Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / s390 / block / dasd_eer.c
diff --git a/kernel/drivers/s390/block/dasd_eer.c b/kernel/drivers/s390/block/dasd_eer.c
new file mode 100644 (file)
index 0000000..21ef63c
--- /dev/null
@@ -0,0 +1,709 @@
+/*
+ *  Character device driver for extended error reporting.
+ *
+ *  Copyright IBM Corp. 2005
+ *  extended error reporting for DASD ECKD devices
+ *  Author(s): Stefan Weinhuber <wein@de.ibm.com>
+ */
+
+#define KMSG_COMPONENT "dasd-eckd"
+
+#include <linux/init.h>
+#include <linux/fs.h>
+#include <linux/kernel.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/device.h>
+#include <linux/poll.h>
+#include <linux/mutex.h>
+#include <linux/err.h>
+#include <linux/slab.h>
+
+#include <asm/uaccess.h>
+#include <linux/atomic.h>
+#include <asm/ebcdic.h>
+
+#include "dasd_int.h"
+#include "dasd_eckd.h"
+
+#ifdef PRINTK_HEADER
+#undef PRINTK_HEADER
+#endif                         /* PRINTK_HEADER */
+#define PRINTK_HEADER "dasd(eer):"
+
+/*
+ * SECTION: the internal buffer
+ */
+
+/*
+ * The internal buffer is meant to store obaque blobs of data, so it does
+ * not know of higher level concepts like triggers.
+ * It consists of a number of pages that are used as a ringbuffer. Each data
+ * blob is stored in a simple record that consists of an integer, which
+ * contains the size of the following data, and the data bytes themselfes.
+ *
+ * To allow for multiple independent readers we create one internal buffer
+ * each time the device is opened and destroy the buffer when the file is
+ * closed again. The number of pages used for this buffer is determined by
+ * the module parmeter eer_pages.
+ *
+ * One record can be written to a buffer by using the functions
+ * - dasd_eer_start_record (one time per record to write the size to the
+ *                          buffer and reserve the space for the data)
+ * - dasd_eer_write_buffer (one or more times per record to write the data)
+ * The data can be written in several steps but you will have to compute
+ * the total size up front for the invocation of dasd_eer_start_record.
+ * If the ringbuffer is full, dasd_eer_start_record will remove the required
+ * number of old records.
+ *
+ * A record is typically read in two steps, first read the integer that
+ * specifies the size of the following data, then read the data.
+ * Both can be done by
+ * - dasd_eer_read_buffer
+ *
+ * For all mentioned functions you need to get the bufferlock first and keep
+ * it until a complete record is written or read.
+ *
+ * All information necessary to keep track of an internal buffer is kept in
+ * a struct eerbuffer. The buffer specific to a file pointer is strored in
+ * the private_data field of that file. To be able to write data to all
+ * existing buffers, each buffer is also added to the bufferlist.
+ * If the user does not want to read a complete record in one go, we have to
+ * keep track of the rest of the record. residual stores the number of bytes
+ * that are still to deliver. If the rest of the record is invalidated between
+ * two reads then residual will be set to -1 so that the next read will fail.
+ * All entries in the eerbuffer structure are protected with the bufferlock.
+ * To avoid races between writing to a buffer on the one side and creating
+ * and destroying buffers on the other side, the bufferlock must also be used
+ * to protect the bufferlist.
+ */
+
+static int eer_pages = 5;
+module_param(eer_pages, int, S_IRUGO|S_IWUSR);
+
+struct eerbuffer {
+       struct list_head list;
+       char **buffer;
+       int buffersize;
+       int buffer_page_count;
+       int head;
+        int tail;
+       int residual;
+};
+
+static LIST_HEAD(bufferlist);
+static DEFINE_SPINLOCK(bufferlock);
+static DECLARE_WAIT_QUEUE_HEAD(dasd_eer_read_wait_queue);
+
+/*
+ * How many free bytes are available on the buffer.
+ * Needs to be called with bufferlock held.
+ */
+static int dasd_eer_get_free_bytes(struct eerbuffer *eerb)
+{
+       if (eerb->head < eerb->tail)
+               return eerb->tail - eerb->head - 1;
+       return eerb->buffersize - eerb->head + eerb->tail -1;
+}
+
+/*
+ * How many bytes of buffer space are used.
+ * Needs to be called with bufferlock held.
