lloop.c

lustre 1.6.5 source code

        int                error;
        loff_t                size;

        if (!try_module_get(THIS_MODULE))
                return -ENODEV;

        error = -EBUSY;
        if (lo->lo_state != LLOOP_UNBOUND)
                goto out;

        mapping = file->f_mapping;
        inode = mapping->host;

        error = -EINVAL;
        if (!S_ISREG(inode->i_mode) || inode->i_sb->s_magic != LL_SUPER_MAGIC)
                goto out;

        if (!(file->f_mode & FMODE_WRITE))
                lo_flags |= LO_FLAGS_READ_ONLY;

        size = get_loop_size(lo, file);

        if ((loff_t)(sector_t)size != size) {
                error = -EFBIG;
                goto out;
        }

        /* remove all pages in cache so as dirty pages not to be existent. */
        truncate_inode_pages(mapping, 0);

        set_device_ro(bdev, (lo_flags & LO_FLAGS_READ_ONLY) != 0);

        lo->lo_blocksize = CFS_PAGE_SIZE;
        lo->lo_device = bdev;
        lo->lo_flags = lo_flags;
        lo->lo_backing_file = file;
        lo->ioctl = NULL;
        lo->lo_sizelimit = 0;
        lo->old_gfp_mask = mapping_gfp_mask(mapping);
        mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));

        lo->lo_bio = lo->lo_biotail = NULL;

        /*
         * set queue make_request_fn, and add limits based on lower level
         * device
         */
        blk_queue_make_request(lo->lo_queue, loop_make_request);
        lo->lo_queue->queuedata = lo;
        lo->lo_queue->unplug_fn = loop_unplug;

        /* queue parameters */
        blk_queue_hardsect_size(lo->lo_queue, CFS_PAGE_SIZE);
        blk_queue_max_sectors(lo->lo_queue, LLOOP_MAX_SEGMENTS);
        blk_queue_max_phys_segments(lo->lo_queue, LLOOP_MAX_SEGMENTS);

        set_capacity(disks[lo->lo_number], size);
        bd_set_size(bdev, size << 9);

        set_blocksize(bdev, lo->lo_blocksize);

        kernel_thread(loop_thread, lo, CLONE_KERNEL);
        down(&lo->lo_sem);
        return 0;

 out:
        /* This is safe: open() is still holding a reference. */
        module_put(THIS_MODULE);
        return error;
}

static int loop_clr_fd(struct lloop_device *lo, struct block_device *bdev, 
                       int count)
{
        struct file *filp = lo->lo_backing_file;
        int gfp = lo->old_gfp_mask;

        if (lo->lo_state != LLOOP_BOUND)
                return -ENXIO;

        if (lo->lo_refcnt > count)        /* we needed one fd for the ioctl */
                return -EBUSY;

        if (filp == NULL)
                return -EINVAL;

        spin_lock_irq(&lo->lo_lock);
        lo->lo_state = LLOOP_RUNDOWN;
        if (atomic_dec_and_test(&lo->lo_pending))
                up(&lo->lo_bh_mutex);
        spin_unlock_irq(&lo->lo_lock);

        down(&lo->lo_sem);
        lo->lo_backing_file = NULL;
        lo->ioctl = NULL;
        lo->lo_device = NULL;
        lo->lo_offset = 0;
        lo->lo_sizelimit = 0;
        lo->lo_flags = 0;
        ll_invalidate_bdev(bdev, 0);
        set_capacity(disks[lo->lo_number], 0);
        bd_set_size(bdev, 0);
        mapping_set_gfp_mask(filp->f_mapping, gfp);
        lo->lo_state = LLOOP_UNBOUND;
        fput(filp);
        /* This is safe: open() is still holding a reference. */
        module_put(THIS_MODULE);
        return 0;
}

static int lo_open(struct inode *inode, struct file *file)
{
        struct lloop_device *lo = inode->i_bdev->bd_disk->private_data;

        down(&lo->lo_ctl_mutex);
        lo->lo_refcnt++;
        up(&lo->lo_ctl_mutex);

