xref: /openbmc/u-boot/doc/README.nand (revision fd697ecf)
1NAND FLASH commands and notes
2
3See NOTE below!!!
4
5# (C) Copyright 2003
6# Dave Ellis, SIXNET, dge@sixnetio.com
7#
8# SPDX-License-Identifier:	GPL-2.0+
9
10Commands:
11
12   nand bad
13      Print a list of all of the bad blocks in the current device.
14
15   nand device
16      Print information about the current NAND device.
17
18   nand device num
19      Make device `num' the current device and print information about it.
20
21   nand erase off|partition size
22   nand erase clean [off|partition size]
23      Erase `size' bytes starting at offset `off'. Alternatively partition
24      name can be specified, in this case size will be eventually limited
25      to not exceed partition size (this behaviour applies also to read
26      and write commands). Only complete erase blocks can be erased.
27
28      If `erase' is specified without an offset or size, the entire flash
29      is erased. If `erase' is specified with partition but without an
30      size, the entire partition is erased.
31
32      If `clean' is specified, a JFFS2-style clean marker is written to
33      each block after it is erased.
34
35      This command will not erase blocks that are marked bad. There is
36      a debug option in cmd_nand.c to allow bad blocks to be erased.
37      Please read the warning there before using it, as blocks marked
38      bad by the manufacturer must _NEVER_ be erased.
39
40   nand info
41      Print information about all of the NAND devices found.
42
43   nand read addr ofs|partition size
44      Read `size' bytes from `ofs' in NAND flash to `addr'.  Blocks that
45      are marked bad are skipped.  If a page cannot be read because an
46      uncorrectable data error is found, the command stops with an error.
47
48   nand read.oob addr ofs|partition size
49      Read `size' bytes from the out-of-band data area corresponding to
50      `ofs' in NAND flash to `addr'. This is limited to the 16 bytes of
51      data for one 512-byte page or 2 256-byte pages. There is no check
52      for bad blocks or ECC errors.
53
54   nand write addr ofs|partition size
55      Write `size' bytes from `addr' to `ofs' in NAND flash.  Blocks that
56      are marked bad are skipped.  If a page cannot be read because an
57      uncorrectable data error is found, the command stops with an error.
58
59      As JFFS2 skips blocks similarly, this allows writing a JFFS2 image,
60      as long as the image is short enough to fit even after skipping the
61      bad blocks.  Compact images, such as those produced by mkfs.jffs2
62      should work well, but loading an image copied from another flash is
63      going to be trouble if there are any bad blocks.
64
65   nand write.trimffs addr ofs|partition size
66      Enabled by the CONFIG_CMD_NAND_TRIMFFS macro. This command will write to
67      the NAND flash in a manner identical to the 'nand write' command
68      described above -- with the additional check that all pages at the end
69      of eraseblocks which contain only 0xff data will not be written to the
70      NAND flash. This behaviour is required when flashing UBI images
71      containing UBIFS volumes as per the UBI FAQ[1].
72
73      [1] http://www.linux-mtd.infradead.org/doc/ubi.html#L_flasher_algo
74
75   nand write.oob addr ofs|partition size
76      Write `size' bytes from `addr' to the out-of-band data area
77      corresponding to `ofs' in NAND flash. This is limited to the 16 bytes
78      of data for one 512-byte page or 2 256-byte pages. There is no check
79      for bad blocks.
80
81   nand read.raw addr ofs|partition [count]
82   nand write.raw addr ofs|partition [count]
83      Read or write one or more pages at "ofs" in NAND flash, from or to
84      "addr" in memory.  This is a raw access, so ECC is avoided and the
85      OOB area is transferred as well.  If count is absent, it is assumed
86      to be one page.  As with .yaffs2 accesses, the data is formatted as
87      a packed sequence of "data, oob, data, oob, ..." -- no alignment of
88      individual pages is maintained.
89
90Configuration Options:
91
92   CONFIG_SYS_NAND_U_BOOT_OFFS
93	NAND Offset from where SPL will read u-boot image. This is the starting
94	address of u-boot MTD partition in NAND.
