| 02:34.25 | *** join/#asterisk-dev mjordan (~mjordan@user-69-1-6-49.knology.net) |
| 02:34.25 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 02:53.06 | *** join/#asterisk-dev frawd (~francois@189.Red-81-38-88.dynamicIP.rima-tde.net) |
| 03:05.18 | *** join/#asterisk-dev tzafrir_laptop (~tzafrir@bzq-218-155-147.cablep.bezeqint.net) |
| 03:05.19 | *** mode/#asterisk-dev [+o tzafrir_laptop] by ChanServ |
| 03:45.32 | *** join/#asterisk-dev otherwiseguy (~otherwise@cpe-69-76-221-79.kc.res.rr.com) |
| 04:02.48 | *** join/#asterisk-dev DocAwesome (~DocAwesom@asterisk/documenteur-extraordinaire/blitzrage) |
| 04:02.48 | *** mode/#asterisk-dev [+o DocAwesome] by ChanServ |
| 04:08.56 | *** join/#asterisk-dev irroot (~irroot@197.168.197.39) |
| 05:00.36 | *** join/#asterisk-dev irroot (~irroot@pbx.distrotech.co.za) |
| 05:44.52 | *** join/#asterisk-dev wdoekes2 (~walter@wjd.osso.nl) |
| 05:51.37 | irroot | yo there |
| 05:51.47 | irroot | wdoekes2 top o the morning |
| 06:30.43 | wdoekes2 | good morning to you |
| 06:36.01 | wdoekes2 | ah.. my internet seems to have been gone between 5:09 and 6:44 |
| 06:45.52 | *** join/#asterisk-dev oej (~olle@ns.webway.se) |
| 06:45.52 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 06:56.42 | irroot | wdoekes2 all your nets are belong to us :P |
| 07:00.30 | irroot | new toy to play with camera door phone / intercom |
| 07:18.18 | *** join/#asterisk-dev apten (~apten@carbon.gonicus.de) |
| 07:27.50 | *** join/#asterisk-dev schmidts (~schmidts@Lo20-rt01.lm33.sil.at) |
| 07:27.51 | schmidts | good morning |
| 08:25.46 | *** part/#asterisk-dev schmidts (~schmidts@Lo20-rt01.lm33.sil.at) |
| 08:25.51 | *** join/#asterisk-dev schmidts (~schmidts@Lo20-rt01.lm33.sil.at) |
| 08:27.34 | *** join/#asterisk-dev schmidts (~schmidts@Lo20-rt01.lm33.sil.at) |
| 08:27.38 | schmidts | good morning |
| 08:27.55 | schmidts | irroot do you still want to see a gdb of my system? |
| 08:28.42 | irroot | oh baby love it when you talk dirty |
| 08:29.03 | irroot | schmidts log a fault and attach it sure |
| 08:29.35 | schmidts | irroot i miss the "sorry wrong window" from your first sentence :D |
| 08:29.39 | schmidts | http://pastebin.com/A0siciqm |
| 08:30.28 | irroot | lol :P |
| 08:31.03 | irroot | when / how you get that trace |
| 08:31.10 | irroot | dont see __ast_read |
| 08:32.22 | schmidts | during a high load time |
| 08:32.45 | schmidts | but i have another with ast_read in it |
| 08:33.17 | schmidts | http://pastebin.com/vwbCdi5V |
| 08:33.23 | oej | Hello folks |
| 08:33.45 | schmidts | morning oej |
| 08:35.10 | *** join/#asterisk-dev tamiel (~tamiel@213.30.183.226) |
| 08:39.42 | irroot | need to work out what is causing the load |
| 08:54.28 | wdoekes2 | oej: regarding that you do not always get feedback on the reviewboard, I do see two very old items from you that could use a Discard or an updated patch ;) (not a complaint, there are probably 2 out of 3 pages on the reviewboard that should be canned) |
| 08:55.10 | irroot | wdoekes2 oej i need to go through the RB sometime but time is short |
| 08:56.22 | wdoekes2 | irroot: I figure the authors could review their year old items to see if they still apply |
| 08:57.44 | wdoekes2 | (guilty myself on having ooold issues on jira too) |
| 09:02.45 | irroot | wdoekes2 there is a chan_h323 patch in there that ill leave for a bit seems like others we have moved to chan_ooh323 |
| 09:06.01 | *** join/#asterisk-dev hehol (~hehol@2001:1438:1009:200:f16b:31a0:f30f:792) |
