1<?xml version="1.0" encoding="UTF-8"?>
2<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"
3	"http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd" []>
4
5<book id="MTD-NAND-Guide">
6 <bookinfo>
7  <title>MTD NAND Driver Programming Interface</title>
8  
9  <authorgroup>
10   <author>
11    <firstname>Thomas</firstname>
12    <surname>Gleixner</surname>
13    <affiliation>
14     <address>
15      <email>tglx@linutronix.de</email>
16     </address>
17    </affiliation>
18   </author>
19  </authorgroup>
20
21  <copyright>
22   <year>2004</year>
23   <holder>Thomas Gleixner</holder>
24  </copyright>
25
26  <legalnotice>
27   <para>
28     This documentation is free software; you can redistribute
29     it and/or modify it under the terms of the GNU General Public
30     License version 2 as published by the Free Software Foundation.
31   </para>
32      
33   <para>
34     This program is distributed in the hope that it will be
35     useful, but WITHOUT ANY WARRANTY; without even the implied
36     warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
37     See the GNU General Public License for more details.
38   </para>
39      
40   <para>
41     You should have received a copy of the GNU General Public
42     License along with this program; if not, write to the Free
43     Software Foundation, Inc., 59 Temple Place, Suite 330, Boston,
44     MA 02111-1307 USA
45   </para>
46      
47   <para>
48     For more details see the file COPYING in the source
49     distribution of Linux.
50   </para>
51  </legalnotice>
52 </bookinfo>
53
54<toc></toc>
55
56  <chapter id="intro">
57      <title>Introduction</title>
58  <para>
59  	The generic NAND driver supports almost all NAND and AG-AND based
60	chips and connects them to the Memory Technology Devices (MTD)
61	subsystem of the Linux Kernel.
62  </para>
63  <para>
64  	This documentation is provided for developers who want to implement
65	board drivers or filesystem drivers suitable for NAND devices.
66  </para>
67  </chapter>
68  
69  <chapter id="bugs">
70     <title>Known Bugs And Assumptions</title>
71  <para>
72	None.	
73  </para>
74  </chapter>
75
76  <chapter id="dochints">
77     <title>Documentation hints</title>
78     <para>
79     The function and structure docs are autogenerated. Each function and 
80     struct member has a short description which is marked with an [XXX] identifier.
81     The following chapters explain the meaning of those identifiers.
82     </para>
83     <sect1 id="Function_identifiers_XXX">
84	<title>Function identifiers [XXX]</title>
85     	<para>
86	The functions are marked with [XXX] identifiers in the short
87	comment. The identifiers explain the usage and scope of the
88	functions. Following identifiers are used:
89     	</para>
90	<itemizedlist>
91		<listitem><para>
92	  	[MTD Interface]</para><para>
93		These functions provide the interface to the MTD kernel API. 
94		They are not replaceable and provide functionality
95		which is complete hardware independent.
96		</para></listitem>
97		<listitem><para>
98	  	[NAND Interface]</para><para>
99		These functions are exported and provide the interface to the NAND kernel API. 
100		</para></listitem>
101		<listitem><para>
102	  	[GENERIC]</para><para>
103		Generic functions are not replaceable and provide functionality
104		which is complete hardware independent.
105		</para></listitem>
106		<listitem><para>
107	  	[DEFAULT]</para><para>
108		Default functions provide hardware related functionality which is suitable
109		for most of the implementations. These functions can be replaced by the
110		board driver if necessary. Those functions are called via pointers in the
111		NAND chip description structure. The board driver can set the functions which
112		should be replaced by board dependent functions before calling nand_scan().
113		If the function pointer is NULL on entry to nand_scan() then the pointer
114		is set to the default function which is suitable for the detected chip type.
115		</para></listitem>
116	</itemizedlist>
117     </sect1>
118     <sect1 id="Struct_member_identifiers_XXX">
119	<title>Struct member identifiers [XXX]</title>
120     	<para>
121	The struct members are marked with [XXX] identifiers in the 
122	comment. The identifiers explain the usage and scope of the
123	members. Following identifiers are used:
124     	</para>
125	<itemizedlist>
126		<listitem><para>
127	  	[INTERN]</para><para>
128		These members are for NAND driver internal use only and must not be
129		modified. Most of these values are calculated from the chip geometry
130		information which is evaluated during nand_scan().
131		</para></listitem>
132		<listitem><para>
133	  	[REPLACEABLE]</para><para>
134		Replaceable members hold hardware related functions which can be 
135		provided by the board driver. The board driver can set the functions which
136		should be replaced by board dependent functions before calling nand_scan().
137		If the function pointer is NULL on entry to nand_scan() then the pointer
138		is set to the default function which is suitable for the detected chip type.
139		</para></listitem>
140		<listitem><para>
141	  	[BOARDSPECIFIC]</para><para>
142		Board specific members hold hardware related information which must
143		be provided by the board driver. The board driver must set the function
144		pointers and datafields before calling nand_scan().
145		</para></listitem>
146		<listitem><para>
147	  	[OPTIONAL]</para><para>
148		Optional members can hold information relevant for the board driver. The
149		generic NAND driver code does not use this information.
150		</para></listitem>
151	</itemizedlist>
152     </sect1>
153  </chapter>   
154
155  <chapter id="basicboarddriver">
156     	<title>Basic board driver</title>
157	<para>
158		For most boards it will be sufficient to provide just the
159		basic functions and fill out some really board dependent
160		members in the nand chip description structure.
161	</para>
162	<sect1 id="Basic_defines">
163		<title>Basic defines</title>
164		<para>
165			At least you have to provide a mtd structure and
166			a storage for the ioremap'ed chip address.
167			You can allocate the mtd structure using kmalloc
168			or you can allocate it statically.
169			In case of static allocation you have to allocate
170			a nand_chip structure too.
171		</para>
172		<para>
173			Kmalloc based example
174		</para>
175		<programlisting>
176static struct mtd_info *board_mtd;
177static void __iomem *baseaddr;
178		</programlisting>
179		<para>
180			Static example
181		</para>
182		<programlisting>
183static struct mtd_info board_mtd;
184static struct nand_chip board_chip;
185static void __iomem *baseaddr;
186		</programlisting>
187	</sect1>
188	<sect1 id="Partition_defines">
189		<title>Partition defines</title>
190		<para>
191			If you want to divide your device into partitions, then
192			define a partitioning scheme suitable to your board.
