summaryrefslogtreecommitdiffstats
path: root/drivers/s390/char/vmur.c
diff options
context:
space:
mode:
authorFrank Munzert <munzert@de.ibm.com>2007-07-17 13:36:06 +0200
committerMartin Schwidefsky <schwidefsky@de.ibm.com>2007-07-17 13:36:19 +0200
commit810cb5b32ded8f28880b502e984d807d03869d3b (patch)
treef3370de9af4974b54983c4bd377e606fca287938 /drivers/s390/char/vmur.c
parentd9d119f12fdb14d57e96e4698ef2ec303b102ae9 (diff)
downloadop-kernel-dev-810cb5b32ded8f28880b502e984d807d03869d3b.zip
op-kernel-dev-810cb5b32ded8f28880b502e984d807d03869d3b.tar.gz
[S390] z/VM unit record device driver
z/VM Unit record character device driver to access VM reader, punch, and printer. Signed-off-by: Frank Munzert <munzert@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Diffstat (limited to 'drivers/s390/char/vmur.c')
-rw-r--r--drivers/s390/char/vmur.c906
1 files changed, 906 insertions, 0 deletions
diff --git a/drivers/s390/char/vmur.c b/drivers/s390/char/vmur.c
new file mode 100644
index 0000000..e90b0f8
--- /dev/null
+++ b/drivers/s390/char/vmur.c
@@ -0,0 +1,906 @@
+/*
+ * Linux driver for System z and s390 unit record devices
+ * (z/VM virtual punch, reader, printer)
+ *
+ * Copyright IBM Corp. 2001, 2007
+ * Authors: Malcolm Beattie <beattiem@uk.ibm.com>
+ * Michael Holzheu <holzheu@de.ibm.com>
+ * Frank Munzert <munzert@de.ibm.com>
+ */
+
+#include <linux/cdev.h>
+
+#include <asm/uaccess.h>
+#include <asm/cio.h>
+#include <asm/ccwdev.h>
+#include <asm/debug.h>
+
+#include "vmur.h"
+
+/*
+ * Driver overview
+ *
+ * Unit record device support is implemented as a character device driver.
+ * We can fit at least 16 bits into a device minor number and use the
+ * simple method of mapping a character device number with minor abcd
+ * to the unit record device with devno abcd.
+ * I/O to virtual unit record devices is handled as follows:
+ * Reads: Diagnose code 0x14 (input spool file manipulation)
+ * is used to read spool data page-wise.
+ * Writes: The CCW used is WRITE_CCW_CMD (0x01). The device's record length
+ * is available by reading sysfs attr reclen. Each write() to the device
+ * must specify an integral multiple (maximal 511) of reclen.
+ */
+
+static char ur_banner[] = "z/VM virtual unit record device driver";
+
+MODULE_AUTHOR("IBM Corporation");
+MODULE_DESCRIPTION("s390 z/VM virtual unit record device driver");
+MODULE_LICENSE("GPL");
+
+#define PRINTK_HEADER "vmur: "
+
+static dev_t ur_first_dev_maj_min;
+static struct class *vmur_class;
+static struct debug_info *vmur_dbf;
+
+/* We put the device's record length (for writes) in the driver_info field */
+static struct ccw_device_id ur_ids[] = {
+ { CCWDEV_CU_DI(READER_PUNCH_DEVTYPE, 80) },
+ { CCWDEV_CU_DI(PRINTER_DEVTYPE, 132) },
+ { /* end of list */ }
+};
+
+MODULE_DEVICE_TABLE(ccw, ur_ids);
+
+static int ur_probe(struct ccw_device *cdev);
+static void ur_remove(struct ccw_device *cdev);
+static int ur_set_online(struct ccw_device *cdev);
+static int ur_set_offline(struct ccw_device *cdev);
+
+static struct ccw_driver ur_driver = {
+ .name = "vmur",
+ .owner = THIS_MODULE,
+ .ids = ur_ids,
+ .probe = ur_probe,
+ .remove = ur_remove,
+ .set_online = ur_set_online,
+ .set_offline = ur_set_offline,
+};
+
+/*
+ * Allocation, freeing, getting and putting of urdev structures
+ */
+static struct urdev *urdev_alloc(struct ccw_device *cdev)
+{
+ struct urdev *urd;
+
+ urd = kzalloc(sizeof(struct urdev), GFP_KERNEL);
+ if (!urd)
+ return NULL;
+ urd->cdev = cdev;
+ urd->reclen = cdev->id.driver_info;
+ ccw_device_get_id(cdev, &urd->dev_id);
+ mutex_init(&urd->io_mutex);
+ mutex_init(&urd->open_mutex);
+ return urd;
+}
+
+static void urdev_free(struct urdev *urd)
+{
+ kfree(urd);
+}
+
+/*
+ * This is how the character device driver gets a reference to a
+ * ur device. When this call returns successfully, a reference has
+ * been taken (by get_device) on the underlying kobject. The recipient
+ * of this urdev pointer must eventually drop it with urdev_put(urd)
+ * which does the corresponding put_device().