| 09:09.10 | oej | wdoekes2: The problem is that if takes two months to get responses, I'm on another project. Can't get funding to complete from the old customer... |
| 09:10.09 | oej | wdoeskes2: All patches are in branches that apply, so that's not a problem really. I keep maintaining the branches, maybe not spend time to update the patch every month on reviewboard |
| 09:10.34 | oej | I've closed the ones I've given up on |
| 09:26.48 | *** join/#asterisk-dev oej (~olle@ns.webway.se) |
| 09:26.48 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 09:39.51 | oej | HELP SET URL |
| 09:39.58 | oej | Oops |
| 09:40.01 | oej | wrong chat |
| 10:12.43 | *** join/#asterisk-dev tzafrir (~tzafrir@212.179.75.202) |
| 10:12.43 | *** mode/#asterisk-dev [+o tzafrir] by ChanServ |
| 10:41.00 | *** join/#asterisk-dev kpfleming (~kpfleming@asterisk/digium-director-of-software-tech/kpfleming) |
| 10:41.00 | *** mode/#asterisk-dev [+o kpfleming] by ChanServ |
| 10:48.09 | irroot | kpfleming go sleep :P morinig |
| 10:48.18 | irroot | s/morinig/morning/ |
| 10:55.04 | kpfleming | no, it's time to get up. an early day today. |
| 10:56.34 | Marquis42 | Early day? I'm only an hour ahead of you, and I've been up for 3 hours and at work for 1. :P |
| 10:59.03 | kpfleming | Marquis42: that's just sick. sick. |
| 11:05.07 | irroot | Marquis42 any objections to penalty rules been in a ao2 container not a linked list |
| 11:05.27 | irroot | and not be copied to the queue entry |
| 11:38.44 | Marquis42 | irroot: None at all |
| 11:39.12 | Marquis42 | kpfleming: Yeah, it's pretty crazy. And to make matters worse I usually work until 7pm or so before I go home. |
| 11:39.44 | kpfleming | and then you eat moderate size balls of wasabi paste |
| 11:39.51 | Marquis42 | Exactly! :D |
| 11:40.08 | irroot | wasabi is great for sinus problems !!! |
| 11:40.09 | Marquis42 | I foresee never living that moment down. :) |
| 11:40.56 | irroot | once convinced someone that wasabi was avocado garnishing with there sushi total sushi n00b |
| 11:41.13 | Marquis42 | irroot: That's just mean! But funny. ;) |
| 11:51.41 | *** join/#asterisk-dev black187 (5d67162a@gateway/web/freenode/ip.93.103.22.42) |
| 12:26.47 | irroot | penalty rules are now a linked list |
| 12:27.08 | irroot | oops a ao2 container but the rules in the list are still a linked list |
| 12:32.50 | *** join/#asterisk-dev mjordan (~mjordan@nat/digium/x-rpexhjlmlelvcsss) |
| 12:32.50 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 12:54.17 | *** part/#asterisk-dev black187 (5d67162a@gateway/web/freenode/ip.93.103.22.42) |
| 13:00.24 | *** join/#asterisk-dev mnicholson (~mnicholso@nat/digium/x-nmtwrpudawvuyroj) |
| 13:00.31 | *** mode/#asterisk-dev [+o mnicholson] by ChanServ |
| 13:03.10 | *** join/#asterisk-dev BlackBishop (~dexter@2001:470:26:45f::1) |
| 13:12.53 | irroot | on way home l8r |
| 13:41.10 | *** join/#asterisk-dev oej (~olle@ns.webway.se) |
| 13:41.11 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 13:41.28 | *** join/#asterisk-dev oej (~olle@ns.webway.se) |
| 13:41.28 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 13:58.05 | *** join/#asterisk-dev oej (~olle@ns.webway.se) |
| 13:58.05 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 13:59.18 | *** join/#asterisk-dev irroot (~irroot@197.175.43.236) |
| 14:16.43 | irroot | the digium folks have scored out of me today |
| 14:17.10 | irroot | 2x quad PRI 2x dual PRI and 2x 4port BRI |
