+
+/**
+ * ata_eh_detach_dev - detach ATA device
+ * @dev: ATA device to detach
+ *
+ * Detach @dev.
+ *
+ * LOCKING:
+ * None.
+ */
+static void ata_eh_detach_dev(struct ata_device *dev)
+{
+ struct ata_port *ap = dev->ap;
+ unsigned long flags;
+
+ ata_dev_disable(dev);
+
+ spin_lock_irqsave(ap->lock, flags);
+
+ dev->flags &= ~ATA_DFLAG_DETACH;
+
+ if (ata_scsi_offline_dev(dev)) {
+ dev->flags |= ATA_DFLAG_DETACHED;
+ ap->flags |= ATA_FLAG_SCSI_HOTPLUG;
+ }
+
+ /* clear per-dev EH actions */
+ ata_eh_clear_action(dev, &ap->eh_info, ATA_EH_PERDEV_MASK);
+ ata_eh_clear_action(dev, &ap->eh_context.i, ATA_EH_PERDEV_MASK);
+
+ spin_unlock_irqrestore(ap->lock, flags);
+}
+
+/**
+ * ata_eh_about_to_do - about to perform eh_action
+ * @ap: target ATA port
+ * @dev: target ATA dev for per-dev action (can be NULL)
+ * @action: action about to be performed
+ *
+ * Called just before performing EH actions to clear related bits
+ * in @ap->eh_info such that eh actions are not unnecessarily
+ * repeated.
+ *
+ * LOCKING:
+ * None.
+ */
+static void ata_eh_about_to_do(struct ata_port *ap, struct ata_device *dev,
+ unsigned int action)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(ap->lock, flags);
+ ata_eh_clear_action(dev, &ap->eh_info, action);
+ ap->flags |= ATA_FLAG_RECOVERED;
+ spin_unlock_irqrestore(ap->lock, flags);
+}
+
+/**
+ * ata_eh_done - EH action complete
+ * @ap: target ATA port
+ * @dev: target ATA dev for per-dev action (can be NULL)
+ * @action: action just completed
+ *
+ * Called right after performing EH actions to clear related bits
+ * in @ap->eh_context.
+ *
+ * LOCKING:
+ * None.
+ */
+static void ata_eh_done(struct ata_port *ap, struct ata_device *dev,
+ unsigned int action)
+{
+ ata_eh_clear_action(dev, &ap->eh_context.i, action);
+}
+
+/**
+ * ata_err_string - convert err_mask to descriptive string
+ * @err_mask: error mask to convert to string
+ *
+ * Convert @err_mask to descriptive string. Errors are
+ * prioritized according to severity and only the most severe
+ * error is reported.
+ *
+ * LOCKING:
+ * None.
+ *
+ * RETURNS:
+ * Descriptive string for @err_mask
+ */
+static const char * ata_err_string(unsigned int err_mask)
+{
+ if (err_mask & AC_ERR_HOST_BUS)
+ return "host bus error";
+ if (err_mask & AC_ERR_ATA_BUS)
+ return "ATA bus error";
+ if (err_mask & AC_ERR_TIMEOUT)
+ return "timeout";
+ if (err_mask & AC_ERR_HSM)
+ return "HSM violation";
+ if (err_mask & AC_ERR_SYSTEM)
+ return "internal error";
+ if (err_mask & AC_ERR_MEDIA)
+ return "media error";
+ if (err_mask & AC_ERR_INVALID)
+ return "invalid argument";
+ if (err_mask & AC_ERR_DEV)
+ return "device error";
+ return "unknown error";
+}
+
+/**
+ * ata_read_log_page - read a specific log page
+ * @dev: target device
+ * @page: page to read
+ * @buf: buffer to store read page
+ * @sectors: number of sectors to read
+ *
+ * Read log page using READ_LOG_EXT command.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ *
+ * RETURNS:
+ * 0 on success, AC_ERR_* mask otherwise.
+ */
+static unsigned int ata_read_log_page(struct ata_device *dev,
+ u8 page, void *buf, unsigned int sectors)
+{
+ struct ata_taskfile tf;
+ unsigned int err_mask;
+
+ DPRINTK("read log page - page %d\n", page);
+
+ ata_tf_init(dev, &tf);
+ tf.command = ATA_CMD_READ_LOG_EXT;
+ tf.lbal = page;
+ tf.nsect = sectors;
+ tf.hob_nsect = sectors >> 8;
+ tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_LBA48 | ATA_TFLAG_DEVICE;
+ tf.protocol = ATA_PROT_PIO;
+
+ err_mask = ata_exec_internal(dev, &tf, NULL, DMA_FROM_DEVICE,
+ buf, sectors * ATA_SECT_SIZE);
+
+ DPRINTK("EXIT, err_mask=%x\n", err_mask);
+ return err_mask;
+}
+
+/**
+ * ata_eh_read_log_10h - Read log page 10h for NCQ error details
+ * @dev: Device to read log page 10h from
+ * @tag: Resulting tag of the failed command
+ * @tf: Resulting taskfile registers of the failed command
+ *
+ * Read log page 10h to obtain NCQ error details and clear error
+ * condition.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ *
+ * RETURNS:
+ * 0 on success, -errno otherwise.
