Rework the pixel computation to use areas

This changes the displayio pixel computation from per-pixel to
per-area. This is precursor work to updating portions of the screen
(#1169). It should provide mild speedups because bounds checks are
done once per area rather than once per pixel. Filling by area also
allows TileGrid to maintain a row-associative fill pattern even when
the display's refresh is orthogonal to it.

ports/atmel-samd/Makefile

@@ -99,7 +99,6 @@ endif
 
 #Debugging/Optimization
 ifeq ($(DEBUG), 1)
-  # Turn on Python modules useful for debugging (e.g. uheap, ustack).
   CFLAGS += -ggdb
   # You may want to disable -flto if it interferes with debugging.
   CFLAGS += -flto

shared-bindings/displayio/Display.h

@@ -53,6 +53,7 @@ void common_hal_displayio_display_refresh_soon(displayio_display_obj_t* self);
 bool displayio_display_begin_transaction(displayio_display_obj_t* self);
 void displayio_display_end_transaction(displayio_display_obj_t* self);
 
+// The second point of the region is exclusive.
 void displayio_display_set_region_to_update(displayio_display_obj_t* self, uint16_t x0, uint16_t y0, uint16_t x1, uint16_t y1);
 bool displayio_display_frame_queued(displayio_display_obj_t* self);
 

shared-module/displayio/Group.c

@@ -134,32 +134,28 @@ void displayio_group_construct(displayio_group_t* self, displayio_group_child_t*
     self->scale = scale;
 }
 
-bool displayio_group_get_pixel(displayio_group_t *self, int16_t x, int16_t y, uint16_t* pixel) {
-    x -= self->x;
-    y -= self->y;
-    // When we are scaled we need to substract all but one to ensure -scale to 0 divide down to -1.
-    // Normally -scale to scale both divide down to 0 because 0 is unsigned.
-    if (x < 0) {
-        x -= self->scale - 1;
-    }
-    if (y < 0) {
-        y -= self->scale - 1;
-    }
-    x /= self->scale;
-    y /= self->scale;
+bool displayio_group_get_area(displayio_group_t *self, displayio_buffer_transform_t* transform, displayio_area_t* area, uint32_t* mask, uint32_t* buffer) {
+    displayio_area_shift(area, -self->x * transform->scale, -self->y * transform->scale);
+    transform->scale *= self->scale;
+
+    bool full_coverage = false;
     for (int32_t i = self->size - 1; i >= 0 ; i--) {
         mp_obj_t layer = self->children[i].native;
         if (MP_OBJ_IS_TYPE(layer, &displayio_tilegrid_type)) {
-            if (displayio_tilegrid_get_pixel(layer, x, y, pixel)) {
-                return true;
+            if (displayio_tilegrid_get_area(layer, transform, area, mask, buffer)) {
+                full_coverage = true;
+                break;
             }
         } else if (MP_OBJ_IS_TYPE(layer, &displayio_group_type)) {
-            if (displayio_group_get_pixel(layer, x, y, pixel)) {
-                return true;
+            if (displayio_group_get_area(layer, transform, area, mask, buffer)) {
+                full_coverage = true;
+                break;
             }
         }
     }
-    return false;
+    transform->scale /= self->scale;
+    displayio_area_shift(area, self->x * transform->scale, self->y * transform->scale);
+    return full_coverage;
 }
 
 bool displayio_group_needs_refresh(displayio_group_t *self) {

shared-module/displayio/Group.h

@@ -31,6 +31,7 @@
 #include <stdint.h>
 
 #include "py/obj.h"
+#include "shared-module/displayio/area.h"
 
 typedef struct {
     mp_obj_t native;
@@ -49,7 +50,7 @@ typedef struct {
 } displayio_group_t;
 
 void displayio_group_construct(displayio_group_t* self, displayio_group_child_t* child_array, uint32_t max_size, uint32_t scale, mp_int_t x, mp_int_t y);
-bool displayio_group_get_pixel(displayio_group_t *group, int16_t x, int16_t y, uint16_t *pixel);
+bool displayio_group_get_area(displayio_group_t *group, displayio_buffer_transform_t* transform, displayio_area_t* area, uint32_t* mask, uint32_t *buffer);
 bool displayio_group_needs_refresh(displayio_group_t *self);
 void displayio_group_finish_refresh(displayio_group_t *self);
 

