raid5.c

linux和2410结合开发 用他可以生成2410所需的zImage文件

		}
	}
	/*
	 * handle errors in spares (during reconstruction)
	 */
	if (conf->spare) {
		disk = conf->spare;
		if (disk->dev == dev) {
			printk (KERN_ALERT
				"raid5: Disk failure on spare %s\n",
				partition_name (dev));
			if (!conf->spare->operational) {
				/* probably a SET_DISK_FAULTY ioctl */
				return -EIO;
			}
			disk->operational = 0;
			disk->write_only = 0;
			conf->spare = NULL;
			mark_disk_faulty(sb->disks+disk->number);
			mark_disk_nonsync(sb->disks+disk->number);
			mark_disk_inactive(sb->disks+disk->number);
			sb->spare_disks--;
			sb->working_disks--;
			sb->failed_disks++;

			mddev->sb_dirty = 1;
			md_wakeup_thread(conf->thread);

			return 0;
		}
	}
	MD_BUG();
	return -EIO;
}	

/*
 * Input: a 'big' sector number,
 * Output: index of the data and parity disk, and the sector # in them.
 */
static unsigned long raid5_compute_sector(unsigned long r_sector, unsigned int raid_disks,
			unsigned int data_disks, unsigned int * dd_idx,
			unsigned int * pd_idx, raid5_conf_t *conf)
{
	unsigned long stripe;
	unsigned long chunk_number;
	unsigned int chunk_offset;
	unsigned long new_sector;
	int sectors_per_chunk = conf->chunk_size >> 9;

	/* First compute the information on this sector */

	/*
	 * Compute the chunk number and the sector offset inside the chunk
	 */
	chunk_number = r_sector / sectors_per_chunk;
	chunk_offset = r_sector % sectors_per_chunk;

	/*
	 * Compute the stripe number
	 */
	stripe = chunk_number / data_disks;

	/*
	 * Compute the data disk and parity disk indexes inside the stripe
	 */
	*dd_idx = chunk_number % data_disks;

	/*
	 * Select the parity disk based on the user selected algorithm.
	 */
	if (conf->level == 4)
		*pd_idx = data_disks;
	else switch (conf->algorithm) {
		case ALGORITHM_LEFT_ASYMMETRIC:
			*pd_idx = data_disks - stripe % raid_disks;
			if (*dd_idx >= *pd_idx)
				(*dd_idx)++;
			break;
		case ALGORITHM_RIGHT_ASYMMETRIC:
			*pd_idx = stripe % raid_disks;
			if (*dd_idx >= *pd_idx)
				(*dd_idx)++;
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
			*pd_idx = data_disks - stripe % raid_disks;
			*dd_idx = (*pd_idx + 1 + *dd_idx) % raid_disks;
			break;
		case ALGORITHM_RIGHT_SYMMETRIC:
			*pd_idx = stripe % raid_disks;
			*dd_idx = (*pd_idx + 1 + *dd_idx) % raid_disks;
			break;
		default:
			printk ("raid5: unsupported algorithm %d\n", conf->algorithm);
	}

	/*
	 * Finally, compute the new sector number
	 */
	new_sector = stripe * sectors_per_chunk + chunk_offset;
	return new_sector;
}

#if 0
static unsigned long compute_blocknr(struct stripe_head *sh, int i)
{
	raid5_conf_t *conf = sh->raid_conf;
	int raid_disks = conf->raid_disks, data_disks = raid_disks - 1;
	unsigned long new_sector = sh->sector, check;
	int sectors_per_chunk = conf->chunk_size >> 9;
	unsigned long stripe = new_sector / sectors_per_chunk;
	int chunk_offset = new_sector % sectors_per_chunk;
	int chunk_number, dummy1, dummy2, dd_idx = i;
	unsigned long r_sector, blocknr;

	switch (conf->algorithm) {
		case ALGORITHM_LEFT_ASYMMETRIC:
		case ALGORITHM_RIGHT_ASYMMETRIC:
			if (i > sh->pd_idx)
				i--;
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
		case ALGORITHM_RIGHT_SYMMETRIC:
			if (i < sh->pd_idx)
				i += raid_disks;
			i -= (sh->pd_idx + 1);
			break;
		default:
			printk ("raid5: unsupported algorithm %d\n", conf->algorithm);
	}

	chunk_number = stripe * data_disks + i;
	r_sector = chunk_number * sectors_per_chunk + chunk_offset;
	blocknr = r_sector / (sh->size >> 9);

	check = raid5_compute_sector (r_sector, raid_disks, data_disks, &dummy1, &dummy2, conf);
	if (check != sh->sector || dummy1 != dd_idx || dummy2 != sh->pd_idx) {
		printk("compute_blocknr: map not correct\n");
		return 0;
	}
	return blocknr;
}
#endif

