Introduce Virtio-loopback epsilon release:

Epsilon release introduces a new compatibility layer which make virtio-loopback
design to work with QEMU and rust-vmm vhost-user backend without require any
changes.

Signed-off-by: Timos Ampelikiotis <t.ampelikiotis@virtualopensystems.com>
Change-Id: I52e57563e08a7d0bdc002f8e928ee61ba0c53dd9

3 files changed, 455 insertions, 0 deletions

diff --git a/scripts/performance/dissect.py b/scripts/performance/dissect.py
new file mode 100755
index 000000000..bf24f5092
--- /dev/null
+++ b/scripts/performance/dissect.py
@@ -0,0 +1,166 @@
+#!/usr/bin/env python3
+
+#  Print the percentage of instructions spent in each phase of QEMU
+#  execution.
+#
+#  Syntax:
+#  dissect.py [-h] -- <qemu executable> [<qemu executable options>] \
+#                   <target executable> [<target executable options>]
+#
+#  [-h] - Print the script arguments help message.
+#
+#  Example of usage:
+#  dissect.py -- qemu-arm coulomb_double-arm
+#
+#  This file is a part of the project "TCG Continuous Benchmarking".
+#
+#  Copyright (C) 2020  Ahmed Karaman <ahmedkhaledkaraman@gmail.com>
+#  Copyright (C) 2020  Aleksandar Markovic <aleksandar.qemu.devel@gmail.com>
+#
+#  This program is free software: you can redistribute it and/or modify
+#  it under the terms of the GNU General Public License as published by
+#  the Free Software Foundation, either version 2 of the License, or
+#  (at your option) any later version.
+#
+#  This program is distributed in the hope that it will be useful,
+#  but WITHOUT ANY WARRANTY; without even the implied warranty of
+#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+#  GNU General Public License for more details.
+#
+#  You should have received a copy of the GNU General Public License
+#  along with this program. If not, see <https://www.gnu.org/licenses/>.
+
+import argparse
+import os
+import subprocess
+import sys
+import tempfile
+
+
+def get_JIT_line(callgrind_data):
+    """
+    Search for the first instance of the JIT call in
+    the callgrind_annotate output when ran using --tree=caller
+    This is equivalent to the self number of instructions of JIT.
+
+    Parameters:
+    callgrind_data (list): callgrind_annotate output
+
+    Returns:
+    (int): Line number
+    """
+    line = -1
+    for i in range(len(callgrind_data)):
+        if callgrind_data[i].strip('\n') and \
+                callgrind_data[i].split()[-1] == "[???]":
+            line = i
+            break
+    if line == -1:
+        sys.exit("Couldn't locate the JIT call ... Exiting.")
+    return line
+
+
+def main():
+    # Parse the command line arguments
+    parser = argparse.ArgumentParser(
+        usage='dissect.py [-h] -- '
+        '<qemu executable> [<qemu executable options>] '
+        '<target executable> [<target executable options>]')
+
+    parser.add_argument('command', type=str, nargs='+', help=argparse.SUPPRESS)
+
+    args = parser.parse_args()
+
+    # Extract the needed variables from the args
+    command = args.command
+
+    # Insure that valgrind is installed
+    check_valgrind = subprocess.run(
+        ["which", "valgrind"], stdout=subprocess.DEVNULL)
+    if check_valgrind.returncode:
+        sys.exit("Please install valgrind before running the script.")
