vlib: address vlib_error_t scaling issuestable/1901

Encoding the vpp node index into the vlib_error_t as a 10-bit quantity
limits us to 1K graph nodes. Unfortunately, a few nodes need 6 bit
per-node error codes. Only a very few nodes have so many counters.

It turns out that there are about 2K total error counters in the system,
which is (approximately) the maximum error heap index.

The current (index,code) encoding limits the number of interfaces to
around 250, since each interface has two associated graph nodes and we
have about 500 "normal, interior" graph node

This patch adds an error-index to node-index map, so we can store
error heap indices directly in the vlib_buffer_t.

Type: refactor

Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: Ic2e91a5b344c9df3b98b264cacda246e06092a94

7 files changed, 48 insertions, 51 deletions

diff --git a/src/plugins/map/map.c b/src/plugins/map/map.c
index 7a44a1905d4..08c67d0ac8d 100644
--- a/src/plugins/map/map.c
+++ b/src/plugins/map/map.c
@@ -1034,7 +1034,7 @@ map_error_counter_get (u32 node_index, map_error_t map_error)
   vlib_node_t *n = vlib_get_node (vm, node_index);
   u32 ci;
 
-  ci = vlib_error_get_code (e);
+  ci = vlib_error_get_code (&vm->node_main, e);
   ASSERT (ci < n->n_errors);
   ci += n->error_heap_index;
 
diff --git a/src/vlib/error.c b/src/vlib/error.c
index 9bd928c6d0f..a0ce0b7eb96 100644
--- a/src/vlib/error.c
+++ b/src/vlib/error.c
@@ -51,8 +51,10 @@ vlib_error_drop_buffers (vlib_main_t * vm,
 {
   u32 n_left_this_frame, n_buffers_left, *args, n_args_left;
   vlib_error_t drop_error;
+  vlib_node_t *n;
 
-  drop_error = vlib_error_set (drop_error_node, drop_error_code);
+  n = vlib_get_node (vm, drop_error_node);
+  drop_error = n->error_heap_index + drop_error_code;
 
   n_buffers_left = n_buffers;
   while (n_buffers_left > 0)
@@ -161,6 +163,8 @@ vlib_register_errors (vlib_main_t * vm,
 		      u32 node_index, u32 n_errors, char *error_strings[])
 {
   vlib_error_main_t *em = &vm->error_main;
+  vlib_node_main_t *nm = &vm->node_main;
+
   vlib_node_t *n = vlib_get_node (vm, node_index);
   uword l;
   void *oldheap;
@@ -220,19 +224,21 @@ vlib_register_errors (vlib_main_t * vm,
   /* (re)register the em->counters base address, switch back to main heap */
   vlib_stats_pop_heap2 (em->counters, vm->thread_index, oldheap);
 
-  if (VLIB_ELOG_MAIN_LOOP > 0)
-    {
-      elog_event_type_t t;
-      uword i;
+  {
+    elog_event_type_t t;
+    uword i;
 
-      clib_memset (&t, 0, sizeof (t));
-      for (i = 0; i < n_errors; i++)
-	{
-	  t.format = (char *) format (0, "%v %s: %%d",
-				      n->name, error_strings[i]);
-	  vm->error_elog_event_types[n->error_heap_index + i] = t;
-	}
-    }
+    clib_memset (&t, 0, sizeof (t));
+    if (n_errors > 0)
+      vec_validate (nm->node_by_error, n->error_heap_index + n_errors - 1);
+    for (i = 0; i < n_errors; i++)
+      {
+	t.format = (char *) format (0, "%v %s: %%d",
+				    n->name, error_strings[i]);
+	vm->error_elog_event_types[n->error_heap_index + i] = t;
+	nm->node_by_error[n->error_heap_index + i] = n->index;
+      }
+  }
 }
 
 static clib_error_t *
diff --git a/src/vlib/error.h b/src/vlib/error.h
index 58352519be2..0da3a18d85d 100644
--- a/src/vlib/error.h
+++ b/src/vlib/error.h
@@ -40,38 +40,8 @@
 #ifndef included_vlib_error_h
 #define included_vlib_error_h
 
-/* Combined 16 bit node & 16 bit code as 32 bit number. */
 typedef u16 vlib_error_t;
 
