raid6main.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 2,095 行 · 第 1/4 页

	 * need to be failed
	 */
	if (failed > 2 && to_read+to_write+written) {
		spin_lock_irq(&conf->device_lock);
		for (i=disks; i--; ) {
			/* fail all writes first */
			bi = sh->dev[i].towrite;
			sh->dev[i].towrite = NULL;
			if (bi) to_write--;

			while (bi && bi->bi_sector < sh->dev[i].sector + STRIPE_SECTORS){
				struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
				clear_bit(BIO_UPTODATE, &bi->bi_flags);
				if (--bi->bi_phys_segments == 0) {
					md_write_end(conf->mddev);
					bi->bi_next = return_bi;
					return_bi = bi;
				}
				bi = nextbi;
			}
			/* and fail all 'written' */
			bi = sh->dev[i].written;
			sh->dev[i].written = NULL;
			while (bi && bi->bi_sector < sh->dev[i].sector + STRIPE_SECTORS) {
				struct bio *bi2 = r5_next_bio(bi, sh->dev[i].sector);
				clear_bit(BIO_UPTODATE, &bi->bi_flags);
				if (--bi->bi_phys_segments == 0) {
					md_write_end(conf->mddev);
					bi->bi_next = return_bi;
					return_bi = bi;
				}
				bi = bi2;
			}

			/* fail any reads if this device is non-operational */
			if (!test_bit(R5_Insync, &sh->dev[i].flags)) {
				bi = sh->dev[i].toread;
				sh->dev[i].toread = NULL;
				if (bi) to_read--;
				while (bi && bi->bi_sector < sh->dev[i].sector + STRIPE_SECTORS){
					struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
					clear_bit(BIO_UPTODATE, &bi->bi_flags);
					if (--bi->bi_phys_segments == 0) {
						bi->bi_next = return_bi;
						return_bi = bi;
					}
					bi = nextbi;
				}
			}
		}
		spin_unlock_irq(&conf->device_lock);
	}
	if (failed > 2 && syncing) {
		md_done_sync(conf->mddev, STRIPE_SECTORS,0);
		clear_bit(STRIPE_SYNCING, &sh->state);
		syncing = 0;
	}

	/*
	 * might be able to return some write requests if the parity blocks
	 * are safe, or on a failed drive
	 */
	pdev = &sh->dev[pd_idx];
	p_failed = (failed >= 1 && failed_num[0] == pd_idx)
		|| (failed >= 2 && failed_num[1] == pd_idx);
	qdev = &sh->dev[qd_idx];
	q_failed = (failed >= 1 && failed_num[0] == qd_idx)
		|| (failed >= 2 && failed_num[1] == qd_idx);

	if ( written &&
	     ( p_failed || ((test_bit(R5_Insync, &pdev->flags)
			     && !test_bit(R5_LOCKED, &pdev->flags)
			     && test_bit(R5_UPTODATE, &pdev->flags))) ) &&
	     ( q_failed || ((test_bit(R5_Insync, &qdev->flags)
			     && !test_bit(R5_LOCKED, &qdev->flags)
			     && test_bit(R5_UPTODATE, &qdev->flags))) ) ) {
		/* any written block on an uptodate or failed drive can be
		 * returned.  Note that if we 'wrote' to a failed drive,
		 * it will be UPTODATE, but never LOCKED, so we don't need
		 * to test 'failed' directly.
		 */
		for (i=disks; i--; )
			if (sh->dev[i].written) {
				dev = &sh->dev[i];
				if (!test_bit(R5_LOCKED, &dev->flags) &&
				    test_bit(R5_UPTODATE, &dev->flags) ) {
					/* We can return any write requests */
					struct bio *wbi, *wbi2;
					PRINTK("Return write for stripe %llu disc %d\n",
					       (unsigned long long)sh->sector, i);
					spin_lock_irq(&conf->device_lock);
					wbi = dev->written;
					dev->written = NULL;
					while (wbi && wbi->bi_sector < dev->sector + STRIPE_SECTORS) {
						wbi2 = r5_next_bio(wbi, dev->sector);
						if (--wbi->bi_phys_segments == 0) {
							md_write_end(conf->mddev);
							wbi->bi_next = return_bi;
							return_bi = wbi;
						}
						wbi = wbi2;
					}
					spin_unlock_irq(&conf->device_lock);
				}
			}
	}

