| --- Log opened Thu Sep 15 00:00:12 2016 | ||
| -!- heroux_ is now known as heroux | 04:24 | |
| shorne | stekern: Now I have the fixes reshared as per this link | 08:12 |
|---|---|---|
| shorne | https://github.com/stffrdhrn/linux/commits/openrisc-fixes-4.8 | 08:12 |
| shorne | *reshaped*, sqaushed the or32 toolchain fix, and removed dts for now | 08:13 |
| shorne | but, I think having the dts in tree does 2 things, makes build easier + makes our support for hardware much better advertised | 08:19 |
| shorne | But I do agree that autogenerating dts by something like looking at wb config in fusesoc would be awesome | 08:20 |
| olofk | shorne: Yeah. I'm fine with having them in the tree. Just that it doesn't scale very well with many different boards and configurations | 08:22 |
| olofk | We still haven't defined some kind of baseline functionality, which would have helped | 08:22 |
| olofk | e.g. 32MB RAM, UART@0x90000000, GPIO@0x91000000, SPI@0xB0000000 | 08:23 |
| shorne | yeah, but have a look at https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/arch/arm/boot/dts?id=refs/tags/v4.8-rc6 | 08:23 |
| shorne | but I think we can do better :) | 08:24 |
| SMDhome | shorne: what's wrong w/ having tons of dts files in kernel dir? | 08:25 |
| shorne | SMDhome: I am ok with it, but since our hardware is open, we have both the SOC config in source and dts in source | 08:26 |
| shorne | they are virtually equivelent it might be nice to generate dts from the source soc (verilog, wb bus config) source | 08:27 |
| shorne | But then again, the kernel driver names etc would be hard to map to | 08:29 |
| shorne | and keep in sync with the kernel, I think keeping out of tree we would not be able to keep up to date with api changes/ refactoring etc | 08:30 |
| SMDhome | shorne: and what if you make some preconfigured SoCs(i.e. for de nano and digilent board) and dts for them? And for enthusiasts there will be generic dts with guide or generator to configure dts | 08:34 |
| olofk | SMDhome: I've been encouraging people to put dts files in the top-level FuseSoC cores | 08:48 |
| olofk | We got for atlys and neek so far | 08:49 |
| SMDhome | olofk: do you have a board w/ mor1kx in it now? Could you run a binary there just for test? | 08:52 |
| olofk | SMDhome: You could, but it would be awkward | 08:53 |
| olofk | But I have been running a system quite recently with just mor1kx+block RAM+JTAG | 08:54 |
| SMDhome | olofk: I'm trying to run coremark, but it takes too long even on verilator | 08:54 |
| olofk | ah ok | 08:55 |
| SMDhome | So I'd like to know if that tests ever ends :) | 08:55 |
| olofk | How much RAM does it need? | 08:55 |
| olofk | Running it on a de0_nano or similar would probably be pretty simple | 08:57 |
| olofk | You need CPU+RAM at least. And it's probably nice to have a UART to watch the output | 08:57 |
| SMDhome | that's a good question! Last addr in binary disasm is 0x13cfc | 08:57 |
| olofk | And JTAG for loading the program at runtime | 08:57 |
| SMDhome | should I get de0-nano dev & educational board instead of de0-nano-soc? Seems like the latter has no ram connected to the fpga | 10:26 |
| --- Log closed Fri Sep 16 00:00:14 2016 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!