md.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

		 */
		if (calc_sb_csum(rdev->sb) != rdev->sb->sb_csum) {
			if (rdev->sb->events_lo || rdev->sb->events_hi)
				if ((rdev->sb->events_lo--)==0)
					rdev->sb->events_hi--;
		}

		printk("%s's event counter: %08lx\n", partition_name(rdev->dev),
			(unsigned long)rdev->sb->events_lo);
		if (!freshest) {
			freshest = rdev;
			continue;
		}
		/*
		 * Find the newest superblock version
		 */
		ev1 = md_event(rdev->sb);
		ev2 = md_event(freshest->sb);
		if (ev1 != ev2) {
			out_of_date = 1;
			if (ev1 > ev2)
				freshest = rdev;
		}
	}
	if (out_of_date) {
		printk(OUT_OF_DATE);
		printk("freshest: %s\n", partition_name(freshest->dev));
	}
	memcpy (sb, freshest->sb, sizeof(*sb));

	/*
	 * at this point we have picked the 'best' superblock
	 * from all available superblocks.
	 * now we validate this superblock and kick out possibly
	 * failed disks.
	 */
	ITERATE_RDEV(mddev,rdev,tmp) {
		/*
		 * Kick all non-fresh devices faulty
		 */
		__u64 ev1, ev2;
		ev1 = md_event(rdev->sb);
		ev2 = md_event(sb);
		++ev1;
		if (ev1 < ev2) {
			printk("md: kicking non-fresh %s from array!\n",
						partition_name(rdev->dev));
			kick_rdev_from_array(rdev);
			continue;
		}
	}

	/*
	 * Fix up changed device names ... but only if this disk has a
	 * recent update time. Use faulty checksum ones too.
	 */
	ITERATE_RDEV(mddev,rdev,tmp) {
		__u64 ev1, ev2, ev3;
		if (rdev->faulty) { /* REMOVEME */
			MD_BUG();
			goto abort;
		}
		ev1 = md_event(rdev->sb);
		ev2 = md_event(sb);
		ev3 = ev2;
		--ev3;
		if ((rdev->dev != rdev->old_dev) &&
		    ((ev1 == ev2) || (ev1 == ev3))) {
			mdp_disk_t *desc;

			printk("md: device name has changed from %s to %s since last import!\n", partition_name(rdev->old_dev), partition_name(rdev->dev));
			if (rdev->desc_nr == -1) {
				MD_BUG();
				goto abort;
			}
			desc = &sb->disks[rdev->desc_nr];
			if (rdev->old_dev != MKDEV(desc->major, desc->minor)) {
				MD_BUG();
				goto abort;
			}
			desc->major = MAJOR(rdev->dev);
			desc->minor = MINOR(rdev->dev);
			desc = &rdev->sb->this_disk;
			desc->major = MAJOR(rdev->dev);
			desc->minor = MINOR(rdev->dev);
		}
	}

	/*
	 * Remove unavailable and faulty devices ...
	 *
	 * note that if an array becomes completely unrunnable due to
	 * missing devices, we do not write the superblock back, so the
	 * administrator has a chance to fix things up. The removal thus
	 * only happens if it's nonfatal to the contents of the array.
	 */
	for (i = 0; i < MD_SB_DISKS; i++) {
		int found;
		mdp_disk_t *desc;
		kdev_t dev;

		desc = sb->disks + i;
		dev = MKDEV(desc->major, desc->minor);

		/*
		 * We kick faulty devices/descriptors immediately.
		 */
		if (disk_faulty(desc)) {
			found = 0;
			ITERATE_RDEV(mddev,rdev,tmp) {
				if (rdev->desc_nr != desc->number)
					continue;
				printk("md%d: kicking faulty %s!\n",
					mdidx(mddev),partition_name(rdev->dev));
				kick_rdev_from_array(rdev);
				found = 1;
				break;
			}
			if (!found) {
				if (dev == MKDEV(0,0))
					continue;
				printk("md%d: removing former faulty %s!\n",
					mdidx(mddev), partition_name(dev));
			}
			remove_descriptor(desc, sb);
			continue;
		}

		if (dev == MKDEV(0,0))
			continue;
		/*
		 * Is this device present in the rdev ring?
		 */
		found = 0;
		ITERATE_RDEV(mddev,rdev,tmp) {
			if (rdev->desc_nr == desc->number) {
				found = 1;
				break;
			}
		}
		if (found)
			continue;

		printk("md%d: former device %s is unavailable, removing from array!\n", mdidx(mddev), partition_name(dev));
		remove_descriptor(desc, sb);
	}