+ */
+static int ata_eh_read_log_10h(struct ata_device *dev,
+ int *tag, struct ata_taskfile *tf)
+{
+ u8 *buf = dev->ap->sector_buf;
+ unsigned int err_mask;
+ u8 csum;
+ int i;
+
+ err_mask = ata_read_log_page(dev, ATA_LOG_SATA_NCQ, buf, 1);
+ if (err_mask)
+ return -EIO;
+
+ csum = 0;
+ for (i = 0; i < ATA_SECT_SIZE; i++)
+ csum += buf[i];
+ if (csum)
+ ata_dev_printk(dev, KERN_WARNING,
+ "invalid checksum 0x%x on log page 10h\n", csum);
+
+ if (buf[0] & 0x80)
+ return -ENOENT;
+
+ *tag = buf[0] & 0x1f;
+
+ tf->command = buf[2];
+ tf->feature = buf[3];
+ tf->lbal = buf[4];
+ tf->lbam = buf[5];
+ tf->lbah = buf[6];
+ tf->device = buf[7];
+ tf->hob_lbal = buf[8];
+ tf->hob_lbam = buf[9];
+ tf->hob_lbah = buf[10];
+ tf->nsect = buf[12];
+ tf->hob_nsect = buf[13];
+
+ return 0;
+}
+
+/**
+ * atapi_eh_request_sense - perform ATAPI REQUEST_SENSE
+ * @dev: device to perform REQUEST_SENSE to
+ * @sense_buf: result sense data buffer (SCSI_SENSE_BUFFERSIZE bytes long)
+ *
+ * Perform ATAPI REQUEST_SENSE after the device reported CHECK
+ * SENSE. This function is EH helper.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ *
+ * RETURNS:
+ * 0 on success, AC_ERR_* mask on failure
+ */
+static unsigned int atapi_eh_request_sense(struct ata_device *dev,
+ unsigned char *sense_buf)
+{
+ struct ata_port *ap = dev->ap;
+ struct ata_taskfile tf;
+ u8 cdb[ATAPI_CDB_LEN];
+
+ DPRINTK("ATAPI request sense\n");
+
+ ata_tf_init(dev, &tf);
+
+ /* FIXME: is this needed? */
+ memset(sense_buf, 0, SCSI_SENSE_BUFFERSIZE);
+
+ /* XXX: why tf_read here? */
+ ap->ops->tf_read(ap, &tf);
+
+ /* fill these in, for the case where they are -not- overwritten */
+ sense_buf[0] = 0x70;
+ sense_buf[2] = tf.feature >> 4;
+
+ memset(cdb, 0, ATAPI_CDB_LEN);
+ cdb[0] = REQUEST_SENSE;
+ cdb[4] = SCSI_SENSE_BUFFERSIZE;
+
+ tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE;
+ tf.command = ATA_CMD_PACKET;
+
+ /* is it pointless to prefer PIO for "safety reasons"? */
+ if (ap->flags & ATA_FLAG_PIO_DMA) {
+ tf.protocol = ATA_PROT_ATAPI_DMA;
+ tf.feature |= ATAPI_PKT_DMA;
+ } else {
+ tf.protocol = ATA_PROT_ATAPI;
+ tf.lbam = (8 * 1024) & 0xff;
+ tf.lbah = (8 * 1024) >> 8;
+ }
+
+ return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
+ sense_buf, SCSI_SENSE_BUFFERSIZE);
+}
+
+/**
+ * ata_eh_analyze_serror - analyze SError for a failed port
+ * @ap: ATA port to analyze SError for
+ *
+ * Analyze SError if available and further determine cause of
+ * failure.
+ *
+ * LOCKING:
+ * None.
