system/freebsd-src
system/freebsd-src
1
0
Fork 0
mirror of https://github.com/freebsd/freebsd-src synced 2024-10-07 00:50:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

zfs: merge openzfs/zfs@8f1b7a6fa

Browse source 
Notable upstream pull request merges:
 #15839 c3f2f1aa2 vdev probe to slow disk can stall mmp write checker
 #15888 5044c4e3f Fast Dedup: ZAP Shrinking
 #15996 db499e68f Overflowing refreservation is bad
 #16118 67d13998b Make more taskq parameters writable
 #16128 21bc066ec Fix updating the zvol_htable when renaming a zvol
 #16130 645b83307 Improve write issue taskqs utilization
 #16131 8fd3a5d02 Slightly improve dnode hash
 #16134 a6edc0adb zio: try to execute TYPE_NULL ZIOs on the current task
 #16141 b28461b7c Fix arcstats for FreeBSD after zfetch support

Obtained from:	OpenZFS
OpenZFS commit:	8f1b7a6fa6
This commit is contained in:
Martin Matuska 2024-05-03 17:56:40 +02:00
parent 04ea5e9f84 8f1b7a6fa6
commit b985c9cafd
49 changed files with 1332 additions and 271 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
sys/contrib/openzfs/cmd/arcstat.in
View file

@ -200,6 +200,8 @@ if sys.platform.startswith('freebsd'):
k = [ctl for ctl in sysctl.filter('kstat.zfs.misc.arcstats')
if ctl.type != sysctl.CTLTYPE_NODE]
k += [ctl for ctl in sysctl.filter('kstat.zfs.misc.zfetchstats')
if ctl.type != sysctl.CTLTYPE_NODE]
if not k:
sys.exit(1)
@ -211,8 +213,12 @@ if sys.platform.startswith('freebsd'):
continue
name, value = s.name, s.value
# Trims 'kstat.zfs.misc.arcstats' from the name
kstat[name[24:]] = int(value)
if "arcstats" in name:
# Trims 'kstat.zfs.misc.arcstats' from the name
kstat[name[24:]] = int(value)
else:
kstat["zfetch_" + name[27:]] = int(value)
elif sys.platform.startswith('linux'):
def kstat_update():

2
sys/contrib/openzfs/cmd/zpool/os/linux/zpool_vdev_os.c
View file

@ -438,7 +438,7 @@ static char *zpool_sysfs_gets(char *path)
return (NULL);
}
buf = calloc(sizeof (*buf), statbuf.st_size + 1);
buf = calloc(statbuf.st_size + 1, sizeof (*buf));
if (buf == NULL) {
close(fd);
return (NULL);

2
sys/contrib/openzfs/cmd/zpool/zpool_main.c
View file

@ -9050,7 +9050,7 @@ status_callback(zpool_handle_t *zhp, void *data)
printf_color(ANSI_BOLD, gettext("action: "));
printf_color(ANSI_YELLOW, gettext("Make sure the pool's devices"
" are connected, then reboot your system and\n\timport the "
"pool.\n"));
"pool or run 'zpool clear' to resume the pool.\n"));
break;
case ZPOOL_STATUS_IO_FAILURE_WAIT:

9
sys/contrib/openzfs/config/always-pyzfs.m4
View file

@ -80,10 +80,11 @@ AC_DEFUN([ZFS_AC_CONFIG_ALWAYS_PYZFS], [
[AC_MSG_ERROR("Python $PYTHON_VERSION unknown")]
)
AX_PYTHON_DEVEL([$PYTHON_REQUIRED_VERSION], [
AS_IF([test "x$enable_pyzfs" = xyes], [
AC_MSG_ERROR("Python $PYTHON_REQUIRED_VERSION development library is not installed")
], [test "x$enable_pyzfs" != xno], [
AS_IF([test "x$enable_pyzfs" = xyes], [
AX_PYTHON_DEVEL([$PYTHON_REQUIRED_VERSION])
], [
AX_PYTHON_DEVEL([$PYTHON_REQUIRED_VERSION], [true])
AS_IF([test "x$ax_python_devel_found" = xno], [
enable_pyzfs=no
])
])