-always_inline u32
-vlib_error_get_node (vlib_error_t e)
-{
-  return e >> 6;
-}
-
-always_inline u32
-vlib_error_get_code (vlib_error_t e)
-{
-  return e & 0x3f;
-}
-
-always_inline vlib_error_t
-vlib_error_set (u32 node_index, u32 code)
-{
-  ASSERT (node_index < (1 << 10));
-  ASSERT (code < (1 << 6));
-  return (node_index << 6) | code;
-}
-
-always_inline vlib_error_t
-vlib_error_set_code (vlib_error_t e, u32 code)
-{
-  ASSERT (vlib_error_get_code (e) == 0);
-  ASSERT (code < (1 << 6));
-  e |= code;
-  return e;
-}
-
 typedef struct
 {
   /* Error counters. */
diff --git a/src/vlib/node.c b/src/vlib/node.c
index 69f505143b2..97ab0444b1b 100644
--- a/src/vlib/node.c
+++ b/src/vlib/node.c
@@ -502,7 +502,7 @@ register_node (vlib_main_t * vm, vlib_node_registration_t * r)
 
     vec_resize (rt->errors, r->n_errors);
     for (i = 0; i < vec_len (rt->errors); i++)
-      rt->errors[i] = vlib_error_set (n->index, i);
+      rt->errors[i] = n->error_heap_index + i;
 
     STATIC_ASSERT_SIZEOF (vlib_node_runtime_t, 128);
     ASSERT (vec_len (n->runtime_data) <= VLIB_NODE_RUNTIME_DATA_SIZE);
diff --git a/src/vlib/node.h b/src/vlib/node.h
index 011709b9705..38ae4ec23c0 100644
--- a/src/vlib/node.h
+++ b/src/vlib/node.h
@@ -784,8 +784,27 @@ typedef struct
 
   /* Node registrations added by constructors */
   vlib_node_registration_t *node_registrations;
+
+  /* Node index from error code */
+  u32 *node_by_error;
 } vlib_node_main_t;
 
+typedef u16 vlib_error_t;
+
+always_inline u32
+vlib_error_get_node (vlib_node_main_t * nm, vlib_error_t e)
+{
+  return nm->node_by_error[e];
+}
+
+always_inline u32
+vlib_error_get_code (vlib_node_main_t * nm, vlib_error_t e)
+{
+  u32 node_index = nm->node_by_error[e];
+  vlib_node_t *n = nm->nodes[node_index];
+  u32 error_code = e - n->error_heap_index;
+  return error_code;
+}
 
 #define FRAME_QUEUE_MAX_NELTS 32
 typedef struct
diff --git a/src/vlib/threads.c b/src/vlib/threads.c
index 56f9d790ee6..5bd94760456 100644
--- a/src/vlib/threads.c
+++ b/src/vlib/threads.c
@@ -856,6 +856,7 @@ start_workers (vlib_main_t * vm)
 	      /* zap the (per worker) frame freelists, etc */
 	      nm_clone->frame_sizes = 0;
 	      nm_clone->frame_size_hash = hash_create (0, sizeof (uword));
+	      nm_clone->node_by_error = nm->node_by_error;
 
 	      /* Packet trace buffers are guaranteed to be empty, nothing to do here */
 
@@ -1175,6 +1176,7 @@ vlib_worker_thread_node_refork (void)
 
   nm_clone->processes = vec_dup_aligned (nm->processes,
 					 CLIB_CACHE_LINE_BYTES);
+  nm_clone->node_by_error = nm->node_by_error;
 }
 
 void
diff --git a/src/vnet/interface_output.c b/src/vnet/interface_output.c
index ba69d36b6c4..db4949d5afc 100644
--- a/src/vnet/interface_output.c
+++ b/src/vnet/interface_output.c
@@ -562,10 +562,10 @@ counter_index (vlib_main_t * vm, vlib_error_t e)
   vlib_node_t *n;
   u32 ci, ni;
 
-  ni = vlib_error_get_node (e);
+  ni = vlib_error_get_node (&vm->node_main, e);
   n = vlib_get_node (vm, ni);
 
-  ci = vlib_error_get_code (e);
+  ci = vlib_error_get_code (&vm->node_main, e);
   ASSERT (ci < n->n_errors);
 
   ci += n->error_heap_index;
@@ -583,8 +583,8 @@ format_vnet_error_trace (u8 * s, va_list * va)
   vlib_error_main_t *em = &vm->error_main;
   u32 i;
 
-  error_node = vlib_get_node (vm, vlib_error_get_node (e[0]));
-  i = counter_index (vm, e[0]);
+  error_node = vlib_get_node (vm, vlib_error_get_node (&vm->node_main, e[0]));
+  i = counter_index (vm, vlib_error_get_code (&vm->node_main, e[0]));
   s = format (s, "%v: %s", error_node->name, em->error_strings_heap[i]);
 
   return s;
@@ -652,8 +652,8 @@ trace_errors_with_buffers (vlib_main_t * vm,
 static u8 *
 validate_error (vlib_main_t * vm, vlib_error_t * e, u32 index)
 {
-  uword node_index = vlib_error_get_node (e[0]);
-  uword code = vlib_error_get_code (e[0]);
+  uword node_index = vlib_error_get_node (&vm->node_main, e[0]);
+  uword code = vlib_error_get_code (&vm->node_main, e[0]);
   vlib_node_t *n;
 
   if (node_index >= vec_len (vm->node_main.nodes))