vlib: add postmortem pcap dispatch trace

Inspired by a real-life conundrum: scenario X involves a vpp crash in
ip4-load-balance because vnet_buffer(b)->ip.adj_index[VLIB_TX] is
(still) set to ~0.

The problem takes most of a day to occur, and we need to see the
broken packet's graph trajectory, metadata, etc. to understand the
problem.

Fix a signed/unsigned ASSERT bug in vlib_get_trace_count().

Rename elog_post_mortem_dump() -> vlib_post_mortem_dump(), add
dispatch trace post-mortem dump.

Add FILTER_FLAG_POST_MORTEM so we can (putatively) capture a ludicrous
number of buffer traces, without actually using more than one dispatch
cycle's worth of memory.

Type: improvement

Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: If093202ef071df46e290370bd9b33bf6560d30e6

1 files changed, 9 insertions, 0 deletions

diff --git a/src/vlib/trace.c b/src/vlib/trace.c
index abd116622c7..f90f275fa87 100644
--- a/src/vlib/trace.c
+++ b/src/vlib/trace.c
@@ -201,6 +201,15 @@ filter_accept (vlib_trace_main_t * tm, vlib_trace_header_t * h)
   if (tm->filter_flag == 0)
     return 1;
 
+  /*
+   * When capturing a post-mortem dispatch trace,
+   * toss all existing traces once per dispatch cycle.
+   * So we can trace 4 billion pkts without running out of
+   * memory...
+   */
+  if (tm->filter_flag == FILTER_FLAG_POST_MORTEM)
+    return 0;
+
   if (tm->filter_flag == FILTER_FLAG_INCLUDE)
     {
       while (h < e)