shared-module/displayio/TileGrid.c

@@ -56,31 +56,40 @@ void common_hal_displayio_tilegrid_construct(displayio_tilegrid_t *self, mp_obj_
     self->bitmap_width_in_tiles = bitmap_width_in_tiles;
     self->width_in_tiles = width;
     self->height_in_tiles = height;
-    self->total_width = width * tile_width;
-    self->total_height = height * tile_height;
+    self->area.x1 = x;
+    self->area.y1 = y;
+    // -1 because areas are inclusive
+    self->area.x2 = x + width * tile_width - 1;
+    self->area.y2 = y + height * tile_height - 1;
     self->tile_width = tile_width;
     self->tile_height = tile_height;
     self->bitmap = bitmap;
     self->pixel_shader = pixel_shader;
-    self->x = x;
-    self->y = y;
 }
 
 
 mp_int_t common_hal_displayio_tilegrid_get_x(displayio_tilegrid_t *self) {
-    return self->x;
+    return self->area.x1;
 }
 void common_hal_displayio_tilegrid_set_x(displayio_tilegrid_t *self, mp_int_t x) {
-    self->needs_refresh = self->x != x;
-    self->x = x;
+    if (self->area.x1 == x) {
+        return;
+    }
+    self->needs_refresh = true;
+    self->area.x2 += (self->area.x1 - x);
+    self->area.x1 = x;
 }
 mp_int_t common_hal_displayio_tilegrid_get_y(displayio_tilegrid_t *self) {
-    return self->y;
+    return self->area.y1;
 }
 
 void common_hal_displayio_tilegrid_set_y(displayio_tilegrid_t *self, mp_int_t y) {
-    self->needs_refresh = self->y != y;
-    self->y = y;
+    if (self->area.y1 == y) {
+        return;
+    }
+    self->needs_refresh = true;
+    self->area.y2 += (self->area.y1 - y);
+    self->area.y1 = y;
 }
 
 mp_obj_t common_hal_displayio_tilegrid_get_pixel_shader(displayio_tilegrid_t *self) {
@@ -128,14 +137,11 @@ void common_hal_displayio_tilegrid_set_tile(displayio_tilegrid_t *self, uint16_t
 void common_hal_displayio_tilegrid_set_top_left(displayio_tilegrid_t *self, uint16_t x, uint16_t y) {
     self->top_left_x = x;
     self->top_left_y = y;
+    self->needs_refresh = true;
 }
-bool displayio_tilegrid_get_pixel(displayio_tilegrid_t *self, int16_t x, int16_t y, uint16_t* pixel) {
-    x -= self->x;
-    y -= self->y;
-    if (y < 0 || y >= self->total_height || x >= self->total_width || x < 0) {
-        return false;
-    }
 
+bool displayio_tilegrid_get_area(displayio_tilegrid_t *self, displayio_buffer_transform_t* transform, displayio_area_t* area, uint32_t* mask, uint32_t *buffer) {
+    // If no tiles are present we have no impact.
     uint8_t* tiles = self->tiles;
     if (self->inline_tiles) {
         tiles = (uint8_t*) &self->tiles;
@@ -143,30 +149,103 @@ bool displayio_tilegrid_get_pixel(displayio_tilegrid_t *self, int16_t x, int16_t
     if (tiles == NULL) {
         return false;
     }
-    uint16_t tile_location = ((y / self->tile_height + self->top_left_y) % self->height_in_tiles) * self->width_in_tiles + (x / self->tile_width + self->top_left_x) % self->width_in_tiles;
-    uint8_t tile = tiles[tile_location];
-    uint16_t tile_x = tile_x = (tile % self->bitmap_width_in_tiles) * self->tile_width + x % self->tile_width;
-    uint16_t tile_y = tile_y = (tile / self->bitmap_width_in_tiles) * self->tile_height + y % self->tile_height;
 
-    uint32_t value = 0;
-    if (MP_OBJ_IS_TYPE(self->bitmap, &displayio_bitmap_type)) {
-        value = common_hal_displayio_bitmap_get_pixel(self->bitmap, tile_x, tile_y);
-    } else if (MP_OBJ_IS_TYPE(self->bitmap, &displayio_shape_type)) {
-        value = common_hal_displayio_shape_get_pixel(self->bitmap, tile_x, tile_y);
-    } else if (MP_OBJ_IS_TYPE(self->bitmap, &displayio_ondiskbitmap_type)) {
-        value = common_hal_displayio_ondiskbitmap_get_pixel(self->bitmap, tile_x, tile_y);
+    displayio_area_t overlap;
+    displayio_area_t scaled_area = {
+        .x1 = self->area.x1 * transform->scale,
+        .y1 = self->area.y1 * transform->scale,
+        .x2 = (self->area.x2 + 1) * transform->scale - 1, // Second point is inclusive.
+        .y2 = (self->area.y2 + 1) * transform->scale - 1
+    };
+    if (!displayio_area_compute_overlap(area, &scaled_area, &overlap)) {
+        return false;
     }
 
