cciss.c

Linux块设备驱动源码

	
		// Fill in the scatter gather information
		if (iocommand.buf_size > 0 ) 
		{
			temp64.val = pci_map_single( host->pdev, buff,
                                        iocommand.buf_size, 
                                PCI_DMA_BIDIRECTIONAL);	
			c->SG[0].Addr.lower = temp64.val32.lower;
			c->SG[0].Addr.upper = temp64.val32.upper;
			c->SG[0].Len = iocommand.buf_size;
			c->SG[0].Ext = 0;  // we are not chaining
		}
		c->waiting = &wait;

		/* Put the request on the tail of the request queue */
		spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
		addQ(&host->reqQ, c);
		host->Qdepth++;
		start_io(host);
		spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);

		wait_for_completion(&wait);

		/* unlock the buffers from DMA */
		temp64.val32.lower = c->SG[0].Addr.lower;
                temp64.val32.upper = c->SG[0].Addr.upper;
                pci_unmap_single( host->pdev, (dma_addr_t) temp64.val,
                	iocommand.buf_size, PCI_DMA_BIDIRECTIONAL);

		/* Copy the error information out */ 
		iocommand.error_info = *(c->err_info);
		if ( copy_to_user(argp, &iocommand, sizeof( IOCTL_Command_struct) ) )
		{
			kfree(buff);
			cmd_free(host, c, 0);
			return( -EFAULT);	
		} 	

		if (iocommand.Request.Type.Direction == XFER_READ)
                {
                        /* Copy the data out of the buffer we created */
                        if (copy_to_user(iocommand.buf, buff, iocommand.buf_size))
			{
                        	kfree(buff);
				cmd_free(host, c, 0);
				return -EFAULT;
			}
                }
                kfree(buff);
		cmd_free(host, c, 0);
                return(0);
	} 
	case CCISS_BIG_PASSTHRU: {
		BIG_IOCTL_Command_struct *ioc;
		CommandList_struct *c;
		unsigned char **buff = NULL;
		int	*buff_size = NULL;
		u64bit	temp64;
		unsigned long flags;
		BYTE sg_used = 0;
		int status = 0;
		int i;
		DECLARE_COMPLETION(wait);
		__u32   left;
		__u32	sz;
		BYTE    __user *data_ptr;

		if (!arg)
			return -EINVAL;
		if (!capable(CAP_SYS_RAWIO))
			return -EPERM;
		ioc = (BIG_IOCTL_Command_struct *) 
			kmalloc(sizeof(*ioc), GFP_KERNEL);
		if (!ioc) {
			status = -ENOMEM;
			goto cleanup1;
		}
		if (copy_from_user(ioc, argp, sizeof(*ioc))) {
			status = -EFAULT;
			goto cleanup1;
		}
		if ((ioc->buf_size < 1) &&
			(ioc->Request.Type.Direction != XFER_NONE)) {
				status = -EINVAL;
				goto cleanup1;
		}
		/* Check kmalloc limits  using all SGs */
		if (ioc->malloc_size > MAX_KMALLOC_SIZE) {
			status = -EINVAL;
			goto cleanup1;
		}
		if (ioc->buf_size > ioc->malloc_size * MAXSGENTRIES) {
			status = -EINVAL;
			goto cleanup1;
		}
		buff = (unsigned char **) kmalloc(MAXSGENTRIES * 
				sizeof(char *), GFP_KERNEL);
		if (!buff) {
			status = -ENOMEM;
			goto cleanup1;
		}
		memset(buff, 0, MAXSGENTRIES);
		buff_size = (int *) kmalloc(MAXSGENTRIES * sizeof(int), 
					GFP_KERNEL);
		if (!buff_size) {
			status = -ENOMEM;
			goto cleanup1;
		}
		left = ioc->buf_size;
		data_ptr = ioc->buf;
		while (left) {
			sz = (left > ioc->malloc_size) ? ioc->malloc_size : left;
			buff_size[sg_used] = sz;
			buff[sg_used] = kmalloc(sz, GFP_KERNEL);
			if (buff[sg_used] == NULL) {
				status = -ENOMEM;
				goto cleanup1;
			}
			if (ioc->Request.Type.Direction == XFER_WRITE &&
				copy_from_user(buff[sg_used], data_ptr, sz)) {
					status = -ENOMEM;
					goto cleanup1;			
			} else {
				memset(buff[sg_used], 0, sz);
			}
			left -= sz;
			data_ptr += sz;
			sg_used++;
		}
		if ((c = cmd_alloc(host , 0)) == NULL) {
			status = -ENOMEM;
			goto cleanup1;	
		}
		c->cmd_type = CMD_IOCTL_PEND;
		c->Header.ReplyQueue = 0;
		
		if( ioc->buf_size > 0) {
			c->Header.SGList = sg_used;
			c->Header.SGTotal= sg_used;
		} else { 
			c->Header.SGList = 0;
			c->Header.SGTotal= 0;
		}
		c->Header.LUN = ioc->LUN_info;
		c->Header.Tag.lower = c->busaddr;
		