341
sys/contrib/openzfs/config/ax_python_devel.m4
View file

@ -4,18 +4,13 @@
#
# SYNOPSIS
#
# AX_PYTHON_DEVEL([version], [action-if-not-found])
# AX_PYTHON_DEVEL([version[,optional]])
#
# DESCRIPTION
#
# Note: Defines as a precious variable "PYTHON_VERSION". Don't override it
# in your configure.ac.
#
# Note: this is a slightly modified version of the original AX_PYTHON_DEVEL
# macro which accepts an additional [action-if-not-found] argument. This
# allow to detect if Python development is available without aborting the
# configure phase with an hard error in case it is not.
#
# This macro checks for Python and tries to get the include path to
# 'Python.h'. It provides the $(PYTHON_CPPFLAGS) and $(PYTHON_LIBS) output
# variables. It also exports $(PYTHON_EXTRA_LIBS) and
@ -28,6 +23,11 @@
# version number. Don't use "PYTHON_VERSION" for this: that environment
# variable is declared as precious and thus reserved for the end-user.
#
# By default this will fail if it does not detect a development version of
# python. If you want it to continue, set optional to true, like
# AX_PYTHON_DEVEL([], [true]). The ax_python_devel_found variable will be
# "no" if it fails.
#
# This macro should work for all versions of Python >= 2.1.0. As an end
# user, you can disable the check for the python version by setting the
# PYTHON_NOVERSIONCHECK environment variable to something else than the
@ -45,7 +45,6 @@
# Copyright (c) 2009 Matteo Settenvini <matteo@member.fsf.org>
# Copyright (c) 2009 Horst Knorr <hk_classes@knoda.org>
# Copyright (c) 2013 Daniel Mullner <muellner@math.stanford.edu>
# Copyright (c) 2018 loli10K <ezomori.nozomu@gmail.com>
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
@ -73,10 +72,18 @@
# modified version of the Autoconf Macro, you may extend this special
# exception to the GPL to apply to your modified version as well.
#serial 21
#serial 36
AU_ALIAS([AC_PYTHON_DEVEL], [AX_PYTHON_DEVEL])
AC_DEFUN([AX_PYTHON_DEVEL],[
# Get whether it's optional
if test -z "$2"; then
ax_python_devel_optional=false
else
ax_python_devel_optional=$2
fi
ax_python_devel_found=yes
#
# Allow the use of a (user set) custom python version
#
@ -87,23 +94,26 @@ AC_DEFUN([AX_PYTHON_DEVEL],[
AC_PATH_PROG([PYTHON],[python[$PYTHON_VERSION]])
if test -z "$PYTHON"; then
m4_ifvaln([$2],[$2],[
AC_MSG_ERROR([Cannot find python$PYTHON_VERSION in your system path])
PYTHON_VERSION=""
])
AC_MSG_WARN([Cannot find python$PYTHON_VERSION in your system path])
if ! $ax_python_devel_optional; then
AC_MSG_ERROR([Giving up, python development not available])
fi
ax_python_devel_found=no
PYTHON_VERSION=""
fi
#
# Check for a version of Python >= 2.1.0
#
AC_MSG_CHECKING([for a version of Python >= '2.1.0'])
ac_supports_python_ver=`$PYTHON -c "import sys; \
if test $ax_python_devel_found = yes; then
#
# Check for a version of Python >= 2.1.0
#
AC_MSG_CHECKING([for a version of Python >= '2.1.0'])
ac_supports_python_ver=`$PYTHON -c "import sys; \
ver = sys.version.split ()[[0]]; \
print (ver >= '2.1.0')"`
if test "$ac_supports_python_ver" != "True"; then
if test "$ac_supports_python_ver" != "True"; then
if test -z "$PYTHON_NOVERSIONCHECK"; then
AC_MSG_RESULT([no])
AC_MSG_FAILURE([
AC_MSG_WARN([
This version of the AC@&t@_PYTHON_DEVEL macro
doesn't work properly with versions of Python before
2.1.0. You may need to re-run configure, setting the
@ -112,20 +122,27 @@ PYTHON_EXTRA_LIBS and PYTHON_EXTRA_LDFLAGS by hand.
Moreover, to disable this check, set PYTHON_NOVERSIONCHECK
to something else than an empty string.
])
if ! $ax_python_devel_optional; then
AC_MSG_FAILURE([Giving up])
fi
ax_python_devel_found=no
PYTHON_VERSION=""
else
AC_MSG_RESULT([skip at user request])
fi
else
else
AC_MSG_RESULT([yes])
fi
fi
#
# If the macro parameter ``version'' is set, honour it.
# A Python shim class, VPy, is used to implement correct version comparisons via
# string expressions, since e.g. a naive textual ">= 2.7.3" won't work for
# Python 2.7.10 (the ".1" being evaluated as less than ".3").
#
if test -n "$1"; then
if test $ax_python_devel_found = yes; then
#
# If the macro parameter ``version'' is set, honour it.
# A Python shim class, VPy, is used to implement correct version comparisons via
# string expressions, since e.g. a naive textual ">= 2.7.3" won't work for
# Python 2.7.10 (the ".1" being evaluated as less than ".3").
#
if test -n "$1"; then
AC_MSG_CHECKING([for a version of Python $1])
cat << EOF > ax_python_devel_vpy.py
class VPy:
@ -133,7 +150,7 @@ class VPy:
return tuple(map(int, s.strip().replace("rc", ".").split(".")))
def __init__(self):
import sys
self.vpy = tuple(sys.version_info)
self.vpy = tuple(sys.version_info)[[:3]]
def __eq__(self, s):
return self.vpy == self.vtup(s)
def __ne__(self, s):
@ -155,25 +172,69 @@ EOF
AC_MSG_RESULT([yes])
else
AC_MSG_RESULT([no])
AC_MSG_ERROR([this package requires Python $1.
AC_MSG_WARN([this package requires Python $1.
If you have it installed, but it isn't the default Python
interpreter in your system path, please pass the PYTHON_VERSION
variable to configure. See ``configure --help'' for reference.
])
if ! $ax_python_devel_optional; then
AC_MSG_ERROR([Giving up])
fi
ax_python_devel_found=no
PYTHON_VERSION=""
fi
fi
fi
#
# Check for Python include path
#
#
AC_MSG_CHECKING([for Python include path])
if test -z "$PYTHON_CPPFLAGS"; then
python_path=`$PYTHON -c "import sysconfig; \
print (sysconfig.get_path('include'));"`
plat_python_path=`$PYTHON -c "import sysconfig; \
print (sysconfig.get_path('platinclude'));"`
if test $ax_python_devel_found = yes; then
#
# Check if you have distutils, else fail
#
AC_MSG_CHECKING([for the sysconfig Python package])
ac_sysconfig_result=`$PYTHON -c "import sysconfig" 2>&1`
if test $? -eq 0; then
AC_MSG_RESULT([yes])
IMPORT_SYSCONFIG="import sysconfig"
else
AC_MSG_RESULT([no])
AC_MSG_CHECKING([for the distutils Python package])
ac_sysconfig_result=`$PYTHON -c "from distutils import sysconfig" 2>&1`
if test $? -eq 0; then
AC_MSG_RESULT([yes])
IMPORT_SYSCONFIG="from distutils import sysconfig"
else
AC_MSG_WARN([cannot import Python module "distutils".
Please check your Python installation. The error was:
$ac_sysconfig_result])
if ! $ax_python_devel_optional; then
AC_MSG_ERROR([Giving up])
fi
ax_python_devel_found=no
PYTHON_VERSION=""
fi
fi
fi
if test $ax_python_devel_found = yes; then
#
# Check for Python include path
#
AC_MSG_CHECKING([for Python include path])
if test -z "$PYTHON_CPPFLAGS"; then
if test "$IMPORT_SYSCONFIG" = "import sysconfig"; then
# sysconfig module has different functions
python_path=`$PYTHON -c "$IMPORT_SYSCONFIG; \
print (sysconfig.get_path ('include'));"`
plat_python_path=`$PYTHON -c "$IMPORT_SYSCONFIG; \
print (sysconfig.get_path ('platinclude'));"`
else
# old distutils way
python_path=`$PYTHON -c "$IMPORT_SYSCONFIG; \
print (sysconfig.get_python_inc ());"`
plat_python_path=`$PYTHON -c "$IMPORT_SYSCONFIG; \
print (sysconfig.get_python_inc (plat_specific=1));"`
fi
if test -n "${python_path}"; then
if test "${plat_python_path}" != "${python_path}"; then
python_path="-I$python_path -I$plat_python_path"
@ -182,15 +243,15 @@ variable to configure. See ``configure --help'' for reference.
fi
fi
PYTHON_CPPFLAGS=$python_path
fi
AC_MSG_RESULT([$PYTHON_CPPFLAGS])
AC_SUBST([PYTHON_CPPFLAGS])
fi
AC_MSG_RESULT([$PYTHON_CPPFLAGS])
AC_SUBST([PYTHON_CPPFLAGS])
#
# Check for Python library path
#
AC_MSG_CHECKING([for Python library path])
if test -z "$PYTHON_LIBS"; then
#
# Check for Python library path
#
AC_MSG_CHECKING([for Python library path])
if test -z "$PYTHON_LIBS"; then
# (makes two attempts to ensure we've got a version number
# from the interpreter)
ac_python_version=`cat<<EOD | $PYTHON -
@ -208,7 +269,7 @@ EOD`
ac_python_version=$PYTHON_VERSION
else
ac_python_version=`$PYTHON -c "import sys; \
print ('.'.join(sys.version.split('.')[[:2]]))"`
print ("%d.%d" % sys.version_info[[:2]])"`
fi
fi
@ -220,7 +281,7 @@ EOD`
ac_python_libdir=`cat<<EOD | $PYTHON -
# There should be only one
import sysconfig
$IMPORT_SYSCONFIG
e = sysconfig.get_config_var('LIBDIR')
if e is not None:
print (e)
@ -229,7 +290,7 @@ EOD`
# Now, for the library:
ac_python_library=`cat<<EOD | $PYTHON -
import sysconfig
$IMPORT_SYSCONFIG
c = sysconfig.get_config_vars()
if 'LDVERSION' in c:
print ('python'+c[['LDVERSION']])
@ -249,88 +310,140 @@ EOD`
else
# old way: use libpython from python_configdir
ac_python_libdir=`$PYTHON -c \
"import sysconfig; \
"from sysconfig import get_python_lib as f; \
import os; \
print (os.path.join(sysconfig.get_path('platstdlib'), 'config'));"`
print (os.path.join(f(plat_specific=1, standard_lib=1), 'config'));"`
PYTHON_LIBS="-L$ac_python_libdir -lpython$ac_python_version"
fi
if test -z "PYTHON_LIBS"; then
m4_ifvaln([$2],[$2],[
AC_MSG_ERROR([
AC_MSG_WARN([
Cannot determine location of your Python DSO. Please check it was installed with
dynamic libraries enabled, or try setting PYTHON_LIBS by hand.
])
])
if ! $ax_python_devel_optional; then
AC_MSG_ERROR([Giving up])
fi
ax_python_devel_found=no
PYTHON_VERSION=""
fi
fi
fi
AC_MSG_RESULT([$PYTHON_LIBS])
AC_SUBST([PYTHON_LIBS])
#
# Check for site packages
#
AC_MSG_CHECKING([for Python site-packages path])
if test -z "$PYTHON_SITE_PKG"; then
PYTHON_SITE_PKG=`$PYTHON -c "import distutils.sysconfig; \
print (distutils.sysconfig.get_python_lib(0,0));" 2>/dev/null || \
$PYTHON -c "import sysconfig; \
print (sysconfig.get_path('purelib'));"`
fi
AC_MSG_RESULT([$PYTHON_SITE_PKG])
AC_SUBST([PYTHON_SITE_PKG])
if test $ax_python_devel_found = yes; then
AC_MSG_RESULT([$PYTHON_LIBS])
AC_SUBST([PYTHON_LIBS])
#
# libraries which must be linked in when embedding
#
AC_MSG_CHECKING(python extra libraries)
if test -z "$PYTHON_EXTRA_LIBS"; then
PYTHON_EXTRA_LIBS=`$PYTHON -c "import sysconfig; \
#
# Check for site packages
#
AC_MSG_CHECKING([for Python site-packages path])
if test -z "$PYTHON_SITE_PKG"; then
if test "$IMPORT_SYSCONFIG" = "import sysconfig"; then
PYTHON_SITE_PKG=`$PYTHON -c "
$IMPORT_SYSCONFIG;
if hasattr(sysconfig, 'get_default_scheme'):
scheme = sysconfig.get_default_scheme()
else:
scheme = sysconfig._get_default_scheme()
if scheme == 'posix_local':
# Debian's default scheme installs to /usr/local/ but we want to find headers in /usr/
scheme = 'posix_prefix'
prefix = '$prefix'
if prefix == 'NONE':
prefix = '$ac_default_prefix'
sitedir = sysconfig.get_path('purelib', scheme, vars={'base': prefix})
print(sitedir)"`
else
# distutils.sysconfig way
PYTHON_SITE_PKG=`$PYTHON -c "$IMPORT_SYSCONFIG; \
print (sysconfig.get_python_lib(0,0));"`
fi
fi
AC_MSG_RESULT([$PYTHON_SITE_PKG])
AC_SUBST([PYTHON_SITE_PKG])
#
# Check for platform-specific site packages
#
AC_MSG_CHECKING([for Python platform specific site-packages path])
if test -z "$PYTHON_PLATFORM_SITE_PKG"; then
if test "$IMPORT_SYSCONFIG" = "import sysconfig"; then
PYTHON_PLATFORM_SITE_PKG=`$PYTHON -c "
$IMPORT_SYSCONFIG;
if hasattr(sysconfig, 'get_default_scheme'):
scheme = sysconfig.get_default_scheme()
else:
scheme = sysconfig._get_default_scheme()
if scheme == 'posix_local':
# Debian's default scheme installs to /usr/local/ but we want to find headers in /usr/
scheme = 'posix_prefix'
prefix = '$prefix'
if prefix == 'NONE':
prefix = '$ac_default_prefix'
sitedir = sysconfig.get_path('platlib', scheme, vars={'platbase': prefix})
print(sitedir)"`
else
# distutils.sysconfig way
PYTHON_PLATFORM_SITE_PKG=`$PYTHON -c "$IMPORT_SYSCONFIG; \
print (sysconfig.get_python_lib(1,0));"`
fi
fi
AC_MSG_RESULT([$PYTHON_PLATFORM_SITE_PKG])
AC_SUBST([PYTHON_PLATFORM_SITE_PKG])
#
# libraries which must be linked in when embedding
#
AC_MSG_CHECKING(python extra libraries)
if test -z "$PYTHON_EXTRA_LIBS"; then
PYTHON_EXTRA_LIBS=`$PYTHON -c "$IMPORT_SYSCONFIG; \
conf = sysconfig.get_config_var; \
print (conf('LIBS') + ' ' + conf('SYSLIBS'))"`
fi
AC_MSG_RESULT([$PYTHON_EXTRA_LIBS])
AC_SUBST(PYTHON_EXTRA_LIBS)
fi
AC_MSG_RESULT([$PYTHON_EXTRA_LIBS])
AC_SUBST(PYTHON_EXTRA_LIBS)
#
# linking flags needed when embedding
#
AC_MSG_CHECKING(python extra linking flags)
if test -z "$PYTHON_EXTRA_LDFLAGS"; then
PYTHON_EXTRA_LDFLAGS=`$PYTHON -c "import sysconfig; \
#
# linking flags needed when embedding
#
AC_MSG_CHECKING(python extra linking flags)
if test -z "$PYTHON_EXTRA_LDFLAGS"; then
PYTHON_EXTRA_LDFLAGS=`$PYTHON -c "$IMPORT_SYSCONFIG; \
conf = sysconfig.get_config_var; \
print (conf('LINKFORSHARED'))"`
fi
AC_MSG_RESULT([$PYTHON_EXTRA_LDFLAGS])
AC_SUBST(PYTHON_EXTRA_LDFLAGS)
# Hack for macos, it sticks this in here.
PYTHON_EXTRA_LDFLAGS=`echo $PYTHON_EXTRA_LDFLAGS | sed 's/CoreFoundation.*$/CoreFoundation/'`
fi
AC_MSG_RESULT([$PYTHON_EXTRA_LDFLAGS])
AC_SUBST(PYTHON_EXTRA_LDFLAGS)
#
# final check to see if everything compiles alright
#
AC_MSG_CHECKING([consistency of all components of python development environment])
# save current global flags
ac_save_LIBS="$LIBS"
ac_save_LDFLAGS="$LDFLAGS"
ac_save_CPPFLAGS="$CPPFLAGS"
LIBS="$ac_save_LIBS $PYTHON_LIBS $PYTHON_EXTRA_LIBS $PYTHON_EXTRA_LIBS"
LDFLAGS="$ac_save_LDFLAGS $PYTHON_EXTRA_LDFLAGS"
CPPFLAGS="$ac_save_CPPFLAGS $PYTHON_CPPFLAGS"
AC_LANG_PUSH([C])
AC_LINK_IFELSE([
#
# final check to see if everything compiles alright
#
AC_MSG_CHECKING([consistency of all components of python development environment])
# save current global flags
ac_save_LIBS="$LIBS"
ac_save_LDFLAGS="$LDFLAGS"
ac_save_CPPFLAGS="$CPPFLAGS"
LIBS="$ac_save_LIBS $PYTHON_LIBS $PYTHON_EXTRA_LIBS"
LDFLAGS="$ac_save_LDFLAGS $PYTHON_EXTRA_LDFLAGS"
CPPFLAGS="$ac_save_CPPFLAGS $PYTHON_CPPFLAGS"
AC_LANG_PUSH([C])
AC_LINK_IFELSE([
AC_LANG_PROGRAM([[#include <Python.h>]],
[[Py_Initialize();]])
],[pythonexists=yes],[pythonexists=no])
AC_LANG_POP([C])
# turn back to default flags
CPPFLAGS="$ac_save_CPPFLAGS"
LIBS="$ac_save_LIBS"
LDFLAGS="$ac_save_LDFLAGS"
AC_LANG_POP([C])
# turn back to default flags
CPPFLAGS="$ac_save_CPPFLAGS"
LIBS="$ac_save_LIBS"
LDFLAGS="$ac_save_LDFLAGS"
AC_MSG_RESULT([$pythonexists])
AC_MSG_RESULT([$pythonexists])
if test ! "x$pythonexists" = "xyes"; then
m4_ifvaln([$2],[$2],[
AC_MSG_FAILURE([
if test ! "x$pythonexists" = "xyes"; then
AC_MSG_WARN([
Could not link test program to Python. Maybe the main Python library has been
installed in some non-standard library path. If so, pass it to configure,
via the LIBS environment variable.
@ -340,9 +453,13 @@ EOD`
You probably have to install the development version of the Python package
for your distribution. The exact name of this package varies among them.
============================================================================
])
PYTHON_VERSION=""
])
])
if ! $ax_python_devel_optional; then
AC_MSG_ERROR([Giving up])
fi
ax_python_devel_found=no
PYTHON_VERSION=""
fi
fi
#

4
sys/contrib/openzfs/config/find_system_library.m4
View file

@ -90,8 +90,8 @@ AC_DEFUN([ZFS_AC_FIND_SYSTEM_LIBRARY], [
AC_DEFINE([HAVE_][$1], [1], [Define if you have [$5]])
$7
],[dnl ELSE
AC_SUBST([$1]_CFLAGS, [])
AC_SUBST([$1]_LIBS, [])
AC_SUBST([$1]_CFLAGS, [""])
AC_SUBST([$1]_LIBS, [""])
AC_MSG_WARN([cannot find [$5] via pkg-config or in the standard locations])
$8
])

14
sys/contrib/openzfs/config/user-backtrace.m4 Normal file
View file

@ -0,0 +1,14 @@
dnl
dnl backtrace(), for userspace assertions. glibc has this directly in libc.
dnl FreeBSD and (sometimes) musl have it in a separate -lexecinfo. It's assumed
dnl that this will also get the companion function backtrace_symbols().
dnl
AC_DEFUN([ZFS_AC_CONFIG_USER_BACKTRACE], [
AX_SAVE_FLAGS
LIBS=""
AC_SEARCH_LIBS([backtrace], [execinfo], [
AC_DEFINE(HAVE_BACKTRACE, 1, [backtrace() is available])
AC_SUBST([BACKTRACE_LIBS], ["$LIBS"])
])
AX_RESTORE_FLAGS
])

