| wpwrak | mwalle: hmm, another nice feature for urjtag would be if -q actually did anything :) the chattiness is a little inconvenient at times ... | 03:43 |
|---|---|---|
| kristianpaul | Fallenou: when i read _any_ register from milkymist soc, i must always use volatile unsigned int? i cant do volatile signed short int reads? | 04:15 |
| kristianpaul | Fallenou: i mean, the reads and writes are fixed to to 32 bits? no more no less? | 04:15 |
| stekern | kristianpaul: CSR doesn't seem to have a datamask, so writes can't work with anything else than 32-bit | 05:13 |
| stekern | why do you want to access them at non 32-bit granularity? | 05:13 |
| kristianpaul | stekern: and that datamask should be related with sel_i i guess | 12:25 |
| kristianpaul | stekern: i'm trying to print (%d) (using milkymist software) some registers stored in a 16 bits register (this goes to the verilog module) but of course | 12:28 |
| kristianpaul | wb_dat_o <= {16'h0, i_16_bit_register}; | 12:29 |
| kristianpaul | so when i read as a volatile signed int, still be printed as a unsigned | 12:29 |
| kristianpaul | and when try it as short, well i just got zeros be printed | 12:29 |
| kristianpaul | bbl | 12:33 |
| stekern | kristianpaul: you could always sign extend it in the verilog module. or in sw before you print it | 13:03 |
| wpwrak | maybe try this: volatile uint32_t *p = whatever; value = (int16_t) *p; | 13:43 |
| stekern | wpwrak: yep, that would be how to do it in sw, let's serve the verilog version on silverplate too: wb_dat_o <= {16{i_16_bit_register[15]}, i_16_bit_register}; | 13:58 |
| stekern | ;) | 13:58 |
| wpwrak | stekern: now you've made it complicated and he has to choose where to do the sign extension ;-) | 14:03 |
| stekern | hehe, freedom of choice! :) | 14:04 |
| scrts | hi guys ;) | 14:36 |
| Alarm | hi | 14:43 |
| Alarm | he following line: "qemu-system-LM32-M-kernel Milkymist flickernoise" must be executed in which folder? | 14:43 |
| Alarm | *the | 14:43 |
| sb0 | er... first you should probably pay more attention to the syntax and spelling of the command | 15:12 |
| Alarm | I did a copy / paste the command in the wiki http://milkymist.org/wiki/index.php?title=Using_QEMU | 15:19 |
| sb0 | your clipboard has an interesting behaviour :) mine gives 'qemu-system-lm32 -M milkymist -kernel ...' instead | 15:20 |
| sb0 | anyway 1. qemu-system-lm32 should be in your PATH 2. the executable passed as parameter to -kernel is relative to the current folder (unless you give an absolute path, i.e. starting with /) | 15:21 |
| Alarm | I can not locate flickernoise? I find flickernoise.fbi, | 15:55 |
| sb0 | http://milkymist.org/updates/current/ | 15:56 |
| Alarm | I had to forget something when I compiled? | 15:59 |
| sb0 | try: make bin/flickernoise | 16:01 |
| Alarm | No rule to make target `bin/flickernoise' | 16:30 |
| Alarm | unless I am mistaken, 'make-compile-flickernoise flickernoise.fbi C' can not create file 'flickernoise "for qemu | 16:53 |
| sb0 | ah, you are using xiangfu's scripts... try going one level below and rebuild only flickernoise | 18:11 |
| Alarm | ok thanks | 18:34 |
| kristianpaul | stekern: wpwrak i choose verilog way, hardware is faster :) | 19:59 |
| kristianpaul | stekern: Thanks !! | 19:59 |
| --- Thu Oct 13 2011 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!