video: Optimise video_flush_copy() for full-line damage
When copying partial framebuffer regions line by line, there is overhead from multiple memcpy() calls. Optimise video_flush_copy() to detect when entire lines are being copied (damage spans full width) and perform a single memcpy() for the whole region instead of looping line by line. Also invert the early-exit check to reduce nesting. Co-developed-by: Claude <noreply@anthropic.com> Signed-off-by: Simon Glass <simon.glass@canonical.com>
This commit is contained in:
@@ -507,16 +507,25 @@ static void video_flush_copy(struct udevice *vid)
|
||||
if (!priv->copy_fb)
|
||||
return;
|
||||
|
||||
if (damage->x1 && damage->y1) {
|
||||
int lstart = damage->x0 * VNBYTES(priv->bpix);
|
||||
int lend = damage->x1 * VNBYTES(priv->bpix);
|
||||
if (!damage->x1 || !damage->y1)
|
||||
return;
|
||||
|
||||
int lstart = damage->x0 * VNBYTES(priv->bpix);
|
||||
int llen = damage->x1 * VNBYTES(priv->bpix) - lstart;
|
||||
|
||||
/* Copy entire region at once if full lines are damaged */
|
||||
if (!lstart && llen == priv->line_length) {
|
||||
ulong offset = damage->y0 * priv->line_length;
|
||||
ulong len = (damage->y1 - damage->y0) * priv->line_length;
|
||||
|
||||
memcpy(priv->copy_fb + offset, priv->fb + offset, len);
|
||||
} else {
|
||||
int y;
|
||||
|
||||
for (y = damage->y0; y < damage->y1; y++) {
|
||||
ulong offset = (y * priv->line_length) + lstart;
|
||||
ulong len = lend - lstart;
|
||||
ulong offset = y * priv->line_length + lstart;
|
||||
|
||||
memcpy(priv->copy_fb + offset, priv->fb + offset, len);
|
||||
memcpy(priv->copy_fb + offset, priv->fb + offset, llen);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user