+ */
+static struct urdev *urdev_get_from_devno(u16 devno)
+{
+ char bus_id[16];
+ struct ccw_device *cdev;
+
+ sprintf(bus_id, "0.0.%04x", devno);
+ cdev = get_ccwdev_by_busid(&ur_driver, bus_id);
+ if (!cdev)
+ return NULL;
+
+ return cdev->dev.driver_data;
+}
+
+static void urdev_put(struct urdev *urd)
+{
+ put_device(&urd->cdev->dev);
+}
+
+/*
+ * Low-level functions to do I/O to a ur device.
+ * alloc_chan_prog
+ * do_ur_io
+ * ur_int_handler
+ *
+ * alloc_chan_prog allocates and builds the channel program
+ *
+ * do_ur_io issues the channel program to the device and blocks waiting
+ * on a completion event it publishes at urd->io_done. The function
+ * serialises itself on the device's mutex so that only one I/O
+ * is issued at a time (and that I/O is synchronous).
+ *
+ * ur_int_handler catches the "I/O done" interrupt, writes the
+ * subchannel status word into the scsw member of the urdev structure
+ * and complete()s the io_done to wake the waiting do_ur_io.
+ *
+ * The caller of do_ur_io is responsible for kfree()ing the channel program
+ * address pointer that alloc_chan_prog returned.
+ */
+
+
+/*
+ * alloc_chan_prog
+ * The channel program we use is write commands chained together
+ * with a final NOP CCW command-chained on (which ensures that CE and DE
+ * are presented together in a single interrupt instead of as separate
+ * interrupts unless an incorrect length indication kicks in first). The
+ * data length in each CCW is reclen. The caller must ensure that count
+ * is an integral multiple of reclen.
+ * The channel program pointer returned by this function must be freed
+ * with kfree. The caller is responsible for checking that
+ * count/reclen is not ridiculously large.
+ */
+static struct ccw1 *alloc_chan_prog(char *buf, size_t count, size_t reclen)
+{
+ size_t num_ccws;
+ struct ccw1 *cpa;
+ int i;
+
+ TRACE("alloc_chan_prog(%p, %zu, %zu)\n", buf, count, reclen);
+
+ /*
+ * We chain a NOP onto the writes to force CE+DE together.
+ * That means we allocate room for CCWs to cover count/reclen
+ * records plus a NOP.
+ */
+ num_ccws = count / reclen + 1;
+ cpa = kmalloc(num_ccws * sizeof(struct ccw1), GFP_KERNEL | GFP_DMA);
+ if (!cpa)
+ return NULL;
+
+ for (i = 0; count; i++) {
+ cpa[i].cmd_code = WRITE_CCW_CMD;
+ cpa[i].flags = CCW_FLAG_CC | CCW_FLAG_SLI;
+ cpa[i].count = reclen;
+ cpa[i].cda = __pa(buf);
+ buf += reclen;
+ count -= reclen;
+ }
+ /* The following NOP CCW forces CE+DE to be presented together */
+ cpa[i].cmd_code = CCW_CMD_NOOP;
+ cpa[i].flags = 0;
+ cpa[i].count = 0;
+ cpa[i].cda = 0;
+
+ return cpa;
+}
+
+static int do_ur_io(struct urdev *urd, struct ccw1 *cpa)
+{
+ int rc;
+ struct ccw_device *cdev = urd->cdev;
+ DECLARE_COMPLETION(event);
+
+ TRACE("do_ur_io: cpa=%p\n", cpa);
+
+ rc = mutex_lock_interruptible(&urd->io_mutex);
+ if (rc)
+ return rc;
+
+ urd->io_done = &event;
+
+ spin_lock_irq(get_ccwdev_lock(cdev));
+ rc = ccw_device_start(cdev, cpa, 1, 0, 0);
