net: time the stall, not the wait, before dropping an unreadable peer

Found auditing the previous commit for the push. Timing from the first
rewind would drop a peer whose message is large and merely slow, which
would be a fault of our own making rather than one we fixed.

Restart the clock whenever the buffer has grown by even one byte since the
last look. A transfer still making progress is never dropped however long
it takes; only one that has stopped growing while the node goes on
expecting more is, which is the actual failure.

(cherry picked from commit 5dddb26bae02013ce87719b11953aadf0bfa2594)

1 files changed, 6 insertions, 2 deletions

diff --git a/src/net.h b/src/net.h
index 112426e..440550c 100644
--- a/src/net.h
+++ b/src/net.h
@@ -670,9 +670,13 @@ public:
     string strBtfAddr;
     string strBtfMeeting;
 
-    // When this peer's message loop first rewound a header to wait for the
-    // rest of a message, or 0 when it is not waiting. See ProcessMessages.
+    // When this peer's message loop last rewound a header without having made
+    // progress, or 0 when it is not waiting, plus how much was buffered then.
+    // The pair times a STALL rather than a duration, so a large message
+    // arriving slowly is never mistaken for one that is never arriving.
+    // See ProcessMessages.
     int64 nMsgWaitSince;
+    unsigned int nMsgWaitSize;
 
     // Height this peer announced in its version message, or -1 if it sent a
     // version message that predates the field. Knowing how far along everyone