pdc.c

/*
 * Promise FastTrak ATARAID metadata format handler.
 *
 * Copyright (C) 2004,2005  Heinz Mauelshagen, Red Hat GmbH.
 *                          All rights reserved.
 *
 * See file LICENSE at the top of this source tree for license information.
 */

/*
 * pdc_read() and pdc_group() profited from
 * Carl-Daniel Hailfinger's raiddetect code.
 */
#define     HANDLER     "pdc"

#include "internal.h"
#define     FORMAT_HANDLER
#include "pdc.h"

#if   BYTE_ORDER != LITTLE_ENDIAN
#  define   DM_BYTEORDER_SWAB
#  include  <datastruct/byteorder.h>
#endif

static const char *handler = HANDLER;

/*
 * Make up Promise RAID device name.
 */
static unsigned int set_number(struct pdc *pdc)
{
      return pdc->raid.disk_number >= (pdc->raid.total_disks / 2);
}

static size_t _name(struct pdc *pdc, char *str, size_t len, int subset)
{
      return snprintf(str, len, subset ? "pdc_%u-%u" : "pdc_%u",
                  pdc->raid.magic_1, set_number(pdc));
}

static char *name(struct lib_context *lc, struct raid_dev *rd,
              unsigned int subset)
{
      struct pdc *pdc = META(rd, pdc);
      size_t len;
      char *ret = NULL;

      if ((ret = dbg_malloc((len = _name(pdc, ret, 0, subset) + 1)))) {
            _name(pdc, ret, len, subset);
            mk_alpha(lc, ret + HANDLER_LEN,
                   len - HANDLER_LEN - (subset ? 2 : 0));
      } else
            log_alloc_err(lc, handler);

      return ret;
}

/*
 * Retrieve status of device.
 *
 * FIXME: need to identify state definitions.
 */
static enum status status(struct pdc *pdc)
{
      return PDC_BROKEN(pdc) ? s_broken : s_ok;
}

#define     PDC_T_RAID10      0x2   /* Not defind by Promise (yet). */
static int is_raid10(struct pdc *pdc)
{
      return pdc->raid.type == PDC_T_RAID10 ||
             (pdc->raid.type == PDC_T_RAID1 && pdc->raid.total_disks > 3);
}

/* Neutralize disk type */
static enum type type(struct pdc *pdc)
{
      /* Mapping of Promise types to generic types. */
      static struct types types[] = {
              { PDC_T_SPAN,   t_linear},
              { PDC_T_RAID0,  t_raid0},
              { PDC_T_RAID1,  t_raid1},
              { PDC_T_RAID10, t_raid0},
              { 0, t_undef}
      };

      if (is_raid10(pdc))
            pdc->raid.type = PDC_T_RAID10;

      return rd_type(types, (unsigned int) pdc->raid.type);
}

/* Calculate checksum on Promise metadata. */
static uint32_t checksum(struct pdc *pdc)
{
      unsigned int i = 511, sum = 0;
      uint32_t *p = (uint32_t*) pdc;

      while (i--)
            sum += *p++;

      return sum == pdc->checksum;
}

/* Calculate metadata offset. */
/*
 * Read a Promise FastTrak RAID device
 */
/* Endianess conversion. */
#if   BYTE_ORDER == LITTLE_ENDIAN
#  define   to_cpu      NULL
#else
static void to_cpu(void *meta)
{
      struct pdc *pdc = meta;
      struct pdc_disk *disk;

      CVT32(pdc->magic_0);
      CVT32(pdc->magic_1);
      CVT32(pdc->raid.flags);
      CVT32(pdc->raid.magic_0);
      CVT32(pdc->raid.disk_secs);
      CVT32(pdc->raid.total_secs);
      CVT16(pdc->raid.cylinders);
      CVT32(pdc->raid.magic_1);

      for (disk = pdc->raid.disk;
           disk < pdc->raid.disk + pdc->raid.total_disks;
           disk++) {
            CVT32(disk->magic_0);
            CVT32(disk->disk_number);
      }
}
#endif

