- when setting datalist to NULL, do it at the end with
g_datalist_unlock_and_set() to avoid the additional atomic
operations.
- when freeing the old "d", do that after releasing the bit
lock.
The previous code was in practice correct.
(((guintptr) ptr) | ((gsize) mask))
works even if gsize is smaller than guintptr. And while the C standard
only guarantees that cast between uintptr_t and void* works, on
reasonable compilers it works to directly cast uintptr_t to GData*.
Still, no need for such uncertainty. Just be clear about the involved
types (use guintptr throughout) and cast first to (gpointer).
g_datalist_clear_i() had only one caller, altough the code comment made
it sound as if in the past there were more.
A function that has only one caller, and then needs a code comment to
explain the context in which it is called, makes the code more complicated
than necessary.
Especially since the function expects to be called with a global lock
held, and then unlocks and relocks. Spreading such things to another
function (which is only used once) makes code harder to follow.
Inline the function, so that we can see all the (trivial) code at one place.
This saves an additional atomic set.
Also, changing the pointer before the unlock can awake a blocked thread,
but at this point, the lock is still held (and the thread may need to
sleep again). This can be avoided.
The existing g_pointer_bit_lock() and g_pointer_bit_unlock() API
requires the user to understand/reimplement how bits of the pointer get
mangled. Add helper functions for that.
The useful thing to do with g_pointer_bit_lock() API is to get/set
pointers while having it locked. For example, to set the pointer a user
can do:
g_pointer_bit_lock (&lockptr, lock_bit);
ptr2 = set_bit_pointer_as_if_locked(ptr, lock_bit);
g_atomic_pointer_set (&lockptr, ptr2);
g_pointer_bit_unlock (&lockptr, lock_bit);
That has several problems:
- it requires one extra atomic operations (3 instead of 2, in the
non-contended case).
- the first g_atomic_pointer_set() already wakes blocked threads,
which find themselves still being locked and needs to go back to
sleep.
- the user needs to re-implement how bit-locking mangles the pointer so
that it looks as if it were locked.
- while the user tries to re-implement what glib does to mangle the
pointer for bitlocking, there is no immediate guarantee that they get
it right.
Now we can do instead:
g_pointer_bit_lock(&lockptr, lock_bit);
g_pointer_bit_unlock_and_set(&lockptr, lock_bit, ptr, 0);
This will also emit a critical if @ptr has the locked bit set.
g_pointer_bit_lock() really only works with pointers that have a certain
alignment, and the lowest bits unset. Otherwise, there is no space to
encode both the locking and all pointer values. The new assertion helps
to catch such bugs.
Also, g_pointer_bit_lock_mask_ptr() is here, so we can do:
g_pointer_bit_lock(&lockptr, lock_bit);
/* set a pointer separately, when g_pointer_bit_unlock_and_set() is unsuitable. */
g_atomic_pointer_set(&lockptr, g_pointer_bit_lock_mask_ptr(ptr, lock_bit, TRUE, 0, NULL));
...
g_pointer_bit_unlock(&lockptr, lock_bit);
and:
g_pointer_bit_lock(&lockptr, lock_bit);
/* read the real pointer after getting the lock. */
ptr = g_pointer_bit_lock_mask_ptr(lockptr, lock_bit, FALSE, 0, NULL));
...
g_pointer_bit_unlock(&lockptr, lock_bit);
While glibc is fine with it (and returns a `NULL` pointer), technically
it’s implementation-defined behaviour according to POSIX, so it’s best
avoided.
See
https://pubs.opengroup.org/onlinepubs/9699919799/functions/posix_memalign.html.
In particular, valgrind will warn about it, which is causing failures of
the gdbus-codegen tests when valgrind is enabled. For example,
https://gitlab.gnome.org/GNOME/glib/-/jobs/3460673 gives
```
==15276== posix_memalign() invalid size value: 0
==15276== at 0x484B7BC: posix_memalign (vg_replace_malloc.c:2099)
==15276== by 0x49320B2: g_variant_new_from_bytes (gvariant-core.c:629)
==15276== by 0x4931853: g_variant_new_from_data (gvariant.c:6226)
==15276== by 0x4B9A951: g_dbus_gvalue_to_gvariant (gdbusutils.c:708)
==15276== by 0x41BD15: _foo_igen_bat_skeleton_handle_get_property (gdbus-test-codegen-generated.c:13503)
==15276== by 0x41BFAF: foo_igen_bat_skeleton_dbus_interface_get_properties (gdbus-test-codegen-generated.c:13582)
…
```
Signed-off-by: Philip Withnall <pwithnall@gnome.org>
Helps: #3228
The "--code" option was removed years ago in d5b8d8d523c3bc26aa9fe6c364d3a17d325b55b6
so remove references to it from README and g-ir-compiler(1)
Remove the "--no-init" option from g-ir-compiler and g-ir-compiler(1)
as it was documented to "can only be used if --code is also specified",
so no reason to keep it around.
