| 00:06.58 | *** join/#asterisk-bugs Corydon76-dig (i=two@pdpc/supporter/bronze/Corydon76-home) |
| 00:06.58 | *** mode/#asterisk-bugs [+o Corydon76-dig] by ChanServ |
| 01:30.41 | *** join/#asterisk-bugs implicit (n=bayan@unaffiliated/implicit) |
| 01:30.41 | *** mode/#asterisk-bugs [+o implicit] by ChanServ |
| 01:38.12 | *** join/#asterisk-bugs Deeewayne (i=dwayne@76.29.245.9) |
| 01:38.12 | *** mode/#asterisk-bugs [+o Deeewayne] by ChanServ |
| 01:53.10 | *** join/#asterisk-bugs Corydon76-dig (i=two@pdpc/supporter/bronze/Corydon76-home) |
| 01:53.10 | *** mode/#asterisk-bugs [+o Corydon76-dig] by ChanServ |
| 02:28.37 | *** join/#asterisk-bugs fakhir (n=fakhir@unaffiliated/fakhir) |
| 03:06.07 | *** join/#asterisk-bugs lmadsen (n=Leif@asterisk/documenteur-extraordinaire/blitzrage) |
| 03:06.07 | *** mode/#asterisk-bugs [+o lmadsen] by ChanServ |
| 03:50.54 | *** join/#asterisk-bugs implicit (n=bayan@unaffiliated/implicit) |
| 03:50.54 | *** mode/#asterisk-bugs [+o implicit] by ChanServ |
| 07:38.15 | *** join/#asterisk-bugs tzafrir_laptop (n=tzafrir@local.xorcom.com) |
| 07:42.40 | *** part/#asterisk-bugs dzajr0 (n=dzajro@ca-8171.fw.nextra.sk) |
| 08:39.30 | *** join/#asterisk-bugs oej (n=olle@193.136.47.146) |
| 08:39.41 | *** join/#asterisk-bugs IsUp (n=nocturne@88.255.77.200) |
| 08:43.59 | *** join/#asterisk-bugs dzajr0 (n=dzajro@ca-8171.fw.nextra.sk) |
| 10:47.10 | *** join/#asterisk-bugs flujan (n=flujan@internet.nube.com.br) |
| 11:05.49 | *** join/#asterisk-bugs caio1982 (i=caio1982@CAcert-br/caio1982) |
| 12:25.58 | *** join/#asterisk-bugs lmadsen (n=Leif@asterisk/documenteur-extraordinaire/blitzrage) |
| 12:25.58 | *** mode/#asterisk-bugs [+o lmadsen] by ChanServ |
| 12:47.37 | *** join/#asterisk-bugs russellb (n=russell@asterisk/digium-open-source-team-lead/russellb) |
| 12:47.37 | *** mode/#asterisk-bugs [+o russellb] by ChanServ |
| 12:48.48 | *** join/#asterisk-bugs jmls1 (n=Julian@host217-36-208-155.in-addr.btopenworld.com) |
| 13:45.16 | *** join/#asterisk-bugs bweschke (n=BJ_Wesch@74.94.3.214) |
| 14:11.16 | *** join/#asterisk-bugs putnopvut (n=putnopvu@nat/digium/x-d8e055d05e9099fd) |
| 14:11.16 | *** mode/#asterisk-bugs [+o putnopvut] by ChanServ |
| 14:17.37 | *** join/#asterisk-bugs bbryant (n=Brett_Br@adsl-153-43-19.chs.bellsouth.net) |
| 14:17.37 | *** mode/#asterisk-bugs [+o bbryant] by ChanServ |
| 14:28.14 | *** join/#asterisk-bugs Deeewayne (n=Deeewayn@nat/digium/x-867a9fdc54cca32c) |
| 14:28.14 | *** mode/#asterisk-bugs [+o Deeewayne] by ChanServ |
| 15:47.35 | flujan | putnopvut: If you need further information about that bug M13538 let me know... I will be here! :) |
| 15:47.45 | putnopvut | flujan: okay, thanks. |
| 15:51.43 | *** join/#asterisk-bugs russellb (n=russellb@asterisk/digium-open-source-team-lead/russellb) |
| 15:51.43 | *** mode/#asterisk-bugs [+o russellb] by ChanServ |
| 16:18.18 | *** part/#asterisk-bugs bbryant (n=Brett_Br@adsl-153-43-19.chs.bellsouth.net) |
| 16:28.58 | *** part/#asterisk-bugs dzajr0 (n=dzajro@ca-8171.fw.nextra.sk) |
| 16:53.47 | *** join/#asterisk-bugs mvanbaak_ (n=michiel@asterisk/contributor-and-bug-marshal/mvanbaak) |
| 17:01.38 | *** join/#asterisk-bugs implicit (n=bayan@unaffiliated/implicit) |
| 17:01.38 | *** mode/#asterisk-bugs [+o implicit] by ChanServ |
| 17:30.36 | *** join/#asterisk-bugs atis_work (n=atis_wor@193.238.212.171) |
| 17:38.27 | *** join/#asterisk-bugs tzafrir_laptop (n=tzafrir@local.xorcom.com) |
| 17:59.35 | *** join/#asterisk-bugs caio1982 (i=caio1982@CAcert-br/caio1982) |
| 18:35.15 | *** join/#asterisk-bugs fakhir (n=fakhir@unaffiliated/fakhir) |
| 19:09.43 | *** join/#asterisk-bugs atis_work (n=atis_wor@193.238.212.171) |
