/*
* Given a file descriptor, clear (zero) the label information.
*/
int
zpool_clear_label(int fd)
{
struct stat statbuf;
int l;
vdev_label_t *label;
uint64_t size;
if (fstat_blk(fd, &statbuf) == -1)
return (0);
size = P2ALIGN_TYPED(statbuf.st_size, sizeof (vdev_label_t), uint64_t);
if ((label = calloc(sizeof (vdev_label_t), 1)) == NULL)
return (-1);
for (l = 0; l < VDEV_LABELS; l++) {
if (pwrite64(fd, label, sizeof (vdev_label_t),
label_offset(size, l)) != sizeof (vdev_label_t)) {
free(label);
return (-1);
}
}
free(label);
return (0);
}
/*
* Given the root disk device name, read the label from
* the device, and construct a configuration nvlist.
*/
int
vdev_disk_read_rootlabel(char *devname, nvlist_t **config)
{
vdev_label_t *label;
struct device *dev;
uint64_t size;
int l;
int error = -1;
char *minor_name;
error = device_open(devname + 5, DO_RDWR, &dev);
if (error)
return error;
size = P2ALIGN_TYPED(dev->size, sizeof (vdev_label_t), uint64_t);
label = kmem_alloc(sizeof (vdev_label_t), KM_SLEEP);
*config = NULL;
for (l = 0; l < VDEV_LABELS; l++) {
uint64_t offset, state, txg = 0;
/* read vdev label */
offset = vdev_label_offset(size, l, 0);
if (vdev_disk_physio(dev, (caddr_t)label,
VDEV_SKIP_SIZE + VDEV_PHYS_SIZE, offset, 0) != 0)
continue;
if (nvlist_unpack(label->vl_vdev_phys.vp_nvlist,
sizeof (label->vl_vdev_phys.vp_nvlist), config, 0) != 0) {
*config = NULL;
continue;
}
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
&state) != 0 || state >= POOL_STATE_DESTROYED) {
nvlist_free(*config);
*config = NULL;
continue;
}
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
&txg) != 0 || txg == 0) {
nvlist_free(*config);
*config = NULL;
continue;
}
break;
}
kmem_free(label, sizeof (vdev_label_t));
device_close(dev);
if (*config == NULL)
error = EIDRM;
return (error);
}
/*
* Get data to generate a TX_WRITE intent log record.
*/
static int
zvol_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio)
{
zvol_state_t *zv = arg;
objset_t *os = zv->zv_objset;
uint64_t offset = lr->lr_offset;
uint64_t size = lr->lr_length;
dmu_buf_t *db;
zgd_t *zgd;
int error;
ASSERT(zio != NULL);
ASSERT(size != 0);
zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_PUSHPAGE);
zgd->zgd_zilog = zv->zv_zilog;
zgd->zgd_rl = zfs_range_lock(&zv->zv_znode, offset, size, RL_READER);
/*
* Write records come in two flavors: immediate and indirect.
* For small writes it's cheaper to store the data with the
* log record (immediate); for large writes it's cheaper to
* sync the data and get a pointer to it (indirect) so that
* we don't have to write the data twice.
*/
if (buf != NULL) { /* immediate write */
error = dmu_read(os, ZVOL_OBJ, offset, size, buf,
DMU_READ_NO_PREFETCH);
} else {
size = zv->zv_volblocksize;
offset = P2ALIGN_TYPED(offset, size, uint64_t);
error = dmu_buf_hold(os, ZVOL_OBJ, offset, zgd, &db,
DMU_READ_NO_PREFETCH);
if (error == 0) {
zgd->zgd_db = db;
zgd->zgd_bp = &lr->lr_blkptr;
ASSERT(db != NULL);
ASSERT(db->db_offset == offset);
ASSERT(db->db_size == size);
error = dmu_sync(zio, lr->lr_common.lrc_txg,
zvol_get_done, zgd);
if (error == 0)
return (0);
}
}
zvol_get_done(zgd, error);
return (error);
}
/*
* Build page tables for the lower 1G of physical memory using 2M
* pages, and prepare page tables for mapping new kernel and boot
* archive pages using 4K pages.