        return 0;
}

static int lo_release(struct inode *inode, struct file *file)
{
        struct lloop_device *lo = inode->i_bdev->bd_disk->private_data;

        down(&lo->lo_ctl_mutex);
        --lo->lo_refcnt;
        up(&lo->lo_ctl_mutex);

        return 0;
}

/* lloop device node's ioctl function. */
static int lo_ioctl(struct inode *inode, struct file *unused, 
        unsigned int cmd, unsigned long arg)
{
        struct lloop_device *lo = inode->i_bdev->bd_disk->private_data;
        struct block_device *bdev = inode->i_bdev;
        int err = 0;

        down(&lloop_mutex);
        switch (cmd) {
        case LL_IOC_LLOOP_DETACH: {
                err = loop_clr_fd(lo, bdev, 2);
                if (err == 0)
                        blkdev_put(bdev); /* grabbed in LLOOP_ATTACH */
                break;
        }

        case LL_IOC_LLOOP_INFO: {
                __u64 ino = 0;

                if (lo->lo_state == LLOOP_BOUND)
                        ino = lo->lo_backing_file->f_dentry->d_inode->i_ino;

                if (put_user(ino, (__u64 *)arg))
                        err = -EFAULT;
                break; 
        }

        default:
                err = -EINVAL;
                break;
        }
        up(&lloop_mutex);

        return err;
}

static struct block_device_operations lo_fops = {
        .owner =        THIS_MODULE,
        .open =         lo_open,
        .release =      lo_release,
        .ioctl =        lo_ioctl,
};

/* dynamic iocontrol callback. 
 * This callback is registered in lloop_init and will be called by 
 * ll_iocontrol_call. 
 * This is a llite regular file ioctl function. It takes the responsibility 
 * of attaching a file, and detaching a file by a lloop's device numner. 
 */
static enum llioc_iter lloop_ioctl(struct inode *unused, struct file *file, 
                unsigned int cmd, unsigned long arg,
                void *magic, int *rcp)
{
        struct lloop_device *lo = NULL;
        struct block_device *bdev = NULL;
        int err = 0;
        dev_t dev;

        if (magic != ll_iocontrol_magic)
                return LLIOC_CONT;

        if (disks == NULL)
                GOTO(out1, err = -ENODEV);

        down(&lloop_mutex);
        switch (cmd) {
        case LL_IOC_LLOOP_ATTACH: {
                struct lloop_device *lo_free = NULL;
                int i;

                for (i = 0; i < max_loop; i++, lo = NULL) {
                        lo = &loop_dev[i];
                        if (lo->lo_state == LLOOP_UNBOUND) {
                                if (!lo_free)
                                        lo_free = lo;
                                continue;
                        }
                        if (lo->lo_backing_file->f_dentry->d_inode == 
                            file->f_dentry->d_inode)
                                break;
                }
                if (lo || !lo_free)
                        GOTO(out, err = -EBUSY);

                lo = lo_free;
                dev = MKDEV(lloop_major, lo->lo_number);

                /* quit if the used pointer is writable */
                if (put_user((long)old_encode_dev(dev), (long*)arg))
                        GOTO(out, err = -EFAULT);

                bdev = open_by_devnum(dev, file->f_mode);
                if (IS_ERR(bdev))
                        GOTO(out, err = PTR_ERR(bdev));

                get_file(file);
                err = loop_set_fd(lo, NULL, bdev, file);
                if (err) {
                        fput(file);
                        blkdev_put(bdev);
                }

                break;
        }

        case LL_IOC_LLOOP_DETACH_BYDEV: {
                int minor;
                
                dev = old_decode_dev(arg);
                if (MAJOR(dev) != lloop_major)
                        GOTO(out, err = -EINVAL);

                minor = MINOR(dev);
                if (minor > max_loop - 1)
                        GOTO(out, err = -EINVAL);

                lo = &loop_dev[minor];
                if (lo->lo_state != LLOOP_BOUND)
                        GOTO(out, err = -EINVAL);

