IRC logs for #openrisc Friday, 2014-10-31

--- Log opened Fri Oct 31 00:00:01 2014
wallentohey all, I use the new or1k-headers script of pete (https://github.com/pgavin/or1k-headers)09:18
wallentoit generates the SPR addresses like this: 0x4800U09:19
wallentounfortunately the GNU assembler does not like it09:19
wallentowhats the best way to cope with it? Let the script remove the unsigned modifier or is there any asm magic?09:19
wallentoor is it pure c?09:23
stekernwallento: good question... I guess the options are 1) remove the U 2) generate separate defines for asm10:16
wallentoyeah, I went with the first option (see my fork where it is optional now), that should be fine for newlib10:17
wallentothanks10:17
stekernah, nice. yes, I think that's the best way to handle it10:23
stekernand you can anyway get 2) from it too10:23
wallentookay, guys, I am nearly approaching the finish of newlib13:58
wallentolast question is what to do with the multicore stuff13:58
wallentountil now there was -mmulticore and GCC selected libc_mc.a and crt0_mc.o instead of the regular ones13:59
wallentokinda multilib, but not really13:59
wallentowhat are the differences: i) Needs 1 shadow register file, ii) uses lwa and swa, iii) something I probably forgot14:00
wallentosorry, libor1k_mc.a14:00
wallentoii) is kinda -marchv1.1 :)14:01
wallentowhat are your opinions? should it be multilib? should it be two different outputs (its mainly a few #ifdef OR1K_MULTICORE)?14:02
olofkIf there aren't any conflicts, I guess that both versions could be built14:08
olofkBut my knowledge of the topic is not very good14:09
olofkBut I think I have optimized wb_sdram_ctrl ~3% in my efforts to make it support RAMs with different widths14:10
stekernspeedwise?14:30
olofkstekern: Yep. I'll clean it up and push to a new repo if you want to take a look19:10
stekerncool, sure19:19
--- Log closed Sat Nov 01 00:00:03 2014

Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!