linux/drivers/mtd/inftlcore.c

968 lines
25 KiB
C
Raw Normal View History

/*
* inftlcore.c -- Linux driver for Inverse Flash Translation Layer (INFTL)
*
* (C) Copyright 2002, Greg Ungerer (gerg@snapgear.com)
*
* Based heavily on the nftlcore.c code which is:
* (c) 1999 Machine Vision Holdings, Inc.
* Author: David Woodhouse <dwmw2@infradead.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/delay.h>
#include <linux/slab.h>
#include <linux/sched.h>
#include <linux/init.h>
#include <linux/kmod.h>
#include <linux/hdreg.h>
#include <linux/mtd/mtd.h>
#include <linux/mtd/nftl.h>
#include <linux/mtd/inftl.h>
#include <linux/mtd/nand.h>
#include <asm/uaccess.h>
#include <asm/errno.h>
#include <asm/io.h>
/*
* Maximum number of loops while examining next block, to have a
* chance to detect consistency problems (they should never happen
* because of the checks done in the mounting.
*/
#define MAX_LOOPS 10000
static void inftl_add_mtd(struct mtd_blktrans_ops *tr, struct mtd_info *mtd)
{
struct INFTLrecord *inftl;
unsigned long temp;
if (mtd->type != MTD_NANDFLASH)
return;
/* OK, this is moderately ugly. But probably safe. Alternatives? */
if (memcmp(mtd->name, "DiskOnChip", 10))
return;
if (!mtd->block_isbad) {
printk(KERN_ERR
"INFTL no longer supports the old DiskOnChip drivers loaded via docprobe.\n"
"Please use the new diskonchip driver under the NAND subsystem.\n");
return;
}
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: add_mtd for %s\n", mtd->name);
inftl = kzalloc(sizeof(*inftl), GFP_KERNEL);
if (!inftl) {
printk(KERN_WARNING "INFTL: Out of memory for data structures\n");
return;
}
inftl->mbd.mtd = mtd;
inftl->mbd.devnum = -1;
inftl->mbd.tr = tr;
if (INFTL_mount(inftl) < 0) {
printk(KERN_WARNING "INFTL: could not mount device\n");
kfree(inftl);
return;
}
/* OK, it's a new one. Set up all the data structures. */
/* Calculate geometry */
inftl->cylinders = 1024;
inftl->heads = 16;
temp = inftl->cylinders * inftl->heads;
inftl->sectors = inftl->mbd.size / temp;
if (inftl->mbd.size % temp) {
inftl->sectors++;
temp = inftl->cylinders * inftl->sectors;
inftl->heads = inftl->mbd.size / temp;
if (inftl->mbd.size % temp) {
inftl->heads++;
temp = inftl->heads * inftl->sectors;
inftl->cylinders = inftl->mbd.size / temp;
}
}
if (inftl->mbd.size != inftl->heads * inftl->cylinders * inftl->sectors) {
/*
Oh no we don't have
mbd.size == heads * cylinders * sectors
*/
printk(KERN_WARNING "INFTL: cannot calculate a geometry to "
"match size of 0x%lx.\n", inftl->mbd.size);
printk(KERN_WARNING "INFTL: using C:%d H:%d S:%d "
"(== 0x%lx sects)\n",
inftl->cylinders, inftl->heads , inftl->sectors,
(long)inftl->cylinders * (long)inftl->heads *
(long)inftl->sectors );
}
if (add_mtd_blktrans_dev(&inftl->mbd)) {
kfree(inftl->PUtable);
kfree(inftl->VUtable);
kfree(inftl);
return;
}
#ifdef PSYCHO_DEBUG
printk(KERN_INFO "INFTL: Found new inftl%c\n", inftl->mbd.devnum + 'a');
#endif
return;
}
static void inftl_remove_dev(struct mtd_blktrans_dev *dev)
{
struct INFTLrecord *inftl = (void *)dev;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: remove_dev (i=%d)\n", dev->devnum);
del_mtd_blktrans_dev(dev);
kfree(inftl->PUtable);
kfree(inftl->VUtable);
kfree(inftl);
}
/*
* Actual INFTL access routines.
