nvs_flash: Add binary creation support for NVS partition.

Ideally suited for generating a binary externally, containing key-value pairs specific
to device manufacturers. Utility allows creation of a binary, compatible
with NVS structure, which can be separately flashed onto a new
partition. This helps device manufacturers set different values for
different devices, e.g. serial numbers, but using a single firmaware
image.

components/nvs_flash/nvs_partition_generator/README.rst

@@ -0,0 +1,60 @@
+NVS Partition Generator Utility
+===============================
+
+Introduction
+------------
+
+:component_file:`nvs_flash/nvs_partition_generator/nvs_partition_gen.py` utility is designed to help create a binary file, compatible with NVS architecture defined in :doc:`Non-Volatile Storage </api-reference/storage/nvs_flash>`, based on user provided key-value pairs in a CSV file.
+Utility is ideally suited for generating a binary blob, containing data specific to ODM/OEM, which can be flashed externally at the time of device manufacturing. This helps manufacturers set unique value for various parameters for each device, e.g. serial number, while using same application firmaware for all devices.
+
+CSV file format
+---------------
+
+Each row of the .csv file should have 4 parameters, separated by comma. Below is the description of each of these parameters:
+
+Key
+	Key of the data. Data can later be accessed from an application via this key.
+
+Type
+	Supported values are ``file``, ``data`` and ``namespace``.
+
+Encoding
+    Supported values are: ``u8``, ``i8``, ``u16``, ``u32``, ``i32``, ``string``, ``hex2bin`` and ``binary``. This specifies how actual data values are encoded in the resultant binary file. Difference between ``string`` and ``binary`` encoding is that ``string`` data is terminated with a NULL character, whereas ``binary`` data is not.
+
+    .. note:: For ``file`` type, only ``hex2bin``, ``string`` and ``binary`` is supported as of now.
+
+Value
+	Data value.
+
+.. note:: Encoding and Value cells for ``namespace`` field type should be empty. Encoding and Value of ``namespace`` is fixed and isn't configurable. Any value in these cells are ignored.
+
+.. note:: First row of the CSV file should always be column header and isn't configurable.
+
+Below is an example dump of such CSV file::
+
+    key,type,encoding,value     <-- column header
+    namespace_name,namespace,,  <-- First entry should be of type "namespace"
+    key1,data,u8,1
+    key2,file,string,/path/to/file
+
+.. note:: Make sure there are no spaces before and after ',' in CSV file.
+
+NVS Entry and Namespace association
+-----------------------------------
+
+When a new namespace entry is encountered in the CSV file, each follow-up entries will be part of that namespace, until next namespace entry is found, in which case all the follow-up entries will be part of the new namespace.
+
+.. note:: First entry in a CSV file should always be ``namespace`` entry.
+
+Running the utility
+-------------------
+
+A sample CSV file provided with the utility. You can run the utility using below command::
+
+    python nvs_partition_generator.py sample.csv sample.bin
+
+Caveats
+-------
+-  Utility doesn't check for duplicate keys and will write data pertaining to both keys. User needs to make sure keys are distinct.
+-  Once a new page is created, no data will be written in the space left in previous page. Fields in the CSV file need to be ordered in such a way so as to optimize memory.
+-  64-bit datatype is not yet supported.

