linux/drivers/mtd/mtdconcat.c

995 lines
23 KiB
C
Raw Normal View History

/*
* MTD device concatenation layer
*
* (C) 2002 Robert Kaiser <rkaiser@sysgo.de>
*
* NAND support by Christian Gan <cgan@iders.ca>
*
* This code is GPL
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/sched.h>
#include <linux/types.h>
#include <linux/backing-dev.h>
#include <linux/mtd/mtd.h>
#include <linux/mtd/concat.h>
#include <asm/div64.h>
/*
* Our storage structure:
* Subdev points to an array of pointers to struct mtd_info objects
* which is allocated along with this structure
*
*/
struct mtd_concat {
struct mtd_info mtd;
int num_subdev;
struct mtd_info **subdev;
};
/*
* how to calculate the size required for the above structure,
* including the pointer array subdev points to:
*/
#define SIZEOF_STRUCT_MTD_CONCAT(num_subdev) \
((sizeof(struct mtd_concat) + (num_subdev) * sizeof(struct mtd_info *)))
/*
* Given a pointer to the MTD object in the mtd_concat structure,
* we can retrieve the pointer to that structure with this macro.
*/
#define CONCAT(x) ((struct mtd_concat *)(x))
/*
* MTD methods which look up the relevant subdevice, translate the
* effective address and pass through to the subdevice.
*/
static int
concat_read(struct mtd_info *mtd, loff_t from, size_t len,
size_t * retlen, u_char * buf)
{
struct mtd_concat *concat = CONCAT(mtd);
int ret = 0, err;
int i;
*retlen = 0;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
size_t size, retsize;
if (from >= subdev->size) {
/* Not destined for this subdev */
size = 0;
from -= subdev->size;
continue;
}
if (from + len > subdev->size)
/* First part goes into this subdev */
size = subdev->size - from;
else
/* Entire transaction goes into this subdev */
size = len;
err = subdev->read(subdev, from, size, &retsize, buf);
/* Save information about bitflips! */
if (unlikely(err)) {
if (err == -EBADMSG) {
mtd->ecc_stats.failed++;
ret = err;
} else if (err == -EUCLEAN) {
mtd->ecc_stats.corrected++;
/* Do not overwrite -EBADMSG !! */
if (!ret)
ret = err;
} else
return err;
}
*retlen += retsize;
len -= size;
if (len == 0)
return ret;
buf += size;
from = 0;
}
return -EINVAL;
}
static int
concat_write(struct mtd_info *mtd, loff_t to, size_t len,
size_t * retlen, const u_char * buf)
{
struct mtd_concat *concat = CONCAT(mtd);
int err = -EINVAL;
int i;
if (!(mtd->flags & MTD_WRITEABLE))
return -EROFS;
*retlen = 0;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
size_t size, retsize;
if (to >= subdev->size) {
size = 0;
to -= subdev->size;
continue;
}
if (to + len > subdev->size)
size = subdev->size - to;
else
size = len;
if (!(subdev->flags & MTD_WRITEABLE))
err = -EROFS;
else
err = subdev->write(subdev, to, size, &retsize, buf);
if (err)
break;
*retlen += retsize;
len -= size;
if (len == 0)
break;
err = -EINVAL;
buf += size;
to = 0;
}
return err;
}
static int
concat_writev(struct mtd_info *mtd, const struct kvec *vecs,
unsigned long count, loff_t to, size_t * retlen)
{
struct mtd_concat *concat = CONCAT(mtd);
struct kvec *vecs_copy;
unsigned long entry_low, entry_high;
size_t total_len = 0;
int i;
int err = -EINVAL;
if (!(mtd->flags & MTD_WRITEABLE))
return -EROFS;
*retlen = 0;
/* Calculate total length of data */
for (i = 0; i < count; i++)
total_len += vecs[i].iov_len;
/* Do not allow write past end of device */
if ((to + total_len) > mtd->size)
return -EINVAL;
/* Check alignment */
if (mtd->writesize > 1) {
uint64_t __to = to;
if (do_div(__to, mtd->writesize) || (total_len % mtd->writesize))
return -EINVAL;
}
/* make a copy of vecs */
vecs_copy = kmalloc(sizeof(struct kvec) * count, GFP_KERNEL);
if (!vecs_copy)
return -ENOMEM;
memcpy(vecs_copy, vecs, sizeof(struct kvec) * count);
entry_low = 0;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
size_t size, wsize, retsize, old_iov_len;
if (to >= subdev->size) {
to -= subdev->size;
continue;
}