		c->Request = ioc->Request;
		if (ioc->buf_size > 0 ) {
			int i;
			for(i=0; i<sg_used; i++) {
				temp64.val = pci_map_single( host->pdev, buff[i],
					buff_size[i],
					PCI_DMA_BIDIRECTIONAL);
				c->SG[i].Addr.lower = temp64.val32.lower;
				c->SG[i].Addr.upper = temp64.val32.upper;
				c->SG[i].Len = buff_size[i];
				c->SG[i].Ext = 0;  /* we are not chaining */
			}
		}
		c->waiting = &wait;
		/* Put the request on the tail of the request queue */
		spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
		addQ(&host->reqQ, c);
		host->Qdepth++;
		start_io(host);
		spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
		wait_for_completion(&wait);
		/* unlock the buffers from DMA */
		for(i=0; i<sg_used; i++) {
			temp64.val32.lower = c->SG[i].Addr.lower;
			temp64.val32.upper = c->SG[i].Addr.upper;
			pci_unmap_single( host->pdev, (dma_addr_t) temp64.val,
				buff_size[i], PCI_DMA_BIDIRECTIONAL);
		}
		/* Copy the error information out */
		ioc->error_info = *(c->err_info);
		if (copy_to_user(argp, ioc, sizeof(*ioc))) {
			cmd_free(host, c, 0);
			status = -EFAULT;
			goto cleanup1;
		}
		if (ioc->Request.Type.Direction == XFER_READ) {
			/* Copy the data out of the buffer we created */
			BYTE __user *ptr = ioc->buf;
	        	for(i=0; i< sg_used; i++) {
				if (copy_to_user(ptr, buff[i], buff_size[i])) {
					cmd_free(host, c, 0);
					status = -EFAULT;
					goto cleanup1;
				}
				ptr += buff_size[i];
			}
		}
		cmd_free(host, c, 0);
		status = 0;
cleanup1:
		if (buff) {
			for(i=0; i<sg_used; i++)
				if(buff[i] != NULL)
					kfree(buff[i]);
			kfree(buff);
		}
		if (buff_size)
			kfree(buff_size);
		if (ioc)
			kfree(ioc);
		return(status);
	}
	default:
		return -ENOTTY;
	}
	
}

/*
 * revalidate_allvol is for online array config utilities.  After a
 * utility reconfigures the drives in the array, it can use this function
 * (through an ioctl) to make the driver zap any previous disk structs for
 * that controller and get new ones.
 *
 * Right now I'm using the getgeometry() function to do this, but this
 * function should probably be finer grained and allow you to revalidate one
 * particualar logical volume (instead of all of them on a particular
 * controller).
 */
static int revalidate_allvol(ctlr_info_t *host)
{
	int ctlr = host->ctlr, i;
	unsigned long flags;

        spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
        if (host->usage_count > 1) {
                spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
                printk(KERN_WARNING "cciss: Device busy for volume"
                        " revalidation (usage=%d)\n", host->usage_count);
                return -EBUSY;
        }
        host->usage_count++;
	spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);

	for(i=0; i< NWD; i++) {
		struct gendisk *disk = host->gendisk[i];
		if (disk->flags & GENHD_FL_UP)
			del_gendisk(disk);
	}

        /*
         * Set the partition and block size structures for all volumes
         * on this controller to zero.  We will reread all of this data
         */
        memset(host->drv,        0, sizeof(drive_info_struct)
						* CISS_MAX_LUN);
        /*
         * Tell the array controller not to give us any interrupts while
         * we check the new geometry.  Then turn interrupts back on when
         * we're done.
         */
        host->access.set_intr_mask(host, CCISS_INTR_OFF);
        cciss_getgeometry(ctlr);
        host->access.set_intr_mask(host, CCISS_INTR_ON);

	/* Loop through each real device */ 
	for (i = 0; i < NWD; i++) {
		struct gendisk *disk = host->gendisk[i];
		drive_info_struct *drv = &(host->drv[i]);
		/* we must register the controller even if no disks exist */
		/* this is for the online array utilities */
		if (!drv->heads && i)
			continue;
		blk_queue_hardsect_size(drv->queue, drv->block_size);
		set_capacity(disk, drv->nr_blocks);
		add_disk(disk);
	}
        host->usage_count--;
        return 0;
}

/* This function will check the usage_count of the drive to be updated/added.
 * If the usage_count is zero then the drive information will be updated and
 * the disk will be re-registered with the kernel.  If not then it will be
 * left alone for the next reboot.  The exception to this is disk 0 which
 * will always be left registered with the kernel since it is also the
 * controller node.  Any changes to disk 0 will show up on the next
 * reboot.
*/
static void cciss_update_drive_info(int ctlr, int drv_index)
  {
	ctlr_info_t *h = hba[ctlr];
	struct gendisk *disk;
	ReadCapdata_struct *size_buff = NULL;
	InquiryData_struct *inq_buff = NULL;
	unsigned int block_size;
	unsigned int total_size;
	unsigned long flags = 0;
	int ret = 0;