193		</para>
194		<programlisting>
195#define NUM_PARTITIONS 2
196static struct mtd_partition partition_info[] = {
197	{ .name = "Flash partition 1",
198	  .offset =  0,
199	  .size =    8 * 1024 * 1024 },
200	{ .name = "Flash partition 2",
201	  .offset =  MTDPART_OFS_NEXT,
202	  .size =    MTDPART_SIZ_FULL },
203};
204		</programlisting>
205	</sect1>
206	<sect1 id="Hardware_control_functions">
207		<title>Hardware control function</title>
208		<para>
209			The hardware control function provides access to the 
210			control pins of the NAND chip(s). 
211			The access can be done by GPIO pins or by address lines.
212			If you use address lines, make sure that the timing
213			requirements are met.
214		</para>
215		<para>
216			<emphasis>GPIO based example</emphasis>
217		</para>
218		<programlisting>
219static void board_hwcontrol(struct mtd_info *mtd, int cmd)
220{
221	switch(cmd){
222		case NAND_CTL_SETCLE: /* Set CLE pin high */ break;
223		case NAND_CTL_CLRCLE: /* Set CLE pin low */ break;
224		case NAND_CTL_SETALE: /* Set ALE pin high */ break;
225		case NAND_CTL_CLRALE: /* Set ALE pin low */ break;
226		case NAND_CTL_SETNCE: /* Set nCE pin low */ break;
227		case NAND_CTL_CLRNCE: /* Set nCE pin high */ break;
228	}
229}
230		</programlisting>
231		<para>
232			<emphasis>Address lines based example.</emphasis> It's assumed that the
233			nCE pin is driven by a chip select decoder.
234		</para>
235		<programlisting>
236static void board_hwcontrol(struct mtd_info *mtd, int cmd)
237{
238	struct nand_chip *this = (struct nand_chip *) mtd->priv;
239	switch(cmd){
240		case NAND_CTL_SETCLE: this->IO_ADDR_W |= CLE_ADRR_BIT;  break;
241		case NAND_CTL_CLRCLE: this->IO_ADDR_W &amp;= ~CLE_ADRR_BIT; break;
242		case NAND_CTL_SETALE: this->IO_ADDR_W |= ALE_ADRR_BIT;  break;
243		case NAND_CTL_CLRALE: this->IO_ADDR_W &amp;= ~ALE_ADRR_BIT; break;
244	}
245}
246		</programlisting>
247	</sect1>
248	<sect1 id="Device_ready_function">
249		<title>Device ready function</title>
250		<para>
251			If the hardware interface has the ready busy pin of the NAND chip connected to a
252			GPIO or other accessible I/O pin, this function is used to read back the state of the
253			pin. The function has no arguments and should return 0, if the device is busy (R/B pin 
254			is low) and 1, if the device is ready (R/B pin is high).
255			If the hardware interface does not give access to the ready busy pin, then
256			the function must not be defined and the function pointer this->dev_ready is set to NULL.		
257		</para>
258	</sect1>
259	<sect1 id="Init_function">
260		<title>Init function</title>
261		<para>
262			The init function allocates memory and sets up all the board
263			specific parameters and function pointers. When everything
264			is set up nand_scan() is called. This function tries to
265			detect and identify then chip. If a chip is found all the
266			internal data fields are initialized accordingly.
267			The structure(s) have to be zeroed out first and then filled with the necessary
268			information about the device.
269		</para>
270		<programlisting>
271static int __init board_init (void)
272{
273	struct nand_chip *this;
274	int err = 0;
275
276	/* Allocate memory for MTD device structure and private data */
277	board_mtd = kzalloc(sizeof(struct mtd_info) + sizeof(struct nand_chip), GFP_KERNEL);
278	if (!board_mtd) {
279		printk ("Unable to allocate NAND MTD device structure.\n");
280		err = -ENOMEM;
281		goto out;
282	}
283
284	/* map physical address */
285	baseaddr = ioremap(CHIP_PHYSICAL_ADDRESS, 1024);
286	if (!baseaddr) {
287		printk("Ioremap to access NAND chip failed\n");
288		err = -EIO;
289		goto out_mtd;
290	}
291
292	/* Get pointer to private data */
293	this = (struct nand_chip *) ();
294	/* Link the private data with the MTD structure */
295	board_mtd->priv = this;
296
297	/* Set address of NAND IO lines */
298	this->IO_ADDR_R = baseaddr;
299	this->IO_ADDR_W = baseaddr;
300	/* Reference hardware control function */
301	this->hwcontrol = board_hwcontrol;
302	/* Set command delay time, see datasheet for correct value */
303	this->chip_delay = CHIP_DEPENDEND_COMMAND_DELAY;
304	/* Assign the device ready function, if available */
305	this->dev_ready = board_dev_ready;
306	this->eccmode = NAND_ECC_SOFT;
307
308	/* Scan to find existence of the device */
309	if (nand_scan (board_mtd, 1)) {
310		err = -ENXIO;
311		goto out_ior;
312	}
313	
314	add_mtd_partitions(board_mtd, partition_info, NUM_PARTITIONS);
315	goto out;
316
317out_ior:
318	iounmap(baseaddr);
319out_mtd:
320	kfree (board_mtd);
321out:
322	return err;
323}
324module_init(board_init);
325		</programlisting>
326	</sect1>
327	<sect1 id="Exit_function">
328		<title>Exit function</title>
329		<para>
330			The exit function is only necessary if the driver is
331			compiled as a module. It releases all resources which
332			are held by the chip driver and unregisters the partitions
333			in the MTD layer.