+ */
+static int dasd_eer_get_filled_bytes(struct eerbuffer *eerb)
+{
+
+       if (eerb->head >= eerb->tail)
+               return eerb->head - eerb->tail;
+       return eerb->buffersize - eerb->tail + eerb->head;
+}
+
+/*
+ * The dasd_eer_write_buffer function just copies count bytes of data
+ * to the buffer. Make sure to call dasd_eer_start_record first, to
+ * make sure that enough free space is available.
+ * Needs to be called with bufferlock held.
+ */
+static void dasd_eer_write_buffer(struct eerbuffer *eerb,
+                                 char *data, int count)
+{
+
+       unsigned long headindex,localhead;
+       unsigned long rest, len;
+       char *nextdata;
+
+       nextdata = data;
+       rest = count;
+       while (rest > 0) {
+               headindex = eerb->head / PAGE_SIZE;
+               localhead = eerb->head % PAGE_SIZE;
+               len = min(rest, PAGE_SIZE - localhead);
+               memcpy(eerb->buffer[headindex]+localhead, nextdata, len);
+               nextdata += len;
+               rest -= len;
+               eerb->head += len;
+               if (eerb->head == eerb->buffersize)
+                       eerb->head = 0; /* wrap around */
+               BUG_ON(eerb->head > eerb->buffersize);
+       }
+}
+
+/*
+ * Needs to be called with bufferlock held.
+ */
+static int dasd_eer_read_buffer(struct eerbuffer *eerb, char *data, int count)
+{
+
+       unsigned long tailindex,localtail;
+       unsigned long rest, len, finalcount;
+       char *nextdata;
+
+       finalcount = min(count, dasd_eer_get_filled_bytes(eerb));
+       nextdata = data;
+       rest = finalcount;
+       while (rest > 0) {
+               tailindex = eerb->tail / PAGE_SIZE;
+               localtail = eerb->tail % PAGE_SIZE;
+               len = min(rest, PAGE_SIZE - localtail);
+               memcpy(nextdata, eerb->buffer[tailindex] + localtail, len);
+               nextdata += len;
+               rest -= len;
+               eerb->tail += len;
+               if (eerb->tail == eerb->buffersize)
+                       eerb->tail = 0; /* wrap around */
+               BUG_ON(eerb->tail > eerb->buffersize);
+       }
+       return finalcount;
+}
+
+/*
+ * Whenever you want to write a blob of data to the internal buffer you
+ * have to start by using this function first. It will write the number
+ * of bytes that will be written to the buffer. If necessary it will remove
+ * old records to make room for the new one.
+ * Needs to be called with bufferlock held.
+ */
+static int dasd_eer_start_record(struct eerbuffer *eerb, int count)
+{
+       int tailcount;
+
+       if (count + sizeof(count) > eerb->buffersize)
+               return -ENOMEM;
+       while (dasd_eer_get_free_bytes(eerb) < count + sizeof(count)) {
+               if (eerb->residual > 0) {
+                       eerb->tail += eerb->residual;
+                       if (eerb->tail >= eerb->buffersize)
+                               eerb->tail -= eerb->buffersize;
+                       eerb->residual = -1;
+               }
+               dasd_eer_read_buffer(eerb, (char *) &tailcount,
+                                    sizeof(tailcount));
+               eerb->tail += tailcount;
+               if (eerb->tail >= eerb->buffersize)
+                       eerb->tail -= eerb->buffersize;
+       }
+       dasd_eer_write_buffer(eerb, (char*) &count, sizeof(count));
+
+       return 0;
+};
+
+/*
+ * Release pages that are not used anymore.
+ */
+static void dasd_eer_free_buffer_pages(char **buf, int no_pages)
+{
+       int i;
+
+       for (i = 0; i < no_pages; i++)
+               free_page((unsigned long) buf[i]);
+}
+
+/*
+ * Allocate a new set of memory pages.
+ */
+static int dasd_eer_allocate_buffer_pages(char **buf, int no_pages)
+{
+       int i;
+
+       for (i = 0; i < no_pages; i++) {
+               buf[i] = (char *) get_zeroed_page(GFP_KERNEL);
+               if (!buf[i]) {
+                       dasd_eer_free_buffer_pages(buf, i);
+                       return -ENOMEM;
+               }
+       }
+       return 0;
+}
+
+/*
+ * SECTION: The extended error reporting functionality
+ */
+
+/*
+ * When a DASD device driver wants to report an error, it calls the
+ * function dasd_eer_write and gives the respective trigger ID as
+ * parameter. Currently there are four kinds of triggers:
+ *
+ * DASD_EER_FATALERROR:  all kinds of unrecoverable I/O problems
+ * DASD_EER_PPRCSUSPEND: PPRC was suspended
+ * DASD_EER_NOPATH:      There is no path to the device left.
