Back
[10:02:01] <jepler> I tried busy-waiting in uspace to improve worst-case latency, but it didn't help much. for posterity:
http://emergent.unpythonic.net/files/sandbox/0001-uspace-bad-idea-busy-waiting.patch
[10:18:10] <jepler> (it maybe helped a little for the one-thread case, but made the two-thread case worse)
[10:39:11] <pcw_home> There seems to be a bug with uspace-hm2-eth or maybe my environment
[10:39:13] <pcw_home> if you reboot it doest work until you "make setuid" again
[10:46:16] <jepler> pcw_home: that's suitably weird.
[10:46:25] <jepler> pcw_home: what does it do instead of working? I'm about to walk out the door, but I'll read back later.
[10:53:12] <pcw_home> It starts but fails during initilization
[10:53:13] <pcw_home> if you run latency test it complains of inability to get I/O permission
[10:53:37] <pcw_home> (but starts and has huge numbers)
[10:53:56] <pcw_home> I can capture the messages on Monday
[11:17:42] <seb_kuzminsky> today is the day
[11:17:48] <seb_kuzminsky> that i finally review uspace
[11:17:53] <seb_kuzminsky> (i think i've said that before)
[11:23:56] <cradek> I've lost track of what I'm supposed to be doing
[11:26:02] <Tom_itx> happen often?
[11:26:23] <cradek> when there are many things happening around me, yes
[11:52:27] <seb_kuzminsky> hurr, i was going to test the new 3.4.96 kernel (without memleak's config changes so far), plugged the wifi in to download it, and the machine panicked
[11:52:31] <seb_kuzminsky> "oh yeah!"
[11:53:03] <cradek> get a horse^Wwire
[11:53:29] <seb_kuzminsky> lucky for me i'm so disorganized there's about a dozen other experimental kernels on that machine, surely one of them works
[12:03:36] <seb_kuzminsky> linux-image-3.4-9-rtai-686-pae_3.4.96-1linuxcnc panics on usb wifi dongle insert in exactly the same way as linux-image-3.4-9-rtai-686-pae_3.4.87-1~linuxcnc.1 did
[12:03:44] <seb_kuzminsky> who makes up these crazy version numbers?
[12:04:01] <seb_kuzminsky> i'll try memfrob's config on 3.4.96 next
[12:04:19] <seb_kuzminsky> but first: pack the family up and go to the swimming pool
[12:04:24] <seb_kuzminsky> bbl
[12:27:20] <skunkworks_> logger[psha]_:
[12:27:20] <logger[psha]_> skunkworks_: Log stored at
http://psha.org.ru/irc/%23linuxcnc-devel/2014-07-04.html
[12:31:18] <cradek> ha, if you poke "Mail reader" on the menu, it offers you the choice of ... mutt
[12:34:08] <cradek> Mousepad is a terrible name for a text editor. It's really not discoverable at all when that's what the menu says.
[12:37:24] <skunkworks_> heh
[12:38:45] <skunkworks_> the 2 motherboards i have tested with weezey and the debian rt kernel have not done anything odd for me..
[12:39:01] <skunkworks_> so far
[12:39:42] <skunkworks_> (I think it is the same MB pcw is using)
[12:50:49] <jepler> pcw_home: that's weird. I don't see anything similar here--after a reboot, I still get
[12:50:52] <jepler> Note: Using POSIX realtime
[12:51:16] <jepler> I am guessing this is the note you are talking about, which might instead say
[12:51:19] <jepler> cannot gain I/O privileges - forgot 'sudo make setuid'?
[12:51:21] <jepler> Note: Using POSIX non-realtime
[12:51:46] <jepler> I'd like to see what 'ls -l bin/rtapi_app' says before and after rebooting. is something taking away the setuid?
[12:54:20] <pcw_home> I'll try again on Monday The system was running the ubc3-7i80 branch, I wonder if that made a difference
[12:54:37] <pcw_home> maybe it will stay fixed...
[12:58:15] <pcw_home> I may go into work today at check again, there are lots of illegal fireworks going on around here which Charlie hates
[12:58:17] <pcw_home> On the other hand maybe I should stay here with the garden hose at the ready...