*/
static void
fastboot_build_pagetables(fastboot_info_t *nk)
{
/*
* Map lower 1G physical memory. Use large pages.
*/
fastboot_map_with_size(nk, 0, 0, ONE_GIG, 1);
/*
* Map one 4K page to get the middle page tables set up.
*/
fake_va = P2ALIGN_TYPED(fake_va, nk->fi_lpagesize, uintptr_t);
fastboot_map_with_size(nk, fake_va,
nk->fi_files[0].fb_pte_list_va[0] & MMU_PAGEMASK, PAGESIZE, 0);
}
/*
* Given a file descriptor, read the label information and return an nvlist
* describing the configuration, if there is one.
*/
int
zpool_read_label(int fd, nvlist_t **config)
{
struct stat64 statbuf;
int l;
vdev_label_t *label;
uint64_t state, txg, size;
*config = NULL;
if (fstat64(fd, &statbuf) == -1)
return (0);
size = P2ALIGN_TYPED(statbuf.st_size, sizeof (vdev_label_t), uint64_t);
if ((label = malloc(sizeof (vdev_label_t))) == NULL)
return (-1);
for (l = 0; l < VDEV_LABELS; l++) {
if (pread64(fd, label, sizeof (vdev_label_t),
label_offset(size, l)) != sizeof (vdev_label_t))
continue;
if (nvlist_unpack(label->vl_vdev_phys.vp_nvlist,
sizeof (label->vl_vdev_phys.vp_nvlist), config, 0) != 0)
continue;
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
&state) != 0 || state > POOL_STATE_L2CACHE) {
nvlist_free(*config);
continue;
}
if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
(nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
&txg) != 0 || txg == 0)) {
nvlist_free(*config);
continue;
}
free(label);
return (0);
}
free(label);
*config = NULL;
return (0);
}
/*
* Given a file descriptor, read the label information and return an nvlist
* describing the configuration, if there is one.
* Return 0 on success, or -1 on failure
*/
int
zpool_read_label(int fd, nvlist_t **config, int *num_labels)
{
struct stat statbuf;
int l, count = 0;
vdev_label_t *label;
nvlist_t *expected_config = NULL;
uint64_t expected_guid = 0, size;
*config = NULL;
if (fstat(fd, &statbuf) == -1)
return (-1);
size = P2ALIGN_TYPED(statbuf.st_size, sizeof (vdev_label_t), uint64_t);
if ((label = malloc(sizeof (vdev_label_t))) == NULL)
return (-1);
for (l = 0; l < VDEV_LABELS; l++) {
uint64_t state, guid, txg;
if (pread(fd, label, sizeof (vdev_label_t),
label_offset(size, l)) != sizeof (vdev_label_t))
continue;
if (nvlist_unpack(label->vl_vdev_phys.vp_nvlist,
sizeof (label->vl_vdev_phys.vp_nvlist), config, 0) != 0)
continue;
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_GUID,
&guid) != 0 || guid == 0) {
nvlist_free(*config);
continue;
}
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
&state) != 0 || state > POOL_STATE_L2CACHE) {
nvlist_free(*config);
continue;
}
if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
(nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
&txg) != 0 || txg == 0)) {
nvlist_free(*config);
continue;
}
if (expected_guid) {
if (expected_guid == guid)
count++;
nvlist_free(*config);
} else {
expected_config = *config;
expected_guid = guid;
count++;
}
}
if (num_labels != NULL)
*num_labels = count;
free(label);
*config = expected_config;
return (0);
}
int
zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx, offset_t off, ssize_t len,
char *addr)
{
dmu_object_info_t doi;
ssize_t nbytes;
itx_t *itx;
lr_write_t *lr;
objset_t *os;
dmu_buf_t *db;
uint64_t txg;
uint64_t boff;
int error;
uint32_t blocksize;
/* handle common case */
if (len <= zvol_immediate_write_sz) {
itx = zvol_immediate_itx(off, len, addr);
(void) zil_itx_assign(zv->zv_zilog, itx, tx);
return (0);
}
txg = dmu_tx_get_txg(tx);
os = zv->zv_objset;
/*
* We need to dmu_sync() each block in the range.