	/* if the disk already exists then deregister it before proceeding*/
	if (h->drv[drv_index].raid_level != -1){
		spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
		h->drv[drv_index].busy_configuring = 1;
		spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
		ret = deregister_disk(h->gendisk[drv_index],
			&h->drv[drv_index], 0);
		h->drv[drv_index].busy_configuring = 0;
	}

	/* If the disk is in use return */
	if (ret)
		return;


	/* Get information about the disk and modify the driver sturcture */
	size_buff = kmalloc(sizeof( ReadCapdata_struct), GFP_KERNEL);
        if (size_buff == NULL)
		goto mem_msg;
	inq_buff = kmalloc(sizeof( InquiryData_struct), GFP_KERNEL);
	if (inq_buff == NULL)
		goto mem_msg;

	cciss_read_capacity(ctlr, drv_index, size_buff, 1,
		&total_size, &block_size);
	cciss_geometry_inquiry(ctlr, drv_index, 1, total_size, block_size,
		inq_buff, &h->drv[drv_index]);

	++h->num_luns;
	disk = h->gendisk[drv_index];
	set_capacity(disk, h->drv[drv_index].nr_blocks);


	/* if it's the controller it's already added */
	if (drv_index){
		disk->queue = blk_init_queue(do_cciss_request, &h->lock);

		/* Set up queue information */
		disk->queue->backing_dev_info.ra_pages = READ_AHEAD;
		blk_queue_bounce_limit(disk->queue, hba[ctlr]->pdev->dma_mask);

		/* This is a hardware imposed limit. */
		blk_queue_max_hw_segments(disk->queue, MAXSGENTRIES);

		/* This is a limit in the driver and could be eliminated. */
		blk_queue_max_phys_segments(disk->queue, MAXSGENTRIES);

		blk_queue_max_sectors(disk->queue, 512);

		disk->queue->queuedata = hba[ctlr];

		blk_queue_hardsect_size(disk->queue,
			hba[ctlr]->drv[drv_index].block_size);

		h->drv[drv_index].queue = disk->queue;
		add_disk(disk);
	}

freeret:
	kfree(size_buff);
	kfree(inq_buff);
	return;
mem_msg:
	printk(KERN_ERR "cciss: out of memory\n");
	goto freeret;
}

/* This function will find the first index of the controllers drive array
 * that has a -1 for the raid_level and will return that index.  This is
 * where new drives will be added.  If the index to be returned is greater
 * than the highest_lun index for the controller then highest_lun is set
 * to this new index.  If there are no available indexes then -1 is returned.
*/
static int cciss_find_free_drive_index(int ctlr)
{
	int i;

	for (i=0; i < CISS_MAX_LUN; i++){
		if (hba[ctlr]->drv[i].raid_level == -1){
			if (i > hba[ctlr]->highest_lun)
				hba[ctlr]->highest_lun = i;
			return i;
		}
	}
	return -1;
}

/* This function will add and remove logical drives from the Logical
 * drive array of the controller and maintain persistancy of ordering
 * so that mount points are preserved until the next reboot.  This allows
 * for the removal of logical drives in the middle of the drive array
 * without a re-ordering of those drives.
 * INPUT
 * h		= The controller to perform the operations on
 * del_disk	= The disk to remove if specified.  If the value given
 *		  is NULL then no disk is removed.
*/
static int rebuild_lun_table(ctlr_info_t *h, struct gendisk *del_disk)
{
	int ctlr = h->ctlr;
	int num_luns;
	ReportLunData_struct *ld_buff = NULL;
	drive_info_struct *drv = NULL;
	int return_code;
	int listlength = 0;
	int i;
	int drv_found;
	int drv_index = 0;
	__u32 lunid = 0;
	unsigned long flags;

	/* Set busy_configuring flag for this operation */
	spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
	if (h->num_luns >= CISS_MAX_LUN){
		spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
		return -EINVAL;
	}

	if (h->busy_configuring){
		spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
		return -EBUSY;
	}
	h->busy_configuring = 1;

	/* if del_disk is NULL then we are being called to add a new disk
	 * and update the logical drive table.  If it is not NULL then
	 * we will check if the disk is in use or not.
	 */
	if (del_disk != NULL){
		drv = get_drv(del_disk);
		drv->busy_configuring = 1;
		spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
		return_code = deregister_disk(del_disk, drv, 1);
		drv->busy_configuring = 0;
		h->busy_configuring = 0;
		return return_code;
	} else {
		spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
		if (!capable(CAP_SYS_RAWIO))
			return -EPERM;

		ld_buff = kzalloc(sizeof(ReportLunData_struct), GFP_KERNEL);
		if (ld_buff == NULL)
			goto mem_msg;

		return_code = sendcmd_withirq(CISS_REPORT_LOG, ctlr, ld_buff,
				sizeof(ReportLunData_struct), 0, 0, 0,
				TYPE_CMD);

		if (return_code == IO_OK){
			listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[0])) << 24;
			listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[1])) << 16;
			listlength |= (0xff & (unsigned int)(ld_buff->LUNListLength[2])) << 8;
			listlength |= 0xff & (unsigned int)(ld_buff->LUNListLength[3]);