/* Read and try to discover Promise signature. */
static void *pdc_read_metadata(struct lib_context *lc, struct dev_info *di,
                         size_t *size, uint64_t *offset,
                         union read_info *info)
{
      /* Assume certain sectors off the end of the RAID device. */
      static unsigned int *s, sectors[] = {
            PDC_CONFIGOFFSETS, 0,
      };
      struct pdc *ret;

      if ((ret = alloc_private(lc, handler, sizeof(*ret)))) {
            for (s = sectors; *s; s++) {
                  if (read_file(lc, handler, di->path, ret, sizeof(*ret),
                              (di->sectors - *s) << 9) &&
                      !strncmp((const char*) ret->promise_id, PDC_MAGIC,
                             PDC_ID_LENGTH)) {
                        info->u32 = *s;
                        return (void*) ret;
                  }
            }

            dbg_free(ret);
      }

      return NULL;
}

/* Magic check. */
static int is_pdc(struct lib_context *lc, struct dev_info *di, void *meta)
{
      struct pdc *pdc = meta;

      /*
       * No we've got the PDC magic string veryfied, we can
       * check that the rest of the metadata is valid.
       */
      if (pdc->raid.total_disks &&
          pdc->raid.total_disks < PDC_MAXDISKS)
            return 1;


      LOG_ERR(lc, 0, "%s: identifying %s, magic_0: 0x%x/0x%x, "
            "magic_1: 0x%x/0x%x, total_disks: %u",
            handler, di->path,
            pdc->magic_0, pdc->raid.magic_0,
            pdc->magic_1, pdc->raid.magic_1, pdc->raid.total_disks);
}

static int setup_rd(struct lib_context *lc, struct raid_dev *rd,
                struct dev_info *di, void *meta, union read_info *info);
static struct raid_dev *pdc_read(struct lib_context *lc, struct dev_info *di)
{ 
      return read_raid_dev(lc, di, pdc_read_metadata, 0, 0, to_cpu, is_pdc,
                       NULL, setup_rd, handler);
}

/*
 * Write a Promise FastTrak RAID device.
 */
static int pdc_write(struct lib_context *lc, struct raid_dev *rd, int erase)
{
      int ret;
#if   BYTE_ORDER != LITTLE_ENDIAN
      struct pdc *pdc = META(rd, pdc);

      to_disk(pdc);
#endif
      ret = write_metadata(lc, handler, rd, -1, erase);
#if   BYTE_ORDER != LITTLE_ENDIAN
      to_cpu(pdc);
#endif
      return ret;
}

/* Decide about ordering sequence of RAID device. */
static int dev_sort(struct list_head *pos, struct list_head *new)
{
      return (META(RD(new), pdc))->raid.disk_number <
             (META(RD(pos), pdc))->raid.disk_number;
}

/* Decide about ordering sequence of RAID subset. */
static int set_sort(struct list_head *pos, struct list_head *new)
{
      return !set_number(META(RD_RS(RS(new)), pdc));
}

/*
 * Group the RAID disk into a Promise set.
 */
static unsigned int stride(struct pdc *pdc)
{
      return pdc->raid.raid0_shift ? 1 << pdc->raid.raid0_shift : 0;
}

static void super_created(struct raid_set *super, void *private)
{
      super->type   = t_raid1;
      super->stride = stride(META((private), pdc));
}

/* Add a PDC RAID device to a set. */
static int group_rd(struct lib_context *lc, struct raid_set *rs,
                struct raid_set **ss, struct raid_dev *rd)
{
      struct pdc *pdc = META(rd, pdc);

      if (!init_raid_set(lc, rs, rd, stride(pdc), pdc->raid.type, handler))
            return 0;

      list_add_sorted(lc, &rs->devs, &rd->devs, dev_sort);

      switch (pdc->raid.type) {
      case PDC_T_SPAN:
      case PDC_T_RAID0:
      case PDC_T_RAID1:
            if (!find_set(lc, NULL, rs->name, FIND_TOP))
                  list_add_tail(&rs->list, LC_RS(lc));

            break;