334		</para>
335		<programlisting>
336#ifdef MODULE
337static void __exit board_cleanup (void)
338{
339	/* Release resources, unregister device */
340	nand_release (board_mtd);
341
342	/* unmap physical address */
343	iounmap(baseaddr);
344	
345	/* Free the MTD device structure */
346	kfree (board_mtd);
347}
348module_exit(board_cleanup);
349#endif
350		</programlisting>
351	</sect1>
352  </chapter>
353
354  <chapter id="boarddriversadvanced">
355     	<title>Advanced board driver functions</title>
356	<para>
357		This chapter describes the advanced functionality of the NAND
358		driver. For a list of functions which can be overridden by the board
359		driver see the documentation of the nand_chip structure.
360	</para>
361	<sect1 id="Multiple_chip_control">
362		<title>Multiple chip control</title>
363		<para>
364			The nand driver can control chip arrays. Therefore the
365			board driver must provide an own select_chip function. This
366			function must (de)select the requested chip.
367			The function pointer in the nand_chip structure must
368			be set before calling nand_scan(). The maxchip parameter
369			of nand_scan() defines the maximum number of chips to
370			scan for. Make sure that the select_chip function can
371			handle the requested number of chips.
372		</para>
373		<para>
374			The nand driver concatenates the chips to one virtual
375			chip and provides this virtual chip to the MTD layer.
376		</para>
377		<para>
378			<emphasis>Note: The driver can only handle linear chip arrays
379			of equally sized chips. There is no support for
380			parallel arrays which extend the buswidth.</emphasis>
381		</para>
382		<para>
383			<emphasis>GPIO based example</emphasis>
384		</para>
385		<programlisting>
386static void board_select_chip (struct mtd_info *mtd, int chip)
387{
388	/* Deselect all chips, set all nCE pins high */
389	GPIO(BOARD_NAND_NCE) |= 0xff;	
390	if (chip >= 0)
391		GPIO(BOARD_NAND_NCE) &amp;= ~ (1 &lt;&lt; chip);
392}
393		</programlisting>
394		<para>
395			<emphasis>Address lines based example.</emphasis>
396			Its assumed that the nCE pins are connected to an
397			address decoder.
398		</para>
399		<programlisting>
400static void board_select_chip (struct mtd_info *mtd, int chip)
401{
402	struct nand_chip *this = (struct nand_chip *) mtd->priv;
403	
404	/* Deselect all chips */
405	this->IO_ADDR_R &amp;= ~BOARD_NAND_ADDR_MASK;
406	this->IO_ADDR_W &amp;= ~BOARD_NAND_ADDR_MASK;
407	switch (chip) {
408	case 0:
409		this->IO_ADDR_R |= BOARD_NAND_ADDR_CHIP0;
410		this->IO_ADDR_W |= BOARD_NAND_ADDR_CHIP0;
411		break;
412	....	
413	case n:
414		this->IO_ADDR_R |= BOARD_NAND_ADDR_CHIPn;
415		this->IO_ADDR_W |= BOARD_NAND_ADDR_CHIPn;
416		break;
417	}	
418}
419		</programlisting>
420	</sect1>
421	<sect1 id="Hardware_ECC_support">
422		<title>Hardware ECC support</title>
423		<sect2 id="Functions_and_constants">
424			<title>Functions and constants</title>
425			<para>
426				The nand driver supports three different types of
427				hardware ECC.
428				<itemizedlist>
429				<listitem><para>NAND_ECC_HW3_256</para><para>
430				Hardware ECC generator providing 3 bytes ECC per
431				256 byte.
432				</para>	</listitem>
433				<listitem><para>NAND_ECC_HW3_512</para><para>
434				Hardware ECC generator providing 3 bytes ECC per
435				512 byte.
436				</para>	</listitem>
437				<listitem><para>NAND_ECC_HW6_512</para><para>
438				Hardware ECC generator providing 6 bytes ECC per
439				512 byte.
440				</para>	</listitem>
441				<listitem><para>NAND_ECC_HW8_512</para><para>
442				Hardware ECC generator providing 6 bytes ECC per
443				512 byte.
444				</para>	</listitem>
445				</itemizedlist>
446				If your hardware generator has a different functionality
447				add it at the appropriate place in nand_base.c
448			</para>
449			<para>
450				The board driver must provide following functions:
451				<itemizedlist>
452				<listitem><para>enable_hwecc</para><para>
453				This function is called before reading / writing to
454				the chip. Reset or initialize the hardware generator
455				in this function. The function is called with an
456				argument which let you distinguish between read 
457				and write operations.
458				</para>	</listitem>
459				<listitem><para>calculate_ecc</para><para>
460				This function is called after read / write from / to
461				the chip. Transfer the ECC from the hardware to
462				the buffer. If the option NAND_HWECC_SYNDROME is set
463				then the function is only called on write. See below.
464				</para>	</listitem>
465				<listitem><para>correct_data</para><para>
466				In case of an ECC error this function is called for
467				error detection and correction. Return 1 respectively 2
468				in case the error can be corrected. If the error is
469				not correctable return -1. If your hardware generator
470				matches the default algorithm of the nand_ecc software
471				generator then use the correction function provided
472				by nand_ecc instead of implementing duplicated code.
473				</para>	</listitem>
474				</itemizedlist>
475			</para>
476		</sect2>
477		<sect2 id="Hardware_ECC_with_syndrome_calculation">
478		<title>Hardware ECC with syndrome calculation</title>
479			<para>
480				Many hardware ECC implementations provide Reed-Solomon
481				codes and calculate an error syndrome on read. The syndrome
482				must be converted to a standard Reed-Solomon syndrome
483				before calling the error correction code in the generic
484				Reed-Solomon library.
485			</para>
486			<para>
487				The ECC bytes must be placed immediately after the data
488				bytes in order to make the syndrome generator work. This
489				is contrary to the usual layout used by software ECC. The
490				separation of data and out of band area is not longer
491				possible. The nand driver code handles this layout and
492				the remaining free bytes in the oob area are managed by 
493				the autoplacement code. Provide a matching oob-layout
494				in this case. See rts_from4.c and diskonchip.c for 
495				implementation reference. In those cases we must also
496				use bad block tables on FLASH, because the ECC layout is
497				interfering with the bad block marker positions.
498				See bad block table support for details.
499			</para>
500		</sect2>
501	</sect1>
502	<sect1 id="Bad_Block_table_support">
503		<title>Bad block table support</title>
504		<para>
505			Most NAND chips mark the bad blocks at a defined
506			position in the spare area. Those blocks must 
507			not be erased under any circumstances as the bad 
508			block information would be lost.
