Add low level event recording support

Add low level event recording infrastructure for system and user
defined events.  The infrastructure is able to record high frequency
events such as

 * SMP lock acquire/release,
 * interrupt entry/exit,
 * thread switches,
 * UMA zone allocate/free, and
 * Ethernet packet input/output, etc.

It allows post-mortem analysis in fatal error handlers, e.g. the last
events are in the record buffer, the newest event overwrites the oldest
event.  It is possible to detect record buffer overflows for consumers
that expect a continuous stream of events, e.g. to display the system
state in real-time.

The implementation supports high-end SMP machines (more than 1GHz
processor frequency, more than four processors).

Add a new API instead. The implementation uses per-processor data
structures and no atomic read-modify-write operations.  It is uses
per-processor ring buffers to record the events.

The CPU counter is used to get the time of events. It is combined with
periodic uptime events to synchronize it with CLOCK_REALTIME.

The existing capture engine tries to solve this problem also, but its
performance is not good enough for high-end production systems.  The
main issues are the variable-size buffers and the use of SMP locks for
synchronization.  To fix this, the API would change significantly.

Update #3665.

1 files changed, 448 insertions, 0 deletions

diff --git a/cpukit/libtrace/record/record-client.c b/cpukit/libtrace/record/record-client.c
new file mode 100644
index 0000000000..585aa3895d
--- /dev/null
+++ b/cpukit/libtrace/record/record-client.c
@@ -0,0 +1,448 @@
+/*
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright (C) 2018, 2019 embedded brains GmbH
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/*
+ * This file must be compatible to general purpose POSIX system, e.g. Linux,
+ * FreeBSD.  It may be used for utility programs.
+ */
+
+#if HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <rtems/recordclient.h>
+
+#include <string.h>
+
+static void set_to_bt_scaler(
+  rtems_record_client_context *ctx,
+  uint32_t                     frequency
+)
+{
+  uint64_t bin_per_s;
+
+  bin_per_s = UINT64_C( 1 ) << 32;
+  ctx->to_bt_scaler = ( ( bin_per_s << 31 ) + frequency - 1 ) / frequency;
+}
+
+static rtems_record_client_status call_handler(
+  const rtems_record_client_context *ctx,
+  uint64_t                           bt,
+  rtems_record_event                 event,
+  uint64_t                           data
+)
+{
+  uint32_t seconds;
+  uint32_t nanosec;
+
+  seconds = (uint32_t) ( bt >> 32 );
+  nanosec = (uint32_t) ( ( UINT64_C( 1000000000 ) * (uint32_t) bt ) >> 32 );
+
+  return ( *ctx->handler )(
+    seconds,
+    nanosec,
+    ctx->cpu,
+    event,
+    data,
+    ctx->handler_arg
+  );
+}
+
+static void check_overflow(
+  const rtems_record_client_context *ctx,
+  const rtems_record_client_per_cpu *per_cpu,
+  uint32_t                           new_head
+)
+{
+  uint32_t last_tail;
+  uint32_t last_head;
+  uint32_t capacity;
+  uint32_t new_content;
+  uint64_t bt;
+
+  last_tail = per_cpu->tail[ per_cpu->index ];
+  last_head = per_cpu->head[ per_cpu->index ];
+
+  if ( last_tail == last_head ) {
+    return;
+  }
+
+  capacity = ( last_tail - last_head - 1 ) & ( ctx->count - 1 );
+  new_content = new_head - last_head;
+
+  if ( new_content <= capacity ) {
+    return;
+  }
+
+  bt = ( per_cpu->uptime.time_accumulated * ctx->to_bt_scaler ) >> 31;
+  bt += per_cpu->uptime.bt;
+
+  call_handler(
+    ctx,
+    bt,
+    RTEMS_RECORD_OVERFLOW,
+    new_content - capacity
+  );
+}
+
+static rtems_record_client_status visit( rtems_record_client_context *ctx )
+{
+  rtems_record_client_per_cpu *per_cpu;
+  uint32_t                     time;
+  rtems_record_event           event;
+  uint64_t                     data;
+  uint64_t                     bt;
+
+  per_cpu = &ctx->per_cpu[ ctx->cpu ];
+  time = RTEMS_RECORD_GET_TIME( ctx->event );
+  event = RTEMS_RECORD_GET_EVENT( ctx->event );
+  data = ctx->data;
+
+  switch ( event ) {
+    case RTEMS_RECORD_PROCESSOR:
+      if ( data >= RTEMS_RECORD_CLIENT_MAXIMUM_CPU_COUNT ) {
+        return RTEMS_RECORD_CLIENT_ERROR_UNSUPPORTED_CPU;
+      }
