Tag zfs-2.2.4

META file and changelog updated. Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Fix updating the zvol_htable when renaming a zvol
2026-05-24 03:08:51 +03:00 · 2024-04-30 10:01:15 -07:00 · 2024-04-30 10:01:15 -07:00 · 2024-04-30 10:01:15 -07:00 · 2024-04-30 10:01:15 -07:00 · 2024-04-29 17:45:53 -07:00
155 changed files with 4783 additions and 1643 deletions
@@ -30,6 +30,7 @@ Andreas Dilger <adilger@dilger.ca>
 Andrew Walker <awalker@ixsystems.com>
 Benedikt Neuffer <github@itfriend.de>
 Chengfei Zhu <chengfeix.zhu@intel.com>
+ChenHao Lu <18302010006@fudan.edu.cn>
 Chris Lindee <chris.lindee+github@gmail.com>
 Colm Buckley <colm@tuatha.org>
 Crag Wang <crag0715@gmail.com>
@@ -43,6 +44,7 @@ Glenn Washburn <development@efficientek.com>
 Gordan Bobic <gordan.bobic@gmail.com>
 Gregory Bartholomew <gregory.lee.bartholomew@gmail.com>
 hedong zhang <h_d_zhang@163.com>
+Ilkka Sovanto <github@ilkka.kapsi.fi>
 InsanePrawn <Insane.Prawny@gmail.com>
 Jason Cohen <jwittlincohen@gmail.com>
 Jason Harmening <jason.harmening@gmail.com>
@@ -57,6 +59,7 @@ KernelOfTruth <kerneloftruth@gmail.com>
 Liu Hua <liu.hua130@zte.com.cn>
 Liu Qing <winglq@gmail.com>
 loli10K <ezomori.nozomu@gmail.com>
+Mart Frauenlob <allkind@fastest.cc>
 Matthias Blankertz <matthias@blankertz.org>
 Michael Gmelin <grembo@FreeBSD.org>
 Olivier Mazouffre <olivier.mazouffre@ims-bordeaux.fr>
@@ -73,6 +76,9 @@ WHR <msl0000023508@gmail.com>
 Yanping Gao <yanping.gao@xtaotech.com>
 Youzhong Yang <youzhong@gmail.com>

+# Signed-off-by: overriding Author:
+Yuxin Wang <yuxinwang9999@gmail.com> <Bi11gates9999@gmail.com>
+
 # Commits from strange places, long ago
 Brian Behlendorf <behlendorf1@llnl.gov> <behlendo@7e1ea52c-4ff2-0310-8f11-9dd32ca42a1c>
 Brian Behlendorf <behlendorf1@llnl.gov> <behlendo@fedora-17-amd64.(none)>
@@ -102,12 +108,15 @@ Brandon Thetford <brandon@dodecatec.com> <dodexahedron@users.noreply.github.com>
 buzzingwires <buzzingwires@outlook.com> <131118055+buzzingwires@users.noreply.github.com>
 Cedric Maunoury <cedric.maunoury@gmail.com> <38213715+cedricmaunoury@users.noreply.github.com>
 Charles Suh <charles.suh@gmail.com> <charlessuh@users.noreply.github.com>
+Chris Peredun <chris.peredun@ixsystems.com> <126915832+chrisperedun@users.noreply.github.com>
 Dacian Reece-Stremtan <dacianstremtan@gmail.com> <35844628+dacianstremtan@users.noreply.github.com>
 Damian Szuberski <szuberskidamian@gmail.com> <30863496+szubersk@users.noreply.github.com>
 Daniel Hiepler <d-git@coderdu.de> <32984777+heeplr@users.noreply.github.com>
 Daniel Kobras <d.kobras@science-computing.de> <sckobras@users.noreply.github.com>
 Daniel Reichelt <hacking@nachtgeist.net> <nachtgeist@users.noreply.github.com>
 David Quigley <david.quigley@intel.com> <dpquigl@users.noreply.github.com>
+Dennis R. Friedrichsen <dennis.r.friedrichsen@gmail.com> <31087738+dennisfriedrichsen@users.noreply.github.com>
+Dex Wood <slash2314@gmail.com> <slash2314@users.noreply.github.com>
 DHE <git@dehacked.net> <DeHackEd@users.noreply.github.com>
 Dmitri John Ledkov <dimitri.ledkov@canonical.com> <19779+xnox@users.noreply.github.com>
 Dries Michiels <driesm.michiels@gmail.com> <32487486+driesmp@users.noreply.github.com>
@@ -128,6 +137,7 @@ Harry Mallon <hjmallon@gmail.com> <1816667+hjmallon@users.noreply.github.com>
 Hiếu Lê <leorize+oss@disroot.org> <alaviss@users.noreply.github.com>
 Jake Howard <git@theorangeone.net> <RealOrangeOne@users.noreply.github.com>
 James Cowgill <james.cowgill@mips.com> <jcowgill@users.noreply.github.com>
+Jaron Kent-Dobias <jaron@kent-dobias.com> <kentdobias@users.noreply.github.com>
 Jason King <jason.king@joyent.com> <jasonbking@users.noreply.github.com>
 Jeff Dike <jdike@akamai.com> <52420226+jdike@users.noreply.github.com>
 Jitendra Patidar <jitendra.patidar@nutanix.com> <53164267+jsai20@users.noreply.github.com>
@@ -137,7 +147,9 @@ John L. Hammond <john.hammond@intel.com> <35266395+jhammond-intel@users.noreply.
 John-Mark Gurney <jmg@funkthat.com> <jmgurney@users.noreply.github.com>
 John Ramsden <johnramsden@riseup.net> <johnramsden@users.noreply.github.com>
 Jonathon Fernyhough <jonathon@m2x.dev> <559369+jonathonf@users.noreply.github.com>
+Jose Luis Duran <jlduran@gmail.com> <jlduran@users.noreply.github.com>
 Justin Hibbits <chmeeedalf@gmail.com> <chmeeedalf@users.noreply.github.com>
+Kevin Greene <kevin.greene@delphix.com> <104801862+kxgreene@users.noreply.github.com>
 Kevin Jin <lostking2008@hotmail.com> <33590050+jxdking@users.noreply.github.com>
 Kevin P. Fleming <kevin@km6g.us> <kpfleming@users.noreply.github.com>
 Krzysztof Piecuch <piecuch@kpiecuch.pl> <3964215+pikrzysztof@users.noreply.github.com>
@@ -148,9 +160,11 @@ Lorenz Hüdepohl <dev@stellardeath.org> <lhuedepohl@users.noreply.github.com>
 Luís Henriques <henrix@camandro.org> <73643340+lumigch@users.noreply.github.com>
 Marcin Skarbek <git@skarbek.name> <mskarbek@users.noreply.github.com>
 Matt Fiddaman <github@m.fiddaman.uk> <81489167+matt-fidd@users.noreply.github.com>
+Maxim Filimonov <che@bein.link> <part1zano@users.noreply.github.com>
 Max Zettlmeißl <max@zettlmeissl.de> <6818198+maxz@users.noreply.github.com>
 Michael Niewöhner <foss@mniewoehner.de> <c0d3z3r0@users.noreply.github.com>
 Michael Zhivich <mzhivich@akamai.com> <33133421+mzhivich@users.noreply.github.com>
+MigeljanImeri <ImeriMigel@gmail.com> <78048439+MigeljanImeri@users.noreply.github.com>
 Mo Zhou <cdluminate@gmail.com> <5723047+cdluminate@users.noreply.github.com>
 Nick Mattis <nickm970@gmail.com> <nmattis@users.noreply.github.com>
 omni <omni+vagant@hack.org> <79493359+omnivagant@users.noreply.github.com>
@@ -164,6 +178,7 @@ Ping Huang <huangping@smartx.com> <101400146+hpingfs@users.noreply.github.com>
 Piotr P. Stefaniak <pstef@freebsd.org> <pstef@users.noreply.github.com>
 Richard Allen <belperite@gmail.com> <33836503+belperite@users.noreply.github.com>
 Rich Ercolani <rincebrain@gmail.com> <214141+rincebrain@users.noreply.github.com>
+Rick Macklem <rmacklem@uoguelph.ca> <64620010+rmacklem@users.noreply.github.com>
 Rob Wing <rob.wing@klarasystems.com> <98866084+rob-wing@users.noreply.github.com>
 Roman Strashkin <roman.strashkin@nexenta.com> <Ramzec@users.noreply.github.com>
 Ryan Hirasaki <ryanhirasaki@gmail.com> <4690732+RyanHir@users.noreply.github.com>
@@ -174,6 +189,8 @@ Scott Colby <scott@scolby.com> <scolby33@users.noreply.github.com>
 Sean Eric Fagan <kithrup@mac.com> <kithrup@users.noreply.github.com>
 Spencer Kinny <spencerkinny1995@gmail.com> <30333052+Spencer-Kinny@users.noreply.github.com>
 Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com> <75025422+nssrikanth@users.noreply.github.com>
+Stefan Lendl <s.lendl@proxmox.com> <1321542+stfl@users.noreply.github.com>
+Thomas Bertschinger <bertschinger@lanl.gov> <101425190+bertschinger@users.noreply.github.com>
 Thomas Geppert <geppi@digitx.de> <geppi@users.noreply.github.com>
 Tim Crawford <tcrawford@datto.com> <crawfxrd@users.noreply.github.com>
 Tom Matthews <tom@axiom-partners.com> <tomtastic@users.noreply.github.com>
@@ -181,6 +198,7 @@ Tony Perkins <tperkins@datto.com> <62951051+tony-zfs@users.noreply.github.com>
 Torsten Wörtwein <twoertwein@gmail.com> <twoertwein@users.noreply.github.com>
 Tulsi Jain <tulsi.jain@delphix.com> <TulsiJain@users.noreply.github.com>
 Václav Skála <skala@vshosting.cz> <33496485+vaclavskala@users.noreply.github.com>
+Vaibhav Bhanawat <vaibhav.bhanawat@delphix.com> <88050553+vaibhav-delphix@users.noreply.github.com>
 Violet Purcell <vimproved@inventati.org> <66446404+vimproved@users.noreply.github.com>
 Vipin Kumar Verma <vipin.verma@hpe.com> <75025470+vermavipinkumar@users.noreply.github.com>
 Wolfgang Bumiller <w.bumiller@proxmox.com> <Blub@users.noreply.github.com>
@@ -88,9 +88,11 @@ CONTRIBUTORS:
    Bassu <bassu@phi9.com>
    Ben Allen <bsallen@alcf.anl.gov>
    Ben Cordero <bencord0@condi.me>
+    Benda Xu <orv@debian.org>
    Benedikt Neuffer <github@itfriend.de>
    Benjamin Albrecht <git@albrecht.io>
    Benjamin Gentil <benjgentil.pro@gmail.com>
+    Benjamin Sherman <benjamin@holyarmy.org>
    Ben McGough <bmcgough@fredhutch.org>
    Ben Rubson <ben.rubson@gmail.com>
    Ben Wolsieffer <benwolsieffer@gmail.com>
@@ -111,6 +113,7 @@ CONTRIBUTORS:
    bzzz77 <bzzz.tomas@gmail.com>
    cable2999 <cable2999@users.noreply.github.com>
    Caleb James DeLisle <calebdelisle@lavabit.com>
+    Cameron Harr <harr1@llnl.gov>
    Cao Xuewen <cao.xuewen@zte.com.cn>
    Carlo Landmeter <clandmeter@gmail.com>
    Carlos Alberto Lopez Perez <clopez@igalia.com>
@@ -120,12 +123,15 @@ CONTRIBUTORS:
    Chen Can <chen.can2@zte.com.cn>
    Chengfei Zhu <chengfeix.zhu@intel.com>
    Chen Haiquan <oc@yunify.com>
+    ChenHao Lu <18302010006@fudan.edu.cn>
    Chip Parker <aparker@enthought.com>
    Chris Burroughs <chris.burroughs@gmail.com>
+    Chris Davidson <christopher.davidson@gmail.com>
    Chris Dunlap <cdunlap@llnl.gov>
    Chris Dunlop <chris@onthe.net.au>
    Chris Lindee <chris.lindee+github@gmail.com>
    Chris McDonough <chrism@plope.com>
+    Chris Peredun <chris.peredun@ixsystems.com>
    Chris Siden <chris.siden@delphix.com>
    Chris Siebenmann <cks.github@cs.toronto.edu>
    Christer Ekholm <che@chrekh.se>
@@ -144,6 +150,7 @@ CONTRIBUTORS:
    Clint Armstrong <clint@clintarmstrong.net>
    Coleman Kane <ckane@colemankane.org>
    Colin Ian King <colin.king@canonical.com>
+    Colin Percival <cperciva@tarsnap.com>
    Colm Buckley <colm@tuatha.org>
    Crag Wang <crag0715@gmail.com>
    Craig Loomis <cloomis@astro.princeton.edu>
@@ -156,6 +163,7 @@ CONTRIBUTORS:
    Damiano Albani <damiano.albani@gmail.com>
    Damian Szuberski <szuberskidamian@gmail.com>
    Damian Wojsław <damian@wojslaw.pl>
+    Daniel Berlin <dberlin@dberlin.org>
    Daniel Hiepler <d-git@coderdu.de>
    Daniel Hoffman <dj.hoffman@delphix.com>
    Daniel Kobras <d.kobras@science-computing.de>
@@ -176,8 +184,10 @@ CONTRIBUTORS:
    David Quigley <david.quigley@intel.com>
    Debabrata Banerjee <dbanerje@akamai.com>
    D. Ebdrup <debdrup@freebsd.org>
+    Dennis R. Friedrichsen <dennis.r.friedrichsen@gmail.com>
    Denys Rtveliashvili <denys@rtveliashvili.name>
    Derek Dai <daiderek@gmail.com>
+    Dex Wood <slash2314@gmail.com>
    DHE <git@dehacked.net>
    Didier Roche <didrocks@ubuntu.com>
    Dimitri John Ledkov <xnox@ubuntu.com>
@@ -235,9 +245,11 @@ CONTRIBUTORS:
    Gionatan Danti <g.danti@assyoma.it>
    Giuseppe Di Natale <guss80@gmail.com>
    Glenn Washburn <development@efficientek.com>
+    gofaster <felix.gofaster@gmail.com>
    Gordan Bobic <gordan@redsleeve.org>
    Gordon Bergling <gbergling@googlemail.com>
    Gordon Ross <gwr@nexenta.com>
+    Gordon Tetlow <gordon@freebsd.org>
    Graham Christensen <graham@grahamc.com>
    Graham Perrin <grahamperrin@gmail.com>
    Gregor Kopka <gregor@kopka.net>
@@ -265,6 +277,7 @@ CONTRIBUTORS:
    Igor Kozhukhov <ikozhukhov@gmail.com>
    Igor Lvovsky <ilvovsky@gmail.com>
    ilbsmart <wgqimut@gmail.com>
+    Ilkka Sovanto <github@ilkka.kapsi.fi>
    illiliti <illiliti@protonmail.com>
    ilovezfs <ilovezfs@icloud.com>
    InsanePrawn <Insane.Prawny@gmail.com>
@@ -280,9 +293,11 @@ CONTRIBUTORS:
    Jan Engelhardt <jengelh@inai.de>
    Jan Kryl <jan.kryl@nexenta.com>
    Jan Sanislo <oystr@cs.washington.edu>
+    Jaron Kent-Dobias <jaron@kent-dobias.com>
    Jason Cohen <jwittlincohen@gmail.com>
    Jason Harmening <jason.harmening@gmail.com>
    Jason King <jason.brian.king@gmail.com>
+    Jason Lee <jasonlee@lanl.gov>
    Jason Zaman <jasonzaman@gmail.com>
    Javen Wu <wu.javen@gmail.com>
    Jean-Baptiste Lallement <jean-baptiste@ubuntu.com>
@@ -313,6 +328,7 @@ CONTRIBUTORS:
    Jonathon Fernyhough <jonathon@m2x.dev>
    Jorgen Lundman <lundman@lundman.net>
    Josef 'Jeff' Sipek <josef.sipek@nexenta.com>
+    Jose Luis Duran <jlduran@gmail.com>
    Josh Soref <jsoref@users.noreply.github.com>
    Joshua M. Clulow <josh@sysmgr.org>
    José Luis Salvador Rufo <salvador.joseluis@gmail.com>
@@ -336,8 +352,10 @@ CONTRIBUTORS:
    Kash Pande <kash@tripleback.net>
    Kay Pedersen <christianpe96@gmail.com>
    Keith M Wesolowski <wesolows@foobazco.org>
+    Kent Ross <k@mad.cash>
    KernelOfTruth <kerneloftruth@gmail.com>
    Kevin Bowling <kevin.bowling@kev009.com>
+    Kevin Greene <kevin.greene@delphix.com>
    Kevin Jin <lostking2008@hotmail.com>
    Kevin P. Fleming <kevin@km6g.us>
    Kevin Tanguy <kevin.tanguy@ovh.net>
@@ -389,6 +407,7 @@ CONTRIBUTORS:
    Mark Shellenbaum <Mark.Shellenbaum@Oracle.COM>
    marku89 <mar42@kola.li>
    Mark Wright <markwright@internode.on.net>
+    Mart Frauenlob <allkind@fastest.cc>
    Martin Matuska <mm@FreeBSD.org>
    Martin Rüegg <martin.rueegg@metaworx.ch>
    Massimo Maggi <me@massimo-maggi.eu>
@@ -405,6 +424,7 @@ CONTRIBUTORS:
    Matus Kral <matuskral@me.com>
    Mauricio Faria de Oliveira <mfo@canonical.com>
    Max Grossman <max.grossman@delphix.com>
+    Maxim Filimonov <che@bein.link>
    Maximilian Mehnert <maximilian.mehnert@gmx.de>
    Max Zettlmeißl <max@zettlmeissl.de>
    Md Islam <mdnahian@outlook.com>
@@ -417,6 +437,7 @@ CONTRIBUTORS:
    Michael Niewöhner <foss@mniewoehner.de>
    Michael Zhivich <mzhivich@akamai.com>
    Michal Vasilek <michal@vasilek.cz>
+    MigeljanImeri <ImeriMigel@gmail.com>
    Mike Gerdts <mike.gerdts@joyent.com>
    Mike Harsch <mike@harschsystems.com>
    Mike Leddy <mike.leddy@gmail.com>
@@ -448,6 +469,7 @@ CONTRIBUTORS:
    Olaf Faaland <faaland1@llnl.gov>
    Oleg Drokin <green@linuxhacker.ru>
    Oleg Stepura <oleg@stepura.com>
+    Olivier Certner <olce.freebsd@certner.fr>
    Olivier Mazouffre <olivier.mazouffre@ims-bordeaux.fr>
    omni <omni+vagant@hack.org>
    Orivej Desh <orivej@gmx.fr>
@@ -479,6 +501,7 @@ CONTRIBUTORS:
    Prasad Joshi <prasadjoshi124@gmail.com>
    privb0x23 <privb0x23@users.noreply.github.com>
    P.SCH <p88@yahoo.com>
+    Quartz <yyhran@163.com>
    Quentin Zdanis <zdanisq@gmail.com>
    Rafael Kitover <rkitover@gmail.com>
    RageLtMan <sempervictus@users.noreply.github.com>
@@ -491,11 +514,15 @@ CONTRIBUTORS:
    Riccardo Schirone <rschirone91@gmail.com>
    Richard Allen <belperite@gmail.com>
    Richard Elling <Richard.Elling@RichardElling.com>
+    Richard Kojedzinszky <richard@kojedz.in>
    Richard Laager <rlaager@wiktel.com>
    Richard Lowe <richlowe@richlowe.net>
    Richard Sharpe <rsharpe@samba.org>
    Richard Yao <ryao@gentoo.org>
    Rich Ercolani <rincebrain@gmail.com>
+    Rick Macklem <rmacklem@uoguelph.ca>
+    rilysh <nightquick@proton.me>
+    Robert Evans <evansr@google.com>
    Robert Novak <sailnfool@gmail.com>
    Roberto Ricci <ricci@disroot.org>
    Rob Norris <robn@despairlabs.com>
@@ -509,7 +536,9 @@ CONTRIBUTORS:
    Ryan Lahfa <masterancpp@gmail.com>
    Ryan Libby <rlibby@FreeBSD.org>
    Ryan Moeller <freqlabs@FreeBSD.org>
+    Sam Atkinson <samatk@amazon.com>
    Sam Hathaway <github.com@munkynet.org>
+    Sam James <sam@gentoo.org>
    Sam Lunt <samuel.j.lunt@gmail.com>
    Samuel VERSCHELDE <stormi-github@ylix.fr>
    Samuel Wycliffe <samuelwycliffe@gmail.com>
@@ -530,6 +559,8 @@ CONTRIBUTORS:
    Shaan Nobee <sniper111@gmail.com>
    Shampavman <sham.pavman@nexenta.com>
    Shaun Tancheff <shaun@aeonazure.com>
+    Shawn Bayern <sbayern@law.fsu.edu>
+    Shengqi Chen <harry-chen@outlook.com>
    Shen Yan <shenyanxxxy@qq.com>
    Simon Guest <simon.guest@tesujimath.org>
    Simon Klinkert <simon.klinkert@gmail.com>
@@ -537,6 +568,7 @@ CONTRIBUTORS:
    Spencer Kinny <spencerkinny1995@gmail.com>
    Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com>
    Stanislav Seletskiy <s.seletskiy@gmail.com>
+    Stefan Lendl <s.lendl@proxmox.com>
    Steffen Müthing <steffen.muething@iwr.uni-heidelberg.de>
    Stephen Blinick <stephen.blinick@delphix.com>
    sterlingjensen <sterlingjensen@users.noreply.github.com>
@@ -557,6 +589,7 @@ CONTRIBUTORS:
    Teodor Spæren <teodor_spaeren@riseup.net>
    TerraTech <TerraTech@users.noreply.github.com>
    Thijs Cramer <thijs.cramer@gmail.com>
+    Thomas Bertschinger <bertschinger@lanl.gov>
    Thomas Geppert <geppi@digitx.de>
    Thomas Lamprecht <guggentom@hotmail.de>
    Till Maas <opensource@till.name>
@@ -586,6 +619,7 @@ CONTRIBUTORS:
    Turbo Fredriksson <turbo@bayour.com>
    Tyler J. Stachecki <stachecki.tyler@gmail.com>
    Umer Saleem <usaleem@ixsystems.com>
+    Vaibhav Bhanawat <vaibhav.bhanawat@delphix.com>
    Valmiky Arquissandas <kayvlim@gmail.com>
    Val Packett <val@packett.cool>
    Vince van Oosten <techhazard@codeforyouand.me>
@@ -614,6 +648,7 @@ CONTRIBUTORS:
    yuina822 <ayuichi@club.kyutech.ac.jp>
    YunQiang Su <syq@debian.org>
    Yuri Pankov <yuri.pankov@gmail.com>
+    Yuxin Wang <yuxinwang9999@gmail.com>
    Yuxuan Shui <yshuiv7@gmail.com>
    Zachary Bedell <zac@thebedells.org>
    Zach Dykstra <dykstra.zachary@gmail.com>
@@ -1,10 +1,10 @@
 Meta:          1
 Name:          zfs
 Branch:        1.0
-Version:       2.2.3
+Version:       2.2.4
 Release:       1
 Release-Tags:  relext
 License:       CDDL
 Author:        OpenZFS
-Linux-Maximum: 6.7
+Linux-Maximum: 6.8
 Linux-Minimum: 3.10
@@ -793,18 +793,27 @@ def section_dmu(kstats_dict):