| 19:52.53 | *** join/#asterisk-bugs wolfelectronic (n=wolfelec@91.112.227.150) |
| 20:09.23 | *** join/#asterisk-bugs atis_work (n=atis_wor@193.238.212.171) |
| 21:05.28 | *** join/#asterisk-bugs tzafrir_laptop (n=tzafrir@bzq-179-75-202.static.bezeqint.net) |
| 21:09.59 | *** join/#asterisk-bugs atis_work (n=atis_wor@193.238.212.171) |
| 21:14.40 | wolfelectronic | seanbright? |
| 21:15.29 | lmadsen | seanbright: does it bother you that your name is an oxymoron? |
| 21:15.33 | lmadsen | ducks :) |
| 21:17.03 | wolfelectronic | seanbright? |
| 21:21.39 | seanbright | what's up? |
| 21:23.17 | seanbright | wolfelectronic: in regards to 13668, i'm not really a chan_sip guy |
| 21:23.37 | wolfelectronic | altough being an asterisk newbe i am an programmer senior, 15 years practise in c/C++. so i wonder if the fix for this bug is working well, i am not shure |
| 21:24.02 | seanbright | ok, let me look at putnopvut's fix |
| 21:24.36 | wolfelectronic | take a look at if (inuse && *inuse > 0 && ast_test_flag(&fup->flags[0], SIP_INC_COUNT)) { |
| 21:24.37 | wolfelectronic | <PROTECTED> |
| 21:24.37 | wolfelectronic | <PROTECTED> |
| 21:24.37 | wolfelectronic | <PROTECTED> |
| 21:24.37 | wolfelectronic | <PROTECTED> |
| 21:24.43 | putnopvut | I see my name! |
| 21:24.48 | putnopvut | M13668 |
| 21:24.50 | MuffinMan | [closed] [Asterisk] Channels/chan_sip/General 0013668: sip show inuse count is negative reported by mjc (Karma: +1.00) http://bugs.digium.com/view.php?id=13668 |
| 21:24.58 | seanbright | ok, i'm looking at it. |
| 21:25.23 | seanbright | what do you think is wrong with it? |
| 21:25.26 | wolfelectronic | first there is a check if unuse is 0 and if it is 0 the *inuse=0 should crash |
| 21:25.53 | seanbright | correct, that is a bug |
| 21:26.04 | seanbright | putnopvut: fix it |
| 21:26.07 | seanbright | heh |
| 21:26.12 | putnopvut | wolfelectronic: the check for inuse being non-zero is just a formality. It should be impossible for inuse to be NULL. |
| 21:26.16 | wolfelectronic | as i understand the function it can be called by more than one task |
| 21:26.53 | wolfelectronic | otherwhise there would be no need for ast_atomic_fetch |
| 21:27.02 | putnopvut | wolfelectronic: that is correct. |
| 21:28.12 | wolfelectronic | but i imaghine what hapens if there is a taskswitch between the if statment and clearing the flag or decrement the counter. should there be al lock and alle that if statement be inside a lock? |
| 21:31.30 | putnopvut | Oh, so you're suggesting that perhaps *inuse may get set to 0 between the if and the ast_atomic_fetchadd_int? |
| 21:31.57 | seanbright | good catch |
| 21:32.01 | seanbright | give him commit access |
| 21:32.17 | putnopvut | lol |
| 21:32.50 | wolfelectronic | no, not only the zero pointer, but the also the state of the flag and the counter |
| 21:33.46 | putnopvut | The state of the flag? What do you mean? |
| 21:34.11 | wolfelectronic | ast_test_flag(&fup->flags[0], SIP_INC_COUNT and ast_clear_flag(&fup->flags[0], SIP_INC_COUNT) |
| 21:34.43 | wolfelectronic | what if another task changes the state of the flag between the compare and the change of the flag |
| 21:35.37 | wolfelectronic | lets say the other task sets the flag between the compare and the clear ogf the flag |
| 21:35.40 | putnopvut | wolfelectronic: ah, I see what you're saying now. |
| 21:36.05 | wolfelectronic | same is should apply for the younter. |
| 21:36.17 | putnopvut | Placing a lock around that section of code would solve all of those problems. |
| 21:36.37 | wolfelectronic | yes, and the atomic_add could be removed |
| 21:36.55 | putnopvut | correct. |
