?20:29
_franck__is atkbd the driver on top of opencores_ps2 ?20:29
_franck__module as Linux module20:29
stekernah, ok20:30
stekernno, I don't think I touched anything like that20:30
_franck__ok let's see if it change something for me20:31
_franck__no, must be something else. ps2_dat stays at '0' after the driver starts. If I hit a key, I can see the clk line clocking.20:36
_franck__if I don't start linux, when I hit a key I can see clk and dat working good20:36
stekernwhat is the output from linux?20:38
stekernabout the keyboard I mean?20:38
_franck__atkbd serio0: keyboard reset failed on 94000000.ocps220:39
_franck__but I signal tapped it and that's true: the reset doesn't work20:40
blueCmdstekern: status update!20:41
_franck__clk goes low, then dat goes low then clk goes high and dat stays low and nothing20:41
_franck__the keyboard is supposed to drive the clock line20:42
stekernblueCmd: making progress20:48
stekernhaving perl problems due to ubuntu distupgrade now though20:49
stekernprobably because I haven't rebooted this machine...20:49
stekernblueCmd: I get a lot of these now: W: line 13 [unstable-or1k] aliases: Alias ‘unstable-or1k’ already associated with ‘unknown’ chroot21:12
stekernis that normal?21:12
blueCmdstekern: just remove the alias-row in your /etc/schroot/something.d/*.conf21:15
blueCmdit's a hack since the mk-sbuild wants it to be named sid instead of unstable21:15
stekernwhy are there always absurd changes after an ubuntu distupgrade?21:20
stekernsuddenly my clock in the taskbar have changed place21:20
stekernE: There are problems and -y was used without --force-yes21:22
stekernI guess the problem is: "WARNING: The following packages cannot be authenticated!"21:24
olofkBinutils mailing list is mighty weird. I just learned that the apple linker has a switch called -why_live. Now that's a suicidal linker21:26
stekernI don't understand how to get away from that...21:50
blueCmdstekern: is it in sbuild?21:52
blueCmdor what are you doing?21:52
stekernI have no clue what I'm doing, hasn't that became evident by now ;)21:54
stekernbuilder/ bash21:54
stekernthat's what I'm trying to get working21:54
blueCmdcan I have the console output?21:54
blueCmdsudo schroot -c source:unstable-or1k -d /tmp21:58
blueCmd# Add these lines to /etc/apt/apt.conf.d/80bootstrap21:58
blueCmdAcquire::Check-Valid-Until false;21:58
blueCmdAPT::Get::AllowUnauthenticated true;21:58
stekernah.. ok, that answer my other question way back in the log22:04
stekernyour github instructions are missing the source in 'sudo schroot -c source:unstable-or1k -d /tmp' btw22:06
stekernok, now it's building22:14
blueCmdstekern: where?22:30
blueCmdstekern: what email do you want to be listed in the maintainers file for ibntuils?22:39
blueCmdthat is the address you are using on github, so I guess that's ok22:49
blueCmdah, right - I probably need to get an account first xD22:54
blueCmdstekern: you should fill in as well22:54
blueCmdHah, I just git push:ed to binutils.23:05
blueCmdstekern: you have a new version of libc in - I just rebuilt it, try that one and see if it solves your sys_connect issues23:17
--- Log closed Fri Apr 25 00:00:20 2014

Generated by 2.15.2 by Marius Gedminas - find it at!