2 * JFFS -- Journaling Flash File System, Linux implementation.
4 * Copyright (C) 1999, 2000 Axis Communications, Inc.
6 * Created by Finn Hakansson <finn@axis.com>.
8 * This is free software; you can redistribute it and/or modify it
9 * under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
13 * $Id: intrep.c,v 1.102 2001/09/23 23:28:36 dwmw2 Exp $
15 * Ported to Linux 2.3.x and MTD:
16 * Copyright (C) 2000 Alexander Larsson (alex@cendio.se), Cendio Systems AB
20 /* This file contains the code for the internal structure of the
21 Journaling Flash File System, JFFS. */
26 * memcpy_to_flash() and memcpy_from_flash() functions.
28 * Implementation of hard links.
30 * Organize the source code in a better way. Against the VFS we could
31 * have jffs_ext.c, and against the block device jffs_int.c.
32 * A better file-internal organization too.
34 * A better checksum algorithm.
36 * Consider endianness stuff. ntohl() etc.
38 * Are we handling the atime, mtime, ctime members of the inode right?
40 * Remove some duplicated code. Take a look at jffs_write_node() and
41 * jffs_rewrite_data() for instance.
43 * Implement more meaning of the nlink member in various data structures.
44 * nlink could be used in conjunction with hard links for instance.
46 * Better memory management. Allocate data structures in larger chunks
49 * If too much meta data is stored, a garbage collect should be issued.
50 * We have experienced problems with too much meta data with for instance
53 * Improve the calls to jffs_ioctl(). We would like to retrieve more
54 * information to be able to debug (or to supervise) JFFS during run-time.
58 #include <linux/types.h>
59 #include <linux/slab.h>
60 #include <linux/jffs.h>
62 #include <linux/stat.h>
63 #include <linux/pagemap.h>
64 #include <linux/mutex.h>
65 #include <asm/byteorder.h>
66 #include <linux/smp_lock.h>
67 #include <linux/time.h>
68 #include <linux/ctype.h>
73 long no_jffs_node
= 0;
74 static long no_jffs_file
= 0;
75 #if defined(JFFS_MEMORY_DEBUG) && JFFS_MEMORY_DEBUG
76 long no_jffs_control
= 0;
77 long no_jffs_raw_inode
= 0;
78 long no_jffs_node_ref
= 0;
80 long no_jffs_fmcontrol
= 0;
85 static int jffs_scan_flash(struct jffs_control
*c
);
86 static int jffs_update_file(struct jffs_file
*f
, struct jffs_node
*node
);
87 static int jffs_build_file(struct jffs_file
*f
);
88 static int jffs_free_file(struct jffs_file
*f
);
89 static int jffs_free_node_list(struct jffs_file
*f
);
90 static int jffs_garbage_collect_now(struct jffs_control
*c
);
91 static int jffs_insert_file_into_hash(struct jffs_file
*f
);
92 static int jffs_remove_redundant_nodes(struct jffs_file
*f
);
94 /* Is there enough space on the flash? */
95 static inline int JFFS_ENOUGH_SPACE(struct jffs_control
*c
, __u32 space
)
97 struct jffs_fmcontrol
*fmc
= c
->fmc
;
100 if ((fmc
->flash_size
- (fmc
->used_size
+ fmc
->dirty_size
))
101 >= fmc
->min_free_size
+ space
) {
104 if (fmc
->dirty_size
< fmc
->sector_size
)
107 if (jffs_garbage_collect_now(c
)) {
108 D1(printk("JFFS_ENOUGH_SPACE: jffs_garbage_collect_now() failed.\n"));
114 #if CONFIG_JFFS_FS_VERBOSE > 0
116 flash_read_u8(struct mtd_info
*mtd
, loff_t from
)
122 res
= MTD_READ(mtd
, from
, 1, &retlen
, &ret
);
124 printk("Didn't read a byte in flash_read_u8(). Returned %d\n", res
);
132 jffs_hexdump(struct mtd_info
*mtd
, loff_t pos
, int size
)
140 printk("%ld:", (long) pos
);
141 for (j
= 0; j
< 16; j
++) {
142 line
[j
] = flash_read_u8(mtd
, pos
++);
144 for (i
= 0; i
< j
; i
++) {
146 printk(" %.2x", line
[i
] & 0xff);
149 printk("%.2x", line
[i
] & 0xff);
153 /* Print empty space */
154 for (; i
< 16; i
++) {
164 for (i
= 0; i
< j
; i
++) {
165 if (isgraph(line
[i
])) {
166 printk("%c", line
[i
]);
177 /* Print the contents of a node. */
179 jffs_print_node(struct jffs_node
*n
)
181 D(printk("jffs_node: 0x%p\n", n
));
183 D(printk(" 0x%08x, /* version */\n", n
->version
));
184 D(printk(" 0x%08x, /* data_offset */\n", n
->data_offset
));
185 D(printk(" 0x%08x, /* data_size */\n", n
->data_size
));
186 D(printk(" 0x%08x, /* removed_size */\n", n
->removed_size
));
187 D(printk(" 0x%08x, /* fm_offset */\n", n
->fm_offset
));
188 D(printk(" 0x%02x, /* name_size */\n", n
->name_size
));
189 D(printk(" 0x%p, /* fm, fm->offset: %u */\n",
190 n
->fm
, (n
->fm
? n
->fm
->offset
: 0)));
191 D(printk(" 0x%p, /* version_prev */\n", n
->version_prev
));
192 D(printk(" 0x%p, /* version_next */\n", n
->version_next
));
193 D(printk(" 0x%p, /* range_prev */\n", n
->range_prev
));
194 D(printk(" 0x%p, /* range_next */\n", n
->range_next
));
200 /* Print the contents of a raw inode. */
202 jffs_print_raw_inode(struct jffs_raw_inode
*raw_inode
)
204 D(printk("jffs_raw_inode: inode number: %u\n", raw_inode
->ino
));
206 D(printk(" 0x%08x, /* magic */\n", raw_inode
->magic
));
207 D(printk(" 0x%08x, /* ino */\n", raw_inode
->ino
));
208 D(printk(" 0x%08x, /* pino */\n", raw_inode
->pino
));
209 D(printk(" 0x%08x, /* version */\n", raw_inode
->version
));
210 D(printk(" 0x%08x, /* mode */\n", raw_inode
->mode
));
211 D(printk(" 0x%04x, /* uid */\n", raw_inode
->uid
));
212 D(printk(" 0x%04x, /* gid */\n", raw_inode
->gid
));
213 D(printk(" 0x%08x, /* atime */\n", raw_inode
->atime
));
214 D(printk(" 0x%08x, /* mtime */\n", raw_inode
->mtime
));
215 D(printk(" 0x%08x, /* ctime */\n", raw_inode
->ctime
));
216 D(printk(" 0x%08x, /* offset */\n", raw_inode
->offset
));
217 D(printk(" 0x%08x, /* dsize */\n", raw_inode
->dsize
));
218 D(printk(" 0x%08x, /* rsize */\n", raw_inode
->rsize
));
219 D(printk(" 0x%02x, /* nsize */\n", raw_inode
->nsize
));
220 D(printk(" 0x%02x, /* nlink */\n", raw_inode
->nlink
));
221 D(printk(" 0x%02x, /* spare */\n",
223 D(printk(" %u, /* rename */\n",
225 D(printk(" %u, /* deleted */\n",
226 raw_inode
->deleted
));
227 D(printk(" 0x%02x, /* accurate */\n",
228 raw_inode
->accurate
));
229 D(printk(" 0x%08x, /* dchksum */\n", raw_inode
->dchksum
));
230 D(printk(" 0x%04x, /* nchksum */\n", raw_inode
->nchksum
));
231 D(printk(" 0x%04x, /* chksum */\n", raw_inode
->chksum
));
235 #define flash_safe_acquire(arg)
236 #define flash_safe_release(arg)
240 flash_safe_read(struct mtd_info
*mtd
, loff_t from
,
241 u_char
*buf
, size_t count
)
246 D3(printk(KERN_NOTICE
"flash_safe_read(%p, %08x, %p, %08x)\n",
247 mtd
, (unsigned int) from
, buf
, count
));
249 res
= mtd
->read(mtd
, from
, count
, &retlen
, buf
);
250 if (retlen
!= count
) {
251 panic("Didn't read all bytes in flash_safe_read(). Returned %d\n", res
);
253 return res
?res
:retlen
;
258 flash_read_u32(struct mtd_info
*mtd
, loff_t from
)
264 res
= mtd
->read(mtd
, from
, 4, &retlen
, (unsigned char *)&ret
);
266 printk("Didn't read all bytes in flash_read_u32(). Returned %d\n", res
);
275 flash_safe_write(struct mtd_info
*mtd
, loff_t to
,
276 const u_char
*buf
, size_t count
)
281 D3(printk(KERN_NOTICE
"flash_safe_write(%p, %08x, %p, %08x)\n",
282 mtd
, (unsigned int) to
, buf
, count
));
284 res
= mtd
->write(mtd
, to
, count
, &retlen
, buf
);
285 if (retlen
!= count
) {
286 printk("Didn't write all bytes in flash_safe_write(). Returned %d\n", res
);
288 return res
?res
:retlen
;
293 flash_safe_writev(struct mtd_info
*mtd
, const struct kvec
*vecs
,
294 unsigned long iovec_cnt
, loff_t to
)
296 size_t retlen
, retlen_a
;
300 D3(printk(KERN_NOTICE
"flash_safe_writev(%p, %08x, %p)\n",
301 mtd
, (unsigned int) to
, vecs
));
304 res
= mtd
->writev(mtd
, vecs
, iovec_cnt
, to
, &retlen
);
305 return res
? res
: retlen
;
307 /* Not implemented writev. Repeatedly use write - on the not so
308 unreasonable assumption that the mtd driver doesn't care how
309 many write cycles we use. */
313 for (i
=0; !res
&& i
<iovec_cnt
; i
++) {
314 res
= mtd
->write(mtd
, to
, vecs
[i
].iov_len
, &retlen_a
,
316 if (retlen_a
!= vecs
[i
].iov_len
) {
317 printk("Didn't write all bytes in flash_safe_writev(). Returned %d\n", res
);
318 if (i
!= iovec_cnt
-1)
321 /* If res is non-zero, retlen_a is undefined, but we don't
322 care because in that case it's not going to be
328 return res
?res
:retlen
;
333 flash_memset(struct mtd_info
*mtd
, loff_t to
,
334 const u_char c
, size_t size
)
336 static unsigned char pattern
[64];
339 /* fill up pattern */
341 for(i
= 0; i
< 64; i
++)
344 /* write as many 64-byte chunks as we can */
347 flash_safe_write(mtd
, to
, pattern
, 64);
355 flash_safe_write(mtd
, to
, pattern
, size
);
362 intrep_erase_callback(struct erase_info
*done
)
364 wait_queue_head_t
*wait_q
;
366 wait_q
= (wait_queue_head_t
*)done
->priv
;
373 flash_erase_region(struct mtd_info
*mtd
, loff_t start
,
376 struct erase_info
*erase
;
377 DECLARE_WAITQUEUE(wait
, current
);
378 wait_queue_head_t wait_q
;
380 erase
= kmalloc(sizeof(struct erase_info
), GFP_KERNEL
);
384 init_waitqueue_head(&wait_q
);
387 erase
->callback
= intrep_erase_callback
;
390 erase
->priv
= (u_long
)&wait_q
;
392 /* FIXME: Use TASK_INTERRUPTIBLE and deal with being interrupted */
393 set_current_state(TASK_UNINTERRUPTIBLE
);
394 add_wait_queue(&wait_q
, &wait
);
396 if (mtd
->erase(mtd
, erase
) < 0) {
397 set_current_state(TASK_RUNNING
);
398 remove_wait_queue(&wait_q
, &wait
);
401 printk(KERN_WARNING
"flash: erase of region [0x%lx, 0x%lx] "
402 "totally failed\n", (long)start
, (long)start
+ size
);
407 schedule(); /* Wait for flash to finish. */
408 remove_wait_queue(&wait_q
, &wait
);
415 /* This routine calculates checksums in JFFS. */
417 jffs_checksum(const void *data
, int size
)
420 __u8
*ptr
= (__u8
*)data
;
424 D3(printk(", result: 0x%08x\n", sum
));
430 jffs_checksum_flash(struct mtd_info
*mtd
, loff_t start
, int size
, __u32
*result
)
437 /* Allocate read buffer */
438 read_buf
= (__u8
*) kmalloc (sizeof(__u8
) * 4096, GFP_KERNEL
);
440 printk(KERN_NOTICE
"kmalloc failed in jffs_checksum_flash()\n");
443 /* Loop until checksum done */
445 /* Get amount of data to read */
451 /* Perform flash read */
452 D3(printk(KERN_NOTICE
"jffs_checksum_flash\n"));
453 flash_safe_read(mtd
, ptr
, &read_buf
[0], length
);
455 /* Compute checksum */
456 for (i
=0; i
< length
; i
++)
459 /* Update pointer and size */
464 /* Free read buffer */
468 D3(printk("checksum result: 0x%08x\n", sum
));
473 static __inline__
void jffs_fm_write_lock(struct jffs_fmcontrol
*fmc
)
475 // down(&fmc->wlock);
478 static __inline__
void jffs_fm_write_unlock(struct jffs_fmcontrol
*fmc
)
484 /* Create and initialize a new struct jffs_file. */
485 static struct jffs_file
*
486 jffs_create_file(struct jffs_control
*c
,
487 const struct jffs_raw_inode
*raw_inode
)
491 if (!(f
= kzalloc(sizeof(*f
), GFP_KERNEL
))) {
492 D(printk("jffs_create_file(): Failed!\n"));
496 f
->ino
= raw_inode
->ino
;
497 f
->pino
= raw_inode
->pino
;
498 f
->nlink
= raw_inode
->nlink
;
499 f
->deleted
= raw_inode
->deleted
;
506 /* Build a control block for the file system. */
507 static struct jffs_control
*
508 jffs_create_control(struct super_block
*sb
)
510 struct jffs_control
*c
;
511 register int s
= sizeof(struct jffs_control
);
515 D2(printk("jffs_create_control()\n"));
517 if (!(c
= kmalloc(s
, GFP_KERNEL
))) {
520 DJM(no_jffs_control
++);
523 c
->hash_len
= JFFS_HASH_SIZE
;
524 s
= sizeof(struct list_head
) * c
->hash_len
;
525 if (!(c
->hash
= kmalloc(s
, GFP_KERNEL
))) {
529 for (i
= 0; i
< c
->hash_len
; i
++)
530 INIT_LIST_HEAD(&c
->hash
[i
]);
531 if (!(c
->fmc
= jffs_build_begin(c
, MINOR(sb
->s_dev
)))) {
534 c
->next_ino
= JFFS_MIN_INO
+ 1;
535 c
->delete_list
= (struct jffs_delete_list
*) 0;
542 DJM(no_jffs_control
--);
543 D(t
= t
? t
: "c->hash");
545 D(t
= t
? t
: "control");
546 D(printk("jffs_create_control(): Allocation failed: (%s)\n", t
));
547 return (struct jffs_control
*)0;
551 /* Clean up all data structures associated with the file system. */
553 jffs_cleanup_control(struct jffs_control
*c
)
555 D2(printk("jffs_cleanup_control()\n"));
558 D(printk("jffs_cleanup_control(): c == NULL !!!\n"));
562 while (c
->delete_list
) {
563 struct jffs_delete_list
*delete_list_element
;
564 delete_list_element
= c
->delete_list
;
565 c
->delete_list
= c
->delete_list
->next
;
566 kfree(delete_list_element
);
569 /* Free all files and nodes. */
571 jffs_foreach_file(c
, jffs_free_node_list
);
572 jffs_foreach_file(c
, jffs_free_file
);
576 jffs_cleanup_fmcontrol(c
->fmc
);
578 DJM(no_jffs_control
--);
579 D3(printk("jffs_cleanup_control(): Leaving...\n"));
583 /* This function adds a virtual root node to the in-RAM representation.