size = min_t(uint64_t, total_len, subdev->size - to);
wsize = size; /* store for future use */
entry_high = entry_low;
while (entry_high < count) {
if (size <= vecs_copy[entry_high].iov_len)
break;
size -= vecs_copy[entry_high++].iov_len;
}
old_iov_len = vecs_copy[entry_high].iov_len;
vecs_copy[entry_high].iov_len = size;
if (!(subdev->flags & MTD_WRITEABLE))
err = -EROFS;
else
err = subdev->writev(subdev, &vecs_copy[entry_low],
entry_high - entry_low + 1, to, &retsize);
vecs_copy[entry_high].iov_len = old_iov_len - size;
vecs_copy[entry_high].iov_base += size;
entry_low = entry_high;
if (err)
break;
*retlen += retsize;
total_len -= wsize;
if (total_len == 0)
break;
err = -EINVAL;
to = 0;
}
kfree(vecs_copy);
return err;
}
static int
[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
concat_read_oob(struct mtd_info *mtd, loff_t from, struct mtd_oob_ops *ops)
{
struct mtd_concat *concat = CONCAT(mtd);
[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
struct mtd_oob_ops devops = *ops;
int i, err, ret = 0;
ops->retlen = ops->oobretlen = 0;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
if (from >= subdev->size) {
from -= subdev->size;
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
/* partial read ? */
if (from + devops.len > subdev->size)
devops.len = subdev->size - from;
[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
err = subdev->read_oob(subdev, from, &devops);
ops->retlen += devops.retlen;
ops->oobretlen += devops.oobretlen;
/* Save information about bitflips! */
if (unlikely(err)) {
if (err == -EBADMSG) {
mtd->ecc_stats.failed++;
ret = err;
} else if (err == -EUCLEAN) {
mtd->ecc_stats.corrected++;
/* Do not overwrite -EBADMSG !! */
if (!ret)
ret = err;
} else
return err;
}
if (devops.datbuf) {
devops.len = ops->len - ops->retlen;
if (!devops.len)
return ret;
[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
devops.datbuf += devops.retlen;
}
if (devops.oobbuf) {
devops.ooblen = ops->ooblen - ops->oobretlen;
if (!devops.ooblen)
return ret;
devops.oobbuf += ops->oobretlen;
}
from = 0;
}
[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 -EINVAL;
}
static int
[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
concat_write_oob(struct mtd_info *mtd, loff_t to, struct mtd_oob_ops *ops)
{
struct mtd_concat *concat = CONCAT(mtd);
[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
struct mtd_oob_ops devops = *ops;
int i, err;
if (!(mtd->flags & MTD_WRITEABLE))
return -EROFS;
[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
ops->retlen = 0;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
if (to >= subdev->size) {
to -= subdev->size;
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
/* partial write ? */
if (to + devops.len > subdev->size)
devops.len = subdev->size - to;
[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
err = subdev->write_oob(subdev, to, &devops);
ops->retlen += devops.retlen;
if (err)
[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 err;
if (devops.datbuf) {
devops.len = ops->len - ops->retlen;
if (!devops.len)
return 0;
[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
devops.datbuf += devops.retlen;
}
if (devops.oobbuf) {
devops.ooblen = ops->ooblen - ops->oobretlen;
if (!devops.ooblen)
return 0;
devops.oobbuf += devops.oobretlen;
}
to = 0;
}
[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 -EINVAL;
}
static void concat_erase_callback(struct erase_info *instr)
{
wake_up((wait_queue_head_t *) instr->priv);
}
static int concat_dev_erase(struct mtd_info *mtd, struct erase_info *erase)
{
int err;
wait_queue_head_t waitq;
DECLARE_WAITQUEUE(wait, current);
/*
* This code was stol^H^H^H^Hinspired by mtdchar.c
*/
init_waitqueue_head(&waitq);
erase->mtd = mtd;
erase->callback = concat_erase_callback;
erase->priv = (unsigned long) &waitq;
/*
* FIXME: Allow INTERRUPTIBLE. Which means
* not having the wait_queue head on the stack.