509			It is possible to check the bad block mark each
510			time when the blocks are accessed by reading the
511			spare area of the first page in the block. This
512			is time consuming so a bad block table is used.
513		</para>
514		<para>
515			The nand driver supports various types of bad block
516			tables.
517			<itemizedlist>
518			<listitem><para>Per device</para><para>
519			The bad block table contains all bad block information
520			of the device which can consist of multiple chips.
521			</para>	</listitem>
522			<listitem><para>Per chip</para><para>
523			A bad block table is used per chip and contains the
524			bad block information for this particular chip.
525			</para>	</listitem>
526			<listitem><para>Fixed offset</para><para>
527			The bad block table is located at a fixed offset
528			in the chip (device). This applies to various
529			DiskOnChip devices.
530			</para>	</listitem>
531			<listitem><para>Automatic placed</para><para>
532			The bad block table is automatically placed and
533			detected either at the end or at the beginning
534			of a chip (device)
535			</para>	</listitem>
536			<listitem><para>Mirrored tables</para><para>
537			The bad block table is mirrored on the chip (device) to
538			allow updates of the bad block table without data loss.
539			</para>	</listitem>
540			</itemizedlist>
541		</para>
542		<para>	
543			nand_scan() calls the function nand_default_bbt(). 
544			nand_default_bbt() selects appropriate default
545			bad block table descriptors depending on the chip information
546			which was retrieved by nand_scan().
547		</para>
548		<para>
549			The standard policy is scanning the device for bad 
550			blocks and build a ram based bad block table which
551			allows faster access than always checking the
552			bad block information on the flash chip itself.
553		</para>
554		<sect2 id="Flash_based_tables">
555			<title>Flash based tables</title>
556			<para>
557				It may be desired or necessary to keep a bad block table in FLASH.
558				For AG-AND chips this is mandatory, as they have no factory marked
559				bad blocks. They have factory marked good blocks. The marker pattern
560				is erased when the block is erased to be reused. So in case of
561				powerloss before writing the pattern back to the chip this block 
562				would be lost and added to the bad blocks. Therefore we scan the 
563				chip(s) when we detect them the first time for good blocks and 
564				store this information in a bad block table before erasing any 
565				of the blocks.
566			</para>
567			<para>
568				The blocks in which the tables are stored are protected against
569				accidental access by marking them bad in the memory bad block
570				table. The bad block table management functions are allowed
571				to circumvent this protection.
572			</para>
573			<para>
574				The simplest way to activate the FLASH based bad block table support 
575				is to set the option NAND_BBT_USE_FLASH in the bbt_option field of
576				the nand chip structure before calling nand_scan(). For AG-AND
577				chips is this done by default.
578				This activates the default FLASH based bad block table functionality 
579				of the NAND driver. The default bad block table options are
580				<itemizedlist>
581				<listitem><para>Store bad block table per chip</para></listitem>
582				<listitem><para>Use 2 bits per block</para></listitem>
583				<listitem><para>Automatic placement at the end of the chip</para></listitem>
584				<listitem><para>Use mirrored tables with version numbers</para></listitem>
585				<listitem><para>Reserve 4 blocks at the end of the chip</para></listitem>
586				</itemizedlist>
587			</para>
588		</sect2>
589		<sect2 id="User_defined_tables">
590			<title>User defined tables</title>
591			<para>
592				User defined tables are created by filling out a 
593				nand_bbt_descr structure and storing the pointer in the
594				nand_chip structure member bbt_td before calling nand_scan(). 
595				If a mirror table is necessary a second structure must be
596				created and a pointer to this structure must be stored
597				in bbt_md inside the nand_chip structure. If the bbt_md 
598				member is set to NULL then only the main table is used
599				and no scan for the mirrored table is performed.
600			</para>
601			<para>
602				The most important field in the nand_bbt_descr structure
603				is the options field. The options define most of the 
604				table properties. Use the predefined constants from
605				nand.h to define the options.
606				<itemizedlist>
607				<listitem><para>Number of bits per block</para>
608				<para>The supported number of bits is 1, 2, 4, 8.</para></listitem>
609				<listitem><para>Table per chip</para>
610				<para>Setting the constant NAND_BBT_PERCHIP selects that
611				a bad block table is managed for each chip in a chip array.
612				If this option is not set then a per device bad block table
613				is used.</para></listitem>
614				<listitem><para>Table location is absolute</para>
615				<para>Use the option constant NAND_BBT_ABSPAGE and
616				define the absolute page number where the bad block
617				table starts in the field pages. If you have selected bad block
618				tables per chip and you have a multi chip array then the start page
619				must be given for each chip in the chip array. Note: there is no scan
620				for a table ident pattern performed, so the fields 
621				pattern, veroffs, offs, len can be left uninitialized</para></listitem>
622				<listitem><para>Table location is automatically detected</para>
623				<para>The table can either be located in the first or the last good
624				blocks of the chip (device). Set NAND_BBT_LASTBLOCK to place
625				the bad block table at the end of the chip (device). The
626				bad block tables are marked and identified by a pattern which
627				is stored in the spare area of the first page in the block which
628				holds the bad block table. Store a pointer to the pattern  
629				in the pattern field. Further the length of the pattern has to be 
630				stored in len and the offset in the spare area must be given
631				in the offs member of the nand_bbt_descr structure. For mirrored
632				bad block tables different patterns are mandatory.</para></listitem>
633				<listitem><para>Table creation</para>
634				<para>Set the option NAND_BBT_CREATE to enable the table creation
635				if no table can be found during the scan. Usually this is done only 
636				once if a new chip is found. </para></listitem>
637				<listitem><para>Table write support</para>
638				<para>Set the option NAND_BBT_WRITE to enable the table write support.
639				This allows the update of the bad block table(s) in case a block has
640				to be marked bad due to wear. The MTD interface function block_markbad
641				is calling the update function of the bad block table. If the write
642				support is enabled then the table is updated on FLASH.</para>
643				<para>
644				Note: Write support should only be enabled for mirrored tables with
645				version control.