+ */
+static void ata_eh_analyze_serror(struct ata_port *ap)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ u32 serror = ehc->i.serror;
+ unsigned int err_mask = 0, action = 0;
+
+ if (serror & SERR_PERSISTENT) {
+ err_mask |= AC_ERR_ATA_BUS;
+ action |= ATA_EH_HARDRESET;
+ }
+ if (serror &
+ (SERR_DATA_RECOVERED | SERR_COMM_RECOVERED | SERR_DATA)) {
+ err_mask |= AC_ERR_ATA_BUS;
+ action |= ATA_EH_SOFTRESET;
+ }
+ if (serror & SERR_PROTOCOL) {
+ err_mask |= AC_ERR_HSM;
+ action |= ATA_EH_SOFTRESET;
+ }
+ if (serror & SERR_INTERNAL) {
+ err_mask |= AC_ERR_SYSTEM;
+ action |= ATA_EH_SOFTRESET;
+ }
+ if (serror & (SERR_PHYRDY_CHG | SERR_DEV_XCHG))
+ ata_ehi_hotplugged(&ehc->i);
+
+ ehc->i.err_mask |= err_mask;
+ ehc->i.action |= action;
+}
+
+/**
+ * ata_eh_analyze_ncq_error - analyze NCQ error
+ * @ap: ATA port to analyze NCQ error for
+ *
+ * Read log page 10h, determine the offending qc and acquire
+ * error status TF. For NCQ device errors, all LLDDs have to do
+ * is setting AC_ERR_DEV in ehi->err_mask. This function takes
+ * care of the rest.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ */
+static void ata_eh_analyze_ncq_error(struct ata_port *ap)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ struct ata_device *dev = ap->device;
+ struct ata_queued_cmd *qc;
+ struct ata_taskfile tf;
+ int tag, rc;
+
+ /* if frozen, we can't do much */
+ if (ap->flags & ATA_FLAG_FROZEN)
+ return;
+
+ /* is it NCQ device error? */
+ if (!ap->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
+ return;
+
+ /* has LLDD analyzed already? */
+ for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
+ qc = __ata_qc_from_tag(ap, tag);
+
+ if (!(qc->flags & ATA_QCFLAG_FAILED))
+ continue;
+
+ if (qc->err_mask)
+ return;
+ }
+
+ /* okay, this error is ours */
+ rc = ata_eh_read_log_10h(dev, &tag, &tf);
+ if (rc) {
+ ata_port_printk(ap, KERN_ERR, "failed to read log page 10h "
+ "(errno=%d)\n", rc);
+ return;
+ }
+
+ if (!(ap->sactive & (1 << tag))) {
+ ata_port_printk(ap, KERN_ERR, "log page 10h reported "
+ "inactive tag %d\n", tag);
+ return;
+ }
+
+ /* we've got the perpetrator, condemn it */
+ qc = __ata_qc_from_tag(ap, tag);
+ memcpy(&qc->result_tf, &tf, sizeof(tf));
+ qc->err_mask |= AC_ERR_DEV;
+ ehc->i.err_mask &= ~AC_ERR_DEV;
+}
+
+/**
+ * ata_eh_analyze_tf - analyze taskfile of a failed qc
+ * @qc: qc to analyze
+ * @tf: Taskfile registers to analyze
+ *
+ * Analyze taskfile of @qc and further determine cause of
+ * failure. This function also requests ATAPI sense data if
+ * avaliable.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ *
+ * RETURNS:
+ * Determined recovery action
+ */
+static unsigned int ata_eh_analyze_tf(struct ata_queued_cmd *qc,
+ const struct ata_taskfile *tf)
+{
+ unsigned int tmp, action = 0;
+ u8 stat = tf->command, err = tf->feature;
+
+ if ((stat & (ATA_BUSY | ATA_DRQ | ATA_DRDY)) != ATA_DRDY) {
+ qc->err_mask |= AC_ERR_HSM;
+ return ATA_EH_SOFTRESET;
+ }
+
+ if (!(qc->err_mask & AC_ERR_DEV))
+ return 0;
+
+ switch (qc->dev->class) {
+ case ATA_DEV_ATA:
+ if (err & ATA_ICRC)
+ qc->err_mask |= AC_ERR_ATA_BUS;
+ if (err & ATA_UNC)
+ qc->err_mask |= AC_ERR_MEDIA;
+ if (err & ATA_IDNF)
+ qc->err_mask |= AC_ERR_INVALID;
+ break;
+
+ case ATA_DEV_ATAPI:
+ tmp = atapi_eh_request_sense(qc->dev,
+ qc->scsicmd->sense_buffer);
+ if (!tmp) {
+ /* ATA_QCFLAG_SENSE_VALID is used to tell
+ * atapi_qc_complete() that sense data is
+ * already valid.
+ *
+ * TODO: interpret sense data and set
+ * appropriate err_mask.
+ */
+ qc->flags |= ATA_QCFLAG_SENSE_VALID;
+ } else
+ qc->err_mask |= tmp;
+ }
+
+ if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
+ action |= ATA_EH_SOFTRESET;
+
+ return action;
+}
+
+static int ata_eh_categorize_ering_entry(struct ata_ering_entry *ent)
+{
+ if (ent->err_mask & (AC_ERR_ATA_BUS | AC_ERR_TIMEOUT))
+ return 1;
+
+ if (ent->is_io) {
+ if (ent->err_mask & AC_ERR_HSM)
+ return 1;
+ if ((ent->err_mask &
+ (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
+ return 2;
+ }
+
+ return 0;
+}
+
+struct speed_down_needed_arg {
+ u64 since;
+ int nr_errors[3];
+};
+
+static int speed_down_needed_cb(struct ata_ering_entry *ent, void *void_arg)
+{
+ struct speed_down_needed_arg *arg = void_arg;
+
+ if (ent->timestamp < arg->since)
+ return -1;
+
+ arg->nr_errors[ata_eh_categorize_ering_entry(ent)]++;
+ return 0;
+}
+
+/**
+ * ata_eh_speed_down_needed - Determine wheter speed down is necessary
+ * @dev: Device of interest
+ *
+ * This function examines error ring of @dev and determines
+ * whether speed down is necessary. Speed down is necessary if
+ * there have been more than 3 of Cat-1 errors or 10 of Cat-2
+ * errors during last 15 minutes.
