IRC logs for #openrisc Tuesday, 2012-12-11

juliusbstekern: cool :)01:13
juliusbI'm seeing an odd issue with cappuccino though01:13
juliusbactually, no i'm not01:14
juliusbI'm getting the cycle counter code working (requires fixing set_gpr in each of the CPU pipelines)01:15
juliusband when I did the l.nop 6 i wasn't getting anything in the regs01:15
juliusbbut, when that nop is in the pipeline where the monitor picks it up, other stuff is writing to that regsiter address01:15
juliusbso i had to put in a couple of nops before it01:15
juliusbso that the pipeline was clear01:16
juliusbhehe01:16
juliusbpesky long pipeliens01:16
mor1kx[mor1kx] juliusbaxter pushed 1 new commit to master: https://github.com/openrisc/mor1kx/commit/2388f379d74bed260a8be01cd7d71102d8a244f701:40
mor1kxmor1kx/master 2388f37 Julius Baxter: Fix set_spr functions...01:40
juliusbwell it took a while but we finally have this: http://opencores.org/or1k/OR1K_CPU_Cores02:57
stekernjuliusb: yeah, the monitor/trace stuff is probably a bit shaky still04:21
stekernFor example, instructions following mfspr/mtspr will not always be shown (since it is assumed that mfspr/mtspr are 2 cycle instructions)04:22
stekernand when a nop bubble is inserted, that instruction will show up twice in the trace04:22
stekernnot 100% about get_gpr neither, since the monitor assumes that all results are ready at the end of execute04:23
stekernbut loads now produce their result in ctrl/mem stage04:24
stekernhmm, doesn't tracing to screen work anymore?04:37
-!- derRicha1d is now known as derRichard13:41
_franck_I managed to get gdb uses p/P packets to read spr registers and to use target description file15:09
_franck_now we can get rid of readspr/writespr commands15:09
_franck_in openocd15:09
_franck_openocd will send an xml file to gdb:15:09
_franck_https://github.com/Franck79/openOCD/blob/p_packet_or1k/or32.xml15:10
_franck_this file match the or1k registers list:15:10
_franck_https://github.com/Franck79/openOCD/blob/p_packet_or1k/src/target/or1k.c15:10
_franck_so if you want to add, modify registers you just have to change the xml file and or1k.c register description15:10
_franck_you don't have to recompile gdb15:10
_franck_the only or1k hack in openocd core is now:15:12
_franck_https://github.com/Franck79/openOCD/commit/d6a0460c7a722e248dacb767d0499dbe2e7bdbf015:12
LoneTech_franck_: cool! :)15:29
_franck_LoneTech: I saw on the openocd channel you still have some hard time playing with your jtag ? :)15:31
LoneTechwell, getting different bits working15:31
LoneTechI found the problem with the ft2232 driver, found I should use the ftdi driver, got configs written for that too, reworked an install script.. the usual15:32
LoneTechI should finalize the virtual jtag patch so I can use one binary with or without virtual jtag, but it seems openocd only has a compile time choice between mohor and adv15:34
_franck_yes it is a compile time choise15:36
_franck_I also submit a patch for the ft2232 driver not so long time ago15:37
_franck_got the same answer you got....15:37
LoneTecha 1-line bugfix for an existing documented driver should make it in, imho15:43
LoneTechthough the ftdi driver does look better15:44
-!- X-Scale is now known as Guest1975216:59
-!- Guest19752 is now known as X-Scale17:00
-!- Netsplit *.net <-> *.split quits: jonmasters, jonibo17:25
-!- Netsplit over, joins: jonibo, jonmasters17:26
-!- Netsplit *.net <-> *.split quits: jonmasters, jonibo17:30
-!- Netsplit over, joins: jonibo, jonmasters17:32
poke53282Does someone know what the exit(34425) message means in or1ksim? I have a compiled a new vmlinux image with the additional driver MTD_PHYSMAP_OF. Without the driver it works.20:14
poke53282It crashes in arch_local_irq_restore function in the last command.20:29
poke53282The machine is in a pretty early state in the boot process. No text output through UART so far.20:40
poke53282I can find out myself but maybe the problem is known.20:41

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