mirror of
https://github.com/AsahiLinux/u-boot
synced 2024-11-26 14:40:41 +00:00
1221ce459d
Now, arch/${ARCH}/include/asm/errno.h and include/linux/errno.h have the same content. (both just wrap <asm-generic/errno.h>) Replace all include directives for <asm/errno.h> with <linux/errno.h>. Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> [trini: Fixup include/clk.] Signed-off-by: Tom Rini <trini@konsulko.com>
905 lines
22 KiB
C
905 lines
22 KiB
C
/*
|
|
* drivers/mtd/nand/nand_util.c
|
|
*
|
|
* Copyright (C) 2006 by Weiss-Electronic GmbH.
|
|
* All rights reserved.
|
|
*
|
|
* @author: Guido Classen <clagix@gmail.com>
|
|
* @descr: NAND Flash support
|
|
* @references: borrowed heavily from Linux mtd-utils code:
|
|
* flash_eraseall.c by Arcom Control System Ltd
|
|
* nandwrite.c by Steven J. Hill (sjhill@realitydiluted.com)
|
|
* and Thomas Gleixner (tglx@linutronix.de)
|
|
*
|
|
* Copyright (C) 2008 Nokia Corporation: drop_ffs() function by
|
|
* Artem Bityutskiy <dedekind1@gmail.com> from mtd-utils
|
|
*
|
|
* Copyright 2010 Freescale Semiconductor
|
|
*
|
|
* SPDX-License-Identifier: GPL-2.0
|
|
*/
|
|
|
|
#include <common.h>
|
|
#include <command.h>
|
|
#include <watchdog.h>
|
|
#include <malloc.h>
|
|
#include <memalign.h>
|
|
#include <div64.h>
|
|
|
|
#include <linux/errno.h>
|
|
#include <linux/mtd/mtd.h>
|
|
#include <nand.h>
|
|
#include <jffs2/jffs2.h>
|
|
|
|
typedef struct erase_info erase_info_t;
|
|
typedef struct mtd_info mtd_info_t;
|
|
|
|
/* support only for native endian JFFS2 */
|
|
#define cpu_to_je16(x) (x)
|
|
#define cpu_to_je32(x) (x)
|
|
|
|
/**
|
|
* nand_erase_opts: - erase NAND flash with support for various options
|
|
* (jffs2 formatting)
|
|
*
|
|
* @param mtd nand mtd instance to erase
|
|
* @param opts options, @see struct nand_erase_options
|
|
* @return 0 in case of success
|
|
*
|
|
* This code is ported from flash_eraseall.c from Linux mtd utils by
|
|
* Arcom Control System Ltd.
|
|
*/
|
|
int nand_erase_opts(struct mtd_info *mtd,
|
|
const nand_erase_options_t *opts)
|
|
{
|
|
struct jffs2_unknown_node cleanmarker;
|
|
erase_info_t erase;
|
|
unsigned long erase_length, erased_length; /* in blocks */
|
|
int result;
|
|
int percent_complete = -1;
|
|
const char *mtd_device = mtd->name;
|
|
struct mtd_oob_ops oob_opts;
|
|
struct nand_chip *chip = mtd_to_nand(mtd);
|
|
|
|
if ((opts->offset & (mtd->erasesize - 1)) != 0) {
|
|
printf("Attempt to erase non block-aligned data\n");
|
|
return -1;
|
|
}
|
|
|
|
memset(&erase, 0, sizeof(erase));
|
|
memset(&oob_opts, 0, sizeof(oob_opts));
|
|
|
|
erase.mtd = mtd;
|
|
erase.len = mtd->erasesize;
|
|
erase.addr = opts->offset;
|
|
erase_length = lldiv(opts->length + mtd->erasesize - 1,
|
|
mtd->erasesize);
|
|
|
|
cleanmarker.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
|
|
cleanmarker.nodetype = cpu_to_je16(JFFS2_NODETYPE_CLEANMARKER);
|
|
cleanmarker.totlen = cpu_to_je32(8);
|
|
|
|
/* scrub option allows to erase badblock. To prevent internal
|
|
* check from erase() method, set block check method to dummy
|
|
* and disable bad block table while erasing.