584 Called by jffs_build_fs(). */
586 jffs_add_virtual_root(struct jffs_control
*c
)
588 struct jffs_file
*root
;
589 struct jffs_node
*node
;
591 D2(printk("jffs_add_virtual_root(): "
592 "Creating a virtual root directory.\n"));
594 if (!(root
= kmalloc(sizeof(struct jffs_file
), GFP_KERNEL
))) {
598 if (!(node
= jffs_alloc_node())) {
604 memset(node
, 0, sizeof(struct jffs_node
));
605 node
->ino
= JFFS_MIN_INO
;
606 memset(root
, 0, sizeof(struct jffs_file
));
607 root
->ino
= JFFS_MIN_INO
;
608 root
->mode
= S_IFDIR
| S_IRWXU
| S_IRGRP
609 | S_IXGRP
| S_IROTH
| S_IXOTH
;
610 root
->atime
= root
->mtime
= root
->ctime
= get_seconds();
613 root
->version_head
= root
->version_tail
= node
;
614 jffs_insert_file_into_hash(root
);
619 /* This is where the file system is built and initialized. */
621 jffs_build_fs(struct super_block
*sb
)
623 struct jffs_control
*c
;
626 D2(printk("jffs_build_fs()\n"));
628 if (!(c
= jffs_create_control(sb
))) {
633 if ((err
= jffs_scan_flash(c
)) < 0) {
635 /* scan_flash() wants us to try once more. A flipping
636 bits sector was detect in the middle of the scan flash.
637 Clean up old allocated memory before going in.
639 D1(printk("jffs_build_fs: Cleaning up all control structures,"
640 " reallocating them and trying mount again.\n"));
641 jffs_cleanup_control(c
);
642 if (!(c
= jffs_create_control(sb
))) {
648 if ((err
= jffs_scan_flash(c
)) < 0) {
649 goto jffs_build_fs_fail
;
652 goto jffs_build_fs_fail
;
656 /* Add a virtual root node if no one exists. */
657 if (!jffs_find_file(c
, JFFS_MIN_INO
)) {
658 if ((err
= jffs_add_virtual_root(c
)) < 0) {
659 goto jffs_build_fs_fail
;
663 while (c
->delete_list
) {
665 struct jffs_delete_list
*delete_list_element
;
667 if ((f
= jffs_find_file(c
, c
->delete_list
->ino
))) {
670 delete_list_element
= c
->delete_list
;
671 c
->delete_list
= c
->delete_list
->next
;
672 kfree(delete_list_element
);
675 /* Remove deleted nodes. */
676 if ((err
= jffs_foreach_file(c
, jffs_possibly_delete_file
)) < 0) {
677 printk(KERN_ERR
"JFFS: Failed to remove deleted nodes.\n");
678 goto jffs_build_fs_fail
;
680 /* Remove redundant nodes. (We are not interested in the
681 return value in this case.) */
682 jffs_foreach_file(c
, jffs_remove_redundant_nodes
);
683 /* Try to build a tree from all the nodes. */
684 if ((err
= jffs_foreach_file(c
, jffs_insert_file_into_tree
)) < 0) {
685 printk("JFFS: Failed to build tree.\n");
686 goto jffs_build_fs_fail
;
688 /* Compute the sizes of all files in the filesystem. Adjust if
690 if ((err
= jffs_foreach_file(c
, jffs_build_file
)) < 0) {
691 printk("JFFS: Failed to build file system.\n");
692 goto jffs_build_fs_fail
;
694 sb
->s_fs_info
= (void *)c
;
697 D1(jffs_print_hash_table(c
));
698 D1(jffs_print_tree(c
->root
, 0));
703 jffs_cleanup_control(c
);
705 } /* jffs_build_fs() */
709 This checks for sectors that were being erased in their previous
710 lifetimes and for some reason or the other (power fail etc.),
711 the erase cycles never completed.
712 As the flash array would have reverted back to read status,
713 these sectors are detected by the symptom of the "flipping bits",
714 i.e. bits being read back differently from the same location in
715 flash if read multiple times.
716 The only solution to this is to re-erase the entire
718 Unfortunately detecting "flipping bits" is not a simple exercise
719 as a bit may be read back at 1 or 0 depending on the alignment
720 of the stars in the universe.
721 The level of confidence is in direct proportion to the number of
722 scans done. By power fail testing I (Vipin) have been able to
723 proove that reading twice is not enough.
724 Maybe 4 times? Change NUM_REREADS to a higher number if you want
725 a (even) higher degree of confidence in your mount process.
726 A higher number would of course slow down your mount.
728 static int check_partly_erased_sectors(struct jffs_fmcontrol
*fmc
){
730 #define NUM_REREADS 4 /* see note above */
731 #define READ_AHEAD_BYTES 4096 /* must be a multiple of 4,
732 usually set to kernel page size */
743 loff_t end
= fmc
->flash_size
;
746 /* Allocate read buffers */
747 read_buf1
= (__u8
*) kmalloc (sizeof(__u8
) * READ_AHEAD_BYTES
, GFP_KERNEL
);
751 read_buf2
= (__u8
*) kmalloc (sizeof(__u8
) * READ_AHEAD_BYTES
, GFP_KERNEL
);
760 D1(printk("check_partly_erased_sector():checking sector which contains"
761 " offset 0x%x for flipping bits..\n", (__u32
)pos
));
763 retlen
= flash_safe_read(fmc
->mtd
, pos
,
764 &read_buf1
[0], READ_AHEAD_BYTES
);
767 for(cnt
= 0; cnt
< NUM_REREADS
; cnt
++){
768 (void)flash_safe_read(fmc
->mtd
, pos
,
769 &read_buf2
[0], READ_AHEAD_BYTES
);
771 for (i
=0 ; i
< retlen
; i
+=4) {
772 /* buffers MUST match, double word for word! */
773 if(*((__u32
*) &read_buf1
[i
]) !=
774 *((__u32
*) &read_buf2
[i
])
776 /* flipping bits detected, time to erase sector */
777 /* This will help us log some statistics etc. */
778 D1(printk("Flipping bits detected in re-read round:%i of %i\n",
780 D1(printk("check_partly_erased_sectors:flipping bits detected"
781 " @offset:0x%x(0x%x!=0x%x)\n",
782 (__u32
)pos
+i
, *((__u32
*) &read_buf1
[i
]),
783 *((__u32
*) &read_buf2
[i
])));
785 /* calculate start of present sector */
786 offset
= (((__u32
)pos
+i
)/(__u32
)fmc
->sector_size
) * (__u32
)fmc
->sector_size
;
788 D1(printk("check_partly_erased_sector():erasing sector starting 0x%x.\n",
791 if (flash_erase_region(fmc
->mtd
,
792 offset
, fmc
->sector_size
) < 0) {
793 printk(KERN_ERR
"JFFS: Erase of flash failed. "
794 "offset = %u, erase_size = %d\n",
795 offset
, fmc
->sector_size
);
801 D1(printk("JFFS: Erase of flash sector @0x%x successful.\n",
803 /* skip ahead to the next sector */
804 pos
= (((__u32
)pos
+i
)/(__u32
)fmc
->sector_size
) * (__u32
)fmc
->sector_size
;
805 pos
+= fmc
->sector_size
;
811 pos
+= READ_AHEAD_BYTES
;
818 D2(printk("check_partly_erased_sector():Done checking all sectors till offset 0x%x for flipping bits.\n",
823 }/* end check_partly_erased_sectors() */
827 /* Scan the whole flash memory in order to find all nodes in the
830 jffs_scan_flash(struct jffs_control
*c
)
832 char name
[JFFS_MAX_NAME_LEN
+ 2];
833 struct jffs_raw_inode raw_inode
;
834 struct jffs_node
*node
= NULL
;
835 struct jffs_fmcontrol
*fmc
= c
->fmc
;
843 loff_t end
= fmc
->flash_size
;
848 __u32 free_chunk_size1
;
849 __u32 free_chunk_size2
;
852 #define NUMFREEALLOWED 2 /* 2 chunks of at least erase size space allowed */
853 int num_free_space
= 0; /* Flag err if more than TWO
854 free blocks found. This is NOT allowed
855 by the current jffs design.
857 int num_free_spc_not_accp
= 0; /* For debugging purposed keep count
858 of how much free space was rejected and
862 D1(printk("jffs_scan_flash(): start pos = 0x%lx, end = 0x%lx\n",
863 (long)pos
, (long)end
));
865 flash_safe_acquire(fmc
->mtd
);
868 check and make sure that any sector does not suffer
869 from the "partly erased, bit flipping syndrome" (TM Vipin :)
870 If so, offending sectors will be erased.
872 if(check_partly_erased_sectors(fmc
) < 0){
874 flash_safe_release(fmc
->mtd
);
875 return -EIO
; /* bad, bad, bad error. Cannot continue.*/
878 /* Allocate read buffer */
879 read_buf
= (__u8
*) kmalloc (sizeof(__u8
) * 4096, GFP_KERNEL
);
881 flash_safe_release(fmc
->mtd
);
885 /* Start the scan. */
889 /* Remember the position from where we started this scan. */
892 switch (flash_read_u32(fmc
->mtd
, pos
)) {
893 case JFFS_EMPTY_BITMASK
:
894 /* We have found 0xffffffff at this position. We have to
895 scan the rest of the flash till the end or till
896 something else than 0xffffffff is found.
897 Keep going till we do not find JFFS_EMPTY_BITMASK
900 D1(printk("jffs_scan_flash(): 0xffffffff at pos 0x%lx.\n",
905 len
= end
- pos
< 4096 ? end
- pos
: 4096;
907 retlen
= flash_safe_read(fmc
->mtd
, pos
,
912 for (i
=0 ; i
< retlen
; i
+=4, pos
+= 4) {
913 if(*((__u32
*) &read_buf
[i
]) !=
923 D1(printk("jffs_scan_flash():0xffffffff ended at pos 0x%lx.\n",
926 /* If some free space ends in the middle of a sector,
927 treat it as dirty rather than clean.
928 This is to handle the case where one thread
929 allocated space for a node, but didn't get to
930 actually _write_ it before power was lost, leaving
931 a gap in the log. Shifting all node writes into
932 a single kernel thread will fix the original problem.
934 if ((__u32
) pos
% fmc
->sector_size
) {
935 /* If there was free space in previous
936 sectors, don't mark that dirty too -
937 only from the beginning of this sector
941 test_start
= pos
& ~(fmc
->sector_size
-1); /* end of last sector */
943 if (start
< test_start
) {
945 /* free space started in the previous sector! */
947 if((num_free_space
< NUMFREEALLOWED
) &&
948 ((unsigned int)(test_start
- start
) >= fmc
->sector_size
)){
951 Count it in if we are still under NUMFREEALLOWED *and* it is
952 at least 1 erase sector in length. This will keep us from
953 picking any little ole' space as "free".
956 D1(printk("Reducing end of free space to 0x%x from 0x%x\n",
957 (unsigned int)test_start
, (unsigned int)pos
));
959 D1(printk("Free space accepted: Starting 0x%x for 0x%x bytes\n",
960 (unsigned int) start
,
961 (unsigned int)(test_start
- start
)));
963 /* below, space from "start" to "pos" will be marked dirty. */
966 /* Being in here means that we have found at least an entire
967 erase sector size of free space ending on a sector boundary.