95
96   CONFIG_CMD_NAND
97      Enables NAND support and commmands.
98
99   CONFIG_CMD_NAND_TORTURE
100      Enables the torture command (see description of this command below).
101
102   CONFIG_MTD_NAND_ECC_JFFS2
103      Define this if you want the Error Correction Code information in
104      the out-of-band data to be formatted to match the JFFS2 file system.
105      CONFIG_MTD_NAND_ECC_YAFFS would be another useful choice for
106      someone to implement.
107
108   CONFIG_SYS_MAX_NAND_DEVICE
109      The maximum number of NAND devices you want to support.
110
111   CONFIG_SYS_NAND_MAX_ECCPOS
112      If specified, overrides the maximum number of ECC bytes
113      supported.  Useful for reducing image size, especially with SPL.
114      This must be at least 48 if nand_base.c is used.
115
116   CONFIG_SYS_NAND_MAX_OOBFREE
117      If specified, overrides the maximum number of free OOB regions
118      supported.  Useful for reducing image size, especially with SPL.
119      This must be at least 2 if nand_base.c is used.
120
121   CONFIG_SYS_NAND_MAX_CHIPS
122      The maximum number of NAND chips per device to be supported.
123
124   CONFIG_SYS_NAND_SELF_INIT
125      Traditionally, glue code in drivers/mtd/nand/nand.c has driven
126      the initialization process -- it provides the mtd and nand
127      structs, calls a board init function for a specific device,
128      calls nand_scan(), and registers with mtd.
129
130      This arrangement does not provide drivers with the flexibility to
131      run code between nand_scan_ident() and nand_scan_tail(), or other
132      deviations from the "normal" flow.
133
134      If a board defines CONFIG_SYS_NAND_SELF_INIT, drivers/mtd/nand/nand.c
135      will make one call to board_nand_init(), with no arguments.  That
136      function is responsible for calling a driver init function for
137      each NAND device on the board, that performs all initialization
138      tasks except setting mtd->name, and registering with the rest of
139      U-Boot.  Those last tasks are accomplished by calling  nand_register()
140      on the new mtd device.
141
142      Example of new init to be added to the end of an existing driver
143      init:
144
145	/*
146	 * devnum is the device number to be used in nand commands
147	 * and in mtd->name.  Must be less than
148	 * CONFIG_SYS_NAND_MAX_DEVICE.
149	 */
150	mtd = &nand_info[devnum];
151
152	/* chip is struct nand_chip, and is now provided by the driver. */
153	mtd->priv = &chip;
154
155	/*
156	 * Fill in appropriate values if this driver uses these fields,
157	 * or uses the standard read_byte/write_buf/etc. functions from
158	 * nand_base.c that use these fields.
159	 */
160	chip.IO_ADDR_R = ...;
161	chip.IO_ADDR_W = ...;
162
163	if (nand_scan_ident(mtd, CONFIG_SYS_MAX_NAND_CHIPS, NULL))
164		error out
165
166	/*
167	 * Insert here any code you wish to run after the chip has been
168	 * identified, but before any other I/O is done.
169	 */
170
171	if (nand_scan_tail(mtd))
172		error out
173
174	if (nand_register(devnum))
175		error out
176
177      In addition to providing more flexibility to the driver, it reduces
178      the difference between a U-Boot driver and its Linux counterpart.
179      nand_init() is now reduced to calling board_nand_init() once, and
180      printing a size summary.  This should also make it easier to
181      transition to delayed NAND initialization.
182
183      Please convert your driver even if you don't need the extra
184      flexibility, so that one day we can eliminate the old mechanism.
185
186
187   CONFIG_SYS_NAND_ONFI_DETECTION
188	Enables detection of ONFI compliant devices during probe.
189	And fetching device parameters flashed on device, by parsing
190	ONFI parameter page.
191
192   CONFIG_BCH
193	Enables software based BCH ECC algorithm present in lib/bch.c
194	This is used by SoC platforms which do not have built-in ELM
195	hardware engine required for BCH ECC correction.
196
197   CONFIG_SYS_NAND_BUSWIDTH_16BIT
198	Indicates that NAND device has 16-bit wide data-bus. In absence of this
199	config, bus-width of NAND device is assumed to be either 8-bit and later
200	determined by reading ONFI params.
