Back
[08:29:51] <KGB-linuxcnc> 03John Thornton 052.7 412d512 06linuxcnc 10docs/src/gui/tooledit.txt Docs: remove note about 2.5.0 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=412d512
[08:29:52] <KGB-linuxcnc> 03John Thornton 052.7 be28890 06linuxcnc 10docs/src/gcode/overview.txt Docs: markup fixes * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=be28890
[08:53:03] <linuxcnc-build> build #1725 of 1404.rip-wheezy-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/1404.rip-wheezy-rtpreempt-amd64/builds/1725 blamelist: John Thornton <bjt128@gmail.com>
[09:08:37] <linuxcnc-build> build #3384 of 0000.checkin is complete: Failure [4failed] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/3384 blamelist: John Thornton <bjt128@gmail.com>
[09:16:20] <JT-Shop> rut-row
[09:21:45] <Tom_itx> JT-Shop, saw one of those yesterday:
http://www.motorcycle.com/blog/wp-content/uploads/2014/12/120414-ReverseTrike-Slingshot-Action-3546.jpg
[09:22:58] <Tom_itx> not up close...
[09:28:57] <JT-Shop> yea, someone here in town has one... I'd rather have a red sports car
[09:30:19] <Tom_itx> that's not can am is it?
[09:35:02] <JT-Shop> no
[09:35:09] <JT-Shop> I think Polaris makes them
[09:35:40] <JT-Shop> http://www.polaris.com/en-us/slingshot
[10:14:35] <jepler> task: 4040 cycles, min=0.000012, max=5.853915, avg=0.002888, 3 latency excursions (> 10x expected cycle time of 0.001000s)
[10:14:38] <jepler> halui: emcCommandSend: no echo from Task after 5.000 seconds
[10:18:50] <jepler> since that's a mode switch it's probably the long pause in rename() when disks are busy. I wonder if qemu's cache=unsafe (disables fsyinc in qemu) would help. task: 4040 cycles, min=0.000012, max=5.853915, avg=0.002888, 3 latency excursions (> 10x expected cycle time of 0.001000s)
[10:18:55] <jepler> halui: emcCommandSend: no echo from Task after 5.000 seconds
[10:18:57] <jepler> https://en.wikibooks.org/wiki/QEMU/Devices/Storage
[13:40:59] <jepler> hm, apparently the 4.1-rt kernel doesn't have aufs support which is needed for live
[14:02:07] <pcw_home> the stock debian rt kernel doesn't include aufs?
[14:08:01] <pcw_home> Hmm -10 usec is not enough DPLL pre-read time on this system (ran for a week or so and then got a random FE)
[14:08:02] <pcw_home> But my default -50 is way too big
[17:23:53] <cradek> I don't understand why people on the forum think a custom M code (which is by definition a running gcode program) can switch modes and reload a gcode program.
[17:25:56] <cradek> "has anyone solved this yet" "I want it fixed in the next release" but he never says what problem he's trying to solve
[17:29:56] <seb_kuzminsky> haha
[17:49:51] <jepler> pcw_home: the debian 4.1 kernel doesn't include it, -rt or not
[18:01:21] <cradek> oh dear, and another guy needs to buy a machine with 4 parallel ports for some reason
[18:02:53] <Tom_itx> hah
[18:03:02] <Tom_itx> doesn't know about mesa?
[18:12:31] <JT-Shop> the failed checkin have anything to do with my commits?
[18:15:29] <cradek> JT-Shop: nope
[18:15:44] <cradek> halui: emcCommandSend: no echo from Task after 5.000 seconds
[18:15:44] <cradek> halui: sendMdiCommand: failed to Set Mode MDI
[18:15:54] <cradek> it took 5.8 seconds to do something that timed out in 5 seconds
[18:16:33] <JT-Shop> ah thanks
[18:16:42] <cradek> jepler suggested a fix for seb to try
[19:16:25] <seb_kuzminsky> cache mode "unsafe" is not supported by wheezy's libvirt
[19:16:37] <seb_kuzminsky> i'll try forcing it to writeback
[19:27:16] <jepler> seb_kuzminsky: not sure whether it'll make a real difference but maybe...
[19:29:37] <jepler> interesting patches on lkml I don't understand, but it seems like some (server class?) Intel CPUs can partition cache so that only certain tasks get to use certain parts of it.
[19:30:06] <jepler> it would be cool to start real-time code with a hot cache
[19:52:33] <skunkworks> zlog_:
[21:29:28] <jepler> hm that's not it either
[21:29:28] <jepler> /home/jepler/src/kernel-deb/linux-4.1.3/debian/patches/features/all/aufs4/aufs4-standalone.patch
[21:29:32] <jepler> there is an aufs patch in there
[21:30:10] <jepler> # changes needed to allow aufs to be built out-of-tree.
[21:34:26] <jepler> so it seems like there should be another source package for this out-of-tree aufs driver, but there's not one that I can find
[22:09:37] <seb_kuzminsky> d
[22:12:53] <seb_kuzminsky> jepler: new intel chips (not sure which) lets you specify a process identifier for the running code, and it annotates your cache lines with that id, and you can specify resource limits on cache lines, which get enforced by the hardware (it'll evict *your* cache lines if you ask for too much)
[22:12:58] <seb_kuzminsky> if i recall correctly
