Index: sys/net/iflib.c
===================================================================
--- sys/net/iflib.c
+++ sys/net/iflib.c
@@ -188,6 +188,8 @@
 	uint16_t ifc_sysctl_qs_eq_override;
 	uint16_t ifc_sysctl_rx_budget;
 	uint16_t ifc_sysctl_tx_abdicate;
+	uint16_t ifc_sysctl_core_offset;
+	uint8_t  ifc_sysctl_separate_txrx;
 
 	qidx_t ifc_sysctl_ntxds[8];
 	qidx_t ifc_sysctl_nrxds[8];
@@ -725,6 +727,16 @@
 static struct mbuf * iflib_fixup_rx(struct mbuf *m);
 #endif
 
+static SLIST_HEAD(cpu_offset_list, cpu_offset) cpu_offsets =
+    SLIST_HEAD_INITIALIZER(cpu_offsets);
+struct cpu_offset {
+	SLIST_ENTRY(cpu_offset) entries;
+	cpuset_t	set;
+	uint16_t	offset;
+};
+static struct mtx cpu_offset_mtx;
+MTX_SYSINIT(iflib_cpu_offset, &cpu_offset_mtx, "iflib_cpu_offset lock", MTX_DEF);
+
 NETDUMP_DEFINE(iflib);
 
 #ifdef DEV_NETMAP
@@ -4366,6 +4378,51 @@
 	}
 }
 
+static uint16_t get_ctx_core_offset(if_ctx_t ctx)
+{
+	if_softc_ctx_t scctx = &ctx->ifc_softc_ctx;
+	uint16_t qc;
+	struct cpu_offset *op;
+	uint16_t ret = ctx->ifc_sysctl_core_offset;
+	char kename[128];
+
+	/* XXX This is terrible... */
+	sprintf(kename, "dev.%s.%d.iflib.core_offset",
+	    device_get_name(ctx->ifc_dev), device_get_unit(ctx->ifc_dev));
+	if (testenv(kename) == 0 && ret == 0)
+		ret = 0xffff;
+	if (ret != 0xffff)
+		return ret;
+
+	if (ctx->ifc_sysctl_separate_txrx)
+		qc = scctx->isc_ntxqsets + scctx->isc_nrxqsets;
+	else
+		qc = max(scctx->isc_ntxqsets, scctx->isc_nrxqsets);
+
+	mtx_lock(&cpu_offset_mtx);
+	SLIST_FOREACH(op, &cpu_offsets, entries) {
+		if (CPU_CMP(&ctx->ifc_cpus, &op->set) == 0) {
+			ret = op->offset;
+			op->offset += qc;
+			break;
+		}
+	}
+	if (ret == 0xffff) {
+		ret = 0;
+		op = malloc(sizeof(struct cpu_offset), M_IFLIB,
+		    M_NOWAIT | M_ZERO);
+		MPASS(op != NULL);
+		if (op != NULL) {
+			op->offset = qc;
+			CPU_COPY(&ctx->ifc_cpus, &op->set);
+			SLIST_INSERT_HEAD(&cpu_offsets, op, entries);
+		}
+	}
+	mtx_unlock(&cpu_offset_mtx);
+
+	return ret;
+}
+
 int
 iflib_device_register(device_t dev, void *sc, if_shared_ctx_t sctx, if_ctx_t *ctxp)
 {
@@ -4518,6 +4575,11 @@
 		goto fail_queues;
 
 	/*
+	 * Now that we know how many queues there are, get the core offset.
+	 */
+	ctx->ifc_sysctl_core_offset = get_ctx_core_offset(ctx);
+
+	/*
 	 * Group taskqueues aren't properly set up until SMP is started,
 	 * so we disable interrupts until we can handle them post
 	 * SI_SUB_SMP.
@@ -5558,7 +5620,7 @@
  * Find the nth "close" core to the specified core
  * "close" is defined as the deepest level that shares
  * at least an L2 cache.  With threads, this will be
- * threads on the same core.  If the sahred cache is L3
+ * threads on the same core.  If the shared cache is L3
  * or higher, simply returns the same core.
  */
 static int
@@ -5642,10 +5704,13 @@
     const char *name)
 {
 	device_t dev;
-	int err, cpuid, tid;
+	int co, cpuid, err, tid;
 
 	dev = ctx->ifc_dev;
-	cpuid = find_nth(ctx, qid);
+	co = ctx->ifc_sysctl_core_offset;
+	if (ctx->ifc_sysctl_separate_txrx && type == IFLIB_INTR_TX)
+		co += ctx->ifc_softc_ctx.isc_nrxqsets;
+	cpuid = find_nth(ctx, qid + co);
 	tid = get_core_offset(ctx, type, qid);
 	MPASS(tid >= 0);
 	cpuid = find_close_core(cpuid, tid);
@@ -6247,6 +6312,12 @@
 	SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "tx_abdicate",
 		       CTLFLAG_RWTUN, &ctx->ifc_sysctl_tx_abdicate, 0,
 		       "cause tx to abdicate instead of running to completion");
+	SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "core_offset",
+		       CTLFLAG_RDTUN, &ctx->ifc_sysctl_core_offset, 0,
+		       "offset to start using cores at");
+	SYSCTL_ADD_U8(ctx_list, oid_list, OID_AUTO, "separate_txrx",
+		       CTLFLAG_RDTUN, &ctx->ifc_sysctl_separate_txrx, 0,
+		       "use separate cores for TX and RX");
 
 	/* XXX change for per-queue sizes */
 	SYSCTL_ADD_PROC(ctx_list, oid_list, OID_AUTO, "override_ntxds",