-    if (self->pixel_shader == mp_const_none) {
-        *pixel = value;
-        return true;
-    } else if (MP_OBJ_IS_TYPE(self->pixel_shader, &displayio_palette_type) && displayio_palette_get_color(self->pixel_shader, value, pixel)) {
-        return true;
-    } else if (MP_OBJ_IS_TYPE(self->pixel_shader, &displayio_colorconverter_type) && common_hal_displayio_colorconverter_convert(self->pixel_shader, value, pixel)) {
-        return true;
+    int16_t x_stride = 1;
+    int16_t y_stride = displayio_area_width(area);
+    if (transform->transpose_xy) {
+        x_stride = displayio_area_height(area);
+        y_stride = 1;
+    }
+    uint16_t start = 0;
+    if (transform->mirror_x) {
+        start += (area->x2 - area->x1) * x_stride;
+        x_stride *= -1;
+    }
+    if (transform->mirror_y) {
+        start += (area->y2 - area->y1) * y_stride;
+        y_stride *= -1;
     }
 
-    return false;
+    bool full_coverage = displayio_area_equal(area, &overlap);
+
+    // TODO(tannewt): Set full coverage to true if all pixels outside the overlap have already been
+    // set as well.
+    bool always_full_coverage = false;
+
+    // TODO(tannewt): Check to see if the pixel_shader has any transparency. If it doesn't then we
+    // can either return full coverage or bulk update the mask.
+    int16_t y = overlap.y1 - scaled_area.y1;
+    if (y < 0) {
+        y = 0;
+    }
+    int16_t x_shift = area->x1 - scaled_area.x1;
+    int16_t y_shift = area->y1 - scaled_area.y1;
+    for (; y <= overlap.y2 - scaled_area.y1; y++) {
+        int16_t x = overlap.x1 - scaled_area.x1;
+        if (x < 0) {
+            x = 0;
+        }
+        int16_t row_start = start + (y - y_shift) * y_stride;
+        int16_t local_y = y / transform->scale;
+        for (; x <= overlap.x2 - scaled_area.x1; x++) {
+            // Compute the destination pixel in the buffer and mask based on the transformations.
+            uint16_t offset = row_start + (x - x_shift) * x_stride;
+
+            // Check the mask first to see if the pixel has already been set.
+            if ((mask[offset / 32] & (1 << (offset % 32))) != 0) {
+                continue;
+            }
+            int16_t local_x = x / transform->scale;
+            uint16_t tile_location = ((local_y / self->tile_height + self->top_left_y) % self->height_in_tiles) * self->width_in_tiles + (local_x / self->tile_width + self->top_left_x) % self->width_in_tiles;
+            uint8_t tile = tiles[tile_location];
+            uint16_t tile_x = (tile % self->bitmap_width_in_tiles) * self->tile_width + local_x % self->tile_width;
+            uint16_t tile_y = (tile / self->bitmap_width_in_tiles) * self->tile_height + local_y % self->tile_height;
+
+            uint32_t value = 0;
+            // We always want to read bitmap pixels by row first and then transpose into the destination
+            // buffer because most bitmaps are row associated.
+            if (MP_OBJ_IS_TYPE(self->bitmap, &displayio_bitmap_type)) {
+                value = common_hal_displayio_bitmap_get_pixel(self->bitmap, tile_x, tile_y);
+            } else if (MP_OBJ_IS_TYPE(self->bitmap, &displayio_shape_type)) {
+                value = common_hal_displayio_shape_get_pixel(self->bitmap, tile_x, tile_y);
+            } else if (MP_OBJ_IS_TYPE(self->bitmap, &displayio_ondiskbitmap_type)) {
+                value = common_hal_displayio_ondiskbitmap_get_pixel(self->bitmap, tile_x, tile_y);
+            }
+
+            uint16_t* pixel = ((uint16_t*) buffer) + offset;
+            if (self->pixel_shader == mp_const_none) {
+                *pixel = value;
+                return true;
+            } else if (MP_OBJ_IS_TYPE(self->pixel_shader, &displayio_palette_type)) {
+                if (!displayio_palette_get_color(self->pixel_shader, value, pixel)) {
+                    // mark the pixel as transparent
+                    full_coverage = false;
+                } else if (!always_full_coverage) {
+                    mask[offset / 32] |= 1 << (offset % 32);
+                }
+            } else if (MP_OBJ_IS_TYPE(self->pixel_shader, &displayio_colorconverter_type)) {
+                if (!common_hal_displayio_colorconverter_convert(self->pixel_shader, value, pixel)) {
+                    // mark the pixel as transparent
+                    full_coverage = false;
+                } else if (!always_full_coverage) {
+                    mask[offset / 32] |= 1 << (offset % 32);
+                }
+            }
+        }
+    }
+
+    return full_coverage;
 }
 