44
sys/contrib/openzfs/config/user-libunwind.m4 Normal file
View file

@ -0,0 +1,44 @@
dnl
dnl Checks for libunwind, which usually does a better job than backtrace() when
dnl resolving symbols in the stack backtrace. Newer versions have support for
dnl getting info about the object file the function came from, so we look for
dnl that too and use it if found.
dnl
AC_DEFUN([ZFS_AC_CONFIG_USER_LIBUNWIND], [
AC_ARG_WITH([libunwind],
AS_HELP_STRING([--with-libunwind],
[use libunwind for backtraces in userspace assertions]),
[],
[with_libunwind=auto])
AS_IF([test "x$with_libunwind" != "xno"], [
ZFS_AC_FIND_SYSTEM_LIBRARY(LIBUNWIND, [libunwind], [libunwind.h], [], [unwind], [], [
dnl unw_get_elf_filename() is sometimes a macro, other
dnl times a proper symbol, so we can't just do a link
dnl check; we need to include the header properly.
AX_SAVE_FLAGS
CFLAGS="$CFLAGS $LIBUNWIND_CFLAGS"
LIBS="$LIBS $LIBUNWIND_LIBS"
AC_MSG_CHECKING([for unw_get_elf_filename in libunwind])
AC_LINK_IFELSE([
AC_LANG_PROGRAM([
#define UNW_LOCAL_ONLY
#include <libunwind.h>
], [
unw_get_elf_filename(0, 0, 0, 0);
])
], [
AC_MSG_RESULT([yes])
AC_DEFINE(HAVE_LIBUNWIND_ELF, 1,
[libunwind has unw_get_elf_filename])
], [
AC_MSG_RESULT([no])
])
AX_RESTORE_FLAGS
], [
AS_IF([test "x$with_libunwind" = "xyes"], [
AC_MSG_FAILURE([--with-libunwind was given, but libunwind is not available, try installing libunwind-devel])
])
])
])
])

4
sys/contrib/openzfs/config/user.m4
View file

@ -26,12 +26,14 @@ AC_DEFUN([ZFS_AC_CONFIG_USER], [
ZFS_AC_CONFIG_USER_AIO_H
ZFS_AC_CONFIG_USER_CLOCK_GETTIME
ZFS_AC_CONFIG_USER_PAM
ZFS_AC_CONFIG_USER_BACKTRACE
ZFS_AC_CONFIG_USER_LIBUNWIND
ZFS_AC_CONFIG_USER_RUNSTATEDIR
ZFS_AC_CONFIG_USER_MAKEDEV_IN_SYSMACROS
ZFS_AC_CONFIG_USER_MAKEDEV_IN_MKDEV
ZFS_AC_CONFIG_USER_ZFSEXEC
AC_CHECK_FUNCS([execvpe issetugid mlockall strlcat strlcpy])
AC_CHECK_FUNCS([execvpe issetugid mlockall strlcat strlcpy gettid])
AC_SUBST(RM)
])

2
sys/contrib/openzfs/contrib/debian/control
View file

@ -189,7 +189,7 @@ Depends: dkms (>> 2.1.1.2-5),
file,
libc6-dev | libc-dev,
lsb-release,
python3-distutils | libpython3-stdlib (<< 3.6.4),
python3 (>> 3.12) | python3-distutils | libpython3-stdlib (<< 3.6.4),
${misc:Depends},
${perl:Depends}
Recommends: openzfs-zfs-zed, openzfs-zfsutils (>= ${source:Version}), ${linux:Recommends}

27
sys/contrib/openzfs/include/os/linux/kernel/linux/blkdev_compat.h
View file

@ -94,6 +94,33 @@ blk_queue_set_write_cache(struct request_queue *q, bool wc, bool fua)
#endif
}
/*
* Detect if a device has a write cache. Used to set the intial value for the
* vdev nowritecache flag.
*
* 4.10: QUEUE_FLAG_WC added. Initialised by the driver, but can be changed
* later by the operator. If not set, kernel will return flush requests
* immediately without doing anything.
* 6.6: QUEUE_FLAG_HW_WC added. Initialised by the driver, can't be changed.
* Only controls if the operator is allowed to change _WC. Initial version
* buggy; aliased to QUEUE_FLAG_FUA, so unuseable.
* 6.6.10, 6.7: QUEUE_FLAG_HW_WC fixed.
*
* Older than 4.10 we just assume write cache, and let the normal flush fail
* detection apply.
*/
static inline boolean_t
zfs_bdev_has_write_cache(struct block_device *bdev)
{
#if defined(QUEUE_FLAG_HW_WC) && QUEUE_FLAG_HW_WC != QUEUE_FLAG_FUA
return (test_bit(QUEUE_FLAG_HW_WC, &bdev_get_queue(bdev)->queue_flags));
#elif defined(QUEUE_FLAG_WC)
return (test_bit(QUEUE_FLAG_WC, &bdev_get_queue(bdev)->queue_flags));
#else
return (B_TRUE);
#endif
}
static inline void
blk_queue_set_read_ahead(struct request_queue *q, unsigned long ra_pages)
{

6
sys/contrib/openzfs/include/sys/spa.h
View file

@ -770,7 +770,7 @@ extern int bpobj_enqueue_free_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx);
#define SPA_ASYNC_CONFIG_UPDATE 0x01
#define SPA_ASYNC_REMOVE 0x02
#define SPA_ASYNC_PROBE 0x04
#define SPA_ASYNC_FAULT_VDEV 0x04
#define SPA_ASYNC_RESILVER_DONE 0x08
#define SPA_ASYNC_RESILVER 0x10
#define SPA_ASYNC_AUTOEXPAND 0x20
@ -829,6 +829,8 @@ extern uint_t zfs_sync_pass_deferred_free;
/* spa sync taskqueues */
taskq_t *spa_sync_tq_create(spa_t *spa, const char *name);
void spa_sync_tq_destroy(spa_t *spa);
uint_t spa_acq_allocator(spa_t *spa);
void spa_rel_allocator(spa_t *spa, uint_t allocator);
void spa_select_allocator(zio_t *zio);
/* spa namespace global mutex */
@ -1123,6 +1125,8 @@ extern uint32_t spa_get_hostid(spa_t *spa);
extern void spa_activate_allocation_classes(spa_t *, dmu_tx_t *);
extern boolean_t spa_livelist_delete_check(spa_t *spa);
extern boolean_t spa_mmp_remote_host_activity(spa_t *spa);
extern spa_mode_t spa_mode(spa_t *spa);
extern uint64_t zfs_strtonum(const char *str, char **nptr);

9
sys/contrib/openzfs/include/sys/spa_impl.h
View file

@ -63,6 +63,12 @@ typedef struct spa_alloc {
avl_tree_t spaa_tree;
} ____cacheline_aligned spa_alloc_t;
typedef struct spa_allocs_use {
kmutex_t sau_lock;
uint_t sau_rotor;
boolean_t sau_inuse[];
} spa_allocs_use_t;
typedef struct spa_error_entry {
zbookmark_phys_t se_bookmark;
char *se_name;
@ -192,7 +198,7 @@ typedef struct spa_taskqs {
/* one for each thread in the spa sync taskq */
typedef struct spa_syncthread_info {
kthread_t *sti_thread;
taskq_t *sti_wr_iss_tq; /* assigned wr_iss taskq */
uint_t sti_allocator;
} spa_syncthread_info_t;
typedef enum spa_all_vdev_zap_action {
@ -270,6 +276,7 @@ struct spa {
* allocation performance in write-heavy workloads.
*/
spa_alloc_t *spa_allocs;
spa_allocs_use_t *spa_allocs_use;
int spa_alloc_count;
int spa_active_allocator; /* selectable allocator */

16
sys/contrib/openzfs/include/sys/uberblock_impl.h
View file

@ -50,20 +50,20 @@ extern "C" {
#define MMP_SEQ_VALID_BIT 0x02
#define MMP_FAIL_INT_VALID_BIT 0x04
#define MMP_VALID(ubp) (ubp->ub_magic == UBERBLOCK_MAGIC && \
ubp->ub_mmp_magic == MMP_MAGIC)
#define MMP_INTERVAL_VALID(ubp) (MMP_VALID(ubp) && (ubp->ub_mmp_config & \
#define MMP_VALID(ubp) ((ubp)->ub_magic == UBERBLOCK_MAGIC && \
(ubp)->ub_mmp_magic == MMP_MAGIC)
#define MMP_INTERVAL_VALID(ubp) (MMP_VALID(ubp) && ((ubp)->ub_mmp_config & \
MMP_INTERVAL_VALID_BIT))
#define MMP_SEQ_VALID(ubp) (MMP_VALID(ubp) && (ubp->ub_mmp_config & \
#define MMP_SEQ_VALID(ubp) (MMP_VALID(ubp) && ((ubp)->ub_mmp_config & \
MMP_SEQ_VALID_BIT))
#define MMP_FAIL_INT_VALID(ubp) (MMP_VALID(ubp) && (ubp->ub_mmp_config & \
#define MMP_FAIL_INT_VALID(ubp) (MMP_VALID(ubp) && ((ubp)->ub_mmp_config & \
MMP_FAIL_INT_VALID_BIT))
#define MMP_INTERVAL(ubp) ((ubp->ub_mmp_config & 0x00000000FFFFFF00) \
#define MMP_INTERVAL(ubp) (((ubp)->ub_mmp_config & 0x00000000FFFFFF00) \
>> 8)
#define MMP_SEQ(ubp) ((ubp->ub_mmp_config & 0x0000FFFF00000000) \
#define MMP_SEQ(ubp) (((ubp)->ub_mmp_config & 0x0000FFFF00000000) \
>> 32)
#define MMP_FAIL_INT(ubp) ((ubp->ub_mmp_config & 0xFFFF000000000000) \
#define MMP_FAIL_INT(ubp) (((ubp)->ub_mmp_config & 0xFFFF000000000000) \
>> 48)
#define MMP_INTERVAL_SET(write) \

2
sys/contrib/openzfs/include/sys/vdev_impl.h
View file

@ -273,7 +273,7 @@ struct vdev {
txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
boolean_t vdev_remove_wanted; /* async remove wanted? */
boolean_t vdev_probe_wanted; /* async probe wanted? */
boolean_t vdev_fault_wanted; /* async faulted wanted? */
list_node_t vdev_config_dirty_node; /* config dirty list */
list_node_t vdev_state_dirty_node; /* state dirty list */
uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */

8
sys/contrib/openzfs/include/sys/zfs_context.h
View file

@ -228,9 +228,9 @@ typedef pthread_t kthread_t;
#define thread_create_named(name, stk, stksize, func, arg, len, \
pp, state, pri) \
zk_thread_create(func, arg, stksize, state)
zk_thread_create(name, func, arg, stksize, state)
#define thread_create(stk, stksize, func, arg, len, pp, state, pri) \
zk_thread_create(func, arg, stksize, state)
zk_thread_create(#func, func, arg, stksize, state)
#define thread_exit() pthread_exit(NULL)
#define thread_join(t) pthread_join((pthread_t)(t), NULL)
@ -246,8 +246,8 @@ extern struct proc p0;
#define PS_NONE -1
extern kthread_t *zk_thread_create(void (*func)(void *), void *arg,
size_t stksize, int state);
extern kthread_t *zk_thread_create(const char *name, void (*func)(void *),
void *arg, size_t stksize, int state);
#define issig(why) (FALSE)
#define ISSIG(thr, why) (FALSE)

3
sys/contrib/openzfs/include/sys/zio.h
View file

@ -528,9 +528,6 @@ struct zio {
/* Taskq dispatching state */
taskq_ent_t io_tqent;
/* write issue taskq selection, based upon sync thread */
taskq_t *io_wr_iss_tq;
};
enum blk_verify_flag {

4
sys/contrib/openzfs/lib/libspl/Makefile.am
View file

@ -1,6 +1,6 @@
include $(srcdir)/%D%/include/Makefile.am
libspl_assert_la_CFLAGS = $(AM_CFLAGS) $(LIBRARY_CFLAGS)
libspl_assert_la_CFLAGS = $(AM_CFLAGS) $(LIBRARY_CFLAGS) $(LIBUNWIND_CFLAGS)
libspl_la_CFLAGS = $(libspl_assert_la_CFLAGS)
noinst_LTLIBRARIES += libspl_assert.la libspl.la
@ -43,3 +43,5 @@ libspl_la_LIBADD = \
libspl_assert.la
libspl_la_LIBADD += $(LIBATOMIC_LIBS) $(LIBCLOCK_GETTIME)
libspl_assert_la_LIBADD = $(BACKTRACE_LIBS) $(LIBUNWIND_LIBS)