*/
err = mtd->erase(mtd, erase);
if (!err) {
set_current_state(TASK_UNINTERRUPTIBLE);
add_wait_queue(&waitq, &wait);
if (erase->state != MTD_ERASE_DONE
&& erase->state != MTD_ERASE_FAILED)
schedule();
remove_wait_queue(&waitq, &wait);
set_current_state(TASK_RUNNING);
err = (erase->state == MTD_ERASE_FAILED) ? -EIO : 0;
}
return err;
}
static int concat_erase(struct mtd_info *mtd, struct erase_info *instr)
{
struct mtd_concat *concat = CONCAT(mtd);
struct mtd_info *subdev;
int i, err;
uint64_t length, offset = 0;
struct erase_info *erase;
if (!(mtd->flags & MTD_WRITEABLE))
return -EROFS;
if (instr->addr > concat->mtd.size)
return -EINVAL;
if (instr->len + instr->addr > concat->mtd.size)
return -EINVAL;
/*
* Check for proper erase block alignment of the to-be-erased area.
* It is easier to do this based on the super device's erase
* region info rather than looking at each particular sub-device
* in turn.
*/
if (!concat->mtd.numeraseregions) {
/* the easy case: device has uniform erase block size */
if (instr->addr & (concat->mtd.erasesize - 1))
return -EINVAL;
if (instr->len & (concat->mtd.erasesize - 1))
return -EINVAL;
} else {
/* device has variable erase size */
struct mtd_erase_region_info *erase_regions =
concat->mtd.eraseregions;
/*
* Find the erase region where the to-be-erased area begins:
*/
for (i = 0; i < concat->mtd.numeraseregions &&
instr->addr >= erase_regions[i].offset; i++) ;
--i;
/*
* Now erase_regions[i] is the region in which the
* to-be-erased area begins. Verify that the starting
* offset is aligned to this region's erase size:
*/
if (instr->addr & (erase_regions[i].erasesize - 1))
return -EINVAL;
/*
* now find the erase region where the to-be-erased area ends:
*/
for (; i < concat->mtd.numeraseregions &&
(instr->addr + instr->len) >= erase_regions[i].offset;
++i) ;
--i;
/*
* check if the ending offset is aligned to this region's erase size
*/
if ((instr->addr + instr->len) & (erase_regions[i].erasesize -
1))
return -EINVAL;
}
instr->fail_addr = MTD_FAIL_ADDR_UNKNOWN;
/* make a local copy of instr to avoid modifying the caller's struct */
erase = kmalloc(sizeof (struct erase_info), GFP_KERNEL);
if (!erase)
return -ENOMEM;
*erase = *instr;
length = instr->len;
/*
* find the subdevice where the to-be-erased area begins, adjust
* starting offset to be relative to the subdevice start
*/
for (i = 0; i < concat->num_subdev; i++) {
subdev = concat->subdev[i];
if (subdev->size <= erase->addr) {
erase->addr -= subdev->size;
offset += subdev->size;
} else {
break;
}
}
/* must never happen since size limit has been verified above */
BUG_ON(i >= concat->num_subdev);
/* now do the erase: */
err = 0;
for (; length > 0; i++) {
/* loop for all subdevices affected by this request */
subdev = concat->subdev[i]; /* get current subdevice */
/* limit length to subdevice's size: */
if (erase->addr + length > subdev->size)
erase->len = subdev->size - erase->addr;
else
erase->len = length;
if (!(subdev->flags & MTD_WRITEABLE)) {
err = -EROFS;
break;
}
length -= erase->len;
if ((err = concat_dev_erase(subdev, erase))) {
/* sanity check: should never happen since
* block alignment has been checked above */
BUG_ON(err == -EINVAL);
if (erase->fail_addr != MTD_FAIL_ADDR_UNKNOWN)
instr->fail_addr = erase->fail_addr + offset;
break;
}
/*
* erase->addr specifies the offset of the area to be
* erased *within the current subdevice*. It can be
* non-zero only the first time through this loop, i.e.
* for the first subdevice where blocks need to be erased.
* All the following erases must begin at the start of the
* current subdevice, i.e. at offset zero.