components/nvs_flash/nvs_partition_generator/nvs_partition_gen.py

@@ -0,0 +1,361 @@
+#!/usr/bin/env python
+#
+# esp-idf NVS partition generation tool. Tool helps in generating NVS-compatible
+# partition binary, with key-value pair entries provided via a CSV file.
+#
+# Copyright 2018 Espressif Systems (Shanghai) PTE LTD
+#
+# Licensed 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.
+#
+import sys
+import argparse
+import binascii
+import getopt
+import struct
+import os
+import array
+import csv
+import zlib
+from os import path
+
+""" Class for standard NVS page structure """
+class Page(object):
+    PAGE_PARAMS = {
+        "max_size": 4096,
+        "max_blob_size": 1984,
+        "max_entries": 126
+    }
+
+    # Item type codes
+    U8   = 0x01
+    I8   = 0x11
+    U16  = 0x02
+    I16  = 0x12
+    U32  = 0x04
+    I32  = 0x14
+    SZ   = 0x21
+    BLOB = 0x41
+
+    # Few Page constants
+    HEADER_SIZE = 32
+    BITMAPARRAY_OFFSET = 32
+    BITMAPARRAY_SIZE_IN_BYTES = 32
+    FIRST_ENTRY_OFFSET = 64
+    SINGLE_ENTRY_SIZE = 32
+
+    def __init__(self, page_num):
+        self.entry_num = 0
+        self.bitmap_array = array.array('B')
+        self.page_buf = bytearray(b'\xff')*Page.PAGE_PARAMS["max_size"]
+        self.bitmap_array = self.create_bitmap_array()
+        self.set_header(page_num)
+
+    def set_header(self, page_num):
+        # set page state to active
+        page_header= bytearray(b'\xff')*32
+        page_state_active_seq = 0xFFFFFFFE
+        page_header[0:4] = struct.pack('<I', page_state_active_seq)
+        # set page sequence number
+        page_header[4:8] = struct.pack('<I', page_num)
+        # set header's CRC
+        crc_data = page_header[4:28]
+        crc = zlib.crc32(buffer(crc_data), 0xFFFFFFFF)
+        page_header[28:32] = struct.pack('<I', crc & 0xFFFFFFFF)
+        self.page_buf[0:len(page_header)] = page_header
+
+    def create_bitmap_array(self):
+        bitarray = array.array('B')
+        charsize = 32 # bitmaparray has 256 bits, hence 32 bytes
+        fill = 255 # Fill all 8 bits with 1's
+        bitarray.extend((fill,) * charsize)
+        return bitarray
+
+    def write_bitmaparray(self):
+        bitnum = self.entry_num * 2
+        byte_idx = bitnum / 8 # Find byte index in the array
+        bit_offset = bitnum & 7 # Find bit offset in given byte index
+        mask = ~(1 << bit_offset)
+        self.bitmap_array[byte_idx] &= mask
+        start_idx = Page.BITMAPARRAY_OFFSET
+        end_idx = Page.BITMAPARRAY_OFFSET + Page.BITMAPARRAY_SIZE_IN_BYTES
+        self.page_buf[start_idx:end_idx] = self.bitmap_array
+
+    def write_entry_to_buf(self, data, entrycount):
+        data_offset = Page.FIRST_ENTRY_OFFSET + (Page.SINGLE_ENTRY_SIZE * self.entry_num)
+        start_idx = data_offset
+        end_idx = data_offset + len(data)
+        self.page_buf[start_idx:end_idx]  = data
+
+        # Set bitmap array for entries in current page
+        for i in range(0, entrycount):
+            self.write_bitmaparray()
+            self.entry_num += 1
+
+    """
+    Low-level function to write variable length data into page buffer. Data should be formatted
+    according to encoding specified.
+    """
+    def write_varlen_data(self, key, data, encoding, ns_index):
+        datalen = len(data)
+        if datalen > Page.PAGE_PARAMS["max_blob_size"]:
+            raise InputError("%s: Size exceeds max allowed length." % key)
+
+        # Calculate no. of entries data will require
+        rounded_size = (datalen + 31) & ~31
+        data_entry_count = rounded_size / 32
+        total_entry_count = data_entry_count + 1 # +1 for the entry header
+
+        # Check if page is already full and new page is needed to be created right away
+        if (self.entry_num + total_entry_count) >= Page.PAGE_PARAMS["max_entries"]:
+            raise PageFullError()
+
+        # Entry header
+        entry_struct = bytearray('\xff')*32
+        entry_struct[0] = ns_index # namespace index
+        entry_struct[2] = data_entry_count + 1 # Span
+
+        # set key
+        key_array = bytearray('\x00')*16
+        entry_struct[8:24] = key_array
+        entry_struct[8:8 + len(key)] = key
+
+        # set Type
+        if encoding == "string":
+            entry_struct[1] = Page.SZ
+        elif encoding == "hex2bin" or encoding == "binary":
+            entry_struct[1] = Page.BLOB
+
+        # compute CRC of data
+        entry_struct[24:26] = struct.pack('<H', datalen)
+        crc = zlib.crc32(data, 0xFFFFFFFF)
+        entry_struct[28:32] = struct.pack('<I', crc & 0xFFFFFFFF)
+
+        # compute crc of entry header
+        crc_data = bytearray(28)
+        crc_data[0:4] = entry_struct[0:4]
+        crc_data[4:28] = entry_struct[8:32]
+        crc = zlib.crc32(buffer(crc_data), 0xFFFFFFFF)
+        entry_struct[4:8] = struct.pack('<I', crc & 0xFFFFFFFF)
+
+        # write entry header
+        self.write_entry_to_buf(entry_struct, 1)
+        # write actual data
+        self.write_entry_to_buf(data, data_entry_count)
+
+    """ Low-level function to write data of primitive type into page buffer. """
+    def write_primitive_data(self, key, data, encoding, ns_index):
+        # Check if entry exceeds max number of entries allowed per page
+        if self.entry_num >= Page.PAGE_PARAMS["max_entries"]:
+            raise PageFullError()
+
+        entry_struct = bytearray('\xff')*32
+        entry_struct[0] = ns_index # namespace index
+        entry_struct[2] = 0x01 # Span
+
+        # write key
+        key_array = bytearray('\x00')*16
+        entry_struct[8:24] = key_array
+        entry_struct[8:8 + len(key)] = key
+
+        if encoding == "u8":
+            entry_struct[1] = Page.U8
+            entry_struct[24] = struct.pack('<B', data)
+        elif encoding == "i8":
+            entry_struct[1] = Page.I8
+            entry_struct[24] = struct.pack('<b', data)
+        elif encoding == "u16":
+            entry_struct[1] = Page.U16
+            entry_struct[24:26] = struct.pack('<H', data)
+        elif encoding == "u32":
+            entry_struct[1] = Page.U32
+            entry_struct[24:28] = struct.pack('<I', data)
+        elif encoding == "i32":
+            entry_struct[1] = Page.I32
+            entry_struct[24:28] = struct.pack('<i', data)
+
+        # Compute CRC
+        crc_data = bytearray(28)
+        crc_data[0:4] = entry_struct[0:4]
+        crc_data[4:28] = entry_struct[8:32]
