super.c
No OneTemporary
Actions

Size

47 KB

Referenced Files

None

Subscribers

None

super.c
View Options

	#include <unistd.h>
	#include <stdio.h>
	#include <stdint.h>
	#include <errno.h>

	#include "osi_list.h"
	#include "osi_user.h"
	#include "bio.h"
	#include "util.h"
	#include "file.h"
	#include "rgrp.h"
	#include "fsck.h"
	#include "ondisk.h"
	#include "super.h"
	#include "fsck_incore.h"

	#ifndef BLKGETSIZE64
	#define BLKGETSIZE64 _IOR(0x12, 114, size_t)
	#endif

	#define DIV_RU(x, y) (((x) + (y) - 1) / (y))
	#define ri_compare(rg, ondisk, expected, field, fmt) \
	if (ondisk.field != expected.field) { \
	log_warn("rgindex #%d " #field " discrepancy: index 0x%" fmt \
	" != expected: 0x%" fmt "\n", \
	rg + 1, ondisk.field, expected.field); \
	ondisk.field = expected.field; \
	rgindex_modified = TRUE; \
	}

	static uint64 total_journal_space;

	/**
	* check_sb - Check superblock
	* @sdp: the filesystem
	* @sb: The superblock
	*
	* Checks the version code of the FS is one that we understand how to
	* read and that the sizes of the various on-disk structures have not
	* changed.
	*
	* Returns: 0 on success, -1 on failure
	*/
	static int check_sb(struct fsck_sb sdp, struct gfs_sb sb)
	{
	int error = 0;
	if (sb->sb_header.mh_magic != GFS_MAGIC \|\|
	sb->sb_header.mh_type != GFS_METATYPE_SB){
	log_crit("Either the super block is corrupted, or this "
	"is not a GFS filesystem\n");
	log_debug("Header magic: %X Header Type: %X\n",
	sb->sb_header.mh_magic,
	sb->sb_header.mh_type);
	error = -EINVAL;
	goto out;
	}

	/* If format numbers match exactly, we're done. */
	if (sb->sb_fs_format != GFS_FORMAT_FS \|\|
	sb->sb_multihost_format != GFS_FORMAT_MULTI){
	log_warn("Old file system detected.\n");
	}

	out:
	return error;
	}


	/*
	* read_sb: read the super block from disk
	* sdp: in-core super block
	*
	* This function reads in the super block from disk and
	* initializes various constants maintained in the super
	* block
	*
	* Returns: 0 on success, -1 on failure.
	*/
	int read_sb(struct fsck_sb *sdp)
	{
	osi_buf_t *bh;
	uint64 space = 0;
	unsigned int x;
	int error;
	error = get_and_read_buf(sdp, GFS_SB_ADDR >> sdp->fsb2bb_shift, &bh, 0);
	if (error){
	log_crit("Unable to read superblock\n");
	goto out;
	}

	gfs_sb_in(&sdp->sb, BH_DATA(bh));

	relse_buf(sdp, bh);

	error = check_sb(sdp, &sdp->sb);
	if (error)
	goto out;

	/* FIXME: Need to verify all this */
	/* FIXME: What's this 9? */
	sdp->fsb2bb_shift = sdp->sb.sb_bsize_shift - 9;
	sdp->diptrs =
	(sdp->sb.sb_bsize - sizeof(struct gfs_dinode)) /
	sizeof(uint64);
	sdp->inptrs =
	(sdp->sb.sb_bsize - sizeof(struct gfs_indirect)) /
	sizeof(uint64);
	sdp->jbsize = sdp->sb.sb_bsize - sizeof(struct gfs_meta_header);
	/* FIXME: Why is this /2 */
	sdp->hash_bsize = sdp->sb.sb_bsize / 2;
	sdp->hash_ptrs = sdp->hash_bsize / sizeof(uint64);
	sdp->heightsize[0] = sdp->sb.sb_bsize -
	sizeof(struct gfs_dinode);
	sdp->heightsize[1] = sdp->sb.sb_bsize * sdp->diptrs;
	for (x = 2; ; x++){
	space = sdp->heightsize[x - 1] * sdp->inptrs;
	/* FIXME: Do we really need this first check?? */
	if (space / sdp->inptrs != sdp->heightsize[x - 1] \|\|
	space % sdp->inptrs != 0)
	break;
	sdp->heightsize[x] = space;
	}
	sdp->max_height = x;
	if(sdp->max_height > GFS_MAX_META_HEIGHT){
	log_err("Bad max metadata height.\n");
	error = -1;
	goto out;
	}

	sdp->jheightsize[0] = sdp->sb.sb_bsize -
	sizeof(struct gfs_dinode);
	sdp->jheightsize[1] = sdp->jbsize * sdp->diptrs;
	for (x = 2; ; x++){
	space = sdp->jheightsize[x - 1] * sdp->inptrs;
	if (space / sdp->inptrs != sdp->jheightsize[x - 1] \|\|
	space % sdp->inptrs != 0)
	break;
	sdp->jheightsize[x] = space;
	}
	sdp->max_jheight = x;
	if(sdp->max_jheight > GFS_MAX_META_HEIGHT){
	log_err("Bad max jheight.\n");
	error = -1;
	}

	out:

	return error;
	}


	/*
	* ji_update - fill in journal info
	* ip: the journal index inode
	*
	* Given the inode for the journal index, read in all
	* the journal indexes.
	*
	* Returns: 0 on success, -1 on failure
	*/
	int ji_update(struct fsck_sb *sdp)
	{
	struct fsck_inode *ip = sdp->jiinode;
	char buf[sizeof(struct gfs_jindex)];
	unsigned int j;
	int error=0;


	if(ip->i_di.di_size % sizeof(struct gfs_jindex) != 0){
	log_err("The size reported in the journal index"
	" inode is not a\n"
	"\tmultiple of the size of a journal index.\n");
	return -1;
	}

	if(!(sdp->jindex = (struct gfs_jindex *)malloc(ip->i_di.di_size))) {
	log_err("Unable to allocate journal index\n");
	return -1;
	}
	if(!memset(sdp->jindex, 0, ip->i_di.di_size)) {
	log_err("Unable to zero journal index\n");
	return -1;
	}
	total_journal_space = 0;

	for (j = 0; ; j++) {
	struct gfs_jindex *journ;
	error = readi(ip, buf, j * sizeof(struct gfs_jindex),
	sizeof(struct gfs_jindex));
	if(!error)
	break;
	if (error != sizeof(struct gfs_jindex)){
	log_err("An error occurred while reading the"
	" journal index file.\n");
	goto fail;
	}

	journ = sdp->jindex + j;
	gfs_jindex_in(journ, buf);
	total_journal_space += journ->ji_nsegment * sdp->sb.sb_seg_size;
	}


