fs: Add Kernel-level VFS Performance Profiler

This adds a kernel-level performance profiler for the VFS.
By enabling CONFIG_FS_PROFILER, the core VFS system calls
(file_read, file_write, file_open, and file_close) are
instrumented to track high-resolution execution times using
clock_systime_timespec() seamlessly.

The collected statistics are exposed dynamically via a new
procfs node at /proc/fs/profile, allowing CI regression
testing without needing external debugging tools.

Signed-off-by: Sumit6307 <sumitkesar6307@gmail.com>

Author: Sumit6307

Documentation/components/filesystem/index.rst

@@ -543,6 +543,7 @@ NuttX provides support for a variety of file systems out of the box.
   nxffs.rst
   partition.rst
   procfs.rst
+  profiler.rst
   romfs.rst
   rpmsgfs.rst
   smartfs.rst

Documentation/components/filesystem/profiler.rst

@@ -0,0 +1,38 @@
+==============================
+VFS Performance Profiler
+==============================
+
+The Virtual File System (VFS) Performance Profiler provides a simple, in-kernel 
+mechanism to track execution times and invocation counts for core VFS operations 
+(read, write, open, close) seamlessly. This is highly suitable for 
+CI/CD automated regression testing and performance bottleneck identification.
+
+Configuration
+=============
+
+To enable the profiler, select ``CONFIG_FS_PROFILER`` in your Kconfig.
+To expose the metrics dynamically via procfs, ensure ``CONFIG_FS_PROCFS`` is enabled, and 
+the profiler node is included via ``CONFIG_FS_PROCFS_PROFILER``.
+
+Usage
+=====
+
+When enabled, the profiler automatically intercepts calls to the underlying 
+inode operations and records the execution elapsed times using ``perf_gettime()``.
+Since no blocking mutexes are used during updates (fast ``atomic.h`` operations 
+are utilized instead), the overhead is extremely minimal and safely scales on SMP.
+
+To view the current statistics collectively from the NuttShell (NSH), simply 
+read the node:
+
+.. code-block:: bash
+
+    nsh> cat /proc/fs/profile
+    VFS Performance Profile:
+      Reads:          12 (Total time: 4500120 ns)
+      Writes:          3 (Total time:   95050 ns)
+      Opens:          15 (Total time: 1005000 ns)
+      Closes:         15 (Total time:   45000 ns)
+
+The reported times are in the raw ticks/units provided by ``perf_gettime()`` on 
+your specific architecture.

boards/sim/sim/sim/configs/citest/defconfig

@@ -152,3 +152,5 @@ CONFIG_TESTS_TESTSUITES=y
 CONFIG_TLS_NCLEANUP=4
 CONFIG_TLS_NELEM=16
 CONFIG_TLS_TASK_NELEM=8
+CONFIG_FS_PROFILER=y
+CONFIG_FS_PROCFS_PROFILER=y

fs/Kconfig

@@ -5,6 +5,14 @@
 
 comment "File system configuration"
 
+config FS_PROFILER
+	bool "VFS Performance Profiler"
+	default n
+	---help---
+		Enable nanosecond/microsecond-level profiling for the Virtual File
+		System (VFS) operations (open, close, read, write). The profile stats
+		can be read via /proc/fs/profile if PROCFS is enabled.
+
 config DISABLE_MOUNTPOINT
 	bool "Disable support for mount points"
 	default n

fs/procfs/CMakeLists.txt

@@ -40,6 +40,10 @@ if(NOT CONFIG_DISABLE_MOUNTPOINT)
       list(APPEND SRCS fs_procfspressure.c)
     endif()
 
+    if(CONFIG_FS_PROCFS_PROFILER)
+      list(APPEND SRCS fs_procfsprofile.c)
+    endif()
+
     target_sources(fs PRIVATE ${SRCS})
 
   endif()

fs/procfs/Kconfig

@@ -148,6 +148,16 @@ config FS_PROCFS_EXCLUDE_VERSION
 	bool "Exclude version"
 	default DEFAULT_SMALL
 
+config FS_PROCFS_PROFILER
+	bool "Include fs/profile information"
+	depends on FS_PROFILER && !DEFAULT_SMALL
+	default n
+	---help---
+		Enable the VFS performance profiler procfs node at /proc/fs/profile.
+		This node provides real-time dynamic statistics (call counts, elapsed
+		time) for core VFS operations (read, write, open, close) to help
+		identify filesystem performance bottlenecks and regressions.
+
 config FS_PROCFS_INCLUDE_PRESSURE
 	bool "Include memory pressure notification"
 	default n

fs/procfs/Make.defs

@@ -32,6 +32,10 @@ ifeq ($(CONFIG_FS_PROCFS_INCLUDE_PRESSURE),y)
 CSRCS += fs_procfspressure.c
 endif
 
+ifeq ($(CONFIG_FS_PROCFS_PROFILER),y)
+CSRCS += fs_procfsprofile.c
+endif
+
 # Include procfs build support
 
