Opened 13 years ago
Closed 13 years ago
#4675 closed defect (fixed)
Crash if using -v upnp
Reported by: | Nigel | Owned by: | Nigel |
---|---|---|---|
Priority: | minor | Milestone: | unknown |
Component: | upnp | Version: | unknown |
Severity: | medium | Keywords: | |
Cc: | Ticket locked: | no |
Description
On OS X, something funky happens if setting -v upnp (I tried all the -v flags to be sure):
MacBook% ../bin/mythshutdown -v mheg mythshutdown: Could not initialize myth context. Exiting. MacBook% ../bin/mythshutdown -v upnp mythshutdown: Could not initialize myth context. Exiting. zsh: bus error ../bin/mythshutdown -v upnp
probably something in libmythupnp, but could also be MythContext. Stack trace outside of GDB looks corrupt. Running under GDB does not crash. Going to have to do this the hard way.
Attachments (1)
Change History (6)
comment:1 Changed 13 years ago by
Resolution: | → worksforme |
---|---|
Status: | new → closed |
comment:2 Changed 13 years ago by
I'm sure there is something wrong somewhere :-(
When looking at ticket #4664 I noticed mythshutdown was segfaulting if I added the upnp verbose flag. I'll attach the bt in case you find it useful. As you can see it looks like something is wrong in the SSDPCache destructor. I works OK so long as I don't add upnp to the verbose flags.
comment:3 Changed 13 years ago by
I think we can problem comment or remove the VERBOSE from the SSDPCache dtor, at least for 0.21?
comment:4 Changed 13 years ago by
Resolution: | worksforme |
---|---|
Status: | closed → new |
In libmythupnp, I changed all the VB_UPNP to VB_EIT. Problem went away. I changed them back to VB_UPNP. Problem returned. Went file by file, narrowed it down to the VERBOSE lines in SSDPCache::~SSDPCache() and UPnpDeviceDesc::~UPnpDeviceDesc().
Several tweaks later, the problem has magically gone away. Will assume this was some sort of freak build/cache/cosmic-ray problem?