raid10.c

来自「linux 内核源代码」· C语言 代码 · 共 2,188 行 · 第 1/4 页

/*
 * raid10.c : Multiple Devices driver for Linux
 *
 * Copyright (C) 2000-2004 Neil Brown
 *
 * RAID-10 support for md.
 *
 * Base on code in raid1.c.  See raid1.c for futher copyright information.
 *
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * You should have received a copy of the GNU General Public License
 * (for example /usr/src/linux/COPYING); if not, write to the Free
 * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

#include "dm-bio-list.h"
#include <linux/raid/raid10.h>
#include <linux/raid/bitmap.h>

/*
 * RAID10 provides a combination of RAID0 and RAID1 functionality.
 * The layout of data is defined by
 *    chunk_size
 *    raid_disks
 *    near_copies (stored in low byte of layout)
 *    far_copies (stored in second byte of layout)
 *    far_offset (stored in bit 16 of layout )
 *
 * The data to be stored is divided into chunks using chunksize.
 * Each device is divided into far_copies sections.
 * In each section, chunks are laid out in a style similar to raid0, but
 * near_copies copies of each chunk is stored (each on a different drive).
 * The starting device for each section is offset near_copies from the starting
 * device of the previous section.
 * Thus they are (near_copies*far_copies) of each chunk, and each is on a different
 * drive.
 * near_copies and far_copies must be at least one, and their product is at most
 * raid_disks.
 *
 * If far_offset is true, then the far_copies are handled a bit differently.
 * The copies are still in different stripes, but instead of be very far apart
 * on disk, there are adjacent stripes.
 */

/*
 * Number of guaranteed r10bios in case of extreme VM load:
 */
#define	NR_RAID10_BIOS 256

static void unplug_slaves(mddev_t *mddev);

static void allow_barrier(conf_t *conf);
static void lower_barrier(conf_t *conf);

static void * r10bio_pool_alloc(gfp_t gfp_flags, void *data)
{
	conf_t *conf = data;
	r10bio_t *r10_bio;
	int size = offsetof(struct r10bio_s, devs[conf->copies]);

	/* allocate a r10bio with room for raid_disks entries in the bios array */
	r10_bio = kzalloc(size, gfp_flags);
	if (!r10_bio)
		unplug_slaves(conf->mddev);

	return r10_bio;
}

static void r10bio_pool_free(void *r10_bio, void *data)
{
	kfree(r10_bio);
}

#define RESYNC_BLOCK_SIZE (64*1024)
//#define RESYNC_BLOCK_SIZE PAGE_SIZE
#define RESYNC_SECTORS (RESYNC_BLOCK_SIZE >> 9)
#define RESYNC_PAGES ((RESYNC_BLOCK_SIZE + PAGE_SIZE-1) / PAGE_SIZE)
#define RESYNC_WINDOW (2048*1024)

/*
 * When performing a resync, we need to read and compare, so
 * we need as many pages are there are copies.
 * When performing a recovery, we need 2 bios, one for read,
 * one for write (we recover only one drive per r10buf)
 *
 */
static void * r10buf_pool_alloc(gfp_t gfp_flags, void *data)
{
	conf_t *conf = data;
	struct page *page;
	r10bio_t *r10_bio;
	struct bio *bio;
	int i, j;
	int nalloc;

	r10_bio = r10bio_pool_alloc(gfp_flags, conf);
	if (!r10_bio) {
		unplug_slaves(conf->mddev);
		return NULL;
	}

	if (test_bit(MD_RECOVERY_SYNC, &conf->mddev->recovery))
		nalloc = conf->copies; /* resync */
	else
		nalloc = 2; /* recovery */

	/*
	 * Allocate bios.
	 */
	for (j = nalloc ; j-- ; ) {
		bio = bio_alloc(gfp_flags, RESYNC_PAGES);
		if (!bio)
			goto out_free_bio;
		r10_bio->devs[j].bio = bio;
	}
	/*
	 * Allocate RESYNC_PAGES data pages and attach them
	 * where needed.
	 */
	for (j = 0 ; j < nalloc; j++) {
		bio = r10_bio->devs[j].bio;
		for (i = 0; i < RESYNC_PAGES; i++) {
			page = alloc_page(gfp_flags);
			if (unlikely(!page))
				goto out_free_pages;

			bio->bi_io_vec[i].bv_page = page;
		}
	}

	return r10_bio;