968 Keep track of free spaces accepted.
972 num_free_spc_not_accp
++;
973 D1(printk("Free space (#%i) found but *Not* accepted: Starting"
974 " 0x%x for 0x%x bytes\n",
975 num_free_spc_not_accp
, (unsigned int)start
,
976 (unsigned int)((unsigned int)(pos
& ~(fmc
->sector_size
-1)) - (unsigned int)start
)));
981 if((((__u32
)(pos
- start
)) != 0)){
983 D1(printk("Dirty space: Starting 0x%x for 0x%x bytes\n",
984 (unsigned int) start
, (unsigned int) (pos
- start
)));
985 jffs_fmalloced(fmc
, (__u32
) start
,
986 (__u32
) (pos
- start
), NULL
);
988 /* "Flipping bits" detected. This means that our scan for them
989 did not catch this offset. See check_partly_erased_sectors() for
993 D1(printk("jffs_scan_flash():wants to allocate dirty flash "
994 "space for 0 bytes.\n"));
995 D1(printk("jffs_scan_flash(): Flipping bits! We will free "
996 "all allocated memory, erase this sector and remount\n"));
998 /* calculate start of present sector */
999 offset
= (((__u32
)pos
)/(__u32
)fmc
->sector_size
) * (__u32
)fmc
->sector_size
;
1001 D1(printk("jffs_scan_flash():erasing sector starting 0x%x.\n",
1004 if (flash_erase_region(fmc
->mtd
,
1005 offset
, fmc
->sector_size
) < 0) {
1006 printk(KERN_ERR
"JFFS: Erase of flash failed. "
1007 "offset = %u, erase_size = %d\n",
1008 offset
, fmc
->sector_size
);
1010 flash_safe_release(fmc
->mtd
);
1012 return -1; /* bad, bad, bad! */
1015 flash_safe_release(fmc
->mtd
);
1018 return -EAGAIN
; /* erased offending sector. Try mount one more time please. */
1021 /* Being in here means that we have found free space that ends on an erase sector
1023 Count it in if we are still under NUMFREEALLOWED *and* it is at least 1 erase
1024 sector in length. This will keep us from picking any little ole' space as "free".
1026 if((num_free_space
< NUMFREEALLOWED
) &&
1027 ((unsigned int)(pos
- start
) >= fmc
->sector_size
)){
1028 /* We really don't do anything to mark space as free, except *not*
1029 mark it dirty and just advance the "pos" location pointer.
1030 It will automatically be picked up as free space.
1033 D1(printk("Free space accepted: Starting 0x%x for 0x%x bytes\n",
1034 (unsigned int) start
, (unsigned int) (pos
- start
)));
1036 num_free_spc_not_accp
++;
1037 D1(printk("Free space (#%i) found but *Not* accepted: Starting "
1038 "0x%x for 0x%x bytes\n", num_free_spc_not_accp
,
1039 (unsigned int) start
,
1040 (unsigned int) (pos
- start
)));
1042 /* Mark this space as dirty. We already have our free space. */
1043 D1(printk("Dirty space: Starting 0x%x for 0x%x bytes\n",
1044 (unsigned int) start
, (unsigned int) (pos
- start
)));
1045 jffs_fmalloced(fmc
, (__u32
) start
,
1046 (__u32
) (pos
- start
), NULL
);
1050 if(num_free_space
> NUMFREEALLOWED
){
1051 printk(KERN_WARNING
"jffs_scan_flash(): Found free space "
1052 "number %i. Only %i free space is allowed.\n",
1053 num_free_space
, NUMFREEALLOWED
);
1057 case JFFS_DIRTY_BITMASK
:
1058 /* We have found 0x00000000 at this position. Scan as far
1059 as possible to find out how much is dirty. */
1060 D1(printk("jffs_scan_flash(): 0x00000000 at pos 0x%lx.\n",
1063 && JFFS_DIRTY_BITMASK
== flash_read_u32(fmc
->mtd
, pos
);
1065 D1(printk("jffs_scan_flash(): 0x00 ended at "
1066 "pos 0x%lx.\n", (long)pos
));
1067 jffs_fmalloced(fmc
, (__u32
) start
,
1068 (__u32
) (pos
- start
), NULL
);
1071 case JFFS_MAGIC_BITMASK
:
1072 /* We have probably found a new raw inode. */
1077 /* We're f*cked. This is not solved yet. We have
1078 to scan for the magic pattern. */
1079 D1(printk("*************** Dirty flash memory or "
1081 "hexdump(pos = 0x%lx, len = 128):\n",
1083 D1(jffs_hexdump(fmc
->mtd
, pos
, 128));
1085 for (pos
+= 4; pos
< end
; pos
+= 4) {
1086 switch (flash_read_u32(fmc
->mtd
, pos
)) {
1087 case JFFS_MAGIC_BITMASK
:
1088 case JFFS_EMPTY_BITMASK
:
1089 /* handle these in the main switch() loop */
1098 /* First, mark as dirty the region
1099 which really does contain crap. */
1100 jffs_fmalloced(fmc
, (__u32
) start
,
1101 (__u32
) (pos
- start
),
1107 /* We have found the beginning of an inode. Create a
1108 node for it unless there already is one available. */
1110 if (!(node
= jffs_alloc_node())) {
1111 /* Free read buffer */
1114 /* Release the flash device */
1115 flash_safe_release(fmc
->mtd
);
1119 DJM(no_jffs_node
++);
1122 /* Read the next raw inode. */
1124 flash_safe_read(fmc
->mtd
, pos
, (u_char
*) &raw_inode
,
1125 sizeof(struct jffs_raw_inode
));
1127 /* When we compute the checksum for the inode, we never
1128 count the 'accurate' or the 'checksum' fields. */
1129 tmp_accurate
= raw_inode
.accurate
;
1130 tmp_chksum
= raw_inode
.chksum
;
1131 raw_inode
.accurate
= 0;
1132 raw_inode
.chksum
= 0;
1133 checksum
= jffs_checksum(&raw_inode
,
1134 sizeof(struct jffs_raw_inode
));
1135 raw_inode
.accurate
= tmp_accurate
;
1136 raw_inode
.chksum
= tmp_chksum
;
1138 D3(printk("*** We have found this raw inode at pos 0x%lx "
1139 "on the flash:\n", (long)pos
));
1140 D3(jffs_print_raw_inode(&raw_inode
));
1142 if (checksum
!= raw_inode
.chksum
) {
1143 D1(printk("jffs_scan_flash(): Bad checksum: "
1145 "raw_inode.chksum = %u\n",
1146 checksum
, raw_inode
.chksum
));
1147 pos
+= sizeof(struct jffs_raw_inode
);
1148 jffs_fmalloced(fmc
, (__u32
) start
,
1149 (__u32
) (pos
- start
), NULL
);
1150 /* Reuse this unused struct jffs_node. */
1154 /* Check the raw inode read so far. Start with the
1155 maximum length of the filename. */
1156 if (raw_inode
.nsize
> JFFS_MAX_NAME_LEN
) {
1157 printk(KERN_WARNING
"jffs_scan_flash: Found a "
1158 "JFFS node with name too large\n");
1162 if (raw_inode
.rename
&& raw_inode
.dsize
!= sizeof(__u32
)) {
1163 printk(KERN_WARNING
"jffs_scan_flash: Found a "
1164 "rename node with dsize %u.\n",
1166 jffs_print_raw_inode(&raw_inode
);
1170 /* The node's data segment should not exceed a
1172 if (raw_inode
.dsize
> fmc
->max_chunk_size
) {
1173 printk(KERN_WARNING
"jffs_scan_flash: Found a "
1174 "JFFS node with dsize (0x%x) > max_chunk_size (0x%x)\n",
1175 raw_inode
.dsize
, fmc
->max_chunk_size
);
1179 pos
+= sizeof(struct jffs_raw_inode
);
1181 /* This shouldn't be necessary because a node that
1182 violates the flash boundaries shouldn't be written
1183 in the first place. */
1188 /* Read the name. */
1190 if (raw_inode
.nsize
) {
1191 flash_safe_read(fmc
->mtd
, pos
, name
, raw_inode
.nsize
);
1192 name
[raw_inode
.nsize
] = '\0';
1193 pos
+= raw_inode
.nsize
1194 + JFFS_GET_PAD_BYTES(raw_inode
.nsize
);
1195 D3(printk("name == \"%s\"\n", name
));
1196 checksum
= jffs_checksum(name
, raw_inode
.nsize
);
1197 if (checksum
!= raw_inode
.nchksum
) {
1198 D1(printk("jffs_scan_flash(): Bad checksum: "
1200 "raw_inode.nchksum = %u\n",
1201 checksum
, raw_inode
.nchksum
));
1202 jffs_fmalloced(fmc
, (__u32
) start
,
1203 (__u32
) (pos
- start
), NULL
);
1204 /* Reuse this unused struct jffs_node. */
1212 /* Read the data, if it exists, in order to be sure it
1213 matches the checksum. */
1214 if (raw_inode
.dsize
) {
1215 if (raw_inode
.rename
) {
1216 deleted_file
= flash_read_u32(fmc
->mtd
, pos
);
1218 if (jffs_checksum_flash(fmc
->mtd
, pos
, raw_inode
.dsize
, &checksum
)) {
1219 printk("jffs_checksum_flash() failed to calculate a checksum\n");
1220 jffs_fmalloced(fmc
, (__u32
) start
,
1221 (__u32
) (pos
- start
), NULL
);
1222 /* Reuse this unused struct jffs_node. */
1225 pos
+= raw_inode
.dsize
1226 + JFFS_GET_PAD_BYTES(raw_inode
.dsize
);
1228 if (checksum
!= raw_inode
.dchksum
) {
1229 D1(printk("jffs_scan_flash(): Bad checksum: "
1231 "raw_inode.dchksum = %u\n",
1232 checksum
, raw_inode
.dchksum
));
1233 jffs_fmalloced(fmc
, (__u32
) start
,
1234 (__u32
) (pos
- start
), NULL
);
1235 /* Reuse this unused struct jffs_node. */
1242 /* Remember the highest inode number in the whole file
1243 system. This information will be used when assigning
1244 new files new inode numbers. */
1245 if (c
->next_ino
<= raw_inode
.ino
) {
1246 c
->next_ino
= raw_inode
.ino
+ 1;
1249 if (raw_inode
.accurate
) {
1251 node
->data_offset
= raw_inode
.offset
;
1252 node
->data_size
= raw_inode
.dsize
;
1253 node
->removed_size
= raw_inode
.rsize
;
1254 /* Compute the offset to the actual data in the
1257 = sizeof(struct jffs_raw_inode
)
1259 + JFFS_GET_PAD_BYTES(raw_inode
.nsize
);
1260 node
->fm
= jffs_fmalloced(fmc
, (__u32
) start
,
1261 (__u32
) (pos
- start
),
1264 D(printk("jffs_scan_flash(): !node->fm\n"));
1265 jffs_free_node(node
);
1266 DJM(no_jffs_node
--);
1268 /* Free read buffer */
1271 /* Release the flash device */
1272 flash_safe_release(fmc
->mtd
);
1276 if ((err
= jffs_insert_node(c
, NULL
, &raw_inode
,
1278 printk("JFFS: Failed to handle raw inode. "
1279 "(err = %d)\n", err
);
1282 if (raw_inode
.rename
) {
1283 struct jffs_delete_list
*dl
1284 = (struct jffs_delete_list
*)
1285 kmalloc(sizeof(struct jffs_delete_list
),
1288 D(printk("jffs_scan_flash: !dl\n"));
1289 jffs_free_node(node
);
1290 DJM(no_jffs_node
--);
1292 /* Release the flash device */
1293 flash_safe_release(fmc
->flash_part
);
1295 /* Free read buffer */
1300 dl
->ino
= deleted_file
;
1301 dl
->next
= c
->delete_list
;
1302 c
->delete_list
= dl
;
1303 node
->data_size
= 0;
1305 D3(jffs_print_node(node
));
1306 node
= NULL
; /* Don't free the node! */
1309 jffs_fmalloced(fmc
, (__u32
) start
,
1310 (__u32
) (pos
- start
), NULL
);
1311 D3(printk("jffs_scan_flash(): Just found an obsolete "
1312 "raw_inode. Continuing the scan...\n"));
1313 /* Reuse this unused struct jffs_node. */
1318 jffs_free_node(node
);
1319 DJM(no_jffs_node
--);
1321 jffs_build_end(fmc
);
1323 /* Free read buffer */
1326 if(!num_free_space
){
1327 printk(KERN_WARNING
"jffs_scan_flash(): Did not find even a single "
1328 "chunk of free space. This is BAD!\n");
1332 D3(printk("jffs_scan_flash(): Leaving...\n"));
1333 flash_safe_release(fmc
->mtd
);
1335 /* This is to trap the "free size accounting screwed error. */
1336 free_chunk_size1
= jffs_free_size1(fmc
);
1337 free_chunk_size2
= jffs_free_size2(fmc
);
1339 if (free_chunk_size1
+ free_chunk_size2
!= fmc
->free_size
) {
1341 printk(KERN_WARNING
"jffs_scan_falsh():Free size accounting screwed\n");
1342 printk(KERN_WARNING
"jfffs_scan_flash():free_chunk_size1 == 0x%x, "
1343 "free_chunk_size2 == 0x%x, fmc->free_size == 0x%x\n",
1344 free_chunk_size1
, free_chunk_size2
, fmc
->free_size
);
1346 return -1; /* Do NOT mount f/s so that we can inspect what happened.
