arch/armv7-r: userspace PMU access

This change enables performance monitoring unit (PMU) access from userspace
on ARMv7-R architecture by adding CONFIG_ARCH_HAVE_PERF_EVENTS_USER_ACCESS
support and building PMU code for userspace when needed.

Signed-off-by: hujun5 <hujun5@xiaomi.com>

Documentation/platforms/arm/qemu/boards/qemu-armv7r/index.rst

@@ -41,7 +41,7 @@ NSH (Flat Build)
 Configuring NuttX and compile::
 
      $ ./tools/configure.sh -l qemu-armv7r:nsh
-     $ make -j`nproc`
+     $ make -j$(nproc)
 
 Running with qemu::
 
@@ -58,7 +58,7 @@ with MPU support::
 
   $ cd nuttx
   $ ./tools/configure.sh qemu-armv7r:pnsh
-  $ make -j`nproc`
+  $ make -j$(nproc)
 
 Running with qemu (note: both nuttx and nuttx_user must be loaded)::
 
@@ -106,4 +106,93 @@ The nuttx ELF image can be debugged with QEMU.
 
      $ arm-none-eabi-gdb -tui --eval-command='target remote localhost:1234' nuttx
      (gdb) add-symbol-file nuttx_user
-     (gdb) c
+     (gdb) c
+
+=====================
+Userspace PMU Access
+=====================
+
+Overview
+========
+
+This document describes how to enable and use Performance Monitoring Unit (PMU)
+access from userspace applications on the ARM v7-R QEMU board. The PMU provides
+hardware performance counters that can be used to analyze and profile application
+performance.
+
+1. ARM v7-R QEMU board support
+2. NuttX kernel with PMU support enabled
+3. User-space access permissions configured
+
+Kernel Configuration
+--------------------
+
+To enable PMU support and userspace access, add the following configuration
+options to your defconfig::
+
+     +CONFIG_ARCH_PERF_EVENTS_USER_ACCESS=y
+     +CONFIG_ARCH_PERF_EVENTS=y
+
+
+Userspace API
+
+**perf_gettime**
+================
+
+Basic Usage
+-----------
+
+.. code-block:: c
+
+    clock_t start, end;
+    unsigned long frequency;
+    unsigned long cycles;
+
+    frequency = perf_getfreq();
+    if (frequency == 0) {
+        printf("ERROR: Performance frequency not available\n");
+        return;
+    }
+
+    printf("Operation Profiling Results:\n");
+    printf("CPU Frequency: %lu Hz\n\n", frequency);
+    printf("%-40s | %-12s | %-12s\n", "Operation", "Cycles", "Time (us)");
+    printf("%-40s-+-%-12s-+-%-12s\n", 
+           "----------------------------------------",
+           "------------", "------------");
+
+    start = perf_gettime();
+    result = 1;
+    for (int i = 1; i < 1000; i++) {
+        result *= i;
+    }
+
+    end = perf_gettime();
+    cycles = end - start;
+    printf("%-40s | %12lu | %12lu\n", "Multiplication (1K times)", cycles, (cycles * 1000000UL) / frequency);
+
+    start = perf_gettime();
+    result = 1000000;
+    for (int i = 1; i < 1000; i++) {
+        result /= (i + 1);
+    }
+    end = perf_gettime();
+    cycles = end - start;
+
+    printf("%-40s | %12lu | %12lu\n", "Division (1K times)", cycles, (cycles * 1000000UL) / frequency);
+
+Testing with QEMU
+=================
+
+PNSH (Protected) Configuration
+-------------------------------
+
+For protected build with userspace PMU access::
+
+     $ ./tools/configure.sh qemu-armv7r:pnsh
+     $ make -j$(nproc)
+
+2. Run QEMU (load both kernel and userspace)::
+
+     $ qemu-system-arm -M virt -semihosting -nographic -cpu cortex-r5f -device loader,file=nuttx_user -device loader,file=nuttx
+