+ *
+ * Cat-1 errors are ATA_BUS, TIMEOUT for any command and HSM
+ * violation for known supported commands.
+ *
+ * Cat-2 errors are unclassified DEV error for known supported
+ * command.
+ *
+ * LOCKING:
+ * Inherited from caller.
+ *
+ * RETURNS:
+ * 1 if speed down is necessary, 0 otherwise
+ */
+static int ata_eh_speed_down_needed(struct ata_device *dev)
+{
+ const u64 interval = 15LLU * 60 * HZ;
+ static const int err_limits[3] = { -1, 3, 10 };
+ struct speed_down_needed_arg arg;
+ struct ata_ering_entry *ent;
+ int err_cat;
+ u64 j64;
+
+ ent = ata_ering_top(&dev->ering);
+ if (!ent)
+ return 0;
+
+ err_cat = ata_eh_categorize_ering_entry(ent);
+ if (err_cat == 0)
+ return 0;
+
+ memset(&arg, 0, sizeof(arg));
+
+ j64 = get_jiffies_64();
+ if (j64 >= interval)
+ arg.since = j64 - interval;
+ else
+ arg.since = 0;
+
+ ata_ering_map(&dev->ering, speed_down_needed_cb, &arg);
+
+ return arg.nr_errors[err_cat] > err_limits[err_cat];
+}
+
+/**
+ * ata_eh_speed_down - record error and speed down if necessary
+ * @dev: Failed device
+ * @is_io: Did the device fail during normal IO?
+ * @err_mask: err_mask of the error
+ *
+ * Record error and examine error history to determine whether
+ * adjusting transmission speed is necessary. It also sets
+ * transmission limits appropriately if such adjustment is
+ * necessary.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ *
+ * RETURNS:
+ * 0 on success, -errno otherwise
+ */
+static int ata_eh_speed_down(struct ata_device *dev, int is_io,
+ unsigned int err_mask)
+{
+ if (!err_mask)
+ return 0;
+
+ /* record error and determine whether speed down is necessary */
+ ata_ering_record(&dev->ering, is_io, err_mask);
+
+ if (!ata_eh_speed_down_needed(dev))
+ return 0;
+
+ /* speed down SATA link speed if possible */
+ if (sata_down_spd_limit(dev->ap) == 0)
+ return ATA_EH_HARDRESET;
+
+ /* lower transfer mode */
+ if (ata_down_xfermask_limit(dev, 0) == 0)
+ return ATA_EH_SOFTRESET;
+
+ ata_dev_printk(dev, KERN_ERR,
+ "speed down requested but no transfer mode left\n");
+ return 0;
+}
+
+/**
+ * ata_eh_autopsy - analyze error and determine recovery action
+ * @ap: ATA port to perform autopsy on
+ *
+ * Analyze why @ap failed and determine which recovery action is
+ * needed. This function also sets more detailed AC_ERR_* values
+ * and fills sense data for ATAPI CHECK SENSE.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ */
+static void ata_eh_autopsy(struct ata_port *ap)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ unsigned int action = ehc->i.action;
+ struct ata_device *failed_dev = NULL;
+ unsigned int all_err_mask = 0;
+ int tag, is_io = 0;
+ u32 serror;
+ int rc;
+
+ DPRINTK("ENTER\n");
+
+ /* obtain and analyze SError */
+ rc = sata_scr_read(ap, SCR_ERROR, &serror);
+ if (rc == 0) {
+ ehc->i.serror |= serror;
+ ata_eh_analyze_serror(ap);
+ } else if (rc != -EOPNOTSUPP)
+ action |= ATA_EH_HARDRESET;
+
+ /* analyze NCQ failure */
+ ata_eh_analyze_ncq_error(ap);
+
+ /* any real error trumps AC_ERR_OTHER */
+ if (ehc->i.err_mask & ~AC_ERR_OTHER)
+ ehc->i.err_mask &= ~AC_ERR_OTHER;
+
+ all_err_mask |= ehc->i.err_mask;
+
+ for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
+ struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
+
+ if (!(qc->flags & ATA_QCFLAG_FAILED))
+ continue;
+
+ /* inherit upper level err_mask */
+ qc->err_mask |= ehc->i.err_mask;
+
+ /* analyze TF */
+ action |= ata_eh_analyze_tf(qc, &qc->result_tf);
+
+ /* DEV errors are probably spurious in case of ATA_BUS error */
+ if (qc->err_mask & AC_ERR_ATA_BUS)
+ qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_MEDIA |
+ AC_ERR_INVALID);
+
+ /* any real error trumps unknown error */
+ if (qc->err_mask & ~AC_ERR_OTHER)
+ qc->err_mask &= ~AC_ERR_OTHER;
+
+ /* SENSE_VALID trumps dev/unknown error and revalidation */
+ if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
+ qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);
