qemu-armv7r

This board configuration will use QEMU to emulate generic ARM v7-R series hardware platform and provides support for these devices:

ARM Generic Timer
PL011 UART controller
VirtIO Device

Note

The upstream QEMU does not support ARM Cortex-R virtual machine. This board uses the OpenVela QEMU fork which adds ARM Cortex-R support. You can find the source code and prebuilt binaries at: https://github.com/open-vela/platform_external_qemu/commits/trunk/

Prerequisites

Download and extract the prebuilt QEMU from OpenVela. For Linux x86_64, download from: https://github.com/open-vela/prebuilts_qemu_linux-x86_64

The QEMU binary path will be referred to as <QEMU_PATH> in the following sections:

# Example path structure after extraction:
# <QEMU_PATH>/bin/qemu-system-arm
# <QEMU_PATH>/share/qemu/

Note

The -L option is required because the QEMU shared resources are not installed to system directories. It specifies the directory for the QEMU BIOS and device tree files.

Getting Started

NSH (Flat Build)

Configuring NuttX and compile:

$ ./tools/configure.sh -l qemu-armv7r:nsh
$ make -j$(nproc)

Running with qemu:

$ <QEMU_PATH>/bin/qemu-system-arm \
  -L <QEMU_PATH>/share/qemu \
  -M virt -semihosting -nographic -cpu cortex-r5f \
  -kernel ./nuttx

PNSH (Protected Build)

This is a configuration for testing the BUILD_PROTECTED configuration with MPU support:

$ cd nuttx
$ ./tools/configure.sh qemu-armv7r:pnsh
$ make -j$(nproc)

Running with qemu (note: both nuttx and nuttx_user must be loaded):

$ <QEMU_PATH>/bin/qemu-system-arm \
  -L <QEMU_PATH>/share/qemu \
  -M virt -semihosting -nographic -cpu cortex-r5f \
  -device loader,file=./nuttx_user \
  -device loader,file=./nuttx

Debugging with QEMU

The nuttx ELF image can be debugged with QEMU.

To debug the nuttx (ELF) with symbols, make sure the following change have been applied to defconfig:
```
+CONFIG_DEBUG_SYMBOLS=y
```

Run QEMU at shell terminal 1 (for NSH flat build):

$ <QEMU_PATH>/bin/qemu-system-arm \
  -L <QEMU_PATH>/share/qemu \
  -M virt -semihosting -nographic -cpu cortex-r5f \
  -kernel ./nuttx -S -s

Or for PNSH protected build:

$ <QEMU_PATH>/bin/qemu-system-arm \
  -L <QEMU_PATH>/share/qemu \
  -M virt -semihosting -nographic -cpu cortex-r5f \
  -device loader,file=./nuttx_user \
  -device loader,file=./nuttx -S -s

Run gdb with TUI, connect to QEMU, load nuttx and continue (at shell terminal 2):

$ arm-none-eabi-gdb -tui --eval-command='target remote localhost:1234' nuttx
(gdb) c
Continuing.
^C
Program received signal SIGINT, Interrupt.
(gdb)

For PNSH protected build, you need to load the user space symbols as well:

$ arm-none-eabi-gdb -tui --eval-command='target remote localhost:1234' nuttx
(gdb) add-symbol-file nuttx_user
(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.

ARM v7-R QEMU board support
NuttX kernel with PMU support enabled
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

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)

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