93
sys/contrib/openzfs/lib/libspl/assert.c
View file

@ -22,8 +22,81 @@
* Copyright 2008 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
/*
* Copyright (c) 2024, Rob Norris <robn@despairlabs.com>
*/
#include <assert.h>
#include <pthread.h>
#if defined(__linux__)
#include <errno.h>
#include <sys/prctl.h>
#ifdef HAVE_GETTID
#define libspl_gettid() gettid()
#else
#include <sys/syscall.h>
#define libspl_gettid() ((pid_t)syscall(__NR_gettid))
#endif
#define libspl_getprogname() (program_invocation_short_name)
#define libspl_getthreadname(buf, len) \
prctl(PR_GET_NAME, (unsigned long)(buf), 0, 0, 0)
#elif defined(__FreeBSD__)
#include <pthread_np.h>
#define libspl_gettid() pthread_getthreadid_np()
#define libspl_getprogname() getprogname()
#define libspl_getthreadname(buf, len) \
pthread_getname_np(pthread_self(), buf, len);
#endif
#if defined(HAVE_LIBUNWIND)
#define UNW_LOCAL_ONLY
#include <libunwind.h>
static inline void
libspl_dump_backtrace(void)
{
unw_context_t uc;
unw_cursor_t cp;
unw_word_t ip, off;
char funcname[128];
#ifdef HAVE_LIBUNWIND_ELF
char objname[128];
unw_word_t objoff;
#endif
fprintf(stderr, "Call trace:\n");
unw_getcontext(&uc);
unw_init_local(&cp, &uc);
while (unw_step(&cp) > 0) {
unw_get_reg(&cp, UNW_REG_IP, &ip);
unw_get_proc_name(&cp, funcname, sizeof (funcname), &off);
#ifdef HAVE_LIBUNWIND_ELF
unw_get_elf_filename(&cp, objname, sizeof (objname), &objoff);
fprintf(stderr, " [0x%08lx] %s+0x%2lx (in %s +0x%2lx)\n",
ip, funcname, off, objname, objoff);
#else
fprintf(stderr, " [0x%08lx] %s+0x%2lx\n", ip, funcname, off);
#endif
}
}
#elif defined(HAVE_BACKTRACE)
#include <execinfo.h>
static inline void
libspl_dump_backtrace(void)
{
void *btptrs[100];
size_t nptrs = backtrace(btptrs, 100);
char **bt = backtrace_symbols(btptrs, nptrs);
fprintf(stderr, "Call trace:\n");
for (size_t i = 0; i < nptrs; i++)
fprintf(stderr, " %s\n", bt[i]);
free(bt);
}
#else
#define libspl_dump_backtrace()
#endif
static boolean_t libspl_assert_ok = B_FALSE;
@ -33,21 +106,37 @@ libspl_set_assert_ok(boolean_t val)
libspl_assert_ok = val;
}
static pthread_mutex_t assert_lock = PTHREAD_MUTEX_INITIALIZER;
/* printf version of libspl_assert */
void
libspl_assertf(const char *file, const char *func, int line,
const char *format, ...)
{
pthread_mutex_lock(&assert_lock);
va_list args;
char tname[64];
libspl_getthreadname(tname, sizeof (tname));
fprintf(stderr, "ASSERT at %s:%d:%s()\n", file, line, func);
va_start(args, format);
vfprintf(stderr, format, args);
fprintf(stderr, "\n");
fprintf(stderr, "ASSERT at %s:%d:%s()", file, line, func);
va_end(args);
fprintf(stderr, "\n"
" PID: %-8u COMM: %s\n"
" TID: %-8u NAME: %s\n",
getpid(), libspl_getprogname(),
libspl_gettid(), tname);
libspl_dump_backtrace();
#if !__has_feature(attribute_analyzer_noreturn) && !defined(__COVERITY__)
if (libspl_assert_ok) {
pthread_mutex_unlock(&assert_lock);
return;
}
#endif

14
sys/contrib/openzfs/lib/libuutil/uu_list.c
View file

@ -505,14 +505,20 @@ uu_list_walk(uu_list_t *lp, uu_walk_fn_t *func, void *private, uint32_t flags)
}
if (lp->ul_debug || robust) {
uu_list_walk_t my_walk;
uu_list_walk_t *my_walk;
void *e;
list_walk_init(&my_walk, lp, flags);
my_walk = uu_zalloc(sizeof (*my_walk));
if (my_walk == NULL)
return (-1);
list_walk_init(my_walk, lp, flags);
while (status == UU_WALK_NEXT &&
(e = uu_list_walk_next(&my_walk)) != NULL)
(e = uu_list_walk_next(my_walk)) != NULL)
status = (*func)(e, private);
list_walk_fini(&my_walk);
list_walk_fini(my_walk);
uu_free(my_walk);
} else {
if (!reverse) {
for (np = lp->ul_null_node.uln_next;

15
sys/contrib/openzfs/lib/libzfs/libzfs_dataset.c
View file

@ -5565,8 +5565,21 @@ volsize_from_vdevs(zpool_handle_t *zhp, uint64_t nblocks, uint64_t blksize)
/*
* Scale this size down as a ratio of 128k / tsize.
* See theory statement above.
*
* Bitshift is to avoid the case of nblocks * asize < tsize
* producing a size of 0.
*/
volsize = nblocks * asize * SPA_OLD_MAXBLOCKSIZE / tsize;
volsize = (nblocks * asize) / (tsize >> SPA_MINBLOCKSHIFT);
/*
* If we would blow UINT64_MAX with this next multiplication,
* don't.
*/
if (volsize >
(UINT64_MAX / (SPA_OLD_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT)))
volsize = UINT64_MAX;
else
volsize *= (SPA_OLD_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT);
if (volsize > ret) {
ret = volsize;
}

5
sys/contrib/openzfs/lib/libzpool/kernel.c
View file

@ -92,7 +92,8 @@ zk_thread_wrapper(void *arg)
}
kthread_t *
zk_thread_create(void (*func)(void *), void *arg, size_t stksize, int state)
zk_thread_create(const char *name, void (*func)(void *), void *arg,
size_t stksize, int state)
{
pthread_attr_t attr;
pthread_t tid;
@ -140,6 +141,8 @@ zk_thread_create(void (*func)(void *), void *arg, size_t stksize, int state)
VERIFY0(pthread_create(&tid, &attr, zk_thread_wrapper, ztw));
VERIFY0(pthread_attr_destroy(&attr));
pthread_setname_np(tid, name);
return ((void *)(uintptr_t)tid);
}

4
sys/contrib/openzfs/lib/libzpool/taskq.c
View file

@ -295,8 +295,8 @@ taskq_create(const char *name, int nthreads, pri_t pri,
}
for (t = 0; t < nthreads; t++)
VERIFY((tq->tq_threadlist[t] = thread_create(NULL, 0,
taskq_thread, tq, 0, &p0, TS_RUN, pri)) != NULL);
VERIFY((tq->tq_threadlist[t] = thread_create_named(tq->tq_name,
NULL, 0, taskq_thread, tq, 0, &p0, TS_RUN, pri)) != NULL);
return (tq);
}

41
sys/contrib/openzfs/man/man4/zfs.4
View file

@ -16,7 +16,7 @@
.\" own identifying information:
.\" Portions Copyright [yyyy] [name of copyright owner]
.\"
.Dd January 9, 2024
.Dd February 14, 2024
.Dt ZFS 4
.Os
.
@ -525,10 +525,17 @@ most ZPL operations (e.g. write, create) will return
.
.It Sy spa_num_allocators Ns = Ns Sy 4 Pq int
Determines the number of block alloctators to use per spa instance.
Capped by the number of actual CPUs in the system.
Capped by the number of actual CPUs in the system via
.Sy spa_cpus_per_allocator .
.Pp
Note that setting this value too high could result in performance
degredation and/or excess fragmentation.
Set value only applies to pools imported/created after that.
.
.It Sy spa_cpus_per_allocator Ns = Ns Sy 4 Pq int
Determines the minimum number of CPUs in a system for block alloctator
per spa instance.
Set value only applies to pools imported/created after that.
.
.It Sy spa_upgrade_errlog_limit Ns = Ns Sy 0 Pq uint
Limits the number of on-disk error log entries that will be converted to the
@ -564,9 +571,8 @@ However, this is limited by
Maximum micro ZAP size.
A micro ZAP is upgraded to a fat ZAP, once it grows beyond the specified size.
.
.It Sy zfetch_hole_shift Ns = Ns Sy 2 Pq uint
Log2 fraction of holes in speculative prefetch stream allowed for it to
proceed.
.It Sy zap_shrink_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
If set, adjacent empty ZAP blocks will be collapsed, reducing disk space.
.
.It Sy zfetch_min_distance Ns = Ns Sy 4194304 Ns B Po 4 MiB Pc Pq uint
Min bytes to prefetch per stream.
@ -2327,8 +2333,8 @@ Prioritize requeued I/O.
.
.It Sy zio_taskq_batch_pct Ns = Ns Sy 80 Ns % Pq uint
Percentage of online CPUs which will run a worker thread for I/O.
These workers are responsible for I/O work such as compression and
checksum calculations.
These workers are responsible for I/O work such as compression, encryption,
checksum and parity calculations.
Fractional number of CPUs will be rounded down.
.Pp
The default value of
@ -2336,33 +2342,36 @@ The default value of
was chosen to avoid using all CPUs which can result in
latency issues and inconsistent application performance,
especially when slower compression and/or checksumming is enabled.
Set value only applies to pools imported/created after that.
.
.It Sy zio_taskq_batch_tpq Ns = Ns Sy 0 Pq uint
Number of worker threads per taskq.
Lower values improve I/O ordering and CPU utilization,
while higher reduces lock contention.
Higher values improve I/O ordering and CPU utilization,
while lower reduce lock contention.
Set value only applies to pools imported/created after that.
.Pp
If
.Sy 0 ,
generate a system-dependent value close to 6 threads per taskq.
Set value only applies to pools imported/created after that.
.
.It Sy zio_taskq_wr_iss_ncpus Ns = Ns Sy 0 Pq uint
Determines the number of CPUs to run write issue taskqs.
.Pp
When 0 (the default), the value to use is computed internally
as the number of actual CPUs in the system divided by the
.Sy spa_num_allocators
value.
.It Sy zio_taskq_write_tpq Ns = Ns Sy 16 Pq uint
Determines the minumum number of threads per write issue taskq.
Higher values improve CPU utilization on high throughput,
while lower reduce taskq locks contention on high IOPS.
Set value only applies to pools imported/created after that.
.
.It Sy zio_taskq_read Ns = Ns Sy fixed,1,8 null scale null Pq charp
Set the queue and thread configuration for the IO read queues.
This is an advanced debugging parameter.
Don't change this unless you understand what it does.
Set values only apply to pools imported/created after that.
.
.It Sy zio_taskq_write Ns = Ns Sy sync fixed,1,5 scale fixed,1,5 Pq charp
Set the queue and thread configuration for the IO write queues.
This is an advanced debugging parameter.
Don't change this unless you understand what it does.
Set values only apply to pools imported/created after that.
.
.It Sy zvol_inhibit_dev Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Do not create zvol device nodes.

7
sys/contrib/openzfs/man/man8/zpool-clear.8
View file

@ -50,9 +50,10 @@ If the pool was suspended it will be brought back online provided the
devices can be accessed.
Pools with
.Sy multihost
enabled which have been suspended cannot be resumed.
While the pool was suspended, it may have been imported on
another host, and resuming I/O could result in pool damage.
enabled which have been suspended cannot be resumed when there is evidence
that the pool was imported by another host.
The same checks performed during an import will be applied before the clear
proceeds.
.Bl -tag -width Ds
.It Fl -power
Power on the devices's slot in the storage enclosure and wait for the device

2
sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c
View file

@ -1259,7 +1259,7 @@ zvol_os_rename_minor(zvol_state_t *zv, const char *newname)
ASSERT(MUTEX_HELD(&zv->zv_state_lock));
/* Move to a new hashtable entry. */
zv->zv_hash = zvol_name_hash(zv->zv_name);
zv->zv_hash = zvol_name_hash(newname);
hlist_del(&zv->zv_hlink);
hlist_add_head(&zv->zv_hlink, ZVOL_HT_HEAD(zv->zv_hash));

7
sys/contrib/openzfs/module/os/linux/zfs/vdev_disk.c
View file

@ -429,8 +429,11 @@ vdev_disk_open(vdev_t *v, uint64_t *psize, uint64_t *max_psize,
/* Determine the logical block size */
int logical_block_size = bdev_logical_block_size(bdev);
/* Clear the nowritecache bit, causes vdev_reopen() to try again. */
v->vdev_nowritecache = B_FALSE;
/*
* If the device has a write cache, clear the nowritecache flag,
* so that we start issuing flush requests again.
*/
v->vdev_nowritecache = !zfs_bdev_has_write_cache(bdev);
/* Set when device reports it supports TRIM. */
v->vdev_has_trim = bdev_discard_supported(bdev);