+
+    # Save all intermediate files in a temporary directory
+    with tempfile.TemporaryDirectory() as tmpdirname:
+        # callgrind output file path
+        data_path = os.path.join(tmpdirname, "callgrind.data")
+        # callgrind_annotate output file path
+        annotate_out_path = os.path.join(tmpdirname, "callgrind_annotate.out")
+
+        # Run callgrind
+        callgrind = subprocess.run((["valgrind",
+                                     "--tool=callgrind",
+                                     "--callgrind-out-file=" + data_path]
+                                    + command),
+                                   stdout=subprocess.DEVNULL,
+                                   stderr=subprocess.PIPE)
+        if callgrind.returncode:
+            sys.exit(callgrind.stderr.decode("utf-8"))
+
+        # Save callgrind_annotate output
+        with open(annotate_out_path, "w") as output:
+            callgrind_annotate = subprocess.run(
+                ["callgrind_annotate", data_path, "--tree=caller"],
+                stdout=output,
+                stderr=subprocess.PIPE)
+            if callgrind_annotate.returncode:
+                sys.exit(callgrind_annotate.stderr.decode("utf-8"))
+
+        # Read the callgrind_annotate output to callgrind_data[]
+        callgrind_data = []
+        with open(annotate_out_path, 'r') as data:
+            callgrind_data = data.readlines()
+
+        # Line number with the total number of instructions
+        total_instructions_line_number = 20
+        # Get the total number of instructions
+        total_instructions_line_data = \
+            callgrind_data[total_instructions_line_number]
+        total_instructions = total_instructions_line_data.split()[0]
+        total_instructions = int(total_instructions.replace(',', ''))
+
+        # Line number with the JIT self number of instructions
+        JIT_self_instructions_line_number = get_JIT_line(callgrind_data)
+        # Get the JIT self number of instructions
+        JIT_self_instructions_line_data = \
+            callgrind_data[JIT_self_instructions_line_number]
+        JIT_self_instructions = JIT_self_instructions_line_data.split()[0]
+        JIT_self_instructions = int(JIT_self_instructions.replace(',', ''))
+
+        # Line number with the JIT self + inclusive number of instructions
+        # It's the line above the first JIT call when running with --tree=caller
+        JIT_total_instructions_line_number = JIT_self_instructions_line_number-1
+        # Get the JIT self + inclusive number of instructions
+        JIT_total_instructions_line_data = \
+            callgrind_data[JIT_total_instructions_line_number]
+        JIT_total_instructions = JIT_total_instructions_line_data.split()[0]
+        JIT_total_instructions = int(JIT_total_instructions.replace(',', ''))
+
+        # Calculate number of instructions in helpers and code generation
+        helpers_instructions = JIT_total_instructions-JIT_self_instructions
+        code_generation_instructions = total_instructions-JIT_total_instructions
+
+        # Print results (Insert commas in large numbers)
+        # Print total number of instructions
+        print('{:<20}{:>20}\n'.
+              format("Total Instructions:",
+                     format(total_instructions, ',')))
+        # Print code generation instructions and percentage
+        print('{:<20}{:>20}\t{:>6.3f}%'.
+              format("Code Generation:",
+                     format(code_generation_instructions, ","),
+                     (code_generation_instructions / total_instructions) * 100))
+        # Print JIT instructions and percentage
+        print('{:<20}{:>20}\t{:>6.3f}%'.
+              format("JIT Execution:",
+                     format(JIT_self_instructions, ","),
+                     (JIT_self_instructions / total_instructions) * 100))
+        # Print helpers instructions and percentage
+        print('{:<20}{:>20}\t{:>6.3f}%'.
+              format("Helpers:",
+                     format(helpers_instructions, ","),
+                     (helpers_instructions/total_instructions)*100))
+
+
+if __name__ == "__main__":
+    main()
diff --git a/scripts/performance/topN_callgrind.py b/scripts/performance/topN_callgrind.py
new file mode 100755
index 000000000..67c59197a
--- /dev/null
+++ b/scripts/performance/topN_callgrind.py
@@ -0,0 +1,140 @@
+#!/usr/bin/env python3
+
+#  Print the top N most executed functions in QEMU using callgrind.
+#  Syntax:
+#  topN_callgrind.py [-h] [-n] <number of displayed top functions>  -- \
+#           <qemu executable> [<qemu executable options>] \
+#           <target executable> [<target execurable options>]
+#
+#  [-h] - Print the script arguments help message.
+#  [-n] - Specify the number of top functions to print.
+#       - If this flag is not specified, the tool defaults to 25.
+#
+#  Example of usage:
+#  topN_callgrind.py -n 20 -- qemu-arm coulomb_double-arm
+#
+#  This file is a part of the project "TCG Continuous Benchmarking".
+#
+#  Copyright (C) 2020  Ahmed Karaman <ahmedkhaledkaraman@gmail.com>
+#  Copyright (C) 2020  Aleksandar Markovic <aleksandar.qemu.devel@gmail.com>
+#
+#  This program is free software: you can redistribute it and/or modify
+#  it under the terms of the GNU General Public License as published by
+#  the Free Software Foundation, either version 2 of the License, or
+#  (at your option) any later version.
+#
+#  This program is distributed in the hope that it will be useful,
+#  but WITHOUT ANY WARRANTY; without even the implied warranty of
+#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+#  GNU General Public License for more details.
+#
+#  You should have received a copy of the GNU General Public License
+#  along with this program. If not, see <https://www.gnu.org/licenses/>.
+
+import argparse
+import os
+import subprocess
+import sys
+
+
+# Parse the command line arguments
+parser = argparse.ArgumentParser(
+    usage='topN_callgrind.py [-h] [-n] <number of displayed top functions>  -- '
+          '<qemu executable> [<qemu executable options>] '
+          '<target executable> [<target executable options>]')
+
+parser.add_argument('-n', dest='top', type=int, default=25,
+                    help='Specify the number of top functions to print.')
