Skip to content

Commit

Permalink
zfs: merge openzfs/zfs@797f55ef1
Browse files Browse the repository at this point in the history 
Notable upstream pull request merges:
 #14378 c0e5899 Large sync writes perform worse with slog
 #14721 797f55e Do not persist user/group/project quota zap objects
                  when unneeded
 #15356 380c25f FreeBSD: Improve taskq wrapper
 #15390 4fbc524 Remove lock from dsl_pool_need_dirty_delay()
 #15397 57b4098 Trust ARC_BUF_SHARED() more
 #15402 b29e98f Properly pad struct tx_cpu to cache line
 #15405 ea30b5a Set spa_ccw_fail_time=0 when expanding a vdev
 #15416 b9384b9 FreeBSD: taskq: Remove unused declaration

Obtained from:	OpenZFS
OpenZFS commit:	797f55e
  • Loading branch information
@mmatuska
mmatuska committed Oct 22, 2023
2 parents 63f5375 + 797f55e commit 22b267e
Show file tree
Hide file tree
Showing 22 changed files with 386 additions and 119 deletions.
55 changes: 37 additions & 18 deletions sys/contrib/openzfs/cmd/zed/agents/zfs_mod.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@
* Copyright 2014 Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2016, 2017, Intel Corporation.
* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
* Copyright (c) 2023, Klara Inc.
*/

/*
Expand Down Expand Up @@ -204,7 +205,7 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
uint64_t is_spare = 0;
const char *physpath = NULL, *new_devid = NULL, *enc_sysfs_path = NULL;
char rawpath[PATH_MAX], fullpath[PATH_MAX];
char devpath[PATH_MAX];
char pathbuf[PATH_MAX];
int ret;
int online_flag = ZFS_ONLINE_CHECKREMOVE | ZFS_ONLINE_UNSPARE;
boolean_t is_sd = B_FALSE;
Expand All @@ -214,6 +215,11 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
char **lines = NULL;
int lines_cnt = 0;

/*
* Get the persistent path, typically under the '/dev/disk/by-id' or
* '/dev/disk/by-vdev' directories. Note that this path can change
* when a vdev is replaced with a new disk.
*/
if (nvlist_lookup_string(vdev, ZPOOL_CONFIG_PATH, &path) != 0)
return;

Expand Down Expand Up @@ -370,15 +376,17 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
(void) snprintf(rawpath, sizeof (rawpath), "%s%s",
is_sd ? DEV_BYVDEV_PATH : DEV_BYPATH_PATH, physpath);

if (realpath(rawpath, devpath) == NULL && !is_mpath_wholedisk) {
if (realpath(rawpath, pathbuf) == NULL && !is_mpath_wholedisk) {
zed_log_msg(LOG_INFO, " realpath: %s failed (%s)",
rawpath, strerror(errno));

(void) zpool_vdev_online(zhp, fullpath, ZFS_ONLINE_FORCEFAULT,
&newstate);
int err = zpool_vdev_online(zhp, fullpath,
ZFS_ONLINE_FORCEFAULT, &newstate);

zed_log_msg(LOG_INFO, " zpool_vdev_online: %s FORCEFAULT (%s)",
fullpath, libzfs_error_description(g_zfshdl));
zed_log_msg(LOG_INFO, " zpool_vdev_online: %s FORCEFAULT (%s) "
"err %d, new state %d",
fullpath, libzfs_error_description(g_zfshdl), err,
err ? (int)newstate : 0);
return;
}

Expand Down Expand Up @@ -428,15 +436,15 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
* to trigger a ZFS fault for the device (and any hot spare
* replacement).
*/
leafname = strrchr(devpath, '/') + 1;
leafname = strrchr(pathbuf, '/') + 1;

/*
* If this is a request to label a whole disk, then attempt to
* write out the label.
*/
if (zpool_prepare_and_label_disk(g_zfshdl, zhp, leafname,
vdev, "autoreplace", &lines, &lines_cnt) != 0) {
zed_log_msg(LOG_INFO,
zed_log_msg(LOG_WARNING,
" zpool_prepare_and_label_disk: could not "
"label '%s' (%s)", leafname,
libzfs_error_description(g_zfshdl));
Expand Down Expand Up @@ -468,7 +476,7 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
sizeof (device->pd_physpath));
list_insert_tail(&g_device_list, device);

zed_log_msg(LOG_INFO, " zpool_label_disk: async '%s' (%llu)",
zed_log_msg(LOG_NOTICE, " zpool_label_disk: async '%s' (%llu)",
leafname, (u_longlong_t)guid);

return; /* resumes at EC_DEV_ADD.ESC_DISK for partition */
Expand All @@ -491,8 +499,8 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
}
if (!found) {
/* unexpected partition slice encountered */
zed_log_msg(LOG_INFO, "labeled disk %s unexpected here",
fullpath);
zed_log_msg(LOG_WARNING, "labeled disk %s was "
"unexpected here", fullpath);
(void) zpool_vdev_online(zhp, fullpath,
ZFS_ONLINE_FORCEFAULT, &newstate);
return;
Expand All @@ -501,8 +509,17 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
zed_log_msg(LOG_INFO, " zpool_label_disk: resume '%s' (%llu)",
physpath, (u_longlong_t)guid);

