diff --git a/gc_bba_test/source/netboot.c b/gc_bba_test/source/netboot.c
index aafe906..871ba66 100644
--- a/gc_bba_test/source/netboot.c
+++ b/gc_bba_test/source/netboot.c
@@ -3,6 +3,7 @@
 #include <stdio.h>
 #include <string.h>
 #include <malloc.h>
+#include <unistd.h>
 #include <ogc/lwp_watchdog.h>
 #include <ogc/cache.h>
 #include <fat.h>
@@ -103,16 +104,42 @@ int netboot_receive(void *dst, u32 maxlen, void (*status)(const char *msg))
 	u64 t0;
 	int have_peer = 0;
 
-	if (bba_read(REG_PAR0, s_our_mac, 6) < 0) return -1;
-
-	/* Make sure RX is actually enabled -- a prior test may have left the
-	 * NCRA SR bit clear, and then this would wait forever for frames the
-	 * gateware is not storing. */
+	/* Bring the device up deterministically rather than hoping someone else
+	 * already did.  Pressing Z straight after launch is the normal case, and
+	 * then NOTHING has issued the NCRA reset that triggers the W5100's own
+	 * init -- MACRAW is not running, no frames are stored, and this would sit
+	 * in "timeout: no frames" for a reason that has nothing to do with the
+	 * transfer.  Swiss may have initialised it, or an earlier test may have,
+	 * or neither; a reset here makes all three cases identical.
+	 *
+	 * The reset also clears the SPRAM ring pointers, which is what we want:
+	 * a clean ring with RRP == RWP == page 1, so the first frame we see is
+	 * genuinely the host's first block. */
+	if (status) status("resetting the adapter...");
+	if (bba_w8(REG_NCRA, NCRA_RESET) < 0) return -1;
 	{
-		int ncra = bba_r8(REG_NCRA);
-		if (ncra >= 0 && !(ncra & NCRA_SR))
-			bba_w8(REG_NCRA, (u8)(ncra | NCRA_SR));
+		int i, v = -1;
+		for (i = 0; i < 50; i++) {
+			v = bba_r8(REG_NCRA);
+			if (v >= 0 && !(v & NCRA_RESET)) break;
+			usleep(5000);
+		}
+		if (i == 50) {
+			if (status) status("NCRA reset never self-cleared");
+			return -1;
+		}
+		/* The W5100 re-inits behind that: RSTn pulse plus a 60.3 ms settle,
+		 * done twice.  Polling it sooner reads all-zeros and looks dead. */
+		usleep(400000);
 	}
+	/* Start receive.  Without the SR bit the assembler never writes the
+	 * ring, so frames arrive at the W5100 and go nowhere. */
+	if (bba_w8(REG_NCRA, NCRA_SR) < 0) return -1;
+
+	/* PAR0-5 survives the reset (it is the gateware's power-on MAC), and it
+	 * is read AFTER the reset so the source address we put in ACKs is the
+	 * one the adapter is actually filtering on. */
+	if (bba_read(REG_PAR0, s_our_mac, 6) < 0) return -1;
 
 	if (status) status("waiting for host (send-dol.py)...");
 	t0 = gettime();