poke53281 | Ok, there are currently 2-3 small errors which prevents a smooth compiling of the experimental version using newlib. But easy to fix. | 06:20 |
---|---|---|
stekern | poke53281: hmm, what errors? | 08:44 |
stekern | should I be worried that I'm raising my kids to be geeks? Our one-year old knows the words: mom, dad, granny and ... C3PO... | 16:54 |
stekern | I'm just waiting for her to want her hair just like Leia... | 16:56 |
juliusb | haha that's awesome | 18:04 |
juliusb | that would make you Darth Vader then wouldn't it? | 18:04 |
juliusb | :P | 18:04 |
stekern | I suppose, I have always greeted the kids with a "kssssh, I am your father, ksssh" right after they have been borned ;) | 18:12 |
juliusb | :) | 18:12 |
juliusb | so I'm putting in OR1k 1.0 updates to mor1kx | 18:13 |
juliusb | We need to figure out what the implementation-specific bits are going to be | 18:14 |
juliusb | i've used up an 8-bit field for pipeline implemenation field (remember we have 8 32-bit registers to play with) | 18:14 |
juliusb | I reckon it'll be fine to make the meaning of the ISRs different depending on the pipeline, but it'll probably be easier to keep the meanings the same across the pipelines | 18:15 |
juliusb | so, for example, what sort of multiplier is in the ALU | 18:15 |
juliusb | indicating that by a bit in the ISRs | 18:15 |
juliusb | any pipeline-specific options (can't think of anything off the top of my head) | 18:16 |
juliusb | bus interface info, perhpas, whether it's the classic interface or a bursty one | 18:16 |
juliusb | anyway, i've got the ISRs assigned in the ctrl units of each pipeline | 18:17 |
stekern | to what pipeline it is? | 18:19 |
juliusb | ??? | 18:24 |
juliusb | I've updated all the pipeline models | 18:24 |
juliusb | i'll hopefully post some patches tonight to the mailing list | 18:30 |
juliusb | just so you can pass an eye over it | 18:30 |
juliusb | also need to figure out the mor1kx versino numbering for the VR2 | 18:40 |
juliusb | I reckon increment the minor number | 18:40 |
juliusb | s we have a 24-bit version field int he VR2, which is CPU version | 18:41 |
juliusb | well, we have the choice of somehow embedding the pipeline ID in that VR2 | 18:41 |
juliusb | so maybe do something like {8'dMajor, 8'Minor, 8'PipeID} | 18:54 |
stekern | no, I meant that is it assigned to an ISR which pipeline it is :) | 19:13 |
stekern | or did you put that elsewhere, but sounds like you haven't decided yet | 19:16 |
juliusb | yeah not sure | 19:22 |
juliusb | I reckon we put an 8-bit pipeline ID in the VR2 field | 19:22 |
juliusb | so 8-bits major version, 8-bits minor version, 8-bit pipeline ID | 19:23 |
stekern | sounds good | 20:15 |
poke53281 | @stekern: I used the manual provided here | 20:21 |
poke53281 | http://opencores.org/or1k/OpenRISC_GNU_tool_chain#Installation_of_development_versions | 20:21 |
poke53281 | Following errors I encountered | 20:21 |
poke53281 | 1. "../../or1k-src/gas../opcodes/or1k-desc.h:260:73: error: this decimal constant is unsigned only in ISO C90 [-Werror]" | 20:21 |
poke53281 | Relatively easy to solve I think. I removed the -Werror from the Makefile but this is of course not a good solution. | 20:21 |
poke53281 | The compiler I used is gcc 4.6.1 | 20:21 |
poke53281 | 2. the third compile step "# build newlib" needs or1k-elf-cc, but the file is missing. So I added a symbolic link to or1k-elf-gcc | 20:21 |
poke53281 | 3. I am missing the target compiler or1k-linux in this manual | 20:21 |
poke53281 | 4. "configure --help" does not show all important options. | 20:21 |
stekern | poke53281: http://pastie.org/5031284 | 20:24 |
stekern | my cheat-sheet | 20:24 |
stekern | on 1. that's on 32-bit, right? | 20:24 |
stekern | 2. I haven't seen | 20:25 |
poke53281 | Yes, actually I am using virtualbox-image from the website | 20:25 |
poke53281 | Thanks for the script. Will try it. | 20:27 |
poke53281 | My real system is 64-Bit. But I don't want to use it for this. | 20:28 |
juliusb | stekern: i'll email you the propopsed patches to mor1kx for OR1k 1.0 compatibility | 21:20 |
juliusb | well, not all | 21:20 |
juliusb | just SPR updates | 21:20 |
poke53281 | stekern: The sheet works perfect. But there are two commands missing | 22:26 |
poke53281 | gcc stage 1: make && make install | 22:26 |
poke53281 | uClibc: make && make PREFIX=$SYSROOT install | 22:26 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!