	if(j * sizeof(struct gfs_jindex) != ip->i_di.di_size){
	log_err("journal inode size invalid\n");
	log_debug("j * sizeof(struct gfs_jindex) !="
	" ip->i_di.di_size\n");
	log_debug("%d != %d\n",
	j * sizeof(struct gfs_jindex), ip->i_di.di_size);
	goto fail;
	}
	sdp->journals = j;
	log_debug("%d journals found.\n", j);

	return 0;

	fail:
	free(sdp->jindex);
	return -1;
	}

	/* Print out debugging information in same format as gfs_edit. */
	int hexdump(uint64 startaddr, const unsigned char *lpBuffer, int len)
	{
	const unsigned char pointer, ptr2;
	int i;
	uint64 l;

	pointer = (unsigned char *)lpBuffer;
	ptr2 = (unsigned char *)lpBuffer;
	l = 0;
	while (l < len) {
	log_info("%.8" PRIX64, startaddr + l);
	for (i = 0; i < 16; i++) { /* first print it in hex */
	if (i % 4 == 0)
	log_info(" ");
	log_info("%02X", *pointer);
	pointer++;
	}
	log_info(" [");
	for (i = 0; i < 16; i++) { /* now print it in character format */
	if ((ptr2 >= ' ') && (ptr2 <= '~'))
	log_info("%c", *ptr2);
	else
	log_info(".");
	ptr2++;
	}
	log_info("] \n");
	l += 16;
	}
	return (len);
	}


	/**
	* rgrplength2bitblocks - Stolen from gfs_mkfs.
	*
	* @sdp: the superblock
	* @length: the number of blocks in a RG
	*
	* Give a number of blocks in a RG, figure out the number of blocks
	* needed for bitmaps.
	*
	* Returns: the number of bitmap blocks
	*/

	uint32 rgrplength2bitblocks(struct fsck_sb *sdp, uint32 length)
	{
	uint32 bitbytes;
	uint32 old_blocks = 0, blocks;
	int tries = 0;

	for (;;) {
	bitbytes = (length - old_blocks) / GFS_NBBY;
	blocks = 1;

	if (bitbytes > sdp->sb.sb_bsize - sizeof(struct gfs_rgrp)) {
	bitbytes -= sdp->sb.sb_bsize - sizeof(struct gfs_rgrp);
	blocks += DIV_RU(bitbytes, (sdp->sb.sb_bsize -
	sizeof(struct gfs_meta_header)));
	}
	if (blocks == old_blocks)
	break;
	old_blocks = blocks;
	if (tries++ > 10) {
	blocks = 0;
	break;
	}
	}
	return blocks;
	}

	/*
	* gfs_rgindex_rebuild - rebuild a corrupt Resource Group (RG) index manually
	* where trust_lvl == distrust
	*
	* If this routine is called, it means we have RGs in odd/unexpected places,
	* and there is a corrupt RG. In other words, we can't trust the RG index
	* is completely sane, and the RGs don't fit on nice neat fs boundaries.
	* So we have no choice but to go through the count them by hand.
	* We've tried twice to recover the RGs and RG index, and failed. This is
	* our last chance to remedy the situation.
	*
	* In some cases, we have no choice but to trust the rgindex file. Since
	* it is completely hidden from the users, if we find an inconsistency,
	* it's safer to assume the index is correct and the RG is corrupt rather
	* than the RG is correct and the index is bad. This routine goes to great
	* lengths to determine which is the case and figure it out regardless.
	*
	* This routine tries to minimize performance impact by:
	* 1. Skipping through the filesystem at known increments when possible.
	* 2. Shuffle through every block when RGs are not found at the predicted
	* locations.
	*
	* Note: A GFS filesystem is built by gfs_mkfs into several "subdevices."
	* These are just logical divisions of the logical volume. The gfs_mkfs
	* program considers two types of subdevices: RG-subdevices and journal
	* subdevices. RG subdevices contain one or more RGs. Journal subdevices
	* contain one or more journals. For the purposes of gfs_fsck, when I talk
	* about subdevices, I'm talking about RG-subdevices only. For a freshly
	* created GFS filesystem, the logical volume will be broken apart like this:
	*
	* RG-subdevice 0: n Resource Groups
	* Journal subdevice
	* RG-subdevice 1: n Resource Groups (same as RG-subdevice 0)
	*
	* If the filesystem has been resized via gfs_grow, there will be more
	* RG-subdevices containing more RGs. However, gfs_fsck treats them all
	* as "subdevice 2."
	*
	* NOTE: When giving messages to the users, I am referring to them as
	* "sections" 1, 2, and 3 because "subdevice" sounds too confusing.
	*
	* If an RG is not found at a predicted location, it either means that
	* there is a corrupted RG, or else the RG has been added after the fact
	* by gfs_grow. We can only predict the locations for RGs within a subdevice,
	* but the subdevice boundaries are not that predictable. (Actually, they
	* are, but since we're dealing with a likely corrupt filesystem, I don't
	* want to rely on good data too much to do it this way.)
	*
	* I am, however, going to rely on the fact that the first original subdevice
	* will have the same number of RGs as the second original subdevice.
	* Other RGs found after that will be considered "extra."
	*/
	int gfs_rgindex_rebuild(struct fsck_sb sdp, osi_list_t ret_list,
	unsigned int *num_rgs)
	{
	osi_buf_t bh; / buffer handle */
	uint64 subdevice_size, fs_total_size;
	int number_of_rgs; /* #RGs this subdevice.
	min of 2 per segment * 2 segments = 4 */
	int rg_number; /* real RG number (0 - x) */
	int subd;
	int error, corrupt_rgs;
	int rgi, rgs_per_subd;
	uint64 blok, block_of_last_rg;
	uint64 block_bump;
	uint64 shortest_dist_btwn_rgs[2]; /* one for each subdevice */
	uint64 first_rg_dist[2], initial_first_rg_dist[2];
	struct fsck_rgrp calc_rgd, prev_rgd;
	struct gfs_rgrp tmp_rgrp;
	osi_list_t *tmp;
	int rg_was_fnd = FALSE;
	struct gfs_rindex buf, tmpndx;
	uint64 fs_size_from_rgindex = 0;
	int index_entries_per_subd = 0, subd_ndx_entry, rg;
	uint64_t last_known_ri_addr = 0, prev_known_ri_addr = 0;
	uint32_t last_known_ri_length = 0;
	uint32_t last_known_ri_data = 0;