* For this we need the blocksize.
*/
error = dmu_object_info(os, ZVOL_OBJ, &doi);
if (error)
return (error);
blocksize = doi.doi_data_block_size;
/*
* We need to immediate write or dmu_sync() each block in the range.
*/
while (len) {
nbytes = MIN(len, blocksize - P2PHASE(off, blocksize));
if (nbytes <= zvol_immediate_write_sz) {
itx = zvol_immediate_itx(off, nbytes, addr);
} else {
boff = P2ALIGN_TYPED(off, blocksize, uint64_t);
itx = zil_itx_create(TX_WRITE, sizeof (*lr));
lr = (lr_write_t *)&itx->itx_lr;
lr->lr_foid = ZVOL_OBJ;
lr->lr_offset = off;
lr->lr_length = nbytes;
lr->lr_blkoff = off - boff;
BP_ZERO(&lr->lr_blkptr);
/* XXX - we should do these IOs in parallel */
VERIFY(0 == dmu_buf_hold(os, ZVOL_OBJ, boff,
FTAG, &db));
ASSERT(boff == db->db_offset);
error = dmu_sync(NULL, db, &lr->lr_blkptr,
txg, NULL, NULL);
dmu_buf_rele(db, FTAG);
if (error) {
kmem_free(itx, offsetof(itx_t, itx_lr));
return (error);
}
itx->itx_wr_state = WR_COPIED;
}
(void) zil_itx_assign(zv->zv_zilog, itx, tx);
len -= nbytes;
off += nbytes;
}
return (0);
}
/*
* Given the root disk device devid or pathname, read the label from
* the device, and construct a configuration nvlist.
*/
int
vdev_disk_read_rootlabel(char *devpath, char *devid, nvlist_t **config)
{
ldi_handle_t vd_lh;
vdev_label_t *label;
uint64_t s, size;
int l;
ddi_devid_t tmpdevid;
int error = -1;
char *minor_name;
/*
* Read the device label and build the nvlist.
*/
if (devid != NULL && ddi_devid_str_decode(devid, &tmpdevid,
&minor_name) == 0) {
error = ldi_open_by_devid(tmpdevid, minor_name,
FREAD, kcred, &vd_lh, zfs_li);
ddi_devid_free(tmpdevid);
ddi_devid_str_free(minor_name);
}
if (error && (error = ldi_open_by_name(devpath, FREAD, kcred, &vd_lh,
zfs_li)))
return (error);
if (ldi_get_size(vd_lh, &s)) {
(void) ldi_close(vd_lh, FREAD, kcred);
return (SET_ERROR(EIO));
}
size = P2ALIGN_TYPED(s, sizeof (vdev_label_t), uint64_t);
label = kmem_alloc(sizeof (vdev_label_t), KM_SLEEP);
*config = NULL;
for (l = 0; l < VDEV_LABELS; l++) {
uint64_t offset, state, txg = 0;
/* read vdev label */
offset = vdev_label_offset(size, l, 0);
if (vdev_disk_ldi_physio(vd_lh, (caddr_t)label,
VDEV_SKIP_SIZE + VDEV_PHYS_SIZE, offset, B_READ) != 0)
continue;
if (nvlist_unpack(label->vl_vdev_phys.vp_nvlist,
sizeof (label->vl_vdev_phys.vp_nvlist), config, 0) != 0) {
*config = NULL;
continue;
}
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
&state) != 0 || state >= POOL_STATE_DESTROYED) {
nvlist_free(*config);
*config = NULL;
continue;
}
if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
&txg) != 0 || txg == 0) {
nvlist_free(*config);
*config = NULL;
continue;
}
break;
}
kmem_free(label, sizeof (vdev_label_t));
(void) ldi_close(vd_lh, FREAD, kcred);
if (*config == NULL)
error = SET_ERROR(EIDRM);
return (error);
}