IRC logs for #openrisc Saturday, 2014-06-21

--- Log opened Sat Jun 21 00:00:43 2014
stekerndalias: yeah, some progress, slow but progress still05:55
stekernI (think) I got the TLS things sorted out05:55
stekernbut I saw you are doiing some changes to reloc.h, so I'll probably need to catch up with that too ;)05:56
stekernfrom what I understood, it was to simplify the arch code, so shouldn't be hard I assume05:56
stekernhttps://github.com/skristiansson/musl-or1k/commit/57ac16331223a6ffeece31bbea2f362f6e4990cc05:59
stekern^ "TLS stuff"05:59
stekernI still have things to implement in ldso/or1k and thread/or1k06:01
stekernthat's the current status I think06:01
stekernoh, and look over setjmp/longjmp06:01
daliasstekern, yes, just need to make reloc.h remap the reloc types rather than actually doing the work10:23
-!- rfajardo_ is now known as rfajardo13:22
benjamincburnshello blueCmd - are you around by any chance? Trying to follow your debian boostrap instructions and I’m hitting a snage13:54
benjamincburnssnag, too13:54
benjamincburnswhen I attempt the second stage of the bootstrap (the chroot command) it first said that the qemu command couldn’t be found - I assume because it was missing from the chroot13:55
benjamincburnsso I copied it into the chroot, then ran the command again, and it returned instantly13:55
benjamincburnswith no output :-/13:55
blueCmdbenjamincburns: hi14:09
benjamincburnshello14:09
blueCmdbenjamincburns: the bootstrap instructions might be a bit out of date, I haven't touched it in a month or so14:09
benjamincburnsah14:10
blueCmdsudo chroot initramfs /usr/bin/qemu-or1k-static  -E PATH=/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin bin/bash /debootstrap/debootstrap --second-stage14:10
blueCmdis it that command that fails?14:10
benjamincburnsyep14:10
benjamincburnswell - it fails silently14:10
blueCmdtry using http://openrisc.debian.net/qemu-or1k-static.atomic14:10
blueCmdinstead of http://openrisc.debian.net/qemu-or1k-static14:10
benjamincburnssure - one sec14:10
benjamincburnsthat’s looking better14:11
benjamincburnsI can get it to run bash now - will try the debootstrap command14:12
blueCmdcool!14:12
blueCmdfyi, that one contains the code to decode atomic instructions that was added to openrisc a couple of months ago14:12
blueCmdand libc uses those, so qemu would crash14:12
benjamincburnsI get a few errors now — can’t execute binary file (cat) and can’t work out current architecture14:13
blueCmdif you feel full of energy a pull request on the README.md with up-to-date instructions would be welcome :P14:13
blueCmdreally? weird14:13
benjamincburnsI’d be glad to once I got it working14:13
blueCmdcan you paste the output?14:13
benjamincburnsE: Couldn't work out current architecture14:14
benjamincburnswhoops - was multiline14:14
benjamincburnsone sec14:14
benjamincburnshttp://pastebin.com/RxSdxSbe14:14
blueCmdhm, did you replace both qemu's?14:15
blueCmdqemus*14:15
blueCmd(both host and chroot)14:15
benjamincburnsoh - didn’t realize I needed both to be replaced… will try that14:15
benjamincburnssame output14:16
blueCmdtry blasting the chroot and try again with the new qemu, sorry - never seen that before14:16
benjamincburnsI suspected binfmt at first - is there anything about that I should check on?14:17
blueCmdwell, you can check /proc/sys/fs/binfmt_misc14:17
blueCmdit should contain a file referring to your or1k-qemu14:18
benjamincburnsdoesn’t appear to14:18
benjamincburnspython2.7 python3.3 register status14:19
benjamincburnscat status => enabled14:19
blueCmdhave a look at "update-binfmts"14:20
blueCmdyou might need to --enable that14:20
benjamincburnsah, thanks14:21
benjamincburnsthat’s looking better now — “I: Installing core packages…”14:22
blueCmdcool!14:22
benjamincburnsout of curiosity, how big is the openrisc.debian.net repo? I was thinking of hosting a mirror on the jor1k network relay for fast access to jor1k clients14:23
benjamincburnsprovided I can get a debian image running on jor1k, that is14:23
blueCmdif you feel even more adventurous, you might want to dist-upgrade from "http://openrisc.debian.net/debian/" when that is finished14:23
blueCmdbenjamincburns: a few gb, not super-huge, it doesn't have the _all packages so that saves a lot14:23
benjamincburnscool14:23
blueCmdyou might want to put a hold on coreutils, last I checked it was quite broken in http://openrisc.debian.net/debian/14:24
blueCmdbut /debian/ is buildbots so they might have fixed it automagically14:24
benjamincburnsthough I wonder if it’ll work on jor1k — I don’t think Sebastian has added them to the core14:24
benjamincburnsthem = the new atomic instructions14:24
blueCmdright14:25
blueCmdpoke53281 = Sebastian, so you might just ask him :)14:25
benjamincburnsRight - looks idle at the moment14:25
benjamincburnsI had a look over the recent commits though and I didn’t see anything14:25
benjamincburnsbut I could be blind14:26
benjamincburnsdepending on my level of ambition maybe I’ll take a crack at adding them myself14:26
benjamincburnsI filed an issue, anyway - lately I haven’t had any time to work on personal projects so it’s anybody’s guess as to when I’ll get to it14:29
blueCmdyeah, I'm not active right now either, doing some AFK stuff14:31
benjamincburnsno worries. pull request submitted. dist-upgrade in progress (holding coreutils)14:52
benjamincburnsand thanks for your help!14:53
wallentostekern: two more pushes for the coherency. I feel I approach it slowly..15:14
wallentowith extra tag memory for snooping it works (which is a better option anyway), the other still has failures15:14
stekernhit'n'run18:16
stekernman, this debug-unit stall business is tricky18:16
stekernand the implementation in mor1kx cappuccino need a proper overhaul...18:17
poke53281Hi benjamincburns21:21
poke53281watched the german soccer game :). But probably no one here is interested and it's completely off topic. But I have already written the sentence, so I will send it ;)21:23
poke53281I don't see a reason why Debian should not work.21:28
poke53281the Debian port is already installed on my computer.21:28
poke53281but there is one major problem. It is big. Very big.21:30
poke53281My goal was to provide stand-alone images as small as possible.21:31
poke53281If we start reloading stuff, the perceived speed will be bad.21:33
poke53281I am wondering, wether a loading on demand of hard drive sectors or a nfs implementation would be better.21:34
poke53281if you do the bootstrap I end up with around 200MB.21:36
poke53281Anyhow, implementing the atomic instructions is no problem.21:43
poke53281benjamincburns: I would suggest to start a nfs on your server which can be accessed by jor1k. Then we try a chroot.22:05
benjamincburnspoke53281: will do - I think as long as there’s enough RAM in the jor1k config that there can be a reasonable disk cache it will appear to be fairly fast for most things23:52
--- Log closed Sun Jun 22 00:00:45 2014

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