|
|
*/
|
|
if (opts->scrub) {
|
|
erase.scrub = opts->scrub;
|
|
/*
|
|
* We don't need the bad block table anymore...
|
|
* after scrub, there are no bad blocks left!
|
|
*/
|
|
if (chip->bbt) {
|
|
kfree(chip->bbt);
|
|
}
|
|
chip->bbt = NULL;
|
|
chip->options &= ~NAND_BBT_SCANNED;
|
|
}
|
|
|
|
for (erased_length = 0;
|
|
erased_length < erase_length;
|
|
erase.addr += mtd->erasesize) {
|
|
|
|
WATCHDOG_RESET();
|
|
|
|
if (opts->lim && (erase.addr >= (opts->offset + opts->lim))) {
|
|
puts("Size of erase exceeds limit\n");
|
|
return -EFBIG;
|
|
}
|
|
if (!opts->scrub) {
|
|
int ret = mtd_block_isbad(mtd, erase.addr);
|
|
if (ret > 0) {
|
|
if (!opts->quiet)
|
|
printf("\rSkipping bad block at "
|
|
"0x%08llx "
|
|
" \n",
|
|
erase.addr);
|
|
|
|
if (!opts->spread)
|
|
erased_length++;
|
|
|
|
continue;
|
|
|
|
} else if (ret < 0) {
|
|
printf("\n%s: MTD get bad block failed: %d\n",
|
|
mtd_device,
|
|
ret);
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
erased_length++;
|
|
|
|
result = mtd_erase(mtd, &erase);
|
|
if (result != 0) {
|
|
printf("\n%s: MTD Erase failure: %d\n",
|
|
mtd_device, result);
|
|
continue;
|
|
}
|
|
|
|
/* format for JFFS2 ? */
|
|
if (opts->jffs2 && chip->ecc.layout->oobavail >= 8) {
|
|
struct mtd_oob_ops ops;
|
|
ops.ooblen = 8;
|
|
ops.datbuf = NULL;
|
|
ops.oobbuf = (uint8_t *)&cleanmarker;
|
|
ops.ooboffs = 0;
|
|
ops.mode = MTD_OPS_AUTO_OOB;
|
|
|
|
result = mtd_write_oob(mtd, erase.addr, &ops);
|
|
if (result != 0) {
|
|
printf("\n%s: MTD writeoob failure: %d\n",
|
|
mtd_device, result);
|
|
continue;
|
|
}
|
|
}
|
|
|
|
if (!opts->quiet) {
|
|
unsigned long long n = erased_length * 100ULL;
|
|
int percent;
|
|
|
|
do_div(n, erase_length);
|
|
percent = (int)n;
|
|
|
|
/* output progress message only at whole percent
|
|
* steps to reduce the number of messages printed
|
|
* on (slow) serial consoles
|
|
*/
|
|
if (percent != percent_complete) {
|
|
percent_complete = percent;
|
|
|
|
printf("\rErasing at 0x%llx -- %3d%% complete.",
|
|
erase.addr, percent);
|
|
|
|
if (opts->jffs2 && result == 0)
|
|
printf(" Cleanmarker written at 0x%llx.",
|
|
erase.addr);
|
|
}
|
|
}
|
|
}
|
|
if (!opts->quiet)
|
|
printf("\n");
|
|
|
|
return 0;
|
|
}
|
|
|
|
#ifdef CONFIG_CMD_NAND_LOCK_UNLOCK
|
|
|
|
#define NAND_CMD_LOCK_TIGHT 0x2c
|
|
#define NAND_CMD_LOCK_STATUS 0x7a
|
|
|
|
/******************************************************************************
|
|
* Support for locking / unlocking operations of some NAND devices
|
|
*****************************************************************************/
|
|
|
|
/**
|
|
* nand_lock: Set all pages of NAND flash chip to the LOCK or LOCK-TIGHT
|
|
* state
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param tight bring device in lock tight mode
|
|
*
|
|
* @return 0 on success, -1 in case of error
|
|
*
|
|
* The lock / lock-tight command only applies to the whole chip. To get some
|
|
* parts of the chip lock and others unlocked use the following sequence:
|
|
*
|
|
* - Lock all pages of the chip using nand_lock(mtd, 0) (or the lockpre pin)
|
|
* - Call nand_unlock() once for each consecutive area to be unlocked
|
|
* - If desired: Bring the chip to the lock-tight state using nand_lock(mtd, 1)
|
|
*
|
|
* If the device is in lock-tight state software can't change the
|
|
* current active lock/unlock state of all pages. nand_lock() / nand_unlock()
|
|
* calls will fail. It is only posible to leave lock-tight state by
|
|
* an hardware signal (low pulse on _WP pin) or by power down.
