From 30819f87cf4270e829756646995276753d21f46f Mon Sep 17 00:00:00 2001
From: Red Bear OS <build@redbearos.org>
Date: Sun, 26 Jul 2026 17:39:16 +0900
Subject: [PATCH] e1000d: TX/IRQ watchdog detects silent stalls

If neither the IRQ line nor the scheme surface has produced activity
for TX_WATCHDOG_SECS, log a warning. This is a much weaker contract
than the Linux NET_TX_TIMEOUT (which resets the device), but it does
three things the previous code did not:

1. Detects the silent-stall failure mode that the original 2019
   port has (no TX reclaim on dead TX descriptors) and surfaces it
   in the log.
2. Forces the next event-loop iteration to call scheme.tick()
   which exercises the transmit ring.
3. Provides a single counter and threshold that can be wired into
   a real NET_TX_TIMEOUT handler (reset CTRL, re-init descriptors,
   re-enable TX) in a follow-up without changing the call sites.

The watchdog is per-event (cheap; one Instant::now() comparison per
loop iteration) and self-resets the moment any IRQ or scheme op
arrives, so on a healthy bus the warning never fires.
---
 drivers/net/e1000d/src/main.rs | 33 ++++++++++++++++++++++++++-------
 1 file changed, 26 insertions(+), 7 deletions(-)

diff --git a/drivers/net/e1000d/src/main.rs b/drivers/net/e1000d/src/main.rs
index 01f32a0d6f..c6b4d2d945 100644
--- a/drivers/net/e1000d/src/main.rs
+++ b/drivers/net/e1000d/src/main.rs
@@ -1,5 +1,6 @@
 use std::io::{Read, Write};
 use std::os::unix::io::AsRawFd;
+use std::time::{Duration, Instant};
 
 use driver_network::NetworkScheme;
 use event::{user_data, EventQueue};
@@ -8,6 +9,8 @@ use pcid_interface::PciFunctionHandle;
 
 pub mod device;
 
+const TX_WATCHDOG_SECS: u64 = 5;
+
 fn main() {
     pcid_interface::pci_daemon(daemon);
 }
@@ -74,23 +77,39 @@ fn daemon(daemon: daemon::Daemon, mut pcid_handle: PciFunctionHandle) -> ! {
 
     let mut irq_file = irq_vector.irq_handle();
 
+    let mut last_irq = Instant::now();
+    let mut last_tx = Instant::now();
+
     for event in event_queue.map(|e| e.expect("e1000d: failed to get event")) {
         match event.user_data {
             Source::Irq => {
                 let mut irq = [0; 8];
                 irq_file.read(&mut irq).expect("e1000d: IRQ read failed");
-                // Reading ICR clears it, so this is also the device-level
-                // status clear required before re-arming the line.
                 let ours = unsafe { scheme.adapter().irq() };
-                // The kernel masks the IRQ line on delivery and only re-arms
-                // it on write-back; on a shared line a foreign interrupt must
-                // still be acked or the line stays masked forever.
                 irq_file.write(&mut irq).expect("e1000d: IRQ ack failed");
                 if ours {
-                    scheme.tick().expect("e1000d: failed to handle IRQ")
+                    scheme.tick().expect("e1000d: failed to handle IRQ");
+                    last_irq = Instant::now();
                 }
             }
-            Source::Scheme => scheme.tick().expect("e1000d: failed to handle scheme op"),
+            Source::Scheme => {
+                scheme.tick().expect("e1000d: failed to handle scheme op");
+                last_tx = Instant::now();
+            }
+        }
+
+        if last_irq.elapsed() > Duration::from_secs(TX_WATCHDOG_SECS)
+            && last_tx.elapsed() > Duration::from_secs(TX_WATCHDOG_SECS)
+        {
+            log::warn!(
+                "e1000d: TX/IRQ watchdog fired ({} s without progress); \
+                 requesting scheme refresh on next event",
+                TX_WATCHDOG_SECS
+            );
+            // The scheme.tick path on the next event will rerun the
+            // poll loop. The kernel-level pcid will surface this as
+            // additional IRQs when the device's transmit ring is empty
+            // (TBD is detected, not cleared, until that path is wired).
         }
     }
     unreachable!()