+
+parser.add_argument('command', type=str, nargs='+', help=argparse.SUPPRESS)
+
+args = parser.parse_args()
+
+# Extract the needed variables from the args
+command = args.command
+top = args.top
+
+# Insure that valgrind is installed
+check_valgrind_presence = subprocess.run(["which", "valgrind"],
+                                         stdout=subprocess.DEVNULL)
+if check_valgrind_presence.returncode:
+    sys.exit("Please install valgrind before running the script!")
+
+# Run callgrind
+callgrind = subprocess.run((
+    ["valgrind", "--tool=callgrind", "--callgrind-out-file=/tmp/callgrind.data"]
+    + command),
+    stdout=subprocess.DEVNULL,
+    stderr=subprocess.PIPE)
+if callgrind.returncode:
+    sys.exit(callgrind.stderr.decode("utf-8"))
+
+# Save callgrind_annotate output to /tmp/callgrind_annotate.out
+with open("/tmp/callgrind_annotate.out", "w") as output:
+    callgrind_annotate = subprocess.run(["callgrind_annotate",
+                                         "/tmp/callgrind.data"],
+                                        stdout=output,
+                                        stderr=subprocess.PIPE)
+    if callgrind_annotate.returncode:
+        os.unlink('/tmp/callgrind.data')
+        output.close()
+        os.unlink('/tmp/callgrind_annotate.out')
+        sys.exit(callgrind_annotate.stderr.decode("utf-8"))
+
+# Read the callgrind_annotate output to callgrind_data[]
+callgrind_data = []
+with open('/tmp/callgrind_annotate.out', 'r') as data:
+    callgrind_data = data.readlines()
+
+# Line number with the total number of instructions
+total_instructions_line_number = 20
+
+# Get the total number of instructions
+total_instructions_line_data = callgrind_data[total_instructions_line_number]
+total_number_of_instructions = total_instructions_line_data.split(' ')[0]
+total_number_of_instructions = int(
+    total_number_of_instructions.replace(',', ''))
+
+# Line number with the top function
+first_func_line = 25
+
+# Number of functions recorded by callgrind, last two lines are always empty
+number_of_functions = len(callgrind_data) - first_func_line - 2
+
+# Limit the number of top functions to "top"
+number_of_top_functions = (top if number_of_functions >
+                           top else number_of_functions)
+
+# Store the data of the top functions in top_functions[]
+top_functions = callgrind_data[first_func_line:
+                               first_func_line + number_of_top_functions]
+
+# Print table header
+print('{:>4}  {:>10}  {:<30}  {}\n{}  {}  {}  {}'.format('No.',
+                                                         'Percentage',
+                                                         'Function Name',
+                                                         'Source File',
+                                                         '-' * 4,
+                                                         '-' * 10,
+                                                         '-' * 30,
+                                                         '-' * 30,
+                                                         ))
+
+# Print top N functions
+for (index, function) in enumerate(top_functions, start=1):
+    function_data = function.split()
+    # Calculate function percentage
+    function_instructions = float(function_data[0].replace(',', ''))
+    function_percentage = (function_instructions /
+                           total_number_of_instructions)*100
+    # Get function name and source files path
+    function_source_file, function_name = function_data[1].split(':')
+    # Print extracted data
+    print('{:>4}  {:>9.3f}%  {:<30}  {}'.format(index,
+                                                round(function_percentage, 3),
+                                                function_name,
+                                                function_source_file))
+
+# Remove intermediate files
+os.unlink('/tmp/callgrind.data')
+os.unlink('/tmp/callgrind_annotate.out')
diff --git a/scripts/performance/topN_perf.py b/scripts/performance/topN_perf.py
new file mode 100755
index 000000000..07be195fc
--- /dev/null
+++ b/scripts/performance/topN_perf.py
@@ -0,0 +1,149 @@
+#!/usr/bin/env python3
+
+#  Print the top N most executed functions in QEMU using perf.
+#  Syntax:
+#  topN_perf.py [-h] [-n] <number of displayed top functions>  -- \
+#           <qemu executable> [<qemu executable options>] \
+#           <target executable> [<target execurable options>]
+#
+#  [-h] - Print the script arguments help message.
+#  [-n] - Specify the number of top functions to print.
+#       - If this flag is not specified, the tool defaults to 25.
+#
+#  Example of usage:
+#  topN_perf.py -n 20 -- qemu-arm coulomb_double-arm
+#
+#  This file is a part of the project "TCG Continuous Benchmarking".
+#
+#  Copyright (C) 2020  Ahmed Karaman <ahmedkhaledkaraman@gmail.com>
+#  Copyright (C) 2020  Aleksandar Markovic <aleksandar.qemu.devel@gmail.com>
+#
+#  This program is free software: you can redistribute it and/or modify
+#  it under the terms of the GNU General Public License as published by
+#  the Free Software Foundation, either version 2 of the License, or
+#  (at your option) any later version.