|
|
*/
|
|
int nand_lock(struct mtd_info *mtd, int tight)
|
|
{
|
|
int ret = 0;
|
|
int status;
|
|
struct nand_chip *chip = mtd_to_nand(mtd);
|
|
|
|
/* select the NAND device */
|
|
chip->select_chip(mtd, 0);
|
|
|
|
/* check the Lock Tight Status */
|
|
chip->cmdfunc(mtd, NAND_CMD_LOCK_STATUS, -1, 0);
|
|
if (chip->read_byte(mtd) & NAND_LOCK_STATUS_TIGHT) {
|
|
printf("nand_lock: Device is locked tight!\n");
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
chip->cmdfunc(mtd,
|
|
(tight ? NAND_CMD_LOCK_TIGHT : NAND_CMD_LOCK),
|
|
-1, -1);
|
|
|
|
/* call wait ready function */
|
|
status = chip->waitfunc(mtd, chip);
|
|
|
|
/* see if device thinks it succeeded */
|
|
if (status & 0x01) {
|
|
ret = -1;
|
|
}
|
|
|
|
out:
|
|
/* de-select the NAND device */
|
|
chip->select_chip(mtd, -1);
|
|
return ret;
|
|
}
|
|
|
|
/**
|
|
* nand_get_lock_status: - query current lock state from one page of NAND
|
|
* flash
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param offset page address to query (must be page-aligned!)
|
|
*
|
|
* @return -1 in case of error
|
|
* >0 lock status:
|
|
* bitfield with the following combinations:
|
|
* NAND_LOCK_STATUS_TIGHT: page in tight state
|
|
* NAND_LOCK_STATUS_UNLOCK: page unlocked
|
|
*
|
|
*/
|
|
int nand_get_lock_status(struct mtd_info *mtd, loff_t offset)
|
|
{
|
|
int ret = 0;
|
|
int chipnr;
|
|
int page;
|
|
struct nand_chip *chip = mtd_to_nand(mtd);
|
|
|
|
/* select the NAND device */
|
|
chipnr = (int)(offset >> chip->chip_shift);
|
|
chip->select_chip(mtd, chipnr);
|
|
|
|
|
|
if ((offset & (mtd->writesize - 1)) != 0) {
|
|
printf("nand_get_lock_status: "
|
|
"Start address must be beginning of "
|
|
"nand page!\n");
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
/* check the Lock Status */
|
|
page = (int)(offset >> chip->page_shift);
|
|
chip->cmdfunc(mtd, NAND_CMD_LOCK_STATUS, -1, page & chip->pagemask);
|
|
|
|
ret = chip->read_byte(mtd) & (NAND_LOCK_STATUS_TIGHT
|
|
| NAND_LOCK_STATUS_UNLOCK);
|
|
|
|
out:
|
|
/* de-select the NAND device */
|
|
chip->select_chip(mtd, -1);
|
|
return ret;
|
|
}
|
|
|
|
/**
|
|
* nand_unlock: - Unlock area of NAND pages
|
|
* only one consecutive area can be unlocked at one time!