1347 Mounting this screwed up f/s will screw us up anyway.
1351 return 0; /* as far as we are concerned, we are happy! */
1352 } /* jffs_scan_flash() */
1355 /* Insert any kind of node into the file system. Take care of data
1356 insertions and deletions. Also remove redundant information. The
1357 memory allocated for the `name' is regarded as "given away" in the
1358 caller's perspective. */
1360 jffs_insert_node(struct jffs_control
*c
, struct jffs_file
*f
,
1361 const struct jffs_raw_inode
*raw_inode
,
1362 const char *name
, struct jffs_node
*node
)
1364 int update_name
= 0;
1365 int insert_into_tree
= 0;
1367 D2(printk("jffs_insert_node(): ino = %u, version = %u, "
1368 "name = \"%s\", deleted = %d\n",
1369 raw_inode
->ino
, raw_inode
->version
,
1370 ((name
&& *name
) ? name
: ""), raw_inode
->deleted
));
1372 /* If there doesn't exist an associated jffs_file, then
1373 create, initialize and insert one into the file system. */
1374 if (!f
&& !(f
= jffs_find_file(c
, raw_inode
->ino
))) {
1375 if (!(f
= jffs_create_file(c
, raw_inode
))) {
1378 jffs_insert_file_into_hash(f
);
1379 insert_into_tree
= 1;
1381 node
->ino
= raw_inode
->ino
;
1382 node
->version
= raw_inode
->version
;
1383 node
->data_size
= raw_inode
->dsize
;
1384 node
->fm_offset
= sizeof(struct jffs_raw_inode
) + raw_inode
->nsize
1385 + JFFS_GET_PAD_BYTES(raw_inode
->nsize
);
1386 node
->name_size
= raw_inode
->nsize
;
1388 /* Now insert the node at the correct position into the file's
1390 if (!f
->version_head
) {
1391 /* This is the first node. */
1392 f
->version_head
= node
;
1393 f
->version_tail
= node
;
1394 node
->version_prev
= NULL
;
1395 node
->version_next
= NULL
;
1396 f
->highest_version
= node
->version
;
1398 f
->mode
= raw_inode
->mode
;
1399 f
->uid
= raw_inode
->uid
;
1400 f
->gid
= raw_inode
->gid
;
1401 f
->atime
= raw_inode
->atime
;
1402 f
->mtime
= raw_inode
->mtime
;
1403 f
->ctime
= raw_inode
->ctime
;
1405 else if ((f
->highest_version
< node
->version
)
1406 || (node
->version
== 0)) {
1407 /* Insert at the end of the list. I.e. this node is the
1408 newest one so far. */
1409 node
->version_prev
= f
->version_tail
;
1410 node
->version_next
= NULL
;
1411 f
->version_tail
->version_next
= node
;
1412 f
->version_tail
= node
;
1413 f
->highest_version
= node
->version
;
1415 f
->pino
= raw_inode
->pino
;
1416 f
->mode
= raw_inode
->mode
;
1417 f
->uid
= raw_inode
->uid
;
1418 f
->gid
= raw_inode
->gid
;
1419 f
->atime
= raw_inode
->atime
;
1420 f
->mtime
= raw_inode
->mtime
;
1421 f
->ctime
= raw_inode
->ctime
;
1423 else if (f
->version_head
->version
> node
->version
) {
1424 /* Insert at the bottom of the list. */
1425 node
->version_prev
= NULL
;
1426 node
->version_next
= f
->version_head
;
1427 f
->version_head
->version_prev
= node
;
1428 f
->version_head
= node
;
1434 struct jffs_node
*n
;
1436 /* Search for the insertion position starting from
1437 the tail (newest node). */
1438 for (n
= f
->version_tail
; n
; n
= n
->version_prev
) {
1439 if (n
->version
< node
->version
) {
1440 node
->version_prev
= n
;
1441 node
->version_next
= n
->version_next
;
1442 node
->version_next
->version_prev
= node
;
1443 n
->version_next
= node
;
1455 /* Deletion is irreversible. If any 'deleted' node is ever
1456 written, the file is deleted */
1457 if (raw_inode
->deleted
)
1458 f
->deleted
= raw_inode
->deleted
;
1460 /* Perhaps update the name. */
1461 if (raw_inode
->nsize
&& update_name
&& name
&& *name
&& (name
!= f
->name
)) {
1466 if (!(f
->name
= (char *) kmalloc(raw_inode
->nsize
+ 1,
1471 memcpy(f
->name
, name
, raw_inode
->nsize
);
1472 f
->name
[raw_inode
->nsize
] = '\0';
1473 f
->nsize
= raw_inode
->nsize
;
1474 D3(printk("jffs_insert_node(): Updated the name of "
1475 "the file to \"%s\".\n", name
));
1478 if (!c
->building_fs
) {
1479 D3(printk("jffs_insert_node(): ---------------------------"
1480 "------------------------------------------- 1\n"));
1481 if (insert_into_tree
) {
1482 jffs_insert_file_into_tree(f
);
1484 /* Once upon a time, we would call jffs_possibly_delete_file()
1485 here. That causes an oops if someone's still got the file
1486 open, so now we only do it in jffs_delete_inode()
1489 if (node
->data_size
|| node
->removed_size
) {
1490 jffs_update_file(f
, node
);
1492 jffs_remove_redundant_nodes(f
);
1494 jffs_garbage_collect_trigger(c
);
1496 D3(printk("jffs_insert_node(): ---------------------------"
1497 "------------------------------------------- 2\n"));
1501 } /* jffs_insert_node() */
1504 /* Unlink a jffs_node from the version list it is in. */
1506 jffs_unlink_node_from_version_list(struct jffs_file
*f
,
1507 struct jffs_node
*node
)
1509 if (node
->version_prev
) {
1510 node
->version_prev
->version_next
= node
->version_next
;
1512 f
->version_head
= node
->version_next
;
1514 if (node
->version_next
) {
1515 node
->version_next
->version_prev
= node
->version_prev
;
1517 f
->version_tail
= node
->version_prev
;
1522 /* Unlink a jffs_node from the range list it is in. */
1524 jffs_unlink_node_from_range_list(struct jffs_file
*f
, struct jffs_node
*node
)
1526 if (node
->range_prev
) {
1527 node
->range_prev
->range_next
= node
->range_next
;
1530 f
->range_head
= node
->range_next
;
1532 if (node
->range_next
) {
1533 node
->range_next
->range_prev
= node
->range_prev
;
1536 f
->range_tail
= node
->range_prev
;
1541 /* Function used by jffs_remove_redundant_nodes() below. This function
1542 classifies what kind of information a node adds to a file. */
1544 jffs_classify_node(struct jffs_node
*node
)
1546 __u8 mod_type
= JFFS_MODIFY_INODE
;
1548 if (node
->name_size
) {
1549 mod_type
|= JFFS_MODIFY_NAME
;
1551 if (node
->data_size
|| node
->removed_size
) {
1552 mod_type
|= JFFS_MODIFY_DATA
;
1558 /* Remove redundant nodes from a file. Mark the on-flash memory
1561 jffs_remove_redundant_nodes(struct jffs_file
*f
)
1563 struct jffs_node
*newest_node
;
1564 struct jffs_node
*cur
;
1565 struct jffs_node
*prev
;
1568 __u8 node_with_name_later
= 0;
1570 if (!(newest_node
= f
->version_tail
)) {
1574 /* What does the `newest_node' modify? */
1575 newest_type
= jffs_classify_node(newest_node
);
1576 node_with_name_later
= newest_type
& JFFS_MODIFY_NAME
;
1578 D3(printk("jffs_remove_redundant_nodes(): ino: %u, name: \"%s\", "
1579 "newest_type: %u\n", f
->ino
, (f
->name
? f
->name
: ""),
1582 /* Traverse the file's nodes and determine which of them that are
1583 superfluous. Yeah, this might look very complex at first
1584 glance but it is actually very simple. */
1585 for (cur
= newest_node
->version_prev
; cur
; cur
= prev
) {
1586 prev
= cur
->version_prev
;
1587 mod_type
= jffs_classify_node(cur
);
1588 if ((mod_type
<= JFFS_MODIFY_INODE
)
1589 || ((newest_type
& JFFS_MODIFY_NAME
)
1591 <= (JFFS_MODIFY_INODE
+ JFFS_MODIFY_NAME
)))
1592 || (cur
->data_size
== 0 && cur
->removed_size
1593 && !cur
->version_prev
&& node_with_name_later
)) {
1594 /* Yes, this node is redundant. Remove it. */
1595 D2(printk("jffs_remove_redundant_nodes(): "
1596 "Removing node: ino: %u, version: %u, "
1597 "mod_type: %u\n", cur
->ino
, cur
->version
,
1599 jffs_unlink_node_from_version_list(f
, cur
);
1600 jffs_fmfree(f
->c
->fmc
, cur
->fm
, cur
);
1601 jffs_free_node(cur
);
1602 DJM(no_jffs_node
--);
1605 node_with_name_later
|= (mod_type
& JFFS_MODIFY_NAME
);
1613 /* Insert a file into the hash table. */
1615 jffs_insert_file_into_hash(struct jffs_file
*f
)
1617 int i
= f
->ino
% f
->c
->hash_len
;
1619 D3(printk("jffs_insert_file_into_hash(): f->ino: %u\n", f
->ino
));
1621 list_add(&f
->hash
, &f
->c
->hash
[i
]);
1626 /* Insert a file into the file system tree. */
1628 jffs_insert_file_into_tree(struct jffs_file
*f
)
1630 struct jffs_file
*parent
;
1632 D3(printk("jffs_insert_file_into_tree(): name: \"%s\"\n",
1633 (f
->name
? f
->name
: "")));
1635 if (!(parent
= jffs_find_file(f
->c
, f
->pino
))) {
1639 f
->sibling_prev
= NULL
;
1640 f
->sibling_next
= NULL
;
1644 D1(printk("jffs_insert_file_into_tree(): Found "
1645 "inode with no parent and pino == %u\n",
1651 f
->sibling_next
= parent
->children
;
1652 if (f
->sibling_next
) {
1653 f
->sibling_next
->sibling_prev
= f
;
1655 f
->sibling_prev
= NULL
;
1656 parent
->children
= f
;
1661 /* Remove a file from the hash table. */
1663 jffs_unlink_file_from_hash(struct jffs_file
*f
)
1665 D3(printk("jffs_unlink_file_from_hash(): f: 0x%p, "
1666 "ino %u\n", f
, f
->ino
));
1673 /* Just remove the file from the parent's children. Don't free
1676 jffs_unlink_file_from_tree(struct jffs_file
*f
)
1678 D3(printk("jffs_unlink_file_from_tree(): ino: %d, pino: %d, name: "
1679 "\"%s\"\n", f
->ino
, f
->pino
, (f
->name
? f
->name
: "")));
1681 if (f
->sibling_prev
) {
1682 f
->sibling_prev
->sibling_next
= f
->sibling_next
;
1684 else if (f
->parent
) {
1685 D3(printk("f->parent=%p\n", f
->parent
));
1686 f
->parent
->children
= f
->sibling_next
;
1688 if (f
->sibling_next
) {
1689 f
->sibling_next
->sibling_prev
= f
->sibling_prev
;
1695 /* Find a file with its inode number. */
1697 jffs_find_file(struct jffs_control
*c
, __u32 ino
)
1699 struct jffs_file
*f
;
1700 int i
= ino
% c
->hash_len
;
1702 D3(printk("jffs_find_file(): ino: %u\n", ino
));
1704 list_for_each_entry(f
, &c
->hash
[i
], hash
) {
1707 D3(printk("jffs_find_file(): Found file with ino "
1708 "%u. (name: \"%s\")\n",
1709 ino
, (f
->name
? f
->name
: ""));
1713 D3(printk("jffs_find_file(): Didn't find file "
1714 "with ino %u.\n", ino
);
1720 /* Find a file in a directory. We are comparing the names. */
1722 jffs_find_child(struct jffs_file
*dir
, const char *name
, int len
)
1724 struct jffs_file
*f
;
1726 D3(printk("jffs_find_child()\n"));
1728 for (f
= dir
->children
; f
; f
= f
->sibling_next
) {
1729 if (!f
->deleted
&& f
->name
1730 && !strncmp(f
->name
, name
, len
)
1731 && f
->name
[len
] == '\0') {
1737 printk("jffs_find_child(): Found \"%s\".\n", f
->name
);
1740 char *copy
= (char *) kmalloc(len
+ 1, GFP_KERNEL
);
1742 memcpy(copy
, name
, len
);
1745 printk("jffs_find_child(): Didn't find the file \"%s\".\n",
1746 (copy
? copy
: ""));
1754 /* Write a raw inode that takes up a certain amount of space in the flash
1755 memory. At the end of the flash device, there is often space that is
1756 impossible to use. At these times we want to mark this space as not
1757 used. In the cases when the amount of space is greater or equal than
1758 a struct jffs_raw_inode, we write a "dummy node" that takes up this
1759 space. The space after the raw inode, if it exists, is left as it is.
1760 Since this space after the raw inode contains JFFS_EMPTY_BITMASK bytes,
1761 we can compute the checksum of it; we don't have to manipulate it any
1764 If the space left on the device is less than the size of a struct
1765 jffs_raw_inode, this space is filled with JFFS_DIRTY_BITMASK bytes.