+ spin_unlock_irq(get_ccwdev_lock(cdev));
+
+ TRACE("do_ur_io: ccw_device_start returned %d\n", rc);
+ if (rc)
+ goto out;
+
+ wait_for_completion(&event);
+ TRACE("do_ur_io: I/O complete\n");
+ rc = 0;
+
+out:
+ mutex_unlock(&urd->io_mutex);
+ return rc;
+}
+
+/*
+ * ur interrupt handler, called from the ccw_device layer
+ */
+static void ur_int_handler(struct ccw_device *cdev, unsigned long intparm,
+ struct irb *irb)
+{
+ struct urdev *urd;
+
+ TRACE("ur_int_handler: intparm=0x%lx cstat=%02x dstat=%02x res=%u\n",
+ intparm, irb->scsw.cstat, irb->scsw.dstat, irb->scsw.count);
+
+ if (!intparm) {
+ TRACE("ur_int_handler: unsolicited interrupt\n");
+ return;
+ }
+ urd = cdev->dev.driver_data;
+ /* On special conditions irb is an error pointer */
+ if (IS_ERR(irb))
+ urd->io_request_rc = PTR_ERR(irb);
+ else if (irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
+ urd->io_request_rc = 0;
+ else
+ urd->io_request_rc = -EIO;
+
+ complete(urd->io_done);
+}
+
+/*
+ * reclen sysfs attribute - The record length to be used for write CCWs
+ */
+static ssize_t ur_attr_reclen_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct urdev *urd = dev->driver_data;
+
+ return sprintf(buf, "%zu\n", urd->reclen);
+}
+
+static DEVICE_ATTR(reclen, 0444, ur_attr_reclen_show, NULL);
+
+static int ur_create_attributes(struct device *dev)
+{
+ return device_create_file(dev, &dev_attr_reclen);
+}
+
+static void ur_remove_attributes(struct device *dev)
+{
+ device_remove_file(dev, &dev_attr_reclen);
+}
+
+/*
+ * diagnose code 0x210 - retrieve device information
+ * cc=0 normal completion, we have a real device
+ * cc=1 CP paging error
+ * cc=2 The virtual device exists, but is not associated with a real device
+ * cc=3 Invalid device address, or the virtual device does not exist
+ */
+static int get_urd_class(struct urdev *urd)
+{
+ static struct diag210 ur_diag210;
+ int cc;
+
+ ur_diag210.vrdcdvno = urd->dev_id.devno;
+ ur_diag210.vrdclen = sizeof(struct diag210);
+
+ cc = diag210(&ur_diag210);
+ switch (cc) {
+ case 0:
+ return -ENOTSUPP;
+ case 2:
+ return ur_diag210.vrdcvcla; /* virtual device class */
+ case 3:
+ return -ENODEV;
+ default:
+ return -EIO;
+ }
+}
+
+/*
+ * Allocation and freeing of urfile structures
+ */
+static struct urfile *urfile_alloc(struct urdev *urd)
+{
+ struct urfile *urf;
+
+ urf = kzalloc(sizeof(struct urfile), GFP_KERNEL);
+ if (!urf)
+ return NULL;
+ urf->urd = urd;
+
+ TRACE("urfile_alloc: urd=%p urf=%p rl=%zu\n", urd, urf,
+ urf->dev_reclen);
+
+ return urf;
+}
+
+static void urfile_free(struct urfile *urf)
+{
+ TRACE("urfile_free: urf=%p urd=%p\n", urf, urf->urd);
+ kfree(urf);
+}
+
+/*
+ * The fops implementation of the character device driver
+ */
+static ssize_t do_write(struct urdev *urd, const char __user *udata,
+ size_t count, size_t reclen, loff_t *ppos)
+{
+ struct ccw1 *cpa;
+ char *buf;
+ int rc;
+
+ /* Data buffer must be under 2GB line for fmt1 CCWs: hence GFP_DMA */
