sd.c

GNU Mach 微内核源代码, 基于美国卡内基美隆大学的 Mach 研究项目

/*
 *      sd.c Copyright (C) 1992 Drew Eckhardt 
 *           Copyright (C) 1993, 1994, 1995 Eric Youngdale
 *
 *      Linux scsi disk driver
 *              Initial versions: Drew Eckhardt 
 *              Subsequent revisions: Eric Youngdale
 *
 *      <[e-mail]>
 *
 *       Modified by Eric Youngdale [e-mail] to
 *       add scatter-gather, multiple outstanding request, and other
 *       enhancements.
 *
 *       Modified by Eric Youngdale [e-mail] to support loadable
 *       low-level scsi drivers.
 */

#include <linux/module.h>
#ifdef MODULE
/*
 * This is a variable in scsi.c that is set when we are processing something
 * after boot time.  By definition, this is true when we are a loadable module
 * ourselves.
 */
#define MODULE_FLAG 1
#else
#define MODULE_FLAG scsi_loadable_module_flag
#endif /* MODULE */

#include <linux/fs.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/mm.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/interrupt.h>

#include <asm/system.h>

#define MAJOR_NR SCSI_DISK_MAJOR
#include <linux/blk.h>
#include "scsi.h"
#include "hosts.h"
#include "sd.h"
#include <scsi/scsi_ioctl.h>
#include "constants.h"

#include <linux/genhd.h>

/*
 *  static const char RCSid[] = "$Header:";
 */

#define MAX_RETRIES 5

/*
 *  Time out in seconds for disks and Magneto-opticals (which are slower).
 */

#define SD_TIMEOUT (20 * HZ)
#define SD_MOD_TIMEOUT (25 * HZ)

#define CLUSTERABLE_DEVICE(SC) (SC->host->use_clustering && \
				SC->device->type != TYPE_MOD)

struct hd_struct * sd;

Scsi_Disk * rscsi_disks = NULL;
static int * sd_sizes;
static int * sd_blocksizes;
static int * sd_hardsizes;              /* Hardware sector size */

extern int sd_ioctl(struct inode *, struct file *, unsigned int, unsigned long);

static int check_scsidisk_media_change(kdev_t);
static int fop_revalidate_scsidisk(kdev_t);

static int sd_init_onedisk(int);

static void requeue_sd_request (Scsi_Cmnd * SCpnt);

static int sd_init(void);
static void sd_finish(void);
static int sd_attach(Scsi_Device *);
static int sd_detect(Scsi_Device *);
static void sd_detach(Scsi_Device *);

struct Scsi_Device_Template sd_template = 
{ NULL, "disk", "sd", NULL, TYPE_DISK, 
      SCSI_DISK_MAJOR, 0, 0, 0, 1,
      sd_detect, sd_init,
      sd_finish, sd_attach, sd_detach
};

static int sd_open(struct inode * inode, struct file * filp)
{
    int target;
    target =  DEVICE_NR(inode->i_rdev);
    
    if(target >= sd_template.dev_max || !rscsi_disks[target].device)
	return -ENXIO;   /* No such device */
    
    /* 
     * Make sure that only one process can do a check_change_disk at one time.
     * This is also used to lock out further access when the partition table 
     * is being re-read. 
     */
    
    while (rscsi_disks[target].device->busy)
        barrier();   
    if(rscsi_disks[target].device->removable) {
	check_disk_change(inode->i_rdev);
	
	/*
	 * If the drive is empty, just let the open fail.
	 */
	if ( !rscsi_disks[target].ready )
	    return -ENXIO;

	/*
	 * Similarly, if the device has the write protect tab set,
	 * have the open fail if the user expects to be able to write
	 * to the thing.
	 */
	if ( (rscsi_disks[target].write_prot) && (filp->f_mode & 2) )
	    return -EROFS;
    }

    /*
     * See if we are requesting a non-existent partition.  Do this
     * after checking for disk change.
     */
    if(sd_sizes[MINOR(inode->i_rdev)] == 0)
	return -ENXIO;
    
    if(rscsi_disks[target].device->removable)
	if(!rscsi_disks[target].device->access_count)
	    sd_ioctl(inode, NULL, SCSI_IOCTL_DOORLOCK, 0);

    rscsi_disks[target].device->access_count++;
    if (rscsi_disks[target].device->host->hostt->usage_count)
	(*rscsi_disks[target].device->host->hostt->usage_count)++;
    if(sd_template.usage_count) (*sd_template.usage_count)++;
    return 0;
}

static void sd_release(struct inode * inode, struct file * file)
{
    int target;
    fsync_dev(inode->i_rdev);
    
    target =  DEVICE_NR(inode->i_rdev);
    
    rscsi_disks[target].device->access_count--;
    if (rscsi_disks[target].device->host->hostt->usage_count)
	(*rscsi_disks[target].device->host->hostt->usage_count)--;
    if(sd_template.usage_count) (*sd_template.usage_count)--;
    