	osi_list_init(ret_list);
	*num_rgs = 0;
	/* Get the total size of the device */
	error = ioctl(sdp->diskfd, BLKGETSIZE64,
	&fs_total_size); /* Size in bytes */
	fs_total_size /= sdp->sb.sb_bsize;
	log_debug("fs_total_size = 0x%" PRIX64 " blocks.\n", fs_total_size);
	block_of_last_rg = 0;
	subdevice_size = 0;
	rgs_per_subd = 0;
	/* ----------------------------------------------------------------- */
	/* First, figure out the exact end of the second subdevice. */
	/* That will tell us where the third RG-subdevice should start. */
	/* We need to keep track of how many entries are in the index before */
	/* we hit the journal blocks. That will tell us how many index */
	/* entries will be in the first subdevice, and the second subdevice */
	/* should have the same number. After that, we don't care. */
	/* Note: we're using values in the rgindex, even though we don't */
	/* trust it. We're relatively okay because we're just trying to */
	/* find the highest RG value for the second subdevice. */
	/* ----------------------------------------------------------------- */
	subd = 0;
	index_entries_per_subd = 0;
	subd_ndx_entry = 0;
	for (rg = 0; ; rg++) {
	uint64 end_of_rg;

	error = readi(sdp->riinode,
	(char )&buf, rg sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (!error) /* if end of file */
	break; /* stop looking */
	gfs_rindex_in(&tmpndx, (char )&buf); / read in the index */
	subd_ndx_entry++;
	if (!subd) { /* if we're still in the first subdevice */
	if (tmpndx.ri_addr >= sdp->jindex->ji_addr + total_journal_space) {
	subd++; /* this rgindex belongs to the second subdevice */
	prev_known_ri_addr = 0;
	last_known_ri_addr = 0;
	subd_ndx_entry = 1;
	}
	else {
	index_entries_per_subd++;
	/* Check if this is the last index entry for subdevice */
	if (tmpndx.ri_addr + tmpndx.ri_length + tmpndx.ri_data >=
	sdp->jindex->ji_addr - GFS_NBBY) {
	subd++; /* NEXT rgindex belongs to the second subdevice */
	subd_ndx_entry = 0;
	prev_known_ri_addr = 0;
	last_known_ri_addr = 0;
	continue;
	}
	}
	}
	end_of_rg = tmpndx.ri_addr + tmpndx.ri_length + tmpndx.ri_data;
	/* ----------------------------------------------------------------- */
	/* Make sure the rgindex looks relatively sane. After all, */
	/* at this stage of the game, we don't trust it. */
	/* ----------------------------------------------------------------- */
	if (subd && end_of_rg > sdp->jindex->ji_addr + total_journal_space &&
	end_of_rg <= fs_total_size) { /* looks relatively sane */
	/* Save some data values we can fall back on: */
	prev_known_ri_addr = last_known_ri_addr;
	last_known_ri_addr = tmpndx.ri_addr;
	last_known_ri_length = tmpndx.ri_length;
	last_known_ri_data = tmpndx.ri_data;
	if (fs_size_from_rgindex < end_of_rg)
	fs_size_from_rgindex = end_of_rg;
	/* Quit after we hit the same number of entries as 1st subdevice */
	if (subd_ndx_entry >= index_entries_per_subd)
	break;
	}
	else if (!subd && end_of_rg < sdp->jindex->ji_addr &&
	end_of_rg > 0) { /* looks relatively sane */
	/* Save some data values we can fall back on: */
	prev_known_ri_addr = last_known_ri_addr;
	last_known_ri_addr = tmpndx.ri_addr;
	last_known_ri_length = tmpndx.ri_length;
	last_known_ri_data = tmpndx.ri_data;
	if (fs_size_from_rgindex < end_of_rg)
	fs_size_from_rgindex = end_of_rg;
	}
	else { /* Otherwise we have a corrupt index entry */
	log_debug("Likely damage to rgindex entry %d.\n",
	subd_ndx_entry + (subd * index_entries_per_subd));
	if (prev_known_ri_addr) {
	/* Try to extrapolate from the previous one */
	tmpndx.ri_addr = last_known_ri_addr +
	(last_known_ri_addr - prev_known_ri_addr);
	tmpndx.ri_length = last_known_ri_length;
	tmpndx.ri_data = last_known_ri_data;
	log_debug("Extrapolating addr=0x%" PRIx64 ", length=0x%x, "
	"data=%x\n", tmpndx.ri_addr,
	tmpndx.ri_length, tmpndx.ri_data);
	end_of_rg = tmpndx.ri_addr + tmpndx.ri_length +
	tmpndx.ri_data;
	if (end_of_rg > sdp->jindex->ji_addr +
	total_journal_space &&
	end_of_rg <= fs_total_size) { /* looks relatively okay */
	/* Adjust data values we can fall back on: */
	last_known_ri_addr = tmpndx.ri_addr;
	if (fs_size_from_rgindex < end_of_rg)
	fs_size_from_rgindex = end_of_rg;
	/* Quit after we hit the same number of entries as
	the first subdevice/section */
	if (subd_ndx_entry >= index_entries_per_subd)
	break;
	}
	} /* if we have a good previous */
	else {
	log_debug("Not enough data to figure it out--skipped.\n");
	}
	} /* corrupt RG index entry */
	} /* for all RGs in the index */
	log_debug("Index entries/section=%d, third section addr = 0x%"PRIx64"\n",
	index_entries_per_subd, fs_size_from_rgindex);
	initial_first_rg_dist[0] = first_rg_dist[0] = sdp->jindex->ji_addr -
	((GFS_SB_ADDR >> sdp->fsb2bb_shift) + 1);
	initial_first_rg_dist[1] = first_rg_dist[1] = sdp->jindex->ji_addr;
	/* ----------------------------------------------------------------- */
	/* Now let's figure out the space between RGs for the first subd, */
	/* and the second subd. Subsequent RGs will be unpredictable. */
	/* We need to know the distance between RGs because if one is */
	/* corrupt or overwritten, we need to salvage it at the correct */
	/* location. For example, if RG #2 is nuked, at first glance, it */
	/* appears as if our RGs are twice as far apart as they should be. */
	/* So we should chase down a couple to get more than one opinion. */
	/* If several RGs are nuked, sorry, I'll only go so far to recover. */
	/* This check will be slower because we have to read blocks 1 by 1. */
	/* Luckily, we only have to do a few this way. */
	/* Later, we can bump ahead by the amount we find here. */
	/* ----------------------------------------------------------------- */
	subdevice_size = sdp->jindex->ji_addr; /* addr of first journal */;
	for (subd = 0; subd < 2; subd++) {
	uint64 start_block;

	if (!subd)
	start_block = (GFS_SB_ADDR >> sdp->fsb2bb_shift) + 1;
	else
	start_block = sdp->jindex->ji_addr + total_journal_space;
	block_of_last_rg = start_block;
	number_of_rgs = 0;
	shortest_dist_btwn_rgs[subd] = subdevice_size;
	for (blok = start_block; blok < fs_total_size; blok++) {
	error = get_and_read_buf(sdp, blok, &bh, 0);
	if (error){
	log_crit("Unable to read block 0x%" PRIX64 "\n", blok);
	return -1;
	}
	if ((blok == start_block) \|\| /* If first RG block or */
	!check_type(bh, GFS_METATYPE_RG)) { /* we found an RG */
	log_debug("%d:RG found at block 0x%" PRIx64 "\n", subd + 1,
	blok);
	/* If we spilled into the next subdevice, quit. */
	if (blok + GFS_NBBY >= start_block + subdevice_size) {
	log_debug("This is in the next subdevice--skipping.\n");
	break;
	}
	gfs_rgrp_in(&tmp_rgrp, BH_DATA(bh));
	if (blok == start_block) {
	shortest_dist_btwn_rgs[subd] = subdevice_size;
	log_debug("Start of section %d.\n", subd + 1);
	}
	else {
	uint64 rgdist;