+ buf = kmalloc(count, GFP_KERNEL | GFP_DMA);
+ if (!buf)
+ return -ENOMEM;
+
+ if (copy_from_user(buf, udata, count)) {
+ rc = -EFAULT;
+ goto fail_kfree_buf;
+ }
+
+ cpa = alloc_chan_prog(buf, count, reclen);
+ if (!cpa) {
+ rc = -ENOMEM;
+ goto fail_kfree_buf;
+ }
+
+ rc = do_ur_io(urd, cpa);
+ if (rc)
+ goto fail_kfree_cpa;
+
+ if (urd->io_request_rc) {
+ rc = urd->io_request_rc;
+ goto fail_kfree_cpa;
+ }
+ *ppos += count;
+ rc = count;
+fail_kfree_cpa:
+ kfree(cpa);
+fail_kfree_buf:
+ kfree(buf);
+ return rc;
+}
+
+static ssize_t ur_write(struct file *file, const char __user *udata,
+ size_t count, loff_t *ppos)
+{
+ struct urfile *urf = file->private_data;
+
+ TRACE("ur_write: count=%zu\n", count);
+
+ if (count == 0)
+ return 0;
+
+ if (count % urf->dev_reclen)
+ return -EINVAL; /* count must be a multiple of reclen */
+
+ if (count > urf->dev_reclen * MAX_RECS_PER_IO)
+ count = urf->dev_reclen * MAX_RECS_PER_IO;
+
+ return do_write(urf->urd, udata, count, urf->dev_reclen, ppos);
+}
+
+static int do_diag_14(unsigned long rx, unsigned long ry1,
+ unsigned long subcode)
+{
+ register unsigned long _ry1 asm("2") = ry1;
+ register unsigned long _ry2 asm("3") = subcode;
+ int rc = 0;
+
+ asm volatile(
+#ifdef CONFIG_64BIT
+ " sam31\n"
+ " diag %2,2,0x14\n"
+ " sam64\n"
+#else
+ " diag %2,2,0x14\n"
+#endif
+ " ipm %0\n"
+ " srl %0,28\n"
+ : "=d" (rc), "+d" (_ry2)
+ : "d" (rx), "d" (_ry1)
+ : "cc");
+
+ TRACE("diag 14: subcode=0x%lx, cc=%i\n", subcode, rc);
+ return rc;
+}
+
+/*
+ * diagnose code 0x14 subcode 0x0028 - position spool file to designated
+ * record
+ * cc=0 normal completion
+ * cc=2 no file active on the virtual reader or device not ready
+ * cc=3 record specified is beyond EOF
+ */
+static int diag_position_to_record(int devno, int record)
+{
+ int cc;
+
+ cc = do_diag_14(record, devno, 0x28);
+ switch (cc) {
+ case 0:
+ return 0;
+ case 2:
+ return -ENOMEDIUM;
+ case 3:
+ return -ENODATA; /* position beyond end of file */
+ default:
+ return -EIO;
+ }
+}
+
+/*
+ * diagnose code 0x14 subcode 0x0000 - read next spool file buffer
+ * cc=0 normal completion
+ * cc=1 EOF reached
+ * cc=2 no file active on the virtual reader, and no file eligible
+ * cc=3 file already active on the virtual reader or specified virtual
+ * reader does not exist or is not a reader
+ */
+static int diag_read_file(int devno, char *buf)
+{
+ int cc;
+
+ cc = do_diag_14((unsigned long) buf, devno, 0x00);
+ switch (cc) {
+ case 0:
+ return 0;
+ case 1:
+ return -ENODATA;
+ case 2:
+ return -ENOMEDIUM;
+ default:
+ return -EIO;
+ }
+}
+
+static ssize_t diag14_read(struct file *file, char __user *ubuf, size_t count,
+ loff_t *offs)
+{
+ size_t len, copied, res;
+ char *buf;
+ int rc;
+ u16 reclen;
+ struct urdev *urd;
+
+ urd = ((struct urfile *) file->private_data)->urd;
+ reclen = ((struct urfile *) file->private_data)->file_reclen;
+