+ * DASD_EER_STATECHANGE: The state of the device has changed.
+ *
+ * For the first three triggers all required information can be supplied by
+ * the caller. For these triggers a record is written by the function
+ * dasd_eer_write_standard_trigger.
+ *
+ * The DASD_EER_STATECHANGE trigger is special since a sense subsystem
+ * status ccw need to be executed to gather the necessary sense data first.
+ * The dasd_eer_snss function will queue the SNSS request and the request
+ * callback will then call dasd_eer_write with the DASD_EER_STATCHANGE
+ * trigger.
+ *
+ * To avoid memory allocations at runtime, the necessary memory is allocated
+ * when the extended error reporting is enabled for a device (by
+ * dasd_eer_probe). There is one sense subsystem status request for each
+ * eer enabled DASD device. The presence of the cqr in device->eer_cqr
+ * indicates that eer is enable for the device. The use of the snss request
+ * is protected by the DASD_FLAG_EER_IN_USE bit. When this flag indicates
+ * that the cqr is currently in use, dasd_eer_snss cannot start a second
+ * request but sets the DASD_FLAG_EER_SNSS flag instead. The callback of
+ * the SNSS request will check the bit and call dasd_eer_snss again.
+ */
+
+#define SNSS_DATA_SIZE 44
+
+#define DASD_EER_BUSID_SIZE 10
+struct dasd_eer_header {
+       __u32 total_size;
+       __u32 trigger;
+       __u64 tv_sec;
+       __u64 tv_usec;
+       char busid[DASD_EER_BUSID_SIZE];
+} __attribute__ ((packed));
+
+/*
+ * The following function can be used for those triggers that have
+ * all necessary data available when the function is called.
+ * If the parameter cqr is not NULL, the chain of requests will be searched
+ * for valid sense data, and all valid sense data sets will be added to
+ * the triggers data.
+ */
+static void dasd_eer_write_standard_trigger(struct dasd_device *device,
+                                           struct dasd_ccw_req *cqr,
+                                           int trigger)
+{
+       struct dasd_ccw_req *temp_cqr;
+       int data_size;
+       struct timeval tv;
+       struct dasd_eer_header header;
+       unsigned long flags;
+       struct eerbuffer *eerb;
+       char *sense;
+
+       /* go through cqr chain and count the valid sense data sets */
+       data_size = 0;
+       for (temp_cqr = cqr; temp_cqr; temp_cqr = temp_cqr->refers)
+               if (dasd_get_sense(&temp_cqr->irb))
+                       data_size += 32;
+
+       header.total_size = sizeof(header) + data_size + 4; /* "EOR" */
+       header.trigger = trigger;
+       do_gettimeofday(&tv);
+       header.tv_sec = tv.tv_sec;
+       header.tv_usec = tv.tv_usec;
+       strncpy(header.busid, dev_name(&device->cdev->dev),
+               DASD_EER_BUSID_SIZE);
+
+       spin_lock_irqsave(&bufferlock, flags);
+       list_for_each_entry(eerb, &bufferlist, list) {
+               dasd_eer_start_record(eerb, header.total_size);
+               dasd_eer_write_buffer(eerb, (char *) &header, sizeof(header));
+               for (temp_cqr = cqr; temp_cqr; temp_cqr = temp_cqr->refers) {
+                       sense = dasd_get_sense(&temp_cqr->irb);
+                       if (sense)
+                               dasd_eer_write_buffer(eerb, sense, 32);
+               }
+               dasd_eer_write_buffer(eerb, "EOR", 4);
+       }
+       spin_unlock_irqrestore(&bufferlock, flags);
+       wake_up_interruptible(&dasd_eer_read_wait_queue);
+}
+
+/*
+ * This function writes a DASD_EER_STATECHANGE trigger.