+ action &= ~ATA_EH_REVALIDATE;
+ }
+
+ /* accumulate error info */
+ failed_dev = qc->dev;
+ all_err_mask |= qc->err_mask;
+ if (qc->flags & ATA_QCFLAG_IO)
+ is_io = 1;
+ }
+
+ /* enforce default EH actions */
+ if (ap->flags & ATA_FLAG_FROZEN ||
+ all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
+ action |= ATA_EH_SOFTRESET;
+ else if (all_err_mask)
+ action |= ATA_EH_REVALIDATE;
+
+ /* if we have offending qcs and the associated failed device */
+ if (failed_dev) {
+ /* speed down */
+ action |= ata_eh_speed_down(failed_dev, is_io, all_err_mask);
+
+ /* perform per-dev EH action only on the offending device */
+ ehc->i.dev_action[failed_dev->devno] |=
+ action & ATA_EH_PERDEV_MASK;
+ action &= ~ATA_EH_PERDEV_MASK;
+ }
+
+ /* record autopsy result */
+ ehc->i.dev = failed_dev;
+ ehc->i.action = action;
+
+ DPRINTK("EXIT\n");
+}
+
+/**
+ * ata_eh_report - report error handling to user
+ * @ap: ATA port EH is going on
+ *
+ * Report EH to user.
+ *
+ * LOCKING:
+ * None.
+ */
+static void ata_eh_report(struct ata_port *ap)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ const char *frozen, *desc;
+ int tag, nr_failed = 0;
+
+ desc = NULL;
+ if (ehc->i.desc[0] != '\0')
+ desc = ehc->i.desc;
+
+ for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
+ struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
+
+ if (!(qc->flags & ATA_QCFLAG_FAILED))
+ continue;
+ if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
+ continue;
+
+ nr_failed++;
+ }
+
+ if (!nr_failed && !ehc->i.err_mask)
+ return;
+
+ frozen = "";
+ if (ap->flags & ATA_FLAG_FROZEN)
+ frozen = " frozen";
+
+ if (ehc->i.dev) {
+ ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
+ "SAct 0x%x SErr 0x%x action 0x%x%s\n",
+ ehc->i.err_mask, ap->sactive, ehc->i.serror,
+ ehc->i.action, frozen);
+ if (desc)
+ ata_dev_printk(ehc->i.dev, KERN_ERR, "(%s)\n", desc);
+ } else {
+ ata_port_printk(ap, KERN_ERR, "exception Emask 0x%x "
+ "SAct 0x%x SErr 0x%x action 0x%x%s\n",
+ ehc->i.err_mask, ap->sactive, ehc->i.serror,
+ ehc->i.action, frozen);
+ if (desc)
+ ata_port_printk(ap, KERN_ERR, "(%s)\n", desc);
+ }
+
+ for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
+ struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
+
+ if (!(qc->flags & ATA_QCFLAG_FAILED) || !qc->err_mask)
+ continue;
+
+ ata_dev_printk(qc->dev, KERN_ERR, "tag %d cmd 0x%x "
+ "Emask 0x%x stat 0x%x err 0x%x (%s)\n",
+ qc->tag, qc->tf.command, qc->err_mask,
+ qc->result_tf.command, qc->result_tf.feature,
+ ata_err_string(qc->err_mask));
+ }
+}
+
+static int ata_do_reset(struct ata_port *ap, ata_reset_fn_t reset,
+ unsigned int *classes)
+{
+ int i, rc;
+
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ classes[i] = ATA_DEV_UNKNOWN;
+
+ rc = reset(ap, classes);
+ if (rc)
+ return rc;
+
+ /* If any class isn't ATA_DEV_UNKNOWN, consider classification
+ * is complete and convert all ATA_DEV_UNKNOWN to
+ * ATA_DEV_NONE.
+ */
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ if (classes[i] != ATA_DEV_UNKNOWN)
+ break;
+
+ if (i < ATA_MAX_DEVICES)
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ if (classes[i] == ATA_DEV_UNKNOWN)
+ classes[i] = ATA_DEV_NONE;
+
+ return 0;
+}
+
+static int ata_eh_followup_srst_needed(int rc, int classify,
+ const unsigned int *classes)
+{
+ if (rc == -EAGAIN)
+ return 1;
+ if (rc != 0)
+ return 0;
+ if (classify && classes[0] == ATA_DEV_UNKNOWN)
+ return 1;
+ return 0;
+}
+
+static int ata_eh_reset(struct ata_port *ap, int classify,
+ ata_prereset_fn_t prereset, ata_reset_fn_t softreset,
+ ata_reset_fn_t hardreset, ata_postreset_fn_t postreset)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ unsigned int *classes = ehc->classes;
+ int tries = ATA_EH_RESET_TRIES;
+ int verbose = !(ap->flags & ATA_FLAG_LOADING);
+ unsigned int action;
+ ata_reset_fn_t reset;
+ int i, did_followup_srst, rc;
+
+ /* Determine which reset to use and record in ehc->i.action.