+#
+#  This program is distributed in the hope that it will be useful,
+#  but WITHOUT ANY WARRANTY; without even the implied warranty of
+#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+#  GNU General Public License for more details.
+#
+#  You should have received a copy of the GNU General Public License
+#  along with this program. If not, see <https://www.gnu.org/licenses/>.
+
+import argparse
+import os
+import subprocess
+import sys
+
+
+# Parse the command line arguments
+parser = argparse.ArgumentParser(
+    usage='topN_perf.py [-h] [-n] <number of displayed top functions >  -- '
+          '<qemu executable> [<qemu executable options>] '
+          '<target executable> [<target executable options>]')
+
+parser.add_argument('-n', dest='top', type=int, default=25,
+                    help='Specify the number of top functions to print.')
+
+parser.add_argument('command', type=str, nargs='+', help=argparse.SUPPRESS)
+
+args = parser.parse_args()
+
+# Extract the needed variables from the args
+command = args.command
+top = args.top
+
+# Insure that perf is installed
+check_perf_presence = subprocess.run(["which", "perf"],
+                                     stdout=subprocess.DEVNULL)
+if check_perf_presence.returncode:
+    sys.exit("Please install perf before running the script!")
+
+# Insure user has previllage to run perf
+check_perf_executability = subprocess.run(["perf", "stat", "ls", "/"],
+                                          stdout=subprocess.DEVNULL,
+                                          stderr=subprocess.DEVNULL)
+if check_perf_executability.returncode:
+    sys.exit(
+"""
+Error:
+You may not have permission to collect stats.
+
+Consider tweaking /proc/sys/kernel/perf_event_paranoid,
+which controls use of the performance events system by
+unprivileged users (without CAP_SYS_ADMIN).
+
+  -1: Allow use of (almost) all events by all users
+      Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK
+   0: Disallow ftrace function tracepoint by users without CAP_SYS_ADMIN
+      Disallow raw tracepoint access by users without CAP_SYS_ADMIN
+   1: Disallow CPU event access by users without CAP_SYS_ADMIN
+   2: Disallow kernel profiling by users without CAP_SYS_ADMIN
+
+To make this setting permanent, edit /etc/sysctl.conf too, e.g.:
+   kernel.perf_event_paranoid = -1
+
+* Alternatively, you can run this script under sudo privileges.
+"""
+)
+
+# Run perf record
+perf_record = subprocess.run((["perf", "record", "--output=/tmp/perf.data"] +
+                              command),
+                             stdout=subprocess.DEVNULL,
+                             stderr=subprocess.PIPE)
+if perf_record.returncode:
+    os.unlink('/tmp/perf.data')
+    sys.exit(perf_record.stderr.decode("utf-8"))
+
+# Save perf report output to /tmp/perf_report.out
+with open("/tmp/perf_report.out", "w") as output:
+    perf_report = subprocess.run(
+        ["perf", "report", "--input=/tmp/perf.data", "--stdio"],
+        stdout=output,
+        stderr=subprocess.PIPE)
+    if perf_report.returncode:
+        os.unlink('/tmp/perf.data')
+        output.close()
+        os.unlink('/tmp/perf_report.out')
+        sys.exit(perf_report.stderr.decode("utf-8"))
+
+# Read the reported data to functions[]
+functions = []
+with open("/tmp/perf_report.out", "r") as data:
+    # Only read lines that are not comments (comments start with #)
+    # Only read lines that are not empty
+    functions = [line for line in data.readlines() if line and line[0]
+                 != '#' and line[0] != "\n"]
+
+# Limit the number of top functions to "top"
+number_of_top_functions = top if len(functions) > top else len(functions)
+
+# Store the data of the top functions in top_functions[]
+top_functions = functions[:number_of_top_functions]
+
+# Print table header
+print('{:>4}  {:>10}  {:<30}  {}\n{}  {}  {}  {}'.format('No.',
+                                                         'Percentage',
+                                                         'Name',
+                                                         'Invoked by',
+                                                         '-' * 4,
+                                                         '-' * 10,
+                                                         '-' * 30,
+                                                         '-' * 25))
+
+# Print top N functions
+for (index, function) in enumerate(top_functions, start=1):
+    function_data = function.split()
+    function_percentage = function_data[0]
+    function_name = function_data[-1]
+    function_invoker = ' '.join(function_data[2:-2])
+    print('{:>4}  {:>10}  {:<30}  {}'.format(index,
+                                             function_percentage,
+                                             function_name,
+                                             function_invoker))
+
+# Remove intermediate files
+os.unlink('/tmp/perf.data')
+os.unlink('/tmp/perf_report.out')