Previously, a C-style cast was used to convert a
Common::Array<Plugin *>, populated with pointers to StaticPlugin
and DynamicPlugin instances, to a
Common::Array<PluginSubclass<T> *>, but PluginSubclass<T> is a
*sibling* class to StaticPlugin/DynamicPlugin, so this cast was
invalid and the results undefined. The methods for retrieving
subclasses of plugins can't be easily changed to just generate an
array of temporary wrapper objects that expose an identical API
which dereferences to the preferred PluginObject subclass because
pointers to these objects are retained by other parts of ScummVM,
so the wrappers would needed to be persisted or they would need to
just re-expose the underlying Plugin object again. This indicated
that a way to solve this problem is to have the callers receive
Plugin objects and get the PluginObject from the Plugin by
explicitly stating their desired type, in a similar manner to
std::get(std::variant), so that the pattern used by this patch to
solve the problem.
Closes gh-1051.
When a 32-bit cursor has the same size as an 8- or 16-bit cursor,
the mouse surfaces were not being regenerated even though the
32-bit cursors have a different memory requirement. This lead to
memory corruption as an inappropriate surface would be used for
the other type of cursor.
The shoe-horned 32-bit cursor support is clearly showing its
scrappy nature here and probably ought to be revisited in the
future if the SurfaceSdl graphics manager sticks around.
Fixes Trac#10349, Trac#10350, Trac#10351.
By default, frac_t is interpreted as signed. The resulting range isn't
large enough to store offsets, so we interpret it as unsigned here
instead. Fixes a crash in QfG1/Mac where instrument->loop_size is 49457.
If a game is doing a screen shake (for example, DOTT when the
stereo is on), and the user does an RTL, the screen shake offset
may get stuck if the engine does not reset it on shutdown. To
avoid this in all cases, just always reset the screen shake
whenever the graphics manager is told to switch to a new graphics
mode.
This situation is triggered normally when _currentScript is 0xFF,
but it could potentially also happen if _currentScript is some
other number >= NUM_SCRIPT_SLOT, so the check is a bit more
conservative than it might appear to need to be.
Compilers optimise these back into shifts on architectures where
shifts of negative integers work the same as mul/div, so this
solves the UB without actually causing any performance issue.