+ rc = diag_position_to_record(urd->dev_id.devno, *offs / PAGE_SIZE + 1);
+ if (rc == -ENODATA)
+ return 0;
+ if (rc)
+ return rc;
+
+ len = min((size_t) PAGE_SIZE, count);
+ buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!buf)
+ return -ENOMEM;
+
+ copied = 0;
+ res = (size_t) (*offs % PAGE_SIZE);
+ do {
+ rc = diag_read_file(urd->dev_id.devno, buf);
+ if (rc == -ENODATA) {
+ break;
+ }
+ if (rc)
+ goto fail;
+ if (reclen)
+ *((u16 *) &buf[FILE_RECLEN_OFFSET]) = reclen;
+ len = min(count - copied, PAGE_SIZE - res);
+ if (copy_to_user(ubuf + copied, buf + res, len)) {
+ rc = -EFAULT;
+ goto fail;
+ }
+ res = 0;
+ copied += len;
+ } while (copied != count);
+
+ *offs += copied;
+ rc = copied;
+fail:
+ kfree(buf);
+ return rc;
+}
+
+static ssize_t ur_read(struct file *file, char __user *ubuf, size_t count,
+ loff_t *offs)
+{
+ struct urdev *urd;
+ int rc;
+
+ TRACE("ur_read: count=%zu ppos=%li\n", count, (unsigned long) *offs);
+
+ if (count == 0)
+ return 0;
+
+ urd = ((struct urfile *) file->private_data)->urd;
+ rc = mutex_lock_interruptible(&urd->io_mutex);
+ if (rc)
+ return rc;
+ rc = diag14_read(file, ubuf, count, offs);
+ mutex_unlock(&urd->io_mutex);
+ return rc;
+}
+
+/*
+ * diagnose code 0x14 subcode 0x0fff - retrieve next file descriptor
+ * cc=0 normal completion
+ * cc=1 no files on reader queue or no subsequent file
+ * cc=2 spid specified is invalid
+ */
+static int diag_read_next_file_info(struct file_control_block *buf, int spid)
+{
+ int cc;
+
+ cc = do_diag_14((unsigned long) buf, spid, 0xfff);
+ switch (cc) {
+ case 0:
+ return 0;
+ default:
+ return -ENODATA;
+ }
+}
+
+static int verify_device(struct urdev *urd)
+{
+ struct file_control_block fcb;
+ char *buf;
+ int rc;
+
+ switch (urd->class) {
+ case DEV_CLASS_UR_O:
+ return 0; /* no check needed here */
+ case DEV_CLASS_UR_I:
+ /* check for empty reader device (beginning of chain) */
+ rc = diag_read_next_file_info(&fcb, 0);
+ if (rc)
+ return rc;
+
+ /* open file on virtual reader */
+ buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!buf)
+ return -ENOMEM;
+ rc = diag_read_file(urd->dev_id.devno, buf);
+ kfree(buf);
+
+ if ((rc != 0) && (rc != -ENODATA)) /* EOF does not hurt */
+ return rc;
+ return 0;
+ default:
+ return -ENOTSUPP;
+ }
+}
+
+static int get_file_reclen(struct urdev *urd)
+{
+ struct file_control_block fcb;
+ int rc;
+
+ switch (urd->class) {
+ case DEV_CLASS_UR_O:
+ return 0;
+ case DEV_CLASS_UR_I:
+ rc = diag_read_next_file_info(&fcb, 0);
+ if (rc)
+ return rc;
+ break;
+ default:
+ return -ENOTSUPP;
+ }
+ if (fcb.file_stat & FLG_CP_DUMP)
+ return 0;
+
+ return fcb.rec_len;
+}
+
+static int ur_open(struct inode *inode, struct file *file)
+{
+ u16 devno;
+ struct urdev *urd;
+ struct urfile *urf;
+ unsigned short accmode;
+ int rc;
+
+ accmode = file->f_flags & O_ACCMODE;
+
+ if (accmode == O_RDWR)
+ return -EACCES;
+
+ /*
+ * We treat the minor number as the devno of the ur device
+ * to find in the driver tree.