      case PDC_T_RAID10:
            /*
             * RAID10:
             *
             * We've got a striped raid set with a mirror on top
             * when we get here.
             * Let's find and optionally allocate the mirror set on top.
             */
            if (!(*ss = join_superset(lc, name, super_created,
                                set_sort, rs, rd)))
                  return 0;
      }

      return 1;
}

static struct raid_set *pdc_group(struct lib_context *lc, struct raid_dev *rd)
{
      struct raid_set *rs, *ss = NULL;

      if (T_SPARE(rd))
            return NULL;

      if ((rs = find_or_alloc_raid_set(lc, rd->name, FIND_ALL, rd,
                               NO_LIST, NO_CREATE, NO_CREATE_ARG)))
            return group_rd(lc, rs, &ss, rd) ? (ss ? ss : rs) : NULL;

      return NULL;
}

/*
 * Check a PDC RAID set.
 *
 * FIXME: more sanity checks.
 */
static unsigned int devices(struct raid_dev *rd, void *context)
{
      struct pdc *pdc = META(rd, pdc);

      if (context && pdc->raid.type != PDC_T_SPAN)
            *((uint64_t*) context) += rd->sectors;

      return pdc->raid.total_disks;
}

static int check_rd(struct lib_context *lc, struct raid_set *rs,
                struct raid_dev *rd, void *context)
{
      return *((uint64_t*) context) >= (META(rd, pdc))->raid.total_secs;
}

static int pdc_check(struct lib_context *lc, struct raid_set *rs)
{
      uint64_t total_secs = 0;

      /*
       * Calculate total_secs in 1st check_raid_set() run and
       * check the device sizes against that in the 2nd run.
       */
      return check_raid_set(lc, rs, devices, &total_secs,
                        NO_CHECK_RD, NULL, handler) &&
             check_raid_set(lc, rs, devices, NULL,
                        check_rd, &total_secs, handler);
}

/*
 * IO error event handler.
 */
static int event_io(struct lib_context *lc, struct event_io *e_io)
{
      struct raid_dev *rd = e_io->rd;
      struct pdc *pdc = META(rd, pdc);

      /* Avoid write trashing. */
      if (status(pdc) & s_broken)
            return 0;

      PDC_SET_BROKEN(pdc);

      return 1;
}

static struct event_handlers pdc_event_handlers = {
      .io = event_io,
      .rd = NULL, /* FIXME: no device add/remove event handler yet. */
};

#ifdef DMRAID_NATIVE_LOG
/* Log native information about a Promise RAID device. */
static void pdc_log(struct lib_context *lc, struct raid_dev *rd)
{
      unsigned int i;
      struct pdc *pdc = META(rd, pdc);
      struct pdc_disk *disk;

      log_print(lc, "%s (%s):", rd->di->path, handler);
      DP("promise_id: \"%s\"", pdc, pdc->promise_id);
      DP("unknown_0: 0x%x", pdc, pdc->unknown_0);
      DP("magic_0: 0x%x", pdc, pdc->magic_0);
      DP("unknown_1: 0x%x", pdc, pdc->unknown_1);
      DP("magic_1: 0x%x", pdc, pdc->magic_1);
      DP("unknown_2: 0x%x", pdc, pdc->unknown_2);
      DP("raid.flags: 0x%x", pdc, pdc->raid.flags);
      DP("raid.unknown_0: 0x%x", pdc, pdc->raid.unknown_0);
      DP("raid.disk_number: %u", pdc, pdc->raid.disk_number);
      DP("raid.channel: %u", pdc, pdc->raid.channel);
      DP("raid.device: %u", pdc, pdc->raid.device);
      DP("raid.magic_0: 0x%x", pdc, pdc->raid.magic_0);
      DP("raid.unknown_1: 0x%x", pdc, pdc->raid.unknown_1);
      DP("raid.unknown_2: 0x%x", pdc, pdc->raid.unknown_2);
      DP("raid.disk_secs: %u", pdc, pdc->raid.disk_secs);
      DP("raid.unknown_3: 0x%x", pdc, pdc->raid.unknown_3);
      DP("raid.unknown_4: 0x%x", pdc, pdc->raid.unknown_4);
      DP("raid.status: 0x%x", pdc, pdc->raid.status);
      DP("raid.type: 0x%x", pdc, pdc->raid.type);
      DP("raid.total_disks: %u", pdc, pdc->raid.total_disks);
      DP("raid.raid0_shift: %u", pdc, pdc->raid.raid0_shift);
      DP("raid.raid0_disks: %u", pdc, pdc->raid.raid0_disks);
      DP("raid.array_number: %u", pdc, pdc->raid.array_number);
      DP("raid.total_secs: %u", pdc, pdc->raid.total_secs);
      DP("raid.cylinders: %u", pdc, pdc->raid.cylinders);
      DP("raid.heads: %u", pdc, pdc->raid.heads);
      DP("raid.sectors: %u", pdc, pdc->raid.sectors);
      DP("raid.magic_1: 0x%x", pdc, pdc->raid.magic_1);
      DP("raid.unknown_5: 0x%x", pdc, pdc->raid.unknown_5);

