IRC logs for #openrisc Saturday, 2013-03-30

--- Log opened Sat Mar 30 00:00:52 2013
-!- Netsplit *.net <-> *.split quits: serp_02:56
-!- Netsplit over, joins: serp_03:00
andresjkanyone awake?03:08
stekernI am now, but barely04:48
andresjkits like 6am. Respect!04:55
andresjkI needed a Makefile for kernel modules but I got it04:55
andresjkI got a simple example of kernel loading with printk that I found in internet04:56
andresjkbut I dont get why this>>> example doesnt compile04:57
andresjkI guess is due to kernel version. but the one of the OR is 3.1.xx so it should04:57
andresjkmaybe is another thing. but with _init and _exit works fine in the fpga04:58
stekernI usually wake up 5am, but today I could have slept longer, our little girl thought otherwise though05:17
stekernwhat does it complain about?05:18
andresjkthis is what I get :
andresjkbut other module that I compile if loaded and I cannot unloaded05:27
andresjkI get rmmod: chdir(/lib/modules) No such file or directory05:28
andresjkand If I try to insert another module, like insmod mod.ko I get: "insmod : cant insert 'mod.ko': File exists"05:30
stekernin the first, it's complaining about not finding gcc05:31
andresjkI dont understand why it says "File exists". Isn't it obvious...05:31
stekerndo you have that in your PATH?05:31
andresjkyes I do05:31
andresjkI can compile user apps fine05:32
andresjkoh, maybe its due to the prototype functions05:33
andresjkbut in the example they dont use those05:33
andresjkin my other module (which does compile) there are the module_init and module_exit prototypes05:34
andresjkyes, its because of the  underscores05:43
andresjkstekern, why do you think rmmod is not responding?05:49
stekernI assume you have checked that there is a /lib/modules directory05:53
stekernis this on the board or in or1ksim?05:53
stekernand I'm still baffled about your first error, but you seem to be able to compile other stuff05:54
andresjkIts in the board. The module called hello-1 appears in /lib/modules as hello_105:55
andresjkyes, my first error was weird but Im pretty sure its what I told you05:56
andresjkinteresting, I cannot rm -rf hello_105:57
stekernwhat happens if you try or1ksim?06:00
andresjkcan I connect or1ksim to internet?06:05
stekernif you enable tun/tap06:12
stekernit's described in the or1ksim documentation06:12
andresjkactually I dont get your point. Would it be the same?06:27
andresjkI flash my fpga and its the same :(06:28
andresjkbtw, stekern , there is a thing I dont get about wishbone. My master peripheral works perfect when it is wired to the memory controller but if I wired to the arbiter1 it fails. I mean the system hangs completely06:29
andresjkI did the same to the Ethernet controller. I wired it to the master slot of the arbiter and it stop working. I thought that the data arbiter was connected to the memory controller and I was going to be the same but its know06:30
andresjkits not*06:31
stekernwhat would be the same?06:39
andresjkto connect the master to a master slot of an arbiter or directly to the MC06:41
andresjkThe same the simulator and the OR120006:42
andresjkisnt the simulator based on the or1k rtl or arch06:43
stekernthe arbiter is a bit dumb, it just grants access to the master with highest priority. what happens if you connect it to a port with lower priority?06:45
stekernno, it wouldn't be the same. they are different implementations of the or1k architecture and not based on each others, but only the architecture spec. or1ksim is however meant to be the golden reference, so if some would be based on the other, or1200 would be based on or1ksim.06:47
stekernbut my point with trying in or1ksim would be to exclude any rtl bugs06:47
stekernbecause, of course you are right, in your case they should function exactly the same. that's what I'd like to find out, if they really do ;)06:54
andresjkyeah, It makes sense. I will recompile the kernel to do it faster06:55
andresjkabout the arbiters,  actually I have to chain them because they have 2 master slots, so actually the highest priority arbiter has only 1 useful slot and the other one is used to chain with the second priority07:01
andresjkin the second priority arbiter in the slot0 I have the debugger and in the slot 1 my peripheral, only for the test because it doesnt work there it has to be in the mc just like the eth07:02
stekernwhat board are you using?07:06
stekern(sorry, I know you've mentioned this, but can't remember right now)07:07
andresjkthe ML50107:07
andresjkits fine :)07:07
andresjkbtw, thanks for helping, I really appreciated it :)07:08
andresjkohh the same module I could insert in the board cannot be inserted07:15
andresjk# insmod hello2.koi07:15
andresjknsmod hello2.ko07:15
andresjkhello2: version magic '3.4.0-00006-g065bba6-dirty ' should be '3.4.0-rc6-11570-gcc5234d-dirty '07:15
andresjkinsmod: can't insert 'hello2.ko': invalid module format07:15
andresjkstekern, so the problem is the version, right?07:17
stekerndidn't you compile it together with your kernel?07:18
andresjkwhat do you mean?07:21
andresjkI compile it with the makefile and then I copy the .ko into /usr07:22
stekerni mean, why is the version different?07:29
andresjkcould it be because I compiled the module separately ? There should be a why to fake the version of the module o a way to make the kernel ignored it07:33
andresjkI don't think that slightly difference in the version can make that07:34
mor1kx[mor1kx] skristiansson pushed 1 new commit to master:
mor1kxmor1kx/master 0cb2ef8 Stefan Kristiansson: icache, dcache: harmonise icache and dcache...08:55
andresjkstekern, when I do uname -r in the board I get : 3.4.0-00006-g065bba6-dirty09:01
andresjkand if the I do a modinfo of a compiled module I get: 3.4.0-00006-g065bba6-dirty as vermagic09:01
stekernaren't you using the same kernel sources for the one you're running in or1ksim?09:02
andresjkIm sorry stekern , the early test in the or1ksim I did it with a backup copy. I thought I was the same but I was a older checkout. Now I try the same version as the board09:15
andresjkand the modules loads09:16
andresjkbut again I cannot remove it09:16
andresjksame error as the board09:16
andresjkrmmod: chdir(/lib/modules): No such file or directory09:16
stekernok, so now we determined that it's a software problem at least09:20
stekernoh great, updating from kubuntu 12.04 to 12.10 broke *everything*10:25
andresjkhow come?13:24
andresjkwhen updating & upgrading from ubuntu 10 to 12.10 my nvidia driver broke several times =)13:25
stekernandresjk: yeah, the (proprietary) ati drivers were one of the things that broke completely. I went back to the open source drivers, could be very well be so that they fulfill my needs completely16:23
stekernthe other thing was this:
andresjkdid you solve it or still broke?16:50
stekernyeah, I got everything sorted out17:12
stekernwell, except that I now got a white background in emacs in console mode for some reason...17:16
andresjkisn't it configurable in the .emacs file?17:36
stekernyeah, it just a matter of fixing it, no big deal17:39
andresjkstekern, are you there?23:53
--- Log closed Sun Mar 31 00:00:54 2013

Generated by 2.15.2 by Marius Gedminas - find it at!