    if(rscsi_disks[target].device->removable) {
	if(!rscsi_disks[target].device->access_count)
	    sd_ioctl(inode, NULL, SCSI_IOCTL_DOORUNLOCK, 0);
    }
}

static void sd_geninit(struct gendisk *);

static struct file_operations sd_fops = {
    NULL,                        /* lseek - default */
    block_read,                  /* read - general block-dev read */
    block_write,                 /* write - general block-dev write */
    NULL,                        /* readdir - bad */
    NULL,                        /* select */
    sd_ioctl,                    /* ioctl */
    NULL,                        /* mmap */
    sd_open,                     /* open code */
    sd_release,                  /* release */
    block_fsync,                 /* fsync */
    NULL,                        /* fasync */
    check_scsidisk_media_change, /* Disk change */
    fop_revalidate_scsidisk      /* revalidate */
};

static struct gendisk sd_gendisk = {
    MAJOR_NR,                    /* Major number */
    "sd",                        /* Major name */
    4,                           /* Bits to shift to get real from partition */
    1 << 4,                      /* Number of partitions per real */
    0,                           /* maximum number of real */
    sd_geninit,                  /* init function */
    NULL,                        /* hd struct */
    NULL,                        /* block sizes */
    0,                           /* number */
    NULL,                        /* internal */
    NULL                         /* next */
};

static void sd_geninit (struct gendisk *ignored)
{
    int i;
    
    for (i = 0; i < sd_template.dev_max; ++i)
	if(rscsi_disks[i].device) 
	    sd[i << 4].nr_sects = rscsi_disks[i].capacity;
#if 0
    /* No longer needed - we keep track of this as we attach/detach */
    sd_gendisk.nr_real = sd_template.dev_max;
#endif
}

/*
 * rw_intr is the interrupt routine for the device driver.
 * It will be notified on the end of a SCSI read / write, and
 * will take one of several actions based on success or failure.
 */

static void rw_intr (Scsi_Cmnd *SCpnt)
{
    int result = SCpnt->result;
    int this_count = SCpnt->bufflen >> 9;
    int good_sectors = (result == 0 ? this_count : 0);
    int block_sectors = 1;
    
#ifdef DEBUG
    printk("sd%c : rw_intr(%d, %d)\n", 'a' + MINOR(SCpnt->request.rq_dev), 
	   SCpnt->host->host_no, result);
#endif

    /*
      Handle MEDIUM ERRORs that indicate partial success.  Since this is a
      relatively rare error condition, no care is taken to avoid unnecessary
      additional work such as memcpy's that could be avoided.
    */

    if (driver_byte(result) != 0 &&		    /* An error occurred */
	SCpnt->sense_buffer[0] == 0xF0 &&	    /* Sense data is valid */
	SCpnt->sense_buffer[2] == MEDIUM_ERROR)
      {
	long error_sector = (SCpnt->sense_buffer[3] << 24) |
			    (SCpnt->sense_buffer[4] << 16) |
			    (SCpnt->sense_buffer[5] << 8) |
			    SCpnt->sense_buffer[6];
	int sector_size =
	  rscsi_disks[DEVICE_NR(SCpnt->request.rq_dev)].sector_size;
	if (SCpnt->request.bh != NULL)
	  block_sectors = SCpnt->request.bh->b_size >> 9;
	if (sector_size == 1024)
	  {
	    error_sector <<= 1;
	    if (block_sectors < 2) block_sectors = 2;
	  }
	else if (sector_size == 256)
	  error_sector >>= 1;
	error_sector -= sd[MINOR(SCpnt->request.rq_dev)].start_sect;
	error_sector &= ~ (block_sectors - 1);
	good_sectors = error_sector - SCpnt->request.sector;
	if (good_sectors < 0 || good_sectors >= this_count)
	  good_sectors = 0;
      }
    
    /*
     * Handle RECOVERED ERRORs that indicate success after recovery action
     * by the target device.
     */

    if (SCpnt->sense_buffer[0] == 0xF0 &&	    /* Sense data is valid */
	SCpnt->sense_buffer[2] == RECOVERED_ERROR)
      {
	printk("scsidisk recovered I/O error: dev %s, sector %lu, absolute sector %lu\n",
	       kdevname(SCpnt->request.rq_dev), SCpnt->request.sector, 
	       SCpnt->request.sector + sd[MINOR(SCpnt->request.rq_dev)].start_sect);
	good_sectors = this_count;
	result = 0;
      }

