freebsd-dev/lib/libzfs/libzfs_fru.c
Brian Behlendorf 6b003d7cda Add linux topology support
Solaris recently introduced the idea of drive topology because
where a drive is located does matter.  I have already handled
this with udev/blkid integration under Linux so I'm hopeful
this case can simply be removed but for now I've just stubbed
out what is needed in libspl and commented out the rest here.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2010-08-31 13:42:00 -07:00

470 lines
12 KiB
C

/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2009 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
#include <dlfcn.h>
#include <errno.h>
#include <libintl.h>
#include <link.h>
#include <pthread.h>
#include <strings.h>
#include <unistd.h>
#include <libzfs.h>
#if defined(HAVE_LIBTOPO)
#include <fm/libtopo.h>
#include <sys/fm/protocol.h>
#include <sys/systeminfo.h>
#include "libzfs_impl.h"
/*
* This file is responsible for determining the relationship between I/O
* devices paths and physical locations. In the world of MPxIO and external
* enclosures, the device path is not synonymous with the physical location.
* If you remove a drive and insert it into a different slot, it will end up
* with the same path under MPxIO. If you recable storage enclosures, the
* device paths may change. All of this makes it difficult to implement the
* 'autoreplace' property, which is supposed to automatically manage disk
* replacement based on physical slot.
*
* In order to work around these limitations, we have a per-vdev FRU property
* that is the libtopo path (minus disk-specific authority information) to the
* physical location of the device on the system. This is an optional
* property, and is only needed when using the 'autoreplace' property or when
* generating FMA faults against vdevs.
*/
/*
* Because the FMA packages depend on ZFS, we have to dlopen() libtopo in case
* it is not present. We only need this once per library instance, so it is
* not part of the libzfs handle.
*/
static void *_topo_dlhandle;
static topo_hdl_t *(*_topo_open)(int, const char *, int *);
static void (*_topo_close)(topo_hdl_t *);
static char *(*_topo_snap_hold)(topo_hdl_t *, const char *, int *);
static void (*_topo_snap_release)(topo_hdl_t *);
static topo_walk_t *(*_topo_walk_init)(topo_hdl_t *, const char *,
topo_walk_cb_t, void *, int *);
static int (*_topo_walk_step)(topo_walk_t *, int);
static void (*_topo_walk_fini)(topo_walk_t *);
static void (*_topo_hdl_strfree)(topo_hdl_t *, char *);
static char *(*_topo_node_name)(tnode_t *);
static int (*_topo_prop_get_string)(tnode_t *, const char *, const char *,
char **, int *);
static int (*_topo_node_fru)(tnode_t *, nvlist_t **, nvlist_t *, int *);
static int (*_topo_fmri_nvl2str)(topo_hdl_t *, nvlist_t *, char **, int *);
static int (*_topo_fmri_strcmp_noauth)(topo_hdl_t *, const char *,
const char *);
#define ZFS_FRU_HASH_SIZE 257
static size_t
fru_strhash(const char *key)
{
ulong_t g, h = 0;
const char *p;
for (p = key; *p != '\0'; p++) {
h = (h << 4) + *p;
if ((g = (h & 0xf0000000)) != 0) {
h ^= (g >> 24);
h ^= g;
}
}
return (h % ZFS_FRU_HASH_SIZE);
}
static int
libzfs_fru_gather(topo_hdl_t *thp, tnode_t *tn, void *arg)
{
libzfs_handle_t *hdl = arg;
nvlist_t *fru;
char *devpath, *frustr;
int err;
libzfs_fru_t *frup;
size_t idx;
/*
* If this is the chassis node, and we don't yet have the system
* chassis ID, then fill in this value now.
*/
if (hdl->libzfs_chassis_id[0] == '\0' &&
strcmp(_topo_node_name(tn), "chassis") == 0) {
if (_topo_prop_get_string(tn, FM_FMRI_AUTHORITY,
FM_FMRI_AUTH_CHASSIS, &devpath, &err) == 0)
(void) strlcpy(hdl->libzfs_chassis_id, devpath,
sizeof (hdl->libzfs_chassis_id));
}
/*
* Skip non-disk nodes.
*/
if (strcmp(_topo_node_name(tn), "disk") != 0)
return (TOPO_WALK_NEXT);
/*
* Get the devfs path and FRU.
*/
if (_topo_prop_get_string(tn, "io", "devfs-path", &devpath, &err) != 0)
return (TOPO_WALK_NEXT);
if (libzfs_fru_lookup(hdl, devpath) != NULL) {
_topo_hdl_strfree(thp, devpath);
return (TOPO_WALK_NEXT);
}
if (_topo_node_fru(tn, &fru, NULL, &err) != 0) {
_topo_hdl_strfree(thp, devpath);
return (TOPO_WALK_NEXT);
}
/*
* Convert the FRU into a string.
*/
if (_topo_fmri_nvl2str(thp, fru, &frustr, &err) != 0) {
nvlist_free(fru);
_topo_hdl_strfree(thp, devpath);
return (TOPO_WALK_NEXT);
}
nvlist_free(fru);
/*
* Finally, we have a FRU string and device path. Add it to the hash.
