| 15:44.22 | *** join/#asterisk-dev infobot (~infobot@rikers.org) |
| 15:44.22 | *** topic/#asterisk-dev is Asterisk Development Discussion -=- http://www.asterisk.org/developers -=- Tier 2 and 3.14159265 support is in #asterisk -=- Tier 4 support is in #2,000 |
| 15:51.56 | *** join/#asterisk-dev Juggie (~Juggie@unaffiliated/juggie) |
| 16:17.01 | kpfleming | nice spam on asterisk-dev list :-) |
| 16:18.51 | seanbright | ass hole |
| 16:18.59 | mjordan | I enjoyed how it got resent to him in the next digest |
| 16:25.20 | pabelanger | lol |
| 16:25.21 | pabelanger | fail |
| 16:25.31 | pabelanger | go to http://astricon.net |
| 16:25.46 | pabelanger | Just got a big pop up |
| 16:25.56 | pabelanger | www.astricon.net contains content from cssglobe.com, a site known to distribute malware. Your computer might catch a virus if you visit this site. |
| 16:25.59 | kpfleming | oh good grief |
| 16:26.15 | kpfleming | notifying marketing now, thanks |
| 16:26.36 | pabelanger | Was trying to see when proposals are going out |
| 16:26.50 | kpfleming | going out? |
| 16:27.00 | kpfleming | do you mean the request for proposals? |
| 16:27.30 | pabelanger | sorry, when speakers are getting notified of acceptance |
| 16:27.35 | kpfleming | ahh |
| 16:27.39 | Kobaz | rfp's! |
| 16:29.39 | file | NEVER! |
| 16:30.33 | file | wouldn't mind knowing too |
| 16:30.54 | kpfleming | i would be surprised if that happened any sooner than a couple of weeks from now |
| 16:31.10 | file | kpfleming, I would be surprised... nevermind |
| 17:29.15 | *** join/#asterisk-dev oej (~olle@h87-96-134-129.dynamic.se.alltele.net) |
| 17:29.15 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 18:02.35 | kpfleming | leifmadsen: you suggested basically the same thing i did :-) |
| 18:05.23 | leifmadsen | oh I didn't see your response :) |
| 18:05.28 | leifmadsen | was it on the mailing list? |
| 18:06.02 | kpfleming | yes |
| 18:18.52 | pabelanger | leifmadsen: ASTERISK-17837 |
| 18:18.58 | pabelanger | just ran into that issue too |
| 18:22.14 | leifmadsen | huh interesting... |
| 18:22.41 | leifmadsen | ya there must be some check in there that stops that which is probably no longer necessary? |
| 18:22.56 | leifmadsen | looks like I tested and confirmed it happens with extconfig.conf, but not with voicemail.conf |
| 18:31.38 | pabelanger | right, extconfig.conf is the only one affected |
| 18:33.52 | mjordan | pabelanger: that's rather odd |
| 18:34.23 | pabelanger | mjordan: Ya, trying to figure it out now. But some crazy logic |
| 18:37.07 | *** join/#asterisk-dev citywok (~citywok@67-134-194-33.dia.static.qwest.net) |
| 19:48.20 | *** join/#asterisk-dev oej (~olle@h87-96-134-129.dynamic.se.alltele.net) |
| 19:48.20 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 19:49.59 | *** join/#asterisk-dev Juggie (~Juggie@unaffiliated/juggie) |
| 19:54.59 | *** join/#asterisk-dev Juggie (~Juggie@unaffiliated/juggie) |
| 20:00.58 | *** join/#asterisk-dev Juggie (~Juggie@unaffiliated/juggie) |
| 20:05.08 | *** join/#asterisk-dev jgowdy (~jgowdy@basharteg.com) |
| 20:05.19 | jgowdy | Am I correct in thinking that AST_LIST_FIRST isn't thread safe? |
| 20:05.22 | jgowdy | Dereferences head which could be NULL in the case of an empty list it seems |
| 20:05.42 | russellb | AST_LIST_ANYTHING isn't thread-safe |
| 20:05.55 | russellb | except AST_LIST_LOCK/UNLOCK/RDLOCK/WRLOCK |
| 20:05.56 | russellb | :) |
| 20:06.09 | jgowdy | that's what it looked like to me |
| 20:06.29 | jgowdy | I'm trying to figure out if app_externalivr.c is doing some thread unsafe operations on lists |
| 20:06.42 | russellb | probably not, only one thread should be running in there |
| 20:06.48 | russellb | it doesn't spawn any threads |
| 20:07.02 | russellb | one thread per-channel/connection/whatever |
| 20:07.27 | jgowdy | because I'll run for a while with many externalivrs running, and then suddenly I get streams of externalivr child process exits |
| 20:07.46 | jgowdy | Nothing in the full log but child process exiting |
| 20:08.07 | russellb | OOM? |
| 20:08.27 | russellb | just a guess |
| 20:08.35 | jgowdy | nope, I have tons of memory on these blades, not out of file handles |
| 20:08.39 | jgowdy | everything looks kosher |