	/*
	 * Double check wether all devices mentioned in the
	 * superblock are in the rdev ring.
	 */
	for (i = 0; i < MD_SB_DISKS; i++) {
		mdp_disk_t *desc;
		kdev_t dev;

		desc = sb->disks + i;
		dev = MKDEV(desc->major, desc->minor);

		if (dev == MKDEV(0,0))
			continue;

		if (disk_faulty(desc)) {
			MD_BUG();
			goto abort;
		}

		rdev = find_rdev(mddev, dev);
		if (!rdev) {
			MD_BUG();
			goto abort;
		}
	}

	/*
	 * Do a final reality check.
	 */
	ITERATE_RDEV(mddev,rdev,tmp) {
		if (rdev->desc_nr == -1) {
			MD_BUG();
			goto abort;
		}
		/*
		 * is the desc_nr unique?
		 */
		ITERATE_RDEV(mddev,rdev2,tmp2) {
			if ((rdev2 != rdev) &&
					(rdev2->desc_nr == rdev->desc_nr)) {
				MD_BUG();
				goto abort;
			}
		}
		/*
		 * is the device unique?
		 */
		ITERATE_RDEV(mddev,rdev2,tmp2) {
			if ((rdev2 != rdev) &&
					(rdev2->dev == rdev->dev)) {
				MD_BUG();
				goto abort;
			}
		}
	}

	/*
	 * Check if we can support this RAID array
	 */
	if (sb->major_version != MD_MAJOR_VERSION ||
			sb->minor_version > MD_MINOR_VERSION) {

		printk (OLD_VERSION, mdidx(mddev), sb->major_version,
				sb->minor_version, sb->patch_version);
		goto abort;
	}

	if ((sb->state != (1 << MD_SB_CLEAN)) && ((sb->level == 1) ||
			(sb->level == 4) || (sb->level == 5)))
		printk (NOT_CLEAN_IGNORE, mdidx(mddev));

	return 0;
abort:
	return 1;
}

#undef INCONSISTENT
#undef OUT_OF_DATE
#undef OLD_VERSION
#undef OLD_LEVEL

static int device_size_calculation (mddev_t * mddev)
{
	int data_disks = 0, persistent;
	unsigned int readahead;
	mdp_super_t *sb = mddev->sb;
	struct md_list_head *tmp;
	mdk_rdev_t *rdev;

	/*
	 * Do device size calculation. Bail out if too small.
	 * (we have to do this after having validated chunk_size,
	 * because device size has to be modulo chunk_size)
	 */
	persistent = !mddev->sb->not_persistent;
	ITERATE_RDEV(mddev,rdev,tmp) {
		if (rdev->faulty)
			continue;
		if (rdev->size) {
			MD_BUG();
			continue;
		}
		rdev->size = calc_dev_size(rdev->dev, mddev, persistent);
		if (rdev->size < sb->chunk_size / 1024) {
			printk (KERN_WARNING
				"Dev %s smaller than chunk_size: %ldk < %dk\n",
				partition_name(rdev->dev),
				rdev->size, sb->chunk_size / 1024);
			return -EINVAL;
		}
	}

	switch (sb->level) {
		case -3:
			data_disks = 1;
			break;
		case -2:
			data_disks = 1;
			break;
		case -1:
			zoned_raid_size(mddev);
			data_disks = 1;
			break;
		case 0:
			zoned_raid_size(mddev);
			data_disks = sb->raid_disks;
			break;
		case 1:
			data_disks = 1;
			break;
		case 4:
		case 5:
			data_disks = sb->raid_disks-1;
			break;
		default:
			printk (UNKNOWN_LEVEL, mdidx(mddev), sb->level);
			goto abort;
	}
	if (!md_size[mdidx(mddev)])
		md_size[mdidx(mddev)] = sb->size * data_disks;

	readahead = MD_READAHEAD;
	if ((sb->level == 0) || (sb->level == 4) || (sb->level == 5)) {
		readahead = (mddev->sb->chunk_size>>PAGE_SHIFT) * 4 * data_disks;
		if (readahead < data_disks * (MAX_SECTORS>>(PAGE_SHIFT-9))*2)
			readahead = data_disks * (MAX_SECTORS>>(PAGE_SHIFT-9))*2;
	} else {
		if (sb->level == -3)
			readahead = 0;
	}
	md_maxreadahead[mdidx(mddev)] = readahead;

	printk(KERN_INFO "md%d: max total readahead window set to %ldk\n",
		mdidx(mddev), readahead*(PAGE_SIZE/1024));

	printk(KERN_INFO
		"md%d: %d data-disks, max readahead per data-disk: %ldk\n",
			mdidx(mddev), data_disks, readahead/data_disks*(PAGE_SIZE/1024));
	return 0;
abort:
	return 1;
}