*/
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
/*
* Read oob data from flash
*/
int inftl_read_oob(struct mtd_info *mtd, loff_t offs, size_t len,
size_t *retlen, uint8_t *buf)
{
struct mtd_oob_ops ops;
int res;
ops.mode = MTD_OOB_PLACE;
ops.ooboffs = offs & (mtd->writesize - 1);
ops.ooblen = len;
ops.oobbuf = buf;
ops.datbuf = NULL;
res = mtd->read_oob(mtd, offs & ~(mtd->writesize - 1), &ops);
*retlen = ops.oobretlen;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
return res;
}
/*
* Write oob data to flash
*/
int inftl_write_oob(struct mtd_info *mtd, loff_t offs, size_t len,
size_t *retlen, uint8_t *buf)
{
struct mtd_oob_ops ops;
int res;
ops.mode = MTD_OOB_PLACE;
ops.ooboffs = offs & (mtd->writesize - 1);
ops.ooblen = len;
ops.oobbuf = buf;
ops.datbuf = NULL;
res = mtd->write_oob(mtd, offs & ~(mtd->writesize - 1), &ops);
*retlen = ops.oobretlen;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
return res;
}
/*
* Write data and oob to flash
*/
static int inftl_write(struct mtd_info *mtd, loff_t offs, size_t len,
size_t *retlen, uint8_t *buf, uint8_t *oob)
{
struct mtd_oob_ops ops;
int res;
ops.mode = MTD_OOB_PLACE;
ops.ooboffs = offs;
ops.ooblen = mtd->oobsize;
ops.oobbuf = oob;
ops.datbuf = buf;
ops.len = len;
res = mtd->write_oob(mtd, offs & ~(mtd->writesize - 1), &ops);
*retlen = ops.retlen;
return res;
}
/*
* INFTL_findfreeblock: Find a free Erase Unit on the INFTL partition.
* This function is used when the give Virtual Unit Chain.
*/
static u16 INFTL_findfreeblock(struct INFTLrecord *inftl, int desperate)
{
u16 pot = inftl->LastFreeEUN;
int silly = inftl->nb_blocks;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: INFTL_findfreeblock(inftl=%p,"
"desperate=%d)\n", inftl, desperate);
/*
* Normally, we force a fold to happen before we run out of free
* blocks completely.
*/
if (!desperate && inftl->numfreeEUNs < 2) {
DEBUG(MTD_DEBUG_LEVEL1, "INFTL: there are too few free "
"EUNs (%d)\n", inftl->numfreeEUNs);
return 0xffff;
}
/* Scan for a free block */
do {
if (inftl->PUtable[pot] == BLOCK_FREE) {
inftl->LastFreeEUN = pot;
return pot;
}
if (++pot > inftl->lastEUN)
pot = 0;
if (!silly--) {
printk(KERN_WARNING "INFTL: no free blocks found! "
"EUN range = %d - %d\n", 0, inftl->LastFreeEUN);
return BLOCK_NIL;
}
} while (pot != inftl->LastFreeEUN);
return BLOCK_NIL;
}
static u16 INFTL_foldchain(struct INFTLrecord *inftl, unsigned thisVUC, unsigned pendingblock)
{
u16 BlockMap[MAX_SECTORS_PER_UNIT];
unsigned char BlockDeleted[MAX_SECTORS_PER_UNIT];
unsigned int thisEUN, prevEUN, status;
struct mtd_info *mtd = inftl->mbd.mtd;
int block, silly;
unsigned int targetEUN;
struct inftl_oob oob;
size_t retlen;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: INFTL_foldchain(inftl=%p,thisVUC=%d,"
"pending=%d)\n", inftl, thisVUC, pendingblock);
memset(BlockMap, 0xff, sizeof(BlockMap));
memset(BlockDeleted, 0, sizeof(BlockDeleted));
thisEUN = targetEUN = inftl->VUtable[thisVUC];
if (thisEUN == BLOCK_NIL) {
printk(KERN_WARNING "INFTL: trying to fold non-existent "
"Virtual Unit Chain %d!\n", thisVUC);
return BLOCK_NIL;
}
/*
* Scan to find the Erase Unit which holds the actual data for each
* 512-byte block within the Chain.