+ */
+ devno = MINOR(file->f_dentry->d_inode->i_rdev);
+
+ urd = urdev_get_from_devno(devno);
+ if (!urd)
+ return -ENXIO;
+
+ if (file->f_flags & O_NONBLOCK) {
+ if (!mutex_trylock(&urd->open_mutex)) {
+ rc = -EBUSY;
+ goto fail_put;
+ }
+ } else {
+ if (mutex_lock_interruptible(&urd->open_mutex)) {
+ rc = -ERESTARTSYS;
+ goto fail_put;
+ }
+ }
+
+ TRACE("ur_open\n");
+
+ if (((accmode == O_RDONLY) && (urd->class != DEV_CLASS_UR_I)) ||
+ ((accmode == O_WRONLY) && (urd->class != DEV_CLASS_UR_O))) {
+ TRACE("ur_open: unsupported dev class (%d)\n", urd->class);
+ rc = -EACCES;
+ goto fail_unlock;
+ }
+
+ rc = verify_device(urd);
+ if (rc)
+ goto fail_unlock;
+
+ urf = urfile_alloc(urd);
+ if (!urf) {
+ rc = -ENOMEM;
+ goto fail_unlock;
+ }
+
+ urf->dev_reclen = urd->reclen;
+ rc = get_file_reclen(urd);
+ if (rc < 0)
+ goto fail_urfile_free;
+ urf->file_reclen = rc;
+ file->private_data = urf;
+ return 0;
+
+fail_urfile_free:
+ urfile_free(urf);
+fail_unlock:
+ mutex_unlock(&urd->open_mutex);
+fail_put:
+ urdev_put(urd);
+ return rc;
+}
+
+static int ur_release(struct inode *inode, struct file *file)
+{
+ struct urfile *urf = file->private_data;
+
+ TRACE("ur_release\n");
+ mutex_unlock(&urf->urd->open_mutex);
+ urdev_put(urf->urd);
+ urfile_free(urf);
+ return 0;
+}
+
+static loff_t ur_llseek(struct file *file, loff_t offset, int whence)
+{
+ loff_t newpos;
+
+ if ((file->f_flags & O_ACCMODE) != O_RDONLY)
+ return -ESPIPE; /* seek allowed only for reader */
+ if (offset % PAGE_SIZE)
+ return -ESPIPE; /* only multiples of 4K allowed */
+ switch (whence) {
+ case 0: /* SEEK_SET */
+ newpos = offset;
+ break;
+ case 1: /* SEEK_CUR */
+ newpos = file->f_pos + offset;
+ break;
+ default:
+ return -EINVAL;
+ }
+ file->f_pos = newpos;
+ return newpos;
+}
+
+static struct file_operations ur_fops = {
+ .owner = THIS_MODULE,
+ .open = ur_open,
+ .release = ur_release,
+ .read = ur_read,
+ .write = ur_write,
+ .llseek = ur_llseek,
+};
+
+/*
+ * ccw_device infrastructure:
+ * ur_probe gets its own ref to the device (i.e. get_device),
+ * creates the struct urdev, the device attributes, sets up
+ * the interrupt handler and validates the virtual unit record device.
+ * ur_remove removes the device attributes, frees the struct urdev
+ * and drops (put_device) the ref to the device we got in ur_probe.
+ */
+static int ur_probe(struct ccw_device *cdev)
+{
+ struct urdev *urd;
+ int rc;
+
+ TRACE("ur_probe: cdev=%p state=%d\n", cdev, *(int *) cdev->private);
+
+ if (!get_device(&cdev->dev))
+ return -ENODEV;
+
+ urd = urdev_alloc(cdev);
+ if (!urd) {
+ rc = -ENOMEM;
+ goto fail;
+ }
+ rc = ur_create_attributes(&cdev->dev);
+ if (rc) {
+ rc = -ENOMEM;
+ goto fail;
+ }
+ cdev->dev.driver_data = urd;
+ cdev->handler = ur_int_handler;
+
+ /* validate virtual unit record device */
+ urd->class = get_urd_class(urd);
+ if (urd->class < 0) {
+ rc = urd->class;
+ goto fail;
+ }
+ if ((urd->class != DEV_CLASS_UR_I) && (urd->class != DEV_CLASS_UR_O)) {
+ rc = -ENOTSUPP;
+ goto fail;
+ }
+
+ return 0;
+
+fail:
+ urdev_free(urd);
+ put_device(&cdev->dev);
+ return rc;
+}
+
+static void ur_remove(struct ccw_device *cdev)
+{
+ struct urdev *urd = cdev->dev.driver_data;
+
+ TRACE("ur_remove\n");
+ if (cdev->online)
+ ur_set_offline(cdev);
+ ur_remove_attributes(&cdev->dev);
+ urdev_free(urd);