1766 No raw inode is written this time. */
1768 jffs_write_dummy_node(struct jffs_control
*c
, struct jffs_fm
*dirty_fm
)
1770 struct jffs_fmcontrol
*fmc
= c
->fmc
;
1773 D1(printk("jffs_write_dummy_node(): dirty_fm->offset = 0x%08x, "
1774 "dirty_fm->size = %u\n",
1775 dirty_fm
->offset
, dirty_fm
->size
));
1777 if (dirty_fm
->size
>= sizeof(struct jffs_raw_inode
)) {
1778 struct jffs_raw_inode raw_inode
;
1779 memset(&raw_inode
, 0, sizeof(struct jffs_raw_inode
));
1780 raw_inode
.magic
= JFFS_MAGIC_BITMASK
;
1781 raw_inode
.dsize
= dirty_fm
->size
1782 - sizeof(struct jffs_raw_inode
);
1783 raw_inode
.dchksum
= raw_inode
.dsize
* 0xff;
1785 = jffs_checksum(&raw_inode
, sizeof(struct jffs_raw_inode
));
1787 if ((err
= flash_safe_write(fmc
->mtd
,
1789 (u_char
*)&raw_inode
,
1790 sizeof(struct jffs_raw_inode
)))
1792 printk(KERN_ERR
"JFFS: jffs_write_dummy_node: "
1793 "flash_safe_write failed!\n");
1798 flash_safe_acquire(fmc
->mtd
);
1799 flash_memset(fmc
->mtd
, dirty_fm
->offset
, 0, dirty_fm
->size
);
1800 flash_safe_release(fmc
->mtd
);
1803 D3(printk("jffs_write_dummy_node(): Leaving...\n"));
1808 /* Write a raw inode, possibly its name and possibly some data. */
1810 jffs_write_node(struct jffs_control
*c
, struct jffs_node
*node
,
1811 struct jffs_raw_inode
*raw_inode
,
1812 const char *name
, const unsigned char *data
,
1814 struct jffs_file
*f
)
1816 struct jffs_fmcontrol
*fmc
= c
->fmc
;
1818 struct kvec node_iovec
[4];
1819 unsigned long iovec_cnt
;
1825 __u32 total_name_size
= raw_inode
->nsize
1826 + JFFS_GET_PAD_BYTES(raw_inode
->nsize
);
1827 __u32 total_data_size
= raw_inode
->dsize
1828 + JFFS_GET_PAD_BYTES(raw_inode
->dsize
);
1829 __u32 total_size
= sizeof(struct jffs_raw_inode
)
1830 + total_name_size
+ total_data_size
;
1832 /* If this node isn't something that will eventually let
1833 GC free even more space, then don't allow it unless
1834 there's at least max_chunk_size space still available
1837 slack
= fmc
->max_chunk_size
;
1840 /* Fire the retrorockets and shoot the fruiton torpedoes, sir! */
1843 printk("jffs_write_node(): node == NULL\n");
1846 ASSERT(if (raw_inode
&& raw_inode
->nsize
&& !name
) {
1847 printk("*** jffs_write_node(): nsize = %u but name == NULL\n",
1852 D1(printk("jffs_write_node(): filename = \"%s\", ino = %u, "
1853 "total_size = %u\n",
1854 (name
? name
: ""), raw_inode
->ino
,
1857 jffs_fm_write_lock(fmc
);
1864 /* Deadlocks suck. */
1865 while(fmc
->free_size
< fmc
->min_free_size
+ total_size
+ slack
) {
1866 jffs_fm_write_unlock(fmc
);
1867 if (!JFFS_ENOUGH_SPACE(c
, total_size
+ slack
))
1869 jffs_fm_write_lock(fmc
);
1872 /* First try to allocate some flash memory. */
1873 err
= jffs_fmalloc(fmc
, total_size
, node
, &fm
);
1875 if (err
== -ENOSPC
) {
1876 /* Just out of space. GC and try again */
1877 if (fmc
->dirty_size
< fmc
->sector_size
) {
1878 D(printk("jffs_write_node(): jffs_fmalloc(0x%p, %u) "
1879 "failed, no dirty space to GC\n", fmc
,
1884 D1(printk(KERN_INFO
"jffs_write_node(): Calling jffs_garbage_collect_now()\n"));
1885 jffs_fm_write_unlock(fmc
);
1886 if ((err
= jffs_garbage_collect_now(c
))) {
1887 D(printk("jffs_write_node(): jffs_garbage_collect_now() failed\n"));
1890 jffs_fm_write_lock(fmc
);
1895 jffs_fm_write_unlock(fmc
);
1897 D(printk("jffs_write_node(): jffs_fmalloc(0x%p, %u) "
1898 "failed!\n", fmc
, total_size
));
1903 /* The jffs_fm struct that we got is not good enough.
1904 Make that space dirty and try again */
1905 if ((err
= jffs_write_dummy_node(c
, fm
)) < 0) {
1908 jffs_fm_write_unlock(fmc
);
1909 D(printk("jffs_write_node(): "
1910 "jffs_write_dummy_node(): Failed!\n"));
1918 ASSERT(if (fm
->nodes
== 0) {
1919 printk(KERN_ERR
"jffs_write_node(): fm->nodes == 0\n");
1922 pos
= node
->fm
->offset
;
1924 /* Increment the version number here. We can't let the caller
1925 set it beforehand, because we might have had to do GC on a node
1926 of this file - and we'd end up reusing version numbers.
1929 raw_inode
->version
= f
->highest_version
+ 1;
1930 D1(printk (KERN_NOTICE
"jffs_write_node(): setting version of %s to %d\n", f
->name
, raw_inode
->version
));
1932 /* if the file was deleted, set the deleted bit in the raw inode */
1934 raw_inode
->deleted
= 1;
1937 /* Compute the checksum for the data and name chunks. */
1938 raw_inode
->dchksum
= jffs_checksum(data
, raw_inode
->dsize
);
1939 raw_inode
->nchksum
= jffs_checksum(name
, raw_inode
->nsize
);
1941 /* The checksum is calculated without the chksum and accurate
1942 fields so set them to zero first. */
1943 raw_inode
->accurate
= 0;
1944 raw_inode
->chksum
= 0;
1945 raw_inode
->chksum
= jffs_checksum(raw_inode
,
1946 sizeof(struct jffs_raw_inode
));
1947 raw_inode
->accurate
= 0xff;
1949 D3(printk("jffs_write_node(): About to write this raw inode to the "
1950 "flash at pos 0x%lx:\n", (long)pos
));
1951 D3(jffs_print_raw_inode(raw_inode
));
1953 /* The actual raw JFFS node */
1954 node_iovec
[0].iov_base
= (void *) raw_inode
;
1955 node_iovec
[0].iov_len
= (size_t) sizeof(struct jffs_raw_inode
);
1958 /* Get name and size if there is one */
1959 if (raw_inode
->nsize
) {
1960 node_iovec
[iovec_cnt
].iov_base
= (void *) name
;
1961 node_iovec
[iovec_cnt
].iov_len
= (size_t) raw_inode
->nsize
;
1964 if (JFFS_GET_PAD_BYTES(raw_inode
->nsize
)) {
1965 static unsigned char allff
[3]={255,255,255};
1966 /* Add some extra padding if necessary */
1967 node_iovec
[iovec_cnt
].iov_base
= allff
;
1968 node_iovec
[iovec_cnt
].iov_len
=
1969 JFFS_GET_PAD_BYTES(raw_inode
->nsize
);
1974 /* Get data and size if there is any */
1975 if (raw_inode
->dsize
) {
1976 node_iovec
[iovec_cnt
].iov_base
= (void *) data
;
1977 node_iovec
[iovec_cnt
].iov_len
= (size_t) raw_inode
->dsize
;
1979 /* No need to pad this because we're not actually putting
1984 if ((err
= flash_safe_writev(fmc
->mtd
, node_iovec
, iovec_cnt
,
1986 jffs_fmfree_partly(fmc
, fm
, 0);
1987 jffs_fm_write_unlock(fmc
);
1988 printk(KERN_ERR
"JFFS: jffs_write_node: Failed to write, "
1989 "requested %i, wrote %i\n", total_size
, err
);
1992 if (raw_inode
->deleted
)
1995 jffs_fm_write_unlock(fmc
);
1996 D3(printk("jffs_write_node(): Leaving...\n"));
1997 return raw_inode
->dsize
;
1998 } /* jffs_write_node() */
2001 /* Read data from the node and write it to the buffer. 'node_offset'
2002 is how much we have read from this particular node before and which
2003 shouldn't be read again. 'max_size' is how much space there is in
2006 jffs_get_node_data(struct jffs_file
*f
, struct jffs_node
*node
,
2007 unsigned char *buf
,__u32 node_offset
, __u32 max_size
)
2009 struct jffs_fmcontrol
*fmc
= f
->c
->fmc
;
2010 __u32 pos
= node
->fm
->offset
+ node
->fm_offset
+ node_offset
;
2011 __u32 avail
= node
->data_size
- node_offset
;
2014 D2(printk(" jffs_get_node_data(): file: \"%s\", ino: %u, "
2015 "version: %u, node_offset: %u\n",
2016 f
->name
, node
->ino
, node
->version
, node_offset
));
2018 r
= min(avail
, max_size
);
2019 D3(printk(KERN_NOTICE
"jffs_get_node_data\n"));
2020 flash_safe_read(fmc
->mtd
, pos
, buf
, r
);
2022 D3(printk(" jffs_get_node_data(): Read %u byte%s.\n",
2023 r
, (r
== 1 ? "" : "s")));
2029 /* Read data from the file's nodes. Write the data to the buffer
2030 'buf'. 'read_offset' tells how much data we should skip. */
2032 jffs_read_data(struct jffs_file
*f
, unsigned char *buf
, __u32 read_offset
,
2035 struct jffs_node
*node
;
2036 __u32 read_data
= 0; /* Total amount of read data. */
2037 __u32 node_offset
= 0;
2038 __u32 pos
= 0; /* Number of bytes traversed. */
2040 D2(printk("jffs_read_data(): file = \"%s\", read_offset = %d, "
2042 (f
->name
? f
->name
: ""), read_offset
, size
));
2044 if (read_offset
>= f
->size
) {
2045 D(printk(" f->size: %d\n", f
->size
));
2049 /* First find the node to read data from. */
2050 node
= f
->range_head
;
2051 while (pos
<= read_offset
) {
2052 node_offset
= read_offset
- pos
;
2053 if (node_offset
>= node
->data_size
) {
2054 pos
+= node
->data_size
;
2055 node
= node
->range_next
;
2062 /* "Cats are living proof that not everything in nature
2064 - Garrison Keilor ('97) */
2066 /* Fill the buffer. */
2067 while (node
&& (read_data
< size
)) {
2070 /* This node does not refer to real data. */
2071 r
= min(size
- read_data
,
2072 node
->data_size
- node_offset
);
2073 memset(&buf
[read_data
], 0, r
);
2075 else if ((r
= jffs_get_node_data(f
, node
, &buf
[read_data
],
2077 size
- read_data
)) < 0) {
2082 node
= node
->range_next
;
2084 D3(printk(" jffs_read_data(): Read %u bytes.\n", read_data
));
2089 /* Used for traversing all nodes in the hash table. */
2091 jffs_foreach_file(struct jffs_control
*c
, int (*func
)(struct jffs_file
*))
2097 for (pos
= 0; pos
< c
->hash_len
; pos
++) {
2098 struct jffs_file
*f
, *next
;
2100 /* We must do _safe, because 'func' might remove the
2101 current file 'f' from the list. */
2102 list_for_each_entry_safe(f
, next
, &c
->hash
[pos
], hash
) {
2114 /* Free all nodes associated with a file. */
2116 jffs_free_node_list(struct jffs_file
*f
)
2118 struct jffs_node
*node
;
2119 struct jffs_node
*p
;
2121 D3(printk("jffs_free_node_list(): f #%u, \"%s\"\n",
2122 f
->ino
, (f
->name
? f
->name
: "")));
2123 node
= f
->version_head
;
2126 node
= node
->version_next
;
2128 DJM(no_jffs_node
--);
2134 /* Free a file and its name. */
2136 jffs_free_file(struct jffs_file
*f
)
2138 D3(printk("jffs_free_file: f #%u, \"%s\"\n",
2139 f
->ino
, (f
->name
? f
->name
: "")));
2151 jffs_get_file_count(void)
2153 return no_jffs_file
;
2156 /* See if a file is deleted. If so, mark that file's nodes as obsolete. */
2158 jffs_possibly_delete_file(struct jffs_file
*f
)
2160 struct jffs_node
*n
;
2162 D3(printk("jffs_possibly_delete_file(): ino: %u\n",
2166 printk(KERN_ERR
"jffs_possibly_delete_file(): f == NULL\n");
2171 /* First try to remove all older versions. Commence with
2173 for (n
= f
->version_head
; n
; n
= n
->version_next
) {
2177 if (jffs_fmfree(f
->c
->fmc
, n
->fm
, n
) < 0) {
2181 /* Unlink the file from the filesystem. */
2182 if (!f
->c
->building_fs
) {
2183 jffs_unlink_file_from_tree(f
);
2185 jffs_unlink_file_from_hash(f
);
2186 jffs_free_node_list(f
);
2193 /* Used in conjunction with jffs_foreach_file() to count the number
2194 of files in the file system. */
2196 jffs_file_count(struct jffs_file
*f
)
2202 /* Build up a file's range list from scratch by going through the
2205 jffs_build_file(struct jffs_file
*f
)
2207 struct jffs_node
*n
;
2209 D3(printk("jffs_build_file(): ino: %u, name: \"%s\"\n",
2210 f
->ino
, (f
->name
? f
->name
: "")));
2212 for (n
= f
->version_head
; n
; n
= n
->version_next
) {
2213 jffs_update_file(f
, n
);
2219 /* Remove an amount of data from a file. If this amount of data is
2220 zero, that could mean that a node should be split in two parts.