*/
silly = MAX_LOOPS;
while (thisEUN < inftl->nb_blocks) {
for (block = 0; block < inftl->EraseSize/SECTORSIZE; block ++) {
if ((BlockMap[block] != 0xffff) || BlockDeleted[block])
continue;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
if (inftl_read_oob(mtd, (thisEUN * inftl->EraseSize)
+ (block * SECTORSIZE), 16, &retlen,
(char *)&oob) < 0)
status = SECTOR_IGNORE;
else
status = oob.b.Status | oob.b.Status1;
switch(status) {
case SECTOR_FREE:
case SECTOR_IGNORE:
break;
case SECTOR_USED:
BlockMap[block] = thisEUN;
continue;
case SECTOR_DELETED:
BlockDeleted[block] = 1;
continue;
default:
printk(KERN_WARNING "INFTL: unknown status "
"for block %d in EUN %d: %x\n",
block, thisEUN, status);
break;
}
}
if (!silly--) {
printk(KERN_WARNING "INFTL: infinite loop in Virtual "
"Unit Chain 0x%x\n", thisVUC);
return BLOCK_NIL;
}
thisEUN = inftl->PUtable[thisEUN];
}
/*
* OK. We now know the location of every block in the Virtual Unit
* Chain, and the Erase Unit into which we are supposed to be copying.
* Go for it.
*/
DEBUG(MTD_DEBUG_LEVEL1, "INFTL: folding chain %d into unit %d\n",
thisVUC, targetEUN);
for (block = 0; block < inftl->EraseSize/SECTORSIZE ; block++) {
unsigned char movebuf[SECTORSIZE];
int ret;
/*
* If it's in the target EUN already, or if it's pending write,
* do nothing.
*/
if (BlockMap[block] == targetEUN || (pendingblock ==
(thisVUC * (inftl->EraseSize / SECTORSIZE) + block))) {
continue;
}
/*
* Copy only in non free block (free blocks can only
* happen in case of media errors or deleted blocks).
*/
if (BlockMap[block] == BLOCK_NIL)
continue;
ret = mtd->read(mtd, (inftl->EraseSize * BlockMap[block]) +
(block * SECTORSIZE), SECTORSIZE, &retlen,
movebuf);
if (ret < 0 && ret != -EUCLEAN) {
ret = mtd->read(mtd,
(inftl->EraseSize * BlockMap[block]) +
(block * SECTORSIZE), SECTORSIZE,
&retlen, movebuf);
if (ret != -EIO)
DEBUG(MTD_DEBUG_LEVEL1, "INFTL: error went "
"away on retry?\n");
}
memset(&oob, 0xff, sizeof(struct inftl_oob));
oob.b.Status = oob.b.Status1 = SECTOR_USED;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
inftl_write(inftl->mbd.mtd, (inftl->EraseSize * targetEUN) +
(block * SECTORSIZE), SECTORSIZE, &retlen,
movebuf, (char *)&oob);
}
/*
* Newest unit in chain now contains data from _all_ older units.
* So go through and erase each unit in chain, oldest first. (This
* is important, by doing oldest first if we crash/reboot then it
* it is relatively simple to clean up the mess).
*/
DEBUG(MTD_DEBUG_LEVEL1, "INFTL: want to erase virtual chain %d\n",
thisVUC);
for (;;) {
/* Find oldest unit in chain. */
thisEUN = inftl->VUtable[thisVUC];
prevEUN = BLOCK_NIL;
while (inftl->PUtable[thisEUN] != BLOCK_NIL) {
prevEUN = thisEUN;
thisEUN = inftl->PUtable[thisEUN];
}
/* Check if we are all done */
if (thisEUN == targetEUN)
break;
if (INFTL_formatblock(inftl, thisEUN) < 0) {
/*
* Could not erase : mark block as reserved.
