Jeremy Soller authored Jun 11, 2019 and Iain Lane committed Jun 28, 2019

(cherry picked from commit f5449adb)

Fixes #112

See also: https://github.com/flatpak/flatpak/issues/2782

If we tried to get a preview thumbnail for a file, but that preview icon
did not exist, then we'd generate a warning trying to add a reference to
that GIcon.

Fixes: e629e46a

Mart Raudsepp authored Dec 13, 2018 and Bastien Nocera committed Jan 30, 2019

This is especially important for libstdc++ on distributions that
don't have it directly in a libdir and the runtime linker doesn't
look where needed without /etc/ld.so.cache (e.g. if libstdc++ is
in a GCC per-version subdirectory handled via /etc/ld.so.conf.d/).

If /etc/ld.so.cache is not available, the runtime linker will look
only at a set of predetermined paths - as seen with LD_DEBUG=libs
added to the bwrap call with "--setenv LD_DEBUG libs":

find library=libstdc++.so.6 [0]; searching
 search cache=/etc/ld.so.cache
 search path=/lib64:/usr/lib64		(system search path)
  trying file=/lib64/libstdc++.so.6
  trying file=/usr/lib64/libstdc++.so.6

followed by:

/usr/bin/totem-video-thumbnailer: error while loading shared libraries: libstdc++.so.6: cannot open shared object file: No such file or directory

If /etc/ld.so.cache is available, it will use that for the paths:

find library=libstdc++.so.6 [0]; searching
 search cache=/etc/ld.so.cache
  trying file=/usr/lib/gcc/x86_64-pc-linux-gnu/8.2.0/libstdc++.so.6

By bind mounting just that file out of /etc, we get it to work on
such a system.

Closes: #81

Iain Lane authored Dec 06, 2018 and Bastien Nocera committed Jan 30, 2019

On systems where /usr-merge hasn't been carried out, /bin (etc) won't
point to /usr/bin. In that case we should --ro-bind the directory
instead of --symlinking it.

This implements the suggestion from Simon McVittie on
https://bugzilla.gnome.org/show_bug.cgi?id=787072.

It also handles source directories not existing, which for example
/lib64 won't on 32-bit systems.

Closes: #4
Closes: #89
(cherry picked from commit 259e7e4e)

Fixes: 775becfb

Fixes: 5e8eed22

g_file_info_get_attribute_object() is transfer none, so when getting a
preview GIcon from a gvfs-backed file that supports it, we need to
reference the preview otherwise we might crash.

