mirrors/tmux
1
0
Fork 0
mirror of https://github.com/tmux/tmux.git synced 2026-09-22 21:07:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

screen-redraw: use the prompt's own screen to clip its damage range

Browse Source 
redraw_damage_draw_pane_prompt() recomposes a pane's separately rendered
prompt (wp->prompt) over a damaged sub-range of a REDRAW_SPAN_PANE span,
reusing the (x, n) range that redraw_client_damage() already clipped and
grew against the pane's own *content* grid (wp->screen) via
redraw_damage_grow_span_clip(). That growing exists specifically to avoid
splitting a wide character at the range's edge - but the prompt is drawn
into its own, freshly allocated one-line screen with no relationship to
the content grid, so a range that's clean (or correctly grown) for the
content can still land mid-character in the prompt's own grid.

This is invisible whenever the pane's content is plain ASCII: the content
grid has no padding cells to find, so redraw_damage_grow_span_clip() never
grows the range at all, and the raw geometric range - however it landed -
is passed straight through to the prompt. If that range's end lands right
after a base cell whose padding half falls just outside it, tty_draw_line()
has no room left for that cell's second column
(tty_draw_line_get_empty()'s gc->data.width > nx check) and blanks it
entirely, even though the pane's own content never needed the fix at all.

Reproduced with a floating pane's CJK prompt straddling the boundary
between two tiled panes underneath it: a palette change (OSC 4) in one of
the tiled panes triggers a redraw of its own rectangle, which is occluded
by the floating pane but still geometrically overlaps its prompt row,
recomposing a partial range of the prompt that cuts through a character.

Fix: extract the grid-probing tail of redraw_span_cell_is_padding() into
redraw_screen_cell_is_padding(), usable against any screen, and add
redraw_damage_grow_screen_clip() - the same left/right one-step growth as
redraw_damage_grow_span_clip(), but against an explicit screen and
column origin. redraw_damage_draw_pane_prompt() now re-derives its own
(x0, x1) range from the caller's (x, n) by growing it against the
prompt's own screen before drawing, clamped to the span so it can't bleed
into a neighbouring one.

New regress/floating-pane-prompt-wide-character.sh constructs the exact
tiled-pane-boundary scenario above and checks the CJK prompt text is
intact after the trigger. Verified failing 3/3 against the pre-fix code
(a character is blanked) and passing 5/5 standalone plus 2/2 in the full
regress suite (twice, since this touches the same damage-composition path
as every other redraw) against the fix, with no other tests newly broken
(prompt-words-history.sh and the untracked image-support scratch tests
are pre-existing, unrelated failures).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Michael Grant
2026-09-22 17:16:25 +01:00
parent 04dbc6d89b
commit d4adce7e00
2 changed files with 229 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