*/
inftl->PUtable[thisEUN] = BLOCK_RESERVED;
} else {
/* Correctly erased : mark it as free */
inftl->PUtable[thisEUN] = BLOCK_FREE;
inftl->PUtable[prevEUN] = BLOCK_NIL;
inftl->numfreeEUNs++;
}
}
return targetEUN;
}
static u16 INFTL_makefreeblock(struct INFTLrecord *inftl, unsigned pendingblock)
{
/*
* This is the part that needs some cleverness applied.
* For now, I'm doing the minimum applicable to actually
* get the thing to work.
* Wear-levelling and other clever stuff needs to be implemented
* and we also need to do some assessment of the results when
* the system loses power half-way through the routine.
*/
u16 LongestChain = 0;
u16 ChainLength = 0, thislen;
u16 chain, EUN;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: INFTL_makefreeblock(inftl=%p,"
"pending=%d)\n", inftl, pendingblock);
for (chain = 0; chain < inftl->nb_blocks; chain++) {
EUN = inftl->VUtable[chain];
thislen = 0;
while (EUN <= inftl->lastEUN) {
thislen++;
EUN = inftl->PUtable[EUN];
if (thislen > 0xff00) {
printk(KERN_WARNING "INFTL: endless loop in "
"Virtual Chain %d: Unit %x\n",
chain, EUN);
/*
* Actually, don't return failure.
* Just ignore this chain and get on with it.
*/
thislen = 0;
break;
}
}
if (thislen > ChainLength) {
ChainLength = thislen;
LongestChain = chain;
}
}
if (ChainLength < 2) {
printk(KERN_WARNING "INFTL: no Virtual Unit Chains available "
"for folding. Failing request\n");
return BLOCK_NIL;
}
return INFTL_foldchain(inftl, LongestChain, pendingblock);
}
static int nrbits(unsigned int val, int bitcount)
{
int i, total = 0;
for (i = 0; (i < bitcount); i++)
total += (((0x1 << i) & val) ? 1 : 0);
return total;
}
/*
* INFTL_findwriteunit: Return the unit number into which we can write
* for this block. Make it available if it isn't already.
*/
static inline u16 INFTL_findwriteunit(struct INFTLrecord *inftl, unsigned block)
{
unsigned int thisVUC = block / (inftl->EraseSize / SECTORSIZE);
unsigned int thisEUN, writeEUN, prev_block, status;
unsigned long blockofs = (block * SECTORSIZE) & (inftl->EraseSize -1);
struct mtd_info *mtd = inftl->mbd.mtd;
struct inftl_oob oob;
struct inftl_bci bci;
unsigned char anac, nacs, parity;
size_t retlen;
int silly, silly2 = 3;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: INFTL_findwriteunit(inftl=%p,"
"block=%d)\n", inftl, block);
do {
/*
* Scan the media to find a unit in the VUC which has
* a free space for the block in question.
*/
writeEUN = BLOCK_NIL;
thisEUN = inftl->VUtable[thisVUC];
silly = MAX_LOOPS;
while (thisEUN <= inftl->lastEUN) {
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
inftl_read_oob(mtd, (thisEUN * inftl->EraseSize) +
blockofs, 8, &retlen, (char *)&bci);
status = bci.Status | bci.Status1;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: status of block %d in "
"EUN %d is %x\n", block , writeEUN, status);
switch(status) {
case SECTOR_FREE:
writeEUN = thisEUN;
break;
case SECTOR_DELETED:
case SECTOR_USED:
/* Can't go any further */
goto hitused;
case SECTOR_IGNORE:
break;
default:
/*
* Invalid block. Don't use it any more.
* Must implement.
*/
break;
}
if (!silly--) {
printk(KERN_WARNING "INFTL: infinite loop in "
"Virtual Unit Chain 0x%x\n", thisVUC);
return 0xffff;
}
/* Skip to next block in chain */
thisEUN = inftl->PUtable[thisEUN];
}
hitused:
if (writeEUN != BLOCK_NIL)
return writeEUN;
/*
* OK. We didn't find one in the existing chain, or there
* is no existing chain. Allocate a new one.
*/
writeEUN = INFTL_findfreeblock(inftl, 0);
if (writeEUN == BLOCK_NIL) {
/*
* That didn't work - there were no free blocks just
* waiting to be picked up. We're going to have to fold
* a chain to make room.