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param start start byte address
|
|
* @param length number of bytes to unlock (must be a multiple of
|
|
* page size mtd->writesize)
|
|
* @param allexcept if set, unlock everything not selected
|
|
*
|
|
* @return 0 on success, -1 in case of error
|
|
*/
|
|
int nand_unlock(struct mtd_info *mtd, loff_t start, size_t length,
|
|
int allexcept)
|
|
{
|
|
int ret = 0;
|
|
int chipnr;
|
|
int status;
|
|
int page;
|
|
struct nand_chip *chip = mtd_to_nand(mtd);
|
|
|
|
debug("nand_unlock%s: start: %08llx, length: %zd!\n",
|
|
allexcept ? " (allexcept)" : "", start, length);
|
|
|
|
/* select the NAND device */
|
|
chipnr = (int)(start >> chip->chip_shift);
|
|
chip->select_chip(mtd, chipnr);
|
|
|
|
/* check the WP bit */
|
|
chip->cmdfunc(mtd, NAND_CMD_STATUS, -1, -1);
|
|
if (!(chip->read_byte(mtd) & NAND_STATUS_WP)) {
|
|
printf("nand_unlock: Device is write protected!\n");
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
/* check the Lock Tight Status */
|
|
page = (int)(start >> chip->page_shift);
|
|
chip->cmdfunc(mtd, NAND_CMD_LOCK_STATUS, -1, page & chip->pagemask);
|
|
if (chip->read_byte(mtd) & NAND_LOCK_STATUS_TIGHT) {
|
|
printf("nand_unlock: Device is locked tight!\n");
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
if ((start & (mtd->erasesize - 1)) != 0) {
|
|
printf("nand_unlock: Start address must be beginning of "
|
|
"nand block!\n");
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
if (length == 0 || (length & (mtd->erasesize - 1)) != 0) {
|
|
printf("nand_unlock: Length must be a multiple of nand block "
|
|
"size %08x!\n", mtd->erasesize);
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
/*
|
|
* Set length so that the last address is set to the
|
|
* starting address of the last block
|
|
*/
|
|
length -= mtd->erasesize;
|
|
|
|
/* submit address of first page to unlock */
|
|
chip->cmdfunc(mtd, NAND_CMD_UNLOCK1, -1, page & chip->pagemask);
|
|
|
|
/* submit ADDRESS of LAST page to unlock */
|
|
page += (int)(length >> chip->page_shift);
|
|
|
|
/*
|
|
* Page addresses for unlocking are supposed to be block-aligned.
|
|
* At least some NAND chips use the low bit to indicate that the
|
|
* page range should be inverted.
|
|
*/
|
|
if (allexcept)
|
|
page |= 1;
|
|
|
|
chip->cmdfunc(mtd, NAND_CMD_UNLOCK2, -1, page & chip->pagemask);
|
|
|
|
/* call wait ready function */
|
|
status = chip->waitfunc(mtd, chip);
|
|
/* see if device thinks it succeeded */
|
|
if (status & 0x01) {
|
|
/* there was an error */
|
|
ret = -1;
|
|
goto out;
|
|
}
|
|
|
|
out:
|
|
/* de-select the NAND device */
|
|
chip->select_chip(mtd, -1);
|
|
return ret;
|
|
}
|
|
#endif
|
|
|
|
/**
|
|
* check_skip_len
|
|
*
|
|
* Check if there are any bad blocks, and whether length including bad
|
|
* blocks fits into device
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param offset offset in flash
|
|
* @param length image length
|
|
* @param used length of flash needed for the requested length
|
|
* @return 0 if the image fits and there are no bad blocks
|
|
* 1 if the image fits, but there are bad blocks
|
|
* -1 if the image does not fit
|
|
*/
|
|
static int check_skip_len(struct mtd_info *mtd, loff_t offset, size_t length,
|
|
size_t *used)
|
|
{
|
|
size_t len_excl_bad = 0;
|
|
int ret = 0;
|
|
|
|
while (len_excl_bad < length) {
|
|
size_t block_len, block_off;
|
|
loff_t block_start;
|
|
|
|
if (offset >= mtd->size)
|
|
return -1;
|
|
|
|
block_start = offset & ~(loff_t)(mtd->erasesize - 1);
|
|
block_off = offset & (mtd->erasesize - 1);
|
|
block_len = mtd->erasesize - block_off;
|
|
|
|
if (!nand_block_isbad(mtd, block_start))
|
|
len_excl_bad += block_len;
|
|
else
|
|
ret = 1;
|
|
|
|
offset += block_len;
|
|
*used += block_len;
|
|
}
|
|
|
|
/* If the length is not a multiple of block_len, adjust. */
|
|
if (len_excl_bad > length)
|
|
*used -= (len_excl_bad - length);
|
|
|
|
return ret;
|
|
}
|
|
|
|
#ifdef CONFIG_CMD_NAND_TRIMFFS
|
|
static size_t drop_ffs(const struct mtd_info *mtd, const u_char *buf,
|
|
const size_t *len)
|
|
{
|
|
size_t l = *len;
|
|
ssize_t i;
|
|
|
|
for (i = l - 1; i >= 0; i--)
|
|
if (buf[i] != 0xFF)
|
|
break;
|
|
|
|
/* The resulting length must be aligned to the minimum flash I/O size */
|
|
l = i + 1;
|
|
l = (l + mtd->writesize - 1) / mtd->writesize;
|
|
l *= mtd->writesize;
|
|
|
|
/*
|
|
* since the input length may be unaligned, prevent access past the end
|
|
* of the buffer
|
|
*/
|
|
return min(l, *len);
|
|
}
|
|
#endif
|
|
|
|
/**
|
|
* nand_verify_page_oob:
|
|
*
|
|
* Verify a page of NAND flash, including the OOB.
