vt: pad double-width code points with a zero-white-space

In the Unicode screen buffer, we follow double-width code points with a
space to maintain proper column alignment. This, however, creates
semantic problems when e.g. using cut and paste or selection.

Let's use a better code point for the column padding's purpose i.e. a
zero-white-space rather than a full space.

Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Link: https://lore.kernel.org/r/20250410011839.64418-12-nico@fluxnic.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/drivers/tty/vt/vt.c b/drivers/tty/vt/vt.c
index e3d35c4f92045a8ed7c18ac2193d2913381249d4..dc84f9c6b7c6238277a2ebb34e567c16649b314e 100644 (file)
--- a/drivers/tty/vt/vt.c
+++ b/drivers/tty/vt/vt.c
@@ -2937,12 +2937,13 @@ static int vc_con_write_normal(struct vc_data *vc, int tc, int c,
                        width = 2;
                } else if (ucs_is_zero_width(c)) {
                        prev_c = vc_uniscr_getc(vc, -1);
-                       if (prev_c == ' ' &&
+                       if (prev_c == 0x200B &&
                            ucs_is_double_width(vc_uniscr_getc(vc, -2))) {
                                /*
                                 * Let's merge this zero-width code point with
                                 * the preceding double-width code point by
-                                * replacing the existing whitespace padding.
+                                * replacing the existing zero-white-space
+                                * padding.
                                 */
                                vc_con_rewind(vc);
                        } else if (c == 0xfe0f && prev_c != 0) {
@@ -3040,7 +3041,11 @@ static int vc_con_write_normal(struct vc_data *vc, int tc, int c,
                tc = conv_uni_to_pc(vc, ' ');
                if (tc < 0)
                        tc = ' ';
-               next_c = ' ';
+               /*
+                * Store a zero-white-space in the Unicode screen given that
+                * the previous code point is semantically double-width.
+                */
+               next_c = 0x200B;
        }
 
 out: