1 files changed, 500 insertions, 21 deletions

diff --git a/src/main.cpp b/src/main.cpp
index 7b3e555..6b8e0de 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -14,6 +14,7 @@
 #include <atomic>
 #include <mutex>
 #include <thread>
+#include <system_error>
 #include "bip32.h"
 #include "btfaddr.h"
 #include "btftunnel.h"
@@ -64,6 +65,137 @@ CBlockIndex* pindexBest = NULL;
 map<uint256, CBlock*> mapOrphanBlocks;
 multimap<uint256, CBlock*> mapOrphanBlocksByPrev;
 
+// Block bodies we have sent a getdata for and not yet received, so the
+// failover pass in SendMessages can tell "requested and being served" from
+// "requested and gone quiet" and re-issue the request to a different peer
+// (#33). Keyed by block hash; setAsked is the peers already tried since the
+// round last restarted, so the retry rotates instead of hammering the
+// announcer that is not serving. Guarded by cs_main, which both the message
+// handlers and SendMessages already hold.
+//
+// Capped small on purpose. At the tip there is one block in flight; during
+// initial download there are legitimately hundreds outstanding to one peer,
+// and rotating those would duplicate the whole download to every other peer.
+//
+// The cap is a shared resource and nothing on the path from inv to insertion
+// can tell a real block hash from 32 random bytes -- AlreadyHave only asks
+// whether we already hold the block, and there is no header to check work
+// against. So the table does not refuse a new hash when it is full; it makes
+// room. See MakeRoomInBlockFetch for the rule, which is what stops one peer
+// filling all eight slots and switching failover off for everybody.
+//
+// The initialisers matter: the insertion path uses operator[] and then tests
+// nTimeFirstAsked == 0 to detect a fresh entry. That worked on
+// value-initialisation alone, but only for as long as nobody gave this struct
+// a constructor.
+struct CBlockFetch
+{
+    int64 nTimeFirstAsked = 0;  // when the first getdata for this block went out
+    int64 nTimeAsked      = 0;  // when the latest getdata went out
+    CAddress addrAnnounced;     // peer whose announcement put this hash here
+    set<CAddress> setAsked;     // peers tried since the round last restarted
+};
+map<uint256, CBlockFetch> mapBlockFetch;
+static const int BLOCK_FETCH_TRACK_MAX = 8;
+// A block nobody has served for this long is not a fetch problem: the hash may
+// have been announced and never existed. Drop the entry rather than getdata
+// forever; a re-announcement recreates it.
+static const int BLOCK_FETCH_GIVEUP_SECS = 10 * 60;
+
+// Failover firing at all is worth saying out loud: it means peers are
+// announcing blocks they then do not serve, which is the pathology of #33 and
+// the thing nobody could otherwise see. Naming the peer on every re-request is
+// a different matter. A peer can drive that line as often as it likes by
+// announcing hashes it will not serve, and what it prints is a peer address,
+// which is the one thing the .btf transport otherwise takes trouble not to
+// spread around. So the per-peer detail sits behind the net category and the
+// unconditional line is a count over a window no peer can shorten: the first
+// failover after a quiet spell prints at once, because the change from "blocks
+// arrive" to "blocks are announced and not served" is the event worth seeing,
+// and everything after it is summarised once a minute.
+static const int BLOCK_FETCH_REPORT_SECS = 60;
+static int64 nBlockFetchWindowStart = 0;
+static int   nBlockFetchFailovers   = 0;
+
+// How stale our own best block may be before we stop calling this the tip.
+// Blocks target two minutes, so this is thirty intervals: far longer than any
+// stall worth failing over from -- the one in #33 ran sixteen minutes -- and
+// far shorter than catching up, which starts hours or days back.
+static const int BLOCK_FETCH_TIP_AGE_SECS = 60 * 60;
+
+// Whether this node is tracking the tip rather than still catching up.
+//
+// Measured off the age of our own best block, not off any height a peer told
+// us. nStartingHeight is optional in the version message and every node
+// released so far sends the short form, so it is -1 for essentially every
+// peer; and where it is filled in later it is filled in from the height of a
+// block that peer served us, which during catch-up tracks our own height and
+// would report "at the tip" for the whole download. Block timestamps are the
+// one clock here that peers cannot move: CheckBlock refuses anything more than
+// two hours ahead and AcceptBlock refuses anything at or before the previous
+// median, so the only way to make this read "at the tip" falsely is to mine,
+// and all that buys is the 15s spacing this code had unconditionally anyway.
+bool IsBlockFetchNearTip()
+{
+    return pindexBest && (int64)pindexBest->nTime > GetTime() - BLOCK_FETCH_TIP_AGE_SECS;
+}
+
+// Drop one entry so a hash announced by addrNew can be tracked.
+//
+// Refusing the insertion instead was a remote off-switch for the whole
+// feature. An inv carrying eight hashes that never existed filled every slot
+// -- nothing between inv and here can tell those from real blocks -- and no
+// genuinely new block was tracked again until the give-up timer expired ten
+// minutes later, re-armable for one message per ten minutes from a single
+// connection.
+//
+// So evict instead, by occupancy: a peer may only take a slot from a peer
+// holding strictly more slots than itself, and otherwise recycles one of its
+// own. That is what keeps eviction from becoming the same attack in a new
+// shape. A peer that has filled the table is by definition the heaviest
+// holder, so its own entries are the only ones it can displace, while a peer
+// announcing its first block holds nothing and therefore always evicts from
+// the heaviest -- a genuine announcement is never refused a slot, whatever the
+// attacker holds or however many connections it holds it from. The one case
+// that costs a real fetch is eight distinct peers holding one entry each,
+// where a ninth takes the oldest; that is a busy healthy node, not an attack,
+// and losing the oldest of eight beats refusing the ninth.
+static void MakeRoomInBlockFetch(const CAddress& addrNew)
+{
+    if (mapBlockFetch.empty())
+        return;
+
+    map<CAddress, int> mapHeld;
+    for (map<uint256, CBlockFetch>::iterator mi = mapBlockFetch.begin(); mi != mapBlockFetch.end(); ++mi)
+        mapHeld[(*mi).second.addrAnnounced]++;
+
+    int nMine = mapHeld.count(addrNew) ? mapHeld[addrNew] : 0;
+
+    int nMax = 0;
+    CAddress addrVictim;
+    for (map<CAddress, int>::iterator mi = mapHeld.begin(); mi != mapHeld.end(); ++mi)
+        if ((*mi).second > nMax)
+        {
+            nMax = (*mi).second;
+            addrVictim = (*mi).first;
+        }
+
+    // Nobody is holding more than this peer, so it recycles its own oldest
+    // rather than taking a slot off an equal.
+    if (nMax <= nMine)
+        addrVictim = addrNew;
+
+    map<uint256, CBlockFetch>::iterator miVictim = mapBlockFetch.end();
+    for (map<uint256, CBlockFetch>::iterator mi = mapBlockFetch.begin(); mi != mapBlockFetch.end(); ++mi)
+        if ((*mi).second.addrAnnounced == addrVictim &&
+            (miVictim == mapBlockFetch.end() ||
+             (*mi).second.nTimeFirstAsked < (*miVictim).second.nTimeFirstAsked))
+            miVictim = mi;
+
+    if (miVictim != mapBlockFetch.end())
+        mapBlockFetch.erase(miVictim);
+}
+
 map<uint256, CDataStream*> mapOrphanTransactions;
 multimap<uint256, CDataStream*> mapOrphanTransactionsByPrev;
 
