comment and test regarding notifications on writes that dont change readiness

3 files changed, 17 insertions, 1 deletions

diff --git a/src/tools/miri/src/shims/unix/linux/epoll.rs b/src/tools/miri/src/shims/unix/linux/epoll.rs
index 1ec2dbe233e..ce91712fb27 100644
--- a/src/tools/miri/src/shims/unix/linux/epoll.rs
+++ b/src/tools/miri/src/shims/unix/linux/epoll.rs
@@ -35,6 +35,7 @@ impl EpollEventInstance {
         EpollEventInstance { events, data }
     }
 }
+
 /// EpollEventInterest registers the file description information to an epoll
 /// instance during a successful `epoll_ctl` call. It also stores additional
 /// information needed to check and update readiness state for `epoll_wait`.
@@ -434,7 +435,9 @@ pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
 
     /// For a specific file description, get its ready events and update
     /// the corresponding ready list. This function is called whenever a file description
-    /// is registered with epoll, or when its readiness *might* have changed.
+    /// is registered with epoll, or the buffer it reads from / writes to changed.
+    /// This *will* report an event if anyone is subscribed to it, without any further
+    /// filtering, so do not call this function when an FD didn't have anything happen to it!
     fn check_and_update_readiness(&self, fd_ref: &FileDescriptionRef) -> InterpResult<'tcx, ()> {
         let this = self.eval_context_ref();
         let id = fd_ref.get_id();
diff --git a/src/tools/miri/src/shims/unix/socket.rs b/src/tools/miri/src/shims/unix/socket.rs
index 0f13c8c35ad..cdc9dd0429b 100644
--- a/src/tools/miri/src/shims/unix/socket.rs
+++ b/src/tools/miri/src/shims/unix/socket.rs
@@ -200,6 +200,7 @@ impl FileDescription for SocketPair {
         drop(writebuf);
 
         // Notification should be provided for peer fd as it became readable.
+        // The kernel does this even if the fd was already readable before, so we follow suit.
         ecx.check_and_update_readiness(&peer_fd)?;
 
         return Ok(Ok(actual_write_size));
diff --git a/src/tools/miri/tests/pass-dep/libc/libc-epoll.rs b/src/tools/miri/tests/pass-dep/libc/libc-epoll.rs
index 364a784574a..c27a6a83a05 100644
--- a/src/tools/miri/tests/pass-dep/libc/libc-epoll.rs
+++ b/src/tools/miri/tests/pass-dep/libc/libc-epoll.rs
@@ -83,6 +83,18 @@ fn test_epoll_socketpair() {
     let expected_value = u64::try_from(fds[1]).unwrap();
     assert!(check_epoll_wait::<8>(epfd, vec![(expected_event, expected_value)]));
 
+    // Check that this is indeed using "ET" (edge-trigger) semantics: a second epoll should return nothing.
+    assert!(check_epoll_wait::<8>(epfd, vec![]));
+
+    // Write some more to fd[0].
+    let data = "abcde".as_bytes().as_ptr();
+    let res = unsafe { libc::write(fds[0], data as *const libc::c_void, 5) };
+    assert_eq!(res, 5);
+
+    // This did not change the readiness of fd[1]. And yet, we're seeing the event reported
+    // again by the kernel, so Miri does the same.
+    assert!(check_epoll_wait::<8>(epfd, vec![(expected_event, expected_value)]));
+
     // Close the peer socketpair.
     let res = unsafe { libc::close(fds[0]) };
     assert_eq!(res, 0);