#define check_xor() 	do { 					\
			   if (count == MAX_XOR_BLOCKS) {	\
				xor_block(count, bh_ptr);	\
				count = 1;			\
			   }					\
			} while(0)


static void compute_block(struct stripe_head *sh, int dd_idx)
{
	raid5_conf_t *conf = sh->raid_conf;
	int i, count, disks = conf->raid_disks;
	struct buffer_head *bh_ptr[MAX_XOR_BLOCKS], *bh;

	PRINTK("compute_block, stripe %lu, idx %d\n", sh->sector, dd_idx);


	memset(sh->bh_cache[dd_idx]->b_data, 0, sh->size);
	bh_ptr[0] = sh->bh_cache[dd_idx];
	count = 1;
	for (i = disks ; i--; ) {
		if (i == dd_idx)
			continue;
		bh = sh->bh_cache[i];
		if (buffer_uptodate(bh))
			bh_ptr[count++] = bh;
		else
			printk("compute_block() %d, stripe %lu, %d not present\n", dd_idx, sh->sector, i);

		check_xor();
	}
	if (count != 1)
		xor_block(count, bh_ptr);
	set_bit(BH_Uptodate, &sh->bh_cache[dd_idx]->b_state);
}

static void compute_parity(struct stripe_head *sh, int method)
{
	raid5_conf_t *conf = sh->raid_conf;
	int i, pd_idx = sh->pd_idx, disks = conf->raid_disks, count;
	struct buffer_head *bh_ptr[MAX_XOR_BLOCKS];
	struct buffer_head *chosen[MD_SB_DISKS];

	PRINTK("compute_parity, stripe %lu, method %d\n", sh->sector, method);
	memset(chosen, 0, sizeof(chosen));

	count = 1;
	bh_ptr[0] = sh->bh_cache[pd_idx];
	switch(method) {
	case READ_MODIFY_WRITE:
		if (!buffer_uptodate(sh->bh_cache[pd_idx]))
			BUG();
		for (i=disks ; i-- ;) {
			if (i==pd_idx)
				continue;
			if (sh->bh_write[i] &&
			    buffer_uptodate(sh->bh_cache[i])) {
				bh_ptr[count++] = sh->bh_cache[i];
				chosen[i] = sh->bh_write[i];
				sh->bh_write[i] = sh->bh_write[i]->b_reqnext;
				chosen[i]->b_reqnext = sh->bh_written[i];
				sh->bh_written[i] = chosen[i];
				check_xor();
			}
		}
		break;
	case RECONSTRUCT_WRITE:
		memset(sh->bh_cache[pd_idx]->b_data, 0, sh->size);
		for (i= disks; i-- ;)
			if (i!=pd_idx && sh->bh_write[i]) {
				chosen[i] = sh->bh_write[i];
				sh->bh_write[i] = sh->bh_write[i]->b_reqnext;
				chosen[i]->b_reqnext = sh->bh_written[i];
				sh->bh_written[i] = chosen[i];
			}
		break;
	case CHECK_PARITY:
		break;
	}
	if (count>1) {
		xor_block(count, bh_ptr);
		count = 1;
	}
	
	for (i = disks; i--;)
		if (chosen[i]) {
			struct buffer_head *bh = sh->bh_cache[i];
			char *bdata;
			bdata = bh_kmap(chosen[i]);
			memcpy(bh->b_data,
			       bdata,sh->size);
			bh_kunmap(chosen[i]);
			set_bit(BH_Lock, &bh->b_state);
			mark_buffer_uptodate(bh, 1);
		}

	switch(method) {
	case RECONSTRUCT_WRITE:
	case CHECK_PARITY:
		for (i=disks; i--;)
			if (i != pd_idx) {
				bh_ptr[count++] = sh->bh_cache[i];
				check_xor();
			}
		break;
	case READ_MODIFY_WRITE:
		for (i = disks; i--;)
			if (chosen[i]) {
				bh_ptr[count++] = sh->bh_cache[i];
				check_xor();
			}
	}
	if (count != 1)
		xor_block(count, bh_ptr);
	
	if (method != CHECK_PARITY) {
		mark_buffer_uptodate(sh->bh_cache[pd_idx], 1);
		set_bit(BH_Lock, &sh->bh_cache[pd_idx]->b_state);
	} else
		mark_buffer_uptodate(sh->bh_cache[pd_idx], 0);
}

static void add_stripe_bh (struct stripe_head *sh, struct buffer_head *bh, int dd_idx, int rw)
{
	struct buffer_head **bhp;
	raid5_conf_t *conf = sh->raid_conf;