| 14:21.05 | *** join/#asterisk-dev putnopvut (~putnopvut@asterisk/master-of-queues/mmichelson) |
| 14:21.05 | *** mode/#asterisk-dev [+o putnopvut] by ChanServ |
| 14:26.33 | Qwell | irroot: \o/ |
| 14:27.09 | irroot | Qwell lol hoping its a upgrade from macdonalds or xmas lunch ?? |
| 14:27.56 | irroot | its lightning season we go through a fair ammount of FXS/FXO modules so dont stress |
| 14:28.50 | Qwell | yeah...mcdonalds... that's it |
| 14:29.19 | *** join/#asterisk-dev BlackBishop (~dexter@ipv6.d3xt3r01.tk) |
| 14:32.02 | *** join/#asterisk-dev The_Boy_Wonder (~vossel@asterisk/batman-developer/dvossel) |
| 14:32.02 | *** mode/#asterisk-dev [+o The_Boy_Wonder] by ChanServ |
| 14:37.11 | *** join/#asterisk-dev kpfleming (~kpfleming@asterisk/digium-director-of-software-tech/kpfleming) |
| 14:37.11 | *** mode/#asterisk-dev [+o kpfleming] by ChanServ |
| 14:41.23 | *** join/#asterisk-dev sruffell (~Adium@asterisk/the-kernel-guy/sruffell) |
| 14:41.23 | *** mode/#asterisk-dev [+o sruffell] by ChanServ |
| 14:55.18 | *** join/#asterisk-dev otherwiseguy (~otherwise@69.76.221.79) |
| 15:09.47 | irroot | got a idea for app_queue and the queue struct / objects may take more memory on the downside when a queue is reconfigured [after cache expire / reload command / .... ] if i unlink the queue and it floats in memory until its un refed i a wont have locking problems and it wont affect those in call |
| 15:11.25 | kpfleming | but then you don't have a single queue any more... if new callers enter the queue, you can't put them into priority order ahead of existing callers if there are two copies of the queue |
| 15:13.27 | irroot | kpfleming there will be common bits the name for one but irrelevant the q->head [callers] and q->members will follow |
| 15:13.51 | irroot | q->members is a ao2 container and q->head is now a linkedlist with lock |
| 15:14.01 | kpfleming | so how does this help with locking then? |
| 15:14.37 | irroot | it eliminates need to lock the queue completely |
| 15:14.47 | irroot | its just a store of settings |
| 15:14.57 | irroot | and will never change once linked |
| 15:15.38 | irroot | the main concern is race conditions with settings changing in mid call |
| 15:15.41 | kpfleming | i suppose that could be useful then. tricky to manage, but doable. |
| 15:15.56 | kpfleming | essentially, you are implementing Read-Copy-Update (RCU) |
| 15:16.40 | kpfleming | and you're still going to have to lock the queue when the settings are being changed... but not when they are only being read |
| 15:17.08 | irroot | yip there enough hacks to the queue entry where settings are copied already to prevent changes |
| 15:18.38 | irroot | there a few settings that need to be preserved ie statistics but only its not available till linked |
| 15:19.15 | irroot | so locking is not needed until there is a change after linking |
| 15:21.02 | irroot | are we to assume that pointers and int are atomic for reading due to glibc or makesure all reads are protected somehow |
| 15:22.00 | kpfleming | well... that's a complex question |
| 15:22.12 | kpfleming | the simple answer is no, but not for the reasons you think |
| 15:22.30 | kpfleming | if you write to an 'int', that write *is* atomic (it will all be written at once, and appear in memory together) |
| 15:22.52 | kpfleming | however, without a memory barrier instruction, if the system has more than one CPU, the other CPUs could still see the old value |