    zfetch_stats = isolate_section('zfetchstats', kstats_dict)

-    zfetch_access_total = int(zfetch_stats['hits'])+int(zfetch_stats['misses'])
+    zfetch_access_total = int(zfetch_stats['hits']) +\
+        int(zfetch_stats['future']) + int(zfetch_stats['stride']) +\
+        int(zfetch_stats['past']) + int(zfetch_stats['misses'])

    prt_1('DMU predictive prefetcher calls:', f_hits(zfetch_access_total))
    prt_i2('Stream hits:',
           f_perc(zfetch_stats['hits'], zfetch_access_total),
           f_hits(zfetch_stats['hits']))
+    future = int(zfetch_stats['future']) + int(zfetch_stats['stride'])
+    prt_i2('Hits ahead of stream:', f_perc(future, zfetch_access_total),
+           f_hits(future))
+    prt_i2('Hits behind stream:',
+           f_perc(zfetch_stats['past'], zfetch_access_total),
+           f_hits(zfetch_stats['past']))
    prt_i2('Stream misses:',
           f_perc(zfetch_stats['misses'], zfetch_access_total),
           f_hits(zfetch_stats['misses']))
    prt_i2('Streams limit reached:',
           f_perc(zfetch_stats['max_streams'], zfetch_stats['misses']),
           f_hits(zfetch_stats['max_streams']))
+    prt_i1('Stream strides:', f_hits(zfetch_stats['stride']))
    prt_i1('Prefetches issued', f_hits(zfetch_stats['io_issued']))
    print()

@@ -157,6 +157,16 @@ cols = {
    "free":       [5, 1024, "ARC free memory"],
    "avail":      [5, 1024, "ARC available memory"],
    "waste":      [5, 1024, "Wasted memory due to round up to pagesize"],
+    "ztotal":     [6, 1000, "zfetch total prefetcher calls per second"],
+    "zhits":      [5, 1000, "zfetch stream hits per second"],
+    "zahead":     [6, 1000, "zfetch hits ahead of streams per second"],
+    "zpast":      [5, 1000, "zfetch hits behind streams per second"],
+    "zmisses":    [7, 1000, "zfetch stream misses per second"],
+    "zmax":       [4, 1000, "zfetch limit reached per second"],
+    "zfuture":    [7, 1000, "zfetch stream future per second"],
+    "zstride":    [7, 1000, "zfetch stream strides per second"],
+    "zissued":    [7, 1000, "zfetch prefetches issued per second"],
+    "zactive":    [7, 1000, "zfetch prefetches active per second"],
 }

 v = {}
@@ -164,6 +174,8 @@ hdr = ["time", "read", "ddread", "ddh%", "dmread", "dmh%", "pread", "ph%",
       "size", "c", "avail"]
 xhdr = ["time", "mfu", "mru", "mfug", "mrug", "unc", "eskip", "mtxmis",
        "dread", "pread", "read"]
+zhdr = ["time", "ztotal", "zhits", "zahead", "zpast", "zmisses", "zmax",
+        "zfuture", "zstride", "zissued", "zactive"]
 sint = 1               # Default interval is 1 second
 count = 1              # Default count is 1
 hdr_intr = 20          # Print header every 20 lines of output
@@ -188,6 +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)
@@ -199,19 +213,28 @@ 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():
        global kstat

-        k = [line.strip() for line in open('/proc/spl/kstat/zfs/arcstats')]
+        k1 = [line.strip() for line in open('/proc/spl/kstat/zfs/arcstats')]

-        if not k:
+        k2 = ["zfetch_" + line.strip() for line in
+             open('/proc/spl/kstat/zfs/zfetchstats')]
+
+        if k1 is None or k2 is None:
            sys.exit(1)

-        del k[0:2]
+        del k1[0:2]
+        del k2[0:2]
+        k = k1 + k2
        kstat = {}

        for s in k:
@@ -239,6 +262,7 @@ def usage():
    sys.stderr.write("\t -v : List all possible field headers and definitions"
                     "\n")
    sys.stderr.write("\t -x : Print extended stats\n")
+    sys.stderr.write("\t -z : Print zfetch stats\n")
    sys.stderr.write("\t -f : Specify specific fields to print (see -v)\n")
    sys.stderr.write("\t -o : Redirect output to the specified file\n")
    sys.stderr.write("\t -s : Override default field separator with custom "
@@ -357,6 +381,7 @@ def init():
    global count
    global hdr
    global xhdr
+    global zhdr
    global opfile
    global sep
    global out
@@ -368,15 +393,17 @@ def init():
    xflag = False
    hflag = False
    vflag = False
+    zflag = False
    i = 1

    try:
        opts, args = getopt.getopt(
            sys.argv[1:],
-            "axo:hvs:f:p",
+            "axzo:hvs:f:p",
            [
                "all",
                "extended",
+                "zfetch",
                "outfile",
                "help",
                "verbose",
@@ -410,13 +437,15 @@ def init():
            i += 1
        if opt in ('-p', '--parsable'):
            pretty_print = False
+        if opt in ('-z', '--zfetch'):
+            zflag = True
        i += 1

    argv = sys.argv[i:]
    sint = int(argv[0]) if argv else sint
    count = int(argv[1]) if len(argv) > 1 else (0 if len(argv) > 0 else 1)

-    if hflag or (xflag and desired_cols):
+    if hflag or (xflag and zflag) or ((zflag or xflag) and desired_cols):
        usage()

    if vflag:
@@ -425,6 +454,9 @@ def init():
    if xflag:
        hdr = xhdr

+    if zflag:
+        hdr = zhdr
+
    update_hdr_intr()

    # check if L2ARC exists
@@ -569,6 +601,17 @@ def calculate():
    v["el2mru"] = d["evict_l2_eligible_mru"] // sint
    v["el2inel"] = d["evict_l2_ineligible"] // sint
    v["mtxmis"] = d["mutex_miss"] // sint
+    v["ztotal"] = (d["zfetch_hits"] + d["zfetch_future"] + d["zfetch_stride"] +
+                   d["zfetch_past"] + d["zfetch_misses"]) // sint
+    v["zhits"] = d["zfetch_hits"] // sint
+    v["zahead"] = (d["zfetch_future"] + d["zfetch_stride"]) // sint
+    v["zpast"] = d["zfetch_past"] // sint
+    v["zmisses"] = d["zfetch_misses"] // sint
+    v["zmax"] = d["zfetch_max_streams"] // sint
+    v["zfuture"] = d["zfetch_future"] // sint
+    v["zstride"] = d["zfetch_stride"] // sint
+    v["zissued"] = d["zfetch_io_issued"] // sint
+    v["zactive"] = d["zfetch_io_active"] // sint

    if l2exist:
        v["l2hits"] = d["l2_hits"] // sint
@@ -22,6 +22,7 @@
 * Copyright (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
 *
 * Copyright (c) 2016, Intel Corporation.
+ * Copyright (c) 2023, Klara Inc.
 */

 /*
@@ -231,28 +232,6 @@ fmd_prop_get_int32(fmd_hdl_t *hdl, const char *name)
 	if (strcmp(name, "spare_on_remove") == 0)
 		return (1);

-	if (strcmp(name, "io_N") == 0 || strcmp(name, "checksum_N") == 0)
-		return (10);	/* N = 10 events */
-
-	return (0);
-}
-
-int64_t
-fmd_prop_get_int64(fmd_hdl_t *hdl, const char *name)
-{
-	(void) hdl;
-
-	/*
-	 * These can be looked up in mp->modinfo->fmdi_props
-	 * For now we just hard code for phase 2. In the
-	 * future, there can be a ZED based override.
-	 */
-	if (strcmp(name, "remove_timeout") == 0)
-		return (15ULL * 1000ULL * 1000ULL * 1000ULL);	/* 15 sec */
-
-	if (strcmp(name, "io_T") == 0 || strcmp(name, "checksum_T") == 0)
-		return (1000ULL * 1000ULL * 1000ULL * 600ULL);	/* 10 min */
-
 	return (0);
 }

@@ -535,6 +514,19 @@ fmd_serd_exists(fmd_hdl_t *hdl, const char *name)
 	return (fmd_serd_eng_lookup(&mp->mod_serds, name) != NULL);
 }

+int
+fmd_serd_active(fmd_hdl_t *hdl, const char *name)
+{
+	fmd_module_t *mp = (fmd_module_t *)hdl;
+	fmd_serd_eng_t *sgp;
+
+	if ((sgp = fmd_serd_eng_lookup(&mp->mod_serds, name)) == NULL) {
+		zed_log_msg(LOG_ERR, "serd engine '%s' does not exist", name);
+		return (0);
+	}
+	return (fmd_serd_eng_fired(sgp) || !fmd_serd_eng_empty(sgp));
+}
+
 void
 fmd_serd_reset(fmd_hdl_t *hdl, const char *name)
 {
@@ -543,12 +535,10 @@ fmd_serd_reset(fmd_hdl_t *hdl, const char *name)

 	if ((sgp = fmd_serd_eng_lookup(&mp->mod_serds, name)) == NULL) {
 		zed_log_msg(LOG_ERR, "serd engine '%s' does not exist", name);
-		return;
+	} else {
+		fmd_serd_eng_reset(sgp);
+		fmd_hdl_debug(hdl, "serd_reset %s", name);
 	}
-
-	fmd_serd_eng_reset(sgp);
-
-	fmd_hdl_debug(hdl, "serd_reset %s", name);
 }

 int
@@ -556,16 +546,21 @@ fmd_serd_record(fmd_hdl_t *hdl, const char *name, fmd_event_t *ep)
 {
 	fmd_module_t *mp = (fmd_module_t *)hdl;
 	fmd_serd_eng_t *sgp;
-	int err;

 	if ((sgp = fmd_serd_eng_lookup(&mp->mod_serds, name)) == NULL) {
 		zed_log_msg(LOG_ERR, "failed to add record to SERD engine '%s'",
 		    name);
 		return (0);
 	}
-	err = fmd_serd_eng_record(sgp, ep->ev_hrt);
+	return (fmd_serd_eng_record(sgp, ep->ev_hrt));
+}

-	return (err);
+void
+fmd_serd_gc(fmd_hdl_t *hdl)
+{
+	fmd_module_t *mp = (fmd_module_t *)hdl;
+
+	fmd_serd_hash_apply(&mp->mod_serds, fmd_serd_eng_gc, NULL);
 }

 /* FMD Timers */
@@ -579,7 +574,7 @@ _timer_notify(union sigval sv)
 	const fmd_hdl_ops_t *ops = mp->mod_info->fmdi_ops;
 	struct itimerspec its;

-	fmd_hdl_debug(hdl, "timer fired (%p)", ftp->ft_tid);
+	fmd_hdl_debug(hdl, "%s timer fired (%p)", mp->mod_name, ftp->ft_tid);

 	/* disarm the timer */
 	memset(&its, 0, sizeof (struct itimerspec));
@@ -151,7 +151,6 @@ extern void fmd_hdl_vdebug(fmd_hdl_t *, const char *, va_list);
 extern void fmd_hdl_debug(fmd_hdl_t *, const char *, ...);

 extern int32_t fmd_prop_get_int32(fmd_hdl_t *, const char *);
-extern int64_t fmd_prop_get_int64(fmd_hdl_t *, const char *);

 #define	FMD_STAT_NOALLOC	0x0	/* fmd should use caller's memory */
 #define	FMD_STAT_ALLOC		0x1	/* fmd should allocate stats memory */
@@ -195,10 +194,12 @@ extern size_t fmd_buf_size(fmd_hdl_t *, fmd_case_t *, const char *);
 extern void fmd_serd_create(fmd_hdl_t *, const char *, uint_t, hrtime_t);
 extern void fmd_serd_destroy(fmd_hdl_t *, const char *);
 extern int fmd_serd_exists(fmd_hdl_t *, const char *);
+extern int fmd_serd_active(fmd_hdl_t *, const char *);
 extern void fmd_serd_reset(fmd_hdl_t *, const char *);
 extern int fmd_serd_record(fmd_hdl_t *, const char *, fmd_event_t *);
 extern int fmd_serd_fired(fmd_hdl_t *, const char *);
 extern int fmd_serd_empty(fmd_hdl_t *, const char *);
+extern void fmd_serd_gc(fmd_hdl_t *);

 extern id_t fmd_timer_install(fmd_hdl_t *, void *, fmd_event_t *, hrtime_t);
 extern void fmd_timer_remove(fmd_hdl_t *, id_t);
@@ -310,8 +310,9 @@ fmd_serd_eng_reset(fmd_serd_eng_t *sgp)
 }

 void
-fmd_serd_eng_gc(fmd_serd_eng_t *sgp)
+fmd_serd_eng_gc(fmd_serd_eng_t *sgp, void *arg)
 {
+	(void) arg;
 	fmd_serd_elem_t *sep, *nep;
 	hrtime_t hrt;

@@ -77,7 +77,7 @@ extern int fmd_serd_eng_fired(fmd_serd_eng_t *);
 extern int fmd_serd_eng_empty(fmd_serd_eng_t *);

 extern void fmd_serd_eng_reset(fmd_serd_eng_t *);
-extern void fmd_serd_eng_gc(fmd_serd_eng_t *);
+extern void fmd_serd_eng_gc(fmd_serd_eng_t *, void *);

 #ifdef	__cplusplus
 }
@@ -23,6 +23,7 @@
 * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright 2015 Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2016, Intel Corporation.
+ * Copyright (c) 2023, Klara Inc.
 */

 #include <stddef.h>
@@ -47,11 +48,16 @@
 #define	DEFAULT_CHECKSUM_T	600	/* seconds */
 #define	DEFAULT_IO_N		10	/* events */
 #define	DEFAULT_IO_T		600	/* seconds */
+#define	DEFAULT_SLOW_IO_N	10	/* events */
+#define	DEFAULT_SLOW_IO_T	30	/* seconds */
+
+#define	CASE_GC_TIMEOUT_SECS	43200	/* 12 hours */

 /*
- * Our serd engines are named 'zfs_<pool_guid>_<vdev_guid>_{checksum,io}'.  This
- * #define reserves enough space for two 64-bit hex values plus the length of
- * the longest string.
+ * Our serd engines are named in the following format:
+ *     'zfs_<pool_guid>_<vdev_guid>_{checksum,io,slow_io}'
+ * This #define reserves enough space for two 64-bit hex values plus the
+ * length of the longest string.
 */
 #define	MAX_SERDLEN	(16 * 2 + sizeof ("zfs___checksum"))

@@ -68,6 +74,7 @@ typedef struct zfs_case_data {
 	int		zc_pool_state;
 	char		zc_serd_checksum[MAX_SERDLEN];
 	char		zc_serd_io[MAX_SERDLEN];
+	char		zc_serd_slow_io[MAX_SERDLEN];
 	int		zc_has_remove_timer;
 } zfs_case_data_t;

@@ -114,7 +121,8 @@ zfs_de_stats_t zfs_stats = {
 	{ "resource_drops", FMD_TYPE_UINT64, "resource related ereports" }
 };

-static hrtime_t zfs_remove_timeout;
+/* wait 15 seconds after a removal */
+static hrtime_t zfs_remove_timeout = SEC2NSEC(15);

 uu_list_pool_t *zfs_case_pool;
 uu_list_t *zfs_cases;
@@ -124,6 +132,8 @@ uu_list_t *zfs_cases;
 #define	ZFS_MAKE_EREPORT(type)	\
    FM_EREPORT_CLASS "." ZFS_ERROR_CLASS "." type

+static void zfs_purge_cases(fmd_hdl_t *hdl);
+
 /*
 * Write out the persistent representation of an active case.
 */
@@ -170,6 +180,42 @@ zfs_case_unserialize(fmd_hdl_t *hdl, fmd_case_t *cp)
 	return (zcp);
 }

+/*
+ * count other unique slow-io cases in a pool
+ */
+static uint_t
+zfs_other_slow_cases(fmd_hdl_t *hdl, const zfs_case_data_t *zfs_case)
+{
+	zfs_case_t *zcp;
+	uint_t cases = 0;
+	static hrtime_t next_check = 0;
+
+	/*
+	 * Note that plumbing in some external GC would require adding locking,
+	 * since most of this module code is not thread safe and assumes there
+	 * is only one thread running against the module. So we perform GC here
+	 * inline periodically so that future delay induced faults will be
+	 * possible once the issue causing multiple vdev delays is resolved.
+	 */
+	if (gethrestime_sec() > next_check) {
+		/* Periodically purge old SERD entries and stale cases */
+		fmd_serd_gc(hdl);
+		zfs_purge_cases(hdl);
+		next_check = gethrestime_sec() + CASE_GC_TIMEOUT_SECS;
+	}
+
+	for (zcp = uu_list_first(zfs_cases); zcp != NULL;
+	    zcp = uu_list_next(zfs_cases, zcp)) {
+		if (zcp->zc_data.zc_pool_guid == zfs_case->zc_pool_guid &&
+		    zcp->zc_data.zc_vdev_guid != zfs_case->zc_vdev_guid &&
+		    zcp->zc_data.zc_serd_slow_io[0] != '\0' &&
+		    fmd_serd_active(hdl, zcp->zc_data.zc_serd_slow_io)) {
+			cases++;
+		}
+	}
+	return (cases);
+}
+
 /*
 * Iterate over any active cases.  If any cases are associated with a pool or
 * vdev which is no longer present on the system, close the associated case.
@@ -376,6 +422,14 @@ zfs_serd_name(char *buf, uint64_t pool_guid, uint64_t vdev_guid,
 	    (long long unsigned int)vdev_guid, type);
 }