*/
if ((frup = calloc(sizeof (libzfs_fru_t), 1)) == NULL) {
_topo_hdl_strfree(thp, devpath);
_topo_hdl_strfree(thp, frustr);
return (TOPO_WALK_NEXT);
}
if ((frup->zf_device = strdup(devpath)) == NULL ||
(frup->zf_fru = strdup(frustr)) == NULL) {
free(frup->zf_device);
free(frup);
_topo_hdl_strfree(thp, devpath);
_topo_hdl_strfree(thp, frustr);
return (TOPO_WALK_NEXT);
}
_topo_hdl_strfree(thp, devpath);
_topo_hdl_strfree(thp, frustr);
idx = fru_strhash(frup->zf_device);
frup->zf_chain = hdl->libzfs_fru_hash[idx];
hdl->libzfs_fru_hash[idx] = frup;
frup->zf_next = hdl->libzfs_fru_list;
hdl->libzfs_fru_list = frup;
return (TOPO_WALK_NEXT);
}
/*
* Called during initialization to setup the dynamic libtopo connection.
*/
#pragma init(libzfs_init_fru)
static void
libzfs_init_fru(void)
{
char path[MAXPATHLEN];
char isa[257];
#if defined(_LP64)
if (sysinfo(SI_ARCHITECTURE_64, isa, sizeof (isa)) < 0)
isa[0] = '\0';
#else
isa[0] = '\0';
#endif
(void) snprintf(path, sizeof (path),
"/usr/lib/fm/%s/libtopo.so", isa);
if ((_topo_dlhandle = dlopen(path, RTLD_LAZY)) == NULL)
return;
_topo_open = (topo_hdl_t *(*)())
dlsym(_topo_dlhandle, "topo_open");
_topo_close = (void (*)())
dlsym(_topo_dlhandle, "topo_close");
_topo_snap_hold = (char *(*)())
dlsym(_topo_dlhandle, "topo_snap_hold");
_topo_snap_release = (void (*)())
dlsym(_topo_dlhandle, "topo_snap_release");
_topo_walk_init = (topo_walk_t *(*)())
dlsym(_topo_dlhandle, "topo_walk_init");
_topo_walk_step = (int (*)())
dlsym(_topo_dlhandle, "topo_walk_step");
_topo_walk_fini = (void (*)())
dlsym(_topo_dlhandle, "topo_walk_fini");
_topo_hdl_strfree = (void (*)())
dlsym(_topo_dlhandle, "topo_hdl_strfree");
_topo_node_name = (char *(*)())
dlsym(_topo_dlhandle, "topo_node_name");
_topo_prop_get_string = (int (*)())
dlsym(_topo_dlhandle, "topo_prop_get_string");
_topo_node_fru = (int (*)())
dlsym(_topo_dlhandle, "topo_node_fru");
_topo_fmri_nvl2str = (int (*)())
dlsym(_topo_dlhandle, "topo_fmri_nvl2str");
_topo_fmri_strcmp_noauth = (int (*)())
dlsym(_topo_dlhandle, "topo_fmri_strcmp_noauth");
if (_topo_open == NULL || _topo_close == NULL ||
_topo_snap_hold == NULL || _topo_snap_release == NULL ||
_topo_walk_init == NULL || _topo_walk_step == NULL ||
_topo_walk_fini == NULL || _topo_hdl_strfree == NULL ||
_topo_node_name == NULL || _topo_prop_get_string == NULL ||
_topo_node_fru == NULL || _topo_fmri_nvl2str == NULL ||
_topo_fmri_strcmp_noauth == NULL) {
(void) dlclose(_topo_dlhandle);
_topo_dlhandle = NULL;
}
}
/*
* Refresh the mappings from device path -> FMRI. We do this by walking the
* hc topology looking for disk nodes, and recording the io/devfs-path and FRU.
* Note that we strip out the disk-specific authority information (serial,
* part, revision, etc) so that we are left with only the identifying
* characteristics of the slot (hc path and chassis-id).
*/
void
libzfs_fru_refresh(libzfs_handle_t *hdl)
{
int err;
char *uuid;
topo_hdl_t *thp;
topo_walk_t *twp;
if (_topo_dlhandle == NULL)
return;
/*
* Clear the FRU hash and initialize our basic structures.
*/
libzfs_fru_clear(hdl, B_FALSE);
if ((hdl->libzfs_topo_hdl = _topo_open(TOPO_VERSION,
NULL, &err)) == NULL)
return;
thp = hdl->libzfs_topo_hdl;
if ((uuid = _topo_snap_hold(thp, NULL, &err)) == NULL)
return;
_topo_hdl_strfree(thp, uuid);
if (hdl->libzfs_fru_hash == NULL &&
(hdl->libzfs_fru_hash =
calloc(ZFS_FRU_HASH_SIZE * sizeof (void *), 1)) == NULL)
return;
/*
* We now have a topo snapshot, so iterate over the hc topology looking
* for disks to add to the hash.
