| --- Log opened Fri Mar 07 00:00:09 2014 | ||
| -!- Netsplit *.net <-> *.split quits: enghong | 02:25 | |
| -!- Netsplit *.net <-> *.split quits: _franck_, dalias, ams, wkoszek_ | 02:35 | |
| --- Log closed Fri Mar 07 03:21:44 2014 | ||
| --- Log opened Fri Mar 07 03:21:59 2014 | ||
| -!- Irssi: #openrisc: Total of 31 nicks [0 ops, 0 halfops, 0 voices, 31 normal] | 03:21 | |
| -!- Irssi: Join to #openrisc was synced in 18 secs | 03:22 | |
| stekern | olofk: it's done, sans working sim environment | 03:24 |
|---|---|---|
| -!- Netsplit *.net <-> *.split quits: juliusb, phoohb, poke53282, bentley`, olofk | 03:24 | |
| stekern | olofk: it's done, sans working sim environment | 03:25 |
| stekern | (don't know if you saw the first, netsplits) | 03:25 |
| stekern | I could submit it in the weekend if you want, I've soiled my version with the sublime synth stuff, I probably want to remove that before submitting. | 03:27 |
| -!- Netsplit over, joins: phoohb | 03:30 | |
| stekern | what's wrong with verilator that it can't handle functions like this: https://github.com/skristiansson/wb_sdram_ctrl/blob/master/rtl/verilog/wb_port_arbiter.v#L187 | 06:13 |
| olofk_ | stekern: That looks like the function in my wb_arbiter that I rewrote. Can't remember exactly what happened, but I debugged the generated C++ code and realized that it got stuck in an infinite loop somewhere deep down | 07:32 |
| olofk_ | Don't know if newer versions work better | 07:32 |
| stekern | yeah, no, newer version doesn't work better | 07:33 |
| olofk_ | And yes, I'm interested in the SoCKIT port. Someone asked for VGA support, and I realized that it is the only fusesoc system that uses VGA | 07:33 |
| olofk_ | jeremy_bennett: has said that he's interested in bug reports | 07:33 |
| olofk_ | It should be quite easy to isolate that to a test case | 07:33 |
| stekern | yup, I'll take a closer look at it and add a bug report to the verilator issue tracker | 07:34 |
| stekern | I have one of those in my eco32f implementation too | 07:34 |
| olofk_ | I have a vague memory that it was because a signal that was used in the always block was written to, so it called itself infinitely, but I can't really remember | 07:35 |
| stekern | I have another, very similar function in another place in eco32f, that works though | 07:35 |
| olofk_ | ah yes. The eco32 port is interesting as well so that we can show the broad range of processors supported by fuessoc :) | 07:35 |
| olofk_ | Hmm... that's interesting | 07:35 |
| stekern | but it has a disticnt "default, not found" value, a bit like your workaround (I think) | 07:36 |
| olofk_ | You might be onto something there | 07:38 |
| stekern | looking closer at my function in wb_sdram_ctrl, that actually seems broken... | 07:39 |
| stekern | it can return with ff1 undefined | 07:39 |
| stekern | I have some improvements to the verilator testbench as well | 07:44 |
| stekern | like gracefully exiting on ctrl-c and $finish() | 07:44 |
| stekern | the ctrl-c thing always annoyed me in the mor1kx-devenv testbench | 07:45 |
| stekern | yup, adding an initial ff1 = 0; before that foor loop fixes it | 07:46 |
| stekern | s/foor/for | 07:46 |
| olofk_ | stekern: Hmm... I'm pretty sure I did that in my wb_arbiter | 08:00 |
| olofk_ | http://git.opencores.org/?a=commitdiff&p=orpsoc&h=37ce1fbc734f0a079f31d04963cd2c66392cf4fa | 08:04 |
| olofk_ | Maybe that has been fixed in verilator since I wrote that code | 08:06 |
| stekern | hmm, ok | 08:29 |
| stekern | at least this works with the latest verilator zip: https://github.com/skristiansson/eco32f/blob/master/rtl/verilog/eco32f_ctrl.v#L236 | 08:30 |
| stekern | s/zip/tar | 08:31 |
| stekern | and this too: https://github.com/skristiansson/eco32f/blob/master/rtl/verilog/eco32f_mmu.v#L114 | 08:31 |
| olofk_ | Yeah, they both look very similar. | 09:12 |
| blueCmd | olofk_: http://sourceware.org/ml/binutils/2012-07/msg00269.html - thats how aarch64 did it | 09:37 |
| olofk_ | Fair enough. Are you interested in chopping up the code into patches? | 09:42 |
| jeremy_bennett | olofk_: Bug reports are indeed to be welcomed. Bugzilla is your friend. | 09:42 |
| blueCmd | olofk_: yeah I can do that | 09:43 |
| blueCmd | olofk_: do you have the set of names that have touch the files? | 09:44 |
| olofk_ | blueCmd: Sure. Where do you want them? | 09:47 |
| blueCmd | olofk_: email me them | 09:47 |
| blueCmd | olofk_: is it true that or1k.opc is contributed by Red Hat? | 09:54 |
| blueCmd | it says so in 'cpu/or1k.opc' | 09:54 |
| olofk_ | How the fuck should I know? | 09:54 |
| olofk_ | oh... maybe I should have spotted that when I went through the patch | 09:54 |
| olofk_ | Hmm.. might need to dig deeper there | 09:55 |
| blueCmd | olofk_: haha yes :) | 09:55 |
| olofk_ | woohoo!! A response from binutils | 13:54 |
| olofk_ | And I got a mail from a RTEMS guy who has a student doing an OpenRISC port for GSoC. He's looking for a mentor. Anyone interested? | 14:00 |
| olofk_ | I'll forward it to the lists | 14:00 |
| blueCmd | olofk_: this is really good news! | 20:03 |
| blueCmd | olofk_: I volounter to be the maintainer if somebody else doesn't feel strongly against that | 20:04 |
| wkoszek | Can we e-mail GSOC and ask them why OpenRISC didn't qualify? | 20:31 |
| blueCmd | wkoszek: I can ask a coworker on monday | 21:38 |
| --- Log closed Sat Mar 08 00:00:11 2014 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!