+        crc = zlib.crc32(buffer(crc_data), 0xFFFFFFFF)
+        entry_struct[4:8] = struct.pack('<I', crc & 0xFFFFFFFF)
+
+        # write to file
+        self.write_entry_to_buf(entry_struct, 1)
+
+    """ Get page buffer data of a given page """
+    def get_data(self):
+        return self.page_buf
+
+"""
+NVS class encapsulates all NVS specific operations to create a binary with given key-value pairs. Binary can later be flashed onto device via a flashing utility.
+"""
+class NVS(object):
+    def __init__(self, fout):
+        self.namespace_idx = 0
+        self.page_num = -1
+        self.pages = []
+        self.cur_page = self.create_new_page()
+        self.fout = fout
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        if exc_type == None and exc_value == None:
+            result = self.get_binary_data()
+            self.fout.write(result)
+
+    def create_new_page(self):
+        self.page_num += 1
+        new_page = Page(self.page_num)
+        self.pages.append(new_page)
+        self.cur_page = new_page
+        return new_page
+
+    """
+    Write namespace entry and subsequently increase namespace count so that all upcoming entries
+    will be mapped to a new namespace.
+    """
+    def write_namespace(self, key):
+        self.namespace_idx += 1
+        try:
+            self.cur_page.write_primitive_data(key, self.namespace_idx, "u8", 0)
+        except PageFullError:
+            new_page = self.create_new_page()
+            new_page.write_primitive_data(key, self.namespace_idx, "u8", 0)
+            pass
+
+    """
+    Write key-value pair. Function accepts value in the form of ascii character and converts
+    it into appropriate format before calling Page class's functions to write entry into NVS format.
+    Function handles PageFullError and creates a new page and re-invokes the function on a new page.
+    We don't have to guard re-invocation with try-except since no entry can span multiple pages.
+    """
+    def write_entry(self, key, value, encoding):
+        if encoding == "hex2bin":
+            if len(value) % 2 != 0:
+                raise InputError("%s: Invalid data length. Should be multiple of 2." % key)
+            value = binascii.a2b_hex(value)
+
+        if encoding == "string":
+            value += '\0'
+
+        encoding = encoding.lower()
+        varlen_encodings = ["string", "binary", "hex2bin"]
+        primitive_encodings = ["u8", "i8", "u16", "u32", "i32"]
+        if encoding in varlen_encodings:
+            try:
+                self.cur_page.write_varlen_data(key, value, encoding, self.namespace_idx)
+            except PageFullError:
+                new_page = self.create_new_page()
+                new_page.write_varlen_data(key, value, encoding, self.namespace_idx)
+                pass
+        elif encoding in primitive_encodings:
+            try:
+                self.cur_page.write_primitive_data(key, int(value), encoding, self.namespace_idx)
+            except PageFullError:
+                new_page = self.create_new_page()
+                new_page.write_primitive_data(key, int(value), encoding, self.namespace_idx)
+                sys.exc_clear()
+                pass
+        else:
+            raise InputError("%s: Unsupported encoding" % encoding)
+
+    """ Return accumulated data of all pages """
+    def get_binary_data(self):
+        data = bytearray()
+        for page in self.pages:
+            data += page.get_data()
+        return data
+
+class PageFullError(RuntimeError):
+    """
+    Represents error when current page doesn't have sufficient entries left
+    to accommodate current request
+    """
+    def __init__(self):
+       super(PageFullError, self).__init__()
+
+class InputError(RuntimeError):
+    """
+    Represents error on the input
+    """
+    def __init__(self, e):
+        super(InputError, self).__init__(e)
+
+def nvs_open(result_obj):
+    """ Wrapper to create and NVS class object. This object can later be used to set key-value pairs
+
+    :param result_obj: File/Stream object to dump resultant binary. If data is to be dumped into memory, one way is to use BytesIO object
+    :return: NVS class instance
+    """
+    return NVS(result_obj)
+
+def write_entry(nvs_instance, key, datatype, encoding, value):
+    """ Wrapper to set key-value pair in NVS format
+
+    :param nvs_instance: Instance of an NVS class returned by nvs_open()
+    :param key: Key of the data
+    :param datatype: Data type. Valid values are "file", "data" and "namespace"
+    :param encoding: Data encoding. Valid values are "u8", "i8", "u16", "u32", "i32", "string", "binary" and "hex2bin"
+    :param value: Data value in ascii encoded string format for "data" datatype and filepath for "file" datatype
+    :return: None
+    """
+    if datatype == "file":
+        abs_file_path = value
+        if os.path.isabs(value) == False:
+            script_dir = os.path.dirname(__file__)
+            abs_file_path = os.path.join(script_dir, value)
+        with open(abs_file_path, 'rb') as f:
+            value = f.read()
+
+    if datatype == "namespace":
+        nvs_instance.write_namespace(key)
+    else:
+        nvs_instance.write_entry(key, value, encoding)
+
+def nvs_close(nvs_instance):
+    """ Wrapper to finish writing to NVS and write data to file/stream object provided to nvs_open method
+
+    :param nvs_instance: Instance of NVS class returned by nvs_open()
+    :return: None
+    """
+    nvs_instance.__exit__(None, None, None)
+
+def main():
+    parser = argparse.ArgumentParser(description="ESP32 NVS partition generation utility")
+    parser.add_argument(
+            "input",
+            help="Path to CSV file to parse. Will use stdin if omitted",
+            type=argparse.FileType('rb'),
+            default=sys.stdin)
+
+    parser.add_argument(
+            "output",
+            help='Path to output converted binary file. Will use stdout if omitted',
+            type=argparse.FileType('wb'),
+            default=sys.stdout)
+
+    args = parser.parse_args()
+    with nvs_open(args.output) as nvs_obj:
+        reader = csv.DictReader(args.input, delimiter=',')
+        for row in reader:
+            try:
+                write_entry(nvs_obj, row["key"], row["type"], row["encoding"], row["value"])
+            except InputError as e:
+                print(e)
+                exit(-2)
+
+if __name__ == "__main__":
+    main()