#define TOO_BIG_CHUNKSIZE KERN_ERR \
"too big chunk_size: %d > %d\n"

#define TOO_SMALL_CHUNKSIZE KERN_ERR \
"too small chunk_size: %d < %ld\n"

#define BAD_CHUNKSIZE KERN_ERR \
"no chunksize specified, see 'man raidtab'\n"

static int do_md_run (mddev_t * mddev)
{
	int pnum, err;
	int chunk_size;
	struct md_list_head *tmp;
	mdk_rdev_t *rdev;


	if (!mddev->nb_dev) {
		MD_BUG();
		return -EINVAL;
	}

	if (mddev->pers)
		return -EBUSY;

	/*
	 * Resize disks to align partitions size on a given
	 * chunk size.
	 */
	md_size[mdidx(mddev)] = 0;

	/*
	 * Analyze all RAID superblock(s)
	 */
	if (analyze_sbs(mddev)) {
		MD_BUG();
		return -EINVAL;
	}

	chunk_size = mddev->sb->chunk_size;
	pnum = level_to_pers(mddev->sb->level);

	mddev->param.chunk_size = chunk_size;
	mddev->param.personality = pnum;

	if (chunk_size > MAX_CHUNK_SIZE) {
		printk(TOO_BIG_CHUNKSIZE, chunk_size, MAX_CHUNK_SIZE);
		return -EINVAL;
	}
	/*
	 * chunk-size has to be a power of 2 and multiples of PAGE_SIZE
	 */
	if ( (1 << ffz(~chunk_size)) != chunk_size) {
		MD_BUG();
		return -EINVAL;
	}
	if (chunk_size < PAGE_SIZE) {
		printk(TOO_SMALL_CHUNKSIZE, chunk_size, PAGE_SIZE);
		return -EINVAL;
	}

	if (pnum >= MAX_PERSONALITY) {
		MD_BUG();
		return -EINVAL;
	}

	if ((pnum != RAID1) && (pnum != LINEAR) && !chunk_size) {
		/*
		 * 'default chunksize' in the old md code used to
		 * be PAGE_SIZE, baaad.
		 * we abort here to be on the safe side. We dont
		 * want to continue the bad practice.
		 */
		printk(BAD_CHUNKSIZE);
		return -EINVAL;
	}

	if (!pers[pnum])
	{
#ifdef CONFIG_KMOD
		char module_name[80];
		sprintf (module_name, "md-personality-%d", pnum);
		request_module (module_name);
		if (!pers[pnum])
#endif
			return -EINVAL;
	}

	if (device_size_calculation(mddev))
		return -EINVAL;

	/*
	 * Drop all container device buffers, from now on
	 * the only valid external interface is through the md
	 * device.
	 * Also find largest hardsector size
	 */
	md_hardsect_sizes[mdidx(mddev)] = 512;
	ITERATE_RDEV(mddev,rdev,tmp) {
		if (rdev->faulty)
			continue;
		fsync_dev(rdev->dev);
		invalidate_buffers(rdev->dev);
		if (get_hardsect_size(rdev->dev)
		    > md_hardsect_sizes[mdidx(mddev)]) 
			md_hardsect_sizes[mdidx(mddev)] =
				get_hardsect_size(rdev->dev);
	}
	md_blocksizes[mdidx(mddev)] = 1024;
	if (md_blocksizes[mdidx(mddev)] < md_hardsect_sizes[mdidx(mddev)])
		md_blocksizes[mdidx(mddev)] = md_hardsect_sizes[mdidx(mddev)];
	mddev->pers = pers[pnum];

	err = mddev->pers->run(mddev);
	if (err) {
		printk("pers->run() failed ...\n");
		mddev->pers = NULL;
		return -EINVAL;
	}

	mddev->sb->state &= ~(1 << MD_SB_CLEAN);
	md_update_sb(mddev);

	/*
	 * md_size has units of 1K blocks, which are
	 * twice as large as sectors.
	 */
	md_hd_struct[mdidx(mddev)].start_sect = 0;
	md_hd_struct[mdidx(mddev)].nr_sects = md_size[mdidx(mddev)] << 1;

	read_ahead[MD_MAJOR] = 1024;
	return (0);
}

#undef TOO_BIG_CHUNKSIZE
#undef BAD_CHUNKSIZE

#define OUT(x) do { err = (x); goto out; } while (0)

static int restart_array (mddev_t *mddev)
{
	int err = 0;