+static void
+zfs_case_retire(fmd_hdl_t *hdl, zfs_case_t *zcp)
+{
+	fmd_hdl_debug(hdl, "retiring case");
+
+	fmd_case_close(hdl, zcp->zc_case);
+}
+
 /*
 * Solve a given ZFS case.  This first checks to make sure the diagnosis is
 * still valid, as well as cleaning up any pending timer associated with the
@@ -632,9 +686,7 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 		if (strcmp(class,
 		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_DATA)) == 0 ||
 		    strcmp(class,
-		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_CONFIG_CACHE_WRITE)) == 0 ||
-		    strcmp(class,
-		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_DELAY)) == 0) {
+		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_CONFIG_CACHE_WRITE)) == 0) {
 			zfs_stats.resource_drops.fmds_value.ui64++;
 			return;
 		}
@@ -702,6 +754,9 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 			if (zcp->zc_data.zc_serd_checksum[0] != '\0')
 				fmd_serd_reset(hdl,
 				    zcp->zc_data.zc_serd_checksum);
+			if (zcp->zc_data.zc_serd_slow_io[0] != '\0')
+				fmd_serd_reset(hdl,
+				    zcp->zc_data.zc_serd_slow_io);
 		} else if (fmd_nvl_class_match(hdl, nvl,
 		    ZFS_MAKE_RSRC(FM_RESOURCE_STATECHANGE))) {
 			uint64_t state = 0;
@@ -730,7 +785,11 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 	if (fmd_case_solved(hdl, zcp->zc_case))
 		return;

-	fmd_hdl_debug(hdl, "error event '%s'", class);
+	if (vdev_guid)
+		fmd_hdl_debug(hdl, "error event '%s', vdev %llu", class,
+		    vdev_guid);
+	else
+		fmd_hdl_debug(hdl, "error event '%s'", class);

 	/*
 	 * Determine if we should solve the case and generate a fault.  We solve
@@ -779,6 +838,8 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 	    fmd_nvl_class_match(hdl, nvl,
 	    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_IO_FAILURE)) ||
 	    fmd_nvl_class_match(hdl, nvl,
+	    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_DELAY)) ||
+	    fmd_nvl_class_match(hdl, nvl,
 	    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_PROBE_FAILURE))) {
 		const char *failmode = NULL;
 		boolean_t checkremove = B_FALSE;
@@ -814,6 +875,51 @@ zfs_fm_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl, const char *class)
 			}
 			if (fmd_serd_record(hdl, zcp->zc_data.zc_serd_io, ep))
 				checkremove = B_TRUE;
+		} else if (fmd_nvl_class_match(hdl, nvl,
+		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_DELAY))) {
+			uint64_t slow_io_n, slow_io_t;
+
+			/*
+			 * Create a slow io SERD engine when the VDEV has the
+			 * 'vdev_slow_io_n' and 'vdev_slow_io_n' properties.
+			 */
+			if (zcp->zc_data.zc_serd_slow_io[0] == '\0' &&
+			    nvlist_lookup_uint64(nvl,
+			    FM_EREPORT_PAYLOAD_ZFS_VDEV_SLOW_IO_N,
+			    &slow_io_n) == 0 &&
+			    nvlist_lookup_uint64(nvl,
+			    FM_EREPORT_PAYLOAD_ZFS_VDEV_SLOW_IO_T,
+			    &slow_io_t) == 0) {
+				zfs_serd_name(zcp->zc_data.zc_serd_slow_io,
+				    pool_guid, vdev_guid, "slow_io");
+				fmd_serd_create(hdl,
+				    zcp->zc_data.zc_serd_slow_io,
+				    slow_io_n,
+				    SEC2NSEC(slow_io_t));
+				zfs_case_serialize(zcp);
+			}
+			/* Pass event to SERD engine and see if this triggers */
+			if (zcp->zc_data.zc_serd_slow_io[0] != '\0' &&
+			    fmd_serd_record(hdl, zcp->zc_data.zc_serd_slow_io,
+			    ep)) {
+				/*
+				 * Ignore a slow io diagnosis when other
+				 * VDEVs in the pool show signs of being slow.
+				 */
+				if (zfs_other_slow_cases(hdl, &zcp->zc_data)) {
+					zfs_case_retire(hdl, zcp);
+					fmd_hdl_debug(hdl, "pool %llu has "
+					    "multiple slow io cases -- skip "
+					    "degrading vdev %llu",
+					    (u_longlong_t)
+					    zcp->zc_data.zc_pool_guid,
+					    (u_longlong_t)
+					    zcp->zc_data.zc_vdev_guid);
+				} else {
+					zfs_case_solve(hdl, zcp,
+					    "fault.fs.zfs.vdev.slow_io");
+				}
+			}
 		} else if (fmd_nvl_class_match(hdl, nvl,
 		    ZFS_MAKE_EREPORT(FM_EREPORT_ZFS_CHECKSUM))) {
 			/*
@@ -924,6 +1030,8 @@ zfs_fm_close(fmd_hdl_t *hdl, fmd_case_t *cs)
 		fmd_serd_destroy(hdl, zcp->zc_data.zc_serd_checksum);
 	if (zcp->zc_data.zc_serd_io[0] != '\0')
 		fmd_serd_destroy(hdl, zcp->zc_data.zc_serd_io);
+	if (zcp->zc_data.zc_serd_slow_io[0] != '\0')
+		fmd_serd_destroy(hdl, zcp->zc_data.zc_serd_slow_io);
 	if (zcp->zc_data.zc_has_remove_timer)
 		fmd_timer_remove(hdl, zcp->zc_remove_timer);

@@ -932,30 +1040,15 @@ zfs_fm_close(fmd_hdl_t *hdl, fmd_case_t *cs)
 	fmd_hdl_free(hdl, zcp, sizeof (zfs_case_t));
 }

-/*
- * We use the fmd gc entry point to look for old cases that no longer apply.
- * This allows us to keep our set of case data small in a long running system.
- */
-static void
-zfs_fm_gc(fmd_hdl_t *hdl)
-{
-	zfs_purge_cases(hdl);
-}
-
 static const fmd_hdl_ops_t fmd_ops = {
 	zfs_fm_recv,	/* fmdo_recv */
 	zfs_fm_timeout,	/* fmdo_timeout */
 	zfs_fm_close,	/* fmdo_close */
 	NULL,		/* fmdo_stats */
-	zfs_fm_gc,	/* fmdo_gc */
+	NULL,	/* fmdo_gc */
 };

 static const fmd_prop_t fmd_props[] = {
-	{ "checksum_N", FMD_TYPE_UINT32, "10" },
-	{ "checksum_T", FMD_TYPE_TIME, "10min" },
-	{ "io_N", FMD_TYPE_UINT32, "10" },
-	{ "io_T", FMD_TYPE_TIME, "10min" },
-	{ "remove_timeout", FMD_TYPE_TIME, "15sec" },
 	{ NULL, 0, NULL }
 };

@@ -996,8 +1089,6 @@ _zfs_diagnosis_init(fmd_hdl_t *hdl)

 	(void) fmd_stat_create(hdl, FMD_STAT_NOALLOC, sizeof (zfs_stats) /
 	    sizeof (fmd_stat_t), (fmd_stat_t *)&zfs_stats);
-
-	zfs_remove_timeout = fmd_prop_get_int64(hdl, "remove_timeout");
 }

 void
@@ -523,6 +523,9 @@ zfs_retire_recv(fmd_hdl_t *hdl, fmd_event_t *ep, nvlist_t *nvl,
 		} else if (fmd_nvl_class_match(hdl, fault,
 		    "fault.fs.zfs.vdev.checksum")) {
 			degrade_device = B_TRUE;
+		} else if (fmd_nvl_class_match(hdl, fault,
+		    "fault.fs.zfs.vdev.slow_io")) {
+			degrade_device = B_TRUE;
 		} else if (fmd_nvl_class_match(hdl, fault,
 		    "fault.fs.zfs.device")) {
 			fault_device = B_FALSE;
@@ -309,7 +309,8 @@ get_usage(zfs_help_t idx)
 		    "[filesystem|volume|snapshot] ...\n"));
 	case HELP_MOUNT:
 		return (gettext("\tmount\n"
-		    "\tmount [-flvO] [-o opts] <-a | filesystem>\n"));
+		    "\tmount [-flvO] [-o opts] <-a|-R filesystem|"
+		    "filesystem>\n"));
 	case HELP_PROMOTE:
 		return (gettext("\tpromote <clone-filesystem>\n"));
 	case HELP_RECEIVE:
@@ -6750,6 +6751,8 @@ zfs_do_holds(int argc, char **argv)
 #define	MOUNT_TIME 1		/* seconds */

 typedef struct get_all_state {
+	char		**ga_datasets;
+	int		ga_count;
 	boolean_t	ga_verbose;
 	get_all_cb_t	*ga_cbp;
 } get_all_state_t;
@@ -6796,19 +6799,35 @@ get_one_dataset(zfs_handle_t *zhp, void *data)
 	return (0);
 }

-static void
-get_all_datasets(get_all_cb_t *cbp, boolean_t verbose)
+static int
+get_recursive_datasets(zfs_handle_t *zhp, void *data)
 {
-	get_all_state_t state = {
-	    .ga_verbose = verbose,
-	    .ga_cbp = cbp
-	};
+	get_all_state_t *state = data;
+	int len = strlen(zfs_get_name(zhp));
+	for (int i = 0; i < state->ga_count; ++i) {
+		if (strcmp(state->ga_datasets[i], zfs_get_name(zhp)) == 0)
+			return (get_one_dataset(zhp, data));
+		else if ((strncmp(state->ga_datasets[i], zfs_get_name(zhp),
+		    len) == 0) && state->ga_datasets[i][len] == '/') {
+			(void) zfs_iter_filesystems_v2(zhp, 0,
+			    get_recursive_datasets, data);
+		}
+	}
+	zfs_close(zhp);
+	return (0);
+}

-	if (verbose)
+static void
+get_all_datasets(get_all_state_t *state)
+{
+	if (state->ga_verbose)
 		set_progress_header(gettext("Reading ZFS config"));
-	(void) zfs_iter_root(g_zfs, get_one_dataset, &state);
+	if (state->ga_datasets == NULL)
+		(void) zfs_iter_root(g_zfs, get_one_dataset, state);
+	else
+		(void) zfs_iter_root(g_zfs, get_recursive_datasets, state);

-	if (verbose)
+	if (state->ga_verbose)
 		finish_progress(gettext("done."));
 }

@@ -7154,18 +7173,22 @@ static int
 share_mount(int op, int argc, char **argv)
 {
 	int do_all = 0;
+	int recursive = 0;
 	boolean_t verbose = B_FALSE;
 	int c, ret = 0;
 	char *options = NULL;
 	int flags = 0;

 	/* check options */
-	while ((c = getopt(argc, argv, op == OP_MOUNT ? ":alvo:Of" : "al"))
+	while ((c = getopt(argc, argv, op == OP_MOUNT ? ":aRlvo:Of" : "al"))
 	    != -1) {
 		switch (c) {
 		case 'a':
 			do_all = 1;
 			break;
+		case 'R':
+			recursive = 1;
+			break;
 		case 'v':
 			verbose = B_TRUE;
 			break;
@@ -7207,7 +7230,7 @@ share_mount(int op, int argc, char **argv)
 	argv += optind;

 	/* check number of arguments */
-	if (do_all) {
+	if (do_all || recursive) {
 		enum sa_protocol protocol = SA_NO_PROTOCOL;

 		if (op == OP_SHARE && argc > 0) {
@@ -7216,14 +7239,38 @@ share_mount(int op, int argc, char **argv)
 			argv++;
 		}

-		if (argc != 0) {
+		if (argc != 0 && do_all) {
 			(void) fprintf(stderr, gettext("too many arguments\n"));
 			usage(B_FALSE);
 		}

+		if (argc == 0 && recursive) {
+			(void) fprintf(stderr,
+			    gettext("no dataset provided\n"));
+			usage(B_FALSE);
+		}
+
 		start_progress_timer();
 		get_all_cb_t cb = { 0 };
-		get_all_datasets(&cb, verbose);
+		get_all_state_t state = { 0 };
+		if (argc == 0) {
+			state.ga_datasets = NULL;
+			state.ga_count = -1;
+		} else {
+			zfs_handle_t *zhp;
+			for (int i = 0; i < argc; i++) {
+				zhp = zfs_open(g_zfs, argv[i],
+				    ZFS_TYPE_FILESYSTEM);
+				if (zhp == NULL)
+					usage(B_FALSE);
+				zfs_close(zhp);
+			}
+			state.ga_datasets = argv;
+			state.ga_count = argc;
+		}
+		state.ga_verbose = verbose;
+		state.ga_cbp = &cb;
+		get_all_datasets(&state);

 		if (cb.cb_used == 0) {
 			free(options);
@@ -1083,6 +1083,22 @@ main(int argc, char **argv)
 			libzfs_fini(g_zfs);
 			return (1);
 		}
+
+		if (record.zi_nlanes) {
+			switch (io_type) {
+			case ZIO_TYPE_READ:
+			case ZIO_TYPE_WRITE:
+			case ZIO_TYPES:
+				break;
+			default:
+				(void) fprintf(stderr, "I/O type for a delay "
+				    "must be 'read' or 'write'\n");
+				usage();
+				libzfs_fini(g_zfs);
+				return (1);
+			}
+		}
+
 		if (!error)
 			error = ENXIO;

@@ -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);
@@ -458,7 +458,7 @@ static char *zpool_sysfs_gets(char *path)
 	}

 	/* Remove trailing newline */
-	if (buf[count - 1] == '\n')
+	if (count > 0 && buf[count - 1] == '\n')
 		buf[count - 1] = 0;

 	close(fd);
@@ -22,7 +22,7 @@
 /*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright 2011 Nexenta Systems, Inc. All rights reserved.
- * Copyright (c) 2011, 2020 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright (c) 2012 by Frederik Wessels. All rights reserved.
 * Copyright (c) 2012 by Cyril Plisko. All rights reserved.
 * Copyright (c) 2013 by Prasad Joshi (sTec). All rights reserved.
@@ -131,6 +131,13 @@ static int zpool_do_help(int argc, char **argv);
 static zpool_compat_status_t zpool_do_load_compat(
    const char *, boolean_t *);

+enum zpool_options {
+	ZPOOL_OPTION_POWER = 1024,
+	ZPOOL_OPTION_ALLOW_INUSE,
+	ZPOOL_OPTION_ALLOW_REPLICATION_MISMATCH,
+	ZPOOL_OPTION_ALLOW_ASHIFT_MISMATCH
+};
+
 /*
 * These libumem hooks provide a reasonable set of defaults for the allocator's
 * debugging facilities.
@@ -347,7 +354,7 @@ get_usage(zpool_help_t idx)
 {
 	switch (idx) {
 	case HELP_ADD:
-		return (gettext("\tadd [-fgLnP] [-o property=value] "
+		return (gettext("\tadd [-afgLnP] [-o property=value] "
 		    "<pool> <vdev> ...\n"));
 	case HELP_ATTACH:
 		return (gettext("\tattach [-fsw] [-o property=value] "
@@ -413,7 +420,7 @@ get_usage(zpool_help_t idx)
 		    "[<device> ...]\n"));
 	case HELP_STATUS:
 		return (gettext("\tstatus [--power] [-c [script1,script2,...]] "
-		    "[-igLpPstvxD]  [-T d|u] [pool] ... \n"
+		    "[-DegiLpPstvx] [-T d|u] [pool] ...\n"
 		    "\t    [interval [count]]\n"));
 	case HELP_UPGRADE:
 		return (gettext("\tupgrade\n"
@@ -1009,8 +1016,9 @@ add_prop_list_default(const char *propname, const char *propval,
 }

 /*
- * zpool add [-fgLnP] [-o property=value] <pool> <vdev> ...
+ * zpool add [-afgLnP] [-o property=value] <pool> <vdev> ...
 *
+ *	-a	Disable the ashift validation checks
 *	-f	Force addition of devices, even if they appear in use
 *	-g	Display guid for individual vdev name.
 *	-L	Follow links when resolving vdev path name.
@@ -1026,8 +1034,11 @@ add_prop_list_default(const char *propname, const char *propval,
 int
 zpool_do_add(int argc, char **argv)
 {
-	boolean_t force = B_FALSE;
+	boolean_t check_replication = B_TRUE;
+	boolean_t check_inuse = B_TRUE;
 	boolean_t dryrun = B_FALSE;
+	boolean_t check_ashift = B_TRUE;
+	boolean_t force = B_FALSE;
 	int name_flags = 0;
 	int c;
 	nvlist_t *nvroot;
@@ -1038,8 +1049,18 @@ zpool_do_add(int argc, char **argv)
 	nvlist_t *props = NULL;
 	char *propval;

+	struct option long_options[] = {
+		{"allow-in-use", no_argument, NULL, ZPOOL_OPTION_ALLOW_INUSE},
+		{"allow-replication-mismatch", no_argument, NULL,
+		    ZPOOL_OPTION_ALLOW_REPLICATION_MISMATCH},
+		{"allow-ashift-mismatch", no_argument, NULL,
+		    ZPOOL_OPTION_ALLOW_ASHIFT_MISMATCH},
+		{0, 0, 0, 0}
+	};
+
 	/* check options */
-	while ((c = getopt(argc, argv, "fgLno:P")) != -1) {
+	while ((c = getopt_long(argc, argv, "fgLno:P", long_options, NULL))
+	    != -1) {
 		switch (c) {
 		case 'f':
 			force = B_TRUE;
@@ -1069,6 +1090,15 @@ zpool_do_add(int argc, char **argv)
 		case 'P':
 			name_flags |= VDEV_NAME_PATH;
 			break;
+		case ZPOOL_OPTION_ALLOW_INUSE:
+			check_inuse = B_FALSE;
+			break;
+		case ZPOOL_OPTION_ALLOW_REPLICATION_MISMATCH:
+			check_replication = B_FALSE;
+			break;
+		case ZPOOL_OPTION_ALLOW_ASHIFT_MISMATCH:
+			check_ashift = B_FALSE;
+			break;
 		case '?':
 			(void) fprintf(stderr, gettext("invalid option '%c'\n"),
 			    optopt);
@@ -1089,6 +1119,19 @@ zpool_do_add(int argc, char **argv)
 		usage(B_FALSE);
 	}

+	if (force) {
+		if (!check_inuse || !check_replication || !check_ashift) {
+			(void) fprintf(stderr, gettext("'-f' option is not "
+			    "allowed with '--allow-replication-mismatch', "
+			    "'--allow-ashift-mismatch', or "
+			    "'--allow-in-use'\n"));
+			usage(B_FALSE);
+		}
+		check_inuse = B_FALSE;
+		check_replication = B_FALSE;
+		check_ashift = B_FALSE;
+	}
+
 	poolname = argv[0];

 	argc--;
@@ -1119,8 +1162,8 @@ zpool_do_add(int argc, char **argv)
 	}

 	/* pass off to make_root_vdev for processing */
-	nvroot = make_root_vdev(zhp, props, force, !force, B_FALSE, dryrun,
-	    argc, argv);
+	nvroot = make_root_vdev(zhp, props, !check_inuse,
+	    check_replication, B_FALSE, dryrun, argc, argv);
 	if (nvroot == NULL) {
 		zpool_close(zhp);
 		return (1);
@@ -1224,7 +1267,7 @@ zpool_do_add(int argc, char **argv)

 		ret = 0;
 	} else {
-		ret = (zpool_add(zhp, nvroot) != 0);
+		ret = (zpool_add(zhp, nvroot, check_ashift) != 0);
 	}

 	nvlist_free(props);
@@ -2246,7 +2289,6 @@ print_status_initialize(vdev_stat_t *vs, boolean_t verbose)
 		    !vs->vs_scan_removing) {
 			char zbuf[1024];
 			char tbuf[256];
-			struct tm zaction_ts;

 			time_t t = vs->vs_initialize_action_time;
 			int initialize_pct = 100;
@@ -2256,8 +2298,8 @@ print_status_initialize(vdev_stat_t *vs, boolean_t verbose)
 				    100 / (vs->vs_initialize_bytes_est + 1));
 			}