components/nvs_flash/nvs_partition_generator/sample.csv

@@ -0,0 +1,12 @@
+key,type,encoding,value
+dummyNamespace,namespace,,
+dummyU8Key,data,u8,127
+dummyI8Key,data,i8,-128
+dummyU16Key,data,u16,32768
+dummyU32Key,data,u32,4294967295
+dummyI32Key,data,i32,-2147483648
+dummyStringKey,data,string,0A:0B:0C:0D:0E:0F
+dummyHex2BinKey,data,hex2bin,010203abcdef
+hexFileKey,file,hex2bin,testdata/sample.hex
+stringFileKey,file,string,testdata/sample.txt
+binFileKey,file,binary,testdata/sample.bin

components/nvs_flash/nvs_partition_generator/testdata/sample.bin

@@ -0,0 +1 @@
+
	<09><><EFBFBD>

components/nvs_flash/nvs_partition_generator/testdata/sample.hex

@@ -0,0 +1 @@
+0123456789abcdef

components/nvs_flash/nvs_partition_generator/testdata/sample.txt

@@ -0,0 +1 @@
+abcdefghijklmnopqrstuvwxyz

components/nvs_flash/test_nvs_host/spi_flash_emulation.h

@@ -39,6 +39,14 @@ public:
         spi_flash_emulator_set(this);
     }
 
+    SpiFlashEmulator(const char *filename)
+    {
+        load(filename);
+        // Atleast one page should be free, hence we create mData of size of 2 sectors.
+        mData.resize(2 * SPI_FLASH_SEC_SIZE / 4, 0xffffffff);
+        spi_flash_emulator_set(this);
+    }
+
     ~SpiFlashEmulator()
     {
         spi_flash_emulator_set(nullptr);

components/nvs_flash/test_nvs_host/test_nvs.cpp

@@ -17,6 +17,8 @@
 #include "spi_flash_emulation.h"
 #include <sstream>
 #include <iostream>
+#include <unistd.h>
+#include <sys/wait.h>
 