*/
erase->addr = 0;
offset += subdev->size;
}
instr->state = erase->state;
kfree(erase);
if (err)
return err;
if (instr->callback)
instr->callback(instr);
return 0;
}
static int concat_lock(struct mtd_info *mtd, loff_t ofs, uint64_t len)
{
struct mtd_concat *concat = CONCAT(mtd);
int i, err = -EINVAL;
if ((len + ofs) > mtd->size)
return -EINVAL;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
uint64_t size;
if (ofs >= subdev->size) {
size = 0;
ofs -= subdev->size;
continue;
}
if (ofs + len > subdev->size)
size = subdev->size - ofs;
else
size = len;
err = subdev->lock(subdev, ofs, size);
if (err)
break;
len -= size;
if (len == 0)
break;
err = -EINVAL;
ofs = 0;
}
return err;
}
static int concat_unlock(struct mtd_info *mtd, loff_t ofs, uint64_t len)
{
struct mtd_concat *concat = CONCAT(mtd);
int i, err = 0;
if ((len + ofs) > mtd->size)
return -EINVAL;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
uint64_t size;
if (ofs >= subdev->size) {
size = 0;
ofs -= subdev->size;
continue;
}
if (ofs + len > subdev->size)
size = subdev->size - ofs;
else
size = len;
err = subdev->unlock(subdev, ofs, size);
if (err)
break;
len -= size;
if (len == 0)
break;
err = -EINVAL;
ofs = 0;
}
return err;
}
static void concat_sync(struct mtd_info *mtd)
{
struct mtd_concat *concat = CONCAT(mtd);
int i;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
subdev->sync(subdev);
}
}
static int concat_suspend(struct mtd_info *mtd)
{
struct mtd_concat *concat = CONCAT(mtd);
int i, rc = 0;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
if ((rc = subdev->suspend(subdev)) < 0)
return rc;
}
return rc;
}
static void concat_resume(struct mtd_info *mtd)
{
struct mtd_concat *concat = CONCAT(mtd);
int i;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
subdev->resume(subdev);
}
}
static int concat_block_isbad(struct mtd_info *mtd, loff_t ofs)
{
struct mtd_concat *concat = CONCAT(mtd);
int i, res = 0;
if (!concat->subdev[0]->block_isbad)
return res;
if (ofs > mtd->size)
return -EINVAL;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
if (ofs >= subdev->size) {
ofs -= subdev->size;
continue;
}
res = subdev->block_isbad(subdev, ofs);
break;
}
return res;
}
static int concat_block_markbad(struct mtd_info *mtd, loff_t ofs)
{
struct mtd_concat *concat = CONCAT(mtd);
int i, err = -EINVAL;
if (!concat->subdev[0]->block_markbad)
return 0;
if (ofs > mtd->size)
return -EINVAL;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
if (ofs >= subdev->size) {
ofs -= subdev->size;
continue;
}
err = subdev->block_markbad(subdev, ofs);
if (!err)
mtd->ecc_stats.badblocks++;
break;
}
return err;
}
/*
* try to support NOMMU mmaps on concatenated devices
* - we don't support subdev spanning as we can't guarantee it'll work
*/
static unsigned long concat_get_unmapped_area(struct mtd_info *mtd,
unsigned long len,
unsigned long offset,
unsigned long flags)
{
struct mtd_concat *concat = CONCAT(mtd);
int i;
for (i = 0; i < concat->num_subdev; i++) {
struct mtd_info *subdev = concat->subdev[i];
if (offset >= subdev->size) {
offset -= subdev->size;
continue;
}
/* we've found the subdev over which the mapping will reside */
if (offset + len > subdev->size)
return (unsigned long) -EINVAL;
if (subdev->get_unmapped_area)
return subdev->get_unmapped_area(subdev, len, offset,
flags);
break;
}
return (unsigned long) -ENOSYS;
}
/*
* This function constructs a virtual MTD device by concatenating
* num_devs MTD devices. A pointer to the new device object is
* stored to *new_dev upon success. This function does _not_
* register any devices: this is the caller's responsibility.