2
sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c
View file

@ -1571,7 +1571,7 @@ zvol_os_rename_minor(zvol_state_t *zv, const char *newname)
strlcpy(zv->zv_name, newname, sizeof (zv->zv_name));
/* move to new hashtable entry */
zv->zv_hash = zvol_name_hash(zv->zv_name);
zv->zv_hash = zvol_name_hash(newname);
hlist_del(&zv->zv_hlink);
hlist_add_head(&zv->zv_hlink, ZVOL_HT_HEAD(zv->zv_hash));

8
sys/contrib/openzfs/module/zfs/dmu_objset.c
View file

@ -400,10 +400,10 @@ dnode_hash(const objset_t *os, uint64_t obj)
ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY);
/*
* The low 6 bits of the pointer don't have much entropy, because
* the objset_t is larger than 2^6 bytes long.
* The lower 11 bits of the pointer don't have much entropy, because
* the objset_t is more than 1KB long and so likely aligned to 2KB.
*/
crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (osv >> 6)) & 0xFF];
crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (osv >> 11)) & 0xFF];
crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (obj >> 0)) & 0xFF];
crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (obj >> 8)) & 0xFF];
crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (obj >> 16)) & 0xFF];
@ -1664,12 +1664,14 @@ sync_dnodes_task(void *arg)
sync_objset_arg_t *soa = sda->sda_soa;
objset_t *os = soa->soa_os;
uint_t allocator = spa_acq_allocator(os->os_spa);
multilist_sublist_t *ms =
multilist_sublist_lock_idx(sda->sda_list, sda->sda_sublist_idx);
dmu_objset_sync_dnodes(ms, soa->soa_tx);
multilist_sublist_unlock(ms);
spa_rel_allocator(os->os_spa, allocator);
kmem_free(sda, sizeof (*sda));

5
sys/contrib/openzfs/module/zfs/mmp.c
View file

@ -664,12 +664,13 @@ mmp_thread(void *arg)
(gethrtime() - mmp->mmp_last_write) > mmp_fail_ns) {
zfs_dbgmsg("MMP suspending pool '%s': gethrtime %llu "
"mmp_last_write %llu mmp_interval %llu "
"mmp_fail_intervals %llu mmp_fail_ns %llu",
"mmp_fail_intervals %llu mmp_fail_ns %llu txg %llu",
spa_name(spa), (u_longlong_t)gethrtime(),
(u_longlong_t)mmp->mmp_last_write,
(u_longlong_t)mmp_interval,
(u_longlong_t)mmp_fail_intervals,
(u_longlong_t)mmp_fail_ns);
(u_longlong_t)mmp_fail_ns,
(u_longlong_t)spa->spa_uberblock.ub_txg);
cmn_err(CE_WARN, "MMP writes to pool '%s' have not "
"succeeded in over %llu ms; suspending pool. "
"Hrtime %llu",

191
sys/contrib/openzfs/module/zfs/spa.c
View file

@ -208,7 +208,7 @@ static const uint_t zio_taskq_basedc = 80; /* base duty cycle */
static const boolean_t spa_create_process = B_TRUE; /* no process => no sysdc */
#endif
static uint_t zio_taskq_wr_iss_ncpus = 0;
static uint_t zio_taskq_write_tpq = 16;
/*
* Report any spa_load_verify errors found, but do not fail spa_load.
@ -1067,17 +1067,16 @@ spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
case ZTI_MODE_SYNC:
/*
* Create one wr_iss taskq for every 'zio_taskq_wr_iss_ncpus',
* not to exceed the number of spa allocators.
* Create one wr_iss taskq for every 'zio_taskq_write_tpq' CPUs,
* not to exceed the number of spa allocators, and align to it.
*/
if (zio_taskq_wr_iss_ncpus == 0) {
count = MAX(boot_ncpus / spa->spa_alloc_count, 1);
} else {
count = MAX(1,
boot_ncpus / MAX(1, zio_taskq_wr_iss_ncpus));
}
cpus = MAX(1, boot_ncpus * zio_taskq_batch_pct / 100);
count = MAX(1, cpus / MAX(1, zio_taskq_write_tpq));
count = MAX(count, (zio_taskq_batch_pct + 99) / 100);
count = MIN(count, spa->spa_alloc_count);
while (spa->spa_alloc_count % count != 0 &&
spa->spa_alloc_count < count * 2)
count--;
/*
* zio_taskq_batch_pct is unbounded and may exceed 100%, but no
@ -1495,15 +1494,11 @@ spa_taskq_dispatch_select(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
ASSERT3P(tqs->stqs_taskq, !=, NULL);
ASSERT3U(tqs->stqs_count, !=, 0);
if ((t == ZIO_TYPE_WRITE) && (q == ZIO_TASKQ_ISSUE) &&
(zio != NULL) && (zio->io_wr_iss_tq != NULL)) {
/* dispatch to assigned write issue taskq */
tq = zio->io_wr_iss_tq;
return (tq);
}
if (tqs->stqs_count == 1) {
tq = tqs->stqs_taskq[0];
} else if ((t == ZIO_TYPE_WRITE) && (q == ZIO_TASKQ_ISSUE) &&
(zio != NULL) && ZIO_HAS_ALLOCATOR(zio)) {
tq = tqs->stqs_taskq[zio->io_allocator % tqs->stqs_count];
} else {
tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count];
}
@ -3594,11 +3589,16 @@ spa_activity_check_duration(spa_t *spa, uberblock_t *ub)
}
/*
* Perform the import activity check. If the user canceled the import or
* we detected activity then fail.
* Remote host activity check.
*
* error results:
* 0 - no activity detected
* EREMOTEIO - remote activity detected
* EINTR - user canceled the operation
*/
static int
spa_activity_check(spa_t *spa, uberblock_t *ub, nvlist_t *config)
spa_activity_check(spa_t *spa, uberblock_t *ub, nvlist_t *config,
boolean_t importing)
{
uint64_t txg = ub->ub_txg;
uint64_t timestamp = ub->ub_timestamp;
@ -3643,19 +3643,23 @@ spa_activity_check(spa_t *spa, uberblock_t *ub, nvlist_t *config)
import_expire = gethrtime() + import_delay;
spa_import_progress_set_notes(spa, "Checking MMP activity, waiting "
"%llu ms", (u_longlong_t)NSEC2MSEC(import_delay));
if (importing) {
spa_import_progress_set_notes(spa, "Checking MMP activity, "
"waiting %llu ms", (u_longlong_t)NSEC2MSEC(import_delay));
}
int interations = 0;
int iterations = 0;
while ((now = gethrtime()) < import_expire) {
if (interations++ % 30 == 0) {
if (importing && iterations++ % 30 == 0) {
spa_import_progress_set_notes(spa, "Checking MMP "
"activity, %llu ms remaining",
(u_longlong_t)NSEC2MSEC(import_expire - now));
}
(void) spa_import_progress_set_mmp_check(spa_guid(spa),
NSEC2SEC(import_expire - gethrtime()));
if (importing) {
(void) spa_import_progress_set_mmp_check(spa_guid(spa),
NSEC2SEC(import_expire - gethrtime()));
}
vdev_uberblock_load(rvd, ub, &mmp_label);
@ -3737,6 +3741,61 @@ spa_activity_check(spa_t *spa, uberblock_t *ub, nvlist_t *config)
return (error);
}
/*
* Called from zfs_ioc_clear for a pool that was suspended
* after failing mmp write checks.
*/
boolean_t
spa_mmp_remote_host_activity(spa_t *spa)
{
ASSERT(spa_multihost(spa) && spa_suspended(spa));
nvlist_t *best_label;
uberblock_t best_ub;
/*
* Locate the best uberblock on disk
*/
vdev_uberblock_load(spa->spa_root_vdev, &best_ub, &best_label);
if (best_label) {
/*
* confirm that the best hostid matches our hostid
*/
if (nvlist_exists(best_label, ZPOOL_CONFIG_HOSTID) &&
spa_get_hostid(spa) !=
fnvlist_lookup_uint64(best_label, ZPOOL_CONFIG_HOSTID)) {
nvlist_free(best_label);
return (B_TRUE);
}
nvlist_free(best_label);
} else {
return (B_TRUE);
}
if (!MMP_VALID(&best_ub) ||
!MMP_FAIL_INT_VALID(&best_ub) ||
MMP_FAIL_INT(&best_ub) == 0) {
return (B_TRUE);
}
if (best_ub.ub_txg != spa->spa_uberblock.ub_txg ||
best_ub.ub_timestamp != spa->spa_uberblock.ub_timestamp) {
zfs_dbgmsg("txg mismatch detected during pool clear "
"txg %llu ub_txg %llu timestamp %llu ub_timestamp %llu",
(u_longlong_t)spa->spa_uberblock.ub_txg,
(u_longlong_t)best_ub.ub_txg,
(u_longlong_t)spa->spa_uberblock.ub_timestamp,
(u_longlong_t)best_ub.ub_timestamp);
return (B_TRUE);
}
/*
* Perform an activity check looking for any remote writer
*/
return (spa_activity_check(spa, &spa->spa_uberblock, spa->spa_config,
B_FALSE) != 0);
}
static int
spa_verify_host(spa_t *spa, nvlist_t *mos_config)
{
@ -4063,7 +4122,8 @@ spa_ld_select_uberblock(spa_t *spa, spa_import_type_t type)
return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
}
int error = spa_activity_check(spa, ub, spa->spa_config);
int error =
spa_activity_check(spa, ub, spa->spa_config, B_TRUE);
if (error) {
nvlist_free(label);
return (error);
@ -8771,15 +8831,16 @@ spa_async_remove(spa_t *spa, vdev_t *vd)
}
static void
spa_async_probe(spa_t *spa, vdev_t *vd)
spa_async_fault_vdev(spa_t *spa, vdev_t *vd)
{
if (vd->vdev_probe_wanted) {
vd->vdev_probe_wanted = B_FALSE;
vdev_reopen(vd); /* vdev_open() does the actual probe */
if (vd->vdev_fault_wanted) {
vd->vdev_fault_wanted = B_FALSE;
vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED,
VDEV_AUX_ERR_EXCEEDED);
}
for (int c = 0; c < vd->vdev_children; c++)
spa_async_probe(spa, vd->vdev_child[c]);
spa_async_fault_vdev(spa, vd->vdev_child[c]);
}
static void
@ -8867,11 +8928,11 @@ spa_async_thread(void *arg)
}
/*
* See if any devices need to be probed.
* See if any devices need to be marked faulted.
*/
if (tasks & SPA_ASYNC_PROBE) {
if (tasks & SPA_ASYNC_FAULT_VDEV) {
spa_vdev_state_enter(spa, SCL_NONE);
spa_async_probe(spa, spa->spa_root_vdev);
spa_async_fault_vdev(spa, spa->spa_root_vdev);
(void) spa_vdev_state_exit(spa, NULL, 0);
}
@ -10167,16 +10228,10 @@ spa_sync_tq_create(spa_t *spa, const char *name)
VERIFY(spa->spa_sync_tq != NULL);
VERIFY(kthreads != NULL);
spa_taskqs_t *tqs =
&spa->spa_zio_taskq[ZIO_TYPE_WRITE][ZIO_TASKQ_ISSUE];
spa_syncthread_info_t *ti = spa->spa_syncthreads;
for (int i = 0, w = 0; i < nthreads; i++, w++, ti++) {
for (int i = 0; i < nthreads; i++, ti++) {
ti->sti_thread = kthreads[i];
if (w == tqs->stqs_count) {
w = 0;
}
ti->sti_wr_iss_tq = tqs->stqs_taskq[w];
ti->sti_allocator = i;
}
kmem_free(kthreads, sizeof (*kthreads) * nthreads);
@ -10195,6 +10250,42 @@ spa_sync_tq_destroy(spa_t *spa)
spa->spa_sync_tq = NULL;
}
uint_t
spa_acq_allocator(spa_t *spa)
{
int i;
if (spa->spa_alloc_count == 1)
return (0);
mutex_enter(&spa->spa_allocs_use->sau_lock);
uint_t r = spa->spa_allocs_use->sau_rotor;
do {
if (++r == spa->spa_alloc_count)
r = 0;
} while (spa->spa_allocs_use->sau_inuse[r]);
spa->spa_allocs_use->sau_inuse[r] = B_TRUE;
spa->spa_allocs_use->sau_rotor = r;
mutex_exit(&spa->spa_allocs_use->sau_lock);
spa_syncthread_info_t *ti = spa->spa_syncthreads;
for (i = 0; i < spa->spa_alloc_count; i++, ti++) {
if (ti->sti_thread == curthread) {
ti->sti_allocator = r;
break;
}
}
ASSERT3S(i, <, spa->spa_alloc_count);
return (r);
}
void
spa_rel_allocator(spa_t *spa, uint_t allocator)
{
if (spa->spa_alloc_count > 1)
spa->spa_allocs_use->sau_inuse[allocator] = B_FALSE;
}
void
spa_select_allocator(zio_t *zio)
{
@ -10222,8 +10313,7 @@ spa_select_allocator(zio_t *zio)
spa_syncthread_info_t *ti = spa->spa_syncthreads;
for (int i = 0; i < spa->spa_alloc_count; i++, ti++) {
if (ti->sti_thread == curthread) {
zio->io_allocator = i;
zio->io_wr_iss_tq = ti->sti_wr_iss_tq;
zio->io_allocator = ti->sti_allocator;
return;
}
}
@ -10240,7 +10330,6 @@ spa_select_allocator(zio_t *zio)
bm->zb_blkid >> 20);
zio->io_allocator = (uint_t)hv % spa->spa_alloc_count;
zio->io_wr_iss_tq = NULL;
}
/*
@ -10811,10 +10900,10 @@ ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_data, INT, ZMOD_RW,
ZFS_MODULE_PARAM(zfs_spa, spa_, load_print_vdev_tree, INT, ZMOD_RW,
"Print vdev tree to zfs_dbgmsg during pool import");
ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_pct, UINT, ZMOD_RD,
ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_pct, UINT, ZMOD_RW,
"Percentage of CPUs to run an IO worker thread");
ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_tpq, UINT, ZMOD_RD,
ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_tpq, UINT, ZMOD_RW,
"Number of threads per IO worker taskqueue");
/* BEGIN CSTYLED */
@ -10845,13 +10934,13 @@ ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, new_alloc, INT,
#ifdef _KERNEL
ZFS_MODULE_VIRTUAL_PARAM_CALL(zfs_zio, zio_, taskq_read,
spa_taskq_read_param_set, spa_taskq_read_param_get, ZMOD_RD,
spa_taskq_read_param_set, spa_taskq_read_param_get, ZMOD_RW,
"Configure IO queues for read IO");
ZFS_MODULE_VIRTUAL_PARAM_CALL(zfs_zio, zio_, taskq_write,
spa_taskq_write_param_set, spa_taskq_write_param_get, ZMOD_RD,
spa_taskq_write_param_set, spa_taskq_write_param_get, ZMOD_RW,
"Configure IO queues for write IO");
#endif
/* END CSTYLED */
ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_wr_iss_ncpus, UINT, ZMOD_RW,
"Number of CPUs to run write issue taskqs");
ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_write_tpq, UINT, ZMOD_RW,
"Number of CPUs per write issue taskq");