out_free_pages:
	for ( ; i > 0 ; i--)
		safe_put_page(bio->bi_io_vec[i-1].bv_page);
	while (j--)
		for (i = 0; i < RESYNC_PAGES ; i++)
			safe_put_page(r10_bio->devs[j].bio->bi_io_vec[i].bv_page);
	j = -1;
out_free_bio:
	while ( ++j < nalloc )
		bio_put(r10_bio->devs[j].bio);
	r10bio_pool_free(r10_bio, conf);
	return NULL;
}

static void r10buf_pool_free(void *__r10_bio, void *data)
{
	int i;
	conf_t *conf = data;
	r10bio_t *r10bio = __r10_bio;
	int j;

	for (j=0; j < conf->copies; j++) {
		struct bio *bio = r10bio->devs[j].bio;
		if (bio) {
			for (i = 0; i < RESYNC_PAGES; i++) {
				safe_put_page(bio->bi_io_vec[i].bv_page);
				bio->bi_io_vec[i].bv_page = NULL;
			}
			bio_put(bio);
		}
	}
	r10bio_pool_free(r10bio, conf);
}

static void put_all_bios(conf_t *conf, r10bio_t *r10_bio)
{
	int i;

	for (i = 0; i < conf->copies; i++) {
		struct bio **bio = & r10_bio->devs[i].bio;
		if (*bio && *bio != IO_BLOCKED)
			bio_put(*bio);
		*bio = NULL;
	}
}

static void free_r10bio(r10bio_t *r10_bio)
{
	conf_t *conf = mddev_to_conf(r10_bio->mddev);

	/*
	 * Wake up any possible resync thread that waits for the device
	 * to go idle.
	 */
	allow_barrier(conf);

	put_all_bios(conf, r10_bio);
	mempool_free(r10_bio, conf->r10bio_pool);
}

static void put_buf(r10bio_t *r10_bio)
{
	conf_t *conf = mddev_to_conf(r10_bio->mddev);

	mempool_free(r10_bio, conf->r10buf_pool);

	lower_barrier(conf);
}

static void reschedule_retry(r10bio_t *r10_bio)
{
	unsigned long flags;
	mddev_t *mddev = r10_bio->mddev;
	conf_t *conf = mddev_to_conf(mddev);

	spin_lock_irqsave(&conf->device_lock, flags);
	list_add(&r10_bio->retry_list, &conf->retry_list);
	conf->nr_queued ++;
	spin_unlock_irqrestore(&conf->device_lock, flags);

	md_wakeup_thread(mddev->thread);
}

/*
 * raid_end_bio_io() is called when we have finished servicing a mirrored
 * operation and are ready to return a success/failure code to the buffer
 * cache layer.
 */
static void raid_end_bio_io(r10bio_t *r10_bio)
{
	struct bio *bio = r10_bio->master_bio;

	bio_endio(bio,
		test_bit(R10BIO_Uptodate, &r10_bio->state) ? 0 : -EIO);
	free_r10bio(r10_bio);
}

/*
 * Update disk head position estimator based on IRQ completion info.
 */
static inline void update_head_pos(int slot, r10bio_t *r10_bio)
{
	conf_t *conf = mddev_to_conf(r10_bio->mddev);

	conf->mirrors[r10_bio->devs[slot].devnum].head_position =
		r10_bio->devs[slot].addr + (r10_bio->sectors);
}

static void raid10_end_read_request(struct bio *bio, int error)
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r10bio_t * r10_bio = (r10bio_t *)(bio->bi_private);
	int slot, dev;
	conf_t *conf = mddev_to_conf(r10_bio->mddev);


	slot = r10_bio->read_slot;
	dev = r10_bio->devs[slot].devnum;
	/*
	 * this branch is our 'one mirror IO has finished' event handler:
	 */
	update_head_pos(slot, r10_bio);

	if (uptodate) {
		/*
		 * Set R10BIO_Uptodate in our master bio, so that
		 * we will return a good error code to the higher
		 * levels even if IO on some other mirrored buffer fails.
		 *
		 * The 'master' represents the composite IO operation to
		 * user-side. So if something waits for IO, then it will
		 * wait for the 'master' bio.
		 */
		set_bit(R10BIO_Uptodate, &r10_bio->state);
		raid_end_bio_io(r10_bio);
	} else {
		/*
		 * oops, read error:
		 */
		char b[BDEVNAME_SIZE];
		if (printk_ratelimit())
			printk(KERN_ERR "raid10: %s: rescheduling sector %llu\n",
			       bdevname(conf->mirrors[dev].rdev->bdev,b), (unsigned long long)r10_bio->sector);
		reschedule_retry(r10_bio);
	}

	rdev_dec_pending(conf->mirrors[dev].rdev, conf->mddev);
}

static void raid10_end_write_request(struct bio *bio, int error)
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r10bio_t * r10_bio = (r10bio_t *)(bio->bi_private);
	int slot, dev;
	conf_t *conf = mddev_to_conf(r10_bio->mddev);

	for (slot = 0; slot < conf->copies; slot++)
		if (r10_bio->devs[slot].bio == bio)
			break;
	dev = r10_bio->devs[slot].devnum;