 bool displayio_tilegrid_needs_refresh(displayio_tilegrid_t *self) {

shared-module/displayio/TileGrid.h

@@ -31,18 +31,16 @@
 #include <stdint.h>
 
 #include "py/obj.h"
+#include "shared-module/displayio/area.h"
 
 typedef struct {
     mp_obj_base_t base;
     mp_obj_t bitmap;
     mp_obj_t pixel_shader;
-    uint16_t x;
-    uint16_t y;
+    displayio_area_t area;
     uint16_t bitmap_width_in_tiles;
     uint16_t width_in_tiles;
     uint16_t height_in_tiles;
-    uint16_t total_width;
-    uint16_t total_height;
     uint16_t tile_width;
     uint16_t tile_height;
     uint16_t top_left_x;
@@ -52,7 +50,7 @@ typedef struct {
     bool inline_tiles;
 } displayio_tilegrid_t;
 
-bool displayio_tilegrid_get_pixel(displayio_tilegrid_t *self, int16_t x, int16_t y, uint16_t *pixel);
+bool displayio_tilegrid_get_area(displayio_tilegrid_t *self, displayio_buffer_transform_t* transform, displayio_area_t* area, uint32_t* mask, uint32_t *buffer);
 bool displayio_tilegrid_needs_refresh(displayio_tilegrid_t *self);
 void displayio_tilegrid_finish_refresh(displayio_tilegrid_t *self);
 

shared-module/displayio/__init__.c

@@ -10,6 +10,7 @@
 #include "shared-bindings/displayio/Display.h"
 #include "shared-bindings/displayio/Group.h"
 #include "shared-bindings/displayio/Palette.h"
+#include "shared-module/displayio/area.h"
 #include "supervisor/shared/autoreload.h"
 #include "supervisor/shared/display.h"
 #include "supervisor/memory.h"
@@ -17,8 +18,8 @@
 
 primary_display_t displays[CIRCUITPY_DISPLAY_LIMIT];
 
-static inline void swap(uint16_t* a, uint16_t* b) {
-    uint16_t temp = *a;
+static inline void swap(int16_t* a, int16_t* b) {
+    int16_t temp = *a;
     *a = *b;
     *b = temp;
 }
@@ -56,102 +57,112 @@ void displayio_refresh_displays(void) {
             continue;
         }
         if (displayio_display_refresh_queued(display)) {
-            // We compute the pixels. r and c are row and column to match the display memory
-            // structure. x and y match location within the groups.
-            uint16_t c0 = 0;
-            uint16_t r0 = 0;
-            uint16_t c1 = display->width;
-            uint16_t r1 = display->height;
-            if (display->transpose_xy) {
-                swap(&c1, &r1);
-            }
-
             if (!displayio_display_begin_transaction(display)) {
                 // Can't acquire display bus; skip updating this display. Try next display.
                 continue;
             }
-            displayio_display_set_region_to_update(display, c0, r0, c1, r1);
             displayio_display_end_transaction(display);
 
