Summary: | ASTERISK-18923: res_fax_spandsp usage counter is wrong | ||||
Reporter: | Grigoriy Puzankin (boroda) | Labels: | |||
Date Opened: | 2011-11-25 02:46:38.000-0600 | Date Closed: | 2014-10-14 11:34:04 | ||
Priority: | Minor | Regression? | |||
Status: | Closed/Complete | Components: | Resources/res_fax_spandsp | ||
Versions: | 1.8.7.1 | Frequency of Occurrence | Constant | ||
Related Issues: |
| ||||
Environment: | Attachments: | ( 0) module_ref.diff ( 1) Screen_Shot_2012-12-17_at_2.17.33_PM_copy.png | |||
Description: | When I send fax using SendFAX (from res_fax_spandsp) it increases module usage counter by 2. When fax completes it decreases module usage by 1. So after each iteration counter grows by 1. {noformat} Module Description Use Count res_fax.so Generic FAX Applications 1 res_fax_spandsp.so Spandsp G.711 and T.38 FAX Technologies 11 2 modules loaded Current FAX Sessions: Channel Tech FAXID Type Operation State File(s) 0 FAX sessions {noformat} | ||||
Comments: | By: ibercom (ibercom) 2012-03-15 14:20:26.061-0500 It is a problem because you can't unload this module. The 'Use Count' always is >0. I don't use SendFAX, I use ReceiveFAX and its value is incremented by one every time I call to ReceiveFax. By: John Hass (john8675309tm) 2012-12-17 14:20:47.572-0600 I am also experiencing this issue, the attached pic shows Current Fax sessions that aren't there at all. What I am afraid of more then anything is switching to the digium res_fax and having all the licenses used up, because the counter is not accurate. By: Niklas Larsson (pnlarsson) 2013-01-31 06:43:06.333-0600 I have the same issue when using res_fax_spandsp in T38 gateway mode. Below are some info from asterisk: {code} sipgwdev*CLI> fax show sessions Current FAX Sessions: Channel Tech FAXID Type Operation State File(s) SIP/siptrunk01-gw-00 Spandsp 11 none gateway Active SIP/siptrunk01-gw-00 Spandsp 3 none gateway Active SIP/siptrunk03-gw-00 Spandsp 15 none gateway Active SIP/siptrunk03-gw-00 Spandsp 7 none gateway Active SIP/siptrunk01-gw-00 Spandsp 19 none gateway Active 5 FAX sessions sipgwdev*CLI> sip show channels Peer User/ANR Call ID Format Hold Last Message Expiry Peer 0 active SIP dialogs sipgwdev*CLI> core show channels Channel Location State Application(Data) 0 active channels 0 active calls 27 calls processed sipgwdev*CLI> core show version Asterisk SVN-branch-10-r379777 built by root @ sipgwdev on a x86_64 running Linux on 2013-01-30 09:29:54 UTC sipgwdev*CLI> fax show version FAX For Asterisk Components: Applications: SVN-branch-10-r379777 Spandsp FAX Driver: 20110122 075024 {code} By: Niklas Larsson (pnlarsson) 2013-02-04 09:55:03.690-0600 I added REF_DEBUG to res_fax.c and res_fax_spandsp.c, sent a fax through (GATEWAY mode) and run refcounter: {code} ============== Problem: net Refcount not zero for object 28102f78 Object 28102f78 history: 0x7f5e28102f78 =1 res_fax.c:981:fax_session_new () 0x7f5e28102f78 +1 res_fax.c:1052:fax_session_new () [@1] 0x7f5e28102f78 +1 res_fax_spandsp.c:750:spandsp_fax_gw_gen_alloc () [@2] 0x7f5e28102f78 +1 res_fax.c:967:fax_session_new () [@3] 0x7f5e28102f78 +1 res_fax.c:1052:fax_session_new () [@4] 0x7f5e28102f78 -1 res_fax.c:2582:fax_gateway_start () [@5] 0x7f5e28102f78 -1 res_fax_spandsp.c:755:spandsp_fax_gw_gen_release () [@4] 0x7f5e28102f78 +1 res_fax_spandsp.c:750:spandsp_fax_gw_gen_alloc () [@3] 0x7f5e28102f78 -1 res_fax_spandsp.c:755:spandsp_fax_gw_gen_release () [@4] 0x7f5e28102f78 -1 res_fax.c:2506:destroy_gateway () [@3] 0x7f5e28102f78 -1 res_fax.c:2509:destroy_gateway () [@2] ============== {code} By: Johann Steinwendtner (steinwej) 2013-08-20 07:32:26.427-0500 I 've the same problem on Asterisk 11.5.0. To me it seems that the module counter gets incremented when a new fax session is reserved and also when it is requested. (so IMHO twice) see function: fax_session_new and fax_session_reserve. My current workaround is in fax_session_new: if (!reserved) ast_module_ref(faxmod->tech->module); But that should be approved. By: Johann Steinwendtner (steinwej) 2014-01-28 10:06:26.151-0600 patch against 11.7.0 By: Matt Jordan (mjordan) 2014-10-14 12:04:51.819-0500 [~steinwej] - Corey Farrell had a slightly different fix for this that went up on review board - but thanks for your patch. Can you confirm that Corey's fix (which just got committed to the various branches) fixes the issue for you as well? |