|
|
* Reads page of NAND and verifies the contents and OOB against the
|
|
* values in ops.
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param ops MTD operations, including data to verify
|
|
* @param ofs offset in flash
|
|
* @return 0 in case of success
|
|
*/
|
|
int nand_verify_page_oob(struct mtd_info *mtd, struct mtd_oob_ops *ops,
|
|
loff_t ofs)
|
|
{
|
|
int rval;
|
|
struct mtd_oob_ops vops;
|
|
size_t verlen = mtd->writesize + mtd->oobsize;
|
|
|
|
memcpy(&vops, ops, sizeof(vops));
|
|
|
|
vops.datbuf = memalign(ARCH_DMA_MINALIGN, verlen);
|
|
|
|
if (!vops.datbuf)
|
|
return -ENOMEM;
|
|
|
|
vops.oobbuf = vops.datbuf + mtd->writesize;
|
|
|
|
rval = mtd_read_oob(mtd, ofs, &vops);
|
|
if (!rval)
|
|
rval = memcmp(ops->datbuf, vops.datbuf, vops.len);
|
|
if (!rval)
|
|
rval = memcmp(ops->oobbuf, vops.oobbuf, vops.ooblen);
|
|
|
|
free(vops.datbuf);
|
|
|
|
return rval ? -EIO : 0;
|
|
}
|
|
|
|
/**
|
|
* nand_verify:
|
|
*
|
|
* Verify a region of NAND flash.
|
|
* Reads NAND in page-sized chunks and verifies the contents against
|
|
* the contents of a buffer. The offset into the NAND must be
|
|
* page-aligned, and the function doesn't handle skipping bad blocks.
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param ofs offset in flash
|
|
* @param len buffer length
|
|
* @param buf buffer to read from
|
|
* @return 0 in case of success
|
|
*/
|
|
int nand_verify(struct mtd_info *mtd, loff_t ofs, size_t len, u_char *buf)
|
|
{
|
|
int rval = 0;
|
|
size_t verofs;
|
|
size_t verlen = mtd->writesize;
|
|
uint8_t *verbuf = memalign(ARCH_DMA_MINALIGN, verlen);
|
|
|
|
if (!verbuf)
|
|
return -ENOMEM;
|
|
|
|
/* Read the NAND back in page-size groups to limit malloc size */
|
|
for (verofs = ofs; verofs < ofs + len;
|
|
verofs += verlen, buf += verlen) {
|
|
verlen = min(mtd->writesize, (uint32_t)(ofs + len - verofs));
|
|
rval = nand_read(mtd, verofs, &verlen, verbuf);
|
|
if (!rval || (rval == -EUCLEAN))
|
|
rval = memcmp(buf, verbuf, verlen);
|
|
|
|
if (rval)
|
|
break;
|
|
}
|
|
|
|
free(verbuf);
|
|
|
|
return rval ? -EIO : 0;
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
* nand_write_skip_bad:
|
|
*
|
|
* Write image to NAND flash.
|
|
* Blocks that are marked bad are skipped and the is written to the next
|
|
* block instead as long as the image is short enough to fit even after
|
|
* skipping the bad blocks. Due to bad blocks we may not be able to
|
|
* perform the requested write. In the case where the write would
|
|
* extend beyond the end of the NAND device, both length and actual (if
|
|
* not NULL) are set to 0. In the case where the write would extend
|
|
* beyond the limit we are passed, length is set to 0 and actual is set
|
|
* to the required length.
