If `g_file_monitor_source_dispatch()` drops the last reference to its
`GLocalFileMonitor`, a deadlock will occur, because disposing the
`GLocalFileMonitor` causes synchronous disposal of the
`GFileMonitorSource`, and hence an attempt to re-lock the already-locked
mutex in the `GFileMonitorSource`.
Fix that by dropping the reference to the `GLocalFileMonitor` after
unlocking.
Diagnosed by Ting-Wei Lan. The bug was originally introduced by me in
commit 592a13b483.
Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
gio/glocalfilemonitor.c: In function ‘g_file_monitor_source_new’:
gio/glocalfilemonitor.c:653:3: error: missing initializer for field ‘closure_callback’ of ‘GSourceFuncs’ {aka ‘struct _GSourceFuncs’}
653 | };
| ^
The g_local_file_is_remote function is misleading as it works only for
NFS filesystem types and only for locations in home directorly. Let's
rename it to g_local_file_is_nfs_home to make it obvious.
Previously we were keeping a pointer to the `GFileMonitor` in a
`GFileMonitorSource` instance, but since we weren’t keeping a strong
reference, that `GFileMonitor` instance could be finalised from another
thread at any point while the source was referring to it. Not good.
Use a weak reference, and upgrade it to a strong reference whenever the
`GFileMonitorSource` is referring to the file monitor.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Helps: #1903
`DesktopFileDir` pointers are passed around between threads: they are
initially created on the main thread, but a pointer to them is passed to
the GLib worker thread in the file monitor callback
(`desktop_file_dir_changed()`).
Accordingly, the `DesktopFileDir` objects either have to be
(1) immutable;
(2) reference counted; or
(3) synchronised between the two threads
to avoid one of them being used by one thread after being freed on
another. Option (1) changed with commit 99bc33b6 and is no longer an
option. Option (3) would mean blocking the main thread on the worker
thread, which would be hard to achieve and is against the point of
having a worker thread. So that leaves option (2), which is implemented
here.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Fixes: #1903
Ensure that source is attached to the context before it migth be used
from another thread, since otherwise operation on source are
unsynchronized and not thread-safe.
In particular there was a data race between g_source_attach and
g_source_set_ready_time (used from g_file_monitor_source_handle_event).
GLib currently tries to use FAM volume monitor for monitoring files
within home on NFS. If FAM support is not available, it fallbacks by
default to GInotifyFileMonitor. I think we should fallback to
GPollFileMonitor instead, because inotify is not reliable on NFS,
which may cause issues for dconf. With this patch, it should be safe to
not build libgiofam and still be sure that dconf works properly if home
is mounted on NFS. I think this might be a first step to remove FAM
support from GLib completely, because gamin is buggy and dead for
several years already. Gamin just polls files on NFS anyway. This
change applies on files only, because GPollFileMonitor seems doesn't
support dirs, however it should be enough for dconf. The other
drawback is that one can't set poll timeout currently. Just a note
that this can still be overwritten by GIO_USE_FILE_MONITOR.
For the purposes of debugging, it is quite useful for every GSource to
have a name set. Ensure that any GSource we construct inside GLib has a
name set. For GSources which are then returned to the caller, this name
can then be overridden with something even more useful by the caller.
Since this data is only used for debugging, avoid doing any allocations
for it; just use static strings.
https://gitlab.gnome.org/GNOME/glib/issues/1175
This gets the G_FILE_MONITOR_WATCH_HARD_LINKS flag to the state where it
doesn’t cause crashes, and essentially acts as a no-op. It will not yet
actually monitor for changes made via hard links.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://bugzilla.gnome.org/show_bug.cgi?id=755721
After the big file monitoring rewrite, we only put the IN_MOVED_FROM event
in the queue for such pairs. It matches INOTIFY_DIR_MASK and thus we call
ip_dispatch_event on it, but that function was filtering it out because
the filename in the 'from' event is the one of the temp file, not the
one we are monitoring. That name is in the 'to' event, so compare it as
well, and let the event passin that case.
There is another instance of this check in glocalfilemonitor.c, which is
corrected here as well.
https://bugzilla.gnome.org/show_bug.cgi?id=751358
The changed variable was previously uninitialised in the path where the
rate limit was actually changed. This could result in the
GObject::notify signal not getting emitted.
Spotted by Coverity.