@@ -1712,7 +1844,7 @@ void CWalletTx::RelayWalletTransaction(CTxDB& txdb)
         uint256 hash = GetHash();
         if (!txdb.ContainsTx(hash))
         {
-            if (LogAcceptsCategory("net")) printf("Relaying wtx %s\n", hash.ToString().substr(0,6).c_str());
+            if (LogAcceptsCategory("relay")) printf("Relaying wtx %s\n", hash.ToString().substr(0,6).c_str());
             RelayMessage(CInv(MSG_TX, hash), (CTransaction)*this);
         }
     }
@@ -2678,7 +2810,18 @@ bool LoadBlockIndex(bool fAllowNew)
         //     CTxOut(nValue=50.00000000, scriptPubKey=0x5F1DF16B2B704C8A578D0B)
         //   vMerkleTree: 4a5e1e
 
-        // Bitflash genesis block -- own header (fair launch, 2026 stable relaunch)
+        // Bitflash genesis block -- own header (fair launch)
+        //
+        // Do not change pszTimestamp, the scriptPubKey, nTime or nNonce. They
+        // are the inputs to the genesis hash, and the genesis hash is
+        // consensus. Changing one on its own trips the assert below, which is
+        // loud and reverts cleanly; changing one and regenerating
+        // GENESIS_NONCE and hashGenesisBlock to match -- which is exactly what
+        // the re-mining branch below prints out and invites -- gives a node
+        // that starts without complaint, builds its own chain from height 0,
+        // and rejects every real peer as being on a different chain. That one
+        // fails silently, and the damage scales with how many people run the
+        // "fixed" build.
         char* pszTimestamp = "Bitflash 24/Jul/2026 Fair launch: one CPU one vote, no premine";
         CTransaction txNew;
         txNew.vin.resize(1);
@@ -2691,7 +2834,14 @@ bool LoadBlockIndex(bool fAllowNew)
         block.hashPrevBlock = 0;
         block.hashMerkleRoot = block.BuildMerkleTree();
         block.nVersion = 1;
-        block.nTime    = 1753315200; // 2026-07-24 00:00:00 UTC (stable relaunch)
+        // 1753315200 is 2025-07-24T00:00:00Z. The headline in pszTimestamp
+        // above reads 2026 and does not match it. This is a known discrepancy
+        // and not a typo in the constant: the value is only consensus-critical
+        // as an input to the genesis hash, the block hashes to
+        // hashGenesisBlock with it, and every node on the network agrees. The
+        // mismatch is in the launched chain and cannot be corrected -- see the
+        // warning above before deciding otherwise.
+        block.nTime    = 1753315200;
         block.nBits    = 0x1f0fffff; // = bnProofOfWorkLimit (Bitflash RandomX floor)
         block.nNonce   = GENESIS_NONCE;
 
@@ -2718,6 +2868,11 @@ bool LoadBlockIndex(bool fAllowNew)
             printf("powhash = %s\n", block.GetPoWHash().ToString().c_str());
             printf("hash    = %s\n", block.GetHash().ToString().c_str());
             printf("merkle  = %s\n", block.hashMerkleRoot.ToString().c_str());
+            printf("----------------------------------------\n");
+            printf("These constants bring up a NEW chain. Do not paste them\n");
+            printf("into a build for this network to make an edited nTime or\n");
+            printf("pszTimestamp fit -- a node whose genesis differs starts\n");
+            printf("cleanly on a chain of its own and rejects every real peer.\n");
             printf("========================================\n");
         }
 
@@ -3263,15 +3418,61 @@ bool ProcessMessage(CNode* pfrom, string strCommand, CDataStream& vRecv)
         vRecv >> *pblock;
 
         //// debug print
-        if (LogAcceptsCategory("net")) { printf("received block:\n"); pblock->print(); }
+        if (LogAcceptsCategory("relay")) { printf("received block:\n"); pblock->print(); }
 
         CInv inv(MSG_BLOCK, pblock->GetHash());
         pfrom->AddInventoryKnown(inv);
 
+        // GetHash() covers the 80-byte header and nothing else, so the hash
+        // coming out right does not make this the block we asked for. Every
+        // peer that saw X announced holds X's header, and can answer our
+        // getdata with that header bolted onto a transaction list of its own
+        // invention -- the reply still hashes to X. hashMerkleRoot is the only
+        // thing binding the header to the body, so check it here, before the
+        // mapBlockFetch entry is touched and before CheckBlock spends a
+        // RandomX hash on a body already known to be forged.
+        //
+        // There is no honest way to arrive here. Peer traffic rides the .btf
+        // channel, which is XChaCha20-Poly1305, so a body that arrives at all
+        // arrived intact: a mismatch is the peer's own construction, not line
+        // noise. Drop the peer and leave the fetch outstanding so the failover
+        // sweep asks somebody else for the real X, rather than recording X as
+        // delivered on the strength of a header anyone could copy.
+        if (pblock->hashMerkleRoot != pblock->BuildMerkleTree())
+        {
+            if (LogAcceptsCategory("net"))
+                printf("peer %s served a forged body for %s\n",
+                       pfrom->addr.ToString().c_str(), inv.ToString().c_str());
+            // One line per connection: ProcessMessages keeps draining the
+            // receive buffer after fDisconnect is set, so a peer that packs
+            // many forged bodies into one buffer must not get one log line
+            // for each of them.
+            if (!pfrom->fDisconnect)
+                printf("block fetch: a peer answered the request for %s with a body that is not "
+                       "that block -- dropping it\n", inv.hash.ToString().substr(0,14).c_str());
+            pfrom->fDisconnect = true;
+            return error("ProcessMessage() : forged block body for %s",
+                         inv.hash.ToString().substr(0,14).c_str());
+        }
+
         if (ProcessBlock(pfrom, pblock.release()))
         {
             mapAlreadyAskedFor.erase(inv);
 
+            // The body arrived and validated, so the fetch really is over:
+            // the block is now in mapBlockIndex, or parked in mapOrphanBlocks,
+            // and either way AlreadyHave answers for it from here on.
+            //
+            // This has to stay inside the success branch, next to the erase
+            // above. When validation rejects a body nothing is stored,
+            // AlreadyHave stays false, and the mapAskFor entry was already
+            // consumed when the getdata went out -- so erasing the tracking
+            // here would leave the block requested from nobody at all until
+            // some peer happened to re-announce it. That is exactly the stall
+            // of #33, and erasing before ProcessBlock made it reachable on
+            // demand.
+            mapBlockFetch.erase(inv.hash);
+
             // A peer that just handed us a block is demonstrably at least at
             // that block's height. Without this, nStartingHeight is a snapshot
             // taken once during the version handshake and never touched again,
@@ -3396,12 +3597,119 @@ bool SendMessages(CNode* pto)
             const CInv& inv = (*pto->mapAskFor.begin()).second;
             if (LogAcceptsCategory("net")) printf("sending getdata: %s\n", inv.ToString().c_str());
             if (!AlreadyHave(txdb, inv))
+            {
                 vAskFor.push_back(inv);
+                // Blocks get their delivery tracked so the failover pass
+                // below can chase the ones that go quiet.
+                if (inv.type == MSG_BLOCK)
+                {
+                    // Full, and this is a hash we are not already tracking:
+                    // make room rather than drop it on the floor, or one peer
+                    // announcing eight hashes turns failover off for everyone.
+                    if (!mapBlockFetch.count(inv.hash) &&
+                        mapBlockFetch.size() >= BLOCK_FETCH_TRACK_MAX)
+                        MakeRoomInBlockFetch(pto->addr);
+
+                    if (mapBlockFetch.count(inv.hash) ||
+                        mapBlockFetch.size() < BLOCK_FETCH_TRACK_MAX)
+                    {
+                        CBlockFetch& fetch = mapBlockFetch[inv.hash];
+                        if (fetch.nTimeFirstAsked == 0)
+                        {
+                            fetch.nTimeFirstAsked = GetTime();
+                            // Who is answerable for this hash existing. Set
+                            // once, on creation: the failover sweep asks other
+                            // peers later, and none of them chose to announce
+                            // it, so none of them should be charged for it.
+                            fetch.addrAnnounced = pto->addr;
+                        }
+                        fetch.nTimeAsked = GetTime();
+                        fetch.setAsked.insert(pto->addr);
+                    }
+                }
+            }
             pto->mapAskFor.erase(pto->mapAskFor.begin());
         }
         if (!vAskFor.empty())
             pto->PushMessage("getdata", vAskFor);
 
+
+        //
+        // Message: getdata (block fetch failover)
+        //
+        // A requested block that has not arrived in BLOCK_FETCH_RETRY_SECS is
+        // not still in transit -- a body is at most 1 MB and somebody
+        // demonstrably has it -- so ask somebody else instead of waiting on
+        // whoever was asked first (#33). Each peer the message handler visits
+        // takes at most one turn per block per round; once every connected
+        // peer has had its turn and a further interval passes with nobody
+        // new to ask, the round restarts. A peer that declared itself a
+        // client cannot serve block bodies and is no use as a target.
+        if (!pto->fClient)
+        {
+            vector<CInv> vFailover;
+            int64 nNowSecs = GetTime();
+            for (map<uint256, CBlockFetch>::iterator mi = mapBlockFetch.begin(); mi != mapBlockFetch.end();)
+            {
+                CBlockFetch& fetch = (*mi).second;
+                CInv inv(MSG_BLOCK, (*mi).first);
+                if (AlreadyHave(txdb, inv))
+                {
+                    // Arrived by some other road -- our own miner, an orphan
+                    // parked in mapOrphanBlocks. Nothing left to chase.
+                    mapBlockFetch.erase(mi++);
+                    continue;
+                }
+                if (nNowSecs - fetch.nTimeFirstAsked > BLOCK_FETCH_GIVEUP_SECS)
+                {
+                    printf("block fetch: giving up on %s -- no peer served it in %d s\n",
+                           inv.ToString().c_str(), (int)(nNowSecs - fetch.nTimeFirstAsked));
+                    mapBlockFetch.erase(mi++);
+                    continue;
+                }
+                if (nNowSecs - fetch.nTimeAsked >= BLOCK_FETCH_RETRY_SECS)
+                {
+                    // The handler loop visits every peer within ~100ms, so a
+                    // request stale by two full intervals means nobody outside
+                    // setAsked was there to ask. Start the round over.
+                    if (nNowSecs - fetch.nTimeAsked >= 2 * BLOCK_FETCH_RETRY_SECS)
+                        fetch.setAsked.clear();
+                    if (!fetch.setAsked.count(pto->addr))
+                    {
+                        // Which block, how long, and which peer we are trying
+                        // next. Behind the category because it names a peer
+                        // and a peer decides how often it is written.
+                        if (LogAcceptsCategory("net"))
+                            printf("block fetch failover: %s unanswered for %d s, re-requesting from %s\n",
+                                   inv.ToString().c_str(), (int)(nNowSecs - fetch.nTimeAsked),
+                                   pto->addr.ToString().c_str());
+
+                        // The unconditional half: no addresses, and at most
+                        // one line per BLOCK_FETCH_REPORT_SECS however hard a
+                        // peer tries. See the counters for the reasoning.
+                        nBlockFetchFailovers++;
+                        if (nNowSecs - nBlockFetchWindowStart >= BLOCK_FETCH_REPORT_SECS)
+                        {
+                            printf("block fetch: %d block request%s went unanswered and %s been "
+                                   "re-asked of other peers (%d tracked)\n",
+                                   nBlockFetchFailovers, nBlockFetchFailovers == 1 ? "" : "s",
+                                   nBlockFetchFailovers == 1 ? "has" : "have",
+                                   (int)mapBlockFetch.size());
+                            nBlockFetchWindowStart = nNowSecs;
+                            nBlockFetchFailovers   = 0;
+                        }
+
+                        vFailover.push_back(inv);
+                        fetch.nTimeAsked = nNowSecs;
+                        fetch.setAsked.insert(pto->addr);
+                    }
+                }
+                ++mi;
+            }
+            if (!vFailover.empty())
+                pto->PushMessage("getdata", vFailover);
+        }
+
     }
     return true;
 }
@@ -3662,6 +3970,14 @@ static void ParticipantHashWorker(ParticipantJobShared* sh, int idx, int nWorker
     RandomXDestroyMinerVM(vm);
 }
 
+// Run one participant session: resolve the pool, open the tunnel, subscribe,
+// hash, submit shares. Returns true only if the pool actually handed out work
+// -- at least one well-formed mining.notify arrived. Everything else is false,
+// including a tunnel that opened and was dropped before any job.
+//
+// The caller's reconnect backoff keys off this return value, so any path added
+// here that leaves the session without work MUST return false. "The session
+// ended tidily" is not the question being asked.
 static bool PoolParticipantMiner()
 {
     // ------------------------------------------------------------------
@@ -3739,6 +4055,33 @@ static bool PoolParticipantMiner()
     }
     LogPrint("worker", "[worker] pool resolved to meeting node %s\n", meetingHostPort.c_str());
 
