When presented with an array of empty arrays of 8-byte-aligned types,
GDBus would incorrectly apply the 8-byte alignment when reading back.
https://bugzilla.gnome.org/show_bug.cgi?id=673612
Signed-off-by: David Zeuthen <davidz@redhat.com>
D-Bus arrays are serialized as follows:
1. align to a 4-byte boundary (for the length)
2. uint32: the length of the serialized body in bytes
3. padding for the alignment of the body type (not included in the length)
4. the body.
Note that 3. is a no-op unless the body type is an 8-byte aligned type
(uint64, int64, double, struct, dict_entry), since you are always on a
4-byte boundary from aligning and writing the length.
So, an empty aax (that is, an array containing zero arrays of int64)
is serialized as follows:
1. align to a 4-byte boundary
2. length of the contents of this (empty) array, in bytes (0)
3. align to a 4-byte boundary (the child array's alignment requirement)
4. there is no body.
But previously, GDBus would recurse in step three to align not just for
the type of the child array, but for the nonexistent child array's
contents. This only affects the algorithm when the grandchild type has
8-byte alignment and the reader happened to not already be on an 8-byte
boundary, in which case 4 bytes were spuriously skipped.
https://bugzilla.gnome.org/show_bug.cgi?id=673612
Signed-off-by: David Zeuthen <davidz@redhat.com>
Transforming a GValue holding flags from a GFlagsValue set that includes the 0
value (no flag bits set) into a string would loop until exhausting all the
available memory.
https://bugzilla.gnome.org/show_bug.cgi?id=670557
When G_DISABLE_ASSERT is not defined, g_hash_table_foreach and
g_hash_table_find dereferences the hash table argument before
checking if it's NULL. This causes a crash when one of this function
is mistakenly called with a NULL argument instead of returning
with a warning through g_return_if_fail.
When set to represent a zoneinfo file, TZ may start with :, therefore
glib needs to check it and ignore the first char when building the
resulting filename, or it won't be found.
Also, the path could be absolute, in which case it is wrong to
append /usr/share/timezone
https://bugzilla.gnome.org/show_bug.cgi?id=664237
In non-UTF-8 locales, the translations and nl_langinfo() return values
must be converted to UTF-8 before being returned to the caller.
Likewise, when making a recursive call to expand a format like '%x',
the format string must first be converted to UTF-8.
https://bugzilla.gnome.org/show_bug.cgi?id=668250
When cross-compiling with linaro toolchain 4.6.2, I got several
linking errors about missing symbols because a couple libraries are
not linked.
This patch adds to the LDADD flags the missed libraries.
This patch applies above the glib-2-30 branch.
Signed-off-by: Víctor Manuel Jáquez Leal <vjaquez@igalia.com>
https://bugzilla.gnome.org/show_bug.cgi?id=671711
gdb by default will only print strings up to 200 characters. After that
it abbreviates them. This affects the run-assert-msg-test.sh script if
the path to the glib installation is too long (in our case it was 133
characters, 132 would still have worked...)
By having gdb execute "set print elements 0" before printing the assert
string, the limit on maximum number of characters to print is set to
unlimited.
Signed-off-by: Peter Kjellerstedt <pkj@axis.com>
https://bugzilla.gnome.org/show_bug.cgi?id=670218
foo_free is conceptually "worth" one unref; not decrementing the
refcount here means the GArray or GPtrArray wrapper (but not its
contents) would leak in the following call sequence:
p = g_ptr_array_new ();
g_ptr_array_ref (p);
g_ptr_array_free (p, TRUE);
g_ptr_array_unref (p);
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Depending how the array is freed, we may want to free the underlying
array (the "segment"), the struct wrapper or both.
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Removing the last thing in a dataset frees the dataset, and if the
datalist was in a dataset, we can't safely unlock it after the dataset
has been freed. Unlock it sooner.
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Matthias Clasen <mclasen@redhat.com>
Calling this function with a NULL argument is considered to be invalid,
but one of the regression tests does it anyway (to watch it crash), which
seems a good indication that it's expected to be somewhat common.
Let's check it rather than segfaulting.
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Reviewed-by: Matthias Clasen <mclasen@redhat.com>
These were leaked. Valgrind was sad.
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Also document why we're not actually using the buffer for anything.
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
In practice, the uninitialized refcount will typically mean that the copy is
never freed, and leaks.
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=666113
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Previously, this would fail the assertion
"connection->initialization_error != NULL" after the label "out".
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=665067
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: David Zeuthen <davidz@redhat.com>
This was a regression in commits f41178c6c and d0d04c09: flush_async_data
wasn't necessarily NULL in the "don't flush" case.
Also move initialization of these variables up so that it's
unconditional, since that's easier to verify than checking
that each branch gets it right.
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=664617
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
These might even make useful public API if they grew a Windows
implementation, but for now they can be Unix-only test API.
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=662395
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Cosimo Alfarano <cosimo.alfarano@collabora.co.uk>
We didn't previously flush in a couple of cases where we should have
done:
* a write is running when flush is called: we should flush after it
finishes
* writes have been made since the last flush, but none are pending or
running right now: we should flush the underlying transport straight
away
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=662395
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Cosimo Alfarano <cosimo.alfarano@collabora.co.uk>
This makes it easier to schedule a flush, by putting it on the same code
path as writing and closing.
Also change message_written to expect the lock to be held, since all
that's left in that function either wants to hold the lock or doesn't
care, and it's silly to release the lock immediately before calling
message_written, which just takes it again.
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=662395
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Cosimo Alfarano <cosimo.alfarano@collabora.co.uk>
When we use this function to schedule a flush, it'll be called
with the lock held. Releasing and immediately re-taking the lock would
be pointless.
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=662395
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Cosimo Alfarano <cosimo.alfarano@collabora.co.uk>
maybe_write_next_message now also closes, and I'm about to make it
consider whether to flush as well, so its name is increasingly
inappropriate. Similarly, write_message_in_idle_cb is a wrapper around
it which could do any of those things.
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=662395
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Cosimo Alfarano <cosimo.alfarano@collabora.co.uk>
If the user calls flush_sync() with no messages in the queue, but an
async write call pending, then we ought to flush after that async write
returns (although we don't currently do that). If it was an async close
or flush that was pending, there's no need to flush (again) afterwards.
So, we need to distinguish.
Bug: https://bugzilla.gnome.org/show_bug.cgi?id=662395
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Cosimo Alfarano <cosimo.alfarano@collabora.co.uk>
The 1 bit mutex tests asserts: ((gsize) ptrs) % 8, ==, 0), which fails
when the platform only aligns porters to 32 bits (e.g. S390 and
powerpc).
I'm not sure why this assertion was placed here, but given
that internally g_pointer_bit_trylock uses g_atomic_int_or internally
change the assertion so it only requires the alignment to be a multiple
of sizeof(int)