22
sys/contrib/openzfs/module/zfs/spa_misc.c
View file

@ -394,6 +394,7 @@ static const uint64_t spa_max_slop = 128ULL * 1024 * 1024 * 1024;
* Number of allocators to use, per spa instance
*/
static int spa_num_allocators = 4;
static int spa_cpus_per_allocator = 4;
/*
* Spa active allocator.
@ -747,8 +748,9 @@ spa_add(const char *name, nvlist_t *config, const char *altroot)
if (altroot)
spa->spa_root = spa_strdup(altroot);
/* Do not allow more allocators than CPUs. */
spa->spa_alloc_count = MIN(MAX(spa_num_allocators, 1), boot_ncpus);
/* Do not allow more allocators than fraction of CPUs. */
spa->spa_alloc_count = MAX(MIN(spa_num_allocators,
boot_ncpus / MAX(spa_cpus_per_allocator, 1)), 1);
spa->spa_allocs = kmem_zalloc(spa->spa_alloc_count *
sizeof (spa_alloc_t), KM_SLEEP);
@ -758,6 +760,12 @@ spa_add(const char *name, nvlist_t *config, const char *altroot)
avl_create(&spa->spa_allocs[i].spaa_tree, zio_bookmark_compare,
sizeof (zio_t), offsetof(zio_t, io_queue_node.a));
}
if (spa->spa_alloc_count > 1) {
spa->spa_allocs_use = kmem_zalloc(offsetof(spa_allocs_use_t,
sau_inuse[spa->spa_alloc_count]), KM_SLEEP);
mutex_init(&spa->spa_allocs_use->sau_lock, NULL, MUTEX_DEFAULT,
NULL);
}
avl_create(&spa->spa_metaslabs_by_flushed, metaslab_sort_by_flushed,
sizeof (metaslab_t), offsetof(metaslab_t, ms_spa_txg_node));
@ -853,6 +861,11 @@ spa_remove(spa_t *spa)
}
kmem_free(spa->spa_allocs, spa->spa_alloc_count *
sizeof (spa_alloc_t));
if (spa->spa_alloc_count > 1) {
mutex_destroy(&spa->spa_allocs_use->sau_lock);
kmem_free(spa->spa_allocs_use, offsetof(spa_allocs_use_t,
sau_inuse[spa->spa_alloc_count]));
}
avl_destroy(&spa->spa_metaslabs_by_flushed);
avl_destroy(&spa->spa_sm_logs_by_txg);
@ -3097,4 +3110,7 @@ ZFS_MODULE_PARAM_CALL(zfs_spa, spa_, slop_shift, param_set_slop_shift,
param_get_uint, ZMOD_RW, "Reserved free space in pool");
ZFS_MODULE_PARAM(zfs, spa_, num_allocators, INT, ZMOD_RW,
"Number of allocators per spa, capped by ncpus");
"Number of allocators per spa");
ZFS_MODULE_PARAM(zfs, spa_, cpus_per_allocator, INT, ZMOD_RW,
"Minimum number of CPUs per allocators");

9
sys/contrib/openzfs/module/zfs/txg.c
View file

@ -550,6 +550,15 @@ txg_sync_thread(void *arg)
timer = (delta > timeout ? 0 : timeout - delta);
}
/*
* When we're suspended, nothing should be changing and for
* MMP we don't want to bump anything that would make it
* harder to detect if another host is changing it when
* resuming after a MMP suspend.
*/
if (spa_suspended(spa))
continue;
/*
* Wait until the quiesce thread hands off a txg to us,
* prompting it to do so if necessary.

22
sys/contrib/openzfs/module/zfs/vdev.c
View file

@ -1664,6 +1664,7 @@ vdev_metaslab_fini(vdev_t *vd)
typedef struct vdev_probe_stats {
boolean_t vps_readable;
boolean_t vps_writeable;
boolean_t vps_zio_done_probe;
int vps_flags;
} vdev_probe_stats_t;
@ -1709,6 +1710,17 @@ vdev_probe_done(zio_t *zio)
(void) zfs_ereport_post(FM_EREPORT_ZFS_PROBE_FAILURE,
spa, vd, NULL, NULL, 0);
zio->io_error = SET_ERROR(ENXIO);
/*
* If this probe was initiated from zio pipeline, then
* change the state in a spa_async_request. Probes that
* were initiated from a vdev_open can change the state
* as part of the open call.
*/
if (vps->vps_zio_done_probe) {
vd->vdev_fault_wanted = B_TRUE;
spa_async_request(spa, SPA_ASYNC_FAULT_VDEV);
}
}
mutex_enter(&vd->vdev_probe_lock);
@ -1759,6 +1771,7 @@ vdev_probe(vdev_t *vd, zio_t *zio)
vps->vps_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_PROBE |
ZIO_FLAG_DONT_AGGREGATE | ZIO_FLAG_TRYHARD;
vps->vps_zio_done_probe = (zio != NULL);
if (spa_config_held(spa, SCL_ZIO, RW_WRITER)) {
/*
@ -1785,15 +1798,6 @@ vdev_probe(vdev_t *vd, zio_t *zio)
vd->vdev_probe_zio = pio = zio_null(NULL, spa, vd,
vdev_probe_done, vps,
vps->vps_flags | ZIO_FLAG_DONT_PROPAGATE);
/*
* We can't change the vdev state in this context, so we
* kick off an async task to do it on our behalf.
*/
if (zio != NULL) {
vd->vdev_probe_wanted = B_TRUE;
spa_async_request(spa, SPA_ASYNC_PROBE);
}
}
if (zio != NULL)

4
sys/contrib/openzfs/module/zfs/vdev_label.c
View file

@ -2027,6 +2027,7 @@ vdev_config_sync(vdev_t **svd, int svdcount, uint64_t txg)
/*
* If this isn't a resync due to I/O errors,
* and nothing changed in this transaction group,
* and multihost protection isn't enabled,
* and the vdev configuration hasn't changed,
* then there's nothing to do.
*/
@ -2034,7 +2035,8 @@ vdev_config_sync(vdev_t **svd, int svdcount, uint64_t txg)
boolean_t changed = uberblock_update(ub, spa->spa_root_vdev,
txg, spa->spa_mmp.mmp_delay);
if (!changed && list_is_empty(&spa->spa_config_dirty_list))
if (!changed && list_is_empty(&spa->spa_config_dirty_list) &&
!spa_multihost(spa))
return (0);
}

5
sys/contrib/openzfs/module/zfs/vdev_raidz.c
View file

@ -1891,8 +1891,9 @@ vdev_raidz_matrix_reconstruct(raidz_row_t *rr, int n, int nmissing,
static void
vdev_raidz_reconstruct_general(raidz_row_t *rr, int *tgts, int ntgts)
{
int n, i, c, t, tt;
int nmissing_rows;
int i, c, t, tt;
unsigned int n;
unsigned int nmissing_rows;
int missing_rows[VDEV_RAIDZ_MAXPARITY];
int parity_map[VDEV_RAIDZ_MAXPARITY];
uint8_t *p, *pp;