Return a non-zero result when opening the output file fails and
don't use g_error() for other failures when writing out the file,
since such errors should not produce a core dump.
We never actually include multiple modules in the compiler,
so just nuke that. Also rather than passing around GIrModule
consistently pass around a GIrTypelibBuild structure which
has various things.
This lets us maintain a stack there which we can walk for
better error messages.
Also, fix up the node lookup in giroffsets.c; previously
it didn't really handle includes correctly. We really need to
switch to always using Foo.Bar (i.e. GIName) names internally...
It can't really work right now because we rely on dumping data at runtime,
which requires the library. If in the future we support static scanning,
we can reinvestigate embedded typelibs.
Take a GError * for typelib loading code, validate the header. This
fixes bizarre errors from gjs where g_irepository_require would happily
load old typelibs.
Add a toplevel GirParser object to hold state that is global across
a compilation. Currently just holds the include path, but will
eventually also keep a cached list of parsed modules.
svn path=/trunk/; revision=901
* gir/Makefile.am: Dep on Makefile
* girepository/ginfo.c: Print out a nicer error
message if we failed to load something.
* girepository/girepository.c: Clean up
default typelib handling; remove global
default_typelib variable. Ensure we handle
NULL repository in more places.
Support dependency resolution.
* tests/Makefile.am: Kill off gobject.gir,
it conflicts with the real one.
* tests/Object.gir: Depend on GObject.
* tools/generate.c: Take --includedir
argument to say which directories to search
for typelibs. Print out dependencies.
svn path=/trunk/; revision=541
* gir/Makefile.am: Use --includedir
* girepository/girparser.c: Recursively parse
includes to pull in aliases and expand them.
We need this to avoid putting unknown names in
the typelibs.
* tools/compiler.c: Add --includedir option.
svn path=/trunk/; revision=512
2008-08-28 Tor Lillqvist <tml@novell.com>
* tools/compiler.c (write_out_typelib): Use binary mode for output
file on Windows.
* girepository/girnode.c: Don't print NULL strings.
* tests/invoke/Makefile.am
* tests/scanner/Makefile.am: Use -no-undefined on Windows to
convince libtool to build shared libraries.
* tests/invoke/invoke.c: Don't needlessly include <dlfcn.h>. Use
g_assert() instead of printing out expected errors.
svn path=/trunk/; revision=509
2008-02-22 Mark Doffman <mark.doffman@codethink.co.uk>
* tools/quote-file.sh
* tools/compiler.c
* tools/generate.c
Move to using the 'C' struct compiler code.
WARNING: This commit does not compile. It is a partial change.
svn path=/trunk/; revision=135
* src/compiler.c, src/generate.c, src/gidlnode.c,
src/gidlparser.c, src/ginfo.c, src/girepository.c,
tests/invoke/invoke.c, test/invoke/testfns.c: Hush compiler
warnings about return values, signedness mismatches, unused
variables, and unhandles enum values in switch statements.
* tests/invoke/Makefile.am: Don't install the invoke test program.
Add -I ../../src to the cflags used for the test functions files.
2005-05-24 Matthias Clasen <mclasen@redhat.com>
* src/compiler.c (main): Use it here to validate
the generated metadata.
* src/gmetadata.[hc]: Add code to validate a
metadata blob.
2005-05-13 Matthias Clasen <mclasen@redhat.com>
* src/compiler.c (format_output): Make the generated
C code compile.
* README: Point to a recent libffi snapshot.
* tests/invoke/*: Some invoke tests.
* src/Makefile: Add ginvoke.c and the necessary
libffi information.
* src/girepository.h (g_function_info_invoke): Add
a GError argument.
* src/ginvoke.c (g_function_info_invoke): Initial
implementation of invoke functionality based on libffi.