| 20:08.49 | jgowdy | I have SNMP metrics on the memory load and whatnot |
| 20:09.08 | russellb | punts |
| 20:09.18 | jgowdy | yeah, I'll dig in some more |
| 20:09.41 | jgowdy | I noticed 10.5-rc1 had added some locking around what was probably previously unsafe list access that isn't present in 10.4 |
| 20:10.01 | seanbright | you're welcome |
| 20:10.02 | seanbright | heh |
| 20:10.29 | jgowdy | thanks |
| 20:10.43 | jgowdy | oh snap, that's not just additional locking, it's a null check that wasn't there either |
| 20:10.46 | jgowdy | hah |
| 20:10.46 | jgowdy | thanks again |
| 20:10.48 | seanbright | totally kidding. but yeah, sending an 'S' command right off would basically guarantee a crash. |
| 20:11.09 | seanbright | now sure if it is related to your issue or not. |
| 20:11.13 | russellb | oh right, it is threaded |
| 20:11.17 | russellb | tcp connection thread |
| 20:11.29 | russellb | spawned outside of app_externalivr.c so i didn't see it |
| 20:11.39 | seanbright | sounds like he is using fork() though |
| 20:12.05 | jgowdy | AST_LIST_LOCK is recursive I'm assuming |
| 20:12.12 | russellb | unfortunately yes |
| 20:13.41 | jgowdy | this app is so small to have issues like this ;) |
| 20:14.03 | russellb | the joys of C code |
| 20:14.10 | jgowdy | yeah, seriously |
| 20:14.43 | jgowdy | We've moved all of our telephony code to C# |
| 20:14.57 | seanbright | now you have 2 problems |
| 20:15.00 | seanbright | i kid. |
| 20:15.02 | jgowdy | hehe |
| 20:15.02 | seanbright | i do that. |
| 20:15.24 | jgowdy | It's a great language. We were a pure C/C++ shop for over 10 years before that. |
| 20:15.44 | seanbright | yeah. used to be my language of choice as well. |
| 20:18.07 | jgowdy | yeah the current app_externalivr.c seems written on the basis that AST_LIST_FIRST on the playlist doesn't require a lock |
| 20:18.11 | jgowdy | I hope to hell that's my issue |
| 20:18.31 | seanbright | what is "current"? |
| 20:18.40 | jgowdy | Well, I'm on 10.4 release |
| 20:19.06 | seanbright | and where are you seeing unsafe access? |
| 20:19.09 | seanbright | what line? |
| 20:19.35 | seanbright | gen_nextfile? |
| 20:19.45 | jgowdy | like 244 uses AST_LIST_FIRST in the if statement prior to locking it |
| 20:19.53 | jgowdy | gen_readframe |
| 20:20.25 | seanbright | true story |
| 20:20.27 | jgowdy | If the function presumes the lock is held the same way gen_nextfile is documented, there's no need to lock the list again a few lines in |
| 20:20.51 | jgowdy | I'll see what I can mop up and make a diff for submission |
| 20:33.12 | jgowdy | Oh nice, gen_nextfile claims that the caller has the playlist locked in a comment, but that's a lie |
| 20:33.40 | seanbright | indeed. |
| 20:34.16 | seanbright | that's an optimization |
| 20:34.23 | seanbright | 1/2 the time it's really fast |
| 20:35.56 | jgowdy | At least the problems here appear to be obvious. Better than if they were subtle. |
| 20:41.54 | *** join/#asterisk-dev russellb_ (~russellb@redhat/russellb) |
| 20:41.54 | *** mode/#asterisk-dev [+o russellb_] by ChanServ |
| 20:47.22 | *** join/#asterisk-dev putnopvut (~putnopvut@asterisk/master-of-queues/mmichelson) |
| 20:47.23 | *** mode/#asterisk-dev [+o putnopvut] by ChanServ |
| 20:47.23 | *** join/#asterisk-dev jrose_atDigium (~jrose_atD@nat/digium/x-epxwybwllcvohgrt) |
| 20:53.34 | *** join/#asterisk-dev seanbright (~sean@asterisk/contributor-and-bug-marshal/seanbright) |
| 20:53.34 | *** mode/#asterisk-dev [+o seanbright] by ChanServ |
| 21:12.20 | *** join/#asterisk-dev WIMPy (~wimpy@e183095026.adsl.alicedsl.de) |
| 21:28.54 | *** join/#asterisk-dev infobot (~infobot@rikers.org) |
| 21:28.54 | *** topic/#asterisk-dev is Asterisk Development Discussion -=- http://www.asterisk.org/developers -=- Tier 2 and 3.14159265 support is in #asterisk -=- Tier 4 support is in #2,000 |
| 22:18.23 | *** join/#asterisk-dev citywok (~citywok@67-134-194-33.dia.static.qwest.net) |
| 22:31.45 | *** part/#asterisk-dev mjordan (~mjordan@nat/digium/x-htenloyurzluzupq) |
| 23:35.17 | *** join/#asterisk-dev seanbright (~sean@asterisk/contributor-and-bug-marshal/seanbright) |
| 23:35.17 | *** mode/#asterisk-dev [+o seanbright] by ChanServ |