-            uint16_t x0 = 0;
-            uint16_t x1 = display->width - 1;
-            uint16_t startx = 0;
-            int8_t dx = 1;
-            if (display->mirror_x) {
-                dx = -1;
-                startx = x1;
-            }
-            uint16_t y0 = 0;
-            uint16_t y1 = display->height - 1;
-            uint16_t starty = 0;
-            int8_t dy = 1;
-            if (display->mirror_y) {
-                dy = -1;
-                starty = y1;
-            }
-
-            bool transpose = false;
+            displayio_area_t whole_screen = {
+                .x1 = 0,
+                .y1 = 0,
+                .x2 = display->width - 1,
+                .y2 = display->height - 1
+            };
             if (display->transpose_xy) {
-                transpose = true;
-                int8_t temp_dx = dx;
-                dx = dy;
-                dy = temp_dx;
-
-                swap(&starty, &startx);
-                swap(&x0, &y0);
-                swap(&x1, &y1);
+                swap(&whole_screen.x2, &whole_screen.y2);
             }
 
-            size_t index = 0;
-            uint16_t buffer_size = 256;
+            uint16_t buffer_size = 512;
+
+            uint16_t subrectangles = 1;
+            uint16_t rows_per_buffer = displayio_area_height(&whole_screen);
+            if (displayio_area_size(&whole_screen) > buffer_size) {
+                rows_per_buffer = buffer_size / displayio_area_width(&whole_screen);
+                subrectangles = displayio_area_height(&whole_screen) / rows_per_buffer;
+                buffer_size = rows_per_buffer * displayio_area_width(&whole_screen);
+            }
             uint32_t buffer[buffer_size / 2];
-            bool skip_this_display = false;
-
-            for (uint16_t y = starty; y0 <= y && y <= y1; y += dy) {
-                for (uint16_t x = startx; x0 <= x && x <= x1; x += dx) {
-                    uint16_t* pixel = &(((uint16_t*)buffer)[index]);
-                    *pixel = 0;
-
-                    if (display->current_group != NULL) {
-                        if (transpose) {
-                            displayio_group_get_pixel(display->current_group, y, x, pixel);
-                        } else {
-                            displayio_group_get_pixel(display->current_group, x, y, pixel);
-                        }
-                    }
 
-                    index += 1;
-                    // The buffer is full, send it.
-                    if (index >= buffer_size) {
-                        if (!displayio_display_begin_transaction(display)) {
-                            // Can't acquire display bus; skip the rest of the data. Try next display.
-                            index = 0;
-                            skip_this_display = true;
-                            break;
+            for (uint16_t j = 0; j < subrectangles; j++) {
+                displayio_area_t subrectangle = {
+                    .x1 = 0,
+                    .y1 = rows_per_buffer * j,
+                    .x2 = displayio_area_width(&whole_screen) - 1,
+                    .y2 = rows_per_buffer * (j + 1) - 1
+                };
+
+                displayio_display_begin_transaction(display);
+                displayio_display_set_region_to_update(display, subrectangle.x1, subrectangle.y1,
+                                                                subrectangle.x2 + 1, subrectangle.y2 + 1);
+                displayio_display_end_transaction(display);
+
+                // Handle display mirroring and transpose.
+                displayio_area_t transformed_subrectangle;
+                displayio_buffer_transform_t transform;
+                if (display->mirror_x) {
+                    uint16_t width = displayio_area_width(&whole_screen);
+                    transformed_subrectangle.x1 = width - subrectangle.x2 - 1;
+                    transformed_subrectangle.x2 = width - subrectangle.x1 - 1;
+                } else {
+                    transformed_subrectangle.x1 = subrectangle.x1;
+                    transformed_subrectangle.x2 = subrectangle.x2;
+                }
+                if (display->mirror_y != display->transpose_xy) {
+                    uint16_t height = displayio_area_height(&whole_screen);
+                    transformed_subrectangle.y1 = height - subrectangle.y2 - 1;
+                    transformed_subrectangle.y2 = height - subrectangle.y1 - 1;
+                } else {
+                    transformed_subrectangle.y1 = subrectangle.y1;
+                    transformed_subrectangle.y2 = subrectangle.y2;
+                }
+                transform.width = transformed_subrectangle.x2 - transformed_subrectangle.x1 + 1;
+                transform.height = transformed_subrectangle.y2 - transformed_subrectangle.y1 + 1;
+                if (display->transpose_xy) {
+                    int16_t y1 = transformed_subrectangle.y1;
+                    int16_t y2 = transformed_subrectangle.y2;
+                    transformed_subrectangle.y1 = transformed_subrectangle.x1;
+                    transformed_subrectangle.y2 = transformed_subrectangle.x2;
+                    transformed_subrectangle.x1 = y1;
+                    transformed_subrectangle.x2 = y2;
+                }
+                transform.transpose_xy = display->transpose_xy;
+                transform.mirror_x = display->mirror_x;
+                transform.mirror_y = display->mirror_y;
+                transform.scale = 1;
+
+                uint32_t mask[(buffer_size / 32) + 1];
+                for (uint16_t k = 0; k < (buffer_size / 32) + 1; k++) {
+                    mask[k] = 0x00000000;
+                }
+                bool full_coverage = displayio_group_get_area(display->current_group, &transform, &transformed_subrectangle, mask, buffer);
+                if (!full_coverage) {
+                    uint32_t index = 0;
+                    uint32_t current_mask = 0;
+                    for (int16_t y = subrectangle.y1; y <= subrectangle.y2; y++) {
+                        for (int16_t x = subrectangle.x1; x <= subrectangle.x2; x++) {
+                            if (index % 32 == 0) {
+                                current_mask = mask[index / 32];
+                            }
+                            if ((current_mask & (1 << (index % 32))) == 0) {
+                                ((uint16_t*) buffer)[index] = 0x0000;
+                            }
+                            index++;
                         }
-                        displayio_display_send_pixels(display, buffer, buffer_size / 2);
-                        displayio_display_end_transaction(display);
-                        // TODO(tannewt): Make refresh displays faster so we don't starve other
-                        // background tasks.
-                        usb_background();
-                        index = 0;
                     }
                 }
-            }
 