+ * prereset() may examine and modify it.
+ */
+ action = ehc->i.action;
+ ehc->i.action &= ~ATA_EH_RESET_MASK;
+ if (softreset && (!hardreset || (!sata_set_spd_needed(ap) &&
+ !(action & ATA_EH_HARDRESET))))
+ ehc->i.action |= ATA_EH_SOFTRESET;
+ else
+ ehc->i.action |= ATA_EH_HARDRESET;
+
+ if (prereset) {
+ rc = prereset(ap);
+ if (rc) {
+ ata_port_printk(ap, KERN_ERR,
+ "prereset failed (errno=%d)\n", rc);
+ return rc;
+ }
+ }
+
+ /* prereset() might have modified ehc->i.action */
+ if (ehc->i.action & ATA_EH_HARDRESET)
+ reset = hardreset;
+ else if (ehc->i.action & ATA_EH_SOFTRESET)
+ reset = softreset;
+ else {
+ /* prereset told us not to reset, bang classes and return */
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ classes[i] = ATA_DEV_NONE;
+ return 0;
+ }
+
+ /* did prereset() screw up? if so, fix up to avoid oopsing */
+ if (!reset) {
+ ata_port_printk(ap, KERN_ERR, "BUG: prereset() requested "
+ "invalid reset type\n");
+ if (softreset)
+ reset = softreset;
+ else
+ reset = hardreset;
+ }
+
+ retry:
+ /* shut up during boot probing */
+ if (verbose)
+ ata_port_printk(ap, KERN_INFO, "%s resetting port\n",
+ reset == softreset ? "soft" : "hard");
+
+ /* reset */
+ ata_eh_about_to_do(ap, NULL, ATA_EH_RESET_MASK);
+ ehc->i.flags |= ATA_EHI_DID_RESET;
+
+ rc = ata_do_reset(ap, reset, classes);
+
+ did_followup_srst = 0;
+ if (reset == hardreset &&
+ ata_eh_followup_srst_needed(rc, classify, classes)) {
+ /* okay, let's do follow-up softreset */
+ did_followup_srst = 1;
+ reset = softreset;
+
+ if (!reset) {
+ ata_port_printk(ap, KERN_ERR,
+ "follow-up softreset required "
+ "but no softreset avaliable\n");
+ return -EINVAL;
+ }
+
+ ata_eh_about_to_do(ap, NULL, ATA_EH_RESET_MASK);
+ rc = ata_do_reset(ap, reset, classes);
+
+ if (rc == 0 && classify &&
+ classes[0] == ATA_DEV_UNKNOWN) {
+ ata_port_printk(ap, KERN_ERR,
+ "classification failed\n");
+ return -EINVAL;
+ }
+ }
+
+ if (rc && --tries) {
+ const char *type;
+
+ if (reset == softreset) {
+ if (did_followup_srst)
+ type = "follow-up soft";
+ else
+ type = "soft";
+ } else
+ type = "hard";
+
+ ata_port_printk(ap, KERN_WARNING,
+ "%sreset failed, retrying in 5 secs\n", type);
+ ssleep(5);
+
+ if (reset == hardreset)
+ sata_down_spd_limit(ap);
+ if (hardreset)
+ reset = hardreset;
+ goto retry;
+ }
+
+ if (rc == 0) {
+ /* After the reset, the device state is PIO 0 and the
+ * controller state is undefined. Record the mode.