	rgdist = blok - block_of_last_rg;
	log_debug("%d:dist 0x%" PRIx64 " = 0x% " PRIx64
	" - 0x%" PRIx64, subd + 1, rgdist,
	blok, block_of_last_rg);
	/* ----------------------------------------------------- */
	/* We found another RG. Check to see if we need to set */
	/* the first_rg_dist based on whether it's still at its */
	/* initial value (i.e. the whole subdevice size). */
	/* The first rg distance is different from the rest */
	/* because of the superblock and 64K dead space */
	/* ----------------------------------------------------- */
	if (first_rg_dist[subd] == initial_first_rg_dist[subd])
	first_rg_dist[subd] = rgdist;
	if (rgdist < shortest_dist_btwn_rgs[subd])
	{
	shortest_dist_btwn_rgs[subd] = rgdist;
	log_debug("(shortest so far)\n");
	}
	else
	log_debug("\n");
	}
	number_of_rgs++; /* number of RGs this subdevice */
	/* --------------------------------------------------------- */
	/* Check to see if we're the last RG we want to examine. */
	/* If so, forget checking the next index entry and exit. */
	/* (The next index entry may be for the next RG anyway). */
	/* --------------------------------------------------------- */
	if (number_of_rgs >= 4 \|\|
	number_of_rgs >= index_entries_per_subd)
	break;
	/* --------------------------------------------------------- */
	/* Read in the index entry for the NEXT RG in line and */
	/* compare the RG size difference with what we know. */
	/* --------------------------------------------------------- */
	rg_number = number_of_rgs + (subd * index_entries_per_subd);
	error = readi(sdp->riinode, (char *)&buf,
	rg_number * sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (error) { /* if we read some data (no error really) */
	gfs_rindex_in(&tmpndx, (char *)&buf);
	if (tmpndx.ri_addr > start_block &&
	tmpndx.ri_addr < fs_total_size &&
	tmpndx.ri_addr != blok &&
	tmpndx.ri_addr - blok <
	shortest_dist_btwn_rgs[subd]) {
	shortest_dist_btwn_rgs[subd] =
	tmpndx.ri_addr - blok;
	log_debug("Section %d RG %d(%d): shortest=0x%"PRIx64
	"\n", subd + 1, number_of_rgs, rg_number,
	shortest_dist_btwn_rgs[subd]);
	}
	}
	block_of_last_rg = blok;
	/* --------------------------------------------------------- */
	/* We can't just check every block because some of the files */
	/* in the fs (i.e. inside an RG) might have data that looks */
	/* exactly like a valid RG. Sounds farfetched, but it's not, */
	/* based on my own experiences. */
	/* In my experience, the RG locations are spaced differently */
	/* from their used and free space numbers because of the way */
	/* gfs_mkfs puts them. In other words, the RG locations */
	/* according to the index will be different from the sum of */
	/* the space they take. Why? I don't know, but maybe it */
	/* has to do with the variable size of bitmaps. */
	/* At any rate, used+free can get us close, but not exact. */
	/* Therefore, we have to search for the RG after that. */
	/* --------------------------------------------------------- */
	if (!error &&
	((tmp_rgrp.rg_useddi + tmp_rgrp.rg_free + 1) >> 2) ==
	(tmpndx.ri_addr >> 2)) {
	blok = tmpndx.ri_addr - 1; /* go by the index */
	log_debug("I(0x%" PRIx64 ")\n", blok);
	}
	else {
	blok += tmp_rgrp.rg_useddi + tmp_rgrp.rg_free;
	log_debug("R(0x%" PRIx64 ")\n", blok);
	}
	} /* If first RG block or RG */
	relse_buf(sdp, bh); /* release the read buffer */
	} /* for blok */
	/* -------------------------------------------------------------- */
	/* Sanity-check our first_rg_dist. If RG #2 got nuked, the */
	/* first_rg_dist would measure from #1 to #3, which would be bad. */
	/* We need to take remedial measures to fix it (from the index). */
	/* -------------------------------------------------------------- */
	if (first_rg_dist[subd] >= shortest_dist_btwn_rgs[subd] +
	(shortest_dist_btwn_rgs[subd] / 4)) {
	log_debug("%d:Shortest dist is: 0x%" PRIx64 "\n", subd + 1,
	shortest_dist_btwn_rgs[subd]);
	/* read in the second RG index entry for this subd. */
	readi(sdp->riinode, (char *)&buf,
	(1 + (subd * index_entries_per_subd)) *
	sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	gfs_rindex_in(&tmpndx, (char *)&buf);
	if (tmpndx.ri_addr > start_block) { /* sanity check */
	log_warn("RG %d is damaged: recomputing RG dist from index: ",
	2 + (subd * index_entries_per_subd));
	first_rg_dist[subd] = tmpndx.ri_addr - start_block;
	log_warn("0x%" PRIx64 "\n", first_rg_dist[subd]);
	}
	else {
	log_warn("RG index %d is damaged: extrapolating RG dist: ",
	2 + (subd * index_entries_per_subd));
	first_rg_dist[subd] = (subdevice_size - start_block) %
	((index_entries_per_subd - 1) *
	shortest_dist_btwn_rgs[subd]);
	log_warn("0x%" PRIx64 "\n", first_rg_dist[subd]);
	}
	} /* if first RG distance is within tolerance */
	log_debug("First RG distance: 0x%" PRIx64 "\n", first_rg_dist[subd]);
	log_debug("Section %d: distance between RGs: 0x%" PRIx64 "\n",
	subd + 1, shortest_dist_btwn_rgs[subd]);
	log_debug("Section size: 0x%" PRIx64 "\n", subdevice_size);
	} /* for subd */
	number_of_rgs = 0; /* reset this because it is reused below */
	/* ----------------------------------------------------------------- */
	/* Start reading the filesystem starting with the block after the */
	/* superblock, which should be the first RG. */
	/* The problem is that gfs_grow puts the RGs at unpredictable */
	/* locations. If the fs was only grown once, that would be */
	/* predictable. But if it grows twice, by different amounts, then */
	/* our RGs could be anywhere. After carefully studying the problem */
	/* I've determined that the best thing we can do is to trust the */
	/* rgindex and hope to God it's correct. That's the only way we're */
	/* going to be able to recover RGs in the third section. */
	/* ----------------------------------------------------------------- */
	prev_rgd = NULL;
	block_bump = first_rg_dist[0];
	corrupt_rgs = 0;
	for (subd = 0; subd < 3; subd++) { /* third subdevice is for all RGs
	extended past the normal 2 with
	gfs_grow, etc. */
	uint64 start_block, end_block = 0;