*/
thisEUN = INFTL_makefreeblock(inftl, 0xffff);
/*
* Hopefully we free something, lets try again.
* This time we are desperate...
*/
DEBUG(MTD_DEBUG_LEVEL1, "INFTL: using desperate==1 "
"to find free EUN to accommodate write to "
"VUC %d\n", thisVUC);
writeEUN = INFTL_findfreeblock(inftl, 1);
if (writeEUN == BLOCK_NIL) {
/*
* Ouch. This should never happen - we should
* always be able to make some room somehow.
* If we get here, we've allocated more storage
* space than actual media, or our makefreeblock
* routine is missing something.
*/
printk(KERN_WARNING "INFTL: cannot make free "
"space.\n");
#ifdef DEBUG
INFTL_dumptables(inftl);
INFTL_dumpVUchains(inftl);
#endif
return BLOCK_NIL;
}
}
/*
* Insert new block into virtual chain. Firstly update the
* block headers in flash...
*/
anac = 0;
nacs = 0;
thisEUN = inftl->VUtable[thisVUC];
if (thisEUN != BLOCK_NIL) {
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
inftl_read_oob(mtd, thisEUN * inftl->EraseSize
+ 8, 8, &retlen, (char *)&oob.u);
anac = oob.u.a.ANAC + 1;
nacs = oob.u.a.NACs + 1;
}
prev_block = inftl->VUtable[thisVUC];
if (prev_block < inftl->nb_blocks)
prev_block -= inftl->firstEUN;
parity = (nrbits(thisVUC, 16) & 0x1) ? 0x1 : 0;
parity |= (nrbits(prev_block, 16) & 0x1) ? 0x2 : 0;
parity |= (nrbits(anac, 8) & 0x1) ? 0x4 : 0;
parity |= (nrbits(nacs, 8) & 0x1) ? 0x8 : 0;
oob.u.a.virtualUnitNo = cpu_to_le16(thisVUC);
oob.u.a.prevUnitNo = cpu_to_le16(prev_block);
oob.u.a.ANAC = anac;
oob.u.a.NACs = nacs;
oob.u.a.parityPerField = parity;
oob.u.a.discarded = 0xaa;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
inftl_write_oob(mtd, writeEUN * inftl->EraseSize + 8, 8,
&retlen, (char *)&oob.u);
/* Also back up header... */
oob.u.b.virtualUnitNo = cpu_to_le16(thisVUC);
oob.u.b.prevUnitNo = cpu_to_le16(prev_block);
oob.u.b.ANAC = anac;
oob.u.b.NACs = nacs;
oob.u.b.parityPerField = parity;
oob.u.b.discarded = 0xaa;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
inftl_write_oob(mtd, writeEUN * inftl->EraseSize +
SECTORSIZE * 4 + 8, 8, &retlen, (char *)&oob.u);
inftl->PUtable[writeEUN] = inftl->VUtable[thisVUC];
inftl->VUtable[thisVUC] = writeEUN;
inftl->numfreeEUNs--;
return writeEUN;
} while (silly2--);
printk(KERN_WARNING "INFTL: error folding to make room for Virtual "
"Unit Chain 0x%x\n", thisVUC);
return 0xffff;
}
/*
* Given a Virtual Unit Chain, see if it can be deleted, and if so do it.
*/
static void INFTL_trydeletechain(struct INFTLrecord *inftl, unsigned thisVUC)
{
struct mtd_info *mtd = inftl->mbd.mtd;
unsigned char BlockUsed[MAX_SECTORS_PER_UNIT];
unsigned char BlockDeleted[MAX_SECTORS_PER_UNIT];
unsigned int thisEUN, status;
int block, silly;
struct inftl_bci bci;
size_t retlen;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: INFTL_trydeletechain(inftl=%p,"
"thisVUC=%d)\n", inftl, thisVUC);
memset(BlockUsed, 0, sizeof(BlockUsed));
memset(BlockDeleted, 0, sizeof(BlockDeleted));
thisEUN = inftl->VUtable[thisVUC];
if (thisEUN == BLOCK_NIL) {
printk(KERN_WARNING "INFTL: trying to delete non-existent "
"Virtual Unit Chain %d!\n", thisVUC);
return;
}
/*
* Scan through the Erase Units to determine whether any data is in
* each of the 512-byte blocks within the Chain.
