| --- Log opened Fri Feb 06 00:00:29 2015 | ||
| stekern | olofk: spi, some obscure nand driver, usb etc. basically all !ethoc | 01:57 |
|---|---|---|
| stekern | ...maybe it's time for me to upstream those... | 02:07 |
| stekern | well, at elaaast the spi one | 02:07 |
| stekern | it's hard to keep up with the instream of mails when you have sporadic internet access... | 02:50 |
| stekern | I pushed what u-boot I had on my hd here: https://github.com/skristiansson/u-boot/tree/openrisc | 02:53 |
| stekern | (and what I had on my hd should reflect what was on openrisc.net) | 02:54 |
| olofk | stekern: Yeah, I know the feeling. That's why I came up with the rule about no talking on the internet when I'm not there | 06:11 |
| wallento | olofk: sorry, need to keep up | 07:42 |
| wallento | i will clean up github.io | 07:43 |
| wallento | thanks for your input | 07:43 |
| wallento | is openrisc.net gone forever now? | 07:43 |
| wallento | are you interested in posting some news there? | 07:45 |
| wallento | I will post some or1k-elf toolchain update | 07:45 |
| wallento | maybe one can add a linux update | 07:45 |
| olofk | wallento: Not sure what will happen to openrisc.net, but people are already asking where to donwload stuff, so we should update the docs for now | 07:58 |
| wallento | okay, we may take the chance and clean up this stuff | 09:23 |
| olofk | Hmm.. I'm getting this error http://pastie.org/9891772 when I compile the kernel with my memset routine. Anyone with better Linux/C skills who can figure out what's wrong? | 09:56 |
| ysionneau | olofk: have you made the symbol global ? | 10:22 |
| olofk | ysionneau: Yep | 10:22 |
| * ysionneau is off | 10:22 | |
| ysionneau | :p | 10:22 |
| olofk | I'm looking at the other arches and there are lots of things I don't really get | 10:23 |
| stekern | it looks like it's used from a module, perhaps extra precaution has to be made when writing asm for modules? | 10:24 |
| olofk | stekern: An update. I added some ifdefs according to how the other arches do it and now I get this instead | 10:25 |
| olofk | arch/openrisc/kernel/built-in.o: In function `copy_thread': | 10:25 |
| olofk | (.text+0x7ec): relocation truncated to fit: R_OR1K_INSN_REL_26 against undefined symbol `memset' | 10:25 |
| olofk | And a lot of other similar | 10:25 |
| stekern | but why is memsetg undefined? | 10:27 |
| stekern | -g | 10:27 |
| olofk | I can't understand that either | 10:28 |
| stekern | I wish my computer workspace always looked like this, on a balcony with view over sunset on a beach | 10:28 |
| olofk | :) | 10:28 |
| stekern | and a beer at my right side | 10:28 |
| stekern | olofk: to get git to send e-mails, just do 'git send-email --to <someone> --cc <someone-else> /path/to/patch' | 10:32 |
| olofk | stekern: Cool | 10:41 |
| olofk | Aha.. I think that my .h file is never picked up | 10:41 |
| olofk | Which I can kind of understand, since I haven't told Linux that it exists | 10:42 |
| olofk | Otoh, I can't see that the other arches do that either | 10:43 |
| olofk | This sucks | 10:53 |
| olofk | I had it working in another kernel tree, but in that case I just rebuilt a kernel, so I probably need to do something more to make it work on a fresh kernel | 10:54 |
| olofk | Hmm... most of the asm stuff for OpenRIS is defined directly in the .h files. Is there a reason for this? | 10:57 |
| olofk | And what's this asm/linkage.h that seems to be used in some places? | 11:07 |
| olofk | Linux is complicated. I should probably learn C some day :( | 11:07 |
| olofk | It works when I disabled the module | 12:04 |
| olofk | Actually, that was the only module. Could it be something weird with modules in general? | 12:04 |
| -!- Netsplit *.net <-> *.split quits: ysangkok, tariq786, ed-jones, clopez__, _franck_, jonmasters, enghong, mboehnert1, stekern, martinboehnert1, (+28 more, use /NETSPLIT to show all of them) | 12:19 | |
| -!- Netsplit over, joins: pecastro, dalias, _franck__, hansfbaier, ed-jones, stekern, martinboehnert1, arokux, heroux, sb0 (+28 more) | 12:23 | |
| olofk | Haha. With the optimized memset it executes 15% less instructions on a kernel boot. Probably a whole lot less writes too, so it should be a lot faster without data cache | 12:31 |
| olofk | Yep. 12% less write accesses | 12:34 |
| olofk | Assembler is fun! | 12:35 |
| olofk | I'll probably rewrite fusesoc in assembler now | 12:44 |
| _franck__ | olofk: while you are working on this you should try the C version of memcpy from musl | 12:48 |
| _franck__ | it's the same king of algorithm you can find in arch/microblaze/lib/memcpy.c | 12:49 |
| _franck__ | s/king/kind | 12:49 |
| stekern | olofk: nice! | 13:06 |
| olofk | _franck__: Yeah, I'm using the same algorithm as the microblaze port | 13:16 |
| olofk | for memset | 13:16 |
| -!- rhythmx_ is now known as rhythmx | 21:51 | |
| --- Log closed Sat Feb 07 00:00:30 2015 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!