 #define TEST_ESP_ERR(rc, res) CHECK((rc) == (res))
 #define TEST_ESP_OK(rc) CHECK((rc) == ESP_OK)
@@ -1452,6 +1454,53 @@ TEST_CASE("Recovery from power-off when the entry being erased is not on active
 /* Add new tests above */
 /* This test has to be the final one */
 
+TEST_CASE("check partition generation utility", "[nvs_part_gen]")
+{
+    int childpid = fork();
+    if (childpid == 0) {
+        exit(execlp("python", "python",
+                "../nvs_partition_generator/nvs_partition_gen.py",
+                "../nvs_partition_generator/sample.csv",
+                "../nvs_partition_generator/partition.bin", NULL));
+    } else {
+        CHECK(childpid > 0);
+        int status;
+        waitpid(childpid, &status, 0);
+        CHECK(WEXITSTATUS(status) != -1);
+    }
+}
+
+TEST_CASE("read data from partition generated via partition generation utility", "[nvs_part_gen]")
+{
+    SpiFlashEmulator emu("../nvs_partition_generator/partition.bin");
+    nvs_handle handle;
+    TEST_ESP_OK( nvs_flash_init_custom("test", 0, 2) );
+    TEST_ESP_OK( nvs_open_from_partition("test", "dummyNamespace", NVS_READONLY, &handle));
+    uint8_t u8v;
+    TEST_ESP_OK( nvs_get_u8(handle, "dummyU8Key", &u8v));
+    CHECK(u8v == 127);
+    int8_t i8v;
+    TEST_ESP_OK( nvs_get_i8(handle, "dummyI8Key", &i8v));
+    CHECK(i8v == -128);
+    uint16_t u16v;
+    TEST_ESP_OK( nvs_get_u16(handle, "dummyU16Key", &u16v));
+    CHECK(u16v == 32768);
+    uint32_t u32v;
+    TEST_ESP_OK( nvs_get_u32(handle, "dummyU32Key", &u32v));
+    CHECK(u32v == 4294967295);
+    int32_t i32v;
+    TEST_ESP_OK( nvs_get_i32(handle, "dummyI32Key", &i32v));
+    CHECK(i32v == -2147483648);
+    char buf[64] = {0};
+    size_t buflen = 64;
+    TEST_ESP_OK( nvs_get_str(handle, "dummyStringKey", buf, &buflen));
+    CHECK(strncmp(buf, "0A:0B:0C:0D:0E:0F", buflen) == 0);
+    buflen = 64;
+    uint8_t hexdata[] = {0x01, 0x02, 0x03, 0xab, 0xcd, 0xef};
+    TEST_ESP_OK( nvs_get_blob(handle, "dummyHex2BinKey", buf, &buflen));
+    CHECK(memcmp(buf, hexdata, buflen) == 0);
+}
+
 TEST_CASE("dump all performance data", "[nvs]")
 {
     std::cout << "====================" << std::endl << "Dumping benchmarks" << std::endl;

docs/en/api-reference/storage/index.rst

@@ -7,6 +7,7 @@ Storage API
    SPI Flash and Partition APIs <spi_flash>
    SD/SDIO/MMC Driver <sdmmc>
    Non-Volatile Storage <nvs_flash>
+   NVS Partition Generation Utility <nvs_partition_gen.rst>
    Virtual Filesystem <vfs>
    FAT Filesystem <fatfs>
    Wear Levelling <wear-levelling>

docs/en/api-reference/storage/nvs_flash.rst

@@ -1,5 +1,10 @@
 .. include:: ../../../../components/nvs_flash/README.rst
 
+NVS Partition Generator Utility
+-------------------------------
+
+This utility helps in generating NVS-esque partition binary file which can be flashed separately on a dedicated partition via a flashing utility. Key-value pairs to be flashed onto the partition can be provided via a CSV file. Refer to :doc:`NVS Partition Generator Utility <nvs_partition_gen>` for more details.
+
 Application Example
 -------------------
 

docs/en/api-reference/storage/nvs_partition_gen.rst

@@ -0,0 +1 @@
+.. include:: /../../components/nvs_flash/nvs_partition_generator/README.rst

docs/zh_CN/api-reference/storage/nvs_partition_gen.rst

@@ -0,0 +1 @@
+.. include:: /../../components/nvs_flash/nvs_partition_generator/README.rst