| --- Log opened Thu Mar 12 00:00:18 2015 | ||
| -!- clopez_ is now known as clopez | 10:23 | |
| olofk | Tired of the 32 (30?) IRQ lines to OpenRISC. Can we throw them out? One IRQ should be enough | 13:54 |
|---|---|---|
| juliusb | olofk: (embARC) yeah, I agree, that thing ain't gonna do muc IoT for you, but it'll do a hell of a lot of embedded softcore microprocessor development for that price :) | 13:58 |
| juliusb | olofk: it ultimately is one IRQ as far as the pipeline knows, but at some point you need a banking mechanism to turn many into one | 13:59 |
| olofk | juliusb: I was thinking about an external Interrupt controller | 14:01 |
| olofk | Haven't thought much of the implications. Backwards compatibility would break of course, and we would need some extra jumps I guess to find the cause of the interrupt | 14:02 |
| olofk | But we should also be able to avoid putting crap at 0x200, 0x300,0x400... | 14:03 |
| olofk | And we already use stekern's fancy IRQ controller for multicore anyway | 14:03 |
| olofk | Bonus points if we can assign an SPR or something to set the interrupt vector from software. Then we can put it wherever we want | 14:04 |
| stekern | olofk: where does the 32 hurt? | 14:06 |
| stekern | you can omit 31 of them in hardware if you have an external interrupt controller | 14:06 |
| olofk | stekern: Right now they hurt because my fancy IP-Xact tool has some problems with assigning between vectors and wires :) | 14:09 |
| olofk | But I guess the main reason would be to free up the space between 0x200 and 0x2000 (or wherever the standard vectors are) | 14:10 |
| stekern | what does that have to do with it? | 14:13 |
| stekern | there's only one interrupt exception for all interrupts | 14:13 |
| stekern | 0x800 | 14:13 |
| olofk | oh.. | 14:27 |
| olofk | ah yes | 14:27 |
| olofk | The other are for exceptions | 14:27 |
| olofk | ok. Never mind then :) | 14:28 |
| bandvig | to all: my another proposal for arch. spec. 1.2 http://opencores.org/or1k/Architecture_Specification#ORFPX64A32 | 14:38 |
| juliusb | I think for interrupts there should be a programmable address, exceptions another programmable address with a register containing the exception reason, and for reset, a synthesis-time parameter | 15:46 |
| juliusb | that 0x100, 0x200 etc. thing is bad | 15:46 |
| -!- FreezingAlt is now known as FreezingCold | 19:42 | |
| olofk | What would be the easiest way to transfer files to an OpenRISC system with ethernet running Linux? | 19:46 |
| olofk | Is there a telnet equivalent to scp? | 19:46 |
| stekern | tftp | 19:46 |
| olofk | Jesus. That was quick :) | 19:46 |
| dalias | or netcat. or an httpd + wget | 19:46 |
| dalias | i don't find tftp very easy | 19:47 |
| stekern | not quick enough, it was an answer to your first question, not the second | 19:47 |
| dalias | seriously tho just use ssh/scp | 19:48 |
| dalias | dropbear is tiny | 19:49 |
| stekern | or just normal ftp | 19:49 |
| olofk | I thought of dropbear, but I remember it as quite slow | 19:51 |
| stekern | that's included in the "prebuilt" initramfs iirc | 19:51 |
| olofk | dropbear or ftp? | 19:53 |
| stekern | ftp | 19:54 |
| olofk | Yeah, there's ftpget and ftput in usr/bin | 19:55 |
| stekern | olofk: I just bought 'the book of unwritten tales' | 19:57 |
| stekern | downloading it now | 19:58 |
| olofk | stekern: Cool! | 19:58 |
| olofk | Great game. Took me 20 hours to play through. | 19:59 |
| stekern | I replayed maniac mansion for the first time in ~25 years the other day too | 20:02 |
| olofk | I never really played maniac mansion much | 20:09 |
| olofk | Just some through DOTT and a little on NES | 20:10 |
| stekern | me neither, but it was more fun than I remembered it | 20:11 |
| olofk | Which version did you play? | 20:14 |
| stekern | DOS V2 | 20:14 |
| olofk | Some VGA remake? | 20:15 |
| stekern | no, it's the original "enhanced version" with slightly prettier EGA | 20:17 |
| stekern | V1 http://upload.wikimedia.org/wikipedia/en/7/76/Maniac_Mansion.png | 20:20 |
| stekern | V2 http://arcadeandretro.com/wp-content/uploads/2012/02/maniac_mansion_01.png | 20:20 |
| olofk | Yeah. Slightly prettier | 20:21 |
| olofk | poke53281: I've tried to run lynx from jor1k.com/packages. After I have added the required libs (zlib, ncurses, openssl...) it first complains on "/tmp/: No such directory", and when I run it again it says "Terminal initialisation failed - unknown terminal type?". Any ideas? | 20:31 |
| stekern | I think it wants TERM=linux | 20:37 |
| olofk | Bus error :( | 20:38 |
| olofk | Is that a wishbone bus error? | 20:39 |
| olofk | Like an unaligned access | 20:39 |
| stekern | bus error is bus error, unaligned error is another thing | 20:41 |
| stekern | apart from that, yes | 20:42 |
| olofk | aha | 20:44 |
| olofk | So what could trigger a bus error in or1ksim? | 20:44 |
| dalias | SIGBUS is usually caused either by misaligned loads/stores or by access to a memory-mapped region with no backing (like past the end of a file) | 20:45 |
| olofk | hmm | 20:50 |
| olofk | Another question. Is there some userspace utility to directly communicate with simple_spi, or is the SPI core only used by other (higher-level) drivers? | 20:50 |
| stekern | olofk: https://www.kernel.org/doc/Documentation/spi/spidev | 21:41 |
| stekern | now fix the sound in my unwritten tales.. | 21:45 |
| --- Log closed Fri Mar 13 00:00:19 2015 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!