	/*
	 * Complain if it has no devices
	 */
	if (!mddev->nb_dev)
		OUT(-ENXIO);

	if (mddev->pers) {
		if (!mddev->ro)
			OUT(-EBUSY);

		mddev->ro = 0;
		set_device_ro(mddev_to_kdev(mddev), 0);

		printk (KERN_INFO
			"md%d switched to read-write mode.\n", mdidx(mddev));
		/*
		 * Kick recovery or resync if necessary
		 */
		md_recover_arrays();
		if (mddev->pers->restart_resync)
			mddev->pers->restart_resync(mddev);
	} else
		err = -EINVAL;

out:
	return err;
}

#define STILL_MOUNTED KERN_WARNING \
"md: md%d still mounted.\n"
#define	STILL_IN_USE \
"md: md%d still in use.\n"

static int do_md_stop (mddev_t * mddev, int ro)
{
	int err = 0, resync_interrupted = 0;
	kdev_t dev = mddev_to_kdev(mddev);

 	if (atomic_read(&mddev->active)>1) {
 		printk(STILL_IN_USE, mdidx(mddev));
 		OUT(-EBUSY);
 	}
 
 	/* this shouldn't be needed as above would have fired */
	if (!ro && get_super(dev)) {
		printk (STILL_MOUNTED, mdidx(mddev));
		OUT(-EBUSY);
	}

	if (mddev->pers) {
		/*
		 * It is safe to call stop here, it only frees private
		 * data. Also, it tells us if a device is unstoppable
		 * (eg. resyncing is in progress)
		 */
		if (mddev->pers->stop_resync)
			if (mddev->pers->stop_resync(mddev))
				resync_interrupted = 1;

		if (mddev->recovery_running)
			md_interrupt_thread(md_recovery_thread);

		/*
		 * This synchronizes with signal delivery to the
		 * resync or reconstruction thread. It also nicely
		 * hangs the process if some reconstruction has not
		 * finished.
		 */
		down(&mddev->recovery_sem);
		up(&mddev->recovery_sem);

		/*
		 *  sync and invalidate buffers because we cannot kill the
		 *  main thread with valid IO transfers still around.
		 *  the kernel lock protects us from new requests being
		 *  added after invalidate_buffers().
		 */
		fsync_dev (mddev_to_kdev(mddev));
		fsync_dev (dev);
		invalidate_buffers (dev);

		if (ro) {
			if (mddev->ro)
				OUT(-ENXIO);
			mddev->ro = 1;
		} else {
			if (mddev->ro)
				set_device_ro(dev, 0);
			if (mddev->pers->stop(mddev)) {
				if (mddev->ro)
					set_device_ro(dev, 1);
				OUT(-EBUSY);
			}
			if (mddev->ro)
				mddev->ro = 0;
		}
		if (mddev->sb) {
			/*
			 * mark it clean only if there was no resync
			 * interrupted.
			 */
			if (!mddev->recovery_running && !resync_interrupted) {
				printk("marking sb clean...\n");
				mddev->sb->state |= 1 << MD_SB_CLEAN;
			}
			md_update_sb(mddev);
		}
		if (ro)
			set_device_ro(dev, 1);
	}

	/*
	 * Free resources if final stop
	 */
	if (!ro) {
		printk (KERN_INFO "md%d stopped.\n", mdidx(mddev));
		free_mddev(mddev);

	} else
		printk (KERN_INFO
			"md%d switched to read-only mode.\n", mdidx(mddev));
out:
	return err;
}

#undef OUT

/*
 * We have to safely support old arrays too.
 */
int detect_old_array (mdp_super_t *sb)
{
	if (sb->major_version > 0)
		return 0;
	if (sb->minor_version >= 90)
		return 0;

	return -EINVAL;
}


static void autorun_array (mddev_t *mddev)
{
	mdk_rdev_t *rdev;
	struct md_list_head *tmp;
	int err;

	if (mddev->disks.prev == &mddev->disks) {
		MD_BUG();
		return;
	}

	printk("running: ");

	ITERATE_RDEV(mddev,rdev,tmp) {
		printk("<%s>", partition_name(rdev->dev));
	}
	printk("\nnow!\n");

	err = do_md_run (mddev);
	if (err) {