-            if (skip_this_display) {
-                // Go on to next display.
-                continue;
-            }
-            // Send the remaining data.
-            if (index) {
                 if (!displayio_display_begin_transaction(display)) {
-                    // Can't get display bus. Skip the rest of the data. Try next display.
-                    continue;
+                    // Can't acquire display bus; skip the rest of the data. Try next display.
+                    break;
                 }
-                displayio_display_send_pixels(display, buffer, index * 2);
+                displayio_display_send_pixels(display, buffer, buffer_size / 2);
+                displayio_display_end_transaction(display);
+
+                // TODO(tannewt): Make refresh displays faster so we don't starve other
+                // background tasks.
+                usb_background();
             }
-            displayio_display_end_transaction(display);
         }
         displayio_display_finish_refresh(display);
     }
@@ -220,3 +231,57 @@ void reset_displays(void) {
     }
     #endif
 }
+
+void displayio_area_shift(displayio_area_t* area, int16_t dx, int16_t dy) {
+    area->x1 += dx;
+    area->y1 += dy;
+    area->x2 += dx;
+    area->y2 += dy;
+}
+
+bool displayio_area_compute_overlap(const displayio_area_t* a,
+                                    const displayio_area_t* b,
+                                    displayio_area_t* overlap) {
+    overlap->x1 = a->x1;
+    if (b->x1 > overlap->x1) {
+        overlap->x1 = b->x1;
+    }
+    overlap->x2 = a->x2;
+    if (b->x2 < overlap->x2) {
+        overlap->x2 = b->x2;
+    }
+    if (overlap->x1 > overlap->x2) {
+        return false;
+    }
+    overlap->y1 = a->y1;
+    if (b->y1 > overlap->y1) {
+        overlap->y1 = b->y1;
+    }
+    overlap->y2 = a->y2;
+    if (b->y2 < overlap->y2) {
+        overlap->y2 = b->y2;
+    }
+    if (overlap->y1 > overlap->y2) {
+        return false;
+    }
+    return true;
+}
+
+uint16_t displayio_area_width(const displayio_area_t* area) {
+    return area->x2 - area->x1 + 1;
+}
+
+uint16_t displayio_area_height(const displayio_area_t* area) {
+    return area->y2 - area->y1 + 1;
+}
+
+uint32_t displayio_area_size(const displayio_area_t* area) {
+    return displayio_area_width(area) * displayio_area_height(area);
+}
+
+bool displayio_area_equal(const displayio_area_t* a, const displayio_area_t* b) {
+    return a->x1 == b->x1 &&
+           a->y1 == b->y1 &&
+           a->x2 == b->x2 &&
+           a->y2 == b->y2;
+}

shared-module/displayio/__init__.h

@@ -24,8 +24,8 @@
  * THE SOFTWARE.
  */
 
-#ifndef MICROPY_INCLUDED_SHARED_BINDINGS_DISPLAYIO___INIT___H
-#define MICROPY_INCLUDED_SHARED_BINDINGS_DISPLAYIO___INIT___H
+#ifndef MICROPY_INCLUDED_SHARED_MODULE_DISPLAYIO___INIT___H
+#define MICROPY_INCLUDED_SHARED_MODULE_DISPLAYIO___INIT___H
 