*/
struct mtd_info *mtd_concat_create(struct mtd_info *subdev[], /* subdevices to concatenate */
int num_devs, /* number of subdevices */
const char *name)
{ /* name for the new device */
int i;
size_t size;
struct mtd_concat *concat;
uint32_t max_erasesize, curr_erasesize;
int num_erase_region;
printk(KERN_NOTICE "Concatenating MTD devices:\n");
for (i = 0; i < num_devs; i++)
printk(KERN_NOTICE "(%d): \"%s\"\n", i, subdev[i]->name);
printk(KERN_NOTICE "into device \"%s\"\n", name);
/* allocate the device structure */
size = SIZEOF_STRUCT_MTD_CONCAT(num_devs);
concat = kzalloc(size, GFP_KERNEL);
if (!concat) {
printk
("memory allocation error while creating concatenated device \"%s\"\n",
name);
return NULL;
}
concat->subdev = (struct mtd_info **) (concat + 1);
/*
* Set up the new "super" device's MTD object structure, check for
* incompatibilites between the subdevices.
*/
concat->mtd.type = subdev[0]->type;
concat->mtd.flags = subdev[0]->flags;
concat->mtd.size = subdev[0]->size;
concat->mtd.erasesize = subdev[0]->erasesize;
concat->mtd.writesize = subdev[0]->writesize;
concat->mtd.subpage_sft = subdev[0]->subpage_sft;
concat->mtd.oobsize = subdev[0]->oobsize;
concat->mtd.oobavail = subdev[0]->oobavail;
if (subdev[0]->writev)
concat->mtd.writev = concat_writev;
if (subdev[0]->read_oob)
concat->mtd.read_oob = concat_read_oob;
if (subdev[0]->write_oob)
concat->mtd.write_oob = concat_write_oob;
if (subdev[0]->block_isbad)
concat->mtd.block_isbad = concat_block_isbad;
if (subdev[0]->block_markbad)
concat->mtd.block_markbad = concat_block_markbad;
concat->mtd.ecc_stats.badblocks = subdev[0]->ecc_stats.badblocks;
concat->mtd.backing_dev_info = subdev[0]->backing_dev_info;
concat->subdev[0] = subdev[0];
for (i = 1; i < num_devs; i++) {
if (concat->mtd.type != subdev[i]->type) {
kfree(concat);
printk("Incompatible device type on \"%s\"\n",
subdev[i]->name);
return NULL;
}
if (concat->mtd.flags != subdev[i]->flags) {
/*
* Expect all flags except MTD_WRITEABLE to be
* equal on all subdevices.
*/
if ((concat->mtd.flags ^ subdev[i]->
flags) & ~MTD_WRITEABLE) {
kfree(concat);
printk("Incompatible device flags on \"%s\"\n",
subdev[i]->name);
return NULL;
} else
/* if writeable attribute differs,
make super device writeable */
concat->mtd.flags |=
subdev[i]->flags & MTD_WRITEABLE;
}
/* only permit direct mapping if the BDIs are all the same
* - copy-mapping is still permitted
*/
if (concat->mtd.backing_dev_info !=
subdev[i]->backing_dev_info)
concat->mtd.backing_dev_info =
&default_backing_dev_info;
concat->mtd.size += subdev[i]->size;
concat->mtd.ecc_stats.badblocks +=
subdev[i]->ecc_stats.badblocks;
if (concat->mtd.writesize != subdev[i]->writesize ||
concat->mtd.subpage_sft != subdev[i]->subpage_sft ||
concat->mtd.oobsize != subdev[i]->oobsize ||
!concat->mtd.read_oob != !subdev[i]->read_oob ||
!concat->mtd.write_oob != !subdev[i]->write_oob) {
kfree(concat);
printk("Incompatible OOB or ECC data on \"%s\"\n",
subdev[i]->name);
return NULL;
}
concat->subdev[i] = subdev[i];
}
concat->mtd.ecclayout = subdev[0]->ecclayout;
concat->num_subdev = num_devs;
concat->mtd.name = name;
concat->mtd.erase = concat_erase;
concat->mtd.read = concat_read;
concat->mtd.write = concat_write;
concat->mtd.sync = concat_sync;
concat->mtd.lock = concat_lock;
concat->mtd.unlock = concat_unlock;
concat->mtd.suspend = concat_suspend;
concat->mtd.resume = concat_resume;
concat->mtd.get_unmapped_area = concat_get_unmapped_area;
/*
* Combine the erase block size info of the subdevices:
*
* first, walk the map of the new device and see how
* many changes in erase size we have
*/
max_erasesize = curr_erasesize = subdev[0]->erasesize;
num_erase_region = 1;
for (i = 0; i < num_devs; i++) {
if (subdev[i]->numeraseregions == 0) {
/* current subdevice has uniform erase size */
if (subdev[i]->erasesize != curr_erasesize) {
/* if it differs from the last subdevice's erase size, count it */
++num_erase_region;
curr_erasesize = subdev[i]->erasesize;
if (curr_erasesize > max_erasesize)
max_erasesize = curr_erasesize;
}
} else {
/* current subdevice has variable erase size */
int j;
for (j = 0; j < subdev[i]->numeraseregions; j++) {
/* walk the list of erase regions, count any changes */
if (subdev[i]->eraseregions[j].erasesize !=
curr_erasesize) {
++num_erase_region;
curr_erasesize =
subdev[i]->eraseregions[j].