+    // The descriptor names ONE meeting node, but the pool parks a
+    // registration at every relay it knows (one accept thread per relay,
+    // rpc.cpp:AcceptOneFn), so any seed relay can pair us. Insisting on the
+    // advertised entry forever turns one bad path into a miner that idles
+    // indefinitely while a working path exists (issue #17); after a few
+    // consecutive tunnel failures, rotate the seed relays into the attempts
+    // too. Only ONE participant session runs (StartMinerThreads), so these
+    // statics are single-threaded.
+    static const int TUNNEL_FAILOVER_AFTER = 3;
+    static int nTunnelFails = 0;      // consecutive tunnel failures, this pool
+    static size_t iFailoverRelay = 0; // rotation cursor over vBtfMeetingRelays
+    static std::string strFailPool;   // pool the counters belong to
+    if (strFailPool != strParticipantPool) {
+        strFailPool = strParticipantPool;
+        nTunnelFails = 0;
+        iFailoverRelay = 0;
+    }
+    if (nTunnelFails >= TUNNEL_FAILOVER_AFTER && !vBtfMeetingRelays.empty()) {
+        std::string alt = vBtfMeetingRelays[iFailoverRelay++ % vBtfMeetingRelays.size()];
+        if (alt != meetingHostPort) {
+            LogPrint("worker", "[worker] advertised meeting node %s failed %d"
+                     " times in a row -- trying seed relay %s instead\n",
+                     meetingHostPort.c_str(), nTunnelFails, alt.c_str());
+            meetingHostPort = alt;
+        }
+    }
+
     size_t colon = meetingHostPort.rfind(':');
     if (colon == std::string::npos) {
         LogPrint("worker", "[worker] bad meeting node format '%s'\n", meetingHostPort.c_str());
@@ -3758,15 +4101,18 @@ static bool PoolParticipantMiner()
     // Step 2b: open encrypted tunnel to pool
     // ------------------------------------------------------------------
     SetParticipantMiningStatus("opening tunnel");
+    btf::TunnelFail tunnelFail = btf::TUNNEL_FAIL_NONE;
     s = btf::BtfClientTunnel(host.c_str(), (unsigned short)port,
-                             target_pubkey, service_enc_pub);
+                             target_pubkey, service_enc_pub, &tunnelFail);
     if (s == INVALID_SOCKET) {
-        LogPrint("worker", "[worker] tunnel connect FAILED via %s"
-                 " (relay unreachable or pool not listening)\n",
-                 meetingHostPort.c_str());
-        SetParticipantMiningStatus("tunnel connect failed");
+        nTunnelFails++;
+        LogPrint("worker", "[worker] tunnel connect FAILED via %s: %s\n",
+                 meetingHostPort.c_str(), btf::TunnelFailText(tunnelFail));
+        SetParticipantMiningStatus(std::string("tunnel connect failed: ")
+                                   + btf::TunnelFailText(tunnelFail));
         return false;
     }
+    nTunnelFails = 0;
     LogPrint("worker", "[worker] tunnel open via %s\n", meetingHostPort.c_str());
     SetParticipantMiningStatus("tunnel open");
     }
@@ -3899,9 +4245,32 @@ connected:
     std::vector<ParticipantShareHit> shareHits;
     std::mutex shareHitsMtx;
     std::vector<std::thread> hashThreads;
-    for (int i = 0; i < nWorkers; i++)
-        hashThreads.emplace_back(ParticipantHashWorker, &jobShared, i, nWorkers,
-                                 &shareHits, &shareHitsMtx);
+    try
+    {
+        for (int i = 0; i < nWorkers; i++)
+            hashThreads.emplace_back(ParticipantHashWorker, &jobShared, i, nWorkers,
+                                     &shareHits, &shareHitsMtx);
+    }
+    catch (const std::system_error& e)
+    {
+        // A refused pthread_create used to reach std::terminate here and
+        // kill the worker mid-session -- observed twice on an otherwise
+        // healthy participant. Stop the workers that did start the same way
+        // the disconnect path below does, drop this pool connection, and let
+        // the caller's reconnect loop try again once the pressure clears.
+        LogPrint("worker", "[worker] cannot spawn hash threads (%s)"
+                 " -- dropping pool connection\n", e.what());
+        SetParticipantMiningStatus("hash thread spawn failed");
+        {
+            std::lock_guard<std::mutex> lk(jobShared.mtx);
+            jobShared.stop = true;
+            jobShared.haveJob = false;
+        }
+        for (std::thread& t : hashThreads)
+            t.join();
+        closesocket(s);
+        return false;
+    }
     LogPrint("worker", "[worker] spawned %d hash threads on shared RandomX dataset\n",
              nWorkers);
     const uint64 sessionStartHashes = gParticipantHashes.load();
@@ -3912,8 +4281,17 @@ connected:
     std::string jobId;
     unsigned char header[80] = {};
     uint256 target;
+    // Cached answer to "which block is this job built on", for the
+    // mining.notify handler below -- see the comment there. Session-local, so
+    // a reconnect re-resolves rather than trusting an answer obtained before
+    // a possible change of pool.
+    uint256 hashPrevResolved;           // prevhash the cached answer describes
+    bool fPrevResolved       = false;   // whether that answer exists yet
+    int  nPrevHeight         = -1;      // its height, or -1 == unknown to us
+    int  nBestHeightResolved = -1;      // nBestHeight when we last resolved
     double currentDifficulty = 1.0; // updated by mining.set_difficulty from pool
     bool haveJob         = false;
+    bool gotAnyJob       = false; // any job at all this session -- the return value
     bool gotSubscribeAck = false;
     bool gotAuthorizeAck = false;
     int64 authWaitStart  = 0;
@@ -4030,8 +4408,9 @@ connected:
                                     jobId = newJobId;
                                     memcpy(header, hdrBytes.data(), 80);
                                     memcpy(&target, tgtBytes.data(), 32);