CID: #1296516https://bugzilla.gnome.org/show_bug.cgi?id=748834
Due to a typo, a rename reported via a pair of delete/create events (due
to the watcher not giving the flag for moves to be paired) was
accidentally reported as being created with the old name instead of the
new name.
Fix that.
Return an "interesting" boolean from the event handler function on
GFileMonitorSource.
An event was "interesting" if it will result in a signal actually being
dispatched to the user. It is "uninteresting" if it only hit an
already-dirty rate limiter.
We will use this information to do some backing off in the backends when
faced with a flood of uninteresting events.
Remove all event merging and dispatch logic from GFileMonitor. The only
implementation of GFileMonitor outside of glib is in gvfs and it already
does these things properly.
Get rid of GLocalDirectoryMonitor. We will use a single class,
GLocalFileMonitor, for both directory and file monitoring. This will
prevent every single backend from having to create two objects
separately (eg: ginotifydirectorymonitor.c and ginotifyfilemonitor.c).
Introduce GFileMonitorSource as a thread-safe cross-context dispatch
mechanism. Put it in GLocalFileMonitor. All backends will be expected
to dispatch via the source and not touch the GFileMonitor object at all
from the worker thread.
Remove all construct properties from GLocalFileMonitor and remove the
"context" construct property from GFileMonitor. All backends must now
get the information about what file to monitor from the ->start() call
which is mandatory to implement.
Remove the implementation of rate limiting in GFileMonitor and add an
implementation in GLocalFileMonitor. gvfs never did anything with this
anyway, but if it wanted to, it would have to implement it for itself.
This was done in order to get the rate_limit field into the
GFileMonitorSource so that it could be safely accessed from the worker
thread.
Expose g_local_file_is_remote() internally for NFS detection.
With the "is_remote" functionality exposed, we can now move all
functions for creating local file monitors to a proper location in
glocalfilemonitor.c
Port the inotify backend to adjust to the changes above. None of the
other backends are ported yet. Those will come in future commits.
Add a convenient and race-free method of watching local files from the
GLib worker thread.
Without this, the race-free way to create a monitor that dispatches
events to the worker thread looked something like this:
- dispatch an idle to the worker thread
- from the idle, create the monitor and connect signals
- from the original thread, wait (on a cond?) until the worker thread
has finished setting up the monitor
- read the file that you were monitoring
which is just ridiculously complicated...
To use the new API:
monitor = g_local_file_monitor_new_in_worker ("/path/to/some/file",
G_FILE_MONITOR_NONE,
&error);
g_assert_no_error (error);
g_signal_connect (monitor, "changed", G_CALLBACK (callback), NULL);
g_local_file_monitor_start (monitor);
'callback' will run from the GLib worker thread.
This is the reason that the start() call was introduced in the previous
commit. The backends that don't use the start() call will have a very
thin race between creating the monitor and connecting the signal, but
hopefully they will be fixed soon.
These new APIs will be used (at least) from gdesktopappinfo to watch for
changes in the desktop file directories.
https://bugzilla.gnome.org/show_bug.cgi?id=704887
Stop abusing constructor() to do startup work, adding _start() calls
instead.
The backends themselves still use constructor() although a patch will be
following to also fix inotify.
The reason for using a separate start() call instead of constructed()
will become apparent in future commits.
https://bugzilla.gnome.org/show_bug.cgi?id=704887
Add a pair of new extension points: 'gio-nfs-file-monitor' and
'gio-nfs-directory-monitor'.
Add a check to GLocalFile when creating a file monitor. If the
requested file is in the user's home directory and the user has an NFS
home directory then attempt to use an implementation of one of the new
extension points. If we don't have any implementations then fall back
to the normal "local" monitors.
https://bugzilla.gnome.org/show_bug.cgi?id=592211
Get rid of the complicated default module detection code in
GLocalFileMonitor and GLocalDirectoryMonitor and use the new
_gio_module_get_default_type() function instead.
This change also adds the ability to override the default file monitor
via the GIO_USE_FILE_MONITOR environment variable in the same way as can
be done for GIO_USE_VFS.
https://bugzilla.gnome.org/show_bug.cgi?id=592211
2008-07-01 Cody Russell <bratsche@gnome.org>
* gio/gioenums.h:
* gio/giotypes.h:
Moved all relevant typedefs into these files.
* gio/*.[ch]:
Updated wrt added files.