336
sys/contrib/openzfs/module/zfs/zap.c
View file

@ -22,6 +22,8 @@
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
* Copyright 2023 Alexander Stetsenko <alex.stetsenko@gmail.com>
* Copyright (c) 2023, Klara Inc.
*/
/*
@ -41,6 +43,7 @@
#include <sys/spa.h>
#include <sys/dmu.h>
#include <sys/dnode.h>
#include <sys/zfs_context.h>
#include <sys/zfs_znode.h>
#include <sys/fs/zfs.h>
@ -78,9 +81,16 @@
*/
static int zap_iterate_prefetch = B_TRUE;
/*
* Enable ZAP shrinking. When enabled, empty sibling leaf blocks will be
* collapsed into a single block.
*/
int zap_shrink_enabled = B_TRUE;
int fzap_default_block_shift = 14; /* 16k blocksize */
static uint64_t zap_allocate_blocks(zap_t *zap, int nblocks);
static int zap_shrink(zap_name_t *zn, zap_leaf_t *l, dmu_tx_t *tx);
void
fzap_byteswap(void *vbuf, size_t size)
@ -586,6 +596,72 @@ zap_set_idx_to_blk(zap_t *zap, uint64_t idx, uint64_t blk, dmu_tx_t *tx)
}
}
static int
zap_set_idx_range_to_blk(zap_t *zap, uint64_t idx, uint64_t nptrs, uint64_t blk,
dmu_tx_t *tx)
{
int bs = FZAP_BLOCK_SHIFT(zap);
int epb = bs >> 3; /* entries per block */
int err = 0;
ASSERT(tx != NULL);
ASSERT(RW_WRITE_HELD(&zap->zap_rwlock));
/*
* Check for i/o errors
*/
for (int i = 0; i < nptrs; i += epb) {
uint64_t blk;
err = zap_idx_to_blk(zap, idx + i, &blk);
if (err != 0) {
return (err);
}
}
for (int i = 0; i < nptrs; i++) {
err = zap_set_idx_to_blk(zap, idx + i, blk, tx);
ASSERT0(err); /* we checked for i/o errors above */
if (err != 0)
break;
}
return (err);
}
#define ZAP_PREFIX_HASH(pref, pref_len) ((pref) << (64 - (pref_len)))
/*
* Each leaf has single range of entries (block pointers) in the ZAP ptrtbl.
* If two leaves are siblings, their ranges are adjecent and contain the same
* number of entries. In order to find out if a leaf has a sibling, we need to
* check the range corresponding to the sibling leaf. There is no need to check
* all entries in the range, we only need to check the frist and the last one.
*/
static uint64_t
check_sibling_ptrtbl_range(zap_t *zap, uint64_t prefix, uint64_t prefix_len)
{
ASSERT(RW_LOCK_HELD(&zap->zap_rwlock));
uint64_t h = ZAP_PREFIX_HASH(prefix, prefix_len);
uint64_t idx = ZAP_HASH_IDX(h, zap_f_phys(zap)->zap_ptrtbl.zt_shift);
uint64_t pref_diff = zap_f_phys(zap)->zap_ptrtbl.zt_shift - prefix_len;
uint64_t nptrs = (1 << pref_diff);
uint64_t first;
uint64_t last;
ASSERT3U(idx+nptrs, <=, (1UL << zap_f_phys(zap)->zap_ptrtbl.zt_shift));
if (zap_idx_to_blk(zap, idx, &first) != 0)
return (0);
if (zap_idx_to_blk(zap, idx + nptrs - 1, &last) != 0)
return (0);
if (first != last)
return (0);
return (first);
}
static int
zap_deref_leaf(zap_t *zap, uint64_t h, dmu_tx_t *tx, krw_t lt, zap_leaf_t **lp)
{
@ -958,6 +1034,10 @@ fzap_remove(zap_name_t *zn, dmu_tx_t *tx)
if (err == 0) {
zap_entry_remove(&zeh);
zap_increment_num_entries(zn->zn_zap, -1, tx);
if (zap_leaf_phys(l)->l_hdr.lh_nentries == 0 &&
zap_shrink_enabled)
return (zap_shrink(zn, l, tx));
}
zap_put_leaf(l);
return (err);
@ -1222,13 +1302,19 @@ fzap_cursor_retrieve(zap_t *zap, zap_cursor_t *zc, zap_attribute_t *za)
ZIO_PRIORITY_ASYNC_READ);
}
if (zc->zc_leaf &&
(ZAP_HASH_IDX(zc->zc_hash,
zap_leaf_phys(zc->zc_leaf)->l_hdr.lh_prefix_len) !=
zap_leaf_phys(zc->zc_leaf)->l_hdr.lh_prefix)) {
if (zc->zc_leaf) {
rw_enter(&zc->zc_leaf->l_rwlock, RW_READER);
zap_put_leaf(zc->zc_leaf);
zc->zc_leaf = NULL;
/*
* The leaf was either shrunk or split.
*/
if ((zap_leaf_phys(zc->zc_leaf)->l_hdr.lh_block_type == 0) ||
(ZAP_HASH_IDX(zc->zc_hash,
zap_leaf_phys(zc->zc_leaf)->l_hdr.lh_prefix_len) !=
zap_leaf_phys(zc->zc_leaf)->l_hdr.lh_prefix)) {
zap_put_leaf(zc->zc_leaf);
zc->zc_leaf = NULL;
}
}
again:
@ -1237,8 +1323,6 @@ fzap_cursor_retrieve(zap_t *zap, zap_cursor_t *zc, zap_attribute_t *za)
&zc->zc_leaf);
if (err != 0)
return (err);
} else {
rw_enter(&zc->zc_leaf->l_rwlock, RW_READER);
}
l = zc->zc_leaf;
@ -1367,6 +1451,242 @@ fzap_get_stats(zap_t *zap, zap_stats_t *zs)
}
}
/*
* Find last allocated block and update freeblk.
*/
static void
zap_trunc(zap_t *zap)
{
uint64_t nentries;
uint64_t lastblk;
ASSERT(RW_WRITE_HELD(&zap->zap_rwlock));
if (zap_f_phys(zap)->zap_ptrtbl.zt_blk > 0) {
/* External ptrtbl */
nentries = (1 << zap_f_phys(zap)->zap_ptrtbl.zt_shift);
lastblk = zap_f_phys(zap)->zap_ptrtbl.zt_blk +
zap_f_phys(zap)->zap_ptrtbl.zt_numblks - 1;
} else {
/* Embedded ptrtbl */
nentries = (1 << ZAP_EMBEDDED_PTRTBL_SHIFT(zap));
lastblk = 0;
}
for (uint64_t idx = 0; idx < nentries; idx++) {
uint64_t blk;
if (zap_idx_to_blk(zap, idx, &blk) != 0)
return;
if (blk > lastblk)
lastblk = blk;
}
ASSERT3U(lastblk, <, zap_f_phys(zap)->zap_freeblk);
zap_f_phys(zap)->zap_freeblk = lastblk + 1;
}
/*
* ZAP shrinking algorithm.
*
* We shrink ZAP recuresively removing empty leaves. We can remove an empty leaf
* only if it has a sibling. Sibling leaves have the same prefix length and
* their prefixes differ only by the least significant (sibling) bit. We require
* both siblings to be empty. This eliminates a need to rehash the non-empty
* remaining leaf. When we have removed one of two empty sibling, we set ptrtbl
* entries of the removed leaf to point out to the remaining leaf. Prefix length
* of the remaining leaf is decremented. As a result, it has a new prefix and it
* might have a new sibling. So, we repeat the process.
*
* Steps:
* 1. Check if a sibling leaf (sl) exists and it is empty.
* 2. Release the leaf (l) if it has the sibling bit (slbit) equal to 1.
* 3. Release the sibling (sl) to derefer it again with WRITER lock.
* 4. Upgrade zapdir lock to WRITER (once).
* 5. Derefer released leaves again.
* 6. If it is needed, recheck whether both leaves are still siblings and empty.
* 7. Set ptrtbl pointers of the removed leaf (slbit 1) to point out to blkid of
* the remaining leaf (slbit 0).
* 8. Free disk block of the removed leaf (dmu_free_range).
* 9. Decrement prefix_len of the remaining leaf.
* 10. Repeat the steps.
*/
static int
zap_shrink(zap_name_t *zn, zap_leaf_t *l, dmu_tx_t *tx)
{
zap_t *zap = zn->zn_zap;
int64_t zt_shift = zap_f_phys(zap)->zap_ptrtbl.zt_shift;
uint64_t hash = zn->zn_hash;
uint64_t prefix = zap_leaf_phys(l)->l_hdr.lh_prefix;
uint64_t prefix_len = zap_leaf_phys(l)->l_hdr.lh_prefix_len;
boolean_t trunc = B_FALSE;
int err = 0;
ASSERT3U(zap_leaf_phys(l)->l_hdr.lh_nentries, ==, 0);
ASSERT3U(prefix_len, <=, zap_f_phys(zap)->zap_ptrtbl.zt_shift);
ASSERT(RW_LOCK_HELD(&zap->zap_rwlock));
ASSERT3U(ZAP_HASH_IDX(hash, prefix_len), ==, prefix);
boolean_t writer = B_FALSE;
/*
* To avoid deadlock always deref leaves in the same order -
* sibling 0 first, then sibling 1.
*/
while (prefix_len) {
zap_leaf_t *sl;
int64_t prefix_diff = zt_shift - prefix_len;
uint64_t sl_prefix = prefix ^ 1;
uint64_t sl_hash = ZAP_PREFIX_HASH(sl_prefix, prefix_len);
int slbit = prefix & 1;
ASSERT3U(zap_leaf_phys(l)->l_hdr.lh_nentries, ==, 0);
/*
* Check if there is a sibling by reading ptrtbl ptrs.
*/
if (check_sibling_ptrtbl_range(zap, sl_prefix, prefix_len) == 0)
break;
/*
* sibling 1, unlock it - we haven't yet dereferenced sibling 0.
*/
if (slbit == 1) {
zap_put_leaf(l);
l = NULL;
}
/*
* Dereference sibling leaf and check if it is empty.
*/
if ((err = zap_deref_leaf(zap, sl_hash, tx, RW_READER,
&sl)) != 0)
break;
ASSERT3U(ZAP_HASH_IDX(sl_hash, prefix_len), ==, sl_prefix);
/*
* Check if we have a sibling and it is empty.
*/
if (zap_leaf_phys(sl)->l_hdr.lh_prefix_len != prefix_len ||
zap_leaf_phys(sl)->l_hdr.lh_nentries != 0) {
zap_put_leaf(sl);
break;
}
zap_put_leaf(sl);
/*
* If there two empty sibling, we have work to do, so
* we need to lock ZAP ptrtbl as WRITER.
*/
if (!writer && (writer = zap_tryupgradedir(zap, tx)) == 0) {
/* We failed to upgrade */
if (l != NULL) {
zap_put_leaf(l);
l = NULL;
}
/*
* Usually, the right way to upgrade from a READER lock
* to a WRITER lock is to call zap_unlockdir() and
* zap_lockdir(), but we do not have a tag. Instead,
* we do it in more sophisticated way.
*/
rw_exit(&zap->zap_rwlock);
rw_enter(&zap->zap_rwlock, RW_WRITER);
dmu_buf_will_dirty(zap->zap_dbuf, tx);
zt_shift = zap_f_phys(zap)->zap_ptrtbl.zt_shift;
writer = B_TRUE;
}
/*
* Here we have WRITER lock for ptrtbl.
* Now, we need a WRITER lock for both siblings leaves.
* Also, we have to recheck if the leaves are still siblings
* and still empty.
*/
if (l == NULL) {
/* sibling 0 */
if ((err = zap_deref_leaf(zap, (slbit ? sl_hash : hash),
tx, RW_WRITER, &l)) != 0)
break;
/*
* The leaf isn't empty anymore or
* it was shrunk/split while our locks were down.
*/
if (zap_leaf_phys(l)->l_hdr.lh_nentries != 0 ||
zap_leaf_phys(l)->l_hdr.lh_prefix_len != prefix_len)
break;
}
/* sibling 1 */
if ((err = zap_deref_leaf(zap, (slbit ? hash : sl_hash), tx,
RW_WRITER, &sl)) != 0)
break;
/*
* The leaf isn't empty anymore or
* it was shrunk/split while our locks were down.
*/
if (zap_leaf_phys(sl)->l_hdr.lh_nentries != 0 ||
zap_leaf_phys(sl)->l_hdr.lh_prefix_len != prefix_len) {
zap_put_leaf(sl);
break;
}
/* If we have gotten here, we have a leaf to collapse */
uint64_t idx = (slbit ? prefix : sl_prefix) << prefix_diff;
uint64_t nptrs = (1ULL << prefix_diff);
uint64_t sl_blkid = sl->l_blkid;
/*
* Set ptrtbl entries to point out to the slibling 0 blkid
*/
if ((err = zap_set_idx_range_to_blk(zap, idx, nptrs, l->l_blkid,
tx)) != 0) {
zap_put_leaf(sl);
break;
}
/*
* Free sibling 1 disk block.
*/
int bs = FZAP_BLOCK_SHIFT(zap);
if (sl_blkid == zap_f_phys(zap)->zap_freeblk - 1)
trunc = B_TRUE;
(void) dmu_free_range(zap->zap_objset, zap->zap_object,
sl_blkid << bs, 1 << bs, tx);
zap_put_leaf(sl);
zap_f_phys(zap)->zap_num_leafs--;
/*
* Update prefix and prefix_len.
*/
zap_leaf_phys(l)->l_hdr.lh_prefix >>= 1;
zap_leaf_phys(l)->l_hdr.lh_prefix_len--;
prefix = zap_leaf_phys(l)->l_hdr.lh_prefix;
prefix_len = zap_leaf_phys(l)->l_hdr.lh_prefix_len;
}
if (trunc)
zap_trunc(zap);
if (l != NULL)
zap_put_leaf(l);
return (err);
}
/* CSTYLED */
ZFS_MODULE_PARAM(zfs, , zap_iterate_prefetch, INT, ZMOD_RW,
"When iterating ZAP object, prefetch it");
/* CSTYLED */
ZFS_MODULE_PARAM(zfs, , zap_shrink_enabled, INT, ZMOD_RW,
"Enable ZAP shrinking");

9
sys/contrib/openzfs/module/zfs/zfs_ioctl.c
View file

@ -5823,10 +5823,13 @@ zfs_ioc_clear(zfs_cmd_t *zc)
/*
* If multihost is enabled, resuming I/O is unsafe as another
* host may have imported the pool.
* host may have imported the pool. Check for remote activity.
*/
if (spa_multihost(spa) && spa_suspended(spa))
return (SET_ERROR(EINVAL));
if (spa_multihost(spa) && spa_suspended(spa) &&
spa_mmp_remote_host_activity(spa)) {
spa_close(spa, FTAG);
return (SET_ERROR(EREMOTEIO));
}
spa_vdev_state_enter(spa, SCL_NONE);