179
regress/floating-pane-prompt-wide-character.sh Executable file
View File

@@ -0,0 +1,179 @@
#!/bin/sh
# A damage rectangle's clip range is grown to avoid splitting a wide
# character, but redraw_damage_grow_span_clip() (screen-redraw.c) only ever
# checks the span's own pane *content* grid (wp->screen) for that. For a
# REDRAW_SPAN_PANE span, that same range is then also handed to
# redraw_damage_draw_pane_prompt() to recompose the pane's separately
# rendered prompt (wp->prompt, e.g. from "command-prompt -P") over the
# damaged sub-range - but the prompt is drawn into its own, freshly
# allocated one-line screen, unrelated to the pane's content grid, so a
# range grown (or left ungrown) against the content is not necessarily
# grown correctly for the prompt's own wide characters.
#
# This is invisible when the pane's own content is plain ASCII (as here):
# redraw_damage_grow_span_clip() never finds anything to grow against, so
# the raw, ungrown geometric range is passed straight through to the
# prompt - and if that range's edge lands mid-character in the *prompt's*
# grid, tty_draw_line() clears the character it cuts through
# (tty_draw_line_get_empty()'s gc->data.width > nx check, for a trailing
# base cell with no room left for its padding half).
#
# The trigger is a palette change (OSC 4) in a tiled pane that is one half
# of a vertical split running the full height of the window - occluded
# under the floating pane, but still geometrically triggering a redraw of
# its own rectangle. Positioned so the split boundary falls inside the
# floating pane's own CJK prompt, this reproduces exactly Codex's report:
# "a floating pane containing a CJK prompt across a tiled-pane boundary -
# a palette update in the tiled pane blanks a prompt character."
PATH=/bin:/usr/bin
TERM=screen
LC_ALL=C.UTF-8
export PATH TERM LC_ALL
[ -z "$TEST_TMUX" ] && TEST_TMUX=$(readlink -f ../tmux)
DIR=$(mktemp -d) || exit 1
cd "$DIR" || exit 1
INNER="$TEST_TMUX -Lpromptwide-inner-$$ -f/dev/null"
OUTER="$TEST_TMUX -Lpromptwide-outer-$$ -f/dev/null"
EMITTER=$DIR/emitter.pl
CAPTURE=$DIR/capture
FLOAT=
fail()
{
echo "$*" >&2
[ -s "$CAPTURE" ] && cat "$CAPTURE" >&2
exit 1
}
cleanup()
{
$OUTER kill-server 2>/dev/null
$INNER kill-server 2>/dev/null
cd /
rm -rf "$DIR"
}
trap cleanup 0 1 15
wait_for_client()
{
i=0
while [ "$i" -lt 50 ]; do
CLIENT=$($INNER list-clients -F '#{client_name}' 2>/dev/null)
[ -n "$CLIENT" ] && return 0
sleep 0.1
i=$((i + 1))
done
fail "inner client did not attach"
}
wait_outer_has()
{
marker=$1
i=0
while [ "$i" -lt 50 ]; do
$OUTER capture-pane -p -t outer:0.0 >"$CAPTURE" 2>/dev/null || true
grep -q "$marker" "$CAPTURE" && return 0
sleep 0.1
i=$((i + 1))
done
fail "outer client did not show $marker"
}
wait_prompt_row_intact()
{
i=0
while [ "$i" -lt 50 ]; do
$OUTER capture-pane -p -t outer:0.0 >"$CAPTURE" 2>/dev/null || true
line=$(sed -n "${PROMPTROW}p" "$CAPTURE")
case $line in
*"$PROMPTTEXT"*) return 0 ;;
esac
sleep 0.1
i=$((i + 1))
done
fail "the CJK prompt was not intact after the palette-triggered damage - got: $line"
}
cat >"$EMITTER" <<'PERL'
use strict;
use warnings;
$| = 1;
my $line = <STDIN>;
print "\e]4;1;rgb:11/22/33\e\\";
sleep 100;
PERL
$INNER new-session -d -s inner -x 60 -y 12 "perl '$EMITTER'" || exit 1
$INNER set-option -g status off || exit 1
$INNER set-option -g window-size manual || exit 1
LEFT=$($INNER list-panes -t inner -F '#{pane_id}') || exit 1
# Split so the boundary between the two tiled panes falls at column 17 -
# used below to pick a floating-pane column that lands the boundary
# mid-character inside the prompt.
RIGHT=$($INNER split-window -t inner -h -l 43 -PF '#{pane_id}' \
'sleep 100') || exit 1
RX=$($INNER display-message -p -t "$RIGHT" '#{pane_left}') || exit 1
# Create the floating pane, then check its actual resulting position (the
# border-framing offset added to -X is not something to hand-compute). Try
# adjacent starting columns until the split boundary lands on an odd
# (padding-half) column of the prompt's own numbering, and within the
# prompt's 12-column width.
startx=10
tries=0
while [ "$tries" -lt 4 ]; do
[ -n "$FLOAT" ] && $INNER kill-pane -t "$FLOAT" 2>/dev/null
FLOAT=$($INNER new-pane -d -PF '#{pane_id}' -x 24 -y 5 -X "$startx" \
-Y 2 "sh -c 'i=0; while [ \$i -lt 10 ]; do \
printf AAAAAAAAAAAAAAAAAAAAAA\\\\n; i=\$((i+1)); done; sleep 100'") ||
exit 1
sleep 0.2
X1=$($INNER display-message -p -t "$FLOAT" '#{pane_left}')
Y1=$($INNER display-message -p -t "$FLOAT" '#{pane_top}')
H1=$($INNER display-message -p -t "$FLOAT" '#{pane_height}')
local=$((RX - X1 - 1))
if [ "$local" -ge 1 ] && [ "$local" -le 11 ] &&
[ $((local % 2)) -eq 1 ]; then
break
fi
startx=$((startx + 1))
tries=$((tries + 1))
done
local=$((RX - X1 - 1))
[ "$local" -ge 1 ] && [ "$local" -le 11 ] && [ $((local % 2)) -eq 1 ] ||
fail "could not find bad-parity starting column"
$OUTER new-session -d -s outer -x 60 -y 12 'sleep 100' || exit 1
$OUTER set-option -g status off || exit 1
$OUTER set-option -g window-size manual || exit 1
$OUTER set-option -g default-terminal screen-256color || exit 1
$OUTER respawn-pane -k -t outer:0.0 \
"$TEST_TMUX -Lpromptwide-inner-$$ -f/dev/null attach-session -t inner" ||
exit 1
wait_for_client
wait_outer_has AAAAAAAAAAAAAAAAAAAAAA
CLIENT=$($INNER list-clients -F '#{client_name}') || exit 1
$INNER select-pane -t "$FLOAT" || exit 1
PROMPTTEXT=$(printf '\344\270\255' | perl -CSD -ne 'print $_ x 6')
$INNER command-prompt -b -P -t "$CLIENT" -p "$PROMPTTEXT" \
'display-message -- %1' || exit 1
wait_outer_has "$PROMPTTEXT"
PROMPTROW=$((Y1 + H1))
# Trigger the damage: unblock the emitter so it fires the palette change in
# the left tiled pane, which is occluded under (but geometrically overlaps)
# the floating pane's prompt row.
$INNER send-keys -t "$LEFT" Enter || exit 1
wait_prompt_row_intact
exit 0

