Fixes Sim City 3000 Building Architect.
--
v5: ddraw/tests: Add tests for multiple devices.
ddraw: Support multiple devices per ddraw object.
ddraw: Sync wined3d render target in d3d_device_sync_rendertarget().
ddraw: Store wined3d state in d3d_device.
wined3d: Factor out wined3d_texture_set_lod() function.
ddraw: Don't apply state in ddraw_surface_blt().
ddraw: Store matrix handle in the global table.
ddraw: Store surface handles in the global table.
ddraw: Store material handles in the global table.
https://gitlab.winehq.org/wine/wine/-/merge_requests/5329
Do we always reallocate to larger size? It is currently reallocated based on the actual message size returned by the server, while buffer_size in peek_message may be bigger (and it is first allocated to 1024 bytes). Granted, I only reproduced the actual memory corruption with Proton, but it seems to me that in this regard the handling is the same.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5392#note_66132
We only reallocate the buffer to larger size, so I don't see how we can write past the allocated memory. Still, we potentially miss some opportunities to reuse the buffer, so better tracking of the size seems good to me, but the commit message could be adjusted.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5392#note_66130
Jacek Caban (@jacek) commented about dlls/win32u/message.c:
> }
>
> /* make sure that there is space for 'size' bytes in buffer, growing it if needed */
> -static inline void *get_buffer_space( void **buffer, size_t size, size_t prev_size )
> +static inline void *get_buffer_space( void **buffer, size_t size, size_t *prev_size )
This is no longer "prev", `buffer_size` would be a better name after this change.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5392#note_66129
Jacek Caban (@jacek) commented about dlls/win32u/message.c:
> {
> if (*prev_size < size)
> {
> - *buffer = malloc( size );
> + *buffer = realloc( *buffer, size );
This leaks on error, we could leave the buffer unchanged in that case.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5392#note_66128
Creating a new process is inefficient and also visible to applications. You don't need to duplicate the code, it can be factored out in a couple of handlers used in both command line and interactive mode. In interactive mode you'd run a loop that accepts commands and executes them, until you hit 'quit' or 'exit'.
wmic not handling spaces in the property list (unless quoted) is a bug, but you should be able to reuse process_property_list().
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5402#note_66122
Udev monitor monitors the whole input subsystem, but not all devices
in the input subsystem have devnodes associated to them.
This MR makes the event processing ignore such devices.
All device handling assumes devices have devnodes, so here we just
simply ignore all devices which do not have one. They are irrelevant.
Previously, udev bus thread aborted when an event for a device without
a devnode was processed:
```
10111.330:0068:0084:trace:hid:process_monitor_event Received action "remove" for udev device (null)
10111.330:0068:0084:warn:hid:bus_main_thread L"UDEV" bus wait returned status 0xc0000005
```
Just plugging in and out a normal mouse was enough cause this.
This was because root input devices (which do not have devnodes) were
handled too and `find_device_from_devnode()` choked on NULL argument.
--
v5: winebus: ignore udev events of devices which do not have devnodes
https://gitlab.winehq.org/wine/wine/-/merge_requests/5385