-			(void) localtime_r(&t, &zaction_ts);
-			(void) strftime(tbuf, sizeof (tbuf), "%c", &zaction_ts);
+			(void) ctime_r(&t, tbuf);
+			tbuf[24] = 0;

 			switch (vs->vs_initialize_state) {
 			case VDEV_INITIALIZE_SUSPENDED:
@@ -2297,7 +2339,6 @@ print_status_trim(vdev_stat_t *vs, boolean_t verbose)
 		    !vs->vs_scan_removing) {
 			char zbuf[1024];
 			char tbuf[256];
-			struct tm zaction_ts;

 			time_t t = vs->vs_trim_action_time;
 			int trim_pct = 100;
@@ -2306,8 +2347,8 @@ print_status_trim(vdev_stat_t *vs, boolean_t verbose)
 				    100 / (vs->vs_trim_bytes_est + 1));
 			}

-			(void) localtime_r(&t, &zaction_ts);
-			(void) strftime(tbuf, sizeof (tbuf), "%c", &zaction_ts);
+			(void) ctime_r(&t, tbuf);
+			tbuf[24] = 0;

 			switch (vs->vs_trim_state) {
 			case VDEV_TRIM_SUSPENDED:
@@ -2569,7 +2610,13 @@ print_status_config(zpool_handle_t *zhp, status_cbdata_t *cb, const char *name,
 			break;

 		case VDEV_AUX_ERR_EXCEEDED:
-			(void) printf(gettext("too many errors"));
+			if (vs->vs_read_errors + vs->vs_write_errors +
+			    vs->vs_checksum_errors == 0 && children == 0 &&
+			    vs->vs_slow_ios > 0) {
+				(void) printf(gettext("too many slow I/Os"));
+			} else {
+				(void) printf(gettext("too many errors"));
+			}
 			break;

 		case VDEV_AUX_IO_FAILURE:
@@ -3396,10 +3443,10 @@ do_import(nvlist_t *config, const char *newname, const char *mntopts,
 		ms_status = zpool_enable_datasets(zhp, mntopts, 0);
 		if (ms_status == EZFS_SHAREFAILED) {
 			(void) fprintf(stderr, gettext("Import was "
-			    "successful, but unable to share some datasets"));
+			    "successful, but unable to share some datasets\n"));
 		} else if (ms_status == EZFS_MOUNTFAILED) {
 			(void) fprintf(stderr, gettext("Import was "
-			    "successful, but unable to mount some datasets"));
+			    "successful, but unable to mount some datasets\n"));
 		}
 	}

@@ -7064,7 +7111,6 @@ zpool_do_split(int argc, char **argv)
 	return (ret);
 }

-#define	POWER_OPT 1024

 /*
 * zpool online [--power] <pool> <device> ...
@@ -7082,7 +7128,7 @@ zpool_do_online(int argc, char **argv)
 	int flags = 0;
 	boolean_t is_power_on = B_FALSE;
 	struct option long_options[] = {
-		{"power", no_argument, NULL, POWER_OPT},
+		{"power", no_argument, NULL, ZPOOL_OPTION_POWER},
 		{0, 0, 0, 0}
 	};

@@ -7092,7 +7138,7 @@ zpool_do_online(int argc, char **argv)
 		case 'e':
 			flags |= ZFS_ONLINE_EXPAND;
 			break;
-		case POWER_OPT:
+		case ZPOOL_OPTION_POWER:
 			is_power_on = B_TRUE;
 			break;
 		case '?':
@@ -7205,7 +7251,7 @@ zpool_do_offline(int argc, char **argv)
 	boolean_t is_power_off = B_FALSE;

 	struct option long_options[] = {
-		{"power", no_argument, NULL, POWER_OPT},
+		{"power", no_argument, NULL, ZPOOL_OPTION_POWER},
 		{0, 0, 0, 0}
 	};

@@ -7218,7 +7264,7 @@ zpool_do_offline(int argc, char **argv)
 		case 't':
 			istmp = B_TRUE;
 			break;
-		case POWER_OPT:
+		case ZPOOL_OPTION_POWER:
 			is_power_off = B_TRUE;
 			break;
 		case '?':
@@ -7318,7 +7364,7 @@ zpool_do_clear(int argc, char **argv)
 	char *pool, *device;

 	struct option long_options[] = {
-		{"power", no_argument, NULL, POWER_OPT},
+		{"power", no_argument, NULL, ZPOOL_OPTION_POWER},
 		{0, 0, 0, 0}
 	};

@@ -7335,7 +7381,7 @@ zpool_do_clear(int argc, char **argv)
 		case 'X':
 			xtreme_rewind = B_TRUE;
 			break;
-		case POWER_OPT:
+		case ZPOOL_OPTION_POWER:
 			is_power_on = B_TRUE;
 			break;
 		case '?':
@@ -8864,7 +8910,7 @@ 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:
@@ -9064,22 +9110,22 @@ status_callback(zpool_handle_t *zhp, void *data)
 }

 /*
- * zpool status [-c [script1,script2,...]] [-igLpPstvx] [--power] [-T d|u] ...
+ * zpool status [-c [script1,script2,...]] [-DegiLpPstvx] [--power] [-T d|u] ...
 *              [pool] [interval [count]]
 *
 *	-c CMD	For each vdev, run command CMD
+ *	-D	Display dedup status (undocumented)
 *	-e	Display only unhealthy vdevs
- *	-i	Display vdev initialization status.
 *	-g	Display guid for individual vdev name.
+ *	-i	Display vdev initialization status.
 *	-L	Follow links when resolving vdev path name.
 *	-p	Display values in parsable (exact) format.
 *	-P	Display full path for vdev name.
 *	-s	Display slow IOs column.
- *	-v	Display complete error logs
- *	-x	Display only pools with potential problems
- *	-D	Display dedup status (undocumented)
 *	-t	Display vdev TRIM status.
 *	-T	Display a timestamp in date(1) or Unix format
+ *	-v	Display complete error logs
+ *	-x	Display only pools with potential problems
 *	--power	Display vdev enclosure slot power status
 *
 * Describes the health status of all pools or some subset.
@@ -9095,12 +9141,12 @@ zpool_do_status(int argc, char **argv)
 	char *cmd = NULL;

 	struct option long_options[] = {
-		{"power", no_argument, NULL, POWER_OPT},
+		{"power", no_argument, NULL, ZPOOL_OPTION_POWER},
 		{0, 0, 0, 0}
 	};

 	/* check options */
-	while ((c = getopt_long(argc, argv, "c:eigLpPsvxDtT:", long_options,
+	while ((c = getopt_long(argc, argv, "c:DegiLpPstT:vx", long_options,
 	    NULL)) != -1) {
 		switch (c) {
 		case 'c':
@@ -9127,15 +9173,18 @@ zpool_do_status(int argc, char **argv)
 			}
 			cmd = optarg;
 			break;
+		case 'D':
+			cb.cb_dedup_stats = B_TRUE;
+			break;
 		case 'e':
 			cb.cb_print_unhealthy = B_TRUE;
 			break;
-		case 'i':
-			cb.cb_print_vdev_init = B_TRUE;
-			break;
 		case 'g':
 			cb.cb_name_flags |= VDEV_NAME_GUID;
 			break;
+		case 'i':
+			cb.cb_print_vdev_init = B_TRUE;
+			break;
 		case 'L':
 			cb.cb_name_flags |= VDEV_NAME_FOLLOW_LINKS;
 			break;
@@ -9148,22 +9197,19 @@ zpool_do_status(int argc, char **argv)
 		case 's':
 			cb.cb_print_slow_ios = B_TRUE;
 			break;
-		case 'v':
-			cb.cb_verbose = B_TRUE;
-			break;
-		case 'x':
-			cb.cb_explain = B_TRUE;
-			break;
-		case 'D':
-			cb.cb_dedup_stats = B_TRUE;
-			break;
 		case 't':
 			cb.cb_print_vdev_trim = B_TRUE;
 			break;
 		case 'T':
 			get_timestamp_arg(*optarg);
 			break;
-		case POWER_OPT:
+		case 'v':
+			cb.cb_verbose = B_TRUE;
+			break;
+		case 'x':
+			cb.cb_explain = B_TRUE;
+			break;
+		case ZPOOL_OPTION_POWER:
 			cb.cb_print_power = B_TRUE;
 			break;
 		case '?':
@@ -9202,7 +9248,6 @@ zpool_do_status(int argc, char **argv)

 		if (cb.vcdl != NULL)
 			free_vdev_cmd_data_list(cb.vcdl);
-
 		if (argc == 0 && cb.cb_count == 0)
 			(void) fprintf(stderr, gettext("no pools available\n"));
 		else if (cb.cb_explain && cb.cb_first && cb.cb_allpools)
@@ -10638,11 +10683,10 @@ found:
 		}
 	} else {
 		/*
-		 * The first arg isn't a pool name,
+		 * The first arg isn't the name of a valid pool.
 		 */
-		fprintf(stderr, gettext("missing pool name.\n"));
-		fprintf(stderr, "\n");
-		usage(B_FALSE);
+		fprintf(stderr, gettext("Cannot get properties of %s: "
+		    "no such pool available.\n"), argv[0]);
 		return (1);
 	}

@@ -20,7 +20,7 @@
 */
 /*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2018 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2013 Steven Hartland. All rights reserved.
 * Copyright (c) 2014 Integros [integros.com]
@@ -3270,7 +3270,7 @@ ztest_vdev_add_remove(ztest_ds_t *zd, uint64_t id)
 		    "log" : NULL, ztest_opts.zo_raid_children, zs->zs_mirrors,
 		    1);

-		error = spa_vdev_add(spa, nvroot);
+		error = spa_vdev_add(spa, nvroot, B_FALSE);
 		fnvlist_free(nvroot);

 		switch (error) {
@@ -3332,7 +3332,7 @@ ztest_vdev_class_add(ztest_ds_t *zd, uint64_t id)
 	nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0,
 	    class, ztest_opts.zo_raid_children, zs->zs_mirrors, 1);

-	error = spa_vdev_add(spa, nvroot);
+	error = spa_vdev_add(spa, nvroot, B_FALSE);
 	fnvlist_free(nvroot);

 	if (error == ENOSPC)
@@ -3439,7 +3439,7 @@ ztest_vdev_aux_add_remove(ztest_ds_t *zd, uint64_t id)
 		 */
 		nvlist_t *nvroot = make_vdev_root(NULL, aux, NULL,
 		    (ztest_opts.zo_vdev_size * 5) / 4, 0, NULL, 0, 0, 1);
-		error = spa_vdev_add(spa, nvroot);
+		error = spa_vdev_add(spa, nvroot, B_FALSE);

 		switch (error) {
 		case 0:
@@ -18,6 +18,7 @@ subst_sed_cmd = \
 	-e 's|@ASAN_ENABLED[@]|$(ASAN_ENABLED)|g' \
 	-e 's|@DEFAULT_INIT_NFS_SERVER[@]|$(DEFAULT_INIT_NFS_SERVER)|g' \
 	-e 's|@DEFAULT_INIT_SHELL[@]|$(DEFAULT_INIT_SHELL)|g' \
+	-e 's|@IS_SYSV_RC[@]|$(IS_SYSV_RC)|g' \
 	-e 's|@LIBFETCH_DYNAMIC[@]|$(LIBFETCH_DYNAMIC)|g' \
 	-e 's|@LIBFETCH_SONAME[@]|$(LIBFETCH_SONAME)|g' \
 	-e 's|@PYTHON[@]|$(PYTHON)|g' \
@@ -43,4 +44,4 @@ SUBSTFILES =
 CLEANFILES += $(SUBSTFILES)
 dist_noinst_DATA += $(SUBSTFILES:=.in)

-$(call SUBST,%,)
+$(SUBSTFILES): $(call SUBST,%,)
@@ -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
 			])
 		])
@@ -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

 	#
@@ -377,6 +377,14 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLK_MQ], [
 		(void) blk_mq_alloc_tag_set(&tag_set);
 		return BLK_STS_OK;
 	], [])
+	ZFS_LINUX_TEST_SRC([blk_mq_rq_hctx], [
+		#include <linux/blk-mq.h>
+		#include <linux/blkdev.h>
+	], [
+		struct request rq = {0};
+		struct blk_mq_hw_ctx *hctx = NULL;
+		rq.mq_hctx = hctx;
+	], [])
 ])

 AC_DEFUN([ZFS_AC_KERNEL_BLK_MQ], [
@@ -384,6 +392,13 @@ AC_DEFUN([ZFS_AC_KERNEL_BLK_MQ], [
 	ZFS_LINUX_TEST_RESULT([blk_mq], [
 		AC_MSG_RESULT(yes)
 		AC_DEFINE(HAVE_BLK_MQ, 1, [block multiqueue is available])
+		AC_MSG_CHECKING([whether block multiqueue hardware context is cached in struct request])
+		ZFS_LINUX_TEST_RESULT([blk_mq_rq_hctx], [
+			AC_MSG_RESULT(yes)
+			AC_DEFINE(HAVE_BLK_MQ_RQ_HCTX, 1, [block multiqueue hardware context is cached in struct request])
+		], [
+			AC_MSG_RESULT(no)
+		])
 	], [
 		AC_MSG_RESULT(no)
 	])
@@ -54,6 +54,26 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_OPEN_BY_PATH], [
 	])
 ])

+dnl #
+dnl # 6.9.x API change
+dnl # bdev_file_open_by_path() replaced bdev_open_by_path(),
+dnl # and returns struct file*
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_BDEV_FILE_OPEN_BY_PATH], [
+	ZFS_LINUX_TEST_SRC([bdev_file_open_by_path], [
+		#include <linux/fs.h>
+		#include <linux/blkdev.h>
+	], [
+		struct file *file __attribute__ ((unused)) = NULL;
+		const char *path = "path";
+		fmode_t mode = 0;
+		void *holder = NULL;
+		struct blk_holder_ops h;
+
+		file = bdev_file_open_by_path(path, mode, holder, &h);
+	])
+])
+
 AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_GET_BY_PATH], [
 	AC_MSG_CHECKING([whether blkdev_get_by_path() exists and takes 3 args])
 	ZFS_LINUX_TEST_RESULT([blkdev_get_by_path], [
@@ -73,7 +93,16 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_GET_BY_PATH], [
 					[bdev_open_by_path() exists])
 				AC_MSG_RESULT(yes)
 			], [
-				ZFS_LINUX_TEST_ERROR([blkdev_get_by_path()])
+				AC_MSG_RESULT(no)
+				AC_MSG_CHECKING([whether bdev_file_open_by_path() exists])
+				ZFS_LINUX_TEST_RESULT([bdev_file_open_by_path], [
+					AC_DEFINE(HAVE_BDEV_FILE_OPEN_BY_PATH, 1,
+						[bdev_file_open_by_path() exists])
+					AC_MSG_RESULT(yes)
+				], [
+					AC_MSG_RESULT(no)
+					ZFS_LINUX_TEST_ERROR([blkdev_get_by_path()])
+				])
 			])
 		])
 	])
@@ -149,10 +178,19 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_RELEASE], [
 	])
 ])

+dnl #
+dnl # 6.9.x API change
+dnl #
+dnl # bdev_release() now private, but because bdev_file_open_by_path() returns
+dnl # struct file*, we can just use fput(). So the blkdev_put test no longer
+dnl # fails if not found.
+dnl #
+
 AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_PUT], [
 	AC_MSG_CHECKING([whether blkdev_put() exists])
 	ZFS_LINUX_TEST_RESULT([blkdev_put], [
 		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BLKDEV_PUT, 1, [blkdev_put() exists])
 	], [
 		AC_MSG_RESULT(no)
 		AC_MSG_CHECKING([whether blkdev_put() accepts void* as arg 2])
@@ -168,7 +206,7 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_PUT], [
 				AC_DEFINE(HAVE_BDEV_RELEASE, 1,
 					[bdev_release() exists])
 			], [
-				ZFS_LINUX_TEST_ERROR([blkdev_put()])
+				AC_MSG_RESULT(no)
 			])
 		])
 	])
@@ -523,12 +561,29 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEVNAME], [
 ])

 dnl #
-dnl # 5.19 API: blkdev_issue_secure_erase()
-dnl # 4.7  API: __blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE)
-dnl # 3.10 API: blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE)
+dnl # TRIM support: discard and secure erase. We make use of asynchronous
+dnl #               functions when available.
 dnl #
-AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE], [
-	ZFS_LINUX_TEST_SRC([blkdev_issue_secure_erase], [
+dnl # 3.10:
+dnl #   sync discard:  blkdev_issue_discard(..., 0)
+dnl #   sync erase:    blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE)
+dnl #   async discard: [not available]
+dnl #   async erase:   [not available]
+dnl #
+dnl # 4.7:
+dnl #   sync discard:  blkdev_issue_discard(..., 0)
+dnl #   sync erase:    blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE)
+dnl #   async discard: __blkdev_issue_discard(..., 0)
+dnl #   async erase:   __blkdev_issue_discard(..., BLKDEV_DISCARD_SECURE)
+dnl #
+dnl # 5.19:
+dnl #   sync discard:  blkdev_issue_discard(...)
+dnl #   sync erase:    blkdev_issue_secure_erase(...)
+dnl #   async discard: __blkdev_issue_discard(...)
+dnl #   async erase:   [not available]
+dnl #
+AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_DISCARD], [
+	ZFS_LINUX_TEST_SRC([blkdev_issue_discard_noflags], [
 		#include <linux/blkdev.h>
 	],[
 		struct block_device *bdev = NULL;
@@ -536,10 +591,33 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE], [
 		sector_t nr_sects = 0;
 		int error __attribute__ ((unused));

-		error = blkdev_issue_secure_erase(bdev,
+		error = blkdev_issue_discard(bdev,
 		    sector, nr_sects, GFP_KERNEL);
 	])
+	ZFS_LINUX_TEST_SRC([blkdev_issue_discard_flags], [
+		#include <linux/blkdev.h>
+	],[
+		struct block_device *bdev = NULL;
+		sector_t sector = 0;
+		sector_t nr_sects = 0;
+		unsigned long flags = 0;
+		int error __attribute__ ((unused));

+		error = blkdev_issue_discard(bdev,
+		    sector, nr_sects, GFP_KERNEL, flags);
+	])
+	ZFS_LINUX_TEST_SRC([blkdev_issue_discard_async_noflags], [
+		#include <linux/blkdev.h>
+	],[
+		struct block_device *bdev = NULL;
+		sector_t sector = 0;
+		sector_t nr_sects = 0;
+		struct bio *biop = NULL;
+		int error __attribute__ ((unused));
+
+		error = __blkdev_issue_discard(bdev,
+		    sector, nr_sects, GFP_KERNEL, &biop);
+	])
 	ZFS_LINUX_TEST_SRC([blkdev_issue_discard_async_flags], [
 		#include <linux/blkdev.h>
 	],[
@@ -553,22 +631,52 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE], [
 		error = __blkdev_issue_discard(bdev,
 		    sector, nr_sects, GFP_KERNEL, flags, &biop);
 	])
-
-	ZFS_LINUX_TEST_SRC([blkdev_issue_discard_flags], [
+	ZFS_LINUX_TEST_SRC([blkdev_issue_secure_erase], [
 		#include <linux/blkdev.h>
 	],[
 		struct block_device *bdev = NULL;
 		sector_t sector = 0;
 		sector_t nr_sects = 0;
-		unsigned long flags = 0;
 		int error __attribute__ ((unused));

-		error = blkdev_issue_discard(bdev,
-		    sector, nr_sects, GFP_KERNEL, flags);
+		error = blkdev_issue_secure_erase(bdev,
+		    sector, nr_sects, GFP_KERNEL);
 	])
 ])

-AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE], [
+AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_ISSUE_DISCARD], [
+	AC_MSG_CHECKING([whether blkdev_issue_discard() is available])
+	ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_noflags], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD_NOFLAGS, 1,
+		    [blkdev_issue_discard() is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
+	AC_MSG_CHECKING([whether blkdev_issue_discard(flags) is available])
+	ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_flags], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD_FLAGS, 1,
+		    [blkdev_issue_discard(flags) is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
+	AC_MSG_CHECKING([whether __blkdev_issue_discard() is available])
+	ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_async_noflags], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD_ASYNC_NOFLAGS, 1,
+		    [__blkdev_issue_discard() is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
+	AC_MSG_CHECKING([whether __blkdev_issue_discard(flags) is available])
+	ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_async_flags], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD_ASYNC_FLAGS, 1,
+		    [__blkdev_issue_discard(flags) is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
 	AC_MSG_CHECKING([whether blkdev_issue_secure_erase() is available])
 	ZFS_LINUX_TEST_RESULT([blkdev_issue_secure_erase], [
 		AC_MSG_RESULT(yes)
@@ -576,24 +684,6 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE], [
 		    [blkdev_issue_secure_erase() is available])
 	],[
 		AC_MSG_RESULT(no)
-
-		AC_MSG_CHECKING([whether __blkdev_issue_discard() is available])
-		ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_async_flags], [
-			AC_MSG_RESULT(yes)
-			AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD_ASYNC, 1,
-			    [__blkdev_issue_discard() is available])
-		],[
-			AC_MSG_RESULT(no)
-
-			AC_MSG_CHECKING([whether blkdev_issue_discard() is available])
-			ZFS_LINUX_TEST_RESULT([blkdev_issue_discard_flags], [
-				AC_MSG_RESULT(yes)
-				AC_DEFINE(HAVE_BLKDEV_ISSUE_DISCARD, 1,
-					[blkdev_issue_discard() is available])
-			],[
-				ZFS_LINUX_TEST_ERROR([blkdev_issue_discard()])
-			])
-		])
 	])
 ])

@@ -645,6 +735,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV], [
 	ZFS_AC_KERNEL_SRC_BLKDEV_GET_BY_PATH
 	ZFS_AC_KERNEL_SRC_BLKDEV_GET_BY_PATH_4ARG
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_OPEN_BY_PATH
+	ZFS_AC_KERNEL_SRC_BDEV_FILE_OPEN_BY_PATH
 	ZFS_AC_KERNEL_SRC_BLKDEV_PUT
 	ZFS_AC_KERNEL_SRC_BLKDEV_PUT_HOLDER
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_RELEASE
@@ -657,7 +748,7 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV], [
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_CHECK_MEDIA_CHANGE
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_WHOLE
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEVNAME
-	ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_SECURE_ERASE
+	ZFS_AC_KERNEL_SRC_BLKDEV_ISSUE_DISCARD
 	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_KOBJ
 	ZFS_AC_KERNEL_SRC_BLKDEV_PART_TO_DEV
 	ZFS_AC_KERNEL_SRC_BLKDEV_DISK_CHECK_MEDIA_CHANGE
@@ -678,7 +769,7 @@ AC_DEFUN([ZFS_AC_KERNEL_BLKDEV], [
 	ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE
 	ZFS_AC_KERNEL_BLKDEV_BDEVNAME
 	ZFS_AC_KERNEL_BLKDEV_GET_ERESTARTSYS
-	ZFS_AC_KERNEL_BLKDEV_ISSUE_SECURE_ERASE
+	ZFS_AC_KERNEL_BLKDEV_ISSUE_DISCARD
 	ZFS_AC_KERNEL_BLKDEV_BDEV_KOBJ
 	ZFS_AC_KERNEL_BLKDEV_PART_TO_DEV
 	ZFS_AC_KERNEL_BLKDEV_DISK_CHECK_MEDIA_CHANGE
@@ -4,6 +4,7 @@ dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_FILEMAP], [
 	ZFS_LINUX_TEST_SRC([filemap_range_has_page], [
 		#include <linux/fs.h>
+		#include <linux/pagemap.h>
 	],[
 		struct address_space *mapping = NULL;
 		loff_t lstart = 0;
@@ -50,6 +50,14 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_MAKE_REQUEST_FN], [
 		disk = blk_alloc_disk(NUMA_NO_NODE);
 	])

+	ZFS_LINUX_TEST_SRC([blk_alloc_disk_2arg], [
+		#include <linux/blkdev.h>
+	],[
+		struct queue_limits *lim = NULL;
+		struct gendisk *disk  __attribute__ ((unused));
+		disk = blk_alloc_disk(lim, NUMA_NO_NODE);
+	])
+
 	ZFS_LINUX_TEST_SRC([blk_cleanup_disk], [
 		#include <linux/blkdev.h>
 	],[
@@ -96,6 +104,31 @@ AC_DEFUN([ZFS_AC_KERNEL_MAKE_REQUEST_FN], [
 		], [
 			AC_MSG_RESULT(no)
 		])
+
+		dnl #
+		dnl # Linux 6.9 API Change:
+		dnl # blk_alloc_queue() takes a nullable queue_limits arg.
+		dnl #
+		AC_MSG_CHECKING([whether blk_alloc_disk() exists and takes 2 args])
+		ZFS_LINUX_TEST_RESULT([blk_alloc_disk_2arg], [
+			AC_MSG_RESULT(yes)
+			AC_DEFINE([HAVE_BLK_ALLOC_DISK_2ARG], 1, [blk_alloc_disk() exists and takes 2 args])
+
+			dnl #
+			dnl # 5.20 API change,
+			dnl # Removed blk_cleanup_disk(), put_disk() should be used.
+			dnl #
+			AC_MSG_CHECKING([whether blk_cleanup_disk() exists])
+			ZFS_LINUX_TEST_RESULT([blk_cleanup_disk], [
+				AC_MSG_RESULT(yes)
+				AC_DEFINE([HAVE_BLK_CLEANUP_DISK], 1,
+				    [blk_cleanup_disk() exists])
+			], [
+				AC_MSG_RESULT(no)
+			])
+		], [
+			AC_MSG_RESULT(no)
+		])
 	],[
 		AC_MSG_RESULT(no)

@@ -0,0 +1,17 @@
+AC_DEFUN([ZFS_AC_KERNEL_SRC_MM_PAGE_SIZE], [
+	ZFS_LINUX_TEST_SRC([page_size], [
+		#include <linux/mm.h>
+	],[
+		unsigned long s;
+		s = page_size(NULL);
+	])
+])
+AC_DEFUN([ZFS_AC_KERNEL_MM_PAGE_SIZE], [
+	AC_MSG_CHECKING([whether page_size() is available])
+	ZFS_LINUX_TEST_RESULT([page_size], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_MM_PAGE_SIZE, 1, [page_size() is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
+])
@@ -16,6 +16,9 @@ dnl #
 dnl # 5.3: VFS copy_file_range() expected to do its own fallback,
 dnl #      generic_copy_file_range() added to support it
 dnl #
+dnl # 6.8: generic_copy_file_range() removed, replaced by
+dnl #      splice_copy_file_range()
+dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_COPY_FILE_RANGE], [
 	ZFS_LINUX_TEST_SRC([vfs_copy_file_range], [
 		#include <linux/fs.h>
@@ -72,6 +75,30 @@ AC_DEFUN([ZFS_AC_KERNEL_VFS_GENERIC_COPY_FILE_RANGE], [
 	])
 ])

+AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_SPLICE_COPY_FILE_RANGE], [
+	ZFS_LINUX_TEST_SRC([splice_copy_file_range], [
+		#include <linux/splice.h>
+	], [
+		struct file *src_file __attribute__ ((unused)) = NULL;
+		loff_t src_off __attribute__ ((unused)) = 0;
+		struct file *dst_file __attribute__ ((unused)) = NULL;
+		loff_t dst_off __attribute__ ((unused)) = 0;
+		size_t len __attribute__ ((unused)) = 0;
+		splice_copy_file_range(src_file, src_off, dst_file, dst_off,
+		    len);
+	])
+])
+AC_DEFUN([ZFS_AC_KERNEL_VFS_SPLICE_COPY_FILE_RANGE], [
+	AC_MSG_CHECKING([whether splice_copy_file_range() is available])
+	ZFS_LINUX_TEST_RESULT([splice_copy_file_range], [
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_VFS_SPLICE_COPY_FILE_RANGE, 1,
+		    [splice_copy_file_range() is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
+])
+
 AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_CLONE_FILE_RANGE], [
 	ZFS_LINUX_TEST_SRC([vfs_clone_file_range], [
 		#include <linux/fs.h>
@@ -118,6 +118,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
 	ZFS_AC_KERNEL_SRC_VFS_IOV_ITER
 	ZFS_AC_KERNEL_SRC_VFS_COPY_FILE_RANGE
 	ZFS_AC_KERNEL_SRC_VFS_GENERIC_COPY_FILE_RANGE
+	ZFS_AC_KERNEL_SRC_VFS_SPLICE_COPY_FILE_RANGE
 	ZFS_AC_KERNEL_SRC_VFS_REMAP_FILE_RANGE
 	ZFS_AC_KERNEL_SRC_VFS_CLONE_FILE_RANGE
 	ZFS_AC_KERNEL_SRC_VFS_DEDUPE_FILE_RANGE
@@ -166,6 +167,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
 	ZFS_AC_KERNEL_SRC_REGISTER_SYSCTL_TABLE
 	ZFS_AC_KERNEL_SRC_COPY_SPLICE_READ
 	ZFS_AC_KERNEL_SRC_SYNC_BDEV
+	ZFS_AC_KERNEL_SRC_MM_PAGE_SIZE
 	case "$host_cpu" in
 		powerpc*)
 			ZFS_AC_KERNEL_SRC_CPU_HAS_FEATURE
@@ -266,6 +268,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
 	ZFS_AC_KERNEL_VFS_IOV_ITER
 	ZFS_AC_KERNEL_VFS_COPY_FILE_RANGE
 	ZFS_AC_KERNEL_VFS_GENERIC_COPY_FILE_RANGE
+	ZFS_AC_KERNEL_VFS_SPLICE_COPY_FILE_RANGE
 	ZFS_AC_KERNEL_VFS_REMAP_FILE_RANGE
 	ZFS_AC_KERNEL_VFS_CLONE_FILE_RANGE
 	ZFS_AC_KERNEL_VFS_DEDUPE_FILE_RANGE
@@ -314,6 +317,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
 	ZFS_AC_KERNEL_REGISTER_SYSCTL_TABLE
 	ZFS_AC_KERNEL_COPY_SPLICE_READ
 	ZFS_AC_KERNEL_SYNC_BDEV
+	ZFS_AC_KERNEL_MM_PAGE_SIZE
 	case "$host_cpu" in
 		powerpc*)
 			ZFS_AC_KERNEL_CPU_HAS_FEATURE
@@ -578,13 +578,15 @@ AC_DEFUN([ZFS_AC_DEFAULT_PACKAGE], [

 	AC_MSG_CHECKING([default shell])
 	case "$VENDOR" in
-		gentoo)     DEFAULT_INIT_SHELL="/sbin/openrc-run";;
-		alpine)     DEFAULT_INIT_SHELL="/sbin/openrc-run";;
-		*)          DEFAULT_INIT_SHELL="/bin/sh"         ;;
+		gentoo|alpine)	DEFAULT_INIT_SHELL=/sbin/openrc-run
+				IS_SYSV_RC=false	;;
+		*)		DEFAULT_INIT_SHELL=/bin/sh
+				IS_SYSV_RC=true		;;
 	esac

 	AC_MSG_RESULT([$DEFAULT_INIT_SHELL])
 	AC_SUBST(DEFAULT_INIT_SHELL)
+	AC_SUBST(IS_SYSV_RC)

 	AC_MSG_CHECKING([default nfs server init script])
 	AS_IF([test "$VENDOR" = "debian"],
@@ -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}
@@ -7,11 +7,7 @@ DESCRIPTION

  They have been tested successfully on:

-    * Debian GNU/Linux Wheezy
-    * Debian GNU/Linux Jessie
-    * Ubuntu Trusty
-    * CentOS 6.0
-    * CentOS 6.6
+    * Debian GNU/Linux Bookworm
    * Gentoo

 SUPPORT
@@ -307,7 +307,7 @@ do_start()

 # ----------------------------------------------------

-if [ ! -e /sbin/openrc-run ]
+if @IS_SYSV_RC@
 then
 	case "$1" in
 		start)
@@ -104,7 +104,7 @@ do_stop()

 # ----------------------------------------------------

-if [ ! -e /sbin/openrc-run ]
+if @IS_SYSV_RC@
 then
 	case "$1" in
 		start)
@@ -114,7 +114,7 @@ do_stop()

 # ----------------------------------------------------

-if [ ! -e /sbin/openrc-run ]
+if @IS_SYSV_RC@
 then
 	case "$1" in
 		start)
@@ -57,7 +57,8 @@ do_stop()

 # ----------------------------------------------------

-if [ ! -e /sbin/openrc-run ]; then
+if @IS_SYSV_RC@
+then
 	case "$1" in
 		start)
 			do_start
@@ -93,7 +93,8 @@ do_reload()

 # ----------------------------------------------------

-if [ ! -e /sbin/openrc-run ]; then
+if @IS_SYSV_RC@
+then
 	case "$1" in
 		start)
 			do_start
@@ -21,7 +21,7 @@

 /*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2022 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright Joyent, Inc.
 * Copyright (c) 2013 Steven Hartland. All rights reserved.
 * Copyright (c) 2016, Intel Corporation.
@@ -157,6 +157,8 @@ typedef enum zfs_error {
 	EZFS_CKSUM,		/* insufficient replicas */
 	EZFS_RESUME_EXISTS,	/* Resume on existing dataset without force */
 	EZFS_SHAREFAILED,	/* filesystem share failed */
+	EZFS_RAIDZ_EXPAND_IN_PROGRESS,	/* a raidz is currently expanding */
+	EZFS_ASHIFT_MISMATCH,   /* can't add vdevs with different ashifts */
 	EZFS_UNKNOWN
 } zfs_error_t;

@@ -260,7 +262,7 @@ _LIBZFS_H boolean_t zpool_skip_pool(const char *);
 _LIBZFS_H int zpool_create(libzfs_handle_t *, const char *, nvlist_t *,
    nvlist_t *, nvlist_t *);
 _LIBZFS_H int zpool_destroy(zpool_handle_t *, const char *);
-_LIBZFS_H int zpool_add(zpool_handle_t *, nvlist_t *);
+_LIBZFS_H int zpool_add(zpool_handle_t *, nvlist_t *, boolean_t check_ashift);

 typedef struct splitflags {
 	/* do not split, but return the config that would be split off */
@@ -4,8 +4,6 @@ noinst_HEADERS = \
 	\
 	%D%/spl/acl/acl_common.h \
 	\
-	%D%/spl/rpc/xdr.h \
-	\
 	%D%/spl/sys/ia32/asm_linkage.h \
 	\
 	%D%/spl/sys/acl.h \
@@ -80,7 +78,9 @@ noinst_HEADERS = \
 	%D%/spl/sys/zmod.h \
 	%D%/spl/sys/zone.h \
 	\
+	%D%/zfs/sys/arc_os.h \
 	%D%/zfs/sys/freebsd_crypto.h \
+	%D%/zfs/sys/freebsd_event.h \
 	%D%/zfs/sys/vdev_os.h \
 	%D%/zfs/sys/zfs_bootenv_os.h \
 	%D%/zfs/sys/zfs_context_os.h \
@@ -1,71 +0,0 @@
-/*
- * Sun RPC is a product of Sun Microsystems, Inc. and is provided for
- * unrestricted use provided that this legend is included on all tape
- * media and as a part of the software program in whole or part.  Users
- * may copy or modify Sun RPC without charge, but are not authorized
- * to license or distribute it to anyone else except as part of a product or
- * program developed by the user.
- *
- * SUN RPC IS PROVIDED AS IS WITH NO WARRANTIES OF ANY KIND INCLUDING THE
- * WARRANTIES OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR
- * PURPOSE, OR ARISING FROM A COURSE OF DEALING, USAGE OR TRADE PRACTICE.
- *
- * Sun RPC is provided with no support and without any obligation on the
- * part of Sun Microsystems, Inc. to assist in its use, correction,
- * modification or enhancement.
- *
- * SUN MICROSYSTEMS, INC. SHALL HAVE NO LIABILITY WITH RESPECT TO THE
- * INFRINGEMENT OF COPYRIGHTS, TRADE SECRETS OR ANY PATENTS BY SUN RPC
- * OR ANY PART THEREOF.
- *
- * In no event will Sun Microsystems, Inc. be liable for any lost revenue
- * or profits or other special, indirect and consequential damages, even if
- * Sun has been advised of the possibility of such damages.
- *
- * Sun Microsystems, Inc.
- * 2550 Garcia Avenue
- * Mountain View, California  94043
- */
-
-#ifndef	_OPENSOLARIS_RPC_XDR_H_
-#define	_OPENSOLARIS_RPC_XDR_H_
-
-#include <rpc/types.h>
-#include_next <rpc/xdr.h>
-
-#if !defined(_KERNEL) && !defined(_STANDALONE)
-
-#include <assert.h>
-
-/*
- * Taken from sys/xdr/xdr_mem.c.
- *
- * FreeBSD's userland XDR doesn't implement control method (only the kernel),
- * but OpenSolaris nvpair still depend on it, so we have to implement it here.
- */
-static __inline bool_t
-xdrmem_control(XDR *xdrs, int request, void *info)
-{
-	xdr_bytesrec *xptr;
-
-	switch (request) {
-	case XDR_GET_BYTES_AVAIL:
-		xptr = (xdr_bytesrec *)info;
-		xptr->xc_is_last_record = TRUE;
-		xptr->xc_num_avail = xdrs->x_handy;
-		return (TRUE);
-	default:
-		assert(!"unexpected request");
-	}
-	return (FALSE);
-}
-
-#undef XDR_CONTROL
-#define	XDR_CONTROL(xdrs, req, op)					\
-	(((xdrs)->x_ops->x_control == NULL) ?				\
-	    xdrmem_control((xdrs), (req), (op)) :			\
-	    (*(xdrs)->x_ops->x_control)(xdrs, req, op))
-
-#endif	/* !_KERNEL && !_STANDALONE */
-
-#endif	/* !_OPENSOLARIS_RPC_XDR_H_ */
@@ -39,12 +39,14 @@
 * ASSERT3U()	- Assert unsigned X OP Y is true, if not panic.
 * ASSERT3P()	- Assert pointer X OP Y is true, if not panic.
 * ASSERT0()	- Assert value is zero, if not panic.
+ * ASSERT0P()	- Assert pointer is null, if not panic.
 * VERIFY()	- Verify X is true, if not panic.
 * VERIFY3B()	- Verify boolean X OP Y is true, if not panic.
 * VERIFY3S()	- Verify signed X OP Y is true, if not panic.
 * VERIFY3U()	- Verify unsigned X OP Y is true, if not panic.
 * VERIFY3P()	- Verify pointer X OP Y is true, if not panic.
 * VERIFY0()	- Verify value is zero, if not panic.
+ * VERIFY0P()	- Verify pointer is null, if not panic.
 */

 #ifndef _SPL_DEBUG_H
@@ -89,8 +91,8 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%d " #OP " %d)\n",				\
-		    (boolean_t)(_verify3_left),				\
-		    (boolean_t)(_verify3_right));			\
+		    (boolean_t)_verify3_left,				\
+		    (boolean_t)_verify3_right);				\
 	} while (0)

 #define	VERIFY3S(LEFT, OP, RIGHT)	do {				\
@@ -100,8 +102,8 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%lld " #OP " %lld)\n",			\
-		    (long long) (_verify3_left),			\
-		    (long long) (_verify3_right));			\
+		    (long long)_verify3_left,				\
+		    (long long)_verify3_right);				\
 	} while (0)

 #define	VERIFY3U(LEFT, OP, RIGHT)	do {				\
@@ -111,8 +113,8 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%llu " #OP " %llu)\n",			\
-		    (unsigned long long) (_verify3_left),		\
-		    (unsigned long long) (_verify3_right));		\
+		    (unsigned long long)_verify3_left,			\
+		    (unsigned long long)_verify3_right);		\
 	} while (0)