Split types into separate file for easier maintainership. (#538564)
svn path=/trunk/; revision=7127
2008-06-16 Michael Natterer <mitch@imendio.com>
* *.c: chain up unconditionally in finalize() and dispose(). Also
don't dereference these function pointers when calling them since
that has no meaning at all.
svn path=/trunk/; revision=7048
2008-01-21 Matthias Clasen <mclasen@redhat.com>
* glocal*.c:
* gvolumemanager.c: Whitespace cleanups.
* glocalfileoutputsteam.c (_g_local_file_output_stream_create):
Use the right mode when creating the file.
svn path=/trunk/; revision=6341
2007-12-19 Alexander Larsson <alexl@redhat.com>
* giomodule.c:
Make g_io_modules_load_all_in_directory not unuse
loaded modules so that users of it can do stuff
before unloading.
Init internal "module" types.
Initialize static prio and name for types so that
we don't have to load modules to get it.
* gnativevolumemonitor.h:
* gvolumemonitor.h:
Move is_supported to parent class so that
non-native monitors can avoid being initialized
too. (For instance GDaemonVolumeMonitor if we're
not using GDaemonVfs.)
* glocaldirectorymonitor.[ch]:
* glocalfilemonitor.[ch]:
* gunionvolumemonitor.c:
* gunixvolumemonitor.c:
* gvfs.c:
Find plugins using the static prio+name to
avoid unnecessarily loading the modules.
svn path=/trunk/; revision=6159
2007-12-17 Alexander Larsson <alexl@redhat.com>
* gnativevolumemonitor.h:
* gunionvolumemonitor.c:
* gunixvolumemonitor.c:
Add is_supported() to GNativeVolumeMonitorClass so
that we can avoid having to create an object to see
if the backend is supported at runtime.
Also add name member and an env var to pick a specific
volume monitor backend.
* gmountprivate.h:
* glocalfile.c:
Add cancellable to _g_mount_get_for_mount_path()
* glocaldirectorymonitor.c:
* glocalfilemonitor.c:
Avoid loading and unloading modules while sorting.
svn path=/trunk/; revision=6144
2007-12-10 15:08:59 Tim Janik <timj@imendio.com>
* let g_warn_if_fail replace g_assert as discussed here:
http://mail.gnome.org/archives/gtk-devel-list/2007-October/msg00089.html
* fix bug #502498: Test framework assertion failures should follow
gcc error format.
* gmessages.h, gmessages.c: deprecated g_assert_warning() which is
unused now. removed g_assert*() definitions whcih are provided by
gtestutils.h now. added g_warn_if_reached() and g_warn_if_fail()
which are recommended as g_assert/g_assert_not_reached replacements
for non-test programs.
added g_warn_message() to implement g_warn_*() macros.
use emacs-next-error friendly formatting for file:line: for warnings.
* gtestutils.h, gtestutils.c: use emacs-next-error friendly formatting.
implement g_assert_not_reached() with g_assertion_message() and
g_assert() in terms of g_assertion_message_expr() so we'll be able to
provide assertion messages in test logs.
* gkeyfile.c, gbookmarkfile.c: changed g_assert*() to g_warn_if_fail()
or g_return_if_fail() where suitable.
* gio/: changed g_assert to g_warn_if_fail.
svn path=/trunk/; revision=6086
2007-12-05 Alexander Larsson <alexl@redhat.com>
* gdatainputstream.c:
Fix warnings
* gio.symbols:
* giomodule.[ch]
* glocaldirectorymonitor.c:
* glocalfilemonitor.c:
* gunionvolumemonitor.c:
* gvfs.c:
Make g_io_modules_ensure_loaded a private function and
don't pass in the dirname. This means we can do magic
directory finding in the win32 version.
Export the actual load-modules-in-directory code so that
gvfs can reuse that.
svn path=/trunk/; revision=6050
2007-11-26 Alexander Larsson <alexl@redhat.com>
* Makefile.am:
* configure.in:
* gio-2.0-uninstalled.pc.in:
* gio-2.0.pc.in:
* gio-unix-2.0-uninstalled.pc.in:
* gio-unix-2.0.pc.in:
* gio/
* docs/reference/gio
Merged gio-standalone into glib.
* glib/glibintl.h:
* glib/gutils.c:
Export glib_gettext so that gio can use it
Add P_ (using same domain for now)
Add I_ as g_intern_static_string
svn path=/trunk/; revision=5941