201	Above config is useful when NAND device's bus-width information cannot
202	be determined from on-chip ONFI params, like in following scenarios:
203	- SPL boot does not support reading of ONFI parameters. This is done to
204	  keep SPL code foot-print small.
205	- In current U-Boot flow using nand_init(), driver initialization
206	  happens in board_nand_init() which is called before any device probe
207	  (nand_scan_ident + nand_scan_tail), thus device's ONFI parameters are
208	  not available while configuring controller. So a static CONFIG_NAND_xx
209	  is needed to know the device's bus-width in advance.
210	Some drivers using above config are:
211	drivers/mtd/nand/mxc_nand.c
212	drivers/mtd/nand/ndfc.c
213	drivers/mtd/nand/omap_gpmc.c
214
215
216Platform specific options
217=========================
218   CONFIG_NAND_OMAP_GPMC
219	Enables omap_gpmc.c driver for OMAPx and AMxxxx platforms.
220	GPMC controller is used for parallel NAND flash devices, and can
221	do ECC calculation (not ECC error detection) for HAM1, BCH4, BCH8
222	and BCH16 ECC algorithms.
223
224   CONFIG_NAND_OMAP_ELM
225	Enables omap_elm.c driver for OMAPx and AMxxxx platforms.
226	ELM controller is used for ECC error detection (not ECC calculation)
227	of BCH4, BCH8 and BCH16 ECC algorithms.
228	Some legacy platforms like OMAP3xx do not have in-built ELM h/w engine,
229	thus such SoC platforms need to depend on software library for ECC error
230	detection. However ECC calculation on such plaforms would still be
231	done by GPMC controller.
232
233   CONFIG_SPL_NAND_AM33XX_BCH
234	Enables SPL-NAND driver (am335x_spl_bch.c) which supports ELM based
235        hardware ECC correction. This is useful for platforms which have ELM
236	hardware engine and use NAND boot mode.
237	Some legacy platforms like OMAP3xx do not have in-built ELM h/w engine,
238	so those platforms should use CONFIG_SPL_NAND_SIMPLE for enabling
239        SPL-NAND driver with software ECC correction support.
240
241   CONFIG_NAND_OMAP_ECCSCHEME
242	On OMAP platforms, this CONFIG specifies NAND ECC scheme.
243	It can take following values:
244	OMAP_ECC_HAM1_CODE_SW
245		1-bit Hamming code using software lib.
246		(for legacy devices only)
247	OMAP_ECC_HAM1_CODE_HW
248		1-bit Hamming code using GPMC hardware.
249		(for legacy devices only)
250	OMAP_ECC_BCH4_CODE_HW_DETECTION_SW
251		4-bit BCH code (unsupported)
252	OMAP_ECC_BCH4_CODE_HW
253		4-bit BCH code (unsupported)
254	OMAP_ECC_BCH8_CODE_HW_DETECTION_SW
255		8-bit BCH code with
256		- ecc calculation using GPMC hardware engine,
257		- error detection using software library.
258		- requires CONFIG_BCH to enable software BCH library
259		(For legacy device which do not have ELM h/w engine)
260	OMAP_ECC_BCH8_CODE_HW
261		8-bit BCH code with
262		- ecc calculation using GPMC hardware engine,
263		- error detection using ELM hardware engine.
264	OMAP_ECC_BCH16_CODE_HW
265		16-bit BCH code with
266		- ecc calculation using GPMC hardware engine,
267		- error detection using ELM hardware engine.
268
269	How to select ECC scheme on OMAP and AMxx platforms ?
270	-----------------------------------------------------
271	Though higher ECC schemes have more capability to detect and correct
272	bit-flips, but still selection of ECC scheme is dependent on following
273	- hardware engines present in SoC.
274		Some legacy OMAP SoC do not have ELM h/w engine thus such
275		SoC cannot support BCHx_HW ECC schemes.
276	- size of OOB/Spare region
277		With higher ECC schemes, more OOB/Spare area is required to
278		store ECC. So choice of ECC scheme is limited by NAND oobsize.
