Also explicitly specify `-mstrict-align` (The current default `-mcpu`
on gcc doesn't support unaligned accesses, so aligned memory accesses
are already implicitly required).
The `-Wcast-align` warning seems to oversensitive as it flags code like
this: https://godbolt.org/z/c8481o8aa
This used to be added for aarch64 in a473cfd71b, but was later removed
in 11896868d6.
termcap has last been updated in 2002, and is relying on a lot of
deprecated C99 behavior. All two current users are fine without it,
so just remove it completely.
Also trap the invalid object index inserted for transferred values in
deserialization. In the future we should avoid inserting that
placeholder value in the data stream at all.
This fixes the issue that occurred when, after clicking an inline
paintable page would always scroll to the top. The problem was that
`scroll_an_element_into_view()` relies on `get_bounding_client_rect()`
to produce the correct scroll position and for inline paintables we
were always returning zero rect before this change.
The construct `adopt_ref(*new Obj(TRY(get_resource())))` is another
manifestation of a classic anti-pattern. In old C++, you would leak the
object's memory if the argument threw an exception, or if a member
initializer threw an exception. In our case, we leak if the MappedFile
returns an Error. This is pretty concerning, and we should avoid this
pattern at all costs, and try to use the "safer" helpers whenever
possible.
This change fixes GC-leak caused by following mutual dependency:
- SVGDecodedImageData owns JS::Handle for Page.
- SVGDecodedImageData is owned by visited objects.
by making everything inherited from HTML::DecodedImageData and
ListOfAvailableImages to be GC-allocated.
Generally, if visited object has a handle, very likely we leak
everything visited from object in a handle.
This requires duplicating some logic from Core::Process::get_name()
into AK, which seems unfortunate. But for now, this greatly improves the
log messages for testing Ladybird on Linux.
The feature is hidden behind a runtime flag with a global setter in the
same way that totally enabling/disabling dbgln is.
It's possible for a malloc inside load_program_headers() to steal the
reserved memory space we created for the program headers. Remove the
reservation later in the method.
The `[[gnu::packed]]` attribute apparently lowered the required
alignment of the structs, which caused the compiler to generate two
1 byte loads/stores on RISC-V. This caused the kernel to read/write
incorrect values, as the device only seems to accept 2 byte operations.
All the keys in a property table are guaranteed to be marked via
Shape::m_property_key in each step of the transition chain that leads
up to the Shape.
We previously had a concept of unique shapes, which meant that they
couldn't be shared between multiple objects.
Object shapes became unique in three situations:
- They were the shape of the global object.
- They had more than 100 properties added to them.
- They had one or more properties deleted from them.
Unfortunately, unique shapes presented an annoying problem for inline
caches, and we added a "unique shape serial number" for being able to
tell that a unique shape had been mutated.
This patch gets rid of the concept of unique shapes, simplifying all
the caching code, since inline caches can now simply perform a shape
check and then we're good.
To make this possible, we now have the concept of delete transitions,
which occur when a property is deleted from a shape.
Note that this patch by itself introduces a performance regression in
some situtations, since we now create a lot more shapes, and marking
their property keys can be very heavy. This will be addressed in a
subsequent patch.
These should really be weakly held by the Shape, but we don't have a
mechanism for weak hashmap keys at the moment, so let's just mark
these for now so they don't go stale.
