IRC logs for #openrisc Tuesday, 2013-04-16

--- Log opened Tue Apr 16 00:00:17 2013
stekernjuliusb: could still be some bus related problem still, I've sometimes had funny behaviour when reading the flash from u-boot01:27
stekernjuliusb: ok, you treehugger ;) let's keep the gate on the operands, that will keep it from "wiggling": http://pastie.org/759526901:38
mor1kx[mor1kx] skristiansson pushed 3 new commits to master: https://github.com/openrisc/mor1kx/compare/5d1dd717c2ac...6c7d7d3ee65603:47
mor1kxmor1kx/master 685982b Stefan Kristiansson: execute_alu: make branch destination calculation optional...03:47
mor1kxmor1kx/master 7f3670e Stefan Kristiansson: cappuccino/cpu: disable branch destination calculation in ALU03:47
mor1kxmor1kx/master 6c7d7d3 Stefan Kristiansson: execute_alu: make alu_result default to adder_result...03:47
jemarchhi11:11
mor1kx[mor1kx] skristiansson pushed 3 new commits to master: https://github.com/openrisc/mor1kx/compare/6c7d7d3ee656...f88dde32385c12:38
mor1kxmor1kx/master 1547a64 Stefan Kristiansson: cappuccino/rf: combine 'last_wb_result_b' and 'wb_to_decode_result'...12:38
mor1kxmor1kx/master a88e42b Stefan Kristiansson: cappuccino/execute_ctrl: remove unused ctrl_op_jal_o signal12:38
mor1kxmor1kx/master f88dde3 Stefan Kristiansson: cappuccino/rf: remove unnecessary decode to decode bypass...12:38
mor1kx[mor1kx] skristiansson pushed 3 new commits to master: https://github.com/openrisc/mor1kx/compare/f88dde32385c...f0a01bedbd3d14:34
mor1kxmor1kx/master 2a8547c Stefan Kristiansson: cappuccino/rf: reset wb_hazard_a/b instead of wb_hazard_result...14:34
mor1kxmor1kx/master ef862d1 Stefan Kristiansson: cappuccino/fetch: stop flushing on du_restart_i14:34
mor1kxmor1kx/master f0a01be Stefan Kristiansson: cappuccino/ctrl: clear branch_step when npc is written...14:34
stekernjuliusb: you're not the only one fixing debugging-related bugs =P14:36
@juliusb:)15:48
@juliusbit's quite good actually, I hadn't noticed these bugs in simulation, or thought about them15:48
@juliusbbut when I was finally using the design ont he board a little bit those things annoyed me greatly (when we step into or break on a l.j 0 it couldn't resume because the fetcher had locked up so had to reset the whoel thing)15:49
@juliusbit was a 2 line fix to stop that, but also added tests to recreate that situation15:52
@juliusbbut usability issues like that really piss you off\15:52
stekernyeah, I should fix bp support in cappuccino and then really try using the debugger with it16:04
stekernwhen he comes back from fat-camp16:04
stekernjuliusb: is the 2nd try on the mul patch something you could live with btw?16:05
@juliusboh didn't see it yet...16:51
@juliusbdoes it work OK with having only mul_op?16:52
stekernyup, I ran the 200000 test against espresso and cappuccino17:05
stekernand a and b should stay stable as long as you have the mul in there, otherwise I reckon you have another bug going on17:06
stekernbut the main purpose of the patch was actually to remove the !rst in this (slightly restructured) original code: http://pastie.org/760162417:25
stekernthen I got carried away with the logic removing ;)17:25
stekernbut I think the latest patch should match the previous behaviour, non-wiggling and everything17:26
@juliusboh right, fair enough, !rst is pretty dirty17:36
@juliusbsoudns good17:36
@juliusb200000 test? you mean GCC?17:36
stekernno, the mul test17:37
@juliusbahh sorry im with you17:39
@juliusbyeah cool17:39
@juliusbhow'd you run that?17:39
@juliusbon HW?17:39
stekernin verilator17:40
stekernI might have a 0 too much there though17:40
@juliusb:)17:40
@juliusbnps17:40
stekernthe default or1k-mul17:41
@juliusbya think it's 20k17:41
stekerncappuccino is doing well at fatcamp btw, he's already lost ~300 LC (in the altera way of measuring things)17:58
stekernstill a chubby boy though, weighs in at 5300 LC17:59
stekernI want that to be < 4000 before I can say that I'm proud of him18:00
@juliusbwow, that's ambitious!20:49
@juliusbyou reckon you can?20:49
@juliusband why 4000?20:49
@juliusbOK I can read the contents of the SPI now22:16
@juliusbthey look good22:16
@juliusbbut that's from a C-code app22:16
@juliusbso why isn't the bootrom working.....22:16
@juliusb?22:16
@juliusbworks 100% in sim22:17
@juliusbi've updated it to support no-delay slot stuff22:17
@juliusbas I said, works fine in sim22:17
@juliusbbootrom.S is getting assembled with -mno-delay22:17
@juliusbhmmm22:18
@juliusbnow it works!!22:18
@juliusbanoying....22:18
@juliusbI made it delay before attempting to read the flash22:18
mor1kx[mor1kx] juliusbaxter pushed 2 new commits to master: https://github.com/openrisc/mor1kx/compare/f0a01bedbd3d...6ad603b8359922:41
mor1kxmor1kx/master 809a1cc Julius Baxter: ctrl_pronto: fix bug in PC checker...22:41
mor1kxmor1kx/master 6ad603b Julius Baxter: fetch_pronto: fix sleep behaviour when debugging...22:41
@juliusbstekern: had to do this to get to compile in modelsim: http://pastie.org/762099822:51
@juliusb(didn't strictly have to make a whitespace mess, though)22:52
--- Log closed Wed Apr 17 00:00:19 2013

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