*/
twp = _topo_walk_init(thp, FM_FMRI_SCHEME_HC,
libzfs_fru_gather, hdl, &err);
if (twp != NULL) {
(void) _topo_walk_step(twp, TOPO_WALK_CHILD);
_topo_walk_fini(twp);
}
}
/*
* Given a devfs path, return the FRU for the device, if known. This will
* automatically call libzfs_fru_refresh() if it hasn't already been called by
* the consumer. The string returned is valid until the next call to
* libzfs_fru_refresh().
*/
const char *
libzfs_fru_lookup(libzfs_handle_t *hdl, const char *devpath)
{
size_t idx = fru_strhash(devpath);
libzfs_fru_t *frup;
if (hdl->libzfs_fru_hash == NULL)
libzfs_fru_refresh(hdl);
if (hdl->libzfs_fru_hash == NULL)
return (NULL);
for (frup = hdl->libzfs_fru_hash[idx]; frup != NULL;
frup = frup->zf_chain) {
if (strcmp(devpath, frup->zf_device) == 0)
return (frup->zf_fru);
}
return (NULL);
}
/*
* Given a fru path, return the device path. This will automatically call
* libzfs_fru_refresh() if it hasn't already been called by the consumer. The
* string returned is valid until the next call to libzfs_fru_refresh().
*/
const char *
libzfs_fru_devpath(libzfs_handle_t *hdl, const char *fru)
{
libzfs_fru_t *frup;
size_t idx;
if (hdl->libzfs_fru_hash == NULL)
libzfs_fru_refresh(hdl);
if (hdl->libzfs_fru_hash == NULL)
return (NULL);
for (idx = 0; idx < ZFS_FRU_HASH_SIZE; idx++) {
for (frup = hdl->libzfs_fru_hash[idx]; frup != NULL;
frup = frup->zf_next) {
if (_topo_fmri_strcmp_noauth(hdl->libzfs_topo_hdl,
fru, frup->zf_fru))
return (frup->zf_device);
}
}
return (NULL);
}
/*
* Change the stored FRU for the given vdev.
*/
int
zpool_fru_set(zpool_handle_t *zhp, uint64_t vdev_guid, const char *fru)
{
zfs_cmd_t zc = { 0 };
(void) strncpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
(void) strncpy(zc.zc_value, fru, sizeof (zc.zc_value));
zc.zc_guid = vdev_guid;
if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_VDEV_SETFRU, &zc) != 0)
return (zpool_standard_error_fmt(zhp->zpool_hdl, errno,
dgettext(TEXT_DOMAIN, "cannot set FRU")));
return (0);
}
/*
* Compare to two FRUs, ignoring any authority information.
*/
boolean_t
libzfs_fru_compare(libzfs_handle_t *hdl, const char *a, const char *b)
{
if (hdl->libzfs_fru_hash == NULL)
libzfs_fru_refresh(hdl);
if (hdl->libzfs_fru_hash == NULL)
return (strcmp(a, b) == 0);
return (_topo_fmri_strcmp_noauth(hdl->libzfs_topo_hdl, a, b));
}
/*
* This special function checks to see whether the FRU indicates it's supposed
* to be in the system chassis, but the chassis-id doesn't match. This can
* happen in a clustered case, where both head nodes have the same logical
* disk, but opening the device on the other head node is meaningless.
*/
boolean_t
libzfs_fru_notself(libzfs_handle_t *hdl, const char *fru)
{
const char *chassisid;
size_t len;
if (hdl->libzfs_fru_hash == NULL)
libzfs_fru_refresh(hdl);
if (hdl->libzfs_chassis_id[0] == '\0')
return (B_FALSE);
if (strstr(fru, "/chassis=0/") == NULL)
return (B_FALSE);
if ((chassisid = strstr(fru, ":chassis-id=")) == NULL)
return (B_FALSE);
chassisid += 12;
len = strlen(hdl->libzfs_chassis_id);
if (strncmp(chassisid, hdl->libzfs_chassis_id, len) == 0 &&
(chassisid[len] == '/' || chassisid[len] == ':'))
return (B_FALSE);
return (B_TRUE);
}
/*
* Clear memory associated with the FRU hash.
*/
void
libzfs_fru_clear(libzfs_handle_t *hdl, boolean_t final)
{
libzfs_fru_t *frup;
while ((frup = hdl->libzfs_fru_list) != NULL) {
hdl->libzfs_fru_list = frup->zf_next;
free(frup->zf_device);
free(frup->zf_fru);
free(frup);
}
hdl->libzfs_fru_list = NULL;
if (hdl->libzfs_topo_hdl != NULL) {
_topo_snap_release(hdl->libzfs_topo_hdl);
_topo_close(hdl->libzfs_topo_hdl);
hdl->libzfs_topo_hdl = NULL;
}
if (final) {
free(hdl->libzfs_fru_hash);
} else if (hdl->libzfs_fru_hash != NULL) {
bzero(hdl->libzfs_fru_hash,
ZFS_FRU_HASH_SIZE * sizeof (void *));
}
}
#else /* HAVE_LIBTOPO */
/*
* Clear memory associated with the FRU hash.
*/
void
libzfs_fru_clear(libzfs_handle_t *hdl, boolean_t final)
{
return;
}
#endif /* HAVE_LIBTOPO */