    /*
     * First case : we assume that the command succeeded.  One of two things 
     * will happen here.  Either we will be finished, or there will be more
     * sectors that we were unable to read last time.
     */

    if (good_sectors > 0) {
	
#ifdef DEBUG
	printk("sd%c : %d sectors remain.\n", 'a' + MINOR(SCpnt->request.rq_dev),
	       SCpnt->request.nr_sectors);
	printk("use_sg is %d\n ",SCpnt->use_sg);
#endif
	if (SCpnt->use_sg) {
	    struct scatterlist * sgpnt;
	    int i;
	    sgpnt = (struct scatterlist *) SCpnt->buffer;
	    for(i=0; i<SCpnt->use_sg; i++) {
#ifdef DEBUG
		printk(":%x %x %d\n",sgpnt[i].alt_address, sgpnt[i].address, 
		       sgpnt[i].length);
#endif
		if (sgpnt[i].alt_address) {
		    if (SCpnt->request.cmd == READ)
			memcpy(sgpnt[i].alt_address, sgpnt[i].address, 
			       sgpnt[i].length);
		    scsi_free(sgpnt[i].address, sgpnt[i].length);
		}
	    }

	    /* Free list of scatter-gather pointers */
	    scsi_free(SCpnt->buffer, SCpnt->sglist_len);  
	} else {
	    if (SCpnt->buffer != SCpnt->request.buffer) {
#ifdef DEBUG
		printk("nosg: %x %x %d\n",SCpnt->request.buffer, SCpnt->buffer,
		       SCpnt->bufflen);
#endif  
		if (SCpnt->request.cmd == READ)
		    memcpy(SCpnt->request.buffer, SCpnt->buffer,
			   SCpnt->bufflen);
		scsi_free(SCpnt->buffer, SCpnt->bufflen);
	    }
	}
	/*
	 * If multiple sectors are requested in one buffer, then
	 * they will have been finished off by the first command.
	 * If not, then we have a multi-buffer command.
	 */
	if (SCpnt->request.nr_sectors > this_count)
	{
	    SCpnt->request.errors = 0;
	    
	    if (!SCpnt->request.bh)
	    {
#ifdef DEBUG
		printk("sd%c : handling page request, no buffer\n",
		       'a' + MINOR(SCpnt->request.rq_dev));
#endif
		/*
		 * The SCpnt->request.nr_sectors field is always done in 
		 * 512 byte sectors, even if this really isn't the case.
		 */
		panic("sd.c: linked page request (%lx %x)",
		      SCpnt->request.sector, this_count);
	    }
	}
	SCpnt = end_scsi_request(SCpnt, 1, good_sectors);
	if (result == 0)
	  {
	    requeue_sd_request(SCpnt);
	    return;
	  }
    }
    
    if (good_sectors == 0) {

    /* Free up any indirection buffers we allocated for DMA purposes. */
    if (SCpnt->use_sg) {
	struct scatterlist * sgpnt;
	int i;
	sgpnt = (struct scatterlist *) SCpnt->buffer;
	for(i=0; i<SCpnt->use_sg; i++) {
#ifdef DEBUG
	    printk("err: %x %x %d\n",SCpnt->request.buffer, SCpnt->buffer,
		   SCpnt->bufflen);
#endif
	    if (sgpnt[i].alt_address) {
		scsi_free(sgpnt[i].address, sgpnt[i].length);
	    }
	}
	scsi_free(SCpnt->buffer, SCpnt->sglist_len);  /* Free list of scatter-gather pointers */
    } else {
#ifdef DEBUG
	printk("nosgerr: %x %x %d\n",SCpnt->request.buffer, SCpnt->buffer,
	       SCpnt->bufflen);
#endif
	if (SCpnt->buffer != SCpnt->request.buffer)
	    scsi_free(SCpnt->buffer, SCpnt->bufflen);
    }
    }

    /*
     * Now, if we were good little boys and girls, Santa left us a request
     * sense buffer.  We can extract information from this, so we
     * can choose a block to remap, etc.
     */

    if (driver_byte(result) != 0) {
	if (suggestion(result) == SUGGEST_REMAP) {
#ifdef REMAP
	    /*
	     * Not yet implemented.  A read will fail after being remapped,
	     * a write will call the strategy routine again.
	     */
	    if rscsi_disks[DEVICE_NR(SCpnt->request.rq_dev)].remap
	    {
		result = 0;
	    }
	    else
#endif
	}
	
	if ((SCpnt->sense_buffer[0] & 0x7f) == 0x70) {
	    if ((SCpnt->sense_buffer[2] & 0xf) == UNIT_ATTENTION) {
		if(rscsi_disks[DEVICE_NR(SCpnt->request.rq_dev)].device->removable) {
		    /* detected disc change.  set a bit and quietly refuse
		     * further access.
		     */  
		    rscsi_disks[DEVICE_NR(SCpnt->request.rq_dev)].device->changed = 1;
		    SCpnt = end_scsi_request(SCpnt, 0, this_count);
		    requeue_sd_request(SCpnt);
		    return;
		}
                else
                {
                    /*
                     * Must have been a power glitch, or a bus reset.
                     * Could not have been a media change, so we just retry
                     * the request and see what happens.
                     */
                    requeue_sd_request(SCpnt);
                    return;
                }
	    }
	}