+ */
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ ap->device[i].pio_mode = XFER_PIO_0;
+
+ if (postreset)
+ postreset(ap, classes);
+
+ /* reset successful, schedule revalidation */
+ ata_eh_done(ap, NULL, ATA_EH_RESET_MASK);
+ ehc->i.action |= ATA_EH_REVALIDATE;
+ }
+
+ return rc;
+}
+
+static int ata_eh_revalidate_and_attach(struct ata_port *ap,
+ struct ata_device **r_failed_dev)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ struct ata_device *dev;
+ unsigned long flags;
+ int i, rc = 0;
+
+ DPRINTK("ENTER\n");
+
+ for (i = 0; i < ATA_MAX_DEVICES; i++) {
+ unsigned int action;
+
+ dev = &ap->device[i];
+ action = ata_eh_dev_action(dev);
+
+ if (action & ATA_EH_REVALIDATE && ata_dev_enabled(dev)) {
+ if (ata_port_offline(ap)) {
+ rc = -EIO;
+ break;
+ }
+
+ ata_eh_about_to_do(ap, dev, ATA_EH_REVALIDATE);
+ rc = ata_dev_revalidate(dev,
+ ehc->i.flags & ATA_EHI_DID_RESET);
+ if (rc)
+ break;
+
+ ata_eh_done(ap, dev, ATA_EH_REVALIDATE);
+
+ /* schedule the scsi_rescan_device() here */
+ queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
+ } else if (dev->class == ATA_DEV_UNKNOWN &&
+ ehc->tries[dev->devno] &&
+ ata_class_enabled(ehc->classes[dev->devno])) {
+ dev->class = ehc->classes[dev->devno];
+
+ rc = ata_dev_read_id(dev, &dev->class, 1, dev->id);
+ if (rc == 0)
+ rc = ata_dev_configure(dev, 1);
+
+ if (rc) {
+ dev->class = ATA_DEV_UNKNOWN;
+ break;
+ }
+
+ spin_lock_irqsave(ap->lock, flags);
+ ap->flags |= ATA_FLAG_SCSI_HOTPLUG;
+ spin_unlock_irqrestore(ap->lock, flags);
+ }
+ }
+
+ if (rc)
+ *r_failed_dev = dev;
+
+ DPRINTK("EXIT\n");
+ return rc;
+}
+
+static int ata_port_nr_enabled(struct ata_port *ap)
+{
+ int i, cnt = 0;
+
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ if (ata_dev_enabled(&ap->device[i]))
+ cnt++;
+ return cnt;
+}
+
+static int ata_port_nr_vacant(struct ata_port *ap)
+{
+ int i, cnt = 0;
+
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ if (ap->device[i].class == ATA_DEV_UNKNOWN)
+ cnt++;
+ return cnt;
+}
+
+static int ata_eh_skip_recovery(struct ata_port *ap)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ int i;
+
+ if (ap->flags & ATA_FLAG_FROZEN || ata_port_nr_enabled(ap))
+ return 0;
+
+ /* skip if class codes for all vacant slots are ATA_DEV_NONE */
+ for (i = 0; i < ATA_MAX_DEVICES; i++) {
+ struct ata_device *dev = &ap->device[i];
+
+ if (dev->class == ATA_DEV_UNKNOWN &&
+ ehc->classes[dev->devno] != ATA_DEV_NONE)
+ return 0;
+ }
+
+ return 1;
+}
+
+/**
+ * ata_eh_recover - recover host port after error
+ * @ap: host port to recover
+ * @prereset: prereset method (can be NULL)
+ * @softreset: softreset method (can be NULL)
+ * @hardreset: hardreset method (can be NULL)
+ * @postreset: postreset method (can be NULL)
+ *
+ * This is the alpha and omega, eum and yang, heart and soul of
+ * libata exception handling. On entry, actions required to
+ * recover the port and hotplug requests are recorded in
+ * eh_context. This function executes all the operations with
+ * appropriate retrials and fallbacks to resurrect failed
+ * devices, detach goners and greet newcomers.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ *
+ * RETURNS:
+ * 0 on success, -errno on failure.
+ */
+static int ata_eh_recover(struct ata_port *ap, ata_prereset_fn_t prereset,
+ ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
+ ata_postreset_fn_t postreset)
+{
+ struct ata_eh_context *ehc = &ap->eh_context;
+ struct ata_device *dev;
+ int down_xfermask, i, rc;
+
+ DPRINTK("ENTER\n");
+
+ /* prep for recovery */
+ for (i = 0; i < ATA_MAX_DEVICES; i++) {
+ dev = &ap->device[i];
+
+ ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
+
+ /* process hotplug request */
+ if (dev->flags & ATA_DFLAG_DETACH)
+ ata_eh_detach_dev(dev);
+
+ if (!ata_dev_enabled(dev) &&
+ ((ehc->i.probe_mask & (1 << dev->devno)) &&
+ !(ehc->did_probe_mask & (1 << dev->devno)))) {
+ ata_eh_detach_dev(dev);
+ ata_dev_init(dev);
+ ehc->did_probe_mask |= (1 << dev->devno);
+ ehc->i.action |= ATA_EH_SOFTRESET;
+ }
+ }
+
+ retry:
+ down_xfermask = 0;