	if (subd == 0) {
	start_block = (GFS_SB_ADDR >> sdp->fsb2bb_shift) + 1;
	end_block = subdevice_size - 1;
	}
	else if (subd == 1) {
	start_block = sdp->jindex->ji_addr + total_journal_space;
	/* Moral dilemma: should we go to the last block or should */
	/* we trust the index? If they're close to one another, */
	/* let's use the index. */
	if ((fs_size_from_rgindex >> 2) ==
	((start_block + subdevice_size - 1) >> 2)) /* if we're close */
	end_block = fs_size_from_rgindex - 1; /* trust the index */
	else /* otherwise */
	end_block = start_block + subdevice_size - 1; /* go to end */
	}
	else {
	start_block = end_block + 1;
	end_block = fs_total_size;
	if (start_block + GFS_NBBY >= end_block)
	break;
	}
	log_warn("Section %d: 0x%" PRIx64 " - 0x%" PRIx64 "\n", subd + 1,
	start_block, end_block);
	for (blok = start_block; blok <= end_block; blok += block_bump) {
	uint64 fwd_block;
	int bitmap_was_fnd;

	log_debug("Block 0x%" PRIx64 "\n", blok);
	error = get_and_read_buf(sdp, blok, &bh, 0);
	if (error) {
	log_crit("Unable to read block 0x%" PRIX64 "\n", blok);
	return -1;
	}
	rg_was_fnd = (!check_type(bh, GFS_METATYPE_RG));
	relse_buf(sdp, bh); /* release the read buffer */
	/* ------------------------------------------------------------- */
	/* For the first and second subdevice, we know the RG size. */
	/* Since we're bumping by that amount, this better be an RG. */
	/* ------------------------------------------------------------- */
	/* Allocate a new RG and index. */
	calc_rgd = (struct fsck_rgrp *)malloc(sizeof(struct fsck_rgrp));
	memset(calc_rgd, 0, sizeof(struct fsck_rgrp));
	calc_rgd->rd_sbd = sdp; /* hopefully this is not used */
	osi_list_add_prev(&calc_rgd->rd_list, ret_list);
	calc_rgd->rd_ri.ri_length = 1;
	calc_rgd->rd_ri.ri_addr = blok;
	if (!rg_was_fnd) { /* if not an RG */
	/* ----------------------------------------------------- */
	/* This SHOULD be an RG but isn't. */
	/* ----------------------------------------------------- */
	corrupt_rgs++;
	if (corrupt_rgs < 5)
	log_debug("Missing or damaged RG at block 0x%" PRIx64 \
	"\n", blok);
	else {
	log_crit("Error: too many bad RGs.\n");
	return -1;
	}
	}
	/* ------------------------------------------------ */
	/* Now go through and count the bitmaps for this RG */
	/* ------------------------------------------------ */
	bitmap_was_fnd = FALSE;
	for (fwd_block = blok + 1; fwd_block < fs_total_size;
	fwd_block++) {
	error = get_and_read_buf(sdp, fwd_block, &bh, 0);
	if (error){
	log_crit("Unable to read block 0x%" PRIX64 "\n",
	fwd_block);
	return -1;
	}
	bitmap_was_fnd = (!check_type(bh, GFS_METATYPE_RB));
	relse_buf(sdp, bh);
	if (bitmap_was_fnd) /* if a bitmap */
	calc_rgd->rd_ri.ri_length++;
	else
	break; /* end of bitmap, so call it quits. */
	} /* for subsequent bitmaps */
	calc_rgd->rd_ri.ri_data1 = calc_rgd->rd_ri.ri_addr +
	calc_rgd->rd_ri.ri_length;
	if (prev_rgd) {
	prev_rgd->rd_ri.ri_data = block_bump -
	rgrplength2bitblocks(sdp, block_bump);
	prev_rgd->rd_ri.ri_data -= prev_rgd->rd_ri.ri_data %
	GFS_NBBY;
	prev_rgd->rd_ri.ri_bitbytes = prev_rgd->rd_ri.ri_data /
	GFS_NBBY;
	log_debug("Prev ri_data set to: %" PRIx32 ".\n",
	prev_rgd->rd_ri.ri_data);
	/prev_rgd->rd_ri.ri_data = block_bump;/
	}
	number_of_rgs++;
	log_warn("%c RG %d at block 0x%" PRIX64 " %s",
	(rg_was_fnd ? ' ' : '*'), number_of_rgs, blok,
	(rg_was_fnd ? "intact" : "* DAMAGED *"));
	rgs_per_subd++;
	prev_rgd = calc_rgd;
	block_of_last_rg = blok;
	if (subd == 2) { /* if beyond the normal RGs into gfs_grow RGs */
	/* -------------------------------------------------------- */
	/* RG location is rounded down to the nearest multiple of */
	/* GFS_NBBY, so RG location is only known within a 4 block */
	/* range. It's better to use the rgindex to figure out */
	/* the address of the next RG and bump by the difference. */
	/* However, there's another complication: gfs_grow has */
	/* been known to add RGs to the index in a non-ascending */
	/* order. Therefore, we can't assume the Nth entry in the */
	/* index corresponds to the Nth RG on disk. Wish it was. */
	/* Instead, we have to read all of the rgindex until we */
	/* find an entry that has the smallest address greater than */
	/* the block we're on (blok). */
	/* -------------------------------------------------------- */
	uint64_t rgndx_next_block;

