diff --git a/tools/build/build b/tools/build/build
index fb3ecf2b31..452bd88d8c 100755
--- a/tools/build/build
+++ b/tools/build/build
@@ -5,3 +5,4 @@ scriptDir=$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
 
 ${scriptDir}/test "${@}"
 ${scriptDir}/make "${@}"
+${scriptDir}/check_elf
diff --git a/tools/build/check_elf b/tools/build/check_elf
new file mode 100755
index 0000000000..b89317fd2d
--- /dev/null
+++ b/tools/build/check_elf
@@ -0,0 +1,5 @@
+#!/usr/bin/env bash
+set -e
+
+scriptDir=$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
+python3 ${scriptDir}/elf_sanity.py ${scriptDir}/../../*.elf
diff --git a/tools/build/elf_sanity.py b/tools/build/elf_sanity.py
new file mode 100644
index 0000000000..b46e8be2fa
--- /dev/null
+++ b/tools/build/elf_sanity.py
@@ -0,0 +1,170 @@
+import argparse
+import struct
+import sys
+
+try:
+    from elftools.elf.elffile import ELFFile
+except ImportError:
+    print('pytelftools missing, install to run this script', file=sys.stderr)
+    print('https://github.com/eliben/pyelftools#installing', file=sys.stderr)
+    sys.exit(1)
+
+
+class Colors:
+    RED = '\033[91m'
+    BLUE = '\033[94m'
+    GREEN = '\033[92m'
+    END = '\033[0m'
+
+
+param_type_to_str_dict = {
+    0x0 | 0x0 << 2 | 0x1 << 3: 'PARAM_UINT8',
+    0x0 | 0x0 << 2 | 0x0 << 3: 'PARAM_INT8',
+    0x1 | 0x0 << 2 | 0x1 << 3: 'PARAM_UIN16',
+    0x1 | 0x0 << 2 | 0x0 << 3: 'PARAM_INT16',
+    0x2 | 0x0 << 2 | 0x1 << 3: 'PARAM_UINT32',
+    0x2 | 0x0 << 2 | 0x0 << 3: 'PARAM_INT32',
+    0x2 | 0x1 << 2 | 0x0 << 3: 'PARAM_FLOAT'
+}
+
+
+def param_type_to_str(t: int) -> str:
+    read_only = str()
+    if t & (1 << 6):  # PARAM_RONLY set
+        read_only = ' | PARAM_RONLY'
+
+    return '{:12}{}'.format(param_type_to_str_dict[t & ~(1 << 6)], read_only)
+
+
+log_type_to_str_dict = {
+    0x1: 'LOG_UINT8',
+    0x2: 'LOG_INT8',
+    0x3: 'LOG_UIN16',
+    0x4: 'LOG_INT16',
+    0x5: 'LOG_UINT32',
+    0x6: 'LOG_INT32',
+    0x7: 'LOG_FLOAT',
+    0x8: 'LOG_FP16'
+}
+
+
+def log_type_to_str(t: int) -> str:
+    by_function = str()
+    if t & (1 << 6):  # BY_FUNCTION set
+        by_function = ' | BY_FUNCTION'
+
+    return '{:12}{}'.format(log_type_to_str_dict[t & ~(1 << 6)], by_function)
+
+
+def process_file(filename, list_params: bool, list_logs: bool):
+    with open(filename, 'rb') as f:
+        parameters = check_structs(f, 'param')
+        if list_params:
+            for key in sorted(parameters.keys()):
+                t = parameters[key]
+                print('{:25}\t{}'.format(key, param_type_to_str(t)))
+
+        logs = check_structs(f, 'log')
+        if list_logs:
+            for key in sorted(logs.keys()):
+                t = logs[key]
+                print('{:25}\t{}'.format(key, log_type_to_str(t)))
+
+        n_logs = Colors.GREEN + str(len(logs.keys())) + Colors.END
+        n_params = Colors.BLUE + str(len(parameters.keys())) + Colors.END
+        print('{} parameters and {} log vars in elf'.format(n_params, n_logs))
+
+
+def get_offset_of(elf, addr):
+    for seg in elf.iter_segments():
+        if seg.header['p_type'] != 'PT_LOAD':
+            continue
+
+        # If the symbol is inside the range of a LOADed segment, calculate the
+        # file offset by subtracting the virtual start address and adding the
+        # file offset of the loaded section(s)
+        if addr >= seg['p_vaddr'] and addr < seg['p_vaddr'] + seg['p_filesz']:
+            return addr - seg['p_vaddr'] + seg['p_offset']
+
+    return None
+
+
+def get_offset_of_symbol(elf, name):
+    section = elf.get_section_by_name('.symtab')
+    sym = section.get_symbol_by_name(name)[0]
+    if not sym:
+        print('symbol %s not found' % name, file=sys.stderr)
+        sys.exit(1)
+
+    return get_offset_of(elf, sym['st_value'])
+
+
+def check_structs(stream, what: str) -> dict:
+    elf = ELFFile(stream)
+    offset = get_offset_of_symbol(elf, '_{}_start'.format(what))
+    stop_offset = get_offset_of_symbol(elf, '_{}_stop'.format(what))
+
+    name_type_dict = {}
+    name_maxlen = 25
+    struct_len = 12
+    group_bit = 0x1 << 7
+    start_bit = 0x1
+
+    while offset < stop_offset:
+        elf.stream.seek(offset)
+        #
+        # Parsing log or param, first unpack the struct:
+        # struct [param_s|log_s] {
+        #   uint8_t type;
+        #   char * name;
+        #   void * address;
+        # };
+        #
+        # We want the type and the name.
+        #
+        buffer = elf.stream.read(struct_len)
+        t, addr = struct.unpack('@Bxxxixxxx', buffer)
+        #
+        # Next, convert address of name to offset in elf
+        #
+        addr = get_offset_of(elf, addr)
+        #
+        # And read the name from that offset
+        #
+        elf.stream.seek(addr)
+        name = ''.join(iter(lambda: stream.read(1).decode('ascii'), '\x00'))
+        #
+        # Check if this is start of a group
+        #
+        if t & group_bit != 0 and t & start_bit != 0:
+            current_group = name
+        elif t & group_bit == 0:
+            name = '%s.%s' % (current_group, name)
+            if name in name_type_dict:
+                print('%sDuplicate parameter detected!%s (%s)' %
+                      (Colors.RED, Colors.END, name), file=sys.stderr)
+                sys.exit(1)
+            else:
+                name_type_dict[name] = t
+
+            if len(name) > name_maxlen:
+                print('%sName too long!%s (%s > %d)' %
+                      (Colors.RED, Colors.END, name, name_maxlen),
+                      file=sys.stderr)
+                sys.exit(1)
+
+        offset += struct_len
+    return name_type_dict
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--list-params', action='store_true')
+    parser.add_argument('--list-logs', action='store_true')
+    parser.add_argument('filename', nargs=argparse.REMAINDER)
+    args = parser.parse_args()
+
+    if args.filename:
+        process_file(args.filename[0], args.list_params, args.list_logs)
+    else:
+        sys.exit(1)