[13:39:14] <skunkworks_> quiet here so far...
[14:30:45] <skunkworks_> running overnight..
http://electronicsam.com/images/KandT/testing/DSC_1814.JPG
[14:30:57] <skunkworks_> (in the scary/messy basement..)
[14:38:06] <KGB-linuxcnc> 03Jeff Epler 05jepler/rtos-uspace 29859e0 06linuxcnc 17 commits pushed, 1014 files changed, 03144(+), 0471(-)
[14:38:07] <KGB-linuxcnc> 03Jeff Epler 05jepler/rtos-uspace-apis 19d7fba 06linuxcnc 33 commits pushed, 1030 files changed, 03366(+), 04642(-)
[14:38:07] <KGB-linuxcnc> 03Jeff Epler 05jepler/rtos-uspace-hm2-eth 197bf22 06linuxcnc 40 commits pushed, 1032 files changed, 03379(+), 04655(-)
[14:39:03] <skunkworks_> ooh.. more testing...
[14:39:08] <cradek> cool
[14:40:01] <skunkworks_> crap - now I have to figure out how I am going to hook this system to the network
[14:40:30] <pcw_home> I used a usb-enet dongle
[14:40:55] <pcw_home> ($0.99 at Goodwill)
[14:41:12] <skunkworks_> I have a netgear one - bot wheeze doesn't seem to run with it... (probably need some closed source firmware)
[14:41:15] <skunkworks_> but
[14:42:17] <pcw_home> PCIE enet card, or J1900-DH3 :-)
[14:42:47] <skunkworks_> heh
[14:42:48] <cradek> on the more recent CD images, I included all? the nonfree firmwares
[14:43:04] <cradek> that makes my laptop's wifi work, for instance
[14:43:47] <skunkworks_> sure - so did my laptop wireless.. netgear wna3100
[14:43:50] <cradek> you can tell if you have that by seeing if non-free is in sources.list
[14:43:58] <cradek> ah ok
[14:44:01] <cradek> so you knew that
[14:44:52] <skunkworks_> looks like it is broadcom..
[14:50:31] <skunkworks_> I will give this one a try
[14:50:47] <skunkworks_> ubuntu says it is a rtl8187
[14:51:33] <cradek> that should work fine
[14:55:03] <skunkworks__> that worked
[14:55:15] <skunkworks__> I run into this from time to time - what do I do?
[14:55:17] <skunkworks__> http://pastebin.com/W4VYfeZ3
[14:55:26] <skunkworks__> http://pastebin.com/W4VYfeZ3
[14:55:30] <skunkworks__> ehe
[14:55:56] <skunkworks__> can you force it some how?
[15:04:19] <jepler> skunkworks__: if you have *no local changes at all*, then you want to "git reset --hard @{u}"
[15:04:51] <jepler> otherwise, you need to commit or stash your changes and then deal with this as described in man git-rebase under the heading RECOVERING FROM UPSTREAM REBASE
[15:04:54] <skunkworks_> I have not changed anything..
[15:05:41] <skunkworks_> I found the --hard - what is @{u} ?
[15:05:57] <jepler> this happens to you because I "rebased" (rewrote the history of) my branches. This is inconvenient for everyone who collaborates with me, but the end result is that when the branch is merged to master, all the problems I realized and fixed before that moment can be erased as though they never existd
[15:06:07] <jepler> rather than carried around in the project's shared history forever
[15:06:34] <skunkworks_> that sounds great
[15:08:18] <skunkworks__> so this is where I want to be? HEAD is now at 197bf22 hm2_eth: fix 64-bit build error
[15:08:33] <jepler> yes
[15:08:50] <skunkworks__> making now
[15:09:22] <skunkworks__> does make now run configure?
[15:12:10] <jepler> if configure.in is changed, it can re-run configure
[15:12:19] <jepler> .. and since configure.in is among the things my branch changes, it does so
[15:13:08] <skunkworks__> neat
[15:15:36] <skunkworks_> anything I should look for?