 DEPPATH += --dep-path procfs

fs/procfs/fs_procfs.c

@@ -74,6 +74,9 @@ extern const struct procfs_operations g_thermal_operations;
 extern const struct procfs_operations g_uptime_operations;
 extern const struct procfs_operations g_version_operations;
 extern const struct procfs_operations g_pressure_operations;
+#if defined(CONFIG_FS_PROFILER) && defined(CONFIG_FS_PROCFS_PROFILER)
+extern const struct procfs_operations g_vfs_profile_operations;
+#endif
 
 /* This is not good.  These are implemented in other sub-systems.  Having to
  * deal with them here is not a good coupling. What is really needed is a
@@ -208,6 +211,9 @@ static const struct procfs_entry_s g_procfs_entries[] =
 #ifndef CONFIG_FS_PROCFS_EXCLUDE_VERSION
   { "version",      &g_version_operations,  PROCFS_FILE_TYPE   },
 #endif
+#if defined(CONFIG_FS_PROFILER) && defined(CONFIG_FS_PROCFS_PROFILER)
+  { "profile",      &g_vfs_profile_operations,  PROCFS_FILE_TYPE   },
+#endif
 };
 
 #ifdef CONFIG_FS_PROCFS_REGISTER

fs/procfs/fs_procfsprofile.c

@@ -0,0 +1,126 @@
+/****************************************************************************
+ * fs/procfs/fs_procfsprofile.c
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.  The
+ * ASF licenses this file to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance with the
+ * License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.  See the
+ * License for the specific language governing permissions and limitations
+ * under the License.
+ *
+ ****************************************************************************/
+
+/****************************************************************************
+ * Included Files
+ ****************************************************************************/
+
+#include <nuttx/config.h>
+
+#if defined(CONFIG_FS_PROCFS) && defined(CONFIG_FS_PROFILER) && \
+    defined(CONFIG_FS_PROCFS_PROFILER)
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <stdio.h>
+#include <string.h>
+
+#include <nuttx/fs/fs.h>
+#include <nuttx/fs/procfs.h>
+#include "../vfs/vfs.h"
+
+/****************************************************************************
+ * Private Functions
+ ****************************************************************************/
+
+static int profile_open(FAR struct file *filep, FAR const char *relpath,
+                        int oflags, mode_t mode)
+{
+  return OK;
+}
+
+static int profile_close(FAR struct file *filep)
+{
+  return OK;
+}
+
+static ssize_t profile_read(FAR struct file *filep, FAR char *buffer,
+                            size_t buflen)
+{
+  char buf[256];
+  size_t linesize;
+  off_t offset = filep->f_pos;
+
+  if (offset > 0)
+    {
+      return 0;
+    }
+
+  procfs_snprintf(buf, sizeof(buf),
+           "VFS Performance Profile:\n"
+           "  Reads:  %10lu (Total time: %llu ns)\n"
+           "  Writes: %10lu (Total time: %llu ns)\n"
+           "  Opens:  %10lu (Total time: %llu ns)\n"
+           "  Closes: %10lu (Total time: %llu ns)\n",
+           (unsigned long)g_vfs_profile.reads,
+           (unsigned long long)g_vfs_profile.total_read_time,
+           (unsigned long)g_vfs_profile.writes,
+           (unsigned long long)g_vfs_profile.total_write_time,
+           (unsigned long)g_vfs_profile.opens,
+           (unsigned long long)g_vfs_profile.total_open_time,
+           (unsigned long)g_vfs_profile.closes,
+           (unsigned long long)g_vfs_profile.total_close_time);
+
+  linesize = strlen(buf);
+  if (linesize > buflen)
+    {
+      linesize = buflen;
+    }
+
+  memcpy(buffer, buf, linesize);
+  filep->f_pos += linesize;
+  return linesize;
+}
+
+static int profile_dup(FAR const struct file *oldp, FAR struct file *newp)
+{
+  return OK;
+}
+
+static int profile_stat(FAR const char *relpath, FAR struct stat *buf)
+{
+  memset(buf, 0, sizeof(struct stat));
+  buf->st_mode = S_IFREG | S_IROTH | S_IRGRP | S_IRUSR;
+  return OK;
+}
+
+/****************************************************************************
+ * Public Data
+ ****************************************************************************/
+
+const struct procfs_operations g_vfs_profile_operations =
+{
+  profile_open,       /* open */
+  profile_close,      /* close */
+  profile_read,       /* read */
+  NULL,               /* write */
+  NULL,               /* poll */
+  profile_dup,        /* dup */
+  NULL,               /* opendir */
+  NULL,               /* closedir */
+  NULL,               /* readdir */
+  NULL,               /* rewinddir */
+  profile_stat        /* stat */
+};
+
+#endif
+

fs/vfs/CMakeLists.txt

@@ -93,4 +93,10 @@ if(CONFIG_SIGNAL_FD)
   list(APPEND SRCS fs_signalfd.c)
 endif()
 
+# Support for profiler
+
+if(CONFIG_FS_PROFILER)
+  list(APPEND SRCS fs_profile.c)
+endif()
+
 target_sources(fs PRIVATE ${SRCS})