*/
silly = MAX_LOOPS;
while (thisEUN < inftl->nb_blocks) {
for (block = 0; block < inftl->EraseSize/SECTORSIZE; block++) {
if (BlockUsed[block] || BlockDeleted[block])
continue;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
if (inftl_read_oob(mtd, (thisEUN * inftl->EraseSize)
+ (block * SECTORSIZE), 8 , &retlen,
(char *)&bci) < 0)
status = SECTOR_IGNORE;
else
status = bci.Status | bci.Status1;
switch(status) {
case SECTOR_FREE:
case SECTOR_IGNORE:
break;
case SECTOR_USED:
BlockUsed[block] = 1;
continue;
case SECTOR_DELETED:
BlockDeleted[block] = 1;
continue;
default:
printk(KERN_WARNING "INFTL: unknown status "
"for block %d in EUN %d: 0x%x\n",
block, thisEUN, status);
}
}
if (!silly--) {
printk(KERN_WARNING "INFTL: infinite loop in Virtual "
"Unit Chain 0x%x\n", thisVUC);
return;
}
thisEUN = inftl->PUtable[thisEUN];
}
for (block = 0; block < inftl->EraseSize/SECTORSIZE; block++)
if (BlockUsed[block])
return;
/*
* For each block in the chain free it and make it available
* for future use. Erase from the oldest unit first.
*/
DEBUG(MTD_DEBUG_LEVEL1, "INFTL: deleting empty VUC %d\n", thisVUC);
for (;;) {
u16 *prevEUN = &inftl->VUtable[thisVUC];
thisEUN = *prevEUN;
/* If the chain is all gone already, we're done */
if (thisEUN == BLOCK_NIL) {
DEBUG(MTD_DEBUG_LEVEL2, "INFTL: Empty VUC %d for deletion was already absent\n", thisEUN);
return;
}
/* Find oldest unit in chain. */
while (inftl->PUtable[thisEUN] != BLOCK_NIL) {
BUG_ON(thisEUN >= inftl->nb_blocks);
prevEUN = &inftl->PUtable[thisEUN];
thisEUN = *prevEUN;
}
DEBUG(MTD_DEBUG_LEVEL3, "Deleting EUN %d from VUC %d\n",
thisEUN, thisVUC);
if (INFTL_formatblock(inftl, thisEUN) < 0) {
/*
* Could not erase : mark block as reserved.
*/
inftl->PUtable[thisEUN] = BLOCK_RESERVED;
} else {
/* Correctly erased : mark it as free */
inftl->PUtable[thisEUN] = BLOCK_FREE;
inftl->numfreeEUNs++;
}
/* Now sort out whatever was pointing to it... */
*prevEUN = BLOCK_NIL;
/* Ideally we'd actually be responsive to new
requests while we're doing this -- if there's
free space why should others be made to wait? */
cond_resched();
}
inftl->VUtable[thisVUC] = BLOCK_NIL;
}
static int INFTL_deleteblock(struct INFTLrecord *inftl, unsigned block)
{
unsigned int thisEUN = inftl->VUtable[block / (inftl->EraseSize / SECTORSIZE)];
unsigned long blockofs = (block * SECTORSIZE) & (inftl->EraseSize - 1);
struct mtd_info *mtd = inftl->mbd.mtd;
unsigned int status;
int silly = MAX_LOOPS;
size_t retlen;
struct inftl_bci bci;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: INFTL_deleteblock(inftl=%p,"
"block=%d)\n", inftl, block);
while (thisEUN < inftl->nb_blocks) {
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
if (inftl_read_oob(mtd, (thisEUN * inftl->EraseSize) +
blockofs, 8, &retlen, (char *)&bci) < 0)
status = SECTOR_IGNORE;
else
status = bci.Status | bci.Status1;
switch (status) {
case SECTOR_FREE:
case SECTOR_IGNORE:
break;
case SECTOR_DELETED:
thisEUN = BLOCK_NIL;
goto foundit;
case SECTOR_USED:
goto foundit;
default:
printk(KERN_WARNING "INFTL: unknown status for "
"block %d in EUN %d: 0x%x\n",
block, thisEUN, status);
break;
}
if (!silly--) {
printk(KERN_WARNING "INFTL: infinite loop in Virtual "
"Unit Chain 0x%x\n",
block / (inftl->EraseSize / SECTORSIZE));
return 1;
}
thisEUN = inftl->PUtable[thisEUN];
}
foundit:
if (thisEUN != BLOCK_NIL) {
loff_t ptr = (thisEUN * inftl->EraseSize) + blockofs;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