17
sys/contrib/openzfs/module/zfs/zio.c
View file

@ -803,9 +803,10 @@ zio_notify_parent(zio_t *pio, zio_t *zio, enum zio_wait_type wait,
/*
* If we can tell the caller to execute this parent next, do
* so. We only do this if the parent's zio type matches the
* child's type. Otherwise dispatch the parent zio in its
* own taskq.
* so. We do this if the parent's zio type matches the child's
* type, or if it's a zio_null() with no done callback, and so
* has no actual work to do. Otherwise dispatch the parent zio
* in its own taskq.
*
* Having the caller execute the parent when possible reduces
* locking on the zio taskq's, reduces context switch
@ -825,7 +826,8 @@ zio_notify_parent(zio_t *pio, zio_t *zio, enum zio_wait_type wait,
* of writes for spa_sync(), and the chain of ZIL blocks.
*/
if (next_to_executep != NULL && *next_to_executep == NULL &&
pio->io_type == zio->io_type) {
(pio->io_type == zio->io_type ||
(pio->io_type == ZIO_TYPE_NULL && !pio->io_done))) {
*next_to_executep = pio;
} else {
zio_taskq_dispatch(pio, type, B_FALSE);
@ -2532,8 +2534,10 @@ zio_suspend(spa_t *spa, zio_t *zio, zio_suspend_reason_t reason)
"failure and the failure mode property for this pool "
"is set to panic.", spa_name(spa));
cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O "
"failure and has been suspended.\n", spa_name(spa));
if (reason != ZIO_SUSPEND_MMP) {
cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable "
"I/O failure and has been suspended.\n", spa_name(spa));
}
(void) zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL,
NULL, NULL, 0);
@ -2921,7 +2925,6 @@ static void
zio_gang_inherit_allocator(zio_t *pio, zio_t *cio)
{
cio->io_allocator = pio->io_allocator;
cio->io_wr_iss_tq = pio->io_wr_iss_tq;
}
static void

6
sys/contrib/openzfs/module/zfs/zio_inject.c
View file

@ -607,9 +607,11 @@ zio_handle_io_delay(zio_t *zio)
if (vd->vdev_guid != handler->zi_record.zi_guid)
continue;
/* also match on I/O type (e.g., -T read) */
if (handler->zi_record.zi_iotype != ZIO_TYPES &&
handler->zi_record.zi_iotype != zio->io_type)
continue;
handler->zi_record.zi_iotype != zio->io_type) {
continue;
}
/*
* Defensive; should never happen as the array allocation

4
sys/contrib/openzfs/tests/runfiles/common.run
View file

@ -643,6 +643,10 @@ tags = ['functional', 'compression']
tests = ['cp_files_001_pos', 'cp_files_002_pos', 'cp_stress']
tags = ['functional', 'cp_files']
[tests/functional/zap_shrink]
tests = ['zap_shrink_001_pos']
tags = ['functional', 'zap_shrink']
[tests/functional/crtime]
tests = ['crtime_001_pos' ]
tags = ['functional', 'crtime']

2
sys/contrib/openzfs/tests/runfiles/linux.run
View file

@ -146,7 +146,7 @@ tags = ['functional', 'mmap']
tests = ['mmp_on_thread', 'mmp_on_uberblocks', 'mmp_on_off', 'mmp_interval',
'mmp_active_import', 'mmp_inactive_import', 'mmp_exported_import',
'mmp_write_uberblocks', 'mmp_reset_interval', 'multihost_history',
'mmp_on_zdb', 'mmp_write_distribution', 'mmp_hostid']
'mmp_on_zdb', 'mmp_write_distribution', 'mmp_hostid', 'mmp_write_slow_disk']
tags = ['functional', 'mmp']
[tests/functional/mount:Linux]

4
sys/contrib/openzfs/tests/zfs-tests/tests/Makefile.am
View file

@ -1593,6 +1593,7 @@ nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \
functional/mmp/mmp_on_zdb.ksh \
functional/mmp/mmp_reset_interval.ksh \
functional/mmp/mmp_write_distribution.ksh \
functional/mmp/mmp_write_slow_disk.ksh \
functional/mmp/mmp_write_uberblocks.ksh \
functional/mmp/multihost_history.ksh \
functional/mmp/setup.ksh \
@ -2074,6 +2075,9 @@ nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \
functional/xattr/xattr_012_pos.ksh \
functional/xattr/xattr_013_pos.ksh \
functional/xattr/xattr_compat.ksh \
functional/zap_shrink/cleanup.ksh \
functional/zap_shrink/zap_shrink_001_pos.ksh \
functional/zap_shrink/setup.ksh \
functional/zpool_influxdb/cleanup.ksh \
functional/zpool_influxdb/setup.ksh \
functional/zpool_influxdb/zpool_influxdb.ksh \

97
sys/contrib/openzfs/tests/zfs-tests/tests/functional/mmp/mmp_write_slow_disk.ksh Executable file
View file

@ -0,0 +1,97 @@
#!/bin/ksh -p
#
# CDDL HEADER START
#
# This file and its contents are supplied under the terms of the
# Common Development and Distribution License ("CDDL"), version 1.0.
# You may only use this file in accordance with the terms of version
# 1.0 of the CDDL.
#
# A full copy of the text of the CDDL should have accompanied this
# source. A copy of the CDDL is also available via the Internet at
# http://www.illumos.org/license/CDDL.
#
# CDDL HEADER END
#
#
# Copyright (c) 2024, Klara Inc
#
# DESCRIPTION:
# Verify that long VDEV probes do not cause MMP checks to suspend pool
# Note: without PR-15839 fix, this test will suspend the pool.
#
# A device that is returning unexpected errors will trigger a vdev_probe.
# When the device additionally has slow response times, the probe can hold
# the spa config lock as a writer for a long period of time such that the
# mmp uberblock updates stall when trying to acquire the spa config lock.
#
# STRATEGY:
# 1. Create a pool with multiple leaf vdevs
# 2. Enable multihost and multihost_history
# 3. Delay for MMP writes to occur
# 4. Verify that a long VDEV probe didn't cause MMP check to suspend pool
#
. $STF_SUITE/include/libtest.shlib
. $STF_SUITE/tests/functional/mmp/mmp.cfg
. $STF_SUITE/tests/functional/mmp/mmp.kshlib
verify_runnable "both"
function cleanup
{
log_must zinject -c all
if [[ $(zpool list -H -o health $MMP_POOL) == "SUSPENDED" ]]; then
log_must zpool clear $MMP_POOL
zpool get state $MMP_POOL $MMP_DIR/file.3
zpool events | grep ".fs.zfs." | grep -v "history_event"
fi
poolexists $MMP_POOL && destroy_pool $MMP_POOL
log_must rm -r $MMP_DIR
log_must mmp_clear_hostid
}
log_assert "A long VDEV probe doesn't cause a MMP check suspend"
log_onexit cleanup
MMP_HISTORY_URL=/proc/spl/kstat/zfs/$MMP_POOL/multihost
# Create a multiple drive pool
log_must zpool events -c
log_must mkdir -p $MMP_DIR
log_must truncate -s 128M $MMP_DIR/file.{0,1,2,3,4,5}
log_must zpool create -f $MMP_POOL \
mirror $MMP_DIR/file.{0,1,2} \
mirror $MMP_DIR/file.{3,4,5}
# Enable MMP
log_must mmp_set_hostid $HOSTID1
log_must zpool set multihost=on $MMP_POOL
clear_mmp_history
# Inject vdev write error along with a delay
log_must zinject -f 33 -e io -L pad2 -T write -d $MMP_DIR/file.3 $MMP_POOL
log_must zinject -f 50 -e io -L uber -T write -d $MMP_DIR/file.3 $MMP_POOL
log_must zinject -D 2000:4 -T write -d $MMP_DIR/file.3 $MMP_POOL
log_must dd if=/dev/urandom of=/$MMP_POOL/data bs=1M count=5
sleep 10
sync_pool $MMP_POOL
# Confirm mmp writes to the non-slow disks have taken place
for x in {0,1,2,4}; do
write_count=$(grep -c file.${x} $MMP_HISTORY_URL)
[[ $write_count -gt 0 ]] || log_fail "expecting mmp writes"
done
# Expect that the pool was not suspended
log_must check_state $MMP_POOL "" "ONLINE"
health=$(zpool list -H -o health $MMP_POOL)
log_note "$MMP_POOL health is $health"
[[ "$health" == "SUSPENDED" ]] && log_fail "$MMP_POOL $health unexpected"
log_pass "A long VDEV probe doesn't cause a MMP check suspend"

34
sys/contrib/openzfs/tests/zfs-tests/tests/functional/zap_shrink/cleanup.ksh Executable file
View file

@ -0,0 +1,34 @@
#!/bin/ksh -p
#
# 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 https://opensource.org/licenses/CDDL-1.0.
# 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 2007 Sun Microsystems, Inc. All rights reserved.
# Use is subject to license terms.
#
#
# Copyright (c) 2013 by Delphix. All rights reserved.
#
. $STF_SUITE/include/libtest.shlib
default_cleanup

35
sys/contrib/openzfs/tests/zfs-tests/tests/functional/zap_shrink/setup.ksh Executable file
View file

@ -0,0 +1,35 @@
#!/bin/ksh -p
#
# 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 https://opensource.org/licenses/CDDL-1.0.
# 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 2007 Sun Microsystems, Inc. All rights reserved.
# Use is subject to license terms.
#
#
# Copyright (c) 2013 by Delphix. All rights reserved.
#
. $STF_SUITE/include/libtest.shlib
DISK=${DISKS%% *}
default_setup $DISK

81
sys/contrib/openzfs/tests/zfs-tests/tests/functional/zap_shrink/zap_shrink_001_pos.ksh Executable file
View file

@ -0,0 +1,81 @@
#! /bin/ksh -p
#
# 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 https://opensource.org/licenses/CDDL-1.0.
# 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 2024, Klara Inc.
#
. $STF_SUITE/include/libtest.shlib
#
# DESCRIPTION:
# Create a large number of files in a directory. Then remove all files and
# check that the directory zap was shrunk. Use zdb to check that the zap object
# contains only one leaf block using zdb.
#
verify_runnable "global"
DIR=largedir
NR_FILES=100000
BATCH=1000
CWD=$PWD
log_assert "Create a large number of files ($NR_FILES) in a directory. " \
"Make sure that the directory ZAP object was shrunk."
log_must mkdir $TESTDIR/$DIR
cd $TESTDIR/$DIR
# In order to prevent arguments overflowing, create NR_FILES in BATCH at once.
for i in $(seq $(($NR_FILES/$BATCH))); do
touch $(seq $((($i-1)*$BATCH+1)) $(($i*$BATCH)));
done
cd $CWD
log_must test $NR_FILES -eq $(ls -U $TESTDIR/$DIR | wc -l)
# remove all files in $DIR directory
cd $TESTDIR/$DIR
for i in $(seq $(($NR_FILES/$BATCH))); do
rm $(seq $((($i-1)*$BATCH+1)) $(($i*$BATCH)))
done
cd $CWD
sync_pool $TESTPOOL
log_must test 0 -eq $(ls -U $TESTDIR/$DIR | wc -l)
# check whether zap_shrink works
zapobj=$(zdb -v -O $TESTPOOL/$TESTFS $DIR)
nleafs=$(echo "$zapobj" | grep "Leaf blocks:" | awk -F\: '{print($2);}')
log_must test 1 -eq $nleafs
log_must zpool export $TESTPOOL
log_must zpool import $TESTPOOL
# check whether zap_shrink works
zapobj=$(zdb -v -O $TESTPOOL/$TESTFS $DIR)
nleafs=$(echo "$zapobj" | grep "Leaf blocks:" | awk -F\: '{print($2);}')
log_must test 1 -eq $nleafs
log_pass

13
sys/modules/zfs/zfs_config.h
View file

@ -418,6 +418,9 @@
/* Define if the GNU gettext() function is already present or preinstalled. */
/* #undef HAVE_GETTEXT */
/* Define to 1 if you have the 'gettid' function. */
/* #undef HAVE_GETTID */
/* iops->get_acl() exists */
/* #undef HAVE_GET_ACL */
@ -655,6 +658,12 @@
/* Define if you have [udev] */
/* #undef HAVE_LIBUDEV */
/* Define if you have [unwind] */
/* #undef HAVE_LIBUNWIND */
/* libunwind has unw_get_elf_filename */
/* #undef HAVE_LIBUNWIND_ELF */
/* Define if you have [uuid] */
/* #undef HAVE_LIBUUID */
@ -1179,7 +1188,7 @@
/* #undef ZFS_IS_GPL_COMPATIBLE */
/* Define the project alias string. */
#define ZFS_META_ALIAS "zfs-2.2.99-456-FreeBSD_g1f940de07"
#define ZFS_META_ALIAS "zfs-2.2.99-474-FreeBSD_g8f1b7a6fa"
/* Define the project author. */
#define ZFS_META_AUTHOR "OpenZFS"
@ -1209,7 +1218,7 @@
#define ZFS_META_NAME "zfs"
/* Define the project release. */
#define ZFS_META_RELEASE "456-FreeBSD_g1f940de07"
#define ZFS_META_RELEASE "474-FreeBSD_g8f1b7a6fa"
/* Define the project version. */
#define ZFS_META_VERSION "2.2.99"

2
sys/modules/zfs/zfs_gitrev.h
View file

@ -1 +1 @@
#define ZFS_META_GITREV "zfs-2.2.99-456-g1f940de07"
#define ZFS_META_GITREV "zfs-2.2.99-474-g8f1b7a6fa"