Version-Release number of selected component: 1:epiphany-3.34.1-1.fc31 Additional info: reporter: libreport-2.10.1 backtrace_rating: 4 cgroup: 0::/user.slice/user-1000.slice/user/dbus\x2d:1.2\x2dorg.gnome.Epiphany.SearchProvider.slice/dbus-:1.2-org.gnome.Epiphany.SearchProvider cmdline: /usr/libexec/epiphany-search-provider crash_function: g_error_free executable: /usr/libexec/epiphany-search-provider journald_cursor: s=0d51de3cdc1e4e6a9da36b6d0a968675;i=aa7c7;b=1b843d0adca74e20afc1cffbd9754cdd;m=4816acd43;t=5965bed64652e;x=10511ed20b0f9304 kernel: 5.3.7-301.fc31.x86_64 rootdir: / runlevel: N 5 type: CCpp uid: 1000
Created attachment 1631829 [details] File: backtrace
Created attachment 1631830 [details] File: core_backtrace
Created attachment 1631831 [details] File: cpuinfo
Created attachment 1631832 [details] File: dso_list
Created attachment 1631833 [details] File: environ
Created attachment 1631834 [details] File: limits
Created attachment 1631835 [details] File: maps
Created attachment 1631836 [details] File: mountinfo
Created attachment 1631837 [details] File: open_fds
Created attachment 1631838 [details] File: proc_pid_status
Hi, any chance you have some idea of how to reproduce this? I see 316 reports and from 174 users on FAF, which is a lot. But the backtrace indicates memory corruption and the crashing code is clearly innocent, so it's not actionable without using valgrind or an asan build. Starting a search provider with valgrind will be slightly awkward as we'll need to edit the D-Bus service file, but if we know how to reproduce the crash then that will be worth doing.
*** Bug 1783431 has been marked as a duplicate of this bug. ***
Hm, we have thousands of reports of this, all occurring in exactly the same place. It would be weird for this to be memory corruption from somewhere else, like I had been thinking. Maybe something is wrong with the error returned by ephy_bookmarks_manager_save_sync(). The crash probably only occurs when you have no ~/.local/share/bookmarks.gvdb, so deleting that file is going to be a requirement to reproduce, but that doesn't seem sufficient.
OK, this turned out to be a simple problem. Sorry I didn't notice the first time I looked at it. This will be solved in 3.34.3.
*** Bug 1787172 has been marked as a duplicate of this bug. ***