NAND Flash Memories Application Note
NAND Flash Memories Using Multiply Partition (with Skip IB) on ELNEC Universal Device Programmers (Quick Guide)
Application Note
May 2011 an_elnec_nand_partitions, version 1.07
Version 1.07/05.2011
Page 1 of 16
NAND Flash Memories Application Note
Increasing demands on embedded devices involve also increased complexity of programming tasks. Embedded memory of today does not store just simple piece of firmware. It carries various types of data used by modern operating system driven appliances. Typically, there are some kind of boot-loader, operating system executables and user data stored in single-chip memory unit. Within NAND flash memory devices, management of these mixed-type data is considerably impacted by invalid blocks presence. Correct partitioning of NAND flash memory therefore becomes crucial element of many embedded devices designs. ELNEC programmers are able to cope with NAND flash partitioning tasks. This application note is aimed to help you to understand how to use their features to fulfil your needs. Actually, NAND flash partitioning is supported on BeeProg2, BeeHive204, BeeHive204AP and BeeHive208S programmers. This application note applies to area under dynamical development. Please, check our application notes section (see http://www.elnec.com/support/application-notes/) regularly and always use the newest version of the document. If you recognize the apparatus described in this application note not suitable for your demand, you can consider also using our Multi-Project feature described in separate application note (see http://www.elnec.com/sw/an_elnec_multi_prj.pdf).
Version 1.07/05.2011
Page 2 of 16
NAND Flash Memories Application Note
OVERVIEW
Figure 1. Access Method dialog window (<Alt+S>) with Multiply Partition set
Version 1.07/05.2011
Page 3 of 16
NAND Flash Memories Application Note
To enable Multi-Partition feature, open Access Method dialog window (easy accessible through clicking the link in Device panel in bottom right of pg4uw software window, or shortcut <Alt+S>) and set Invalid Block Management option to Multiply Partition (with Skip IB).
ACCESS METHOD OPTIONS VALIDITY
With Skip IB means that invalid blocks are skipped and next valid blocks are used instead. This is the only invalid blocks management method available in partition and is used automatically.
Setting of Spare Area Usage option is ignored. Spare area is always used and input data file(s) must contain correct data for spare area. Spare area is programmed using method corresponding to User Data setting. If your input data file doesn't contain spare area data (e.g. your system doesn't use spare area), you are still required to add empty (blank) spare area data. You can do it on your own, or you can use Add blank spare area feature available in Load file dialog window.
Setting of Required Good Blocks Area is accepted and is applied globally (i. e. over whole device, not over individual partitions).
Setting of Max. Allowed Number of Invalid Blocks in Device is accepted and is applied globally.
Setting of Quick Program option is accepted and is applied globally.
Reserved Block Area Options are irrelevant when skipping invalid blocks and respective setting is not taken into account.
Setting of Invalid Block Indication Options is accepted and is applied globally.
Setting of Tolerant Verification Options is accepted and is applied globally.
Version 1.07/05.2011
Page 4 of 16
NAND Flash Memories Application Note
OPERATIONS EXECUTION
Blank check
The device is always blank checked entirely, from the first up to the last address location, including spare area. If Blank check before programming is enabled (and is not skipped due to enabling both, Blank check and Erase before programming), it will be performed in the same way, not taking partitioning into account.
Read
Read operation is performed on per partition basis, i. e. individual partitions are processed step by step, from the first partition up to the last one. Note: Partition unused (padding) area is not read. Buffer data at corresponding locations are not altered.
Verify
Verify operation is performed on per partition basis, i. e. individual partitions are processed step by step, from the first partition up to the last one. If Verify after reading or Verify after programming is used, the partition will be read or programmed firstly, and after then it will be verified, see Figure 2. Note: Partition unused (padding) area is not verified. Buffer data at corresponding locations are ignored.
Program
Program operation is performed on per partition basis, i. e. individual partitions are processed step by step, from the first partition up to the last one. See Figure 2 for programming round example. Note: Partition unused (padding) area is not programmed. Buffer data at corresponding locations are ignored.
Version 1.07/05.2011
Page 5 of 16
NAND Flash Memories Application Note
Erase
The device is always erased entirely, from the first up to the last block If Erase before programming is enabled, it will be performed in the same way, not taking partitioning into account.
Figure 2. Program operation round example
Version 1.07/05.2011
Page 6 of 16
NAND Flash Memories Application Note
GENERAL PROCEDURE DESCRIPTION
Setting-up the Multi-Partition mode consists of three basic steps:
Selection of Multiply Partition
Open Access Method dialog window and set Invalid Block Management option to Multiply Partition (with Skip IB), see Figure 1. Set other options, if necessary. After clicking OK, internal software environment will be prepared for Multi-Partition mode. Also, special buffer for Partition Table will be created, see Figure 3. The buffer is not editable, the only way how to define partition table is loading relevant definition file. After Multiply Partition method is deselected, internal software environment is switched back to standard mode and special buffer is destroyed. In consequence, partition table data are lost and must be load again after reenabling Multi-Partition mode again.