+ put_device(&cdev->dev);
+}
+
+static int ur_set_online(struct ccw_device *cdev)
+{
+ struct urdev *urd;
+ int minor, major, rc;
+ char node_id[16];
+
+ TRACE("ur_set_online: cdev=%p state=%d\n", cdev,
+ *(int *) cdev->private);
+
+ if (!try_module_get(ur_driver.owner))
+ return -EINVAL;
+
+ urd = (struct urdev *) cdev->dev.driver_data;
+ minor = urd->dev_id.devno;
+ major = MAJOR(ur_first_dev_maj_min);
+
+ urd->char_device = cdev_alloc();
+ if (!urd->char_device) {
+ rc = -ENOMEM;
+ goto fail_module_put;
+ }
+
+ cdev_init(urd->char_device, &ur_fops);
+ urd->char_device->dev = MKDEV(major, minor);
+ urd->char_device->owner = ur_fops.owner;
+
+ rc = cdev_add(urd->char_device, urd->char_device->dev, 1);
+ if (rc)
+ goto fail_free_cdev;
+ if (urd->cdev->id.cu_type == READER_PUNCH_DEVTYPE) {
+ if (urd->class == DEV_CLASS_UR_I)
+ sprintf(node_id, "vmrdr-%s", cdev->dev.bus_id);
+ if (urd->class == DEV_CLASS_UR_O)
+ sprintf(node_id, "vmpun-%s", cdev->dev.bus_id);
+ } else if (urd->cdev->id.cu_type == PRINTER_DEVTYPE) {
+ sprintf(node_id, "vmprt-%s", cdev->dev.bus_id);
+ } else {
+ rc = -ENOTSUPP;
+ goto fail_free_cdev;
+ }
+
+ urd->device = device_create(vmur_class, NULL, urd->char_device->dev,
+ "%s", node_id);
+ if (IS_ERR(urd->device)) {
+ rc = PTR_ERR(urd->device);
+ TRACE("ur_set_online: device_create rc=%d\n", rc);
+ goto fail_free_cdev;
+ }
+
+ return 0;
+
+fail_free_cdev:
+ cdev_del(urd->char_device);
+fail_module_put:
+ module_put(ur_driver.owner);
+
+ return rc;
+}
+
+static int ur_set_offline(struct ccw_device *cdev)
+{
+ struct urdev *urd;
+
+ TRACE("ur_set_offline: cdev=%p cdev->private=%p state=%d\n",
+ cdev, cdev->private, *(int *) cdev->private);
+ urd = (struct urdev *) cdev->dev.driver_data;
+ device_destroy(vmur_class, urd->char_device->dev);
+ cdev_del(urd->char_device);
+ module_put(ur_driver.owner);
+
+ return 0;
+}
+
+/*
+ * Module initialisation and cleanup
+ */
+static int __init ur_init(void)
+{
+ int rc;
+ dev_t dev;
+
+ if (!MACHINE_IS_VM) {
+ PRINT_ERR("%s is only available under z/VM.\n", ur_banner);
+ return -ENODEV;
+ }
+
+ vmur_dbf = debug_register("vmur", 4, 1, 4 * sizeof(long));
+ if (!vmur_dbf)
+ return -ENOMEM;
+ rc = debug_register_view(vmur_dbf, &debug_sprintf_view);
+ if (rc)
+ goto fail_free_dbf;
+
+ debug_set_level(vmur_dbf, 6);
+
+ rc = ccw_driver_register(&ur_driver);
+ if (rc)
+ goto fail_free_dbf;
+
+ rc = alloc_chrdev_region(&dev, 0, NUM_MINORS, "vmur");
+ if (rc) {
+ PRINT_ERR("alloc_chrdev_region failed: err = %d\n", rc);
+ goto fail_unregister_driver;
+ }
+ ur_first_dev_maj_min = MKDEV(MAJOR(dev), 0);
+
+ vmur_class = class_create(THIS_MODULE, "vmur");
+ if (IS_ERR(vmur_class)) {
+ rc = PTR_ERR(vmur_class);
+ goto fail_unregister_region;
+ }
+ PRINT_INFO("%s loaded.\n", ur_banner);
+ return 0;
+
+fail_unregister_region:
+ unregister_chrdev_region(ur_first_dev_maj_min, NUM_MINORS);
+fail_unregister_driver:
+ ccw_driver_unregister(&ur_driver);
+fail_free_dbf:
+ debug_unregister(vmur_dbf);
+ return rc;
+}
+
+static void __exit ur_exit(void)
+{
+ class_destroy(vmur_class);
+ unregister_chrdev_region(ur_first_dev_maj_min, NUM_MINORS);
+ ccw_driver_unregister(&ur_driver);
+ debug_unregister(vmur_dbf);
+ PRINT_INFO("%s unloaded.\n", ur_banner);
+}
+
+module_init(ur_init);
+module_exit(ur_exit);
OpenPOWER on IntegriCloud