2221 We remove or change the appropriate nodes in the lists.
2223 Starting offset of area to be removed is node->data_offset,
2224 and the length of the area is in node->removed_size. */
2226 jffs_delete_data(struct jffs_file
*f
, struct jffs_node
*node
)
2228 struct jffs_node
*n
;
2229 __u32 offset
= node
->data_offset
;
2230 __u32 remove_size
= node
->removed_size
;
2232 D3(printk("jffs_delete_data(): offset = %u, remove_size = %u\n",
2233 offset
, remove_size
));
2235 if (remove_size
== 0
2237 && f
->range_tail
->data_offset
+ f
->range_tail
->data_size
2239 /* A simple append; nothing to remove or no node to split. */
2243 /* Find the node where we should begin the removal. */
2244 for (n
= f
->range_head
; n
; n
= n
->range_next
) {
2245 if (n
->data_offset
+ n
->data_size
> offset
) {
2250 /* If there's no data in the file there's no data to
2255 if (n
->data_offset
> offset
) {
2256 /* XXX: Not implemented yet. */
2257 printk(KERN_WARNING
"JFFS: An unexpected situation "
2258 "occurred in jffs_delete_data.\n");
2260 else if (n
->data_offset
< offset
) {
2261 /* See if the node has to be split into two parts. */
2262 if (n
->data_offset
+ n
->data_size
> offset
+ remove_size
) {
2264 struct jffs_node
*new_node
;
2265 D3(printk("jffs_delete_data(): Split node with "
2266 "version number %u.\n", n
->version
));
2268 if (!(new_node
= jffs_alloc_node())) {
2269 D(printk("jffs_delete_data(): -ENOMEM\n"));
2272 DJM(no_jffs_node
++);
2274 new_node
->ino
= n
->ino
;
2275 new_node
->version
= n
->version
;
2276 new_node
->data_offset
= offset
;
2277 new_node
->data_size
= n
->data_size
- (remove_size
+ (offset
- n
->data_offset
));
2278 new_node
->fm_offset
= n
->fm_offset
+ (remove_size
+ (offset
- n
->data_offset
));
2279 new_node
->name_size
= n
->name_size
;
2280 new_node
->fm
= n
->fm
;
2281 new_node
->version_prev
= n
;
2282 new_node
->version_next
= n
->version_next
;
2283 if (new_node
->version_next
) {
2284 new_node
->version_next
->version_prev
2288 f
->version_tail
= new_node
;
2290 n
->version_next
= new_node
;
2291 new_node
->range_prev
= n
;
2292 new_node
->range_next
= n
->range_next
;
2293 if (new_node
->range_next
) {
2294 new_node
->range_next
->range_prev
= new_node
;
2297 f
->range_tail
= new_node
;
2299 /* A very interesting can of worms. */
2300 n
->range_next
= new_node
;
2301 n
->data_size
= offset
- n
->data_offset
;
2303 jffs_add_node(new_node
);
2305 D1(printk(KERN_WARNING
"jffs_delete_data(): Splitting an empty node (file hold).\n!"));
2306 D1(printk(KERN_WARNING
"FIXME: Did dwmw2 do the right thing here?\n"));
2308 n
= new_node
->range_next
;
2312 /* No. No need to split the node. Just remove
2313 the end of the node. */
2314 int r
= min(n
->data_offset
+ n
->data_size
2315 - offset
, remove_size
);
2322 /* Remove as many nodes as necessary. */
2323 while (n
&& remove_size
) {
2324 if (n
->data_size
<= remove_size
) {
2325 struct jffs_node
*p
= n
;
2326 remove_size
-= n
->data_size
;
2328 D3(printk("jffs_delete_data(): Removing node: "
2329 "ino: %u, version: %u%s\n",
2331 (p
->fm
? "" : " (virtual)")));
2333 jffs_fmfree(f
->c
->fmc
, p
->fm
, p
);
2335 jffs_unlink_node_from_range_list(f
, p
);
2336 jffs_unlink_node_from_version_list(f
, p
);
2338 DJM(no_jffs_node
--);
2341 n
->data_size
-= remove_size
;
2342 n
->fm_offset
+= remove_size
;
2343 n
->data_offset
-= (node
->removed_size
- remove_size
);
2349 /* Adjust the following nodes' information about offsets etc. */
2350 while (n
&& node
->removed_size
) {
2351 n
->data_offset
-= node
->removed_size
;
2355 if (node
->removed_size
> (f
->size
- node
->data_offset
)) {
2356 /* It's possible that the removed_size is in fact
2357 * greater than the amount of data we actually thought
2358 * were present in the first place - some of the nodes
2359 * which this node originally obsoleted may already have
2360 * been deleted from the flash by subsequent garbage
2363 * If this is the case, don't let f->size go negative.
2364 * Bad things would happen :)
2366 f
->size
= node
->data_offset
;
2368 f
->size
-= node
->removed_size
;
2370 D3(printk("jffs_delete_data(): f->size = %d\n", f
->size
));
2372 } /* jffs_delete_data() */
2375 /* Insert some data into a file. Prior to the call to this function,
2376 jffs_delete_data should be called. */
2378 jffs_insert_data(struct jffs_file
*f
, struct jffs_node
*node
)
2380 D3(printk("jffs_insert_data(): node->data_offset = %u, "
2381 "node->data_size = %u, f->size = %u\n",
2382 node
->data_offset
, node
->data_size
, f
->size
));
2384 /* Find the position where we should insert data. */
2386 if (node
->data_offset
== f
->size
) {
2387 /* A simple append. This is the most common operation. */
2388 node
->range_next
= NULL
;
2389 node
->range_prev
= f
->range_tail
;
2390 if (node
->range_prev
) {
2391 node
->range_prev
->range_next
= node
;
2393 f
->range_tail
= node
;
2394 f
->size
+= node
->data_size
;
2395 if (!f
->range_head
) {
2396 f
->range_head
= node
;
2399 else if (node
->data_offset
< f
->size
) {
2400 /* Trying to insert data into the middle of the file. This
2401 means no problem because jffs_delete_data() has already
2402 prepared the range list for us. */
2403 struct jffs_node
*n
;
2405 /* Find the correct place for the insertion and then insert
2407 for (n
= f
->range_head
; n
; n
= n
->range_next
) {
2408 D2(printk("Cool stuff's happening!\n"));
2410 if (n
->data_offset
== node
->data_offset
) {
2411 node
->range_prev
= n
->range_prev
;
2412 if (node
->range_prev
) {
2413 node
->range_prev
->range_next
= node
;
2416 f
->range_head
= node
;
2418 node
->range_next
= n
;
2419 n
->range_prev
= node
;
2422 ASSERT(else if (n
->data_offset
+ n
->data_size
>
2423 node
->data_offset
) {
2424 printk(KERN_ERR
"jffs_insert_data(): "
2425 "Couldn't find a place to insert "
2431 /* Adjust later nodes' offsets etc. */
2432 n
= node
->range_next
;
2434 n
->data_offset
+= node
->data_size
;
2437 f
->size
+= node
->data_size
;
2439 else if (node
->data_offset
> f
->size
) {
2440 /* Okay. This is tricky. This means that we want to insert
2441 data at a place that is beyond the limits of the file as
2442 it is constructed right now. This is actually a common
2443 event that for instance could occur during the mounting
2444 of the file system if a large file have been truncated,
2445 rewritten and then only partially garbage collected. */
2447 struct jffs_node
*n
;
2449 /* We need a place holder for the data that is missing in
2450 front of this insertion. This "virtual node" will not
2451 be associated with any space on the flash device. */
2452 struct jffs_node
*virtual_node
;
2453 if (!(virtual_node
= jffs_alloc_node())) {
2457 D(printk("jffs_insert_data: Inserting a virtual node.\n"));
2458 D(printk(" node->data_offset = %u\n", node
->data_offset
));
2459 D(printk(" f->size = %u\n", f
->size
));
2461 virtual_node
->ino
= node
->ino
;
2462 virtual_node
->version
= node
->version
;
2463 virtual_node
->removed_size
= 0;
2464 virtual_node
->fm_offset
= 0;
2465 virtual_node
->name_size
= 0;
2466 virtual_node
->fm
= NULL
; /* This is a virtual data holder. */
2467 virtual_node
->version_prev
= NULL
;
2468 virtual_node
->version_next
= NULL
;
2469 virtual_node
->range_next
= NULL
;
2471 /* Are there any data at all in the file yet? */
2472 if (f
->range_head
) {
2473 virtual_node
->data_offset
2474 = f
->range_tail
->data_offset
2475 + f
->range_tail
->data_size
;
2476 virtual_node
->data_size
2477 = node
->data_offset
- virtual_node
->data_offset
;
2478 virtual_node
->range_prev
= f
->range_tail
;
2479 f
->range_tail
->range_next
= virtual_node
;
2482 virtual_node
->data_offset
= 0;
2483 virtual_node
->data_size
= node
->data_offset
;
2484 virtual_node
->range_prev
= NULL
;
2485 f
->range_head
= virtual_node
;
2488 f
->range_tail
= virtual_node
;
2489 f
->size
+= virtual_node
->data_size
;
2491 /* Insert this virtual node in the version list as well. */
2492 for (n
= f
->version_head
; n
; n
= n
->version_next
) {
2493 if (n
->version
== virtual_node
->version
) {
2494 virtual_node
->version_prev
= n
->version_prev
;
2495 n
->version_prev
= virtual_node
;
2496 if (virtual_node
->version_prev
) {
2497 virtual_node
->version_prev
2498 ->version_next
= virtual_node
;
2501 f
->version_head
= virtual_node
;
2503 virtual_node
->version_next
= n
;
2508 D(jffs_print_node(virtual_node
));
2510 /* Make a new try to insert the node. */
2514 D3(printk("jffs_insert_data(): f->size = %d\n", f
->size
));
2519 /* A new node (with data) has been added to the file and now the range
2520 list has to be modified. */
2522 jffs_update_file(struct jffs_file
*f
, struct jffs_node
*node
)
2526 D3(printk("jffs_update_file(): ino: %u, version: %u\n",
2527 f
->ino
, node
->version
));
2529 if (node
->data_size
== 0) {
2530 if (node
->removed_size
== 0) {
2531 /* data_offset == X */
2532 /* data_size == 0 */
2533 /* remove_size == 0 */
2536 /* data_offset == X */
2537 /* data_size == 0 */
2538 /* remove_size != 0 */
2539 if ((err
= jffs_delete_data(f
, node
)) < 0) {
2545 /* data_offset == X */
2546 /* data_size != 0 */
2547 /* remove_size == Y */
2548 if ((err
= jffs_delete_data(f
, node
)) < 0) {
2551 if ((err
= jffs_insert_data(f
, node
)) < 0) {
2558 /* Print the contents of a file. */
2561 jffs_print_file(struct jffs_file
*f
)
2564 D(printk("jffs_file: 0x%p\n", f
));
2566 D(printk(" 0x%08x, /* ino */\n", f
->ino
));
2567 D(printk(" 0x%08x, /* pino */\n", f
->pino
));
2568 D(printk(" 0x%08x, /* mode */\n", f
->mode
));
2569 D(printk(" 0x%04x, /* uid */\n", f
->uid
));
2570 D(printk(" 0x%04x, /* gid */\n", f
->gid
));
2571 D(printk(" 0x%08x, /* atime */\n", f
->atime
));
2572 D(printk(" 0x%08x, /* mtime */\n", f
->mtime
));
2573 D(printk(" 0x%08x, /* ctime */\n", f
->ctime
));
2574 D(printk(" 0x%02x, /* nsize */\n", f
->nsize
));
2575 D(printk(" 0x%02x, /* nlink */\n", f
->nlink
));
2576 D(printk(" 0x%02x, /* deleted */\n", f
->deleted
));
2577 D(printk(" \"%s\", ", (f
->name
? f
->name
: "")));
2578 D(for (i
= strlen(f
->name
? f
->name
: ""); i
< 8; ++i
) {
2581 D(printk("/* name */\n"));
2582 D(printk(" 0x%08x, /* size */\n", f
->size
));
2583 D(printk(" 0x%08x, /* highest_version */\n",
2584 f
->highest_version
));
2585 D(printk(" 0x%p, /* c */\n", f
->c
));
2586 D(printk(" 0x%p, /* parent */\n", f
->parent
));
2587 D(printk(" 0x%p, /* children */\n", f
->children
));
2588 D(printk(" 0x%p, /* sibling_prev */\n", f
->sibling_prev
));
2589 D(printk(" 0x%p, /* sibling_next */\n", f
->sibling_next
));
2590 D(printk(" 0x%p, /* hash_prev */\n", f
->hash
.prev
));
2591 D(printk(" 0x%p, /* hash_next */\n", f
->hash
.next
));
2592 D(printk(" 0x%p, /* range_head */\n", f
->range_head
));
2593 D(printk(" 0x%p, /* range_tail */\n", f
->range_tail
));
2594 D(printk(" 0x%p, /* version_head */\n", f
->version_head
));
2595 D(printk(" 0x%p, /* version_tail */\n", f
->version_tail
));
2602 jffs_print_hash_table(struct jffs_control
*c
)
2606 printk("JFFS: Dumping the file system's hash table...\n");
2607 for (i
= 0; i
< c
->hash_len
; i
++) {
2608 struct jffs_file
*f
;
2609 list_for_each_entry(f
, &c
->hash
[i
], hash
) {
2610 printk("*** c->hash[%u]: \"%s\" "
2611 "(ino: %u, pino: %u)\n",
2612 i
, (f
->name
? f
->name
: ""),
2620 jffs_print_tree(struct jffs_file
*first_file
, int indent
)
2622 struct jffs_file
*f
;
2630 if (!(space
= (char *) kmalloc(indent
+ 1, GFP_KERNEL
))) {
2631 printk("jffs_print_tree(): Out of memory!\n");
2635 memset(space
, ' ', indent
);
2636 space
[indent
] = '\0';
2638 for (f
= first_file
; f
; f
= f
->sibling_next
) {
2639 dir
= S_ISDIR(f
->mode
);
2640 printk("%s%s%s (ino: %u, highest_version: %u, size: %u)\n",
2641 space
, (f
->name
? f
->name
: ""), (dir
? "/" : ""),
2642 f
->ino
, f
->highest_version
, f
->size
);
2644 jffs_print_tree(f
->children
, indent
+ 2);
2652 #if defined(JFFS_MEMORY_DEBUG) && JFFS_MEMORY_DEBUG
2654 jffs_print_memory_allocation_statistics(void)
2656 static long printout