Figure 3. Partition Table buffer example
Version 1.07/05.2011
Page 7 of 16
NAND Flash Memories Application Note
Loading partition table definition file
After enabling Multi-Partition mode, the partition table must be defined. There is only one way how to define the partition table - using menu File >> Load Partition table..., see Figure 4, Figure 5 and Figure 6. However, there are several partition definition file formats supported, see later chapters for more details.
Figure 4. Load Partition table menu
Figure 5. Load Partition table dialog
Version 1.07/05.2011
Page 8 of 16
NAND Flash Memories Application Note
Figure 6. Partition table definition file load example
Loading data image file
If relevant chapter intended to individual partition definition file formats doesn't specify otherwise, data image file should correspond with a copy of NAND flash device without any invalid blocks. It must contain data for all partitions placed at correct locations. Also, it must contain spare area data (or you can add spare area data automatically on file load using Add blank spare area feature).
To load input data image file, use standard Load command (menu File >> Load, shortcut <F3> or Load command from Main toolbar).
Version 1.07/05.2011
Page 9 of 16
NAND Flash Memories Application Note
Figure 7: Load file dialog window
If you haven't a single input data image file but you have several input data files for individual partitions instead, you will need to use Positive offset setting in File load dialog. You can compute the offset using the following formula:
positive_offset = partition_start_block_number x number_of_pages_in_block x page_size_including_spare
In this way, you can load multiple files into buffer, each one at respective offset corresponding to partition start block address offset. Example: Partition should start from block 30, devices has 64 pages in block and one page contains 2048 bytes of data area and 64 bytes of spare area. positive_offset = 30 x 64 x (2048+64) = 4055040dec = 3DE000h
Version 1.07/05.2011
Page 10 of 16
NAND Flash Memories Application Note
USING QUALCOMM MULTIPLY PARTITION FORMAT (*.MBN)
Generally, there are two versions of Qualcomm Multiply Partition format used. They can be simply distinguished by the number of programming files: Procedure for two input files (Qualcomm 80-VE594-1 B) If you have two input files available, they are generally named FactoryImage.bin and PartitionTable.mbn. PartitionTable.mbn is rather small (256 bytes) and contains partition table definition. Load this file using menu File >> Load Partition table..., see chapter Loading partition table definition file. FactoryImage.bin may be rather huge and contains binary data image. Load this file using standard Load procedure, see chapter Loading data image file. It is possible to save your data using this format. To save buffer content in binary format, use standard Save procedure (menu File >> Save, shortcut <F2> or Save command from Main toolbar). To save partition table in Qualcomm Multiply Partition compatible form, use menu File >> Save Partition table..., see Figure 4. Procedure for single input file (Qualcomm 80-VF498-1 A) If you have single input file available, it is generally named FactoryImage2.mbn. The file is rather huge and contains both, partition table definition and binary data image, plus header. The file can be simply identified using hex-viewer: you must identify text Image file with header at file start. The header specifies also block validity indication byte position. This parameter is also accepted and used for proper reading and/or verifying the device. The value overwrites manual setting in Invalid Block Indication Options section of Access Method dialog.1 Load this file using standard Load procedure, see chapter Loading data image file. It is not possible to save your data using this form.
Note: Qualcomm Multiply Partition format was invented by Qualcomm Incorporated (U.S.A.), not by ELNEC. The owner of all potential legal rights is Qualcomm Incorporated. Please, contact Qualcomm CDMA Technologies (http://www.qctconnect.com/) for technical specification.
1 The support released in pg4uw version 2.66. Version 1.07/05.2011 Page 11 of 16
NAND Flash Memories Application Note
USING COMMA SEPARATED VALUES FORMAT (*.CSV)
This mode uses two input files. You can simply prepare both of them on your own: Partition table definition file Partition table definition file uses widely used comma separated values file format. The file should contain the number of rows corresponding to the number of partitions. Each row specifies one partition. Values in row should be separated by separator - comma (,) or semicolon (;) can be used. Space characters (ASCII code 0x20) are ignored and should not be used in place of values separator. Each row should contain several values (both, decimal and/or hexadecimal values can be used):
partition start (mandatory) - specifies the block in device, where partition should start. Enter the block number here.
partition end (mandatory) - specifies the block in device, where partition should end. Enter the block number here.
used partition size (mandatory) - specifies the number of blocks really occupied by partition data. Typically, there are some reserve blocks added for invalid blocks replacement, therefore partition end - partition start > used partition size. Enter number of blocks here.