==19044== Invalid read of size 8
==19044==    at 0x48607E7: get_preview_thumbnail (gnome-desktop-thumbnail.c:978)
==19044==    by 0x48607E7: gnome_desktop_thumbnail_factory_generate_thumbnail (gnome-desktop-thumbnail.c:1058)
==19044==    by 0x401181: main (test-desktop-thumbnail.c:51)
==19044==  Address 0x700f750 is 0 bytes inside a block of size 40 free'd
==19044==    at 0x4839A0C: free (vg_replace_malloc.c:530)
==19044==    by 0x48DFCD0: g_type_free_instance (gtype.c:1943)
==19044==    by 0x4E7F7B5: _g_file_attribute_value_clear (gfileattribute.c:176)
==19044==    by 0x4E83D46: g_file_info_finalize (gfileinfo.c:327)
==19044==    by 0x48C1C61: g_object_unref (gobject.c:3346)
==19044==    by 0x48607D5: get_preview_thumbnail (gnome-desktop-thumbnail.c:974)
==19044==    by 0x48607D5: gnome_desktop_thumbnail_factory_generate_thumbnail (gnome-desktop-thumbnail.c:1058)
==19044==    by 0x401181: main (test-desktop-thumbnail.c:51)
==19044==  Block was alloc'd at
==19044==    at 0x483880B: malloc (vg_replace_malloc.c:299)
==19044==    by 0x4B54F20: g_malloc (gmem.c:99)
==19044==    by 0x4B6C3C2: g_slice_alloc (gslice.c:1024)
==19044==    by 0x4B6C9F8: g_slice_alloc0 (gslice.c:1050)
==19044==    by 0x48DFA33: g_type_create_instance (gtype.c:1846)
==19044==    by 0x48C2397: g_object_new_internal (gobject.c:1805)
==19044==    by 0x48C4113: g_object_new_valist (gobject.c:2128)
==19044==    by 0x48C443B: g_object_new (gobject.c:1648)
==19044==    by 0x7451CF7: g_vfs_icon_new (gvfsicon.c:178)
==19044==    by 0x7451D47: g_vfs_icon_from_tokens (gvfsicon.c:268)
==19044==    by 0x4E8BA45: g_icon_new_from_tokens (gicon.c:381)
==19044==    by 0x4E8BA45: g_icon_new_for_string (gicon.c:462)
==19044==    by 0x7450C5F: _g_dbus_get_file_attribute (gvfsdaemonprotocol.c:300)
==19044==    by 0x7450D26: _g_dbus_get_file_info (gvfsdaemonprotocol.c:340)
==19044==    by 0x867A74C: g_daemon_file_query_info (gdaemonfile.c:830)
==19044==    by 0x486078D: get_preview_thumbnail (gnome-desktop-thumbnail.c:960)
==19044==    by 0x486078D: gnome_desktop_thumbnail_factory_generate_thumbnail (gnome-desktop-thumbnail.c:1058)
==19044==    by 0x401181: main (test-desktop-thumbnail.c:51)
==19044==
==19044== Invalid read of size 8
==19044==    at 0x48607F0: get_preview_thumbnail (gnome-desktop-thumbnail.c:978)
==19044==    by 0x48607F0: gnome_desktop_thumbnail_factory_generate_thumbnail (gnome-desktop-thumbnail.c:1058)
==19044==    by 0x401181: main (test-desktop-thumbnail.c:51)
==19044==  Address 0xaaaaaaaaaaaaaaaa is not stack'd, malloc'd or (recently) free'd

Root-caused by "Just Me"

Closes: #87

Florian Müllner authored Feb 26, 2018 and Carlos Garnacho committed Apr 17, 2018

https://bugzilla.gnome.org/show_bug.cgi?id=793854

As mutter guards against 0 msec idle watches in
meta_idle_monitor_add_idle_watch()

https://bugzilla.gnome.org/show_bug.cgi?id=793999

It's not used anywhere in the stack now that mutter handles hiding the
cursor when touchscreens are used, and doesn't work under Wayland.

https://bugzilla.gnome.org/show_bug.cgi?id=793737

So return an error.

https://bugzilla.gnome.org/show_bug.cgi?id=793737

I found this new nasal demon spray.

(cherry picked from commit 20b6b46a)

In some places it’s used as an unsigned int, in other cases it’s signed,
so this just casts the -1 in a ternary expression to uint32_t, which
makes no difference.

This also requires some casting when using due to legacy requirements.
The string is never modified, so it’s fine.

https://bugzilla.gnome.org/show_bug.cgi?id=793195

Not much to do other than some clever casting, since Xlib insists on
receiving char pointers.

https://bugzilla.gnome.org/show_bug.cgi?id=793195

This additionally removes the ability to pass deprecation flags to
configure.ac. Revert when GTK/GDK warnings are fixed, or just don’t define
the magic macros.

https://bugzilla.gnome.org/show_bug.cgi?id=793195

https://bugzilla.gnome.org/show_bug.cgi?id=793195

Especially if the types don’t match.

https://bugzilla.gnome.org/show_bug.cgi?id=793195

https://bugzilla.gnome.org/show_bug.cgi?id=793195

The pointers in DefaultInputSource don’t correctly express that they are
const and not just point to const data.

https://bugzilla.gnome.org/show_bug.cgi?id=793195