if (inftl_read_oob(mtd, ptr, 8, &retlen, (char *)&bci) < 0)
return -EIO;
bci.Status = bci.Status1 = SECTOR_DELETED;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
if (inftl_write_oob(mtd, ptr, 8, &retlen, (char *)&bci) < 0)
return -EIO;
INFTL_trydeletechain(inftl, block / (inftl->EraseSize / SECTORSIZE));
}
return 0;
}
static int inftl_writeblock(struct mtd_blktrans_dev *mbd, unsigned long block,
char *buffer)
{
struct INFTLrecord *inftl = (void *)mbd;
unsigned int writeEUN;
unsigned long blockofs = (block * SECTORSIZE) & (inftl->EraseSize - 1);
size_t retlen;
struct inftl_oob oob;
char *p, *pend;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: inftl_writeblock(inftl=%p,block=%ld,"
"buffer=%p)\n", inftl, block, buffer);
/* Is block all zero? */
pend = buffer + SECTORSIZE;
for (p = buffer; p < pend && !*p; p++)
;
if (p < pend) {
writeEUN = INFTL_findwriteunit(inftl, block);
if (writeEUN == BLOCK_NIL) {
printk(KERN_WARNING "inftl_writeblock(): cannot find "
"block to write to\n");
/*
* If we _still_ haven't got a block to use,
* we're screwed.
*/
return 1;
}
memset(&oob, 0xff, sizeof(struct inftl_oob));
oob.b.Status = oob.b.Status1 = SECTOR_USED;
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
inftl_write(inftl->mbd.mtd, (writeEUN * inftl->EraseSize) +
blockofs, SECTORSIZE, &retlen, (char *)buffer,
(char *)&oob);
/*
* need to write SECTOR_USED flags since they are not written
* in mtd_writeecc
*/
} else {
INFTL_deleteblock(inftl, block);
}
return 0;
}
static int inftl_readblock(struct mtd_blktrans_dev *mbd, unsigned long block,
char *buffer)
{
struct INFTLrecord *inftl = (void *)mbd;
unsigned int thisEUN = inftl->VUtable[block / (inftl->EraseSize / SECTORSIZE)];
unsigned long blockofs = (block * SECTORSIZE) & (inftl->EraseSize - 1);
struct mtd_info *mtd = inftl->mbd.mtd;
unsigned int status;
int silly = MAX_LOOPS;
struct inftl_bci bci;
size_t retlen;
DEBUG(MTD_DEBUG_LEVEL3, "INFTL: inftl_readblock(inftl=%p,block=%ld,"
"buffer=%p)\n", inftl, block, buffer);
while (thisEUN < inftl->nb_blocks) {
[MTD] Rework the out of band handling completely Hopefully the last iteration on this! The handling of out of band data on NAND was accompanied by tons of fruitless discussions and halfarsed patches to make it work for a particular problem. Sufficiently annoyed by I all those "I know it better" mails and the resonable amount of discarded "it solves my problem" patches, I finally decided to go for the big rework. After removing the _ecc variants of mtd read/write functions the solution to satisfy the various requirements was to refactor the read/write _oob functions in mtd. The major change is that read/write_oob now takes a pointer to an operation descriptor structure "struct mtd_oob_ops".instead of having a function with at least seven arguments. read/write_oob which should probably renamed to a more descriptive name, can do the following tasks: - read/write out of band data - read/write data content and out of band data - read/write raw data content and out of band data (ecc disabled) struct mtd_oob_ops has a mode field, which determines the oob handling mode. Aside of the MTD_OOB_RAW mode, which is intended to be especially for diagnostic purposes and some internal functions e.g. bad block table creation, the other two modes are for mtd clients: MTD_OOB_PLACE puts/gets the given oob data exactly to/from the place which