646				</para></listitem>
647				<listitem><para>Table version control</para>
648				<para>Set the option NAND_BBT_VERSION to enable the table version control.
649				It's highly recommended to enable this for mirrored tables with write
650				support. It makes sure that the risk of losing the bad block
651				table information is reduced to the loss of the information about the
652				one worn out block which should be marked bad. The version is stored in
653				4 consecutive bytes in the spare area of the device. The position of
654				the version number is defined by the member veroffs in the bad block table
655				descriptor.</para></listitem>
656				<listitem><para>Save block contents on write</para>
657				<para>
658				In case that the block which holds the bad block table does contain
659				other useful information, set the option NAND_BBT_SAVECONTENT. When
660				the bad block table is written then the whole block is read the bad
661				block table is updated and the block is erased and everything is 
662				written back. If this option is not set only the bad block table
663				is written and everything else in the block is ignored and erased.
664				</para></listitem>
665				<listitem><para>Number of reserved blocks</para>
666				<para>
667				For automatic placement some blocks must be reserved for
668				bad block table storage. The number of reserved blocks is defined 
669				in the maxblocks member of the bad block table description structure.
670				Reserving 4 blocks for mirrored tables should be a reasonable number. 
671				This also limits the number of blocks which are scanned for the bad
672				block table ident pattern.
673				</para></listitem>
674				</itemizedlist>
675			</para>
676		</sect2>
677	</sect1>
678	<sect1 id="Spare_area_placement">
679		<title>Spare area (auto)placement</title>
680		<para>
681			The nand driver implements different possibilities for
682			placement of filesystem data in the spare area, 
683			<itemizedlist>
684			<listitem><para>Placement defined by fs driver</para></listitem>
685			<listitem><para>Automatic placement</para></listitem>
686			</itemizedlist>
687			The default placement function is automatic placement. The
688			nand driver has built in default placement schemes for the
689			various chiptypes. If due to hardware ECC functionality the
690			default placement does not fit then the board driver can
691			provide a own placement scheme.
692		</para>
693		<para>
694			File system drivers can provide a own placement scheme which
695			is used instead of the default placement scheme.
696		</para>
697		<para>
698			Placement schemes are defined by a nand_oobinfo structure
699	     		<programlisting>
700struct nand_oobinfo {
701	int	useecc;
702	int	eccbytes;
703	int	eccpos[24];
704	int	oobfree[8][2];
705};
706	     		</programlisting>
707			<itemizedlist>
708			<listitem><para>useecc</para><para>
709				The useecc member controls the ecc and placement function. The header
710				file include/mtd/mtd-abi.h contains constants to select ecc and
711				placement. MTD_NANDECC_OFF switches off the ecc complete. This is
712				not recommended and available for testing and diagnosis only.
713				MTD_NANDECC_PLACE selects caller defined placement, MTD_NANDECC_AUTOPLACE
714				selects automatic placement.
715			</para></listitem>
716			<listitem><para>eccbytes</para><para>
717				The eccbytes member defines the number of ecc bytes per page.
718			</para></listitem>
719			<listitem><para>eccpos</para><para>
720				The eccpos array holds the byte offsets in the spare area where
721				the ecc codes are placed.
722			</para></listitem>
723			<listitem><para>oobfree</para><para>
724				The oobfree array defines the areas in the spare area which can be
725				used for automatic placement. The information is given in the format
726				{offset, size}. offset defines the start of the usable area, size the
727				length in bytes. More than one area can be defined. The list is terminated
728				by an {0, 0} entry.
729			</para></listitem>
730			</itemizedlist>
731		</para>
732		<sect2 id="Placement_defined_by_fs_driver">
733			<title>Placement defined by fs driver</title>
734			<para>
735				The calling function provides a pointer to a nand_oobinfo
736				structure which defines the ecc placement. For writes the
737				caller must provide a spare area buffer along with the
738				data buffer. The spare area buffer size is (number of pages) *
739				(size of spare area). For reads the buffer size is
740				(number of pages) * ((size of spare area) + (number of ecc
741				steps per page) * sizeof (int)). The driver stores the
742				result of the ecc check for each tuple in the spare buffer.
743				The storage sequence is 
744			</para>
745			<para>
746				&lt;spare data page 0&gt;&lt;ecc result 0&gt;...&lt;ecc result n&gt;
747			</para>
748			<para>
749				...
750			</para>
751			<para>
752				&lt;spare data page n&gt;&lt;ecc result 0&gt;...&lt;ecc result n&gt;
753			</para>
754			<para>
755				This is a legacy mode used by YAFFS1.
756			</para>
757			<para>
758				If the spare area buffer is NULL then only the ECC placement is
759				done according to the given scheme in the nand_oobinfo structure.
760			</para>
761		</sect2>
762		<sect2 id="Automatic_placement">
763			<title>Automatic placement</title>
764			<para>
765				Automatic placement uses the built in defaults to place the
766				ecc bytes in the spare area. If filesystem data have to be stored /
767				read into the spare area then the calling function must provide a
768				buffer. The buffer size per page is determined by the oobfree array in
769				the nand_oobinfo structure.
770			</para>
771			<para>
772				If the spare area buffer is NULL then only the ECC placement is
773				done according to the default builtin scheme.
774			</para>
775		</sect2>
776	</sect1>	
777	<sect1 id="Spare_area_autoplacement_default">
778		<title>Spare area autoplacement default schemes</title>
779		<sect2 id="pagesize_256">
780			<title>256 byte pagesize</title>
781<informaltable><tgroup cols="3"><tbody>
782<row>
783<entry>Offset</entry>
784<entry>Content</entry>
785<entry>Comment</entry>
786</row>
787<row>
788<entry>0x00</entry>
789<entry>ECC byte 0</entry>
790<entry>Error correction code byte 0</entry>
791</row>
792<row>
793<entry>0x01</entry>
794<entry>ECC byte 1</entry>
795<entry>Error correction code byte 1</entry>
796</row>
797<row>
798<entry>0x02</entry>
799<entry>ECC byte 2</entry>
800<entry>Error correction code byte 2</entry>
801</row>
802<row>
803<entry>0x03</entry>
804<entry>Autoplace 0</entry>
805<entry></entry>
806</row>
807<row>
808<entry>0x04</entry>
809<entry>Autoplace 1</entry>
810<entry></entry>
811</row>
812<row>
813<entry>0x05</entry>
814<entry>Bad block marker</entry>
815<entry>If any bit in this byte is zero, then this block is bad.