	rgndx_next_block = end_block;
	for (rgi = 0; ; rgi++) {
	error = readi(sdp->riinode, (char *)&buf,
	rgi * sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (!error) /* if end of the rgindex */
	break; /* stop processing for more RGs */
	gfs_rindex_in(&tmpndx, (char *)&buf);
	/* if this index entry is the next RG physically */
	if (tmpndx.ri_addr > blok &&
	tmpndx.ri_addr < rgndx_next_block) {
	rgndx_next_block = tmpndx.ri_addr; /* remember it */
	}
	}
	block_bump = rgndx_next_block - blok;
	if (rgndx_next_block == end_block) { /* if no more RGs */
	log_warn(" [length 0x%" PRIx64 "]\n", block_bump);
	break; /* stop processing */
	}
	}
	else {
	if (blok == start_block)
	block_bump = first_rg_dist[subd];
	else
	block_bump = shortest_dist_btwn_rgs[subd];
	}
	if (block_bump != 1)
	log_warn(" [length 0x%" PRIx64 "]\n", block_bump);
	} /* for blocks in subdevice */
	} /* for subdevices */
	/* ------------------------------------------------------------------- */
	/* if we got to the end of the fs, we still need to fix the allocation */
	/* information for the very last RG. */
	/* ------------------------------------------------------------------- */
	if (prev_rgd && !prev_rgd->rd_ri.ri_data) {
	log_debug("Prev ri_data set to: %" PRIx32 ".\n", block_bump);
	prev_rgd->rd_ri.ri_data = block_bump -
	rgrplength2bitblocks(sdp, block_bump);
	prev_rgd->rd_ri.ri_data -= prev_rgd->rd_ri.ri_data % GFS_NBBY;
	prev_rgd->rd_ri.ri_bitbytes = prev_rgd->rd_ri.ri_data / GFS_NBBY;
	prev_rgd = NULL; /* make sure we don't use it later */
	}
	/* else No previous to fix. */
	/* ---------------------------- */
	/* Now dump out the information */
	/* ---------------------------- */
	log_debug("RG index rebuilt as follows:\n");
	for (tmp = ret_list->next, rgi = 0; tmp != ret_list;
	tmp = tmp->next, rgi++) {
	calc_rgd = osi_list_entry(tmp, struct fsck_rgrp, rd_list);
	log_debug("%d: %x / 0x%" PRIx64 " / 0x%08X / 0x%08X\n",
	rgi + 1, calc_rgd->rd_ri.ri_length, calc_rgd->rd_ri.ri_data1,
	calc_rgd->rd_ri.ri_data, calc_rgd->rd_ri.ri_bitbytes);
	/memset(rgindex_buf_ondisk, 0, sizeof(rgindex_buf_ondisk));/
	/gfs_rindex_out(&calc_rgd->rd_ri, rgindex_buf_ondisk);/
	/* Note: rgindex_buf_ondisk is ONLY used for debug to see what
	the entry would look like on disk. */
	/hexdump(rgisizeof(struct gfs_rindex), rgindex_buf_ondisk,
	sizeof(struct gfs_rindex));*/
	}
	*num_rgs = number_of_rgs;
	log_debug("Number of RGs = %d.\n", number_of_rgs);
	return 0;
	}

	/*
	* gfs_rgindex_calculate - calculate what the rgindex should look like
	* in a perfect world (trust_lvl == open_minded)
	*
	* Calculate what the rgindex should look like if no gfs_grow-like operations
	* were performed, so we can later check if all RG index entries are sane.
	*
	* This function goes in blind, assuming the entire rgindex is destroyed.
	* That way, we can rebuild it if really is trashed.
	*
	* However, this won't work if the filesystem has been extended or shrunk.
	* If the RGs aren't where we expect them to be, we have to take more drastic
	* measures to recover them.
	*
	* Assumes: journal index file is minimally sane.
	*
	* We need to check if the address and length values are okay.
	* First RG should start after the superblock at block #x11
	* Number of RGs=subdevice size / (2^rgsize/block size)
	*
	* Returns: 0 on success, -1 on failure
	* Sets: ret_list to a linked list of fsck_rgrp structs representing
	* what we think the rgindex should really look like.
	*/
	int gfs_rgindex_calculate(struct fsck_sb sdp, osi_list_t ret_list,
	unsigned int *num_rgs)
	{
	osi_buf_t bh; / buffer handle */
	uint64 subdevice_size, adjust_subdevice_size, fs_total_size;
	int number_of_rgs; /* min of 4 per segment * 2 segments = 8 */
	int rgnum_within_subdevice;
	int first_half;
	int error;
	int rgi, rgs_per_subd;
	uint64 subdevice_start;
	uint64 addr = 0, prev_addr, length = 0, prev_length;
	uint64 blocks;
	struct fsck_rgrp *calc_rgd;
	char rgindex_buf_ondisk[sizeof(struct gfs_rindex)];
	struct gfs_rindex buf, tmpndx;

	osi_list_init(ret_list);
	*num_rgs = 0;
	/* Get the total size of the device */
	error = ioctl(sdp->diskfd, BLKGETSIZE64,
	&fs_total_size); /* Size in bytes */
	fs_total_size /= sdp->sb.sb_bsize;
	log_debug("fs_total_size = 0x%" PRIX64 " blocks.\n", fs_total_size);

	/* The end of the first subdevice is also where the first journal is.*/
	subdevice_size = sdp->jindex->ji_addr; /* addr of 1st journal (blks) */
	log_debug("subdevice_size = 0x%" PRIX64 ".\n", subdevice_size);

	/* ----------------------------------------------------------------- */
	/* Read the first block of the subdevice and make sure it's an RG. */
	/* ----------------------------------------------------------------- */
	subdevice_start = fs_total_size - subdevice_size;
	error = get_and_read_buf(sdp, subdevice_start, &bh, 0);
	if (error){
	log_crit("Unable to read start of last subdevice.\n");
	return -1;
	}
	if(check_type(bh, GFS_METATYPE_RG)){
	log_warn("The middle RG is not on an even boundary (fs has grown?)\n");
	relse_buf(sdp, bh);
	return -1;
	}
	log_debug("First RG is okay.\n");
	/* --------------------------------------------------------------------- */
	/* Calculate how many RGs there are supposed to be based on the */
	/* rgindex filesize. Remember that our trust level is open-minded here. */
	/* If the filesize of the rgindex file is not a multiple of our rgindex */
	/* structures, then something's wrong and we can't trust the index. */
	/* --------------------------------------------------------------------- */
	number_of_rgs = sdp->riinode->i_di.di_size / sizeof(struct gfs_rindex);
	*num_rgs = number_of_rgs;
	log_warn("number_of_rgs = %d.\n", number_of_rgs);
	if (sdp->riinode->i_di.di_size % sizeof(struct gfs_rindex)) {
	log_warn("WARNING: rgindex file is corrupt.\n");
	return -1;
	}
	/* --------------------------------------------------------------------- */
	/* Check to see if the filesystem has been extended via gfs_grow. */
	/* If so, our assumptions will be wrong and we can't continue. */
	/* Instead, we need to progress to level 3 and dig deeper for the RGs. */
	/* We'll know if the filesystem has been extended by whether or not the */
	/* RG that's midway is on the wrong side of the journals. */
	/* For example, if we have 50 RGs, we'd expect 25 to be on one side of */
	/* the journals, and 25 to be on the other side. If we find out that */
	/* RG number 25 (index 1, or 24 index 0) is on the other side, we grew. */
	/* --------------------------------------------------------------------- */
	rgi = (number_of_rgs / 2) - 1;
	error = readi(sdp->riinode,
	(char )&buf, rgi sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (!error) { /* if end of file */
	log_warn("Error reading RG index.\n");
	return -1; /* stop looking */
	}
	gfs_rindex_in(&tmpndx, (char )&buf); / read in the index entry. */
	if (tmpndx.ri_addr >= sdp->jindex->ji_addr) { /* wrong side of journals */
	log_warn("This filesystem has probably been resized by gfs_grow.\n");
	return -1; /* stop looking */
	}
	/* --------------------------------------------------------------------- */
	/* Now that we know how many RGs there should be, we can calculate */
	/* exactly where we think they should be and build our index with it. */
	/* --------------------------------------------------------------------- */
	rgs_per_subd = (number_of_rgs / 2);
	for (rgi = 0; rgi < number_of_rgs; rgi++) {