 #define	VERIFY3P(LEFT, OP, RIGHT)	do {				\
@@ -121,19 +123,27 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		if (unlikely(!(_verify3_left OP _verify3_right)))	\
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
-		    "failed (%px " #OP " %px)\n",			\
-		    (void *) (_verify3_left),				\
-		    (void *) (_verify3_right));				\
+		    "failed (%p " #OP " %p)\n",				\
+		    (void *)_verify3_left,				\
+		    (void *)_verify3_right);				\
 	} while (0)

 #define	VERIFY0(RIGHT)	do {						\
-		const int64_t _verify3_left = (int64_t)(0);		\
-		const int64_t _verify3_right = (int64_t)(RIGHT);	\
-		if (unlikely(!(_verify3_left == _verify3_right)))	\
+		const int64_t _verify0_right = (int64_t)(RIGHT);	\
+		if (unlikely(!(0 == _verify0_right)))			\
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
-		    "VERIFY0(0 == " #RIGHT ") "				\
+		    "VERIFY0(" #RIGHT ") "				\
 		    "failed (0 == %lld)\n",				\
-		    (long long) (_verify3_right));			\
+		    (long long)_verify0_right);				\
+	} while (0)
+
+#define	VERIFY0P(RIGHT)	do {						\
+		const uintptr_t _verify0_right = (uintptr_t)(RIGHT);	\
+		if (unlikely(!(0 == _verify0_right)))			\
+		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
+		    "VERIFY0P(" #RIGHT ") "				\
+		    "failed (NULL == %p)\n",				\
+		    (void *)_verify0_right);				\
 	} while (0)

 /*
@@ -151,6 +161,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 #define	ASSERT3P(x, y, z)						\
 	((void) sizeof ((uintptr_t)(x)), (void) sizeof ((uintptr_t)(z)))
 #define	ASSERT0(x)		((void) sizeof ((uintptr_t)(x)))
+#define	ASSERT0P(x)		((void) sizeof ((uintptr_t)(x)))
 #define	IMPLY(A, B)							\
 	((void) sizeof ((uintptr_t)(A)), (void) sizeof ((uintptr_t)(B)))
 #define	EQUIV(A, B)		\
@@ -166,6 +177,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 #define	ASSERT3U	VERIFY3U
 #define	ASSERT3P	VERIFY3P
 #define	ASSERT0		VERIFY0
+#define	ASSERT0P	VERIFY0P
 #define	ASSERT		VERIFY
 #define	IMPLY(A, B) \
 	((void)(likely((!(A)) || (B)) ||				\
@@ -5,6 +5,7 @@ kernel_linux_HEADERS = \
 	%D%/kernel/linux/compiler_compat.h \
 	%D%/kernel/linux/dcache_compat.h \
 	%D%/kernel/linux/kmap_compat.h \
+	%D%/kernel/linux/mm_compat.h \
 	%D%/kernel/linux/mod_compat.h \
 	%D%/kernel/linux/page_compat.h \
 	%D%/kernel/linux/percpu_compat.h \
@@ -46,6 +47,7 @@ kernel_sys_HEADERS = \

 kernel_spl_rpcdir = $(kerneldir)/spl/rpc
 kernel_spl_rpc_HEADERS = \
+	%D%/spl/rpc/types.h \
 	%D%/spl/rpc/xdr.h

 kernel_spl_sysdir = $(kerneldir)/spl/sys
@@ -563,9 +563,11 @@ static inline boolean_t
 bdev_discard_supported(struct block_device *bdev)
 {
 #if defined(HAVE_BDEV_MAX_DISCARD_SECTORS)
-	return (!!bdev_max_discard_sectors(bdev));
+	return (bdev_max_discard_sectors(bdev) > 0 &&
+	    bdev_discard_granularity(bdev) > 0);
 #elif defined(HAVE_BLK_QUEUE_DISCARD)
-	return (!!blk_queue_discard(bdev_get_queue(bdev)));
+	return (blk_queue_discard(bdev_get_queue(bdev)) > 0 &&
+	    bdev_get_queue(bdev)->limits.discard_granularity > 0);
 #else
 #error "Unsupported kernel"
 #endif
@@ -0,0 +1,36 @@
+/*
+ * 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 (c) 2023, 2024, Klara Inc.
+ */
+
+#ifndef _ZFS_MM_COMPAT_H
+#define	_ZFS_MM_COMPAT_H
+
+#include <linux/mm.h>
+
+/* 5.4 introduced page_size(). Older kernels can use a trivial macro instead */
+#ifndef HAVE_MM_PAGE_SIZE
+#define	page_size(p) ((unsigned long)(PAGE_SIZE << compound_order(p)))
+#endif
+
+#endif /* _ZFS_MM_COMPAT_H */
@@ -68,6 +68,7 @@ enum scope_prefix_types {
 	zfs_trim,
 	zfs_txg,
 	zfs_vdev,
+	zfs_vdev_disk,
 	zfs_vdev_file,
 	zfs_vdev_mirror,
 	zfs_vnops,
@@ -0,0 +1,30 @@
+/*
+ *  Copyright (c) 2008 Sun Microsystems, Inc.
+ *  Written by Ricardo Correia <Ricardo.M.Correia@Sun.COM>
+ *
+ *  This file is part of the SPL, Solaris Porting Layer.
+ *
+ *  The SPL is free software; you can redistribute it and/or modify it
+ *  under the terms of the GNU General Public License as published by the
+ *  Free Software Foundation; either version 2 of the License, or (at your
+ *  option) any later version.
+ *
+ *  The SPL is distributed in the hope that it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ *  for more details.
+ *
+ *  You should have received a copy of the GNU General Public License along
+ *  with the SPL.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _SPL_RPC_TYPES_H
+#define	_SPL_RPC_TYPES_H
+
+#include <sys/types.h>
+
+/* Just enough to support rpc/xdr.h */
+
+typedef int bool_t;
+
+#endif /* SPL_RPC_TYPES_H */
@@ -23,8 +23,6 @@

 #include <sys/types.h>

-typedef int bool_t;
-
 /*
 * XDR enums and types.
 */
@@ -34,12 +34,14 @@
 * ASSERT3U()	- Assert unsigned X OP Y is true, if not panic.
 * ASSERT3P()	- Assert pointer X OP Y is true, if not panic.
 * ASSERT0()	- Assert value is zero, if not panic.
+ * ASSERT0P()	- Assert pointer is null, if not panic.
 * VERIFY()	- Verify X is true, if not panic.
 * VERIFY3B()	- Verify boolean X OP Y is true, if not panic.
 * VERIFY3S()	- Verify signed X OP Y is true, if not panic.
 * VERIFY3U()	- Verify unsigned X OP Y is true, if not panic.
 * VERIFY3P()	- Verify pointer X OP Y is true, if not panic.
 * VERIFY0()	- Verify value is zero, if not panic.
+ * VERIFY0P()	- Verify pointer is null, if not panic.
 */

 #ifndef _SPL_DEBUG_H
@@ -93,8 +95,8 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%d " #OP " %d)\n",				\
-		    (boolean_t)(_verify3_left),				\
-		    (boolean_t)(_verify3_right));			\
+		    (boolean_t)_verify3_left,				\
+		    (boolean_t)_verify3_right);				\
 	} while (0)

 #define	VERIFY3S(LEFT, OP, RIGHT)	do {				\
@@ -104,8 +106,8 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%lld " #OP " %lld)\n",			\
-		    (long long)(_verify3_left),				\
-		    (long long)(_verify3_right));			\
+		    (long long)_verify3_left,				\
+		    (long long)_verify3_right);				\
 	} while (0)

 #define	VERIFY3U(LEFT, OP, RIGHT)	do {				\
@@ -115,8 +117,8 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%llu " #OP " %llu)\n",			\
-		    (unsigned long long)(_verify3_left),		\
-		    (unsigned long long)(_verify3_right));		\
+		    (unsigned long long)_verify3_left,			\
+		    (unsigned long long)_verify3_right);		\
 	} while (0)

 #define	VERIFY3P(LEFT, OP, RIGHT)	do {				\
@@ -126,18 +128,26 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
 		    "VERIFY3(" #LEFT " "  #OP " "  #RIGHT ") "		\
 		    "failed (%px " #OP " %px)\n",			\
-		    (void *) (_verify3_left),				\
-		    (void *) (_verify3_right));				\
+		    (void *)_verify3_left,				\
+		    (void *)_verify3_right);				\
 	} while (0)

 #define	VERIFY0(RIGHT)	do {						\
-		const int64_t _verify3_left = (int64_t)(0);		\
-		const int64_t _verify3_right = (int64_t)(RIGHT);	\
-		if (unlikely(!(_verify3_left == _verify3_right)))	\
+		const int64_t _verify0_right = (int64_t)(RIGHT);	\
+		if (unlikely(!(0 == _verify0_right)))			\
 		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
-		    "VERIFY0(0 == " #RIGHT ") "				\
+		    "VERIFY0(" #RIGHT ") "				\
 		    "failed (0 == %lld)\n",				\
-		    (long long) (_verify3_right));			\
+		    (long long)_verify0_right);				\
+	} while (0)
+
+#define	VERIFY0P(RIGHT)	do {						\
+		const uintptr_t _verify0_right = (uintptr_t)(RIGHT);	\
+		if (unlikely(!(0 == _verify0_right)))			\
+		    spl_panic(__FILE__, __FUNCTION__, __LINE__,		\
+		    "VERIFY0P(" #RIGHT ") "				\
+		    "failed (NULL == %px)\n",				\
+		    (void *)_verify0_right);				\
 	} while (0)

 #define	VERIFY_IMPLY(A, B) \
@@ -165,6 +175,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 #define	ASSERT3P(x, y, z)						\
 	((void) sizeof ((uintptr_t)(x)), (void) sizeof ((uintptr_t)(z)))
 #define	ASSERT0(x)		((void) sizeof ((uintptr_t)(x)))
+#define	ASSERT0P(x)		((void) sizeof ((uintptr_t)(x)))
 #define	IMPLY(A, B)							\
 	((void) sizeof ((uintptr_t)(A)), (void) sizeof ((uintptr_t)(B)))
 #define	EQUIV(A, B)		\
@@ -180,6 +191,7 @@ spl_assert(const char *buf, const char *file, const char *func, int line)
 #define	ASSERT3U	VERIFY3U
 #define	ASSERT3P	VERIFY3P
 #define	ASSERT0		VERIFY0
+#define	ASSERT0P	VERIFY0P
 #define	ASSERT		VERIFY
 #define	IMPLY		VERIFY_IMPLY
 #define	EQUIV		VERIFY_EQUIV
@@ -104,7 +104,7 @@ typedef struct taskq {
 	/* list node for the cpu hotplug callback */
 	struct hlist_node	tq_hp_cb_node;
 	boolean_t		tq_hp_support;
-	unsigned long		lastshouldstop; /* when to purge dynamic */
+	unsigned long		lastspawnstop;	/* when to purge dynamic */
 } taskq_t;

 typedef struct taskq_ent {
@@ -51,7 +51,9 @@
 		__field(uint64_t,	zl_parse_lr_seq)		    \
 		__field(uint64_t,	zl_parse_blk_count)		    \
 		__field(uint64_t,	zl_parse_lr_count)		    \
-		__field(uint64_t,	zl_cur_used)			    \
+		__field(uint64_t,	zl_cur_size)			    \
+		__field(uint64_t,	zl_cur_left)			    \
+		__field(uint64_t,	zl_cur_max)			    \
 		__field(clock_t,	zl_replay_time)			    \
 		__field(uint64_t,	zl_replay_blks)

@@ -72,7 +74,9 @@
 		__entry->zl_parse_lr_seq	= zilog->zl_parse_lr_seq;   \
 		__entry->zl_parse_blk_count	= zilog->zl_parse_blk_count;\
 		__entry->zl_parse_lr_count	= zilog->zl_parse_lr_count; \
-		__entry->zl_cur_used	= zilog->zl_cur_used;		    \
+		__entry->zl_cur_size	= zilog->zl_cur_size;		    \
+		__entry->zl_cur_left	= zilog->zl_cur_left;		    \
+		__entry->zl_cur_max	= zilog->zl_cur_max;		    \
 		__entry->zl_replay_time	= zilog->zl_replay_time;	    \
 		__entry->zl_replay_blks	= zilog->zl_replay_blks;

@@ -82,7 +86,8 @@
 	"replay %u stop_sync %u logbias %u sync %u "			    \
 	"parse_error %u parse_blk_seq %llu parse_lr_seq %llu "		    \
 	"parse_blk_count %llu parse_lr_count %llu "			    \
-	"cur_used %llu replay_time %lu replay_blks %llu }"
+	"cur_size %llu cur_left %llu cur_max %llu replay_time %lu "	    \
+	"replay_blks %llu }"

 #define	ZILOG_TP_PRINTK_ARGS						    \
 	    __entry->zl_lr_seq, __entry->zl_commit_lr_seq,		    \
@@ -92,7 +97,8 @@
 	    __entry->zl_stop_sync, __entry->zl_logbias, __entry->zl_sync,   \
 	    __entry->zl_parse_error, __entry->zl_parse_blk_seq,		    \
 	    __entry->zl_parse_lr_seq, __entry->zl_parse_blk_count,	    \
-	    __entry->zl_parse_lr_count, __entry->zl_cur_used,		    \
+	    __entry->zl_parse_lr_count, __entry->zl_cur_size,		    \
+	    __entry->zl_cur_left, __entry->zl_cur_max,			    \
 	    __entry->zl_replay_time, __entry->zl_replay_blks

 #define	ITX_TP_STRUCT_ENTRY						    \
@@ -79,6 +79,9 @@ typedef struct abd {

 typedef int abd_iter_func_t(void *buf, size_t len, void *priv);
 typedef int abd_iter_func2_t(void *bufa, void *bufb, size_t len, void *priv);
+#if defined(__linux__) && defined(_KERNEL)
+typedef int abd_iter_page_func_t(struct page *, size_t, size_t, void *);
+#endif

 extern int zfs_abd_scatter_enabled;

@@ -125,6 +128,10 @@ void abd_release_ownership_of_buf(abd_t *);
 int abd_iterate_func(abd_t *, size_t, size_t, abd_iter_func_t *, void *);
 int abd_iterate_func2(abd_t *, abd_t *, size_t, size_t, size_t,
    abd_iter_func2_t *, void *);
+#if defined(__linux__) && defined(_KERNEL)
+int abd_iterate_page_func(abd_t *, size_t, size_t, abd_iter_page_func_t *,
+    void *);
+#endif
 void abd_copy_off(abd_t *, abd_t *, size_t, size_t, size_t);
 void abd_copy_from_buf_off(abd_t *, const void *, size_t, size_t);
 void abd_copy_to_buf_off(void *, abd_t *, size_t, size_t);
@@ -213,6 +220,8 @@ void abd_fini(void);

 /*
 * Linux ABD bio functions
+ * Note: these are only needed to support vdev_classic. See comment in
+ * vdev_disk.c.
 */
 #if defined(__linux__) && defined(_KERNEL)
 unsigned int abd_bio_map_off(struct bio *, abd_t *, unsigned int, size_t);
@@ -21,6 +21,7 @@
 /*
 * Copyright (c) 2014 by Chunwei Chen. All rights reserved.
 * Copyright (c) 2016, 2019 by Delphix. All rights reserved.
+ * Copyright (c) 2023, 2024, Klara Inc.
 */

 #ifndef _ABD_IMPL_H
@@ -38,12 +39,30 @@ typedef enum abd_stats_op {
 	ABDSTAT_DECR  /* Decrease abdstat values */
 } abd_stats_op_t;

-struct scatterlist; /* forward declaration */
+/* forward declarations */
+struct scatterlist;
+struct page;

 struct abd_iter {
 	/* public interface */
-	void		*iter_mapaddr;	/* addr corresponding to iter_pos */
-	size_t		iter_mapsize;	/* length of data valid at mapaddr */
+	union {
+		/* for abd_iter_map()/abd_iter_unmap() */
+		struct {
+			/* addr corresponding to iter_pos */
+			void		*iter_mapaddr;
+			/* length of data valid at mapaddr */
+			size_t		iter_mapsize;
+		};
+		/* for abd_iter_page() */
+		struct {
+			/* current page */
+			struct page	*iter_page;
+			/* offset of data in page */
+			size_t		iter_page_doff;
+			/* size of data in page */
+			size_t		iter_page_dsize;
+		};
+	};

 	/* private */
 	abd_t		*iter_abd;	/* ABD being iterated through */
@@ -78,6 +97,7 @@ boolean_t abd_iter_at_end(struct abd_iter *);
 void abd_iter_advance(struct abd_iter *, size_t);
 void abd_iter_map(struct abd_iter *);
 void abd_iter_unmap(struct abd_iter *);
+void abd_iter_page(struct abd_iter *);

 /*
 * Helper macros
@@ -739,8 +739,6 @@ void *dmu_buf_remove_user(dmu_buf_t *db, dmu_buf_user_t *user);
 void *dmu_buf_get_user(dmu_buf_t *db);

 objset_t *dmu_buf_get_objset(dmu_buf_t *db);
-dnode_t *dmu_buf_dnode_enter(dmu_buf_t *db);
-void dmu_buf_dnode_exit(dmu_buf_t *db);

 /* Block until any in-progress dmu buf user evictions complete. */
 void dmu_buf_user_evict_wait(void);
@@ -889,6 +887,9 @@ extern uint_t zfs_max_recordsize;
 */
 void dmu_prefetch(objset_t *os, uint64_t object, int64_t level, uint64_t offset,
 	uint64_t len, enum zio_priority pri);
+void dmu_prefetch_by_dnode(dnode_t *dn, int64_t level, uint64_t offset,
+	uint64_t len, enum zio_priority pri);
+void dmu_prefetch_dnode(objset_t *os, uint64_t object, enum zio_priority pri);

 typedef struct dmu_object_info {
 	/* All sizes are in bytes unless otherwise indicated. */
@@ -132,6 +132,7 @@ struct objset {
 	zfs_logbias_op_t os_logbias;
 	zfs_cache_type_t os_primary_cache;
 	zfs_cache_type_t os_secondary_cache;
+	zfs_prefetch_type_t os_prefetch;
 	zfs_sync_type_t os_sync;
 	zfs_redundant_metadata_type_t os_redundant_metadata;
 	uint64_t os_recordsize;
@@ -45,18 +45,24 @@ typedef struct zfetch {
 	int		zf_numstreams;	/* number of zstream_t's */
 } zfetch_t;

+typedef struct zsrange {
+	uint16_t	start;
+	uint16_t	end;
+} zsrange_t;
+
+#define	ZFETCH_RANGES	9		/* Fits zstream_t into 128 bytes */
+
 typedef struct zstream {
+	list_node_t	zs_node;	/* link for zf_stream */
 	uint64_t	zs_blkid;	/* expect next access at this blkid */
+	uint_t		zs_atime;	/* time last prefetch issued */
+	zsrange_t	zs_ranges[ZFETCH_RANGES]; /* ranges from future */
 	unsigned int	zs_pf_dist;	/* data prefetch distance in bytes */
 	unsigned int	zs_ipf_dist;	/* L1 prefetch distance in bytes */
 	uint64_t	zs_pf_start;	/* first data block to prefetch */
 	uint64_t	zs_pf_end;	/* data block to prefetch up to */
 	uint64_t	zs_ipf_start;	/* first data block to prefetch L1 */
 	uint64_t	zs_ipf_end;	/* data block to prefetch L1 up to */
-
-	list_node_t	zs_node;	/* link for zf_stream */
-	hrtime_t	zs_atime;	/* time last prefetch issued */
-	zfetch_t	*zs_fetch;	/* parent fetch */
 	boolean_t	zs_missed;	/* stream saw cache misses */
 	boolean_t	zs_more;	/* need more distant prefetch */
 	zfs_refcount_t	zs_callers;	/* number of pending callers */
@@ -74,7 +80,7 @@ void		dmu_zfetch_init(zfetch_t *, struct dnode *);
 void		dmu_zfetch_fini(zfetch_t *);
 zstream_t	*dmu_zfetch_prepare(zfetch_t *, uint64_t, uint64_t, boolean_t,
    boolean_t);
-void		dmu_zfetch_run(zstream_t *, boolean_t, boolean_t);
+void		dmu_zfetch_run(zfetch_t *, zstream_t *, boolean_t, boolean_t);
 void		dmu_zfetch(zfetch_t *, uint64_t, uint64_t, boolean_t, boolean_t,
    boolean_t);

