IRC logs for #openrisc Monday, 2014-09-08

--- Log opened Mon Sep 08 00:00:43 2014
poke53281stekern: Switched to 16bpp. scummvm is really much much faster.03:50
poke53281I also added toppler, a jump and run game.03:50
stekernpoke53281: not really, I have never seen that. What are you compiling? Just looking at the code, the assert is of the "can't happen type", so something must have gone terribly wrong04:42
poke53281the funtion is in mmath.cpp04:45
stekernhardfloat have got a lot less testing than softfloat though, so I won't be surprised if you stumble on some bugs in that area04:45
poke53281they try to figure out, whether an extended floating point format is used. (long double)04:45
poke53281No, I don't activate hard-float04:46
poke53281That's the point04:46
stekernhmm, ok04:46
stekernthen that's even weirder04:46
poke53281No, problem. I investigate myself. I just decided to wait and ask you.04:47
poke53281the code breaks anyhow with an unaligned access. :)04:47
stekernyou can check from where it's called, or1k_output_bf, or1k_output_cmov or or1k_expand_int_compare04:48
stekernas a start04:48
poke53281Will do the next eays04:51
stekernI wonder if a smaller testcase with just that will expose the bug04:55
poke53281I guess. At least for the whole function. It's kind of crazy:
poke53281remove the "return false". That's by me04:59
stekernyeah, I already copy-pasted the relevant parts from sf05:03
stekernbut that compiles fine for me05:05
poke53281with -O3 ?05:05
poke53281Oh sorry. Looks like, there is a -mhard-float.05:06
poke53281But not by me05:06
stekernor1k-linux-musl-g++ -O3 isinf.cpp -S -o -05:06
poke53281add -mhard-float05:06
stekernok, that makes more sense then05:07
stekernyes, then it crashes here too05:07
poke53281Ahh, the -mhard-float comes from sdl-config probably.05:07
poke53281Yes, I should remove hard-float from sdl-config05:08
stekernbut, it's kind of good that you run stuff with hard-float, exposing bugs like this ;)05:09
poke53281I always wonder, if the fpu we implement is really sufficient.05:11
poke53281E. g. division by zero exception or so.05:11
jagadeeshhi im new to openrisc, i got led blink working from this link  how can i proceed further with c program07:23
jagadeeshsorry i am new here is there anyone could help me out07:26
stekernjagadeesh: you'll need a toolchain with a c compiler07:28
stekerninstructions on how to build one can be found here:
jagadeeshi installed the toolchain and the compiler07:29
stekernok, great07:29
stekernthen you just need some code to compile ;)07:29
jagadeeshcan you help me how to access gpios07:30
stekernthey are just memory mapped devices, so you access them as regular memory addresses07:30
jagadeeshsir can u point me to some example or reference designs07:31
stekernusually they are located at 0x91000000, so to set all 8 bits to outputs:07:32
stekernand set them to 107:32
stekern*((uint8_t *)0x91000000) = 0xff; *((uint8_t *)0x91000001) = 0xff;07:32
jagadeesh#include <stdio.h>07:33
jagadeesh#define writeGPIO(addr,val) (*(unsigned char*) (addr)=(val))07:33
jagadeeshint main ()07:33
jagadeeshreturn 0;07:33
jagadeeshsir i tried with the above code07:33
stekernplease use or similar07:33
jagadeeshbut i could not get it working.. please bare with me07:33
stekernwhat does writeGPIO do?07:33
stekernah, sorry it was in what you pasted07:34
stekernyou need to set them to outputs07:34
stekerni.e.: *((uint8_t *)0x91000001) = 0xff;07:35
poke53281I guess, the mmu is off :)07:36
stekernpoke53281: in response to? ;)07:37
jagadeeshsir i'm very new07:37
jagadeeshsome basic depth tutorial would help me....... i was using fusesoc to build the system07:38
stekernto modify your program07:38
stekernthe gpio is located at address 0x91000000 and register 0 is the data register and register 1 the direction register07:39
stekernand 1 means output and 0 input07:39
stekerndefault is input, so you must set the bits connected to the gpio you want as outputs to 1 in register 107:40
jagadeeshpls a example would help me07:41
jagadeeshok i got it working  thank you stekern07:44
jagadeeshcan u give me a document for the registers info, something like a datasheet07:45
jagadeeshHow to store my oprsoc software.elf and system.sof in altera epcs memory08:32
stekernpoke53281: lol09:09
stekernjagadeesh: wait a sec and I'll dig up some notes on the subject09:39
stekernjagadeesh: this is the commands to put linux and u-boot on the epcs of de0 nano09:43
stekernand this is how the .cof looked like:
stekernfor baremetal, skip Linux and threat the baremetal program as u-boot09:44
jagadeeshk thank you alot09:45
jagadeeshill try this out09:45
olofkAaarrggghhh... stupid windows drivers! You are driving me crazy10:33
heshamDoes the vmlinux image (git:// size only 5.4M ?11:29
heshamWell yes, seems like it's only 5.4M size, and running on or1ksim now. Will this image run on Atlys board?11:32
stekernhesham: yes11:39
heshamstekern: Thanks, for or1ksim it got stuck after "Configuring loopback device" line, and when I hit "Ctrl + C" I got, "^CPlease press Enter to activate this console. (sim)" and it returns to or1ksim11:41
stekernyes, but that's just because you are looking at the "debug output" of or1ksim, and you are not connected to the console11:42
stekerntry running 'telnet localhost 10084' from another terminal11:43
heshamConnection refused :/11:44
stekernperhaps 'monitor' is a better word than 'debug output'11:44
stekernhesham: are you running or1ksim at the same time?11:45
heshamWell, I was not, then I ran them at the same time, and I got the command prompt, thanks stekern :)11:45
stekernI have exactly this problem now:
olofkAre you using the gtk backend, readline or something else?13:32
stekernI experience it in -nw mode13:38
--- Log closed Tue Sep 09 00:00:44 2014

Generated by 2.15.2 by Marius Gedminas - find it at!