| 15:24.44 | kpfleming | mutex locks use memory barriers to ensure memory consistency across all the CPUs/cores in the system |
| 15:25.26 | irroot | indeed so realtime sync is a problem in this case but not data coruption |
| 15:25.43 | *** join/#asterisk-dev oron_laptop (~oron_lapt@line25-214.adsl.actcom.co.il) |
| 15:25.51 | kpfleming | correct... no CPU will see a partial value. it will either be old or new. |
| 15:26.03 | kpfleming | the same goes for pointer variables and various other variables as well. |
| 15:26.15 | *** join/#asterisk-dev BlackBishop (~dexter@ipv6.d3xt3r01.tk) |
| 15:27.37 | irroot | seems to be a big performance hit making queue 100% atomic for read and write |
| 15:45.57 | *** join/#asterisk-dev Devon_ (~chatzilla@63.214.236.169) |
| 15:59.03 | *** join/#asterisk-dev BlackBishop (~dexter@ipv6.d3xt3r01.tk) |
| 16:00.20 | *** part/#asterisk-dev pnlarsson (~niklas@fw1.gml.g.icnet.infracom.se) |
| 16:05.38 | *** join/#asterisk-dev anthm (~anthm@freeswitch/developer/anthm) |
| 16:26.40 | *** join/#asterisk-dev elred_ (~elred_@unaffiliated/elred-/x-5010831) |
| 16:32.37 | *** part/#asterisk-dev apten (~apten@carbon.gonicus.de) |
| 16:40.08 | *** join/#asterisk-dev bklang (~bklang@2001:470:8:5de:225:ff:fe49:f8cd) |
| 16:51.53 | wdoekes2 | atomic ops should be available for integer updates, methinks. that's cheaper than mutexes, right? |
| 16:52.13 | kpfleming | yes, they are |
| 16:52.22 | Qwell | wdoekes2: there are a few atomic add/sub functions in use in Asterisk |
| 16:52.28 | kpfleming | they are compiler extensions though, not part of the C standard |
| 16:52.29 | Qwell | or inc/dec, if you will |
| 16:52.36 | kpfleming | but they are provided by GCC, so we can use them |
| 16:52.44 | kpfleming | as long as you don't build for an i386 :-) |
| 16:53.04 | Qwell | hmm. really? |
| 16:53.09 | wdoekes2 | perhaps something for irroot |
| 16:53.18 | Qwell | kpfleming: is it just more expensive on i386, or is it not actually atomic? |
| 16:53.34 | kpfleming | the GCC implementations aren't available on i386, they use instructions that appear on i586 |
| 16:53.47 | kpfleming | s/appear/appeared/ |
| 16:54.18 | irroot | ok folks thx im outta here gotta run .... have some ideas to toy with |
| 16:55.01 | irroot | oops logged onto a customer box about to log off lightning strike and this appeared |
| 16:55.08 | irroot | [Nov 4 18:54:17] NOTICE[3123]: chan_dahdi.c:3059 my_handle_dchan_exception: PRI got event: HDLC Bad FCS (8) on D-channel of span 1 |
| 19:09.14 | *** join/#asterisk-dev irroot (~irroot@41-135-133-79.dsl.mweb.co.za) |
| 19:20.46 | *** join/#asterisk-dev bklang (~bklang@tesla.alkaloid.net) |
| 19:40.07 | *** join/#asterisk-dev Devon_ (~chatzilla@198.211.95.6) |
| 20:10.18 | *** join/#asterisk-dev oej (~oej@2001:470:1f15:d79:9488:5011:62e:403c) |
| 20:10.19 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 21:01.22 | irroot | hey guys intresing Q in #ast a tool for converting conf to RT should be simple enough to write a tool for that |
| 21:02.06 | irroot | read a config file and write it to a rt engine |
| 21:49.43 | *** join/#asterisk-dev serafie (~erin@nat/digium/x-tusjycmfmgragbec) |
| 21:51.45 | *** part/#asterisk-dev mjordan (~mjordan@nat/digium/x-rpexhjlmlelvcsss) |
| 23:11.54 | *** join/#asterisk-dev CoderForLife (~Miranda@unaffiliated/coderforlife) |
| 23:30.34 | *** join/#asterisk-dev elred_ (~elred_@unaffiliated/elred-/x-5010831) |
| 23:44.09 | *** join/#asterisk-dev Tim_Toady (~fuzzy@195.74.224.253.dsl.dyn.forthnet.gr) |