816This applies only to the first page in a block. In the remaining
817pages this byte is reserved</entry>
818</row>
819<row>
820<entry>0x06</entry>
821<entry>Autoplace 2</entry>
822<entry></entry>
823</row>
824<row>
825<entry>0x07</entry>
826<entry>Autoplace 3</entry>
827<entry></entry>
828</row>
829</tbody></tgroup></informaltable>
830		</sect2>
831		<sect2 id="pagesize_512">
832			<title>512 byte pagesize</title>
833<informaltable><tgroup cols="3"><tbody>
834<row>
835<entry>Offset</entry>
836<entry>Content</entry>
837<entry>Comment</entry>
838</row>
839<row>
840<entry>0x00</entry>
841<entry>ECC byte 0</entry>
842<entry>Error correction code byte 0 of the lower 256 Byte data in
843this page</entry>
844</row>
845<row>
846<entry>0x01</entry>
847<entry>ECC byte 1</entry>
848<entry>Error correction code byte 1 of the lower 256 Bytes of data
849in this page</entry>
850</row>
851<row>
852<entry>0x02</entry>
853<entry>ECC byte 2</entry>
854<entry>Error correction code byte 2 of the lower 256 Bytes of data
855in this page</entry>
856</row>
857<row>
858<entry>0x03</entry>
859<entry>ECC byte 3</entry>
860<entry>Error correction code byte 0 of the upper 256 Bytes of data
861in this page</entry>
862</row>
863<row>
864<entry>0x04</entry>
865<entry>reserved</entry>
866<entry>reserved</entry>
867</row>
868<row>
869<entry>0x05</entry>
870<entry>Bad block marker</entry>
871<entry>If any bit in this byte is zero, then this block is bad.
872This applies only to the first page in a block. In the remaining
873pages this byte is reserved</entry>
874</row>
875<row>
876<entry>0x06</entry>
877<entry>ECC byte 4</entry>
878<entry>Error correction code byte 1 of the upper 256 Bytes of data
879in this page</entry>
880</row>
881<row>
882<entry>0x07</entry>
883<entry>ECC byte 5</entry>
884<entry>Error correction code byte 2 of the upper 256 Bytes of data
885in this page</entry>
886</row>
887<row>
888<entry>0x08 - 0x0F</entry>
889<entry>Autoplace 0 - 7</entry>
890<entry></entry>
891</row>
892</tbody></tgroup></informaltable>
893		</sect2>
894		<sect2 id="pagesize_2048">
895			<title>2048 byte pagesize</title>
896<informaltable><tgroup cols="3"><tbody>
897<row>
898<entry>Offset</entry>
899<entry>Content</entry>
900<entry>Comment</entry>
901</row>
902<row>
903<entry>0x00</entry>
904<entry>Bad block marker</entry>
905<entry>If any bit in this byte is zero, then this block is bad.
906This applies only to the first page in a block. In the remaining
907pages this byte is reserved</entry>
908</row>
909<row>
910<entry>0x01</entry>
911<entry>Reserved</entry>
912<entry>Reserved</entry>
913</row>
914<row>
915<entry>0x02-0x27</entry>
916<entry>Autoplace 0 - 37</entry>
917<entry></entry>
918</row>
919<row>
920<entry>0x28</entry>
921<entry>ECC byte 0</entry>
922<entry>Error correction code byte 0 of the first 256 Byte data in
923this page</entry>
924</row>
925<row>
926<entry>0x29</entry>
927<entry>ECC byte 1</entry>
928<entry>Error correction code byte 1 of the first 256 Bytes of data
929in this page</entry>
930</row>
931<row>
932<entry>0x2A</entry>
933<entry>ECC byte 2</entry>
934<entry>Error correction code byte 2 of the first 256 Bytes data in
935this page</entry>
936</row>
937<row>
938<entry>0x2B</entry>
939<entry>ECC byte 3</entry>
940<entry>Error correction code byte 0 of the second 256 Bytes of data
941in this page</entry>
942</row>
943<row>
944<entry>0x2C</entry>
945<entry>ECC byte 4</entry>
946<entry>Error correction code byte 1 of the second 256 Bytes of data
947in this page</entry>
948</row>
949<row>
950<entry>0x2D</entry>
951<entry>ECC byte 5</entry>
952<entry>Error correction code byte 2 of the second 256 Bytes of data
953in this page</entry>
954</row>
955<row>
956<entry>0x2E</entry>
957<entry>ECC byte 6</entry>
958<entry>Error correction code byte 0 of the third 256 Bytes of data
959in this page</entry>
960</row>
961<row>
962<entry>0x2F</entry>
963<entry>ECC byte 7</entry>
964<entry>Error correction code byte 1 of the third 256 Bytes of data
965in this page</entry>
966</row>
967<row>
968<entry>0x30</entry>
969<entry>ECC byte 8</entry>
970<entry>Error correction code byte 2 of the third 256 Bytes of data
971in this page</entry>
972</row>
973<row>
974<entry>0x31</entry>
975<entry>ECC byte 9</entry>
976<entry>Error correction code byte 0 of the fourth 256 Bytes of data
977in this page</entry>
978</row>
979<row>
980<entry>0x32</entry>
981<entry>ECC byte 10</entry>
982<entry>Error correction code byte 1 of the fourth 256 Bytes of data
983in this page</entry>
984</row>
985<row>
986<entry>0x33</entry>
987<entry>ECC byte 11</entry>
988<entry>Error correction code byte 2 of the fourth 256 Bytes of data
989in this page</entry>
990</row>
991<row>
992<entry>0x34</entry>
993<entry>ECC byte 12</entry>
994<entry>Error correction code byte 0 of the fifth 256 Bytes of data
995in this page</entry>
996</row>
997<row>
998<entry>0x35</entry>
999<entry>ECC byte 13</entry>
1000<entry>Error correction code byte 1 of the fifth 256 Bytes of data
1001in this page</entry>
1002</row>
1003<row>