	/* Now we might consider reading some blocks, either to check/generate
	 * parity, or to satisfy requests
	 * or to load a block that is being partially written.
	 */
	if (to_read || non_overwrite || (syncing && (uptodate < disks))) {
		for (i=disks; i--;) {
			dev = &sh->dev[i];
			if (!test_bit(R5_LOCKED, &dev->flags) && !test_bit(R5_UPTODATE, &dev->flags) &&
			    (dev->toread ||
			     (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags)) ||
			     syncing ||
			     (failed >= 1 && (sh->dev[failed_num[0]].toread ||
					 (sh->dev[failed_num[0]].towrite && !test_bit(R5_OVERWRITE, &sh->dev[failed_num[0]].flags)))) ||
			     (failed >= 2 && (sh->dev[failed_num[1]].toread ||
					 (sh->dev[failed_num[1]].towrite && !test_bit(R5_OVERWRITE, &sh->dev[failed_num[1]].flags))))
				    )
				) {
				/* we would like to get this block, possibly
				 * by computing it, but we might not be able to
				 */
				if (uptodate == disks-1) {
					PRINTK("Computing stripe %llu block %d\n",
					       (unsigned long long)sh->sector, i);
					compute_block_1(sh, i);
					uptodate++;
				} else if ( uptodate == disks-2 && failed >= 2 ) {
					/* Computing 2-failure is *very* expensive; only do it if failed >= 2 */
					int other;
					for (other=disks; other--;) {
						if ( other == i )
							continue;
						if ( !test_bit(R5_UPTODATE, &sh->dev[other].flags) )
							break;
					}
					BUG_ON(other < 0);
					PRINTK("Computing stripe %llu blocks %d,%d\n",
					       (unsigned long long)sh->sector, i, other);
					compute_block_2(sh, i, other);
					uptodate += 2;
				} else if (test_bit(R5_Insync, &dev->flags)) {
					set_bit(R5_LOCKED, &dev->flags);
					set_bit(R5_Wantread, &dev->flags);
#if 0
					/* 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_read[i] && !sh->bh_read[i]->b_reqnext &&
					    ! syncing && !failed && !to_write) {
						sh->bh_cache[i]->b_page =  sh->bh_read[i]->b_page;
						sh->bh_cache[i]->b_data =  sh->bh_read[i]->b_data;
					}
#endif
					locked++;
					PRINTK("Reading block %d (sync=%d)\n",
						i, syncing);
					if (syncing)
						md_sync_acct(conf->disks[i].rdev->bdev,
							     STRIPE_SECTORS);
				}
			}
		}
		set_bit(STRIPE_HANDLE, &sh->state);
	}

	/* now to consider writing and what else, if anything should be read */
	if (to_write) {
		int rcw=0, must_compute=0;
		for (i=disks ; i--;) {
			dev = &sh->dev[i];
			/* Would I have to read this buffer for reconstruct_write */
			if (!test_bit(R5_OVERWRITE, &dev->flags)
			    && i != pd_idx && i != qd_idx
			    && (!test_bit(R5_LOCKED, &dev->flags)
#if 0
				|| sh->bh_page[i] != bh->b_page
#endif
				    ) &&
			    !test_bit(R5_UPTODATE, &dev->flags)) {
				if (test_bit(R5_Insync, &dev->flags)) rcw++;
				else {
					PRINTK("raid6: must_compute: disk %d flags=%#lx\n", i, dev->flags);
					must_compute++;
				}
			}
		}
		PRINTK("for sector %llu, rcw=%d, must_compute=%d\n",
		       (unsigned long long)sh->sector, rcw, must_compute);
		set_bit(STRIPE_HANDLE, &sh->state);