+ */
+static void dasd_eer_write_snss_trigger(struct dasd_device *device,
+                                       struct dasd_ccw_req *cqr,
+                                       int trigger)
+{
+       int data_size;
+       int snss_rc;
+       struct timeval tv;
+       struct dasd_eer_header header;
+       unsigned long flags;
+       struct eerbuffer *eerb;
+
+       snss_rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
+       if (snss_rc)
+               data_size = 0;
+       else
+               data_size = SNSS_DATA_SIZE;
+
+       header.total_size = sizeof(header) + data_size + 4; /* "EOR" */
+       header.trigger = DASD_EER_STATECHANGE;
+       do_gettimeofday(&tv);
+       header.tv_sec = tv.tv_sec;
+       header.tv_usec = tv.tv_usec;
+       strncpy(header.busid, dev_name(&device->cdev->dev),
+               DASD_EER_BUSID_SIZE);
+
+       spin_lock_irqsave(&bufferlock, flags);
+       list_for_each_entry(eerb, &bufferlist, list) {
+               dasd_eer_start_record(eerb, header.total_size);
+               dasd_eer_write_buffer(eerb, (char *) &header , sizeof(header));
+               if (!snss_rc)
+                       dasd_eer_write_buffer(eerb, cqr->data, SNSS_DATA_SIZE);
+               dasd_eer_write_buffer(eerb, "EOR", 4);
+       }
+       spin_unlock_irqrestore(&bufferlock, flags);
+       wake_up_interruptible(&dasd_eer_read_wait_queue);
+}
+
+/*
+ * This function is called for all triggers. It calls the appropriate
+ * function that writes the actual trigger records.
+ */
+void dasd_eer_write(struct dasd_device *device, struct dasd_ccw_req *cqr,
+                   unsigned int id)
+{
+       if (!device->eer_cqr)
+               return;
+       switch (id) {
+       case DASD_EER_FATALERROR:
+       case DASD_EER_PPRCSUSPEND:
+               dasd_eer_write_standard_trigger(device, cqr, id);
+               break;
+       case DASD_EER_NOPATH:
+               dasd_eer_write_standard_trigger(device, NULL, id);
+               break;
+       case DASD_EER_STATECHANGE:
+               dasd_eer_write_snss_trigger(device, cqr, id);
+               break;
+       default: /* unknown trigger, so we write it without any sense data */
+               dasd_eer_write_standard_trigger(device, NULL, id);
+               break;
+       }
+}
+EXPORT_SYMBOL(dasd_eer_write);
+
+/*
+ * Start a sense subsystem status request.
+ * Needs to be called with the device held.
+ */
+void dasd_eer_snss(struct dasd_device *device)
+{
+       struct dasd_ccw_req *cqr;
+
+       cqr = device->eer_cqr;
+       if (!cqr)       /* Device not eer enabled. */
+               return;
+       if (test_and_set_bit(DASD_FLAG_EER_IN_USE, &device->flags)) {
+               /* Sense subsystem status request in use. */
+               set_bit(DASD_FLAG_EER_SNSS, &device->flags);
+               return;
+       }
+       /* cdev is already locked, can't use dasd_add_request_head */
+       clear_bit(DASD_FLAG_EER_SNSS, &device->flags);
+       cqr->status = DASD_CQR_QUEUED;
+       list_add(&cqr->devlist, &device->ccw_queue);
+       dasd_schedule_device_bh(device);
+}
+
+/*
+ * Callback function for use with sense subsystem status request.
+ */
+static void dasd_eer_snss_cb(struct dasd_ccw_req *cqr, void *data)
+{
+       struct dasd_device *device = cqr->startdev;
+       unsigned long flags;
+
+       dasd_eer_write(device, cqr, DASD_EER_STATECHANGE);
+       spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+       if (device->eer_cqr == cqr) {
+               clear_bit(DASD_FLAG_EER_IN_USE, &device->flags);
+               if (test_bit(DASD_FLAG_EER_SNSS, &device->flags))
+                       /* Another SNSS has been requested in the meantime. */
+                       dasd_eer_snss(device);
+               cqr = NULL;
+       }
+       spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+       if (cqr)
+               /*
+                * Extended error recovery has been switched off while
+                * the SNSS request was running. It could even have
+                * been switched off and on again in which case there
+                * is a new ccw in device->eer_cqr. Free the "old"
+                * snss request now.
+                */
+               dasd_kfree_request(cqr, device);
+}
+
+/*
+ * Enable error reporting on a given device.