1004<entry>0x36</entry>
1005<entry>ECC byte 14</entry>
1006<entry>Error correction code byte 2 of the fifth 256 Bytes of data
1007in this page</entry>
1008</row>
1009<row>
1010<entry>0x37</entry>
1011<entry>ECC byte 15</entry>
1012<entry>Error correction code byte 0 of the sixt 256 Bytes of data
1013in this page</entry>
1014</row>
1015<row>
1016<entry>0x38</entry>
1017<entry>ECC byte 16</entry>
1018<entry>Error correction code byte 1 of the sixt 256 Bytes of data
1019in this page</entry>
1020</row>
1021<row>
1022<entry>0x39</entry>
1023<entry>ECC byte 17</entry>
1024<entry>Error correction code byte 2 of the sixt 256 Bytes of data
1025in this page</entry>
1026</row>
1027<row>
1028<entry>0x3A</entry>
1029<entry>ECC byte 18</entry>
1030<entry>Error correction code byte 0 of the seventh 256 Bytes of
1031data in this page</entry>
1032</row>
1033<row>
1034<entry>0x3B</entry>
1035<entry>ECC byte 19</entry>
1036<entry>Error correction code byte 1 of the seventh 256 Bytes of
1037data in this page</entry>
1038</row>
1039<row>
1040<entry>0x3C</entry>
1041<entry>ECC byte 20</entry>
1042<entry>Error correction code byte 2 of the seventh 256 Bytes of
1043data in this page</entry>
1044</row>
1045<row>
1046<entry>0x3D</entry>
1047<entry>ECC byte 21</entry>
1048<entry>Error correction code byte 0 of the eighth 256 Bytes of data
1049in this page</entry>
1050</row>
1051<row>
1052<entry>0x3E</entry>
1053<entry>ECC byte 22</entry>
1054<entry>Error correction code byte 1 of the eighth 256 Bytes of data
1055in this page</entry>
1056</row>
1057<row>
1058<entry>0x3F</entry>
1059<entry>ECC byte 23</entry>
1060<entry>Error correction code byte 2 of the eighth 256 Bytes of data
1061in this page</entry>
1062</row>
1063</tbody></tgroup></informaltable>
1064		</sect2>
1065     	</sect1>
1066  </chapter>
1067
1068  <chapter id="filesystems">
1069     	<title>Filesystem support</title>
1070	<para>
1071		The NAND driver provides all necessary functions for a
1072		filesystem via the MTD interface.
1073	</para>
1074	<para>
1075		Filesystems must be aware of the NAND peculiarities and
1076		restrictions. One major restrictions of NAND Flash is, that you cannot 
1077		write as often as you want to a page. The consecutive writes to a page, 
1078		before erasing it again, are restricted to 1-3 writes, depending on the 
1079		manufacturers specifications. This applies similar to the spare area. 
1080	</para>
1081	<para>
1082		Therefore NAND aware filesystems must either write in page size chunks
1083		or hold a writebuffer to collect smaller writes until they sum up to 
1084		pagesize. Available NAND aware filesystems: JFFS2, YAFFS. 		
1085	</para>
1086	<para>
1087		The spare area usage to store filesystem data is controlled by
1088		the spare area placement functionality which is described in one
1089		of the earlier chapters.
1090	</para>
1091  </chapter>	
1092  <chapter id="tools">
1093     	<title>Tools</title>
1094	<para>
1095		The MTD project provides a couple of helpful tools to handle NAND Flash.
1096		<itemizedlist>
1097		<listitem><para>flasherase, flasheraseall: Erase and format FLASH partitions</para></listitem>
1098		<listitem><para>nandwrite: write filesystem images to NAND FLASH</para></listitem>
1099		<listitem><para>nanddump: dump the contents of a NAND FLASH partitions</para></listitem>
1100		</itemizedlist>
1101	</para>
1102	<para>
1103		These tools are aware of the NAND restrictions. Please use those tools
1104		instead of complaining about errors which are caused by non NAND aware
1105		access methods.
1106	</para>
1107  </chapter>	
1108
1109  <chapter id="defines">
1110     <title>Constants</title>
1111     <para>
1112     This chapter describes the constants which might be relevant for a driver developer.
1113     </para>
1114     <sect1 id="Chip_option_constants">
1115	<title>Chip option constants</title>
1116     	<sect2 id="Constants_for_chip_id_table">
1117		<title>Constants for chip id table</title>
1118     		<para>
1119		These constants are defined in nand.h. They are ored together to describe
1120		the chip functionality.
1121     		<programlisting>
1122/* Buswitdh is 16 bit */
1123#define NAND_BUSWIDTH_16	0x00000002
1124/* Device supports partial programming without padding */
1125#define NAND_NO_PADDING		0x00000004
1126/* Chip has cache program function */
1127#define NAND_CACHEPRG		0x00000008
1128/* Chip has copy back function */
1129#define NAND_COPYBACK		0x00000010
1130/* AND Chip which has 4 banks and a confusing page / block 
1131 * assignment. See Renesas datasheet for further information */
1132#define NAND_IS_AND		0x00000020
1133/* Chip has a array of 4 pages which can be read without
1134 * additional ready /busy waits */
1135#define NAND_4PAGE_ARRAY	0x00000040 
1136		</programlisting>
1137     		</para>
1138     	</sect2>
1139     	<sect2 id="Constants_for_runtime_options">
1140		<title>Constants for runtime options</title>
1141     		<para>
1142		These constants are defined in nand.h. They are ored together to describe
1143		the functionality.