	first_half = (rgi < rgs_per_subd ? 1 : 0);
	adjust_subdevice_size = subdevice_size;
	if (first_half) {
	adjust_subdevice_size -= ((GFS_SB_ADDR >> sdp->fsb2bb_shift) + 1);
	rgnum_within_subdevice = rgi;
	}
	else
	rgnum_within_subdevice = rgi - rgs_per_subd;
	prev_length = length;
	if (rgnum_within_subdevice)
	length = adjust_subdevice_size / rgs_per_subd;
	else
	length = adjust_subdevice_size -
	(rgs_per_subd - 1) * (adjust_subdevice_size / rgs_per_subd);

	calc_rgd = (struct fsck_rgrp *)malloc(sizeof(struct fsck_rgrp));
	memset(calc_rgd, 0, sizeof(struct fsck_rgrp));
	calc_rgd->rd_sbd = sdp; /* hopefully this is not used */
	osi_list_add_prev(&calc_rgd->rd_list, ret_list);
	prev_addr = addr;
	if (!rgnum_within_subdevice) {
	if (!rgi) {
	/* The first RG immediately follows the superblock */
	addr = (GFS_SB_ADDR >> sdp->fsb2bb_shift) + 1;
	}
	else /* First RG on second subdevice is at the beginning of it */
	addr = subdevice_start;
	}
	else
	addr = prev_addr + prev_length;
	calc_rgd->rd_ri.ri_addr = addr;
	log_debug("ri_addr[%d] = 0x%"PRIX64 " / ", rgi,
	calc_rgd->rd_ri.ri_addr);
	blocks = length - rgrplength2bitblocks(sdp, length);
	blocks -= blocks % GFS_NBBY;
	calc_rgd->rd_ri.ri_length = rgrplength2bitblocks(sdp, length);
	calc_rgd->rd_ri.ri_data1 = calc_rgd->rd_ri.ri_addr +
	calc_rgd->rd_ri.ri_length;
	calc_rgd->rd_ri.ri_data = blocks;
	calc_rgd->rd_ri.ri_bitbytes = calc_rgd->rd_ri.ri_data / GFS_NBBY;
	log_info("%d / %08X / %08X / %08X\n", calc_rgd->rd_ri.ri_length,
	calc_rgd->rd_ri.ri_data1, calc_rgd->rd_ri.ri_data,
	calc_rgd->rd_ri.ri_bitbytes);
	memset(rgindex_buf_ondisk, 0, sizeof(rgindex_buf_ondisk));
	gfs_rindex_out(&calc_rgd->rd_ri, rgindex_buf_ondisk);
	/* Note: rgindex_buf_ondisk is ONLY used for debug to see what the
	entry would look like on disk. */
	hexdump(rgi*sizeof(struct gfs_rindex),
	(unsigned char *)rgindex_buf_ondisk,
	sizeof(struct gfs_rindex));
	} /* for */
	relse_buf(sdp, bh); /* release the read buffer if we have one */
	return 0;
	}

	/*
	* ri_cleanup - free up the memory we previously allocated.
	*/
	void ri_cleanup(osi_list_t *rglist)
	{
	struct fsck_rgrp *rgd;

	while(!osi_list_empty(rglist)){
	rgd = osi_list_entry(rglist->next, struct fsck_rgrp, rd_list);
	if(rgd->rd_bits)
	free(rgd->rd_bits);
	if(rgd->rd_bh)
	free(rgd->rd_bh);
	osi_list_del(&rgd->rd_list);
	free(rgd);
	}
	}

	/**
	* ri_update - attach rgrps to the super block
	* @sdp:
	*
	* Given the rgrp index inode, link in all rgrps into the super block
	* and be sure that they can be read.
	*
	* If we encounter problems with any RGs, it either means we have a corrupt
	* RG or a corrupt RG file entry (which is less likely). We make up to three
	* attempts to do this. First, we trust that the RG index is correct and
	* read the RGs. If that fails, we become a little less trusting and
	* try to calculate what the RG index should look like in a perfect world.
	* If that doesn't work, we become even less trusting and go to great lengths
	* to figure out exactly where those RGs should be and what the index should
	* look like.
	*
	* Returns: 0 on success, -1 on failure.
	*/
	int ri_update(struct fsck_sb *sdp)
	{
	struct fsck_rgrp rgd, expected_rgd;
	osi_list_t expected_rglist; /* List of expected resource groups */
	osi_list_t *tmp;
	struct gfs_rindex buf;
	unsigned int rg, calc_rg_count;
	int error, count1 = 0, count2 = 0;
	int fix_grow_problems = 0, grow_problems = 0;
	enum rgindex_trust_level { /* how far can we trust our RG index? */
	blind_faith = 0, /* We'd like to trust the rgindex. We always used to
	before bz 179069. This should cover most cases. */
	open_minded = 1, /* At least 1 RG is corrupt. Try to calculate what it
	should be, in a perfect world where our RGs are all
	on even boundaries. Blue sky. Chirping birds. */
	distrust = 2 /* The world isn't perfect, our RGs are not on nice neat
	boundaries. The fs must have been messed with by
	gfs_grow or something. Count the RGs by hand. */
	} trust_lvl;

	log_info("Validating Resource Group index.\n");
	for (trust_lvl = blind_faith; trust_lvl <= distrust; trust_lvl++) {
	log_info("Level %d check.\n", trust_lvl + 1);
	count1 = count2 = 0;
	/* ---------------------------------------------------------------- */
	/* Step 1 - Calculate or figure out our own RG index */
	/* ---------------------------------------------------------------- */
	if (trust_lvl == blind_faith) { /* For now, assume rgindex is gospel */
	osi_list_init(&expected_rglist);
	error = FALSE;
	}
	else if (trust_lvl == open_minded) { /* If we can't trust RG index */
	/* Calculate our own RG index for comparison */
	error = gfs_rgindex_calculate(sdp, &expected_rglist,
	&calc_rg_count);
	if (error) { /* If calculated RGs don't reasonably match the fs */
	log_info("(failed--trying again at level 3)\n");
	ri_cleanup(&sdp->rglist);
	continue; /* Try again, this time counting them manually */
	}
	}
	else if (trust_lvl == distrust) { /* If we can't trust RG index */
	error = gfs_rgindex_rebuild(sdp, &expected_rglist,
	&calc_rg_count); /* count the RGs. */
	if (error) { /* If calculated RGs don't reasonably match the fs */
	log_info("(failed--giving up)\n");
	goto fail; /* try again, this time counting them manually */
	}
	}
	/* ---------------------------------------------------------------- */
	/* Step 2 - Read the real RG index and check its integrity */
	/* ---------------------------------------------------------------- */
	for (rg = 0; ; rg++) {
	int rgindex_modified;