;
2657 printk("________ Memory printout #%ld ________\n", ++printout
);
2658 printk("no_jffs_file = %ld\n", no_jffs_file
);
2659 printk("no_jffs_node = %ld\n", no_jffs_node
);
2660 printk("no_jffs_control = %ld\n", no_jffs_control
);
2661 printk("no_jffs_raw_inode = %ld\n", no_jffs_raw_inode
);
2662 printk("no_jffs_node_ref = %ld\n", no_jffs_node_ref
);
2663 printk("no_jffs_fm = %ld\n", no_jffs_fm
);
2664 printk("no_jffs_fmcontrol = %ld\n", no_jffs_fmcontrol
);
2665 printk("no_hash = %ld\n", no_hash
);
2666 printk("no_name = %ld\n", no_name
);
2672 /* Rewrite `size' bytes, and begin at `node'. */
2674 jffs_rewrite_data(struct jffs_file
*f
, struct jffs_node
*node
, __u32 size
)
2676 struct jffs_control
*c
= f
->c
;
2677 struct jffs_fmcontrol
*fmc
= c
->fmc
;
2678 struct jffs_raw_inode raw_inode
;
2679 struct jffs_node
*new_node
;
2683 __u32 total_name_size
;
2684 __u32 total_data_size
;
2688 D1(printk("***jffs_rewrite_data(): node: %u, name: \"%s\", size: %u\n",
2689 f
->ino
, (f
->name
? f
->name
: "(null)"), size
));
2691 /* Create and initialize the new node. */
2692 if (!(new_node
= jffs_alloc_node())) {
2693 D(printk("jffs_rewrite_data(): "
2694 "Failed to allocate node.\n"));
2697 DJM(no_jffs_node
++);
2698 new_node
->data_offset
= node
->data_offset
;
2699 new_node
->removed_size
= size
;
2700 total_name_size
= JFFS_PAD(f
->nsize
);
2701 total_data_size
= JFFS_PAD(size
);
2702 total_size
= sizeof(struct jffs_raw_inode
)
2703 + total_name_size
+ total_data_size
;
2704 new_node
->fm_offset
= sizeof(struct jffs_raw_inode
)
2708 jffs_fm_write_lock(fmc
);
2711 if ((err
= jffs_fmalloc(fmc
, total_size
, new_node
, &fm
)) < 0) {
2712 DJM(no_jffs_node
--);
2713 jffs_fm_write_unlock(fmc
);
2714 D(printk("jffs_rewrite_data(): Failed to allocate fm.\n"));
2715 jffs_free_node(new_node
);
2718 else if (!fm
->nodes
) {
2719 /* The jffs_fm struct that we got is not big enough. */
2720 /* This should never happen, because we deal with this case
2721 in jffs_garbage_collect_next().*/
2722 printk(KERN_WARNING
"jffs_rewrite_data(): Allocated node is too small (%d bytes of %d)\n", fm
->size
, total_size
);
2723 if ((err
= jffs_write_dummy_node(c
, fm
)) < 0) {
2724 D(printk("jffs_rewrite_data(): "
2725 "jffs_write_dummy_node() Failed!\n"));
2730 jffs_fm_write_unlock(fmc
);
2737 /* Initialize the raw inode. */
2738 raw_inode
.magic
= JFFS_MAGIC_BITMASK
;
2739 raw_inode
.ino
= f
->ino
;
2740 raw_inode
.pino
= f
->pino
;
2741 raw_inode
.version
= f
->highest_version
+ 1;
2742 raw_inode
.mode
= f
->mode
;
2743 raw_inode
.uid
= f
->uid
;
2744 raw_inode
.gid
= f
->gid
;
2745 raw_inode
.atime
= f
->atime
;
2746 raw_inode
.mtime
= f
->mtime
;
2747 raw_inode
.ctime
= f
->ctime
;
2748 raw_inode
.offset
= node
->data_offset
;
2749 raw_inode
.dsize
= size
;
2750 raw_inode
.rsize
= size
;
2751 raw_inode
.nsize
= f
->nsize
;
2752 raw_inode
.nlink
= f
->nlink
;
2753 raw_inode
.spare
= 0;
2754 raw_inode
.rename
= 0;
2755 raw_inode
.deleted
= f
->deleted
;
2756 raw_inode
.accurate
= 0xff;
2757 raw_inode
.dchksum
= 0;
2758 raw_inode
.nchksum
= 0;
2760 pos
= new_node
->fm
->offset
;
2761 pos_dchksum
= pos
+JFFS_RAW_INODE_DCHKSUM_OFFSET
;
2763 D3(printk("jffs_rewrite_data(): Writing this raw inode "
2764 "to pos 0x%ul.\n", pos
));
2765 D3(jffs_print_raw_inode(&raw_inode
));
2767 if ((err
= flash_safe_write(fmc
->mtd
, pos
,
2768 (u_char
*) &raw_inode
,
2769 sizeof(struct jffs_raw_inode
)
2771 - sizeof(__u16
) - sizeof(__u16
))) < 0) {
2772 jffs_fmfree_partly(fmc
, fm
,
2773 total_name_size
+ total_data_size
);
2774 jffs_fm_write_unlock(fmc
);
2775 printk(KERN_ERR
"JFFS: jffs_rewrite_data: Write error during "
2776 "rewrite. (raw inode)\n");
2777 printk(KERN_ERR
"JFFS: jffs_rewrite_data: Now retrying "
2778 "rewrite. (raw inode)\n");
2781 pos
+= sizeof(struct jffs_raw_inode
);
2783 /* Write the name to the flash memory. */
2785 D3(printk("jffs_rewrite_data(): Writing name \"%s\" to "
2786 "pos 0x%ul.\n", f
->name
, (unsigned int) pos
));
2787 if ((err
= flash_safe_write(fmc
->mtd
, pos
,
2790 jffs_fmfree_partly(fmc
, fm
, total_data_size
);
2791 jffs_fm_write_unlock(fmc
);
2792 printk(KERN_ERR
"JFFS: jffs_rewrite_data: Write "
2793 "error during rewrite. (name)\n");
2794 printk(KERN_ERR
"JFFS: jffs_rewrite_data: Now retrying "
2795 "rewrite. (name)\n");
2798 pos
+= total_name_size
;
2799 raw_inode
.nchksum
= jffs_checksum(f
->name
, f
->nsize
);
2802 /* Write the data. */
2805 unsigned char *page
;
2806 __u32 offset
= node
->data_offset
;
2808 if (!(page
= (unsigned char *)__get_free_page(GFP_KERNEL
))) {
2809 jffs_fmfree_partly(fmc
, fm
, 0);
2814 __u32 s
= min(size
, (__u32
)PAGE_SIZE
);
2815 if ((r
= jffs_read_data(f
, (char *)page
,
2817 free_page((unsigned long)page
);
2818 jffs_fmfree_partly(fmc
, fm
, 0);
2819 jffs_fm_write_unlock(fmc
);
2820 printk(KERN_ERR
"JFFS: jffs_rewrite_data: "
2822 "failed! (r = %d)\n", r
);
2825 if ((err
= flash_safe_write(fmc
->mtd
,
2826 pos
, page
, r
)) < 0) {
2827 free_page((unsigned long)page
);
2828 jffs_fmfree_partly(fmc
, fm
, 0);
2829 jffs_fm_write_unlock(fmc
);
2830 printk(KERN_ERR
"JFFS: jffs_rewrite_data: "
2831 "Write error during rewrite. "
2838 raw_inode
.dchksum
+= jffs_checksum(page
, r
);
2841 free_page((unsigned long)page
);
2844 raw_inode
.accurate
= 0;
2845 raw_inode
.chksum
= jffs_checksum(&raw_inode
,
2846 sizeof(struct jffs_raw_inode
)
2849 /* Add the checksum. */
2851 = flash_safe_write(fmc
->mtd
, pos_dchksum
,
2853 &raw_inode
)[JFFS_RAW_INODE_DCHKSUM_OFFSET
],
2854 sizeof(__u32
) + sizeof(__u16
)
2855 + sizeof(__u16
))) < 0) {
2856 jffs_fmfree_partly(fmc
, fm
, 0);
2857 jffs_fm_write_unlock(fmc
);
2858 printk(KERN_ERR
"JFFS: jffs_rewrite_data: Write error during "
2859 "rewrite. (checksum)\n");
2863 /* Now make the file system aware of the newly written node. */
2864 jffs_insert_node(c
, f
, &raw_inode
, f
->name
, new_node
);
2865 jffs_fm_write_unlock(fmc
);
2867 D3(printk("jffs_rewrite_data(): Leaving...\n"));
2869 } /* jffs_rewrite_data() */
2872 /* jffs_garbage_collect_next implements one step in the garbage collect
2873 process and is often called multiple times at each occasion of a
2877 jffs_garbage_collect_next(struct jffs_control
*c
)
2879 struct jffs_fmcontrol
*fmc
= c
->fmc
;
2880 struct jffs_node
*node
;
2881 struct jffs_file
*f
;
2885 __u32 total_name_size
;
2886 __u32 extra_available
;
2888 __u32 free_chunk_size1
= jffs_free_size1(fmc
);
2889 D2(__u32 free_chunk_size2
= jffs_free_size2(fmc
));
2891 /* Get the oldest node in the flash. */
2892 node
= jffs_get_oldest_node(fmc
);
2894 printk(KERN_ERR
"JFFS: jffs_garbage_collect_next: "
2895 "No oldest node found!\n");
2897 goto jffs_garbage_collect_next_end
;
2902 /* Find its corresponding file too. */
2903 f
= jffs_find_file(c
, node
->ino
);
2906 printk (KERN_ERR
"JFFS: jffs_garbage_collect_next: "
2907 "No file to garbage collect! "
2908 "(ino = 0x%08x)\n", node
->ino
);
2909 /* FIXME: Free the offending node and recover. */
2911 goto jffs_garbage_collect_next_end
;
2914 /* We always write out the name. Theoretically, we don't need
2915 to, but for now it's easier - because otherwise we'd have
2916 to keep track of how many times the current name exists on
2917 the flash and make sure it never reaches zero.
2919 The current approach means that would be possible to cause
2920 the GC to end up eating its tail by writing lots of nodes
2921 with no name for it to garbage-collect. Hence the change in
2922 inode.c to write names with _every_ node.
2924 It sucks, but it _should_ work.
2926 total_name_size
= JFFS_PAD(f
->nsize
);
2928 D1(printk("jffs_garbage_collect_next(): \"%s\", "
2929 "ino: %u, version: %u, location 0x%x, dsize %u\n",
2930 (f
->name
? f
->name
: ""), node
->ino
, node
->version
,
2931 node
->fm
->offset
, node
->data_size
));
2933 /* Compute how many data it's possible to rewrite at the moment. */
2934 data_size
= f
->size
- node
->data_offset
;
2936 /* And from that, the total size of the chunk we want to write */
2937 size
= sizeof(struct jffs_raw_inode
) + total_name_size
2938 + data_size
+ JFFS_GET_PAD_BYTES(data_size
);
2940 /* If that's more than max_chunk_size, reduce it accordingly */
2941 if (size
> fmc
->max_chunk_size
) {
2942 size
= fmc
->max_chunk_size
;
2943 data_size
= size
- sizeof(struct jffs_raw_inode
)
2947 /* If we're asking to take up more space than free_chunk_size1
2948 but we _could_ fit in it, shrink accordingly.
2950 if (size
> free_chunk_size1
) {
2952 if (free_chunk_size1
<
2953 (sizeof(struct jffs_raw_inode
) + total_name_size
+ BLOCK_SIZE
)){
2954 /* The space left is too small to be of any
2956 struct jffs_fm
*dirty_fm
2957 = jffs_fmalloced(fmc
,
2958 fmc
->tail
->offset
+ fmc
->tail
->size
,
2959 free_chunk_size1
, NULL
);
2961 printk(KERN_ERR
"JFFS: "
2962 "jffs_garbage_collect_next: "
2963 "Failed to allocate `dirty' "
2966 goto jffs_garbage_collect_next_end
;
2968 D1(printk("Dirtying end of flash - too small\n"));
2969 jffs_write_dummy_node(c
, dirty_fm
);
2971 goto jffs_garbage_collect_next_end
;
2973 D1(printk("Reducing size of new node from %d to %d to avoid "
2974 " exceeding free_chunk_size1\n",
2975 size
, free_chunk_size1
));
2977 size
= free_chunk_size1
;
2978 data_size
= size
- sizeof(struct jffs_raw_inode
)
2983 /* Calculate the amount of space needed to hold the nodes
2984 which are remaining in the tail */
2985 space_needed
= fmc
->min_free_size
- (node
->fm
->offset
% fmc
->sector_size
);
2987 /* From that, calculate how much 'extra' space we can use to
2988 increase the size of the node we're writing from the size
2989 of the node we're obsoleting
2991 if (space_needed
> fmc
->free_size
) {
2992 /* If we've gone below min_free_size for some reason,
2993 don't fuck up. This is why we have
2994 min_free_size > sector_size. Whinge about it though,
2995 just so I can convince myself my maths is right.
2997 D1(printk(KERN_WARNING
"jffs_garbage_collect_next(): "
2998 "space_needed %d exceeded free_size %d\n",
2999 space_needed
, fmc
->free_size
));
3000 extra_available
= 0;
3002 extra_available
= fmc
->free_size
- space_needed
;
3005 /* Check that we don't use up any more 'extra' space than
3007 if (size
> JFFS_PAD(node
->data_size
) + total_name_size
+
3008 sizeof(struct jffs_raw_inode
) + extra_available
) {
3009 D1(printk("Reducing size of new node from %d to %ld to avoid "
3010 "catching our tail\n", size
,
3011 (long) (JFFS_PAD(node
->data_size
) + JFFS_PAD(node
->name_size
) +
3012 sizeof(struct jffs_raw_inode
) + extra_available
)));
3013 D1(printk("space_needed = %d, extra_available = %d\n",
3014 space_needed
, extra_available
));
3016 size
= JFFS_PAD(node
->data_size
) + total_name_size
+
3017 sizeof(struct jffs_raw_inode
) + extra_available
;
3018 data_size
= size
- sizeof(struct jffs_raw_inode
)
3022 D2(printk(" total_name_size: %u\n", total_name_size
));
3023 D2(printk(" data_size: %u\n", data_size
));
3024 D2(printk(" size: %u\n", size
));
3025 D2(printk(" f->nsize: %u\n", f
->nsize
));
3026 D2(printk(" f->size: %u\n", f
->size
));
3027 D2(printk(" node->data_offset: %u\n", node
->data_offset
));
3028 D2(printk(" free_chunk_size1: %u\n", free_chunk_size1
));
3029 D2(printk(" free_chunk_size2: %u\n", free_chunk_size2
));
3030 D2(printk(" node->fm->offset: 0x%08x\n", node
->fm
->offset
));
3032 if ((err
= jffs_rewrite_data(f
, node
, data_size
))) {
3033 printk(KERN_WARNING
"jffs_rewrite_data() failed: %d\n", err
);
3037 jffs_garbage_collect_next_end
:
3038 D3(printk("jffs_garbage_collect_next: Leaving...\n"));
3040 } /* jffs_garbage_collect_next */
3043 /* If an obsolete node is partly going to be erased due to garbage
3044 collection, the part that isn't going to be erased must be filled
3045 with zeroes so that the scan of the flash will work smoothly next
3046 time. (The data in the file could for instance be a JFFS image
3047 which could cause enormous confusion during a scan of the flash
3048 device if we didn't do this.)
