[Xfce-bugs] [Bug 15184] Crash on mount/unmount on FreeBSD 11.2, 12.0, starting from 1.6.16

[bugzilla-daemon at xfce.org]

https://bugzilla.xfce.org/show_bug.cgi?id=15184

alexxcons <alexxcons at xfce.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEEDINFO                    |ASSIGNED
           Assignee|xfce-bugs at xfce.org          |alexxcons at xfce.org

--- Comment #18 from alexxcons <alexxcons at xfce.org> ---
Created attachment 8554
  --> https://bugzilla.xfce.org/attachment.cgi?id=8554&action=edit
possibly different fix

Glad that you fixed it ! The risk of calling "thunar_file_reload" was a good
spot, I did not know that.

Though I am a bit unhappy with the fix itself.

The guarding of thunar_file_reload really should be improved.  It is called ~20
times cross different files.  So if we are going to fix it, I would fix it
everywhere.

1. As far as I understood,"folder->corresponding_file" can get corrupt, not
"folder" itself ?  If so, I would use THUNAR_IS_FILE as direct check to see if
it got destroyed, rather than indirectly checking the a cancelled exo job.

2. You added "exo_job_cancel" before "g_signal_handlers_disconnect_matched " in
order to check if the job got cancelled later. My hope would be that this will
not be required any more, when using THUNAR_IS_FILE as check.
Though if still required, I guess we should add "exo_job_cancel" to all
occurances of "g_signal_handlers_disconnect_matched(folder->job,...)".

Attached a patch which uses THUNAR_IS_FILE to check ... does it as well fix
your crash ? If so, I would add a check + the comment as well in all other
locations where thunar_file_reload is called.

-- 
You are receiving this mail because:
You are the assignee for the bug.