	rgindex_modified = FALSE;
	error = readi(sdp->riinode, (char *)&buf,
	rg * sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (!error) /* if no data was read */
	break; /* we found the end of the rg index file */
	if (error != sizeof(struct gfs_rindex)) {
	log_err("Unable to read resource group index #%u.\n", rg);
	goto fail;
	}

	rgd = (struct fsck_rgrp *)malloc(sizeof(struct fsck_rgrp));
	memset(rgd, 0, sizeof(struct fsck_rgrp));
	rgd->rd_sbd = sdp;
	osi_list_add_prev(&rgd->rd_list, &sdp->rglist);
	gfs_rindex_in(&rgd->rd_ri, (char *)&buf);
	if (trust_lvl != blind_faith) {
	expected_rgd = osi_list_entry(expected_rglist.next,
	struct fsck_rgrp, rd_list);
	/* --------------------------------------------------------- */
	/* Now compare the index to the one we calculated / rebuilt */
	/* Since this is fsck and fsck's job is to fix filesystem */
	/* corruption, it's probably better to trust the calculated */
	/* value and discard what's reported on disk. */
	/* --------------------------------------------------------- */
	ri_compare(rg, rgd->rd_ri, expected_rgd->rd_ri,
	ri_addr, PRIx64);
	ri_compare(rg, rgd->rd_ri, expected_rgd->rd_ri,
	ri_length, PRIx32);
	ri_compare(rg, rgd->rd_ri, expected_rgd->rd_ri,
	ri_data1, PRIx64);
	ri_compare(rg, rgd->rd_ri, expected_rgd->rd_ri,
	ri_data, PRIx32);
	ri_compare(rg, rgd->rd_ri, expected_rgd->rd_ri,
	ri_bitbytes, PRIx32);
	/* If we modified the index, write it back to disk. */
	if (rgindex_modified) {
	if(query(sdp, "Fix the index? (y/n)")) {
	gfs_rindex_out(&rgd->rd_ri, (char *)&buf);
	error = writei(sdp->riinode, (char *)&buf,
	rg * sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (error != sizeof(struct gfs_rindex)) {
	log_err("Unable to fix resource group index %u.\n",
	rg + 1);
	goto fail;
	}
	}
	else
	log_err("RG index not fixed.\n");
	}
	osi_list_del(&expected_rgd->rd_list);
	free(expected_rgd);
	} /* if we can't trust the rg index */
	else { /* blind faith -- just check for the gfs_grow problem */
	if (rgd->rd_ri.ri_data == (uint32_t)-4) {
	if (!fix_grow_problems) {
	log_err("A problem with the rindex file caused by gfs_grow was detected.\n");
	if(query(sdp, "Fix the rindex problem? (y/n)"))
	fix_grow_problems = 1;
	}
	/* Keep a counter in case we hit it more than once. */
	grow_problems++;
	osi_list_del(&rgd->rd_list); /* take it out of the equation */
	free(rgd);
	continue;
	} else if (fix_grow_problems) {
	/* Once we detect the gfs_grow rindex problem, we have to */
	/* rewrite the entire rest of the rindex file, starting */
	/* with the entry AFTER the one that has the problem. */
	gfs_rindex_out(&rgd->rd_ri, (char *)&buf);
	error = writei(sdp->riinode, (char *)&buf,
	(rg - grow_problems) *
	sizeof(struct gfs_rindex),
	sizeof(struct gfs_rindex));
	if (error != sizeof(struct gfs_rindex)) {
	log_err("Unable to fix rindex entry %u.\n",
	rg + 1);
	goto fail;
	}
	}
	}
	error = fs_compute_bitstructs(rgd);
	if (error)
	break;
	rgd->rd_open_count = 0;
	count1++;
	} /* for all RGs in the index */
	rg -= grow_problems;
	if (!error) {
	log_info("%u resource groups found.\n", rg);
	if (trust_lvl != blind_faith && rg != calc_rg_count)
	log_warn("Resource group count discrepancy. Index says %d. " \
	"Should be %d.\n", rg, calc_rg_count);
	/* ------------------------------------------------------------- */
	/* Step 3 - Read the real RGs and check their integrity. */
	/* Now we can somewhat trust the rgindex and the RG addresses, */
	/* so let's read them in, check them and optionally fix them. */
	/* ------------------------------------------------------------- */
	error = FALSE;
	for (tmp = sdp->rglist.next; !error && tmp != &sdp->rglist;
	tmp = tmp->next) {
	rgd = osi_list_entry(tmp, struct fsck_rgrp, rd_list);
	error = fs_rgrp_read(rgd, trust_lvl);
	if (error)
	log_err("Unable to read in rgrp descriptor.\n");
	else
	fs_rgrp_relse(rgd);
	count2++;
	}
	if (!error && count1 != count2){
	log_err("Rgrps allocated (%d) does not equal"
	" rgrps read (%d).\n", count1, count2);
	error = -1;
	}
	sdp->rgcount = count1;
	}
	if (fix_grow_problems) {
	osi_buf_t *dibh;

	get_and_read_buf(sdp, sdp->sb.sb_rindex_di.no_addr, &dibh, 0);
	sdp->riinode->i_di.di_size = rg * sizeof(struct gfs_rindex);
	gfs_dinode_out(&sdp->riinode->i_di, BH_DATA(dibh));
	write_buf(sdp, dibh, 0);
	grow_problems = fix_grow_problems = 0;
	relse_buf(sdp, dibh);
	}
	if (!error) { /* if no problems encountered with the rgs */
	log_info("(passed)\n");
	break; /* no reason to distrust what we saw. Otherwise, we
	reiterate and become a little less trusting. */
	}
	else {
	if (trust_lvl < distrust)
	log_info("(failed--trying again at level 2)\n");
	else
	log_info("(failed--recovery impossible)\n");
	}
	ri_cleanup(&sdp->rglist);
	} /* for trust_lvl */
	return 0;

	fail:
	ri_cleanup(&sdp->rglist);
	return -1;
	}

	int write_sb(struct fsck_sb *sbp)
	{
	int error = 0;
	osi_buf_t *bh;

	error = get_and_read_buf(sbp, GFS_SB_ADDR >> sbp->fsb2bb_shift, &bh, 0);
	if (error){
	log_crit("Unable to read superblock\n");
	goto out;
	}

	gfs_sb_out(&sbp->sb, BH_DATA(bh));

	if((error = write_buf(sbp, bh, BW_WAIT))) {
	stack;
	goto out;
	}

	relse_buf(sbp, bh);
	out:
	return error;

	}

File Metadata

Mime Type: text/x-c
Expires: Tue, Feb 25, 2:11 PM (1 d, 10 h)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 1464859
Default Alt Text: super.c (47 KB)

super.cNo OneTemporaryActions

super.cView Options

File Metadata

Event Timeline

super.c
No OneTemporary
Actions

super.c
View Options