is described by the ooboffs and ooblen fields of the mtd_oob_ops strcuture. It's up to the caller to make sure that the byte positions are not used by the ECC placement algorithms. MTD_OOB_AUTO puts/gets the given oob data automaticaly to/from the places in the out of band area which are described by the oobfree tuples in the ecclayout data structre which is associated to the devicee. The decision whether data plus oob or oob only handling is done depends on the setting of the datbuf member of the data structure. When datbuf == NULL then the internal read/write_oob functions are selected, otherwise the read/write data routines are invoked. Tested on a few platforms with all variants. Please be aware of possible regressions for your particular device / application scenario Disclaimer: Any whining will be ignored from those who just contributed "hot air blurb" and never sat down to tackle the underlying problem of the mess in the NAND driver grown over time and the big chunk of work to fix up the existing users. The problem was not the holiness of the existing MTD interfaces. The problems was the lack of time to go for the big overhaul. It's easy to add more mess to the existing one, but it takes alot of effort to go for a real solution. Improvements and bugfixes are welcome! Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2006-05-29 09:26:58 +08:00
if (inftl_read_oob(mtd, (thisEUN * inftl->EraseSize) +
blockofs, 8, &retlen, (char *)&bci) < 0)
status = SECTOR_IGNORE;
else
status = bci.Status | bci.Status1;
switch (status) {
case SECTOR_DELETED:
thisEUN = BLOCK_NIL;
goto foundit;
case SECTOR_USED:
goto foundit;
case SECTOR_FREE:
case SECTOR_IGNORE:
break;
default:
printk(KERN_WARNING "INFTL: unknown status for "
"block %ld in EUN %d: 0x%04x\n",
block, thisEUN, status);
break;
}
if (!silly--) {
printk(KERN_WARNING "INFTL: infinite loop in "
"Virtual Unit Chain 0x%lx\n",
block / (inftl->EraseSize / SECTORSIZE));
return 1;
}
thisEUN = inftl->PUtable[thisEUN];
}
foundit:
if (thisEUN == BLOCK_NIL) {
/* The requested block is not on the media, return all 0x00 */
memset(buffer, 0, SECTORSIZE);
} else {
size_t retlen;
loff_t ptr = (thisEUN * inftl->EraseSize) + blockofs;
int ret = mtd->read(mtd, ptr, SECTORSIZE, &retlen, buffer);
/* Handle corrected bit flips gracefully */
if (ret < 0 && ret != -EUCLEAN)
return -EIO;
}
return 0;
}
static int inftl_getgeo(struct mtd_blktrans_dev *dev, struct hd_geometry *geo)
{
struct INFTLrecord *inftl = (void *)dev;
geo->heads = inftl->heads;
geo->sectors = inftl->sectors;
geo->cylinders = inftl->cylinders;
return 0;
}
static struct mtd_blktrans_ops inftl_tr = {
.name = "inftl",
.major = INFTL_MAJOR,
.part_bits = INFTL_PARTN_BITS,
.blksize = 512,
.getgeo = inftl_getgeo,
.readsect = inftl_readblock,
.writesect = inftl_writeblock,
.add_mtd = inftl_add_mtd,
.remove_dev = inftl_remove_dev,
.owner = THIS_MODULE,
};
static int __init init_inftl(void)
{
return register_mtd_blktrans(&inftl_tr);
}
static void __exit cleanup_inftl(void)
{
deregister_mtd_blktrans(&inftl_tr);
}
module_init(init_inftl);
module_exit(cleanup_inftl);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Greg Ungerer <gerg@snapgear.com>, David Woodhouse <dwmw2@infradead.org>, Fabrice Bellard <fabrice.bellard@netgem.com> et al.");
MODULE_DESCRIPTION("Support code for Inverse Flash Translation Layer, used on M-Systems DiskOnChip 2000, Millennium and Millennium Plus");