	/*
	 * this branch is our 'one mirror IO has finished' event handler:
	 */
	if (!uptodate) {
		md_error(r10_bio->mddev, conf->mirrors[dev].rdev);
		/* an I/O failed, we can't clear the bitmap */
		set_bit(R10BIO_Degraded, &r10_bio->state);
	} else
		/*
		 * Set R10BIO_Uptodate in our master bio, so that
		 * we will return a good error code for to the higher
		 * levels even if IO on some other mirrored buffer fails.
		 *
		 * The 'master' represents the composite IO operation to
		 * user-side. So if something waits for IO, then it will
		 * wait for the 'master' bio.
		 */
		set_bit(R10BIO_Uptodate, &r10_bio->state);

	update_head_pos(slot, r10_bio);

	/*
	 *
	 * Let's see if all mirrored write operations have finished
	 * already.
	 */
	if (atomic_dec_and_test(&r10_bio->remaining)) {
		/* clear the bitmap if all writes complete successfully */
		bitmap_endwrite(r10_bio->mddev->bitmap, r10_bio->sector,
				r10_bio->sectors,
				!test_bit(R10BIO_Degraded, &r10_bio->state),
				0);
		md_write_end(r10_bio->mddev);
		raid_end_bio_io(r10_bio);
	}

	rdev_dec_pending(conf->mirrors[dev].rdev, conf->mddev);
}


/*
 * RAID10 layout manager
 * Aswell as the chunksize and raid_disks count, there are two
 * parameters: near_copies and far_copies.
 * near_copies * far_copies must be <= raid_disks.
 * Normally one of these will be 1.
 * If both are 1, we get raid0.
 * If near_copies == raid_disks, we get raid1.
 *
 * Chunks are layed out in raid0 style with near_copies copies of the
 * first chunk, followed by near_copies copies of the next chunk and
 * so on.
 * If far_copies > 1, then after 1/far_copies of the array has been assigned
 * as described above, we start again with a device offset of near_copies.
 * So we effectively have another copy of the whole array further down all
 * the drives, but with blocks on different drives.
 * With this layout, and block is never stored twice on the one device.
 *
 * raid10_find_phys finds the sector offset of a given virtual sector
 * on each device that it is on.
 *
 * raid10_find_virt does the reverse mapping, from a device and a
 * sector offset to a virtual address
 */

static void raid10_find_phys(conf_t *conf, r10bio_t *r10bio)
{
	int n,f;
	sector_t sector;
	sector_t chunk;
	sector_t stripe;
	int dev;

	int slot = 0;

	/* now calculate first sector/dev */
	chunk = r10bio->sector >> conf->chunk_shift;
	sector = r10bio->sector & conf->chunk_mask;

	chunk *= conf->near_copies;
	stripe = chunk;
	dev = sector_div(stripe, conf->raid_disks);
	if (conf->far_offset)
		stripe *= conf->far_copies;

	sector += stripe << conf->chunk_shift;

	/* and calculate all the others */
	for (n=0; n < conf->near_copies; n++) {
		int d = dev;
		sector_t s = sector;
		r10bio->devs[slot].addr = sector;
		r10bio->devs[slot].devnum = d;
		slot++;

		for (f = 1; f < conf->far_copies; f++) {
			d += conf->near_copies;
			if (d >= conf->raid_disks)
				d -= conf->raid_disks;
			s += conf->stride;
			r10bio->devs[slot].devnum = d;
			r10bio->devs[slot].addr = s;
			slot++;
		}
		dev++;
		if (dev >= conf->raid_disks) {
			dev = 0;
			sector += (conf->chunk_mask + 1);
		}
	}
	BUG_ON(slot != conf->copies);
}

static sector_t raid10_find_virt(conf_t *conf, sector_t sector, int dev)
{
	sector_t offset, chunk, vchunk;

	offset = sector & conf->chunk_mask;
	if (conf->far_offset) {
		int fc;
		chunk = sector >> conf->chunk_shift;
		fc = sector_div(chunk, conf->far_copies);
		dev -= fc * conf->near_copies;
		if (dev < 0)
			dev += conf->raid_disks;
	} else {
		while (sector >= conf->stride) {
			sector -= conf->stride;
			if (dev < conf->near_copies)
				dev += conf->raid_disks - conf->near_copies;
			else
				dev -= conf->near_copies;
		}
		chunk = sector >> conf->chunk_shift;
	}
	vchunk = chunk * conf->raid_disks + dev;
	sector_div(vchunk, conf->near_copies);
	return (vchunk << conf->chunk_shift) + offset;
}