| 21:37.31 | wolfelectronic | i am not quite shure if a lock around the whole function would suffer, locks arround many lines is often a performance issue |
| 21:38.34 | wolfelectronic | i also have no idea how much runtime a lock costs. (i am programming in industrial automation where performa´nce and runtime is a big isue) |
| 21:38.44 | putnopvut | In this case the only operations occurring inside the lock are a few comparisons, incrementing or decrementing a value, and then clearing or setting a flag. I can work at making the locking as granular as possible, but to be honest, a lock around the entire switch statement would probably be fine. |
| 21:38.55 | seanbright | the only possibly expensive operation in that code block is sip_pee_hold |
| 21:38.57 | seanbright | heh... |
| 21:39.03 | seanbright | sip_peeR_hold |
| 21:39.50 | seanbright | but i haven't looked at that function |
| 21:40.50 | wolfelectronic | well i see some ast_debug lines, not shure if its good to call this inside a lock |
| 21:40.52 | putnopvut | seanbright: yeah, probably can release the lock for that function. |
| 21:41.30 | seanbright | yeah... does a find_peer |
| 21:41.38 | seanbright | lame. |
| 21:41.38 | putnopvut | wolfelectronic: ast_debug in 1.6.x is not nearly as bad as in 1.4 since the actual I/O occurs in a separate thread. |
| 21:42.17 | wolfelectronic | well, as stated i am a asterisk-dev newbe, i just don't like calling logging stuff inside a lock |
| 21:42.31 | seanbright | newbie or not, you identified a problem |
| 21:42.40 | wolfelectronic | i am glad |
| 21:43.10 | putnopvut | I have to say, though, if we made the policy not to make logging calls when a lock is held, we'd have to change a LOT of code :) |
| 21:43.17 | seanbright | yeah |
| 21:43.39 | wolfelectronic | the main reason is that i have just the decribed problem (inuse < 1) at a customer site |
| 21:43.43 | seanbright | i think off-loading to another thread is going to have to be good enough |
| 21:44.00 | seanbright | you mean inuse < 0, right? |
| 21:44.17 | wolfelectronic | yes, sorry for the typo |
| 21:44.17 | putnopvut | Yeah, it's valid for inuse to be == 0. |
| 21:44.53 | putnopvut | All right. I'll re-open the issue and note that you're still having this problem. |
| 21:45.11 | wolfelectronic | it's the same site where i fixed taht misdn outtgoing problem, seanbright |
| 21:46.25 | wolfelectronic | well, at the site the change is not applied, i have running 1.6 RC6 there, and the patch was only applied to the trunk |
| 21:46.52 | wolfelectronic | i just checdked the changes before paching the system |
| 21:47.46 | wolfelectronic | so i cannot say for shure if the fix from putnopvut works on a real system. i just examined the code |
| 21:48.09 | putnopvut | I already re-opened the issue. |
| 21:49.42 | *** join/#asterisk-bugs Deeewayne (n=Deeewayn@nat/digium/x-b7ce0cf963dade95) |
| 21:49.42 | *** mode/#asterisk-bugs [+o Deeewayne] by ChanServ |
| 21:53.02 | wolfelectronic | putnopvut: is there anything else i can do to assist you regarinding this issue? |
| 21:53.27 | putnopvut | wolfelectronic: no. Thanks for bringing the problem to my attention, though. |
| 21:54.13 | wolfelectronic | no problem. i will monitor the issue in bugtracker, so just ad some note an i will get an email. |
| 21:55.28 | wolfelectronic | so then, seanbright, putnopvut have i nice day, it's 11:57 pm local time and i have some other things to fix, so i will leave now |
| 21:55.45 | putnopvut | Have a good night! |
| 21:55.53 | wolfelectronic | bye |
| 22:54.35 | *** part/#asterisk-bugs wolfelectronic (n=wolfelec@91.112.227.150) |
| 23:39.51 | De_Mon | ast_pee_hold, priceless! |