-                                    haveJob = true;
-                                    nNonce  = 1;
+                                    haveJob   = true;
+                                    gotAnyJob = true;
+                                    nNonce    = 1;
                                     {
                                         std::lock_guard<std::mutex> lk(jobShared.mtx);
                                         memcpy(jobShared.header, header, 80);
@@ -4040,8 +4419,64 @@ connected:
                                         jobShared.jobSeq++;
                                         jobShared.haveJob = true;
                                     }
-                                    LogPrint("worker", "[pool->worker] new job %s received\n",
-                                             jobId.c_str());
+                                    // Log the chain position the pool is
+                                    // building on, so a stale or private
+                                    // fork is auditable from worker logs.
+                                    // The prevhash comes from the header we
+                                    // actually hash (== p[2] of the notify);
+                                    // a brief cs_main lookup is safe here --
+                                    // this thread holds no other lock.
+                                    //
+                                    // Resolving it is the expensive half, and
+                                    // this is the socket-reading thread: block
+                                    // validation can hold cs_main for a long
+                                    // time, and every moment spent waiting on
+                                    // it here is a moment set_difficulty and
+                                    // share responses are not being handled.
+                                    // Most notifies in a round are a new job
+                                    // on the SAME tip, so the answer is cached
+                                    // and re-resolved only when it could come
+                                    // out differently: a prevhash we have not
+                                    // resolved, or one we could NOT resolve
+                                    // last time while the chain has since
+                                    // moved (unknown -> known). A block's
+                                    // height is fixed once mapBlockIndex knows
+                                    // it, so a cached "known" never goes
+                                    // stale. nBestHeight is read unlocked on
+                                    // purpose -- it only decides whether to
+                                    // repeat a lookup, never what gets
+                                    // printed, so a stale read costs at most
+                                    // one deferred or one extra lookup.
+                                    // The line still prints in full for every
+                                    // job, cached or not.
+                                    uint256 hashPrev;
+                                    memcpy(&hashPrev, header + 4, 32);
+                                    if (LogAcceptsCategory("worker"))
+                                    {
+                                        int nBestHeightNow = nBestHeight;
+                                        if (!fPrevResolved
+                                            || hashPrev != hashPrevResolved
+                                            || (nPrevHeight < 0 && nBestHeightNow != nBestHeightResolved))
+                                        {
+                                            nPrevHeight = -1;
+                                            CRITICAL_BLOCK(cs_main)
+                                            {
+                                                map<uint256, CBlockIndex*>::iterator mi = mapBlockIndex.find(hashPrev);
+                                                if (mi != mapBlockIndex.end())
+                                                    nPrevHeight = (*mi).second->nHeight;
+                                            }
+                                            hashPrevResolved    = hashPrev;
+                                            nBestHeightResolved = nBestHeightNow;
+                                            fPrevResolved       = true;
+                                        }
+                                        if (nPrevHeight >= 0)
+                                            LogPrint("worker", "[pool->worker] new job %s received, prevhash %s (known, job height %d)\n",
+                                                     jobId.c_str(), hashPrev.GetHex().c_str(),
+                                                     nPrevHeight + 1);
+                                        else
+                                            LogPrint("worker", "[pool->worker] new job %s received, prevhash %s (unknown to our chain)\n",
+                                                     jobId.c_str(), hashPrev.GetHex().c_str());
+                                    }
                                     SetParticipantMiningStatus("hashing job " + jobId);
                                 }
                             }
@@ -4167,7 +4602,11 @@ connected:
              (unsigned long long)hashesSinceSample,
              (unsigned long long)sessionSharesSent,
              (unsigned long long)sessionSharesAccepted);
-    return true;
+    // Not "we got here", which is true of a tunnel the pool accepted and then
+    // dropped without ever sending a job -- exactly the case the caller's
+    // backoff exists for. Reaching the end of the session is not evidence the
+    // session was worth anything; a job is.
+    return gotAnyJob;
 }
 