59
screen-redraw.c
View File

@@ -2069,6 +2069,18 @@ redraw_damage_refresh_status(struct redraw_draw_ctx *dctx,
wp->status_serial = redraw_status_serial;
}
/* Whether the cell at (px, py) in screen s is a padding cell. */
static int
redraw_screen_cell_is_padding(struct screen *s, u_int px, u_int py)
{
struct grid_cell gc;
if (px >= screen_size_x(s))
return (0);
grid_view_get_cell(s->grid, px, py, &gc);
return ((gc.flags & GRID_FLAG_PADDING) != 0);
}
/*
* Whether the cell at scene x-coordinate x within this span is the second
* (padding) half of a wide character - the condition under which growing a
@@ -2092,7 +2104,6 @@ static int
redraw_span_cell_is_padding(struct redraw_span *span, u_int x)
{
struct screen *s;
struct grid_cell gc;
u_int px, py;
switch (span->data.type) {
@@ -2114,10 +2125,7 @@ redraw_span_cell_is_padding(struct redraw_span *span, u_int x)
default:
return (1);
}
if (px >= screen_size_x(s))
return (0);
grid_view_get_cell(s->grid, px, py, &gc);
return ((gc.flags & GRID_FLAG_PADDING) != 0);
return (redraw_screen_cell_is_padding(s, px, py));
}
/*
@@ -2144,6 +2152,26 @@ redraw_damage_grow_span_clip(struct redraw_span *span, u_int *xp, u_int *endp)
(*endp)++;
}
/*
* As redraw_damage_grow_span_clip(), but against an explicit screen: px0 is
* the column in that screen corresponding to span->x, py the row. Used for
* a span's separately rendered content (e.g. a pane's prompt) that isn't
* span->data.p.wp->screen (or whichever grid redraw_span_cell_is_padding()
* would otherwise consult for this span's type), and so has its own,
* unrelated wide-character boundaries at the same columns.
*/
static void
redraw_damage_grow_screen_clip(struct redraw_span *span, struct screen *s,
u_int px0, u_int py, u_int *xp, u_int *endp)
{
if (*xp > span->x &&
redraw_screen_cell_is_padding(s, px0 + (*xp - span->x), py))
(*xp)--;
if (*endp < span->x + span->width &&
redraw_screen_cell_is_padding(s, px0 + (*endp - span->x), py))
(*endp)++;
}
/* Recompose a pane's prompt over a damaged section of its display row. */
static void
redraw_damage_draw_pane_prompt(struct redraw_draw_ctx *dctx,
@@ -2153,7 +2181,7 @@ redraw_damage_draw_pane_prompt(struct redraw_draw_ctx *dctx,
struct window_pane *wp = span->data.p.wp;
struct tty *tty = &scene->c->tty;
struct screen screen;
u_int px, width, prompt_y;
u_int px, width, prompt_y, x0, x1;
if (wp->prompt == NULL || wp->sx == 0 || wp->sy == 0)
return;
@@ -2165,12 +2193,25 @@ redraw_damage_draw_pane_prompt(struct redraw_draw_ctx *dctx,
return;
redraw_make_pane_prompt(wp, &screen);
px = span->data.p.px + (x - span->x);
/*
* x and n were clipped and grown against wp->screen, whose character
* boundaries have nothing to do with the prompt's separately
* rendered screen - realign the range on the prompt's own grid
* instead, clamped to this span so it cannot bleed into a
* neighbouring one.
*/
x0 = x;
x1 = x + n;
redraw_damage_grow_screen_clip(span, &screen, span->data.p.px, 0, &x0,
&x1);
px = span->data.p.px + (x0 - span->x);
if (px < screen_size_x(&screen)) {
width = n;
width = x1 - x0;
if (width > screen_size_x(&screen) - px)
width = screen_size_x(&screen) - px;
tty_draw_line(tty, &screen, px, 0, width, x, y, NULL);
tty_draw_line(tty, &screen, px, 0, width, x0, y, NULL);
}
screen_free(&screen);
}