+ */
+int dasd_eer_enable(struct dasd_device *device)
+{
+       struct dasd_ccw_req *cqr;
+       unsigned long flags;
+       struct ccw1 *ccw;
+
+       if (device->eer_cqr)
+               return 0;
+
+       if (!device->discipline || strcmp(device->discipline->name, "ECKD"))
+               return -EPERM;  /* FIXME: -EMEDIUMTYPE ? */
+
+       cqr = dasd_kmalloc_request(DASD_ECKD_MAGIC, 1 /* SNSS */,
+                                  SNSS_DATA_SIZE, device);
+       if (IS_ERR(cqr))
+               return -ENOMEM;
+
+       cqr->startdev = device;
+       cqr->retries = 255;
+       cqr->expires = 10 * HZ;
+       clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
+       set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
+
+       ccw = cqr->cpaddr;
+       ccw->cmd_code = DASD_ECKD_CCW_SNSS;
+       ccw->count = SNSS_DATA_SIZE;
+       ccw->flags = 0;
+       ccw->cda = (__u32)(addr_t) cqr->data;
+
+       cqr->buildclk = get_tod_clock();
+       cqr->status = DASD_CQR_FILLED;
+       cqr->callback = dasd_eer_snss_cb;
+
+       spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+       if (!device->eer_cqr) {
+               device->eer_cqr = cqr;
+               cqr = NULL;
+       }
+       spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+       if (cqr)
+               dasd_kfree_request(cqr, device);
+       return 0;
+}
+
+/*
+ * Disable error reporting on a given device.
+ */
+void dasd_eer_disable(struct dasd_device *device)
+{
+       struct dasd_ccw_req *cqr;
+       unsigned long flags;
+       int in_use;
+
+       if (!device->eer_cqr)
+               return;
+       spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+       cqr = device->eer_cqr;
+       device->eer_cqr = NULL;
+       clear_bit(DASD_FLAG_EER_SNSS, &device->flags);
+       in_use = test_and_clear_bit(DASD_FLAG_EER_IN_USE, &device->flags);
+       spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+       if (cqr && !in_use)
+               dasd_kfree_request(cqr, device);
+}
+
+/*
+ * SECTION: the device operations
+ */
+
+/*
+ * On the one side we need a lock to access our internal buffer, on the
+ * other side a copy_to_user can sleep. So we need to copy the data we have
+ * to transfer in a readbuffer, which is protected by the readbuffer_mutex.
+ */
+static char readbuffer[PAGE_SIZE];
+static DEFINE_MUTEX(readbuffer_mutex);
+
+static int dasd_eer_open(struct inode *inp, struct file *filp)
+{
+       struct eerbuffer *eerb;
+       unsigned long flags;
+
+       eerb = kzalloc(sizeof(struct eerbuffer), GFP_KERNEL);
+       if (!eerb)
+               return -ENOMEM;
+       eerb->buffer_page_count = eer_pages;
+       if (eerb->buffer_page_count < 1 ||
+           eerb->buffer_page_count > INT_MAX / PAGE_SIZE) {
+               kfree(eerb);
+               DBF_EVENT(DBF_WARNING, "can't open device since module "
+                       "parameter eer_pages is smaller than 1 or"
+                       " bigger than %d", (int)(INT_MAX / PAGE_SIZE));
+               return -EINVAL;
+       }
+       eerb->buffersize = eerb->buffer_page_count * PAGE_SIZE;
+       eerb->buffer = kmalloc(eerb->buffer_page_count * sizeof(char *),
+                              GFP_KERNEL);
+        if (!eerb->buffer) {
+               kfree(eerb);
+                return -ENOMEM;
+       }
+       if (dasd_eer_allocate_buffer_pages(eerb->buffer,
+                                          eerb->buffer_page_count)) {
+               kfree(eerb->buffer);
+               kfree(eerb);
+               return -ENOMEM;
+       }
+       filp->private_data = eerb;
+       spin_lock_irqsave(&bufferlock, flags);
+       list_add(&eerb->list, &bufferlist);
+       spin_unlock_irqrestore(&bufferlock, flags);
+
+       return nonseekable_open(inp,filp);
+}
+
+static int dasd_eer_close(struct inode *inp, struct file *filp)
+{
+       struct eerbuffer *eerb;
+       unsigned long flags;
+
+       eerb = (struct eerbuffer *) filp->private_data;
+       spin_lock_irqsave(&bufferlock, flags);
+       list_del(&eerb->list);
+       spin_unlock_irqrestore(&bufferlock, flags);
+       dasd_eer_free_buffer_pages(eerb->buffer, eerb->buffer_page_count);
+       kfree(eerb->buffer);
+       kfree(eerb);
+
+       return 0;
+}
+
+static ssize_t dasd_eer_read(struct file *filp, char __user *buf,