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param offset offset in flash
|
|
* @param length buffer length
|
|
* @param actual set to size required to write length worth of
|
|
* buffer or 0 on error, if not NULL
|
|
* @param lim maximum size that actual may be in order to not
|
|
* exceed the buffer
|
|
* @param buffer buffer to read from
|
|
* @param flags flags modifying the behaviour of the write to NAND
|
|
* @return 0 in case of success
|
|
*/
|
|
int nand_write_skip_bad(struct mtd_info *mtd, loff_t offset, size_t *length,
|
|
size_t *actual, loff_t lim, u_char *buffer, int flags)
|
|
{
|
|
int rval = 0, blocksize;
|
|
size_t left_to_write = *length;
|
|
size_t used_for_write = 0;
|
|
u_char *p_buffer = buffer;
|
|
int need_skip;
|
|
|
|
if (actual)
|
|
*actual = 0;
|
|
|
|
blocksize = mtd->erasesize;
|
|
|
|
/*
|
|
* nand_write() handles unaligned, partial page writes.
|
|
*
|
|
* We allow length to be unaligned, for convenience in
|
|
* using the $filesize variable.
|
|
*
|
|
* However, starting at an unaligned offset makes the
|
|
* semantics of bad block skipping ambiguous (really,
|
|
* you should only start a block skipping access at a
|
|
* partition boundary). So don't try to handle that.
|
|
*/
|
|
if ((offset & (mtd->writesize - 1)) != 0) {
|
|
printf("Attempt to write non page-aligned data\n");
|
|
*length = 0;
|
|
return -EINVAL;
|
|
}
|
|
|
|
need_skip = check_skip_len(mtd, offset, *length, &used_for_write);
|
|
|
|
if (actual)
|
|
*actual = used_for_write;
|
|
|
|
if (need_skip < 0) {
|
|
printf("Attempt to write outside the flash area\n");
|
|
*length = 0;
|
|
return -EINVAL;
|
|
}
|
|
|
|
if (used_for_write > lim) {
|
|
puts("Size of write exceeds partition or device limit\n");
|
|
*length = 0;
|
|
return -EFBIG;
|
|
}
|
|
|
|
if (!need_skip && !(flags & WITH_DROP_FFS)) {
|
|
rval = nand_write(mtd, offset, length, buffer);
|
|
|
|
if ((flags & WITH_WR_VERIFY) && !rval)
|
|
rval = nand_verify(mtd, offset, *length, buffer);
|
|
|
|
if (rval == 0)
|
|
return 0;
|
|
|
|
*length = 0;
|
|
printf("NAND write to offset %llx failed %d\n",
|
|
offset, rval);
|
|
return rval;
|
|
}
|
|
|
|
while (left_to_write > 0) {
|
|
size_t block_offset = offset & (mtd->erasesize - 1);
|
|
size_t write_size, truncated_write_size;
|
|
|
|
WATCHDOG_RESET();
|
|
|
|
if (nand_block_isbad(mtd, offset & ~(mtd->erasesize - 1))) {
|
|
printf("Skip bad block 0x%08llx\n",
|
|
offset & ~(mtd->erasesize - 1));
|
|
offset += mtd->erasesize - block_offset;
|
|
continue;
|
|
}
|
|
|
|
if (left_to_write < (blocksize - block_offset))
|
|
write_size = left_to_write;
|
|
else
|
|
write_size = blocksize - block_offset;
|
|
|
|
truncated_write_size = write_size;
|
|
#ifdef CONFIG_CMD_NAND_TRIMFFS
|
|
if (flags & WITH_DROP_FFS)
|
|
truncated_write_size = drop_ffs(mtd, p_buffer,
|
|
&write_size);
|
|
#endif
|
|
|
|
rval = nand_write(mtd, offset, &truncated_write_size,
|
|
p_buffer);
|
|
|
|
if ((flags & WITH_WR_VERIFY) && !rval)
|
|
rval = nand_verify(mtd, offset,
|
|
truncated_write_size, p_buffer);
|
|
|
|
offset += write_size;
|
|
p_buffer += write_size;
|
|
|
|
if (rval != 0) {
|
|
printf("NAND write to offset %llx failed %d\n",
|
|
offset, rval);
|
|
*length -= left_to_write;
|
|
return rval;
|
|
}
|
|
|
|
left_to_write -= write_size;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
/**
|
|
* nand_read_skip_bad:
|
|
*
|
|
* Read image from NAND flash.
