IRC logs for #openrisc Sunday, 2013-01-06

stekernwhee, dtlb_permission_test passes08:15
stekerni.e. pagefaults seems to work08:15
stekernhaven't had the time to look at this stuff for a while08:23
stekernpesky non-openrisc projects have been in the way, but now they are off the table ;)08:23
juliusb_franck_: awesome work on the GDB Stuff man08:34
juliusbstekern: ah, nice stuff witht he (I asume) HW DTLB stuff in mor1kx?08:35
stekernah, no, still only soft-tlb08:38
juliusb_franck_: your submits to the OpenOCD list look pretty cool, I kinda don't get it though, it's enabling OpenOCD to request the list of registers (something which is now specified in XML For the or1k GDB port iirc)?08:39
stekernbut the whole dmmu test works now (I just fixed the last failing test, the dc inhihibit test)08:39
juliusbah sure, software-manager DMMU?08:39
juliusbsoftware-manage*d*08:39
juliusblike what OR1200 has?08:39
stekernyup08:40
juliusbcool08:40
stekernwanna get that working before starting to look at hw walkers08:40
juliusbya, I bet :)08:40
juliusbseems like there's been a bit of libc action lately08:40
juliusbi've been busy for a couple weeks here in Aus, just catching back up08:40
stekernyeah, blueCmd have been busy ;)08:41
juliusbis it just me or is the opnrisc.net list archive not working? or have there just been no emails to the list this year yet?08:41
juliusbno, frank sent one jan 308:41
juliusbso.... hmm, the archive is not showing it: http://lists.openrisc.net/pipermail/openrisc/08:42
juliusbwhereas http://lists.opencores.org/pipermail/openrisc/08:42
juliusbhas it08:42
stekernmmm08:42
stekernI found a syscall bug in cappuccino btw08:44
stekernI've added a test for it in the syscall test, espresso pass that so it seems cappuccino specific08:44
stekernthis sequence: l.sys 0; l.j; would fail, the branch address of the l.j would be used instead of the exception vector address08:45
juliusbOK pinged Jonas08:46
juliusbohh nasty bug :)08:46
stekernnow I just have to implement immu, that's gonna be a bit more of a headache I suppose08:47
juliusbi'm working on getting the carry/overflow stuff up to spec08:48
juliusbhaha yep I bet08:48
juliusbbut only for 1 pipeline08:48
juliusbyou could just integrate it into the fetch stage quite easily08:48
stekernyeah, the basic logic is simple, it's all the corner cases that's gonna be a pain ;)08:48
stekernmost of the time with the dmmu went to hooking up the exceptions correctly08:49
juliusbsweet09:11
mor1kx[mor1kx] skristiansson pushed 2 new commits to master: https://github.com/openrisc/mor1kx/compare/ac978a5ad300...2b8d2fa995d009:28
mor1kxmor1kx/master 0c165d1 Stefan Kristiansson: cappuccino/lsu: connect execute lsu address...09:28
mor1kxmor1kx/master 2b8d2fa Stefan Kristiansson: cappuccino: fix a bug related to syscall exception...09:28
juliusb:)09:28
stekernI pushed the dmmu wip to: https://github.com/skristiansson/mor1kx09:36
stekernmostly to have a backup, it's still rough, to say the least ;)09:36
_franck_juliusb: the target descriptor file (the XML one) is given to GDB by OpenOCD. So OpenOCD gives the list of registers to GDB.09:49
stekernis that what was in: http://openocd.zylin.com/#/c/1089/09:53
stekern?09:53
stekernI can't open that link here09:54
_franck_I just did it and ths link is working for me....09:54
juliusb_franck_: ah is that how it works? OK I thought it was GDB describing to OpenOCD the registers it'll want to access09:55
juliusbI guess it works either way - is it meant to allow the proxy to give further information to the debugger about what target it's connected to?09:55
_franck_yes, it describes the target with it's registers09:56
_franck_however, other ports (appart from openrisc one) are using the XML file in order to enable "feature"09:57
stekernmaybe it's just something with chromium09:57
_franck_without extracting registers names from the XML file09:57
juliusbstekern: chrome being used here09:57
_franck_me too09:58
juliusb(or are you on some mobile thing?)09:58
stekernno, my desktop machine09:58
juliusbhmm odd OK09:58
stekernit just stalls and says: Loading Gerrit Code Review ...09:58
juliusbdodge10:00
stekernworks on my firefox/win7 setup10:00
juliusbi'm on version23.0.1271.91 on Ubuntu10:01
_franck_openocd guys are not very fast when it's about accepting other's patches...10:01
juliusbcan't be any worse than OpenRISC ;)10:01
_franck_:)10:01
stekern;)10:01
_franck_http://openocd.zylin.com/#/c/467/ this is here since a year...I'm using it all the time (it's on our openocd)10:03
stekernmine is 20.0.1132.4710:05
stekernisn't that the latest in ubuntu?10:05
_franck_23.0.1271.97 under centos10:06
_franck_scientific linux to be precise10:07
juliusbso on something like an overflow exception, only the EPCR has the OV bit set? the SR for handling the exception doesn't have it set?10:38
stekernyou mean ESR?11:25
stekerngerrit comments field is very confusing to me11:29
juliusbsorry, yes ESR12:17
stekernI'm not sure, but it kind of makes sense if it's only in ESR12:19
stekernotherwise there no way to check for overflows inside an exception handler12:19
juliusbyep12:20
juliusbagreed12:20
stekernI'm not sure why/when you'd need that, but it's a potential use case at least ;)12:20
juliusbyep12:20
juliusbis zero / zero = 1?12:34
juliusbinteger-wise12:35
juliusbno, it's not12:36
juliusbdoesn't matter what the numerator is12:36
juliusbit's just bbad news, regardless12:36
_franck_stekern: what do yo mean by "gerrit comments field is very confusing" ?12:50
stekernah, just a reflection... my first encounter with gerrit, I thought it was hard to follow the conversion13:01
stekernand annoying that you couldn't click on the "x comments on the code" or what it said13:07
juliusbok, range exception works (according to the test) :)14:46
juliusbpronto only14:46
juliusbof course ;)14:46
juliusbothers to follow14:46
stekernnice15:38

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