+                            size_t count, loff_t *ppos)
+{
+       int tc,rc;
+       int tailcount,effective_count;
+        unsigned long flags;
+       struct eerbuffer *eerb;
+
+       eerb = (struct eerbuffer *) filp->private_data;
+       if (mutex_lock_interruptible(&readbuffer_mutex))
+               return -ERESTARTSYS;
+
+       spin_lock_irqsave(&bufferlock, flags);
+
+       if (eerb->residual < 0) { /* the remainder of this record */
+                                 /* has been deleted             */
+               eerb->residual = 0;
+               spin_unlock_irqrestore(&bufferlock, flags);
+               mutex_unlock(&readbuffer_mutex);
+               return -EIO;
+       } else if (eerb->residual > 0) {
+               /* OK we still have a second half of a record to deliver */
+               effective_count = min(eerb->residual, (int) count);
+               eerb->residual -= effective_count;
+       } else {
+               tc = 0;
+               while (!tc) {
+                       tc = dasd_eer_read_buffer(eerb, (char *) &tailcount,
+                                                 sizeof(tailcount));
+                       if (!tc) {
+                               /* no data available */
+                               spin_unlock_irqrestore(&bufferlock, flags);
+                               mutex_unlock(&readbuffer_mutex);
+                               if (filp->f_flags & O_NONBLOCK)
+                                       return -EAGAIN;
+                               rc = wait_event_interruptible(
+                                       dasd_eer_read_wait_queue,
+                                       eerb->head != eerb->tail);
+                               if (rc)
+                                       return rc;
+                               if (mutex_lock_interruptible(&readbuffer_mutex))
+                                       return -ERESTARTSYS;
+                               spin_lock_irqsave(&bufferlock, flags);
+                       }
+               }
+               WARN_ON(tc != sizeof(tailcount));
+               effective_count = min(tailcount,(int)count);
+               eerb->residual = tailcount - effective_count;
+       }
+
+       tc = dasd_eer_read_buffer(eerb, readbuffer, effective_count);
+       WARN_ON(tc != effective_count);
+
+       spin_unlock_irqrestore(&bufferlock, flags);
+
+       if (copy_to_user(buf, readbuffer, effective_count)) {
+               mutex_unlock(&readbuffer_mutex);
+               return -EFAULT;
+       }
+
+       mutex_unlock(&readbuffer_mutex);
+       return effective_count;
+}
+
+static unsigned int dasd_eer_poll(struct file *filp, poll_table *ptable)
+{
+       unsigned int mask;
+       unsigned long flags;
+       struct eerbuffer *eerb;
+
+       eerb = (struct eerbuffer *) filp->private_data;
+       poll_wait(filp, &dasd_eer_read_wait_queue, ptable);
+       spin_lock_irqsave(&bufferlock, flags);
+       if (eerb->head != eerb->tail)
+               mask = POLLIN | POLLRDNORM ;
+       else
+               mask = 0;
+       spin_unlock_irqrestore(&bufferlock, flags);
+       return mask;
+}
+
+static const struct file_operations dasd_eer_fops = {
+       .open           = &dasd_eer_open,
+       .release        = &dasd_eer_close,
+       .read           = &dasd_eer_read,
+       .poll           = &dasd_eer_poll,
+       .owner          = THIS_MODULE,
+       .llseek         = noop_llseek,
+};
+
+static struct miscdevice *dasd_eer_dev = NULL;
+
+int __init dasd_eer_init(void)
+{
+       int rc;
+
+       dasd_eer_dev = kzalloc(sizeof(*dasd_eer_dev), GFP_KERNEL);
+       if (!dasd_eer_dev)
+               return -ENOMEM;
+
+       dasd_eer_dev->minor = MISC_DYNAMIC_MINOR;
+       dasd_eer_dev->name  = "dasd_eer";
+       dasd_eer_dev->fops  = &dasd_eer_fops;
+
+       rc = misc_register(dasd_eer_dev);
+       if (rc) {
+               kfree(dasd_eer_dev);
+               dasd_eer_dev = NULL;
+               DBF_EVENT(DBF_ERR, "%s", "dasd_eer_init could not "
+                      "register misc device");
+               return rc;
+       }
+
+       return 0;
+}
+
+void dasd_eer_exit(void)
+{
+       if (dasd_eer_dev) {
+               misc_deregister(dasd_eer_dev);
+               kfree(dasd_eer_dev);
+               dasd_eer_dev = NULL;
+       }
+}