+ rc = 0;
+
+ /* if UNLOADING, finish immediately */
+ if (ap->flags & ATA_FLAG_UNLOADING)
+ goto out;
+
+ /* skip EH if possible. */
+ if (ata_eh_skip_recovery(ap))
+ ehc->i.action = 0;
+
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ ehc->classes[i] = ATA_DEV_UNKNOWN;
+
+ /* reset */
+ if (ehc->i.action & ATA_EH_RESET_MASK) {
+ ata_eh_freeze_port(ap);
+
+ rc = ata_eh_reset(ap, ata_port_nr_vacant(ap), prereset,
+ softreset, hardreset, postreset);
+ if (rc) {
+ ata_port_printk(ap, KERN_ERR,
+ "reset failed, giving up\n");
+ goto out;
+ }
+
+ ata_eh_thaw_port(ap);
+ }
+
+ /* revalidate existing devices and attach new ones */
+ rc = ata_eh_revalidate_and_attach(ap, &dev);
+ if (rc)
+ goto dev_fail;
+
+ /* configure transfer mode if the port has been reset */
+ if (ehc->i.flags & ATA_EHI_DID_RESET) {
+ rc = ata_set_mode(ap, &dev);
+ if (rc) {
+ down_xfermask = 1;
+ goto dev_fail;
+ }
+ }
+
+ goto out;
+
+ dev_fail:
+ switch (rc) {
+ case -ENODEV:
+ /* device missing, schedule probing */
+ ehc->i.probe_mask |= (1 << dev->devno);
+ case -EINVAL:
+ ehc->tries[dev->devno] = 0;
+ break;
+ case -EIO:
+ sata_down_spd_limit(ap);
+ default:
+ ehc->tries[dev->devno]--;
+ if (down_xfermask &&
+ ata_down_xfermask_limit(dev, ehc->tries[dev->devno] == 1))
+ ehc->tries[dev->devno] = 0;
+ }
+
+ if (ata_dev_enabled(dev) && !ehc->tries[dev->devno]) {
+ /* disable device if it has used up all its chances */
+ ata_dev_disable(dev);
+
+ /* detach if offline */
+ if (ata_port_offline(ap))
+ ata_eh_detach_dev(dev);
+
+ /* probe if requested */
+ if ((ehc->i.probe_mask & (1 << dev->devno)) &&
+ !(ehc->did_probe_mask & (1 << dev->devno))) {
+ ata_eh_detach_dev(dev);
+ ata_dev_init(dev);
+
+ ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
+ ehc->did_probe_mask |= (1 << dev->devno);
+ ehc->i.action |= ATA_EH_SOFTRESET;
+ }
+ } else {
+ /* soft didn't work? be haaaaard */
+ if (ehc->i.flags & ATA_EHI_DID_RESET)
+ ehc->i.action |= ATA_EH_HARDRESET;
+ else
+ ehc->i.action |= ATA_EH_SOFTRESET;
+ }
+
+ if (ata_port_nr_enabled(ap)) {
+ ata_port_printk(ap, KERN_WARNING, "failed to recover some "
+ "devices, retrying in 5 secs\n");
+ ssleep(5);
+ } else {
+ /* no device left, repeat fast */
+ msleep(500);
+ }
+
+ goto retry;
+
+ out:
+ if (rc) {
+ for (i = 0; i < ATA_MAX_DEVICES; i++)
+ ata_dev_disable(&ap->device[i]);
+ }
+
+ DPRINTK("EXIT, rc=%d\n", rc);
+ return rc;
+}
+
+/**
+ * ata_eh_finish - finish up EH
+ * @ap: host port to finish EH for
+ *
+ * Recovery is complete. Clean up EH states and retry or finish
+ * failed qcs.
+ *
+ * LOCKING:
+ * None.
+ */
+static void ata_eh_finish(struct ata_port *ap)
+{
+ int tag;
+
+ /* retry or finish qcs */
+ for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
+ struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
+
+ if (!(qc->flags & ATA_QCFLAG_FAILED))
+ continue;
+
+ if (qc->err_mask) {
+ /* FIXME: Once EH migration is complete,
+ * generate sense data in this function,
+ * considering both err_mask and tf.
+ */
+ if (qc->err_mask & AC_ERR_INVALID)
+ ata_eh_qc_complete(qc);
+ else
+ ata_eh_qc_retry(qc);
+ } else {
+ if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
+ ata_eh_qc_complete(qc);
+ } else {
+ /* feed zero TF to sense generation */
+ memset(&qc->result_tf, 0, sizeof(qc->result_tf));
+ ata_eh_qc_retry(qc);
+ }
+ }
+ }
+}
+
+/**
+ * ata_do_eh - do standard error handling
+ * @ap: host port to handle error for
+ * @prereset: prereset method (can be NULL)
+ * @softreset: softreset method (can be NULL)
+ * @hardreset: hardreset method (can be NULL)
+ * @postreset: postreset method (can be NULL)
+ *
+ * Perform standard error handling sequence.
+ *
+ * LOCKING:
+ * Kernel thread context (may sleep).
+ */
+void ata_do_eh(struct ata_port *ap, ata_prereset_fn_t prereset,
+ ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
+ ata_postreset_fn_t postreset)
+{
+ if (!(ap->flags & ATA_FLAG_LOADING)) {
+ ata_eh_autopsy(ap);
+ ata_eh_report(ap);
+ }
+
+ ata_eh_recover(ap, prereset, softreset, hardreset, postreset);
+ ata_eh_finish(ap);
+}