3049 There are two phases in this procedure: First, the clearing of
3050 the name and data parts of the node. Second, possibly also clearing
3051 a part of the raw inode as well. If the box is power cycled during
3052 the first phase, only the checksum of this node-to-be-cleared-at-
3053 the-end will be wrong. If the box is power cycled during, or after,
3054 the clearing of the raw inode, the information like the length of
3055 the name and data parts are zeroed. The next time the box is
3056 powered up, the scanning algorithm manages this faulty data too
3059 - The checksum is invalid and thus the raw inode must be discarded
3061 - If the lengths of the data part or the name part are zeroed, the
3062 scanning just continues after the raw inode. But after the inode
3063 the scanning procedure just finds zeroes which is the same as
3066 So, in the end, this could never fail. :-) Even if it does fail,
3067 the scanning algorithm should manage that too. */
3070 jffs_clear_end_of_node(struct jffs_control
*c
, __u32 erase_size
)
3073 struct jffs_fmcontrol
*fmc
= c
->fmc
;
3076 __u32 zero_offset_data
;
3077 __u32 zero_size_data
;
3078 __u32 cutting_raw_inode
= 0;
3080 if (!(fm
= jffs_cut_node(fmc
, erase_size
))) {
3081 D3(printk("jffs_clear_end_of_node(): fm == NULL\n"));
3085 /* Where and how much shall we clear? */
3086 zero_offset
= fmc
->head
->offset
+ erase_size
;
3087 zero_size
= fm
->offset
+ fm
->size
- zero_offset
;
3089 /* Do we have to clear the raw_inode explicitly? */
3090 if (fm
->size
- zero_size
< sizeof(struct jffs_raw_inode
)) {
3091 cutting_raw_inode
= sizeof(struct jffs_raw_inode
)
3092 - (fm
->size
- zero_size
);
3095 /* First, clear the name and data fields. */
3096 zero_offset_data
= zero_offset
+ cutting_raw_inode
;
3097 zero_size_data
= zero_size
- cutting_raw_inode
;
3098 flash_safe_acquire(fmc
->mtd
);
3099 flash_memset(fmc
->mtd
, zero_offset_data
, 0, zero_size_data
);
3100 flash_safe_release(fmc
->mtd
);
3102 /* Should we clear a part of the raw inode? */
3103 if (cutting_raw_inode
) {
3104 /* I guess it is ok to clear the raw inode in this order. */
3105 flash_safe_acquire(fmc
->mtd
);
3106 flash_memset(fmc
->mtd
, zero_offset
, 0,
3108 flash_safe_release(fmc
->mtd
);
3112 } /* jffs_clear_end_of_node() */
3114 /* Try to erase as much as possible of the dirt in the flash memory. */
3116 jffs_try_to_erase(struct jffs_control
*c
)
3118 struct jffs_fmcontrol
*fmc
= c
->fmc
;
3123 D3(printk("jffs_try_to_erase()\n"));
3125 erase_size
= jffs_erasable_size(fmc
);
3127 D2(printk("jffs_try_to_erase(): erase_size = %ld\n", erase_size
));
3129 if (erase_size
== 0) {
3132 else if (erase_size
< 0) {
3133 printk(KERN_ERR
"JFFS: jffs_try_to_erase: "
3134 "jffs_erasable_size returned %ld.\n", erase_size
);
3138 if ((err
= jffs_clear_end_of_node(c
, erase_size
)) < 0) {
3139 printk(KERN_ERR
"JFFS: jffs_try_to_erase: "
3140 "Clearing of node failed.\n");
3144 offset
= fmc
->head
->offset
;
3146 /* Now, let's try to do the erase. */
3147 if ((err
= flash_erase_region(fmc
->mtd
,
3148 offset
, erase_size
)) < 0) {
3149 printk(KERN_ERR
"JFFS: Erase of flash failed. "
3150 "offset = %u, erase_size = %ld\n",
3151 offset
, erase_size
);
3152 /* XXX: Here we should allocate this area as dirty
3153 with jffs_fmalloced or something similar. Now
3154 we just report the error. */
3159 /* Check if the erased sectors really got erased. */
3164 pos
= (__u32
)flash_get_direct_pointer(to_kdev_t(c
->sb
->s_dev
), offset
);
3165 end
= pos
+ erase_size
;
3167 D2(printk("JFFS: Checking erased sector(s)...\n"));
3169 flash_safe_acquire(fmc
->mtd
);
3171 for (; pos
< end
; pos
+= 4) {
3172 if (*(__u32
*)pos
!= JFFS_EMPTY_BITMASK
) {
3173 printk("JFFS: Erase failed! pos = 0x%lx\n",
3175 jffs_hexdump(fmc
->mtd
, pos
,
3176 jffs_min(256, end
- pos
));
3182 flash_safe_release(fmc
->mtd
);
3185 D2(printk("JFFS: Erase succeeded.\n"));
3188 /* XXX: Here we should allocate the memory
3189 with jffs_fmalloced() in order to prevent
3190 JFFS from using this area accidentally. */
3196 /* Update the flash memory data structures. */
3197 jffs_sync_erase(fmc
, erase_size
);
3203 /* There are different criteria that should trigger a garbage collect:
3205 1. There is too much dirt in the memory.
3206 2. The free space is becoming small.
3207 3. There are many versions of a node.
3209 The garbage collect should always be done in a manner that guarantees
3210 that future garbage collects cannot be locked. E.g. Rewritten chunks
3211 should not be too large (span more than one sector in the flash memory
3212 for exemple). Of course there is a limit on how intelligent this garbage
3213 collection can be. */
3217 jffs_garbage_collect_now(struct jffs_control
*c
)
3219 struct jffs_fmcontrol
*fmc
= c
->fmc
;
3223 D2(printk("***jffs_garbage_collect_now(): fmc->dirty_size = %u, fmc->free_size = 0x%x\n, fcs1=0x%x, fcs2=0x%x",
3224 fmc
->dirty_size
, fmc
->free_size
, jffs_free_size1(fmc
), jffs_free_size2(fmc
)));
3225 D2(jffs_print_fmcontrol(fmc
));
3227 // down(&fmc->gclock);
3229 /* If it is possible to garbage collect, do so. */
3231 while (erased
== 0) {
3232 D1(printk("***jffs_garbage_collect_now(): round #%u, "
3233 "fmc->dirty_size = %u\n", i
++, fmc
->dirty_size
));
3234 D2(jffs_print_fmcontrol(fmc
));
3236 if ((erased
= jffs_try_to_erase(c
)) < 0) {
3237 printk(KERN_WARNING
"JFFS: Error in "
3238 "garbage collector.\n");
3245 if (fmc
->free_size
== 0) {
3247 printk(KERN_ERR
"jffs_garbage_collect_now(): free_size == 0. This is BAD.\n");
3252 if (fmc
->dirty_size
< fmc
->sector_size
) {
3253 /* Actually, we _may_ have been able to free some,
3254 * if there are many overlapping nodes which aren't
3255 * actually marked dirty because they still have
3256 * some valid data in each.
3262 /* Let's dare to make a garbage collect. */
3263 if ((result
= jffs_garbage_collect_next(c
)) < 0) {
3264 printk(KERN_ERR
"JFFS: Something "
3265 "has gone seriously wrong "
3266 "with a garbage collect.\n");
3270 D1(printk(" jffs_garbage_collect_now(): erased: %ld\n", erased
));
3271 DJM(jffs_print_memory_allocation_statistics());
3275 // up(&fmc->gclock);
3277 D3(printk(" jffs_garbage_collect_now(): Leaving...\n"));
3279 printk("jffs_g_c_now(): erased = %ld\n", erased
);
3280 jffs_print_fmcontrol(fmc
);
3283 if (!erased
&& !result
)
3287 } /* jffs_garbage_collect_now() */
3290 /* Determine if it is reasonable to start garbage collection.
3291 We start a gc pass if either:
3292 - The number of free bytes < MIN_FREE_BYTES && at least one
3294 - The number of dirty bytes > MAX_DIRTY_BYTES
3296 static inline int thread_should_wake (struct jffs_control
*c
)
3298 D1(printk (KERN_NOTICE
"thread_should_wake(): free=%d, dirty=%d, blocksize=%d.\n",
3299 c
->fmc
->free_size
, c
->fmc
->dirty_size
, c
->fmc
->sector_size
));
3301 /* If there's not enough dirty space to free a block, there's no point. */
3302 if (c
->fmc
->dirty_size
< c
->fmc
->sector_size
) {
3303 D2(printk(KERN_NOTICE
"thread_should_wake(): Not waking. Insufficient dirty space\n"));
3307 /* If there is too much RAM used by the various structures, GC */
3308 if (jffs_get_node_inuse() > (c
->fmc
->used_size
/c
->fmc
->max_chunk_size
* 5 + jffs_get_file_count() * 2 + 50)) {
3309 /* FIXME: Provide proof that this test can be satisfied. We
3310 don't want a filesystem doing endless GC just because this
3311 condition cannot ever be false.
3313 D2(printk(KERN_NOTICE
"thread_should_wake(): Waking due to number of nodes\n"));
3317 /* If there are fewer free bytes than the threshold, GC */
3318 if (c
->fmc
->free_size
< c
->gc_minfree_threshold
) {
3319 D2(printk(KERN_NOTICE
"thread_should_wake(): Waking due to insufficent free space\n"));
3322 /* If there are more dirty bytes than the threshold, GC */
3323 if (c
->fmc
->dirty_size
> c
->gc_maxdirty_threshold
) {
3324 D2(printk(KERN_NOTICE
"thread_should_wake(): Waking due to excessive dirty space\n"));
3327 /* FIXME: What about the "There are many versions of a node" condition? */
3333 void jffs_garbage_collect_trigger(struct jffs_control
*c
)
3335 /* NOTE: We rely on the fact that we have the BKL here.
3336 * Otherwise, the gc_task could go away between the check
3337 * and the wake_up_process()
3339 if (c
->gc_task
&& thread_should_wake(c
))
3340 send_sig(SIGHUP
, c
->gc_task
, 1);
3344 /* Kernel threads take (void *) as arguments. Thus we pass
3345 the jffs_control data as a (void *) and then cast it. */
3347 jffs_garbage_collect_thread(void *ptr
)
3349 struct jffs_control
*c
= (struct jffs_control
*) ptr
;
3350 struct jffs_fmcontrol
*fmc
= c
->fmc
;
3355 daemonize("jffs_gcd");
3357 c
->gc_task
= current
;
3360 init_completion(&c
->gc_thread_comp
); /* barrier */
3361 spin_lock_irq(¤t
->sighand
->siglock
);
3362 siginitsetinv (¤t
->blocked
, sigmask(SIGHUP
) | sigmask(SIGKILL
) | sigmask(SIGSTOP
) | sigmask(SIGCONT
));
3363 recalc_sigpending();
3364 spin_unlock_irq(¤t
->sighand
->siglock
);
3366 D1(printk (KERN_NOTICE
"jffs_garbage_collect_thread(): Starting infinite loop.\n"));
3370 /* See if we need to start gc. If we don't, go to sleep.
3372 Current implementation is a BAD THING(tm). If we try
3373 to unmount the FS, the unmount operation will sleep waiting
3374 for this thread to exit. We need to arrange to send it a
3375 sig before the umount process sleeps.
3378 if (!thread_should_wake(c
))
3379 set_current_state (TASK_INTERRUPTIBLE
);
3381 schedule(); /* Yes, we do this even if we want to go
3382 on immediately - we're a low priority
3385 /* Put_super will send a SIGKILL and then wait on the sem.
3387 while (signal_pending(current
)) {
3389 unsigned long signr
= 0;
3391 if (try_to_freeze())
3394 spin_lock_irq(¤t
->sighand
->siglock
);
3395 signr
= dequeue_signal(current
, ¤t
->blocked
, &info
);
3396 spin_unlock_irq(¤t
->sighand
->siglock
);
3400 D1(printk("jffs_garbage_collect_thread(): SIGSTOP received.\n"));
3401 set_current_state(TASK_STOPPED
);
3406 D1(printk("jffs_garbage_collect_thread(): SIGKILL received.\n"));
3408 complete_and_exit(&c
->gc_thread_comp
, 0);
3413 D1(printk (KERN_NOTICE
"jffs_garbage_collect_thread(): collecting.\n"));
3415 D3(printk (KERN_NOTICE
"g_c_thread(): down biglock\n"));
3416 mutex_lock(&fmc
->biglock
);
3418 D1(printk("***jffs_garbage_collect_thread(): round #%u, "
3419 "fmc->dirty_size = %u\n", i
++, fmc
->dirty_size
));
3420 D2(jffs_print_fmcontrol(fmc
));
3422 if ((erased
= jffs_try_to_erase(c
)) < 0) {
3423 printk(KERN_WARNING
"JFFS: Error in "
3424 "garbage collector: %ld.\n", erased
);
3430 if (fmc
->free_size
== 0) {
3431 /* Argh. Might as well commit suicide. */
3432 printk(KERN_ERR
"jffs_garbage_collect_thread(): free_size == 0. This is BAD.\n");
3433 send_sig(SIGQUIT
, c
->gc_task
, 1);
3438 /* Let's dare to make a garbage collect. */
3439 if ((result
= jffs_garbage_collect_next(c
)) < 0) {
3440 printk(KERN_ERR
"JFFS: Something "
3441 "has gone seriously wrong "
3442 "with a garbage collect: %d\n", result
);
3446 D3(printk (KERN_NOTICE
"g_c_thread(): up biglock\n"));
3447 mutex_unlock(&fmc
->biglock
);
3449 } /* jffs_garbage_collect_thread() */