| --- Log opened Fri Nov 25 00:00:58 2016 | ||
| mithro | olofk: Any chance you are coming to 33c3? | 00:10 |
|---|---|---|
| olofk | shorne: I see you got some replies to the patches already | 02:20 |
| olofk | mithro: Unfortunately not. Would have loved to go, but I don't have the time. Are you going? | 02:21 |
| olofk | mithro: We were looking for you at the GSoC mentor summit btw. Thought you were going | 02:21 |
| mithro | Where in the world are you located | 02:36 |
| mithro | olofk: Sadly, none of our team could make it to the mentor summit | 02:48 |
| mithro | olofk: Trying to keep my international travel to one trip a year | 02:48 |
| mithro | olofk: I'll be speaking at 33C3, which will be awesome hopefully | 02:48 |
| mithro | olofk: I'm also visiting _florent_ and the apertus guys before hand | 02:58 |
| mithro | Are there any other or1k FPGA implementations apart from mor1k ? | 03:15 |
| olofk | mithro: I'm in Sweden. I will have to keep my travel to a minimum for some time now as well, so unless you're nearby I probably won't see you :) | 03:33 |
| olofk | Say hi to _florent_ and the Gielda brothers if you meet them | 03:33 |
| olofk | And regarding or1k, there is or1200 (the original one), or10n (from the PULP project), altor32 (a VHDL version) and a few more that I can't think of right now | 03:35 |
| olofk | Hope you have a good time at 33c3. I think Clifford will be talking there as well. Haven't looked at the speaker list | 03:35 |
| mithro | Yeah, he is | 05:14 |
| mithro | olofk: has there been any progress on gcc upstream support? I assume nobody is working on it? | 05:18 |
| olofk | mithro: On the technical side, shorne and other have been trying to keep the code aligned with upstream. On the legal side, we have tried to talk to FSF about a way forward | 06:31 |
| olofk | But no resolution yet | 06:31 |
| wbx | olofk: what is the legal issue with that? binutils is upstream, why not gcc? | 06:32 |
| mithro | Copyright assignment | 06:32 |
| mithro | Someone said no, IIRC? | 06:33 |
| wbx | someone with code in the gcc port? | 06:33 |
| olofk | wbx: Everyone, except for one guy has given permission for copyright assignment. Unfortunately, his work is very early in the development, so technically the rest is based upin that | 06:33 |
| wbx | and this guy is no longer interested in or1k? | 06:34 |
| olofk | The latest idea we had was to see if the stuff he wrote has actually been replaced by other patches | 06:34 |
| olofk | wbx: He is actually running very much involved, as he works for a company that makes proprietary versions of OpenRISC | 06:34 |
| wbx | olofk: isn't is possible to convince the guy to just offer his code as public domain, so no special fsf agreement required. | 06:34 |
| olofk | wbx: No. His standpoint is that he doesn't want to give up his ownership of the code | 06:35 |
| olofk | Which of course is just pure fucking bogus | 06:35 |
| wbx | so he thinks his company benefits from these actions so that toolchain support isn't upstream or what? | 06:36 |
| wbx | i don't understand such actions from people working with open source. | 06:38 |
| olofk | wbx: Well, me neither. But there is not much more we can do to convince him :/ | 06:55 |
| olofk | Anyone remembers how to use the adv_debug_sys UART with OpenOCD? | 07:02 |
| olofk | For the documentation database: jsp port is 7777 (adv_debug_sys jtag serial port) | 07:19 |
| olofk | Not getting any data there unfortunatley :/ | 07:21 |
| olofk | Maybe I should use Richard Herveille's version instead, but it uses a bit too much SystemVerilog for Icarus to handle simulations | 07:26 |
| olofk | Works like a charm now! | 08:06 |
| olofk | In polling mode at least. As I remember it, there is still some bug that stops it from working on Linux as soon as the interrupt is enabled | 08:07 |
| shorne | olofk: that mean you are using uart to linux over usb blaster? | 09:22 |
| shorne | FYI I got a mail from wallento saying the gcc copyright issue has been sorted. I wonder if we should start getting the patches ready. But we are quite a bit behind with gcc rebase | 09:23 |
| wallento | shorne: not sorted, but another approach to solve it :) | 09:23 |
| olofk | shorne: Yes, except that I'm running bare metal in this case | 09:24 |
| olofk | That's fantastic! | 09:24 |
| shorne | wallento: what is that approach? you mentioned you talked to fsf? | 09:26 |
| wallento | I approached them, but things are going slow | 09:26 |
| wallento | I will try to get a fast yes/no | 09:27 |
| olofk | wallento: Just be careful. In some places it's actually illegal to take people hostage | 09:29 |
| olofk | Has anyone looked at this btw? https://github.com/openrisc/newlib/issues/14 | 09:31 |
| olofk | It feels like a real bug, but my C skills are not good enough to figure out if it is a real problem | 09:31 |
| olofk | Actually, looking at the generated asm it looks strange | 09:38 |
| olofk | There are two identical lwz instructions in a row | 09:38 |
| shorne | olofk: I havent looked, looking now | 09:38 |
| olofk | But otoh, it looks like a program bug. The 32-bit pointer isn't aligned to a 32-bit word | 09:43 |
| shorne | this part | 09:49 |
| shorne | 2428: 18 60 00 01 l.movhi r3,0x1 | 09:49 |
| shorne | 242c: a8 63 4d dc l.ori r3,r3,0x4ddc | 09:49 |
| shorne | 2430: 84 63 00 00 l.lwz r3,0(r3) | 09:49 |
| shorne | im using the second program | 09:51 |
| shorne | in simulator I see | 10:02 |
| shorne | S 00002430: 84630000 l.lwz r3,0x0(r3) r3 = 007fdfed flag: 0 | 10:02 |
| shorne | S 00002434: 84630000 l.lwz r3,0x0(r3) r3 = 007fdfed flag: 0 | 10:02 |
| shorne | S 00000600: d4000804 l.sw 0x4(r0),r1 [00000004] = 007fdfe0 flag: 0 | 10:02 |
| shorne | 0x600 | 10:03 |
| shorne | Load/store access to naturally not aligned location. | 10:03 |
| shorne | so definitely bad code | 10:03 |
| shorne | I am guessing gcc | 10:04 |
| shorne | or we could detect in newlix via the 0x600 routing and then access the correct byte | 10:12 |
| shorne | routine | 10:13 |
| olofk | But are we supposed to handle unaligned accesses gracefully? | 13:51 |
| olofk | Or is it ok to error out? I still think we should add a handler in newlib in any case | 13:51 |
| olofk | shorne_: Or did you mean that GCC generates wrong code here? | 14:14 |
| shorne_ | olofk: well maybe, maybe not, I am not sure should gcc be protecting against unaligned accesses? | 17:25 |
| shorne_ | clearly this code is trying to so a byte access | 17:25 |
| shorne_ | newlib doesnt seem to have code for hanlding unaligned byte access | 17:34 |
| kc5tja | olofk: yt? | 17:48 |
| kc5tja | olofk: Need advice on how to package support parts for the KCP53000. Wondering if they should go with the KCP53000 or in separate packages. They (together) provide the Wishbone bus bridge for the CPU. | 17:52 |
| shorne_ | FYI, http://elinux.org/Japan_Technical_Jamboree_59 | 17:57 |
| shorne_ | there is a little embedded linux day conf in Japan, ill be speaking on openrisc | 17:58 |
| -!- shorne_ is now known as shorne | 18:00 | |
| --- Log closed Sat Nov 26 00:00:00 2016 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!