		if (rcw > 0)
			/* want reconstruct write, but need to get some data */
			for (i=disks; i--;) {
				dev = &sh->dev[i];
				if (!test_bit(R5_OVERWRITE, &dev->flags)
				    && !(failed == 0 && (i == pd_idx || i == qd_idx))
				    && !test_bit(R5_LOCKED, &dev->flags) && !test_bit(R5_UPTODATE, &dev->flags) &&
				    test_bit(R5_Insync, &dev->flags)) {
					if (test_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
					{
						PRINTK("Read_old stripe %llu block %d for Reconstruct\n",
						       (unsigned long long)sh->sector, i);
						set_bit(R5_LOCKED, &dev->flags);
						set_bit(R5_Wantread, &dev->flags);
						locked++;
					} else {
						PRINTK("Request delayed stripe %llu block %d for Reconstruct\n",
						       (unsigned long long)sh->sector, i);
						set_bit(STRIPE_DELAYED, &sh->state);
						set_bit(STRIPE_HANDLE, &sh->state);
					}
				}
			}
		/* now if nothing is locked, and if we have enough data, we can start a write request */
		if (locked == 0 && rcw == 0) {
			if ( must_compute > 0 ) {
				/* We have failed blocks and need to compute them */
				switch ( failed ) {
				case 0:	BUG();
				case 1: compute_block_1(sh, failed_num[0]); break;
				case 2: compute_block_2(sh, failed_num[0], failed_num[1]); break;
				default: BUG();	/* This request should have been failed? */
				}
			}

			PRINTK("Computing parity for stripe %llu\n", (unsigned long long)sh->sector);
			compute_parity(sh, RECONSTRUCT_WRITE);
			/* now every locked buffer is ready to be written */
			for (i=disks; i--;)
				if (test_bit(R5_LOCKED, &sh->dev[i].flags)) {
					PRINTK("Writing stripe %llu block %d\n",
					       (unsigned long long)sh->sector, i);
					locked++;
					set_bit(R5_Wantwrite, &sh->dev[i].flags);
#if 0 /**** FIX: I don't understand the logic here... ****/
					if (!test_bit(R5_Insync, &sh->dev[i].flags)
					    || ((i==pd_idx || i==qd_idx) && failed == 0)) /* FIX? */
						set_bit(STRIPE_INSYNC, &sh->state);
#endif
				}
			if (test_and_clear_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
				atomic_dec(&conf->preread_active_stripes);
				if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD)
					md_wakeup_thread(conf->mddev->thread);
			}
		}
	}

	/* maybe we need to check and possibly fix the parity for this stripe
	 * Any reads will already have been scheduled, so we just see if enough data
	 * is available
	 */
	if (syncing && locked == 0 &&
	    !test_bit(STRIPE_INSYNC, &sh->state) && failed <= 2) {
		set_bit(STRIPE_HANDLE, &sh->state);
#if 0 /* RAID-6: Don't support CHECK PARITY yet */
		if (failed == 0) {
			char *pagea;
			if (uptodate != disks)
				BUG();
			compute_parity(sh, CHECK_PARITY);
			uptodate--;
			pagea = page_address(sh->dev[pd_idx].page);
			if ((*(u32*)pagea) == 0 &&
			    !memcmp(pagea, pagea+4, STRIPE_SIZE-4)) {
				/* parity is correct (on disc, not in buffer any more) */
				set_bit(STRIPE_INSYNC, &sh->state);
			}
		}
#endif
		if (!test_bit(STRIPE_INSYNC, &sh->state)) {
			int failed_needupdate[2];
			struct r5dev *adev, *bdev;

			if ( failed < 1 )
				failed_num[0] = pd_idx;
			if ( failed < 2 )
				failed_num[1] = (failed_num[0] == qd_idx) ? pd_idx : qd_idx;

			failed_needupdate[0] = !test_bit(R5_UPTODATE, &sh->dev[failed_num[0]].flags);
			failed_needupdate[1] = !test_bit(R5_UPTODATE, &sh->dev[failed_num[1]].flags);

			PRINTK("sync: failed=%d num=%d,%d fnu=%u%u\n",
			       failed, failed_num[0], failed_num[1], failed_needupdate[0], failed_needupdate[1]);