	PRINTK("adding bh b#%lu to stripe s#%lu\n", bh->b_blocknr, sh->sector);


	spin_lock(&sh->lock);
	spin_lock_irq(&conf->device_lock);
	bh->b_reqnext = NULL;
	if (rw == READ)
		bhp = &sh->bh_read[dd_idx];
	else
		bhp = &sh->bh_write[dd_idx];
	while (*bhp) {
		printk(KERN_NOTICE "raid5: multiple %d requests for sector %ld\n", rw, sh->sector);
		bhp = & (*bhp)->b_reqnext;
	}
	*bhp = bh;
	spin_unlock_irq(&conf->device_lock);
	spin_unlock(&sh->lock);

	PRINTK("added bh b#%lu to stripe s#%lu, disk %d.\n", bh->b_blocknr, sh->sector, dd_idx);
}





/*
 * handle_stripe - do things to a stripe.
 *
 * We lock the stripe and then examine the state of various bits
 * to see what needs to be done.
 * Possible results:
 *    return some read request which now have data
 *    return some write requests which are safely on disc
 *    schedule a read on some buffers
 *    schedule a write of some buffers
 *    return confirmation of parity correctness
 *
 * Parity calculations are done inside the stripe lock
 * buffers are taken off read_list or write_list, and bh_cache buffers
 * get BH_Lock set before the stripe lock is released.
 *
 */
 
static void handle_stripe(struct stripe_head *sh)
{
	raid5_conf_t *conf = sh->raid_conf;
	int disks = conf->raid_disks;
	struct buffer_head *return_ok= NULL, *return_fail = NULL;
	int action[MD_SB_DISKS];
	int i;
	int syncing;
	int locked=0, uptodate=0, to_read=0, to_write=0, failed=0, written=0;
	int failed_num=0;
	struct buffer_head *bh;

	PRINTK("handling stripe %ld, cnt=%d, pd_idx=%d\n", sh->sector, atomic_read(&sh->count), sh->pd_idx);
	memset(action, 0, sizeof(action));

	spin_lock(&sh->lock);
	clear_bit(STRIPE_HANDLE, &sh->state);
	clear_bit(STRIPE_DELAYED, &sh->state);

	syncing = test_bit(STRIPE_SYNCING, &sh->state);
	/* Now to look around and see what can be done */

	for (i=disks; i--; ) {
		bh = sh->bh_cache[i];
		PRINTK("check %d: state 0x%lx read %p write %p written %p\n", i, bh->b_state, sh->bh_read[i], sh->bh_write[i], sh->bh_written[i]);
		/* maybe we can reply to a read */
		if (buffer_uptodate(bh) && sh->bh_read[i]) {
			struct buffer_head *rbh, *rbh2;
			PRINTK("Return read for disc %d\n", i);
			spin_lock_irq(&conf->device_lock);
			rbh = sh->bh_read[i];
			sh->bh_read[i] = NULL;
			spin_unlock_irq(&conf->device_lock);
			while (rbh) {
				char *bdata;
				bdata = bh_kmap(rbh);
				memcpy(bdata, bh->b_data, bh->b_size);
				bh_kunmap(rbh);
				rbh2 = rbh->b_reqnext;
				rbh->b_reqnext = return_ok;
				return_ok = rbh;
				rbh = rbh2;
			}
		}

		/* now count some things */
		if (buffer_locked(bh)) locked++;
		if (buffer_uptodate(bh)) uptodate++;

		
		if (sh->bh_read[i]) to_read++;
		if (sh->bh_write[i]) to_write++;
		if (sh->bh_written[i]) written++;
		if (!conf->disks[i].operational) {
			failed++;
			failed_num = i;
		}
	}
	PRINTK("locked=%d uptodate=%d to_read=%d to_write=%d failed=%d failed_num=%d\n",
	       locked, uptodate, to_read, to_write, failed, failed_num);
	/* check if the array has lost two devices and, if so, some requests might
	 * need to be failed
	 */
	if (failed > 1 && to_read+to_write) {
		for (i=disks; i--; ) {
			/* fail all writes first */
			if (sh->bh_write[i]) to_write--;
			while ((bh = sh->bh_write[i])) {
				sh->bh_write[i] = bh->b_reqnext;
				bh->b_reqnext = return_fail;
				return_fail = bh;
			}
			/* fail any reads if this device is non-operational */
			if (!conf->disks[i].operational) {
				spin_lock_irq(&conf->device_lock);
				if (sh->bh_read[i]) to_read--;
				while ((bh = sh->bh_read[i])) {
					sh->bh_read[i] = bh->b_reqnext;
					bh->b_reqnext = return_fail;
					return_fail = bh;
				}
				spin_unlock_irq(&conf->device_lock);
			}
		}
	}
	if (failed > 1 && syncing) {
		md_done_sync(conf->mddev, (sh->size>>9) - sh->sync_redone,0);
		clear_bit(STRIPE_SYNCING, &sh->state);
		syncing = 0;
	}