|
|
* Blocks that are marked bad are skipped and the next block is read
|
|
* instead as long as the image is short enough to fit even after
|
|
* skipping the bad blocks. Due to bad blocks we may not be able to
|
|
* perform the requested read. In the case where the read would extend
|
|
* beyond the end of the NAND device, both length and actual (if not
|
|
* NULL) are set to 0. In the case where the read would extend beyond
|
|
* the limit we are passed, length is set to 0 and actual is set to the
|
|
* required length.
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param offset offset in flash
|
|
* @param length buffer length, on return holds number of read bytes
|
|
* @param actual set to size required to read length worth of buffer or 0
|
|
* on error, if not NULL
|
|
* @param lim maximum size that actual may be in order to not exceed the
|
|
* buffer
|
|
* @param buffer buffer to write to
|
|
* @return 0 in case of success
|
|
*/
|
|
int nand_read_skip_bad(struct mtd_info *mtd, loff_t offset, size_t *length,
|
|
size_t *actual, loff_t lim, u_char *buffer)
|
|
{
|
|
int rval;
|
|
size_t left_to_read = *length;
|
|
size_t used_for_read = 0;
|
|
u_char *p_buffer = buffer;
|
|
int need_skip;
|
|
|
|
if ((offset & (mtd->writesize - 1)) != 0) {
|
|
printf("Attempt to read non page-aligned data\n");
|
|
*length = 0;
|
|
if (actual)
|
|
*actual = 0;
|
|
return -EINVAL;
|
|
}
|
|
|
|
need_skip = check_skip_len(mtd, offset, *length, &used_for_read);
|
|
|
|
if (actual)
|
|
*actual = used_for_read;
|
|
|
|
if (need_skip < 0) {
|
|
printf("Attempt to read outside the flash area\n");
|
|
*length = 0;
|
|
return -EINVAL;
|
|
}
|
|
|
|
if (used_for_read > lim) {
|
|
puts("Size of read exceeds partition or device limit\n");
|
|
*length = 0;
|
|
return -EFBIG;
|
|
}
|
|
|
|
if (!need_skip) {
|
|
rval = nand_read(mtd, offset, length, buffer);
|
|
if (!rval || rval == -EUCLEAN)
|
|
return 0;
|
|
|
|
*length = 0;
|
|
printf("NAND read from offset %llx failed %d\n",
|
|
offset, rval);
|
|
return rval;
|
|
}
|
|
|
|
while (left_to_read > 0) {
|
|
size_t block_offset = offset & (mtd->erasesize - 1);
|
|
size_t read_length;
|
|
|
|
WATCHDOG_RESET();
|
|
|
|
if (nand_block_isbad(mtd, offset & ~(mtd->erasesize - 1))) {
|
|
printf("Skipping bad block 0x%08llx\n",
|
|
offset & ~(mtd->erasesize - 1));
|
|
offset += mtd->erasesize - block_offset;
|
|
continue;
|
|
}
|
|
|
|
if (left_to_read < (mtd->erasesize - block_offset))
|
|
read_length = left_to_read;
|
|
else
|
|
read_length = mtd->erasesize - block_offset;
|
|
|
|
rval = nand_read(mtd, offset, &read_length, p_buffer);
|
|
if (rval && rval != -EUCLEAN) {
|
|
printf("NAND read from offset %llx failed %d\n",
|
|
offset, rval);
|
|
*length -= left_to_read;
|
|
return rval;
|
|
}
|
|
|
|
left_to_read -= read_length;
|
|
offset += read_length;
|
|
p_buffer += read_length;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
#ifdef CONFIG_CMD_NAND_TORTURE
|
|
|
|
/**
|
|
* check_pattern:
|
|
*
|
|
* Check if buffer contains only a certain byte pattern.