#if 0  /* RAID-6: This code seems to require that CHECK_PARITY destroys the uptodateness of the parity */
			/* should be able to compute the missing block(s) and write to spare */
			if ( failed_needupdate[0] ^ failed_needupdate[1] ) {
				if (uptodate+1 != disks)
					BUG();
				compute_block_1(sh, failed_needupdate[0] ? failed_num[0] : failed_num[1]);
				uptodate++;
			} else if ( failed_needupdate[0] & failed_needupdate[1] ) {
				if (uptodate+2 != disks)
					BUG();
				compute_block_2(sh, failed_num[0], failed_num[1]);
				uptodate += 2;
			}
#else
			compute_block_2(sh, failed_num[0], failed_num[1]);
			uptodate += failed_needupdate[0] + failed_needupdate[1];
#endif

			if (uptodate != disks)
				BUG();

			PRINTK("Marking for sync stripe %llu blocks %d,%d\n",
			       (unsigned long long)sh->sector, failed_num[0], failed_num[1]);

			/**** FIX: Should we really do both of these unconditionally? ****/
			adev = &sh->dev[failed_num[0]];
			locked += !test_bit(R5_LOCKED, &adev->flags);
			set_bit(R5_LOCKED, &adev->flags);
			set_bit(R5_Wantwrite, &adev->flags);
			bdev = &sh->dev[failed_num[1]];
			locked += !test_bit(R5_LOCKED, &bdev->flags);
			set_bit(R5_LOCKED, &bdev->flags);
			set_bit(R5_Wantwrite, &bdev->flags);

			set_bit(STRIPE_INSYNC, &sh->state);
			set_bit(R5_Syncio, &adev->flags);
			set_bit(R5_Syncio, &bdev->flags);
		}
	}
	if (syncing && locked == 0 && test_bit(STRIPE_INSYNC, &sh->state)) {
		md_done_sync(conf->mddev, STRIPE_SECTORS,1);
		clear_bit(STRIPE_SYNCING, &sh->state);
	}

	spin_unlock(&sh->lock);

	while ((bi=return_bi)) {
		int bytes = bi->bi_size;

		return_bi = bi->bi_next;
		bi->bi_next = NULL;
		bi->bi_size = 0;
		bi->bi_end_io(bi, bytes, 0);
	}
	for (i=disks; i-- ;) {
		int rw;
		struct bio *bi;
		mdk_rdev_t *rdev;
		if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags))
			rw = 1;
		else if (test_and_clear_bit(R5_Wantread, &sh->dev[i].flags))
			rw = 0;
		else
			continue;

		bi = &sh->dev[i].req;

		bi->bi_rw = rw;
		if (rw)
			bi->bi_end_io = raid6_end_write_request;
		else
			bi->bi_end_io = raid6_end_read_request;

		spin_lock_irq(&conf->device_lock);
		rdev = conf->disks[i].rdev;
		if (rdev && rdev->faulty)
			rdev = NULL;
		if (rdev)
			atomic_inc(&rdev->nr_pending);
		spin_unlock_irq(&conf->device_lock);

		if (rdev) {
			if (test_bit(R5_Syncio, &sh->dev[i].flags))
				md_sync_acct(rdev->bdev, STRIPE_SECTORS);

			bi->bi_bdev = rdev->bdev;
			PRINTK("for %llu schedule op %ld on disc %d\n",
				(unsigned long long)sh->sector, bi->bi_rw, i);
			atomic_inc(&sh->count);
			bi->bi_sector = sh->sector + rdev->data_offset;
			bi->bi_flags = 1 << BIO_UPTODATE;
			bi->bi_vcnt = 1;
			bi->bi_idx = 0;
			bi->bi_io_vec = &sh->dev[i].vec;
			bi->bi_io_vec[0].bv_len = STRIPE_SIZE;
			bi->bi_io_vec[0].bv_offset = 0;
			bi->bi_size = STRIPE_SIZE;
			bi->bi_next = NULL;
			generic_make_request(bi);
		} else {
			PRINTK("skip op %ld on disc %d for sector %llu\n",
				bi->bi_rw, i, (unsigned long long)sh->sector);
			clear_bit(R5_LOCKED, &sh->dev[i].flags);
			set_bit(STRIPE_HANDLE, &sh->state);
		}
	}
}