	/* might be able to return some write requests if the parity block
	 * is safe, or on a failed drive
	 */
	bh = sh->bh_cache[sh->pd_idx];
	if ( written &&
	     ( (conf->disks[sh->pd_idx].operational && !buffer_locked(bh) && buffer_uptodate(bh))
	       || (failed == 1 && failed_num == sh->pd_idx))
	    ) {
	    /* any written block on a uptodate or failed drive can be returned */
	    for (i=disks; i--; )
		if (sh->bh_written[i]) {
		    bh = sh->bh_cache[i];
		    if (!conf->disks[sh->pd_idx].operational ||
			(!buffer_locked(bh) && buffer_uptodate(bh)) ) {
			/* maybe we can return some write requests */
			struct buffer_head *wbh, *wbh2;
			PRINTK("Return write for disc %d\n", i);
			wbh = sh->bh_written[i];
			sh->bh_written[i] = NULL;
			while (wbh) {
			    wbh2 = wbh->b_reqnext;
			    wbh->b_reqnext = return_ok;
			    return_ok = wbh;
			    wbh = wbh2;
			}
		    }
		}
	}
		
	/* Now we might consider reading some blocks, either to check/generate
	 * parity, or to satisfy requests
	 */
	if (to_read || (syncing && (uptodate+failed < disks))) {
		for (i=disks; i--;) {
			bh = sh->bh_cache[i];
			if (!buffer_locked(bh) && !buffer_uptodate(bh) &&
			    (sh->bh_read[i] || syncing || (failed && sh->bh_read[failed_num]))) {
				/* we would like to get this block, possibly
				 * by computing it, but we might not be able to
				 */
				if (uptodate == disks-1) {
					PRINTK("Computing block %d\n", i);
					compute_block(sh, i);
					uptodate++;
				} else if (conf->disks[i].operational) {
					set_bit(BH_Lock, &bh->b_state);
					action[i] = READ+1;
					/* if I am just reading this block and we don't have
					   a failed drive, or any pending writes then sidestep the cache */
					if (sh->bh_page[i]) BUG();
					if (sh->bh_read[i] && !sh->bh_read[i]->b_reqnext &&
					    ! syncing && !failed && !to_write) {
						sh->bh_page[i] = sh->bh_cache[i]->b_page;
						sh->bh_cache[i]->b_page =  sh->bh_read[i]->b_page;
						sh->bh_cache[i]->b_data =  sh->bh_read[i]->b_data;
					}
					locked++;
					PRINTK("Reading block %d (sync=%d)\n", i, syncing);
					if (syncing)
						md_sync_acct(conf->disks[i].dev, bh->b_size>>9);
				}
			}
		}
		set_bit(STRIPE_HANDLE, &sh->state);
	}

	/* now to consider writing and what else, if anything should be read */
	if (to_write) {
		int rmw=0, rcw=0;
		for (i=disks ; i--;) {
			/* would I have to read this buffer for read_modify_write */
			bh = sh->bh_cache[i];
			if ((sh->bh_write[i] || i == sh->pd_idx) &&
			    (!buffer_locked(bh) || sh->bh_page[i]) &&
			    !buffer_uptodate(bh)) {
				if (conf->disks[i].operational 
/*				    && !(conf->resync_parity && i == sh->pd_idx) */
					)
					rmw++;
				else rmw += 2*disks;  /* cannot read it */
			}
			/* Would I have to read this buffer for reconstruct_write */
			if (!sh->bh_write[i] && i != sh->pd_idx &&
			    (!buffer_locked(bh) || sh->bh_page[i]) &&
			    !buffer_uptodate(bh)) {
				if (conf->disks[i].operational) rcw++;
				else rcw += 2*disks;
			}
		}
		PRINTK("for sector %ld, rmw=%d rcw=%d\n", sh->sector, rmw, rcw);
		set_bit(STRIPE_HANDLE, &sh->state);
		if (rmw < rcw && rmw > 0)
			/* prefer read-modify-write, but need to get some data */
			for (i=disks; i--;) {
				bh = sh->bh_cache[i];
				if ((sh->bh_write[i] || i == sh->pd_idx) &&
				    !buffer_locked(bh) && !buffer_uptodate(bh) &&
				    conf->disks[i].operational) {
					if (test_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
					{
						PRINTK("Read_old block %d for r-m-w\n", i);
						set_bit(BH_Lock, &bh->b_state);
						action[i] = READ+1;
						locked++;