279
280	In general following expression can help:
281		NAND_OOBSIZE >= 2 + (NAND_PAGESIZE / 512) * ECC_BYTES
282	where
283		NAND_OOBSIZE	= number of bytes available in
284				OOB/spare area per NAND page.
285		NAND_PAGESIZE	= bytes in main-area of NAND page.
286		ECC_BYTES	= number of ECC bytes generated to
287				protect 512 bytes of data, which is:
288				3 for HAM1_xx ecc schemes
289				7 for BCH4_xx ecc schemes
290				14 for BCH8_xx ecc schemes
291				26 for BCH16_xx ecc schemes
292
293		example to check for BCH16 on 2K page NAND
294		NAND_PAGESIZE = 2048
295		NAND_OOBSIZE = 64
296		2 + (2048 / 512) * 26 = 106 > NAND_OOBSIZE
297		Thus BCH16 cannot be supported on 2K page NAND.
298
299		However, for 4K pagesize NAND
300		NAND_PAGESIZE = 4096
301		NAND_OOBSIZE = 64
302		ECC_BYTES = 26
303		2 + (4096 / 512) * 26 = 210 < NAND_OOBSIZE
304		Thus BCH16 can be supported on 4K page NAND.
305
306
307    CONFIG_NAND_OMAP_GPMC_PREFETCH
308	On OMAP platforms that use the GPMC controller
309	(CONFIG_NAND_OMAP_GPMC_PREFETCH), this options enables the code that
310	uses the prefetch mode to speed up read operations.
311
312NOTE:
313=====
314
315The current NAND implementation is based on what is in recent
316Linux kernels.  The old legacy implementation has been removed.
317
318If you have board code which used CONFIG_NAND_LEGACY, you'll need
319to convert to the current NAND interface for it to continue to work.
320
321The Disk On Chip driver is currently broken and has been for some time.
322There is a driver in drivers/mtd/nand, taken from Linux, that works with
323the current NAND system but has not yet been adapted to the u-boot
324environment.
325
326Additional improvements to the NAND subsystem by Guido Classen, 10-10-2006
327
328JFFS2 related commands:
329
330  implement "nand erase clean" and old "nand erase"
331  using both the new code which is able to skip bad blocks
332  "nand erase clean" additionally writes JFFS2-cleanmarkers in the oob.
333
334Miscellaneous and testing commands:
335  "markbad [offset]"
336  create an artificial bad block (for testing bad block handling)
337
338  "scrub [offset length]"
339  like "erase" but don't skip bad block. Instead erase them.
340  DANGEROUS!!! Factory set bad blocks will be lost. Use only
341  to remove artificial bad blocks created with the "markbad" command.
342
343  "torture offset"
344  Torture block to determine if it is still reliable.
345  Enabled by the CONFIG_CMD_NAND_TORTURE configuration option.
346  This command returns 0 if the block is still reliable, else 1.
347  If the block is detected as unreliable, it is up to the user to decide to
348  mark this block as bad.
349  The analyzed block is put through 3 erase / write cycles (or less if the block
350  is detected as unreliable earlier).
351  This command can be used in scripts, e.g. together with the markbad command to
352  automate retries and handling of possibly newly detected bad blocks if the
353  nand write command fails.
354  It can also be used manually by users having seen some NAND errors in logs to
355  search the root cause of these errors.
356  The underlying nand_torture() function is also useful for code willing to
357  automate actions following a nand->write() error. This would e.g. be required
358  in order to program or update safely firmware to NAND, especially for the UBI
359  part of such firmware.
360
361
362NAND locking command (for chips with active LOCKPRE pin)
363
364  "nand lock"
365  set NAND chip to lock state (all pages locked)
366
367  "nand lock tight"
368  set NAND chip to lock tight state (software can't change locking anymore)
369
370  "nand lock status"
371  displays current locking status of all pages
372
373  "nand unlock [offset] [size]"
374  unlock consecutive area (can be called multiple times for different areas)
375
376  "nand unlock.allexcept [offset] [size]"
377  unlock all except specified consecutive area
378
379I have tested the code with board containing 128MiB NAND large page chips
380and 32MiB small page chips.
381