|
|
*
|
|
* @param buf buffer to check
|
|
* @param patt the pattern to check
|
|
* @param size buffer size in bytes
|
|
* @return 1 if there are only patt bytes in buf
|
|
* 0 if something else was found
|
|
*/
|
|
static int check_pattern(const u_char *buf, u_char patt, int size)
|
|
{
|
|
int i;
|
|
|
|
for (i = 0; i < size; i++)
|
|
if (buf[i] != patt)
|
|
return 0;
|
|
return 1;
|
|
}
|
|
|
|
/**
|
|
* nand_torture:
|
|
*
|
|
* Torture a block of NAND flash.
|
|
* This is useful to determine if a block that caused a write error is still
|
|
* good or should be marked as bad.
|
|
*
|
|
* @param mtd nand mtd instance
|
|
* @param offset offset in flash
|
|
* @return 0 if the block is still good
|
|
*/
|
|
int nand_torture(struct mtd_info *mtd, loff_t offset)
|
|
{
|
|
u_char patterns[] = {0xa5, 0x5a, 0x00};
|
|
struct erase_info instr = {
|
|
.mtd = mtd,
|
|
.addr = offset,
|
|
.len = mtd->erasesize,
|
|
};
|
|
size_t retlen;
|
|
int err, ret = -1, i, patt_count;
|
|
u_char *buf;
|
|
|
|
if ((offset & (mtd->erasesize - 1)) != 0) {
|
|
puts("Attempt to torture a block at a non block-aligned offset\n");
|
|
return -EINVAL;
|
|
}
|
|
|
|
if (offset + mtd->erasesize > mtd->size) {
|
|
puts("Attempt to torture a block outside the flash area\n");
|
|
return -EINVAL;
|
|
}
|
|
|
|
patt_count = ARRAY_SIZE(patterns);
|
|
|
|
buf = malloc_cache_aligned(mtd->erasesize);
|
|
if (buf == NULL) {
|
|
puts("Out of memory for erase block buffer\n");
|
|
return -ENOMEM;
|
|
}
|
|
|
|
for (i = 0; i < patt_count; i++) {
|
|
err = mtd_erase(mtd, &instr);
|
|
if (err) {
|
|
printf("%s: erase() failed for block at 0x%llx: %d\n",
|
|
mtd->name, instr.addr, err);
|
|
goto out;
|
|
}
|
|
|
|
/* Make sure the block contains only 0xff bytes */
|
|
err = mtd_read(mtd, offset, mtd->erasesize, &retlen, buf);
|
|
if ((err && err != -EUCLEAN) || retlen != mtd->erasesize) {
|
|
printf("%s: read() failed for block at 0x%llx: %d\n",
|
|
mtd->name, instr.addr, err);
|
|
goto out;
|
|
}
|
|
|
|
err = check_pattern(buf, 0xff, mtd->erasesize);
|
|
if (!err) {
|
|
printf("Erased block at 0x%llx, but a non-0xff byte was found\n",
|
|
offset);
|
|
ret = -EIO;
|
|
goto out;
|
|
}
|
|
|
|
/* Write a pattern and check it */
|
|
memset(buf, patterns[i], mtd->erasesize);
|
|
err = mtd_write(mtd, offset, mtd->erasesize, &retlen, buf);
|
|
if (err || retlen != mtd->erasesize) {
|
|
printf("%s: write() failed for block at 0x%llx: %d\n",
|
|
mtd->name, instr.addr, err);
|
|
goto out;
|
|
}
|
|
|
|
err = mtd_read(mtd, offset, mtd->erasesize, &retlen, buf);
|
|
if ((err && err != -EUCLEAN) || retlen != mtd->erasesize) {
|
|
printf("%s: read() failed for block at 0x%llx: %d\n",
|
|
mtd->name, instr.addr, err);
|
|
goto out;
|
|
}
|
|
|
|
err = check_pattern(buf, patterns[i], mtd->erasesize);
|
|
if (!err) {
|
|
printf("Pattern 0x%.2x checking failed for block at "
|
|
"0x%llx\n", patterns[i], offset);
|
|
ret = -EIO;
|
|
goto out;
|
|
}
|
|
}
|
|
|
|
ret = 0;
|
|
|
|
out:
|
|
free(buf);
|
|
return ret;
|
|
}
|
|
|
|
#endif
|