(void) snprintf(devpath, sizeof (devpath), "%s%s",
DEV_BYID_PATH, new_devid);
/*
* Paths that begin with '/dev/disk/by-id/' will change and so
* they must be updated before calling zpool_vdev_attach().
*/
if (strncmp(path, DEV_BYID_PATH, strlen(DEV_BYID_PATH)) == 0) {
(void) snprintf(pathbuf, sizeof (pathbuf), "%s%s",
DEV_BYID_PATH, new_devid);
zed_log_msg(LOG_INFO, " zpool_label_disk: path '%s' "
"replaced by '%s'", path, pathbuf);
path = pathbuf;
}
}

libzfs_free_str_array(lines, lines_cnt);
Expand Down Expand Up @@ -545,9 +562,11 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
* Wait for udev to verify the links exist, then auto-replace
* the leaf disk at same physical location.
*/
if (zpool_label_disk_wait(path, 3000) != 0) {
zed_log_msg(LOG_WARNING, "zfs_mod: expected replacement "
"disk %s is missing", path);
if (zpool_label_disk_wait(path, DISK_LABEL_WAIT) != 0) {
zed_log_msg(LOG_WARNING, "zfs_mod: pool '%s', after labeling "
"replacement disk, the expected disk partition link '%s' "
"is missing after waiting %u ms",
zpool_get_name(zhp), path, DISK_LABEL_WAIT);
nvlist_free(nvroot);
return;
}
Expand All @@ -562,7 +581,7 @@ zfs_process_add(zpool_handle_t *zhp, nvlist_t *vdev, boolean_t labeled)
B_TRUE, B_FALSE);
}

zed_log_msg(LOG_INFO, " zpool_vdev_replace: %s with %s (%s)",
zed_log_msg(LOG_WARNING, " zpool_vdev_replace: %s with %s (%s)",
fullpath, path, (ret == 0) ? "no errors" :
libzfs_error_description(g_zfshdl));

Expand Down Expand Up @@ -660,7 +679,7 @@ zfs_iter_vdev(zpool_handle_t *zhp, nvlist_t *nvl, void *data)
dp->dd_prop, path);
dp->dd_found = B_TRUE;

/* pass the new devid for use by replacing code */
/* pass the new devid for use by auto-replacing code */
if (dp->dd_new_devid != NULL) {
(void) nvlist_add_string(nvl, "new_devid",
dp->dd_new_devid);
Expand Down
2 changes: 1 addition & 1 deletion sys/contrib/openzfs/include/libzutil.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ extern "C" {
#endif

/*
* Default wait time for a device name to be created.
* Default wait time in milliseconds for a device name to be created.
*/
#define DISK_LABEL_WAIT (30 * 1000) /* 30 seconds */

Expand Down
18 changes: 9 additions & 9 deletions sys/contrib/openzfs/include/os/freebsd/spl/sys/taskq.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,9 +30,9 @@

#include <sys/types.h>
#include <sys/proc.h>
#include <sys/queue.h>
#include <sys/taskqueue.h>
#include <sys/thread.h>
#include <sys/ck.h>

#ifdef __cplusplus
extern "C" {
Expand All @@ -48,16 +48,16 @@ typedef uintptr_t taskqid_t;
typedef void (task_func_t)(void *);

typedef struct taskq_ent {
struct task tqent_task;
struct timeout_task tqent_timeout_task;
union {
struct task tqent_task;
struct timeout_task tqent_timeout_task;
};
task_func_t *tqent_func;
void *tqent_arg;
taskqid_t tqent_id;
CK_LIST_ENTRY(taskq_ent) tqent_hash;
uint8_t tqent_type;
uint8_t tqent_registered;
uint8_t tqent_cancelled;
volatile uint32_t tqent_rc;
taskqid_t tqent_id;
LIST_ENTRY(taskq_ent) tqent_hash;
uint_t tqent_type;
volatile uint_t tqent_rc;
} taskq_ent_t;

/*
Expand Down
3 changes: 1 addition & 2 deletions sys/contrib/openzfs/include/sys/txg_impl.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -73,8 +73,7 @@ struct tx_cpu {
kcondvar_t tc_cv[TXG_SIZE];
uint64_t tc_count[TXG_SIZE]; /* tx hold count on each txg */
list_t tc_callbacks[TXG_SIZE]; /* commit cb list */
char tc_pad[8]; /* pad to fill 3 cache lines */
};
} ____cacheline_aligned;

/*
* The tx_state structure maintains the state information about the different
Expand Down
5 changes: 2 additions & 3 deletions sys/contrib/openzfs/lib/libzutil/os/linux/zutil_import_os.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -582,9 +582,8 @@ zfs_device_get_physical(struct udev_device *dev, char *bufptr, size_t buflen)
* Wait up to timeout_ms for udev to set up the device node. The device is
* considered ready when libudev determines it has been initialized, all of
* the device links have been verified to exist, and it has been allowed to
* settle. At this point the device the device can be accessed reliably.
* Depending on the complexity of the udev rules this process could take
* several seconds.
* settle. At this point the device can be accessed reliably. Depending on
* the complexity of the udev rules this process could take several seconds.
*/
int
zpool_label_disk_wait(const char *path, int timeout_ms)
Expand Down
2 changes: 1 addition & 1 deletion sys/contrib/openzfs/man/man4/zfs.4
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2172,7 +2172,7 @@ if a volatile out-of-order write cache is enabled.
Disable intent logging replay.
Can be disabled for recovery from corrupted ZIL.
.
.It Sy zil_slog_bulk Ns = Ns Sy 786432 Ns B Po 768 KiB Pc Pq u64
.It Sy zil_slog_bulk Ns = Ns Sy 67108864 Ns B Po 64 MiB Pc Pq u64
Limit SLOG write size per commit executed with synchronous priority.
Any writes above that will be executed with lower (asynchronous) priority
to limit potential SLOG device abuse by single active ZIL writer.
Expand Down
Loading

0 comments on commit 22b267e

Please sign in to comment.