@@ -82,6 +82,8 @@ extern "C" {
 #define	FM_EREPORT_PAYLOAD_ZFS_VDEV_CKSUM_T	"vdev_cksum_t"
 #define	FM_EREPORT_PAYLOAD_ZFS_VDEV_IO_N	"vdev_io_n"
 #define	FM_EREPORT_PAYLOAD_ZFS_VDEV_IO_T	"vdev_io_t"
+#define	FM_EREPORT_PAYLOAD_ZFS_VDEV_SLOW_IO_N	"vdev_slow_io_n"
+#define	FM_EREPORT_PAYLOAD_ZFS_VDEV_SLOW_IO_T	"vdev_slow_io_t"
 #define	FM_EREPORT_PAYLOAD_ZFS_VDEV_DELAYS	"vdev_delays"
 #define	FM_EREPORT_PAYLOAD_ZFS_PARENT_GUID	"parent_guid"
 #define	FM_EREPORT_PAYLOAD_ZFS_PARENT_TYPE	"parent_type"
@@ -21,7 +21,7 @@

 /*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2020 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2013, 2017 Joyent, Inc. All rights reserved.
 * Copyright (c) 2014 Integros [integros.com]
@@ -191,6 +191,7 @@ typedef enum {
 	ZFS_PROP_REDACTED,
 	ZFS_PROP_REDACT_SNAPS,
 	ZFS_PROP_SNAPSHOTS_CHANGED,
+	ZFS_PROP_PREFETCH,
 	ZFS_NUM_PROPS
 } zfs_prop_t;

@@ -363,6 +364,9 @@ typedef enum {
 	VDEV_PROP_CHECKSUM_T,
 	VDEV_PROP_IO_N,
 	VDEV_PROP_IO_T,
+	VDEV_PROP_RAIDZ_EXPANDING,
+	VDEV_PROP_SLOW_IO_N,
+	VDEV_PROP_SLOW_IO_T,
 	VDEV_NUM_PROPS
 } vdev_prop_t;

@@ -543,6 +547,12 @@ typedef enum zfs_key_location {
 	ZFS_KEYLOCATION_LOCATIONS
 } zfs_keylocation_t;

+typedef enum {
+	ZFS_PREFETCH_NONE = 0,
+	ZFS_PREFETCH_METADATA = 1,
+	ZFS_PREFETCH_ALL = 2
+} zfs_prefetch_type_t;
+
 #define	DEFAULT_PBKDF2_ITERATIONS 350000
 #define	MIN_PBKDF2_ITERATIONS 100000

@@ -1569,6 +1579,8 @@ typedef enum {
 	ZFS_ERR_NOT_USER_NAMESPACE,
 	ZFS_ERR_RESUME_EXISTS,
 	ZFS_ERR_CRYPTO_NOTSUP,
+	ZFS_ERR_RAIDZ_EXPAND_IN_PROGRESS,
+	ZFS_ERR_ASHIFT_MISMATCH,
 } zfs_errno_t;

 /*
@@ -82,12 +82,15 @@ int  multilist_is_empty(multilist_t *);
 unsigned int multilist_get_num_sublists(multilist_t *);
 unsigned int multilist_get_random_index(multilist_t *);

-multilist_sublist_t *multilist_sublist_lock(multilist_t *, unsigned int);
+void multilist_sublist_lock(multilist_sublist_t *);
+multilist_sublist_t *multilist_sublist_lock_idx(multilist_t *, unsigned int);
 multilist_sublist_t *multilist_sublist_lock_obj(multilist_t *, void *);
 void multilist_sublist_unlock(multilist_sublist_t *);

 void multilist_sublist_insert_head(multilist_sublist_t *, void *);
 void multilist_sublist_insert_tail(multilist_sublist_t *, void *);
+void multilist_sublist_insert_after(multilist_sublist_t *, void *, void *);
+void multilist_sublist_insert_before(multilist_sublist_t *, void *, void *);
 void multilist_sublist_move_forward(multilist_sublist_t *mls, void *obj);
 void multilist_sublist_remove(multilist_sublist_t *, void *);
 int  multilist_sublist_is_empty(multilist_sublist_t *);
@@ -20,7 +20,7 @@
 */
 /*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2021 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
 * Copyright 2013 Saso Kiselkov. All rights reserved.
@@ -769,7 +769,7 @@ 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
@@ -784,7 +784,7 @@ extern int bpobj_enqueue_free_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx);
 #define	SPA_ASYNC_DETACH_SPARE			0x4000

 /* device manipulation */
-extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
+extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot, boolean_t ashift_check);
 extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
    int replacing, int rebuild);
 extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
@@ -966,6 +966,10 @@ extern int spa_import_progress_set_max_txg(uint64_t pool_guid,
    uint64_t max_txg);
 extern int spa_import_progress_set_state(uint64_t pool_guid,
    spa_load_state_t spa_load_state);
+extern void spa_import_progress_set_notes(spa_t *spa,
+    const char *fmt, ...) __printflike(2, 3);
+extern void spa_import_progress_set_notes_nolog(spa_t *spa,
+    const char *fmt, ...) __printflike(2, 3);

 /* Pool configuration locks */
 extern int spa_config_tryenter(spa_t *spa, int locks, const void *tag,
@@ -1109,6 +1113,8 @@ 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);

@@ -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) \
@@ -22,6 +22,7 @@
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2011, 2020 by Delphix. All rights reserved.
 * Copyright (c) 2017, Intel Corporation.
+ * Copyright (c) 2023, Klara Inc.
 */

 #ifndef _SYS_VDEV_IMPL_H
@@ -273,7 +274,7 @@ 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)	*/
@@ -453,12 +454,14 @@ struct vdev {
 	zfs_ratelimit_t vdev_checksum_rl;

 	/*
-	 * Checksum and IO thresholds for tuning ZED
+	 * Vdev properties for tuning ZED or zfsd
 	 */
 	uint64_t	vdev_checksum_n;
 	uint64_t	vdev_checksum_t;
 	uint64_t	vdev_io_n;
 	uint64_t	vdev_io_t;
+	uint64_t	vdev_slow_io_n;
+	uint64_t	vdev_slow_io_t;
 };

 #define	VDEV_PAD_SIZE		(8 << 10)
@@ -253,6 +253,9 @@ int zap_add_by_dnode(dnode_t *dn, const char *key,
 int zap_add_uint64(objset_t *ds, uint64_t zapobj, const uint64_t *key,
    int key_numints, int integer_size, uint64_t num_integers,
    const void *val, dmu_tx_t *tx);
+int zap_add_uint64_by_dnode(dnode_t *dn, const uint64_t *key,
+    int key_numints, int integer_size, uint64_t num_integers,
+    const void *val, dmu_tx_t *tx);

 /*
 * Set the attribute with the given name to the given value.  If an
@@ -267,6 +270,9 @@ int zap_update(objset_t *ds, uint64_t zapobj, const char *name,
 int zap_update_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
    int key_numints,
    int integer_size, uint64_t num_integers, const void *val, dmu_tx_t *tx);
+int zap_update_uint64_by_dnode(dnode_t *dn, const uint64_t *key,
+    int key_numints,
+    int integer_size, uint64_t num_integers, const void *val, dmu_tx_t *tx);

 /*
 * Get the length (in integers) and the integer size of the specified
@@ -292,6 +298,8 @@ int zap_remove_norm(objset_t *ds, uint64_t zapobj, const char *name,
 int zap_remove_by_dnode(dnode_t *dn, const char *name, dmu_tx_t *tx);
 int zap_remove_uint64(objset_t *os, uint64_t zapobj, const uint64_t *key,
    int key_numints, dmu_tx_t *tx);
+int zap_remove_uint64_by_dnode(dnode_t *dn, const uint64_t *key,
+    int key_numints, dmu_tx_t *tx);

 /*
 * Returns (in *count) the number of attributes in the specified zap
@@ -145,6 +145,7 @@ typedef struct zap {
 	dmu_buf_user_t zap_dbu;
 	objset_t *zap_objset;
 	uint64_t zap_object;
+	dnode_t *zap_dnode;
 	struct dmu_buf *zap_dbuf;
 	krwlock_t zap_rwlock;
 	boolean_t zap_ismicro;
@@ -47,7 +47,7 @@ struct zap_stats;
 * entries - header space (2*chunksize)
 */
 #define	ZAP_LEAF_NUMCHUNKS_BS(bs) \
-	(((1<<(bs)) - 2*ZAP_LEAF_HASH_NUMENTRIES_BS(bs)) / \
+	(((1U << (bs)) - 2 * ZAP_LEAF_HASH_NUMENTRIES_BS(bs)) / \
 	ZAP_LEAF_CHUNKSIZE - 2)

 #define	ZAP_LEAF_NUMCHUNKS(l) (ZAP_LEAF_NUMCHUNKS_BS(((l)->l_bs)))
@@ -80,7 +80,7 @@ struct zap_stats;
 * chunks per entry (3).
 */
 #define	ZAP_LEAF_HASH_SHIFT_BS(bs) ((bs) - 5)
-#define	ZAP_LEAF_HASH_NUMENTRIES_BS(bs) (1 << ZAP_LEAF_HASH_SHIFT_BS(bs))
+#define	ZAP_LEAF_HASH_NUMENTRIES_BS(bs) (1U << ZAP_LEAF_HASH_SHIFT_BS(bs))
 #define	ZAP_LEAF_HASH_SHIFT(l) (ZAP_LEAF_HASH_SHIFT_BS(((l)->l_bs)))
 #define	ZAP_LEAF_HASH_NUMENTRIES(l) (ZAP_LEAF_HASH_NUMENTRIES_BS(((l)->l_bs)))

@@ -132,7 +132,7 @@ typedef struct zap_leaf_phys {
 	 * with the ZAP_LEAF_CHUNK() macro.
 	 */

-	uint16_t l_hash[1];
+	uint16_t l_hash[];
 } zap_leaf_phys_t;

 typedef union zap_leaf_chunk {
@@ -163,7 +163,7 @@ typedef struct zap_leaf {
 	dmu_buf_user_t l_dbu;
 	krwlock_t l_rwlock;
 	uint64_t l_blkid;		/* 1<<ZAP_BLOCK_SHIFT byte block off */
-	int l_bs;			/* block size shift */
+	uint_t l_bs;			/* block size shift */
 	dmu_buf_t *l_dbuf;
 } zap_leaf_t;

@@ -243,7 +243,7 @@ extern boolean_t zap_entry_normalization_conflict(zap_entry_handle_t *zeh,
 */

 extern void zap_leaf_init(zap_leaf_t *l, boolean_t sort);
-extern void zap_leaf_byteswap(zap_leaf_phys_t *buf, int len);
+extern void zap_leaf_byteswap(zap_leaf_phys_t *buf, size_t len);
 extern void zap_leaf_split(zap_leaf_t *l, zap_leaf_t *nl, boolean_t sort);
 extern void zap_leaf_stats(struct zap *zap, zap_leaf_t *l,
    struct zap_stats *zs);
@@ -181,7 +181,7 @@ typedef struct zil_vdev_node {
 	avl_node_t	zv_node;	/* AVL tree linkage */
 } zil_vdev_node_t;

-#define	ZIL_PREV_BLKS 16
+#define	ZIL_BURSTS 8

 /*
 * Stable storage intent log management structure.  One per dataset.
@@ -216,14 +216,18 @@ struct zilog {
 	uint64_t	zl_parse_lr_count; /* number of log records parsed */
 	itxg_t		zl_itxg[TXG_SIZE]; /* intent log txg chains */
 	list_t		zl_itx_commit_list; /* itx list to be committed */
-	uint64_t	zl_cur_used;	/* current commit log size used */
+	uint64_t	zl_cur_size;	/* current burst full size */
+	uint64_t	zl_cur_left;	/* current burst remaining size */
+	uint64_t	zl_cur_max;	/* biggest record in current burst */
 	list_t		zl_lwb_list;	/* in-flight log write list */
 	avl_tree_t	zl_bp_tree;	/* track bps during log parse */
 	clock_t		zl_replay_time;	/* lbolt of when replay started */
 	uint64_t	zl_replay_blks;	/* number of log blocks replayed */
 	zil_header_t	zl_old_header;	/* debugging aid */
-	uint_t		zl_prev_blks[ZIL_PREV_BLKS]; /* size - sector rounded */
+	uint_t		zl_parallel;	/* workload is multi-threaded */
 	uint_t		zl_prev_rotor;	/* rotor for zl_prev[] */
+	uint_t		zl_prev_opt[ZIL_BURSTS]; /* optimal block size */
+	uint_t		zl_prev_min[ZIL_BURSTS]; /* minimal first block size */
 	txg_node_t	zl_dirty_link;	/* protected by dp_dirty_zilogs list */
 	uint64_t	zl_dirty_max_txg; /* highest txg used to dirty zilog */

@@ -110,8 +110,8 @@ do {									\
 	const uintptr_t __right = (uintptr_t)(RIGHT);			\
 	if (!(__left OP __right))					\
 		libspl_assertf(__FILE__, __FUNCTION__, __LINE__,	\
-		    "%s %s %s (0x%llx %s 0x%llx)", #LEFT, #OP, #RIGHT,	\
-		    (u_longlong_t)__left, #OP, (u_longlong_t)__right);	\
+		    "%s %s %s (%p %s %p)", #LEFT, #OP, #RIGHT,		\
+		    (void *)__left, #OP, (void *)__right);		\
 } while (0)

 #define	VERIFY0(LEFT)							\
@@ -123,6 +123,15 @@ do {									\
 		    (u_longlong_t)__left);				\
 } while (0)