special options/reserved (optional/mandatory) - this value enables to specify some special options. If you use it just due to comment option, enter the value of 0xFFFFFFFF (4 bytes size) here to ensure future compatibility. Special options format specification: MSB(bit 31) LSB(bit 0)
xxxx.xxxx.xxxx.xxxx.xxxx.xxxx.xxxx.xxxx bits 11:0 - Maximum allowed number of invalid blocks in partition: 0xFFF = option disabled any other value specifies the number of invalid blocks that can be accepted in partition bits 15:12 - Invalid blocks management method: 0x0 = Treat All Blocks 0x1 or 0xFF = Skip IB (default) 0x2 = Skip IB with Excess Abandon 0x3 = Check IB without Access Note: Check IB with Skip IB can be specified using Skip IB (0x1 or 0xF) method and non 0xFFF value for Max. allowed number of invalid blocks in partition. bits 23:16 - reserved for future use, consider 0xFF value for future compatibility
Version 1.07/05.2011
Page 12 of 16
NAND Flash Memories Application Note bits 31:24 - File system preparation: 0xFF = option not used 0x00 = JFFS2 Clean Markers using MSB byte ordering (big endian) 0x01 = JFFS2 Clean Markers using LSB byte ordering (little endian) Using values not specified here will cause partition table load error.
comment (optional) - you can enter any text here. Primarily, this item is intended for your notes that will help you to orientate in the file. It may contain e. g. partition name. If you use comments, reserved item must be also used.
Partition table definition file listening example: 0;100;20;0xffffffff;boot 101;200;50;0xffffffff;exec 201;300;0;0xffffffff;res1 301;400;50;0xffffffff;fsys 401;500;0;0xffffffff;res2 501;1000;50;0xffffffff;data
Load this partition table definition file using menu File >> Load Partition table..., see chapter Loading partition table definition file. It is possible to save your partition table definition using this format. To save the partition table data, use menu File >> Save Partition table..., see Figure 4. The table is saved using all items in raw. A partition number is used for comment.
Saved partition table definition file listening, using the above example: 0;100;20;0xFFFFFFFF;Partition 0 101;200;50;0xFFFFFFFF;Partition 1 201;300;0;0xFFFFFFFF;Partition 2 301;400;50;0xFFFFFFFF;Partition 3 401;500;0;0xFFFFFFFF;Partition 4 501;1000;50;0xFFFFFFFF;Partition 5
Data image file Input data image file should be a binary file (recommended) that meets all requirements specified in chapter Loading data image file. Use standard Load procedure to load this file (see chapter Loading data image file).
Version 1.07/05.2011
Page 13 of 16
NAND Flash Memories Application Note
USING GROUP DEFINE FILE FORMAT (*.DEF)
Important note! The support of this format is implemented based only on fragment of specification available from customer. Therefore it cannot be considered full and reliable. We do not recommend to use it, until you exactly know what you do. If you observe any problems, please, contact our technical support with full Group Define file format specification.
Partition table definition file This file consists of file header and group records. Each group record specifies one partition.
Load this partition table definition file using menu File >> Load Partition table..., see chapter Loading partition table definition file. It is possible to save your partition table definition using this format. To save the partition table data, use menu File >> Save Partition table..., see Figure 4.
Data image file Input data image file should be a binary file (recommended) that meets all requirements specified in chapter Loading data image file. Use standard Load procedure to load this file (see chapter Loading data image file).
Version 1.07/05.2011
Page 14 of 16
NAND Flash Memories Application Note
ERROR CODES
During input file(s) loading, several errors can occur. Errors are always displayed in pg4uw log-window. Error message consists of error code and error description in the following form: File loading problem! Error code: #xxyy - error description where xx stands for file format: 00 - binary file 01 - *.mbn file containing just partition table 02 - *.mbn file containing both, partition table as well as partitions data 03 - *.csv file containing partition table 04 - reserved for future use 05 - *.def file containing partition table and yy stands for error type: 10 - disk i/o error (disk i/o error, file access error, ...) 11 - maximum buffer limit exceeded (file size grater than max. supported buffer size 12 - unable to re-allocate buffer (file size is grater than buffer size and there is some problem when reallocating the buffer (e. g. not enough disk space)) 13 - unknown separator (for *.csv files, nor comma (,) nor semicolon (;) were detected as separator) 14 - file does not specify any partition (none partition definition read) 15 - too many partitions specified in file (max. 16 partitions are supported for Qualcomm Multiply Partition, max. 64 partitions generally) 16 - incorrect numeric values format (typo in text-oriented files (*.csv), e. g. @34 instead of 234) 17 - version not supported (not supported version of algorithm specification detected) 18 - invalid file header (damaged header, some mandatory item missing, ...)
Version 1.07/05.2011
Page 15 of 16
NAND Flash Memories Application Note
HISTORY
Version 1.00 - December 2009
initial release
Version 1.01 - January 2010
block validity indication byte position acceptance for Qualcomm Multiply Partition added Error codes section added
Version 1.02 - February 2010
Add blank spare area feature incorporated
Version 1.03 - June 2010
Special options added for CSV partition definition file (max. allowed number of invalid blocks in partition, invalid blocks management method, file system preparation)
Version 1.04 - August 2010
Qualcomm Multiply Partition format specification refined
Version 1.05 - September 2010
added passage on loading multiple input data files
Version 1.06 - February 2011
some minor changes in values alignment for CSV format description
Version 1.07 - May 2011
Group Define file format added
Version 1.07/05.2011
Page 16 of 16