 // ---------------------------------------------------------------------------
@@ -4364,11 +4803,51 @@ bool BitcoinMiner(int nThreadId)
     // Participant mode: connect to external pool instead of mining solo/operator
     if (nMineMode == MINE_PARTICIPANT)
     {
+        // An attempt that never got work backs off (10s doubling to 120s)
+        // instead of hammering the same path every flat 10s, and a streak
+        // lasting minutes is said at printf level: the identical once-per-10s
+        // LogPrint line is how a worker idled all night unnoticed (issue #17).
+        //
+        // Only a session the pool actually sent a job to resets either. The
+        // weaker test -- "the session returned" -- was gameable: a pool that
+        // completes the tunnel and drops without ever sending work leaves by
+        // the same tidy cleanup path a healthy one does, which would pin the
+        // retry at 10s forever and defeat the backoff in exactly the scenario
+        // it exists for.
+        int   nRetrySecs    = 10;
+        int64 nFailingSince = 0;
         while (fGenerateBitcoins && nMineMode == MINE_PARTICIPANT && !fShutdown) {
-            PoolParticipantMiner();
+            bool fGotWork = PoolParticipantMiner();
             if (fGenerateBitcoins && nMineMode == MINE_PARTICIPANT && !fShutdown) {
-                LogPrint("worker", "PoolParticipantMiner: reconnecting in 10s\n");
-                for (int i = 0; i < 10 && !fShutdown; i++) Sleep(1000);
+                if (fGotWork) {
+                    nRetrySecs    = 10;
+                    nFailingSince = 0;
+                } else {
+                    if (nFailingSince == 0)
+                        nFailingSince = GetTime();
+                    int64 nMins = (GetTime() - nFailingSince) / 60;
+                    if (nMins >= 1)
+                        printf("[worker] connection to pool %s has been failing"
+                               " for %lld minutes\n",
+                               strParticipantPool.c_str(), (long long)nMins);
+                }
+                LogPrint("worker", "PoolParticipantMiner: reconnecting in %ds\n",
+                         nRetrySecs);
+                // Poll everything the enclosing while polls, not just
+                // fShutdown. Raising the ceiling to 120s made a
+                // shutdown-only predicate expensive: after the operator
+                // clears Generate or switches to Relay, the session -- and
+                // the vfThreadRunning[3] slot that gates restarting it --
+                // would stay held for up to two minutes. Measured 119s
+                // before, 1s after.
+                for (int i = 0; i < nRetrySecs && !fShutdown && fGenerateBitcoins
+                                && nMineMode == MINE_PARTICIPANT; i++)
+                    Sleep(1000);
+                if (!fGotWork && nRetrySecs < 120) {
+                    nRetrySecs *= 2;
+                    if (nRetrySecs > 120)
+                        nRetrySecs = 120;
+                }
             }
         }
         return true;
@@ -4555,7 +5034,7 @@ bool BitcoinMiner(int nThreadId)
                     //// debug print
                     if (LogAcceptsCategory("net")) printf("BitcoinMiner:\n");
                     if (LogAcceptsCategory("net")) printf("RandomX proof-of-work found  \n  powhash: %s  \ntarget: %s\n", hash.GetHex().c_str(), hashTarget.GetHex().c_str());
-                    if (LogAcceptsCategory("net")) pblock->print();
+                    if (LogAcceptsCategory("relay")) pblock->print();
 
                 SetThreadPriority(GetCurrentThread(), THREAD_PRIORITY_NORMAL);
                 CRITICAL_BLOCK(cs_main)