fix: stacks at top of user VA in own 1 MiB band (guard, heap kept below), 8/16 pages default; kernel fault msg names pid, kind, flags stack overflow; flaky OVERFLOW + spawntest/test-qemu check it

crates/apps/spawntest/src/main.rs

@@ -267,6 +267,16 @@ fn restarted_after_hanging() -> Result<(), &'static str> {
     Ok(())
 }
 
+/// A server that runs off its stack faults on the guard below it, and is
+/// restarted like any other that died.
+fn restarted_after_overflowing() -> Result<(), &'static str> {
+    let _ = syscall::request(FLAKY, [flaky::op::OVERFLOW, 0, 0, 0]);
+    if flaky()? != 1 {
+        return Err("the overflowed server was not restarted");
+    }
+    Ok(())
+}
+
 fn check() -> Result<(), &'static str> {
     let mut fs = Storage::attach(STORAGE, STORAGE_BUF_VA).map_err(|_| "cannot attach to storage")?;
     write_a_file(&mut fs)?;
@@ -305,6 +315,9 @@ fn check() -> Result<(), &'static str> {
 
     restarted_after_hanging()?;
     libkos::println!("spawntest: a hung server is noticed, stopped and restarted: ok");
+
+    restarted_after_overflowing()?;
+    libkos::println!("spawntest: a server that overflows its stack is caught and restarted: ok");
     Ok(())
 }