--- Log opened Mon Apr 29 00:00:36 2013 | ||
stekern | hmm, I think I've found a syscall related bug in the kernel | 03:34 |
---|---|---|
ysangkok | hi, does anyone here know how i can run https://github.com/s-macke/jor1k/blob/master/bin/vmlinux.bin in Qemu or or1ksim? | 20:15 |
@juliusb | ysangkok: hi, I'm not sure - are there no tutorials online about this? | 20:57 |
ysangkok | i didn't find any | 20:58 |
@juliusb | hmm, perhaps you're right | 20:59 |
@juliusb | I'm surprised there's no guide to running Linux on or1ksim | 20:59 |
@juliusb | although, to be honest, I would be more inclined to compile your own kernel + busybox | 20:59 |
ysangkok | there is this: http://kevinmehall.net/openrisc/guide/ | 21:00 |
ysangkok | but i just wanted to run the image from jor1k, i thought it'd be easy. i want to play with jor1k, so it doesn't make much sense for me to build a new image if that won't run on jor1k. and if the existing image doesn't run in qemu, i guess it wouldn't | 21:01 |
@juliusb | unfortunately we don't keep much in the way of pre-built kernels around for people to play with :-/ | 21:02 |
@juliusb | AFAIK anyway | 21:03 |
ysangkok | is s-macke still active in this community? | 21:06 |
@juliusb | stekern: basically have my flop-based cache working now :) | 21:06 |
@juliusb | ysangkok: not sure, he was on IRC a little bit, I believe | 21:06 |
@juliusb | have you tried emailing him? | 21:06 |
ysangkok | nope, but i filed a github issue | 21:06 |
@juliusb | ok, cool | 21:07 |
@juliusb | you could try posting to the openrisc mailing lists | 21:11 |
stekern | ysangkok: not the most convinient way, but you can attach gdb to or1ksim and 'restore' the bin straight into mem | 21:14 |
stekern | juliusb: cool, have done any performance comparisons with it? | 21:15 |
ysangkok | stekern: sounds neat, but I don't know the addresses. actually i never used or1ksim before. | 21:20 |
stekern | I'm trying to run coremark under Linux, but it crashes with mor1kx... | 21:20 |
stekern | ysangkok: just load into address 0 | 21:21 |
ysangkok | how can gdb know where the guest operating system's address 0 is? | 21:22 |
ysangkok | s/operating system/memory/ | 21:22 |
stekern | the .bin starts at address 0, so it doesn't need to now anything | 21:24 |
stekern | ah, now I managed to run coremark with mor1kx under linux | 21:31 |
ysangkok | should i use "rsp_enabled"? | 21:33 |
stekern | are you speaking about or1ksim options? | 21:35 |
stekern | juliusb: I think I was to quick to judge the "level" interrupts as working... | 21:37 |
stekern | either that, or the B3_REGISTERED_FEEDBACK option is bringing out some bug | 21:40 |
ysangkok | stekern: yes | 21:41 |
stekern | because coremark is working fine on an older .bit, with only those two differences | 21:41 |
stekern | ysangkok: I think so | 21:41 |
stekern | was a while since I hooked up gdb to or1ksim, just remember that it's possible | 21:41 |
stekern | http://www.embecosm.com/appnotes/ean2/html/ch05s01.html | 21:42 |
stekern | it's a bit outdated, but the main idea is there | 21:42 |
ysangkok | ah ok then i think i understand what you meant earlier. i thought i should attach gdb to the simulator process as I would attach it to any other process. but if i connect over the special port, then of course the addresses will be in the simuled system and not the host. | 21:43 |
stekern | ah, yes, then your question makes much more sense too ;) | 21:44 |
stekern | sorry for not being verbose enough | 21:44 |
ysangkok | no problem, thanks for helping | 21:44 |
stekern | got to go to bed now, good look with the gdb loading | 21:45 |
stekern | good luck too... | 21:45 |
ysangkok | thanks | 21:45 |
@juliusb | stekern: not yet, just tidying up some bits and pieces | 21:49 |
@juliusb | what seems to be the issue running coremark on mor1kx on Linux? | 21:50 |
poke53281 | Oh, too late. ysangkok has left | 23:58 |
--- Log closed Tue Apr 30 00:00:37 2013 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!