1144     		<programlisting>
1145/* The hw ecc generator provides a syndrome instead a ecc value on read 
1146 * This can only work if we have the ecc bytes directly behind the 
1147 * data bytes. Applies for DOC and AG-AND Renesas HW Reed Solomon generators */
1148#define NAND_HWECC_SYNDROME	0x00020000
1149		</programlisting>
1150     		</para>
1151     	</sect2>
1152     </sect1>	
1153
1154     <sect1 id="EEC_selection_constants">
1155	<title>ECC selection constants</title>
1156	<para>
1157	Use these constants to select the ECC algorithm.
1158  	<programlisting>
1159/* No ECC. Usage is not recommended ! */
1160#define NAND_ECC_NONE		0
1161/* Software ECC 3 byte ECC per 256 Byte data */
1162#define NAND_ECC_SOFT		1
1163/* Hardware ECC 3 byte ECC per 256 Byte data */
1164#define NAND_ECC_HW3_256	2
1165/* Hardware ECC 3 byte ECC per 512 Byte data */
1166#define NAND_ECC_HW3_512	3
1167/* Hardware ECC 6 byte ECC per 512 Byte data */
1168#define NAND_ECC_HW6_512	4
1169/* Hardware ECC 6 byte ECC per 512 Byte data */
1170#define NAND_ECC_HW8_512	6
1171	</programlisting>
1172	</para>
1173     </sect1>	
1174
1175     <sect1 id="Hardware_control_related_constants">
1176	<title>Hardware control related constants</title>
1177	<para>
1178	These constants describe the requested hardware access function when
1179	the boardspecific hardware control function is called
1180  	<programlisting>
1181/* Select the chip by setting nCE to low */
1182#define NAND_CTL_SETNCE 	1
1183/* Deselect the chip by setting nCE to high */
1184#define NAND_CTL_CLRNCE		2
1185/* Select the command latch by setting CLE to high */
1186#define NAND_CTL_SETCLE		3
1187/* Deselect the command latch by setting CLE to low */
1188#define NAND_CTL_CLRCLE		4
1189/* Select the address latch by setting ALE to high */
1190#define NAND_CTL_SETALE		5
1191/* Deselect the address latch by setting ALE to low */
1192#define NAND_CTL_CLRALE		6
1193/* Set write protection by setting WP to high. Not used! */
1194#define NAND_CTL_SETWP		7
1195/* Clear write protection by setting WP to low. Not used! */
1196#define NAND_CTL_CLRWP		8
1197	</programlisting>
1198	</para>
1199     </sect1>	
1200
1201     <sect1 id="Bad_block_table_constants">
1202	<title>Bad block table related constants</title>
1203	<para>
1204	These constants describe the options used for bad block
1205	table descriptors.
1206  	<programlisting>
1207/* Options for the bad block table descriptors */
1208
1209/* The number of bits used per block in the bbt on the device */
1210#define NAND_BBT_NRBITS_MSK	0x0000000F
1211#define NAND_BBT_1BIT		0x00000001
1212#define NAND_BBT_2BIT		0x00000002
1213#define NAND_BBT_4BIT		0x00000004
1214#define NAND_BBT_8BIT		0x00000008
1215/* The bad block table is in the last good block of the device */
1216#define	NAND_BBT_LASTBLOCK	0x00000010
1217/* The bbt is at the given page, else we must scan for the bbt */
1218#define NAND_BBT_ABSPAGE	0x00000020
1219/* bbt is stored per chip on multichip devices */
1220#define NAND_BBT_PERCHIP	0x00000080
1221/* bbt has a version counter at offset veroffs */
1222#define NAND_BBT_VERSION	0x00000100
1223/* Create a bbt if none axists */
1224#define NAND_BBT_CREATE		0x00000200
1225/* Write bbt if necessary */
1226#define NAND_BBT_WRITE		0x00001000
1227/* Read and write back block contents when writing bbt */
1228#define NAND_BBT_SAVECONTENT	0x00002000
1229	</programlisting>
1230	</para>
1231     </sect1>	
1232
1233  </chapter>
1234  	
1235  <chapter id="structs">
1236     <title>Structures</title>
1237     <para>
1238     This chapter contains the autogenerated documentation of the structures which are
1239     used in the NAND driver and might be relevant for a driver developer. Each  
1240     struct member has a short description which is marked with an [XXX] identifier.
1241     See the chapter "Documentation hints" for an explanation.
1242     </para>
1243!Iinclude/linux/mtd/nand.h
1244  </chapter>
1245
1246  <chapter id="pubfunctions">
1247     <title>Public Functions Provided</title>
1248     <para>
1249     This chapter contains the autogenerated documentation of the NAND kernel API functions
1250      which are exported. Each function has a short description which is marked with an [XXX] identifier.
1251     See the chapter "Documentation hints" for an explanation.
1252     </para>
1253!Edrivers/mtd/nand/nand_base.c
1254!Edrivers/mtd/nand/nand_bbt.c
1255!Edrivers/mtd/nand/nand_ecc.c
1256  </chapter>
1257  
1258  <chapter id="intfunctions">
1259     <title>Internal Functions Provided</title>
1260     <para>
1261     This chapter contains the autogenerated documentation of the NAND driver internal functions.
1262     Each function has a short description which is marked with an [XXX] identifier.
1263     See the chapter "Documentation hints" for an explanation.
1264     The functions marked with [DEFAULT] might be relevant for a board driver developer.
1265     </para>
1266!Idrivers/mtd/nand/nand_base.c
1267!Idrivers/mtd/nand/nand_bbt.c
1268<!-- No internal functions for kernel-doc:
1269X!Idrivers/mtd/nand/nand_ecc.c
1270-->
1271  </chapter>
1272
1273  <chapter id="credits">
1274     <title>Credits</title>
1275	<para>
1276		The following people have contributed to the NAND driver:
1277		<orderedlist>
1278			<listitem><para>Steven J. Hill<email>sjhill@realitydiluted.com</email></para></listitem>
1279			<listitem><para>David Woodhouse<email>dwmw2@infradead.org</email></para></listitem>
1280			<listitem><para>Thomas Gleixner<email>tglx@linutronix.de</email></para></listitem>
1281		</orderedlist>
1282		A lot of users have provided bugfixes, improvements and helping hands for testing.
1283		Thanks a lot.
1284	</para>
1285	<para>
1286		The following people have contributed to this document:
1287		<orderedlist>
1288			<listitem><para>Thomas Gleixner<email>tglx@linutronix.de</email></para></listitem>
1289		</orderedlist>
1290	</para>
1291  </chapter>
1292</book>
1293