 #include "shared-bindings/displayio/Display.h"
 #include "shared-bindings/displayio/FourWire.h"
@@ -47,4 +47,4 @@ extern displayio_group_t circuitpython_splash;
 void displayio_refresh_displays(void);
 void reset_displays(void);
 
-#endif // MICROPY_INCLUDED_SHARED_BINDINGS_DISPLAYIO___INIT___H
+#endif // MICROPY_INCLUDED_SHARED_MODULE_DISPLAYIO___INIT___H

shared-module/displayio/area.h

@@ -0,0 +1,57 @@
+/*
+ * This file is part of the Micro Python project, http://micropython.org/
+ *
+ * The MIT License (MIT)
+ *
+ * Copyright (c) 2019 Scott Shawcroft for Adafruit Industries
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef MICROPY_INCLUDED_SHARED_MODULE_DISPLAYIO_AREA_H
+#define MICROPY_INCLUDED_SHARED_MODULE_DISPLAYIO_AREA_H
+
+// Implementations are in __init__.c
+
+typedef struct {
+    int16_t x1;
+    int16_t y1;
+    int16_t x2; // Second point is inclusive.
+    int16_t y2;
+} displayio_area_t;
+
+typedef struct {
+    uint16_t width;
+    uint16_t height;
+    uint8_t scale;
+    bool mirror_x;
+    bool mirror_y;
+    bool transpose_xy;
+} displayio_buffer_transform_t;
+
+void displayio_area_shift(displayio_area_t* area, int16_t dx, int16_t dy);
+bool displayio_area_compute_overlap(const displayio_area_t* a,
+                                    const displayio_area_t* b,
+                                    displayio_area_t* overlap);
+uint16_t displayio_area_width(const displayio_area_t* area);
+uint16_t displayio_area_height(const displayio_area_t* area);
+uint32_t displayio_area_size(const displayio_area_t* area);
+bool displayio_area_equal(const displayio_area_t* a, const displayio_area_t* b);
+
+#endif // MICROPY_INCLUDED_SHARED_MODULE_DISPLAYIO_AREA_H

supervisor/shared/display.c

@@ -70,8 +70,8 @@ void supervisor_start_terminal(uint16_t width_px, uint16_t height_px) {
 
     grid->width_in_tiles = width_in_tiles;
     grid->height_in_tiles = height_in_tiles;
-    grid->total_width = width_in_tiles * grid->tile_width;
-    grid->total_height = height_in_tiles * grid->tile_height;
+    grid->area.x2 = grid->area.x1 + width_in_tiles * grid->tile_width - 1;
+    grid->area.y2 = grid->area.y1 + height_in_tiles * grid->tile_height - 1;
     grid->tiles = tiles;
 
     supervisor_terminal.cursor_x = 0;
@@ -157,13 +157,15 @@ displayio_tilegrid_t blinka_sprite = {
     .base = {.type = &displayio_tilegrid_type },
     .bitmap = &blinka_bitmap,
     .pixel_shader = &blinka_palette,
-    .x = 0,
-    .y = 0,
+    .area = {
+        .x1 = 0,
+        .y1 = 0,
+        .x2 = 16,
+        .y2 = 16
+    },
     .bitmap_width_in_tiles = 1,
     .width_in_tiles = 1,
     .height_in_tiles = 1,
-    .total_width = 16,
-    .total_height = 16,
     .tile_width = 16,
     .tile_height = 16,
     .top_left_x = 16,

tools/gen_display_resources.py

@@ -120,13 +120,15 @@ displayio_tilegrid_t supervisor_terminal_text_grid = {{
     .base = {{ .type = &displayio_tilegrid_type }},
     .bitmap = (displayio_bitmap_t*) &supervisor_terminal_font_bitmap,
     .pixel_shader = &supervisor_terminal_color,
-    .x = 16,
-    .y = 0,
+    .area = {{
+        .x1 = 16,
+        .y1 = 0,
+        .x2 = {1} + 16,
+        .y2 = {2},
+    }},
     .bitmap_width_in_tiles = {0},
     .width_in_tiles = 1,
     .height_in_tiles = 1,
-    .total_width = {1},
-    .total_height = {2},
     .tile_width = {1},
     .tile_height = {2},
     .tiles = NULL,