+
+      ctx->cpu = (uint32_t) data;
+      per_cpu = &ctx->per_cpu[ ctx->cpu ];
+      break;
+    case RTEMS_RECORD_UPTIME_LOW:
+      per_cpu->uptime.bt = (uint32_t) data;
+      per_cpu->uptime.time_at_bt = time;
+      per_cpu->uptime.time_last = time;
+      per_cpu->uptime.time_accumulated = 0;
+      time = 0;
+      break;
+    case RTEMS_RECORD_UPTIME_HIGH:
+      per_cpu->uptime.bt += (int64_t) data << 32;
+      time = 0;
+      break;
+    case RTEMS_RECORD_TAIL:
+      per_cpu->tail[ per_cpu->index ] = (uint32_t) data;
+      break;
+    case RTEMS_RECORD_HEAD:
+      per_cpu->head[ per_cpu->index ]= (uint32_t) data;
+      per_cpu->index ^= 1;
+      check_overflow( ctx, per_cpu, (uint32_t) data );
+      break;
+    case RTEMS_RECORD_COUNT:
+      ctx->count = (uint32_t) data;
+      break;
+    case RTEMS_RECORD_FREQUENCY:
+      set_to_bt_scaler( ctx, (uint32_t) data );
+      break;
+    case RTEMS_RECORD_VERSION:
+      if ( data != RTEMS_RECORD_THE_VERSION ) {
+        return RTEMS_RECORD_CLIENT_ERROR_UNSUPPORTED_VERSION;
+      }
+
+      break;
+    default:
+      break;
+  }
+
+  if ( time != 0 ) {
+    uint32_t delta;
+
+    delta = ( time - per_cpu->uptime.time_last )
+      & ( ( UINT32_C( 1 ) << RTEMS_RECORD_TIME_BITS ) - 1 );
+    per_cpu->uptime.time_last = time;
+    per_cpu->uptime.time_accumulated += delta;
+    bt = ( per_cpu->uptime.time_accumulated * ctx->to_bt_scaler ) >> 31;
+    bt += per_cpu->uptime.bt;
+  } else {
+    bt = 0;
+  }
+
+  return call_handler( ctx, bt, event, data );
+}
+
+static rtems_record_client_status consume_32(
+  rtems_record_client_context *ctx,
+  const void                  *buf,
+  size_t                       n
+)
+{
+  while ( n > 0 ) {
+    size_t m;
+    char *pos;
+
+    m = ctx->todo < n ? ctx->todo : n;
+    pos = ctx->pos;
+    pos = memcpy( pos, buf, m );
+    n -= m;
+    buf = (char *) buf + m;
+
+    if ( m == ctx->todo ) {
+      rtems_record_client_status status;
+
+      ctx->todo = sizeof( ctx->item.format_32 );
+      ctx->pos = &ctx->item.format_32;
+      ctx->event = ctx->item.format_32.event;
+      ctx->data = ctx->item.format_32.data;
+
+      status = visit( ctx );
+
+      if ( status != RTEMS_RECORD_CLIENT_SUCCESS ) {
+        return status;
+      }
+    } else {
+      ctx->todo -= m;
+      ctx->pos = pos + m;
+    }
+  }
+
+  return RTEMS_RECORD_CLIENT_SUCCESS;
+}
+
+static rtems_record_client_status consume_64(
+  rtems_record_client_context *ctx,
+  const void                  *buf,
+  size_t                       n
+)
+{
+  while ( n > 0 ) {
+    size_t m;
+    char *pos;
+
+    m = ctx->todo < n ? ctx->todo : n;
+    pos = ctx->pos;
+    pos = memcpy( pos, buf, m );
+    n -= m;
+    buf = (char *) buf + m;
+
+    if ( m == ctx->todo ) {
+      rtems_record_client_status status;
+
+      ctx->todo = sizeof( ctx->item.format_64 );
+      ctx->pos = &ctx->item.format_64;
+      ctx->event = ctx->item.format_64.event;
+      ctx->data = ctx->item.format_64.data;
+
+      status = visit( ctx );
+
+      if ( status != RTEMS_RECORD_CLIENT_SUCCESS ) {
+        return status;
+      }
+    } else {
+      ctx->todo -= m;
+      ctx->pos = pos + m;
+    }
+  }
+
+  return RTEMS_RECORD_CLIENT_SUCCESS;
+}
+
+static rtems_record_client_status consume_swap_32(
+  rtems_record_client_context *ctx,
+  const void                  *buf,
+  size_t                       n
+)
+{
+  while ( n > 0 ) {
+    size_t m;
+    char *pos;
+
+    m = ctx->todo < n ? ctx->todo : n;
+    pos = ctx->pos;
+    pos = memcpy( pos, buf, m );
+    n -= m;
+    buf = (char *) buf + m;
+
+    if ( m == ctx->todo ) {
+      rtems_record_client_status status;
+
+      ctx->todo = sizeof( ctx->item.format_32 );
+      ctx->pos = &ctx->item.format_32;
+      ctx->event = __builtin_bswap32( ctx->item.format_32.event );
+      ctx->data = __builtin_bswap32( ctx->item.format_32.data );
+
+      status = visit( ctx );
+
+      if ( status != RTEMS_RECORD_CLIENT_SUCCESS ) {
+        return status;
+      }
+    } else {
+      ctx->todo -= m;
+      ctx->pos = pos + m;
+    }
+  }
+
+  return RTEMS_RECORD_CLIENT_SUCCESS;
+}
+
+static rtems_record_client_status consume_swap_64(
