tab_stop unguardedSeverity: LOW
Status: ✅ FIXED
Files: src/input.zig:690-696, src/screen.zig:759
Verification status: identified by static analysis; not reproduced.
// src/input.zig:690-696 (abridged)
'Z' => {
const count = @min(self.paramDefault(0, 1), 1000);
var i: u32 = 0;
while (i < count) : (i += 1) {
if (self.screen.cursor.x > 0) {
const ts = self.screen.tab_stop;
const remainder = self.screen.cursor.x % ts;
Bug #463 (“Screen TAB handling underflows on zero width and can divide by zero”) added a guard to the TAB path:
// src/screen.zig:759
if (ts == 0) break;
The CBT path performs the same modulo without it. tab_stop defaults to 8 and
is only assigned 0 by tests, so this is unreachable in production today — but it
is a latent trap and the two handlers of the same concept now disagree about
whether tab_stop can be zero.
const ts = self.screen.tab_stop;
if (ts == 0) break;
Or, better, make tab_stop a non-zero type so the invariant is enforced once
rather than checked at each use site.
// src/input.zig — CBT
const ts = self.screen.tab_stop;
if (ts == 0) break; // bug #526, matching the #463 guard in the TAB path
const remainder = self.screen.cursor.x % ts;
CBT (CSI Z) survives a zero tab_stop — bug #526 (src/input.zig) sets
tab_stop = 0, feeds CSI 1 Z, and asserts the cursor does not move. Without
the guard that is a divide-by-zero trap in Debug and ReleaseSafe.
Verified with the full suite on Zig 0.16.0 and 0.17.0: 1050/1050 passing on both.
Bugs #463 (TAB zero-width underflow / divide-by-zero), #321 (uncapped CSI Z
loop — the @min(count, 1000) cap above is that fix).