drm/vmwgfx: use check_add_overflow for shader size+offset bound

In the Linux kernel, the following vulnerability has been resolved: drm/vmwgfx: use check_add_overflow for shader size+offset bound vmw_shader_define() validates the user-supplied shader window against its backing buffer with (u64)buffer->tbo.base.size < (u64)size + (u64)offset drm_vmw_shader_create_arg::offset is __u64 in the uapi; when it is near U64_MAX the unsigned addition wraps and the resulting tiny value passes the check. The unbounded offset is then stored in res->guest_memory_offset and forwarded to host SVGA shader-create commands. Use check_add_overflow() to detect the wrap and compare the resulting endpoint against the buffer size.