erasesize;
if (curr_erasesize > max_erasesize)
max_erasesize = curr_erasesize;
}
}
}
}
if (num_erase_region == 1) {
/*
* All subdevices have the same uniform erase size.
* This is easy:
*/
concat->mtd.erasesize = curr_erasesize;
concat->mtd.numeraseregions = 0;
} else {
uint64_t tmp64;
/*
* erase block size varies across the subdevices: allocate
* space to store the data describing the variable erase regions
*/
struct mtd_erase_region_info *erase_region_p;
uint64_t begin, position;
concat->mtd.erasesize = max_erasesize;
concat->mtd.numeraseregions = num_erase_region;
concat->mtd.eraseregions = erase_region_p =
kmalloc(num_erase_region *
sizeof (struct mtd_erase_region_info), GFP_KERNEL);
if (!erase_region_p) {
kfree(concat);
printk
("memory allocation error while creating erase region list"
" for device \"%s\"\n", name);
return NULL;
}
/*
* walk the map of the new device once more and fill in
* in erase region info:
*/
curr_erasesize = subdev[0]->erasesize;
begin = position = 0;
for (i = 0; i < num_devs; i++) {
if (subdev[i]->numeraseregions == 0) {
/* current subdevice has uniform erase size */
if (subdev[i]->erasesize != curr_erasesize) {
/*
* fill in an mtd_erase_region_info structure for the area
* we have walked so far:
*/
erase_region_p->offset = begin;
erase_region_p->erasesize =
curr_erasesize;
tmp64 = position - begin;
do_div(tmp64, curr_erasesize);
erase_region_p->numblocks = tmp64;
begin = position;
curr_erasesize = subdev[i]->erasesize;
++erase_region_p;
}
position += subdev[i]->size;
} else {
/* current subdevice has variable erase size */
int j;
for (j = 0; j < subdev[i]->numeraseregions; j++) {
/* walk the list of erase regions, count any changes */
if (subdev[i]->eraseregions[j].
erasesize != curr_erasesize) {
erase_region_p->offset = begin;
erase_region_p->erasesize =
curr_erasesize;
tmp64 = position - begin;
do_div(tmp64, curr_erasesize);
erase_region_p->numblocks = tmp64;
begin = position;
curr_erasesize =
subdev[i]->eraseregions[j].
erasesize;
++erase_region_p;
}
position +=
subdev[i]->eraseregions[j].
numblocks * (uint64_t)curr_erasesize;
}
}
}
/* Now write the final entry */
erase_region_p->offset = begin;
erase_region_p->erasesize = curr_erasesize;
tmp64 = position - begin;
do_div(tmp64, curr_erasesize);
erase_region_p->numblocks = tmp64;
}
return &concat->mtd;
}
/*
* This function destroys an MTD object obtained from concat_mtd_devs()
*/
void mtd_concat_destroy(struct mtd_info *mtd)
{
struct mtd_concat *concat = CONCAT(mtd);
if (concat->mtd.numeraseregions)
kfree(concat->mtd.eraseregions);
kfree(concat);
}
EXPORT_SYMBOL(mtd_concat_create);
EXPORT_SYMBOL(mtd_concat_destroy);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Robert Kaiser <rkaiser@sysgo.de>");
MODULE_DESCRIPTION("Generic support for concatenating of MTD devices");