[15:17:13] <skunkworks_> (besides bugs.. ;) )
[15:20:54] <jepler> well .. there is realtime delay notification now
[15:21:19] <jepler> but hopefully you won't see a message about it :)
[15:21:50] <jepler> I don't think anything else that has changed is user-visible
[15:22:03] <skunkworks_> yay!
[15:22:06] <skunkworks_> ok
[15:29:16] <skunkworks_> jepler: so far so good.. :)
http://electronicsam.com/images/KandT/testing/ethlatestbuild.png
[15:30:20] <skunkworks_> that is with a usb network adaptor and steaming youtube...
[15:30:27] <jepler> I certainly haven
[15:30:37] <jepler> I certainly haven't tried to stress the -rt kernel yet
[15:30:44] <skunkworks_> *wireless network adaptor
[15:31:37] <skunkworks_> I will just let it run
[15:38:06] <skunkworks_> the interface on wheezy is a lot closer to gnome (in 10.04) just a little less flashy...
[15:55:17] <cradek> there are actually several choices - that one is called xfce
[15:55:42] <cradek> I picked it because it doesn't seem to have any glaring problems
[16:05:33] <jepler> its default pdf viewer stinks
[16:05:50] <jepler> .. it can't view pdfs that are scans, somehow
[16:06:00] <jepler> .. I install evince
[16:08:18] <skunkworks_> cradek: what is the link to wlo livecd
[16:10:04] <cradek> http://linuxcnc.org/binary.hybrid.iso ec2e8151b617426df30d384440656ada
[16:10:52] <cradek> er no it's 0189184d4636939f7a31c3b3af8119df
[16:11:29] <skunkworks_> thank you
[16:47:21] <MrHindsight> skunkworks_: 8unless you have the cpu set to max core speed the on chip power management will throttle the core speed if if idle
[16:47:54] <MrHindsight> how long depends on the microcode and stepping of the cpu
[16:48:43] <MrHindsight> thats why latency goes down when you keep the cpu core speed at MAX and power consumption goes to near max
[16:49:34] <MrHindsight> skunkworks_: which AMD cpu were you testing last night?
[17:12:22] <skunkworks_> that was a intel based board
[17:14:37] <skunkworks_> GA-j1800
[17:14:46] <skunkworks_> hmm - that doesn't sound right
[17:26:16] <skunkworks_> http://www.newegg.com/Product/Product.aspx?Item=N82E16813128688
[17:33:05] <CaptHindsight> i must have read the backlog wrong
[18:29:55] <skunkworks_> the other system is a amd/asus motherboard.
[19:04:54] <CaptHindsight> http://linuxgizmos.com/open-linux-hummingboard-sbc-goes-on-sale-starting-at-45/
[19:30:54] <CaptHindsight> sorry wrong channel
[19:46:54] <cradek> jepler: I want to try max/5i25 on uspace. should I be building jepler/rtos-uspace-aps?
[19:47:37] <cradek> apis
[19:50:51] <cradek> the answer is apparently yes because it just works
[20:03:25] <cradek> seb_kuzminsky: 3.4.87-1~linuxcnc.1 works fine on this real hardware too
[20:38:10] <cradek> jepler: I built and installed uspace packages, and it all works
[20:50:59] <skunkworks> isn't that awesome - just works
[21:00:05] <skunkworks> you should also try the -rt kernel
[21:22:42] <seb_kuzminsky> cradek: thanks for that report
[21:22:59] <seb_kuzminsky> i like that that kernel config is close to the upstream debian "normal" kernel config
[22:10:20] <cradek> skunkworks: I was running uspace/5i25 on the -rt kernel.
[22:13:38] <skunkworks> awesome!
[22:38:35] <skunkworks> duh - I didn't read what you wrote
[22:44:22] <jepler> cradek: cool!
[22:45:21] <cradek> yep, it's really a great thing.
[22:56:14] <skunkworks> jepler: cradek:
http://electronicsam.com/images/KandT/testing/realtimedelay.png
[22:56:30] <skunkworks> (after I pulled the network cable to the 7i80) :)
[22:56:51] <skunkworks> ran all after noon - restarted it and letting it run overnight
[23:05:47] <skunkworks> although it does give the unexpected realtime delay on task 0 twice in the terminal..