                bdev = lo->lo_device;
                err = loop_clr_fd(lo, bdev, 1);
                if (err == 0)
                        blkdev_put(bdev); /* grabbed in LLOOP_ATTACH */

                break;
        }

        default:
                err = -EINVAL;
                break;
        }

out:
        up(&lloop_mutex);
out1:
        if (rcp)
                *rcp = err;
        return LLIOC_STOP;
}

static int __init lloop_init(void)
{
        int        i;
        unsigned int cmdlist[] = {
                LL_IOC_LLOOP_ATTACH,
                LL_IOC_LLOOP_DETACH_BYDEV,
        };

        if (max_loop < 1 || max_loop > 256) {
                CWARN("lloop: invalid max_loop (must be between"
                      " 1 and 256), using default (8)\n");
                max_loop = 8;
        }

        lloop_major = register_blkdev(0, "lloop");
        if (lloop_major < 0)
                return -EIO;

        ll_iocontrol_magic = ll_iocontrol_register(lloop_ioctl, 2, cmdlist);
        if (ll_iocontrol_magic == NULL)
                goto out_mem1;

        loop_dev = kmalloc(max_loop * sizeof(struct lloop_device), GFP_KERNEL);
        if (!loop_dev)
                goto out_mem1;
        memset(loop_dev, 0, max_loop * sizeof(struct lloop_device));

        disks = kmalloc(max_loop * sizeof(struct gendisk *), GFP_KERNEL);
        if (!disks)
                goto out_mem2;

        for (i = 0; i < max_loop; i++) {
                disks[i] = alloc_disk(1);
                if (!disks[i])
                        goto out_mem3;
        }

        init_MUTEX(&lloop_mutex);

        for (i = 0; i < max_loop; i++) {
                struct lloop_device *lo = &loop_dev[i];
                struct gendisk *disk = disks[i];

                memset(lo, 0, sizeof(*lo));
                lo->lo_queue = blk_alloc_queue(GFP_KERNEL);
                if (!lo->lo_queue)
                        goto out_mem4;

                init_MUTEX(&lo->lo_ctl_mutex);
                init_MUTEX_LOCKED(&lo->lo_sem);
                init_MUTEX_LOCKED(&lo->lo_bh_mutex);
                lo->lo_number = i;
                spin_lock_init(&lo->lo_lock);
                disk->major = lloop_major;
                disk->first_minor = i;
                disk->fops = &lo_fops;
                sprintf(disk->disk_name, "lloop%d", i);
                disk->private_data = lo;
                disk->queue = lo->lo_queue;
        }

        /* We cannot fail after we call this, so another loop!*/
        for (i = 0; i < max_loop; i++)
                add_disk(disks[i]);
        return 0;

out_mem4:
        while (i--)
                blk_put_queue(loop_dev[i].lo_queue);
        i = max_loop;
out_mem3:
        while (i--)
                put_disk(disks[i]);
        kfree(disks);
out_mem2:
        kfree(loop_dev);
out_mem1:
        unregister_blkdev(lloop_major, "lloop");
        ll_iocontrol_unregister(ll_iocontrol_magic);
        CERROR("lloop: ran out of memory\n");
        return -ENOMEM;
}

static void lloop_exit(void)
{
        int i;

        ll_iocontrol_unregister(ll_iocontrol_magic);
        for (i = 0; i < max_loop; i++) {
                del_gendisk(disks[i]);
                blk_put_queue(loop_dev[i].lo_queue);
                put_disk(disks[i]);
        }
        if (ll_unregister_blkdev(lloop_major, "lloop"))
                CWARN("lloop: cannot unregister blkdev\n");

        kfree(disks);
        kfree(loop_dev);
}

module_init(lloop_init);
module_exit(lloop_exit);

CFS_MODULE_PARM(max_loop, "i", int, 0444, "maximum of lloop_device");
MODULE_AUTHOR("Cluster File Systems, Inc. <[e-mail]>");
MODULE_DESCRIPTION("Lustre virtual block device");
MODULE_LICENSE("GPL");