FreeSWITCH: Runtime issues (Aborted)

[BlueMax]

Starting 'freeswitch' just echoes 'Aborted' (uClibc message) and the process dies. I've never seen Freeswitch in action but that probably is not correct. Arguments '-h' and '-version' do work though.

Edit: The two commits (94fd5f5 163874a) fix building but do not lead to sane binaries. The same compiling issues are handled already in 'configure.ac' but they are currently limited to ARM hosts(?). I will test how that affects the build (as there are even more flags set).

case "$host" in
     arm*-linux-gnueabi*|arm*-*-linux-*)
[...]
apr_cv_tcp_nodelay_with_cork=yes
ac_cv_file_dbd_apr_dbd_mysql_c=no
[...]

Edit: Bingo! Seems to work. Will revoke the two commits and prepare some new...

[BlueMax]

The $hosts=arm* templates/overrides seem to be ok. Now its running (tested on x86).

[sageahn]

I got the upstream repo today(Apr 23, 2015) and still got the same "aborted" message from x86 and VirtualBox image environment. any comment please.

[BlueMax]

Mazi Lo (maintainer) took a different approach. You can try this patch (put in feeds/telephony/net/freeswitch/patches) that actually reflects my patches (and worked for me). Also create this patch (BlueMax@5d07b2b) to fix this bug.
Please report if it works.

[sageahn]

I am at c5e7c03 in feeds/telephony repo. those two patches you mentioned are already applied. still doesn't work. (aborted)

[BlueMax]

The sofia.xml patch might have been patched upstream (idk) but my configure.ac patches are not included for sure. See here: (https://github.com/openwrt/telephony/blob/master/net/freeswitch/patches/configure_ac.patch). Compare with version above.

[sageahn]

Hi BlueMax,

I noticed that your patch and Mazi's patch are different. however. those patches do not fix this issue. still found the aborted message unfortunately.

[BlueMax]

Then i don't know, sorry. It worked for me at that time.

[skyblaster]

Just testing FreeSWITCH from the 15.05-rc1 package repo on a WNDR3700v4 and am getting the Aborted message right away. Not sure how to provide any useful information here since nothing is written to the logs.
I see the package revisions are the same between rc1, rc2 (just released today), and snapshot, so am guessing they're all broken....at least for the ar71xx target.

[skyblaster]

Just confirmed the same error with 15.05-rc2.

[micmac1]

Hello @BlueMax @sageahn @skyblaster

freeswitch-stable got added to the tree as a drop-in replacement and it doesn't have this issue. Can you please try it and consider closing this issue?

Kind regards,
Sebastian

[skyblaster]

I've been running a snapshot build of freeswitch-stable v1.6.17 from a late May/early June on a RPi2 without a hitch.
Would love to see a cherry picked build added to branch:lede-17.01

[jslachta]

Hi @skyblaster

freeswitch-stable is not the same package as freeswitch, the branch:lede-17.01 should receive only security&bug fixes.

[skyblaster]

I'm well aware of the freeswitch vs freeswitch-stable packages. It's freeswitch-stable that I've been running successfully since June 2017.
Due to #118, I don't think freeswitch even builds. It may still suffer from the aborted message that this ticket covers.

Regarding branch:lede-17.01, the last commit was over a year ago. I see you and @jow- were set to fix that with #132, but no actual backporting occurred. Should that ticket be reopened?

[skyblaster]

I don't know when FreeSWITCH 1.8 will be finalised, but perhaps @micmac1 could cherry pick a 1.6.19 build to backport to branch:lede-17.01 ? Then when 1.8 comes out you could start a new/separate package?