      for (disk = pdc->raid.disk, i = 0;
           i < pdc->raid.total_disks;
           disk++, i++) {
            P2("raid.disk[%d].unknown_0: 0x%x", pdc, i, disk->unknown_0);
            P2("raid.disk[%d].channel: %u", pdc, i, disk->channel);
            P2("raid.disk[%d].device: %u", pdc, i, disk->device);
            P2("raid.disk[%d].magic_0: 0x%x", pdc, i, disk->magic_0);
            P2("raid.disk[%d].disk_number: %u", pdc, i, disk->disk_number);
      }

      P("checksum: 0x%x %s", pdc, pdc->checksum, pdc->checksum,
         checksum(pdc) ? "Ok" : "BAD");
}
#endif

static struct dmraid_format pdc_format = {
      .name = HANDLER,
      .descr      = "Promise FastTrack",
      .caps = "S,0,1,10",
      .format = FMT_RAID,
      .read = pdc_read,
      .write      = pdc_write,
      .group      = pdc_group,
      .check      = pdc_check,
      .events     = &pdc_event_handlers,
#ifdef DMRAID_NATIVE_LOG
      .log  = pdc_log,
#endif
};

/* Register this format handler with the format core. */
int register_pdc(struct lib_context *lc)
{
      return register_format_handler(lc, &pdc_format);
}

/* Calculate RAID device size in sectors depending on RAID type. */
static uint64_t sectors(struct raid_dev *rd, unsigned int meta_sector)
{
      struct pdc *pdc = META(rd, pdc);

      switch (pdc->raid.type) {
      case PDC_T_RAID10:
               return pdc->raid.total_secs / (pdc->raid.total_disks / 2);

      case PDC_T_RAID1:
               return pdc->raid.total_secs;

      case PDC_T_RAID0:
               return pdc->raid.total_secs / pdc->raid.total_disks;

      case PDC_T_SPAN:
             return rd->di->sectors - meta_sector;
      }

      return 0;
}

/* Set the RAID device contents up derived from the PDC ones */
static int setup_rd(struct lib_context *lc, struct raid_dev *rd,
                struct dev_info *di, void *meta, union read_info *info)
{
      unsigned int meta_sector;
      struct pdc *pdc = meta;

      if (!checksum(pdc))
            LOG_ERR(lc, 0, "%s: invalid checksum on %s", handler, di->path);

      if (!(rd->meta_areas = alloc_meta_areas(lc, rd, handler, 1)))
            return 0;

      meta_sector = info->u32;
      rd->meta_areas->offset = di->sectors - meta_sector;
      rd->meta_areas->size = sizeof(*pdc);
      rd->meta_areas->area = (void*) pdc;

        rd->di = di;
      rd->fmt = &pdc_format;

      rd->status = status(pdc);
      /*
       * Type needs to be set before sectors(), because we need
       * to set the RAID10 type used there!
       */
      rd->type   = type(pdc);

      rd->offset = PDC_DATAOFFSET;
      if (!(rd->sectors = sectors(rd, meta_sector)))
            return log_zero_sectors(lc, di->path, handler);

        return (rd->name = name(lc, rd, is_raid10(pdc))) ? 1 : 0;
}