Writes held before the shell's clock is first seen no longer overflow its first step

A 9P write served before a detached host's first pump holds its batch stamped at time 0, so nextWake said to wake at 20 ms while advance, seeing the clock for the first time, set stepped_ns days later and counted due - stepped_ns back past zero: an integer-overflow panic. It was what the 64 KiB msize trial hit in fs-test (restore detached space, the connection closing at the first write's clunk), a race any msize can lose. A batch's wake is now never before the next frame.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

1 files changed, 4 insertions, 2 deletions

diff --git a/src/pardes.zig b/src/pardes.zig
index c1b12a8b..cb39b5d1 100644
--- a/src/pardes.zig
+++ b/src/pardes.zig
@@ -7129,8 +7129,10 @@ pub const Pardes = struct {
     /// sleep exactly as long as nothing changes.
     pub fn nextWake(p: *const Pardes) ?u64 {
         const next = p.stepped_ns + animation.frame_ns;
-        // Held 9P writes go in when they pause: wake for it.
-        if (p.fs.batch.bytes.items.len > 0) return p.fs.batch.last_ns + ctlfs.pane.batch_pause_ns;
+        // Held 9P writes go in when they pause: wake for it, never before
+        // the next frame (writes served before the shell's clock was first
+        // seen are stamped 0, which `advance` would count back from).
+        if (p.fs.batch.bytes.items.len > 0) return @max(next, p.fs.batch.last_ns + ctlfs.pane.batch_pause_ns);
         if (p.chrome_animation.isActive() or p.presentation.animating()) return next;
         var frames: ?u64 = null;
         if (p.look_hover_wait) |waiting| {