/**
 *	raid10_mergeable_bvec -- tell bio layer if a two requests can be merged
 *	@q: request queue
 *	@bio: the buffer head that's been built up so far
 *	@biovec: the request that could be merged to it.
 *
 *	Return amount of bytes we can accept at this offset
 *      If near_copies == raid_disk, there are no striping issues,
 *      but in that case, the function isn't called at all.
 */
static int raid10_mergeable_bvec(struct request_queue *q, struct bio *bio,
				struct bio_vec *bio_vec)
{
	mddev_t *mddev = q->queuedata;
	sector_t sector = bio->bi_sector + get_start_sect(bio->bi_bdev);
	int max;
	unsigned int chunk_sectors = mddev->chunk_size >> 9;
	unsigned int bio_sectors = bio->bi_size >> 9;

	max =  (chunk_sectors - ((sector & (chunk_sectors - 1)) + bio_sectors)) << 9;
	if (max < 0) max = 0; /* bio_add cannot handle a negative return */
	if (max <= bio_vec->bv_len && bio_sectors == 0)
		return bio_vec->bv_len;
	else
		return max;
}

/*
 * This routine returns the disk from which the requested read should
 * be done. There is a per-array 'next expected sequential IO' sector
 * number - if this matches on the next IO then we use the last disk.
 * There is also a per-disk 'last know head position' sector that is
 * maintained from IRQ contexts, both the normal and the resync IO
 * completion handlers update this position correctly. If there is no
 * perfect sequential match then we pick the disk whose head is closest.
 *
 * If there are 2 mirrors in the same 2 devices, performance degrades
 * because position is mirror, not device based.
 *
 * The rdev for the device selected will have nr_pending incremented.
 */

/*
 * FIXME: possibly should rethink readbalancing and do it differently
 * depending on near_copies / far_copies geometry.
 */
static int read_balance(conf_t *conf, r10bio_t *r10_bio)
{
	const unsigned long this_sector = r10_bio->sector;
	int disk, slot, nslot;
	const int sectors = r10_bio->sectors;
	sector_t new_distance, current_distance;
	mdk_rdev_t *rdev;

	raid10_find_phys(conf, r10_bio);
	rcu_read_lock();
	/*
	 * Check if we can balance. We can balance on the whole
	 * device if no resync is going on (recovery is ok), or below
	 * the resync window. We take the first readable disk when
	 * above the resync window.
	 */
	if (conf->mddev->recovery_cp < MaxSector
	    && (this_sector + sectors >= conf->next_resync)) {
		/* make sure that disk is operational */
		slot = 0;
		disk = r10_bio->devs[slot].devnum;

		while ((rdev = rcu_dereference(conf->mirrors[disk].rdev)) == NULL ||
		       r10_bio->devs[slot].bio == IO_BLOCKED ||
		       !test_bit(In_sync, &rdev->flags)) {
			slot++;
			if (slot == conf->copies) {
				slot = 0;
				disk = -1;
				break;
			}
			disk = r10_bio->devs[slot].devnum;
		}
		goto rb_out;
	}


	/* make sure the disk is operational */
	slot = 0;
	disk = r10_bio->devs[slot].devnum;
	while ((rdev=rcu_dereference(conf->mirrors[disk].rdev)) == NULL ||
	       r10_bio->devs[slot].bio == IO_BLOCKED ||
	       !test_bit(In_sync, &rdev->flags)) {
		slot ++;
		if (slot == conf->copies) {
			disk = -1;
			goto rb_out;
		}
		disk = r10_bio->devs[slot].devnum;
	}


	current_distance = abs(r10_bio->devs[slot].addr -
			       conf->mirrors[disk].head_position);

	/* Find the disk whose head is closest */

	for (nslot = slot; nslot < conf->copies; nslot++) {
		int ndisk = r10_bio->devs[nslot].devnum;


		if ((rdev=rcu_dereference(conf->mirrors[ndisk].rdev)) == NULL ||
		    r10_bio->devs[nslot].bio == IO_BLOCKED ||