+  rtems_record_client_context *ctx,
+  const void                  *buf,
+  size_t                       n
+)
+{
+  while ( n > 0 ) {
+    size_t m;
+    char *pos;
+
+    m = ctx->todo < n ? ctx->todo : n;
+    pos = ctx->pos;
+    pos = memcpy( pos, buf, m );
+    n -= m;
+    buf = (char *) buf + m;
+
+    if ( m == ctx->todo ) {
+      rtems_record_client_status status;
+
+      ctx->todo = sizeof( ctx->item.format_64 );
+      ctx->pos = &ctx->item.format_64;
+      ctx->event = __builtin_bswap32( ctx->item.format_64.event );
+      ctx->data = __builtin_bswap64( ctx->item.format_64.data );
+
+      status = visit( ctx );
+
+      if ( status != RTEMS_RECORD_CLIENT_SUCCESS ) {
+        return status;
+      }
+    } else {
+      ctx->todo -= m;
+      ctx->pos = pos + m;
+    }
+  }
+
+  return RTEMS_RECORD_CLIENT_SUCCESS;
+}
+
+static rtems_record_client_status consume_init(
+  rtems_record_client_context *ctx,
+  const void                  *buf,
+  size_t                       n
+)
+{
+  while ( n > 0 ) {
+    size_t m;
+    char *pos;
+
+    m = ctx->todo < n ? ctx->todo : n;
+    pos = ctx->pos;
+    pos = memcpy( pos, buf, m );
+    n -= m;
+    buf = (char *) buf + m;
+
+    if ( m == ctx->todo ) {
+      uint32_t magic;
+
+      magic = ctx->header[ 1 ];
+
+      switch ( ctx->header[ 0 ] ) {
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+        case RTEMS_RECORD_FORMAT_LE_32:
+          ctx->todo = sizeof( ctx->item.format_32 );
+          ctx->pos = &ctx->item.format_32;
+          ctx->consume = consume_32;
+          break;
+        case RTEMS_RECORD_FORMAT_LE_64:
+          ctx->todo = sizeof( ctx->item.format_64 );
+          ctx->pos = &ctx->item.format_64;
+          ctx->consume = consume_64;
+          break;
+        case RTEMS_RECORD_FORMAT_BE_32:
+          ctx->todo = sizeof( ctx->item.format_32 );
+          ctx->pos = &ctx->item.format_32;
+          ctx->consume = consume_swap_32;
+          magic = __builtin_bswap32( magic );
+          break;
+        case RTEMS_RECORD_FORMAT_BE_64:
+          ctx->todo = sizeof( ctx->item.format_64 );
+          ctx->pos = &ctx->item.format_64;
+          ctx->consume = consume_swap_64;
+          magic = __builtin_bswap32( magic );
+          break;
+#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+        case RTEMS_RECORD_FORMAT_LE_32:
+          ctx->todo = sizeof( ctx->item.format_32 );
+          ctx->pos = &ctx->item.format_32;
+          ctx->consume = consume_swap_32;
+          magic = __builtin_bswap32( magic );
+          break;
+        case RTEMS_RECORD_FORMAT_LE_64:
+          ctx->todo = sizeof( ctx->item.format_64 );
+          ctx->pos = &ctx->item.format_64;
+          ctx->consume = consume_swap_64;
+          magic = __builtin_bswap32( magic );
+          break;
+        case RTEMS_RECORD_FORMAT_BE_32:
+          ctx->todo = sizeof( ctx->item.format_32 );
+          ctx->pos = &ctx->item.format_32;
+          ctx->consume = consume_32;
+          break;
+        case RTEMS_RECORD_FORMAT_BE_64:
+          ctx->todo = sizeof( ctx->item.format_64 );
+          ctx->pos = &ctx->item.format_64;
+          ctx->consume = consume_64;
+          break;
+#else
+#error "unexpected __BYTE_ORDER__"
+#endif
+        default:
+          return RTEMS_RECORD_CLIENT_ERROR_UNKNOWN_FORMAT;
+      }
+
+      if ( magic != RTEMS_RECORD_MAGIC ) {
+        return RTEMS_RECORD_CLIENT_ERROR_INVALID_MAGIC;
+      }
+
+      return rtems_record_client_run( ctx, buf, n );
+    } else {
+      ctx->todo -= m;
+      ctx->pos = pos + m;
+    }
+  }
+
+  return RTEMS_RECORD_CLIENT_SUCCESS;
+}
+
+void rtems_record_client_init(
+  rtems_record_client_context *ctx,
+  rtems_record_client_handler  handler,
+  void                        *arg
+)
+{
+  ctx = memset( ctx, 0, sizeof( *ctx ) );
+  ctx->to_bt_scaler = UINT64_C( 1 ) << 31;
+  ctx->handler = handler;
+  ctx->handler_arg = arg;
+  ctx->todo = sizeof( ctx->header );
+  ctx->pos = &ctx->header;
+  ctx->consume = consume_init;
+}
+
+rtems_record_client_status rtems_record_client_run(
+  rtems_record_client_context *ctx,
+  const void                  *buf,
+  size_t                       n
+)
+{
+  return ( *ctx->consume )( ctx, buf, n );
+}