static inline void raid6_activate_delayed(raid6_conf_t *conf)
{
	if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD) {
		while (!list_empty(&conf->delayed_list)) {
			struct list_head *l = conf->delayed_list.next;
			struct stripe_head *sh;
			sh = list_entry(l, struct stripe_head, lru);
			list_del_init(l);
			clear_bit(STRIPE_DELAYED, &sh->state);
			if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
				atomic_inc(&conf->preread_active_stripes);
			list_add_tail(&sh->lru, &conf->handle_list);
		}
	}
}

static void unplug_slaves(mddev_t *mddev)
{
	raid6_conf_t *conf = mddev_to_conf(mddev);
	int i;
	unsigned long flags;

	spin_lock_irqsave(&conf->device_lock, flags);
	for (i=0; i<mddev->raid_disks; i++) {
		mdk_rdev_t *rdev = conf->disks[i].rdev;
		if (rdev && atomic_read(&rdev->nr_pending)) {
			request_queue_t *r_queue = bdev_get_queue(rdev->bdev);

			atomic_inc(&rdev->nr_pending);
			spin_unlock_irqrestore(&conf->device_lock, flags);

			if (r_queue && r_queue->unplug_fn)
				r_queue->unplug_fn(r_queue);

			spin_lock_irqsave(&conf->device_lock, flags);
			atomic_dec(&rdev->nr_pending);
		}
	}
	spin_unlock_irqrestore(&conf->device_lock, flags);
}

static void raid6_unplug_device(request_queue_t *q)
{
	mddev_t *mddev = q->queuedata;
	raid6_conf_t *conf = mddev_to_conf(mddev);
	unsigned long flags;

	spin_lock_irqsave(&conf->device_lock, flags);

	if (blk_remove_plug(q))
		raid6_activate_delayed(conf);
	md_wakeup_thread(mddev->thread);

	spin_unlock_irqrestore(&conf->device_lock, flags);

	unplug_slaves(mddev);
}

static int raid6_issue_flush(request_queue_t *q, struct gendisk *disk,
			     sector_t *error_sector)
{
	mddev_t *mddev = q->queuedata;
	raid6_conf_t *conf = mddev_to_conf(mddev);
	int i, ret = 0;

	for (i=0; i<mddev->raid_disks; i++) {
		mdk_rdev_t *rdev = conf->disks[i].rdev;
		if (rdev && !rdev->faulty) {
			struct block_device *bdev = rdev->bdev;
			request_queue_t *r_queue;

			if (!bdev)
				continue;

			r_queue = bdev_get_queue(bdev);
			if (!r_queue)
				continue;

			if (!r_queue->issue_flush_fn) {
				ret = -EOPNOTSUPP;
				break;
			}

			ret = r_queue->issue_flush_fn(r_queue, bdev->bd_disk, error_sector);
			if (ret)
				break;
		}
	}
	return ret;
}

static inline void raid6_plug_device(raid6_conf_t *conf)
{
	spin_lock_irq(&conf->device_lock);
	blk_plug_device(conf->mddev->queue);
	spin_unlock_irq(&conf->device_lock);
}

static int make_request (request_queue_t *q, struct bio * bi)
{
	mddev_t *mddev = q->queuedata;
	raid6_conf_t *conf = mddev_to_conf(mddev);
	const unsigned int raid_disks = conf->raid_disks;
	const unsigned int data_disks = raid_disks - 2;
	unsigned int dd_idx, pd_idx;
	sector_t new_sector;
	sector_t logical_sector, last_sector;
	struct stripe_head *sh;

	if (bio_data_dir(bi)==WRITE) {
		disk_stat_inc(mddev->gendisk, writes);
		disk_stat_add(mddev->gendisk, write_sectors, bio_sectors(bi));
	} else {
		disk_stat_inc(mddev->gendisk, reads);
		disk_stat_add(mddev->gendisk, read_sectors, bio_sectors(bi));
	}

	logical_sector = bi->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
	last_sector = bi->bi_sector + (bi->bi_size>>9);

	bi->bi_next = NULL;
	bi->bi_phys_segments = 1;	/* over-loaded to count active stripes */
	if ( bio_data_dir(bi) == WRITE )
		md_write_start(mddev);
	for (;logical_sector < last_sector; logical_sector += STRIPE_SECTORS) {