+#define	VERIFY0P(LEFT)							\
+do {									\
+	const uintptr_t __left = (uintptr_t)(LEFT);			\
+	if (!(__left == 0))						\
+		libspl_assertf(__FILE__, __FUNCTION__, __LINE__,	\
+		    "%s == 0 (%p == 0)", #LEFT,				\
+		    (void *)__left);					\
+} while (0)
+
 #ifdef assert
 #undef assert
 #endif
@@ -137,6 +146,7 @@ do {									\
 #define	ASSERT3P(x, y, z)						\
 	((void) sizeof ((uintptr_t)(x)), (void) sizeof ((uintptr_t)(z)))
 #define	ASSERT0(x)		((void) sizeof ((uintptr_t)(x)))
+#define	ASSERT0P(x)		((void) sizeof ((uintptr_t)(x)))
 #define	ASSERT(x)		((void) sizeof ((uintptr_t)(x)))
 #define	assert(x)		((void) sizeof ((uintptr_t)(x)))
 #define	IMPLY(A, B)							\
@@ -149,6 +159,7 @@ do {									\
 #define	ASSERT3U	VERIFY3U
 #define	ASSERT3P	VERIFY3P
 #define	ASSERT0		VERIFY0
+#define	ASSERT0P	VERIFY0P
 #define	ASSERT		VERIFY
 #define	assert		VERIFY
 #define	IMPLY(A, B) \
@@ -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;
@@ -1112,14 +1112,11 @@
        <var-decl name='prev' type-id='b03eadb4' visibility='default'/>
      </data-member>
    </class-decl>
-    <class-decl name='list' size-in-bits='256' is-struct='yes' visibility='default' id='e824dae9'>
+    <class-decl name='list' size-in-bits='192' is-struct='yes' visibility='default' id='e824dae9'>
      <data-member access='public' layout-offset-in-bits='0'>
-        <var-decl name='list_size' type-id='b59d7dce' visibility='default'/>
-      </data-member>
-      <data-member access='public' layout-offset-in-bits='64'>
        <var-decl name='list_offset' type-id='b59d7dce' visibility='default'/>
      </data-member>
-      <data-member access='public' layout-offset-in-bits='128'>
+      <data-member access='public' layout-offset-in-bits='64'>
        <var-decl name='list_head' type-id='b0b5e45e' visibility='default'/>
      </data-member>
    </class-decl>
@@ -1870,7 +1867,8 @@
      <enumerator name='ZFS_PROP_REDACTED' value='93'/>
      <enumerator name='ZFS_PROP_REDACT_SNAPS' value='94'/>
      <enumerator name='ZFS_PROP_SNAPSHOTS_CHANGED' value='95'/>
-      <enumerator name='ZFS_NUM_PROPS' value='96'/>
+      <enumerator name='ZFS_PROP_PREFETCH' value='96'/>
+      <enumerator name='ZFS_NUM_PROPS' value='97'/>
    </enum-decl>
    <typedef-decl name='zfs_prop_t' type-id='4b000d60' id='58603c44'/>
    <enum-decl name='zprop_source_t' naming-typedef-id='a2256d42' id='5903f80e'>
@@ -2878,6 +2876,9 @@
    </function-type>
  </abi-instr>
  <abi-instr address-size='64' path='lib/libzfs/libzfs_crypto.c' language='LANG_C99'>
+    <array-type-def dimensions='1' type-id='38b51b3c' size-in-bits='832' id='02b72c00'>
+      <subrange length='13' type-id='7359adad' id='487fded1'/>
+    </array-type-def>
    <array-type-def dimensions='1' type-id='fb7c6451' size-in-bits='256' id='64177143'>
      <subrange length='32' type-id='7359adad' id='ae5bde82'/>
    </array-type-def>
@@ -2890,6 +2891,10 @@
    <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
    <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
    <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
+    <class-decl name='__locale_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='23de8b96'/>
+    <array-type-def dimensions='1' type-id='80f4b756' size-in-bits='832' id='39e6f84a'>
+      <subrange length='13' type-id='7359adad' id='487fded1'/>
+    </array-type-def>
    <array-type-def dimensions='1' type-id='95e97e5e' size-in-bits='896' id='47394ee0'>
      <subrange length='28' type-id='7359adad' id='3db583d7'/>
    </array-type-def>
@@ -3010,6 +3015,24 @@
    <typedef-decl name='__clock_t' type-id='bd54fe1a' id='4d66c6d7'/>
    <typedef-decl name='__ssize_t' type-id='bd54fe1a' id='41060289'/>
    <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
+    <class-decl name='__locale_struct' size-in-bits='1856' is-struct='yes' visibility='default' id='90cc1ce3'>
+      <data-member access='public' layout-offset-in-bits='0'>
+        <var-decl name='__locales' type-id='02b72c00' visibility='default'/>
+      </data-member>
+      <data-member access='public' layout-offset-in-bits='832'>
+        <var-decl name='__ctype_b' type-id='31347b7a' visibility='default'/>
+      </data-member>
+      <data-member access='public' layout-offset-in-bits='896'>
+        <var-decl name='__ctype_tolower' type-id='6d60f45d' visibility='default'/>
+      </data-member>
+      <data-member access='public' layout-offset-in-bits='960'>
+        <var-decl name='__ctype_toupper' type-id='6d60f45d' visibility='default'/>
+      </data-member>
+      <data-member access='public' layout-offset-in-bits='1024'>
+        <var-decl name='__names' type-id='39e6f84a' visibility='default'/>
+      </data-member>
+    </class-decl>
+    <typedef-decl name='__locale_t' type-id='f01e1813' id='b7ac9b5f'/>
    <class-decl name='__sigset_t' size-in-bits='1024' is-struct='yes' naming-typedef-id='b9c97942' visibility='default' id='2616147f'>
      <data-member access='public' layout-offset-in-bits='0'>
        <var-decl name='__val' type-id='d2baa450' visibility='default'/>
@@ -3025,6 +3048,7 @@
      </data-member>
    </union-decl>
    <typedef-decl name='__sigval_t' type-id='a094b870' id='eabacd01'/>
+    <typedef-decl name='locale_t' type-id='b7ac9b5f' id='973a4f8d'/>
    <class-decl name='siginfo_t' size-in-bits='1024' is-struct='yes' naming-typedef-id='cb681f62' visibility='default' id='d8149419'>
      <data-member access='public' layout-offset-in-bits='0'>
        <var-decl name='si_signo' type-id='95e97e5e' visibility='default'/>
@@ -3260,9 +3284,13 @@
    <pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
    <pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
    <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
+    <pointer-type-def type-id='23de8b96' size-in-bits='64' id='38b51b3c'/>
+    <pointer-type-def type-id='90cc1ce3' size-in-bits='64' id='f01e1813'/>
    <qualified-type-def type-id='9b23c9ad' restrict='yes' id='8c85230f'/>
    <qualified-type-def type-id='80f4b756' restrict='yes' id='9d26089a'/>
    <pointer-type-def type-id='80f4b756' size-in-bits='64' id='7d3cd834'/>
+    <qualified-type-def type-id='95e97e5e' const='yes' id='2448a865'/>
+    <pointer-type-def type-id='2448a865' size-in-bits='64' id='6d60f45d'/>
    <qualified-type-def type-id='aca3bac8' const='yes' id='2498fd78'/>
    <pointer-type-def type-id='2498fd78' size-in-bits='64' id='eed6c816'/>
    <qualified-type-def type-id='eed6c816' restrict='yes' id='a431a9da'/>
@@ -3295,6 +3323,7 @@
    <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
    <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
    <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
+    <class-decl name='__locale_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='23de8b96'/>
    <function-decl name='zpool_get_prop_int' mangled-name='zpool_get_prop_int' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_prop_int'>
      <parameter type-id='4c81de99'/>
      <parameter type-id='5d0c23fb'/>
@@ -3399,6 +3428,10 @@
    <function-decl name='dlerror' visibility='default' binding='global' size-in-bits='64'>
      <return type-id='26a90f95'/>
    </function-decl>
+    <function-decl name='uselocale' visibility='default' binding='global' size-in-bits='64'>
+      <parameter type-id='973a4f8d'/>
+      <return type-id='973a4f8d'/>
+    </function-decl>
    <function-decl name='PKCS5_PBKDF2_HMAC_SHA1' visibility='default' binding='global' size-in-bits='64'>
      <parameter type-id='80f4b756'/>
      <parameter type-id='95e97e5e'/>
@@ -3482,8 +3515,9 @@
      <parameter type-id='80f4b756'/>
      <return type-id='26a90f95'/>
    </function-decl>
-    <function-decl name='strerror' visibility='default' binding='global' size-in-bits='64'>
+    <function-decl name='strerror_l' visibility='default' binding='global' size-in-bits='64'>
      <parameter type-id='95e97e5e'/>
+      <parameter type-id='973a4f8d'/>
      <return type-id='26a90f95'/>
    </function-decl>
    <function-decl name='tcgetattr' visibility='default' binding='global' size-in-bits='64'>
@@ -3840,12 +3874,18 @@
    <qualified-type-def type-id='9c313c2d' const='yes' id='c3b7ba7d'/>
    <pointer-type-def type-id='c3b7ba7d' size-in-bits='64' id='713a56f5'/>
    <pointer-type-def type-id='01a1b934' size-in-bits='64' id='566b3f52'/>
+    <qualified-type-def type-id='566b3f52' restrict='yes' id='c878edd6'/>
+    <pointer-type-def type-id='566b3f52' size-in-bits='64' id='82d4e9e8'/>
+    <qualified-type-def type-id='82d4e9e8' restrict='yes' id='aa19c230'/>
    <pointer-type-def type-id='7e291ce6' size-in-bits='64' id='ca64ff60'/>
    <pointer-type-def type-id='9da381c4' size-in-bits='64' id='cb785ebf'/>
    <pointer-type-def type-id='1b055409' size-in-bits='64' id='9d424d31'/>
    <pointer-type-def type-id='8e0af06e' size-in-bits='64' id='053457bd'/>
    <pointer-type-def type-id='857bb57e' size-in-bits='64' id='75be733c'/>
    <pointer-type-def type-id='a63d15a3' size-in-bits='64' id='a195f4a3'/>
+    <qualified-type-def type-id='a195f4a3' restrict='yes' id='33518961'/>
+    <pointer-type-def type-id='a195f4a3' size-in-bits='64' id='e80ff3ab'/>
+    <qualified-type-def type-id='e80ff3ab' restrict='yes' id='8f2c7109'/>
    <pointer-type-def type-id='eae6431d' size-in-bits='64' id='0d41d328'/>
    <pointer-type-def type-id='7a6844eb' size-in-bits='64' id='18c91f9e'/>
    <pointer-type-def type-id='dddf6ca2' size-in-bits='64' id='d915a820'/>
@@ -4278,9 +4318,13 @@
      <parameter type-id='9d424d31'/>
      <return type-id='95e97e5e'/>
    </function-decl>
-    <function-decl name='getgrnam' visibility='default' binding='global' size-in-bits='64'>
-      <parameter type-id='80f4b756'/>
-      <return type-id='566b3f52'/>
+    <function-decl name='getgrnam_r' visibility='default' binding='global' size-in-bits='64'>
+      <parameter type-id='9d26089a'/>
+      <parameter type-id='c878edd6'/>
+      <parameter type-id='266fe297'/>
+      <parameter type-id='b59d7dce'/>
+      <parameter type-id='aa19c230'/>
+      <return type-id='95e97e5e'/>
    </function-decl>
    <function-decl name='hasmntopt' visibility='default' binding='global' size-in-bits='64'>
      <parameter type-id='48bea5ec'/>
@@ -4304,9 +4348,13 @@
      <parameter type-id='18c91f9e'/>
      <return type-id='95e97e5e'/>
    </function-decl>
-    <function-decl name='getpwnam' visibility='default' binding='global' size-in-bits='64'>
-      <parameter type-id='80f4b756'/>
-      <return type-id='a195f4a3'/>
+    <function-decl name='getpwnam_r' visibility='default' binding='global' size-in-bits='64'>
+      <parameter type-id='9d26089a'/>
+      <parameter type-id='33518961'/>
+      <parameter type-id='266fe297'/>
+      <parameter type-id='b59d7dce'/>
+      <parameter type-id='8f2c7109'/>
+      <return type-id='95e97e5e'/>
    </function-decl>
    <function-decl name='strtol' visibility='default' binding='global' size-in-bits='64'>
      <parameter type-id='9d26089a'/>
@@ -5671,7 +5719,10 @@
      <enumerator name='VDEV_PROP_CHECKSUM_T' value='43'/>
      <enumerator name='VDEV_PROP_IO_N' value='44'/>
      <enumerator name='VDEV_PROP_IO_T' value='45'/>
-      <enumerator name='VDEV_NUM_PROPS' value='46'/>
+      <enumerator name='VDEV_PROP_RAIDZ_EXPANDING' value='46'/>
+      <enumerator name='VDEV_PROP_SLOW_IO_N' value='47'/>
+      <enumerator name='VDEV_PROP_SLOW_IO_T' value='48'/>
+      <enumerator name='VDEV_NUM_PROPS' value='49'/>
    </enum-decl>
    <typedef-decl name='vdev_prop_t' type-id='1573bec8' id='5aa5c90c'/>
    <class-decl name='zpool_load_policy' size-in-bits='256' is-struct='yes' visibility='default' id='2f65b36f'>
@@ -6242,6 +6293,7 @@
    <function-decl name='zpool_add' mangled-name='zpool_add' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_add'>
      <parameter type-id='4c81de99' name='zhp'/>
      <parameter type-id='5ce45b60' name='nvroot'/>
+      <parameter type-id='c19b74c3' name='ashift_check'/>
      <return type-id='95e97e5e'/>
    </function-decl>
    <function-decl name='zpool_export' mangled-name='zpool_export' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_export'>
@@ -6693,7 +6745,7 @@
      <enumerator name='LZC_SEND_FLAG_RAW' value='8'/>
      <enumerator name='LZC_SEND_FLAG_SAVED' value='16'/>
    </enum-decl>
-    <class-decl name='ddt_key' size-in-bits='320' is-struct='yes' visibility='default' id='e0a4a1cb'>
+    <class-decl name='ddt_key_t' size-in-bits='320' is-struct='yes' naming-typedef-id='67f6d2cf' visibility='default' id='5fae1718'>
      <data-member access='public' layout-offset-in-bits='0'>
        <var-decl name='ddk_cksum' type-id='39730d0b' visibility='default'/>
      </data-member>
@@ -6701,7 +6753,7 @@
        <var-decl name='ddk_prop' type-id='9c313c2d' visibility='default'/>
      </data-member>
    </class-decl>
-    <typedef-decl name='ddt_key_t' type-id='e0a4a1cb' id='67f6d2cf'/>
+    <typedef-decl name='ddt_key_t' type-id='5fae1718' id='67f6d2cf'/>
    <enum-decl name='dmu_object_type' id='04b3b0b9'>
      <underlying-type type-id='9cac1fee'/>
      <enumerator name='DMU_OT_NONE' value='0'/>
@@ -22,7 +22,7 @@
 /*
 * Copyright 2015 Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2020 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
 * Copyright (c) 2018 Datto Inc.
 * Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
@@ -1724,7 +1724,7 @@ zpool_discard_checkpoint(zpool_handle_t *zhp)
 * necessary verification to ensure that the vdev specification is well-formed.
 */
 int
-zpool_add(zpool_handle_t *zhp, nvlist_t *nvroot)
+zpool_add(zpool_handle_t *zhp, nvlist_t *nvroot, boolean_t check_ashift)
 {
 	zfs_cmd_t zc = {"\0"};
 	int ret;
@@ -1756,6 +1756,7 @@ zpool_add(zpool_handle_t *zhp, nvlist_t *nvroot)

 	zcmd_write_conf_nvlist(hdl, &zc, nvroot);
 	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
+	zc.zc_flags = check_ashift;

 	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_ADD, &zc) != 0) {
 		switch (errno) {
@@ -1899,7 +1900,8 @@ zpool_rewind_exclaim(libzfs_handle_t *hdl, const char *name, boolean_t dryrun,
 	(void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss);

 	if (localtime_r((time_t *)&rewindto, &t) != NULL &&
-	    strftime(timestr, 128, "%c", &t) != 0) {
+	    ctime_r((time_t *)&rewindto, timestr) != NULL) {
+		timestr[24] = 0;
 		if (dryrun) {
 			(void) printf(dgettext(TEXT_DOMAIN,
 			    "Would be able to return %s "
@@ -1961,7 +1963,8 @@ zpool_explain_recover(libzfs_handle_t *hdl, const char *name, int reason,
 	    "Recovery is possible, but will result in some data loss.\n"));

 	if (localtime_r((time_t *)&rewindto, &t) != NULL &&
-	    strftime(timestr, 128, "%c", &t) != 0) {
+	    ctime_r((time_t *)&rewindto, timestr) != NULL) {
+		timestr[24] = 0;
 		(void) printf(dgettext(TEXT_DOMAIN,
 		    "\tReturning the pool to its state as of %s\n"
 		    "\tshould correct the problem.  "),
@@ -5224,6 +5227,8 @@ zpool_get_vdev_prop_value(nvlist_t *nvprop, vdev_prop_t prop, char *prop_name,
 		case VDEV_PROP_CHECKSUM_T:
 		case VDEV_PROP_IO_N:
 		case VDEV_PROP_IO_T:
+		case VDEV_PROP_SLOW_IO_N:
+		case VDEV_PROP_SLOW_IO_T:
 			if (intval == UINT64_MAX) {
 				(void) strlcpy(buf, "-", len);
 			} else {
@@ -1053,6 +1053,7 @@ send_progress_thread(void *arg)
 		}
 	}
 	pthread_cleanup_pop(B_TRUE);
+	return (NULL);
 }

 static boolean_t
@@ -22,7 +22,7 @@
 /*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright 2020 Joyent, Inc. All rights reserved.
- * Copyright (c) 2011, 2020 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2024 by Delphix. All rights reserved.
 * Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
 * Copyright (c) 2017 Datto Inc.
 * Copyright (c) 2020 The FreeBSD Foundation
@@ -317,6 +317,9 @@ libzfs_error_description(libzfs_handle_t *hdl)
 	case EZFS_RESUME_EXISTS:
 		return (dgettext(TEXT_DOMAIN, "Resuming recv on existing "
 		    "dataset without force"));
+	case EZFS_ASHIFT_MISMATCH:
+		return (dgettext(TEXT_DOMAIN, "adding devices with "
+		    "different physical sector sizes is not allowed"));
 	case EZFS_UNKNOWN:
 		return (dgettext(TEXT_DOMAIN, "unknown error"));
 	default:
@@ -763,6 +766,9 @@ zpool_standard_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...)
 	case ZFS_ERR_IOC_ARG_BADTYPE:
 		zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
 		break;
+	case ZFS_ERR_ASHIFT_MISMATCH:
+		zfs_verror(hdl, EZFS_ASHIFT_MISMATCH, fmt, ap);
+		break;
 	default:
 		zfs_error_aux(hdl, "%s", strerror(error));
 		zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);
@@ -1699,7 +1705,9 @@ zprop_parse_value(libzfs_handle_t *hdl, nvpair_t *elem, int prop,
 		    (prop == VDEV_PROP_CHECKSUM_N ||
 		    prop == VDEV_PROP_CHECKSUM_T ||
 		    prop == VDEV_PROP_IO_N ||
-		    prop == VDEV_PROP_IO_T)) {
+		    prop == VDEV_PROP_IO_T ||
+		    prop == VDEV_PROP_SLOW_IO_N ||
+		    prop == VDEV_PROP_SLOW_IO_T)) {
 			*ivalp = UINT64_MAX;
 		}

@@ -273,6 +273,16 @@ zpool_label_disk(libzfs_handle_t *hdl, zpool_handle_t *zhp, const char *name)
 	vtoc->efi_parts[0].p_start = start_block;
 	vtoc->efi_parts[0].p_size = slice_size;

+	if (vtoc->efi_parts[0].p_size * vtoc->efi_lbasize < SPA_MINDEVSIZE) {
+		(void) close(fd);
+		efi_free(vtoc);
+
+		zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot "
+		    "label '%s': partition would be less than the minimum "
+		    "device size (64M)"), path);
+		return (zfs_error(hdl, EZFS_LABELFAILED, errbuf));
+	}
+
 	/*
 	 * Why we use V_USR: V_BACKUP confuses users, and is considered
 	 * disposable by some EFI utilities (since EFI doesn't have a backup
@@ -62,7 +62,6 @@ dist_man_MANS = \
 	%D%/man8/zfs-userspace.8 \
 	%D%/man8/zfs-wait.8 \
 	%D%/man8/zfs_ids_to_path.8 \
-	%D%/man8/zfs_prepare_disk.8 \
 	%D%/man8/zgenhostid.8 \
 	%D%/man8/zinject.8 \
 	%D%/man8/zpool.8 \
@@ -115,7 +114,8 @@ endif

 nodist_man_MANS = \
 	%D%/man8/zed.8 \
-	%D%/man8/zfs-mount-generator.8
+	%D%/man8/zfs-mount-generator.8 \
+	%D%/man8/zfs_prepare_disk.8

 dist_noinst_DATA += $(dist_noinst_man_MANS) $(dist_man_MANS)

@@ -186,18 +186,8 @@ reading it could cause a lock-up if the list grow too large
 without limiting the output.
 "(truncated)" will be shown if the list is larger than the limit.
 .
-.It Sy spl_taskq_thread_timeout_ms Ns = Ns Sy 10000 Pq uint
-(Linux-only)
-How long a taskq has to have had no work before we tear it down.
-Previously, we would tear down a dynamic taskq worker as soon
-as we noticed it had no work, but it was observed that this led
-to a lot of churn in tearing down things we then immediately
-spawned anew.
-In practice, it seems any nonzero value will remove the vast
-majority of this churn, while the nontrivially larger value
-was chosen to help filter out the little remaining churn on
-a mostly idle system.
-Setting this value to
-.Sy 0
-will revert to the previous behavior.
+.It Sy spl_taskq_thread_timeout_ms Ns = Ns Sy 5000 Pq uint
+Minimum idle threads exit interval for dynamic taskqs.
+Smaller values allow idle threads exit more often and potentially be
+respawned again on demand, causing more churn.
 .El
@@ -2,6 +2,7 @@
 .\" Copyright (c) 2013 by Turbo Fredriksson <turbo@bayour.com>. All rights reserved.
 .\" Copyright (c) 2019, 2021 by Delphix. All rights reserved.
 .\" Copyright (c) 2019 Datto Inc.
+.\" Copyright (c) 2023, 2024 Klara, Inc.
 .\" 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
@@ -15,7 +16,7 @@
 .\" own identifying information:
 .\" Portions Copyright [yyyy] [name of copyright owner]
 .\"
-.Dd July 21, 2023
+.Dd January 9, 2024
 .Dt ZFS 4
 .Os
 .
@@ -244,12 +245,25 @@ For blocks that could be forced to be a gang block (due to
 .Sy metaslab_force_ganging ) ,
 force this many of them to be gang blocks.
 .
-.It Sy zfs_ddt_zap_default_bs Ns = Ns Sy 15 Po 32 KiB Pc Pq int
+.It Sy brt_zap_prefetch Ns = Ns Sy 1 Ns | Ns 0 Pq int
+Controls prefetching BRT records for blocks which are going to be cloned.
+.
+.It Sy brt_zap_default_bs Ns = Ns Sy 12 Po 4 KiB Pc Pq int
+Default BRT ZAP data block size as a power of 2. Note that changing this after
+creating a BRT on the pool will not affect existing BRTs, only newly created
+ones.
+.
+.It Sy brt_zap_default_ibs Ns = Ns Sy 12 Po 4 KiB Pc Pq int
+Default BRT ZAP indirect block size as a power of 2. Note that changing this
+after creating a BRT on the pool will not affect existing BRTs, only newly
+created ones.
+.
+.It Sy ddt_zap_default_bs Ns = Ns Sy 15 Po 32 KiB Pc Pq int
 Default DDT ZAP data block size as a power of 2. Note that changing this after
 creating a DDT on the pool will not affect existing DDTs, only newly created
 ones.
 .
-.It Sy zfs_ddt_zap_default_ibs Ns = Ns Sy 15 Po 32 KiB Pc Pq int
+.It Sy ddt_zap_default_ibs Ns = Ns Sy 15 Po 32 KiB Pc Pq int
 Default DDT ZAP indirect block size as a power of 2. Note that changing this
 after creating a DDT on the pool will not affect existing DDTs, only newly
 created ones.
@@ -530,6 +544,10 @@ 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 zfetch_min_distance Ns = Ns Sy 4194304 Ns B Po 4 MiB Pc Pq uint
 Min bytes to prefetch per stream.
 Prefetch distance starts from the demand access size and quickly grows to
@@ -544,6 +562,13 @@ Max bytes to prefetch per stream.
 .It Sy zfetch_max_idistance Ns = Ns Sy 67108864 Ns B Po 64 MiB Pc Pq uint
 Max bytes to prefetch indirects for per stream.
 .
+.It Sy zfetch_max_reorder Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq uint
+Requests within this byte distance from the current prefetch stream position
+are considered parts of the stream, reordered due to parallel processing.
+Such requests do not advance the stream position immediately unless
+.Sy zfetch_hole_shift
+fill threshold is reached, but saved to fill holes in the stream later.
+.
 .It Sy zfetch_max_streams Ns = Ns Sy 8 Pq uint
 Max number of streams per zfetch (prefetch streams per file).
 .
@@ -798,7 +823,7 @@ Note that this should not be set below the ZED thresholds
 (currently 10 checksums over 10 seconds)
 or else the daemon may not trigger any action.
 .
-.It Sy zfs_commit_timeout_pct Ns = Ns Sy 5 Ns % Pq uint
+.It Sy zfs_commit_timeout_pct Ns = Ns Sy 10 Ns % Pq uint
 This controls the amount of time that a ZIL block (lwb) will remain "open"
 when it isn't "full", and it has a thread waiting for it to be committed to
 stable storage.
@@ -1345,6 +1370,42 @@ _
 	4	Driver	No driver retries on driver errors.
 .TE
 .
+.It Sy zfs_vdev_disk_max_segs Ns = Ns Sy 0 Pq uint
+Maximum number of segments to add to a BIO (min 4).
+If this is higher than the maximum allowed by the device queue or the kernel
+itself, it will be clamped.
+Setting it to zero will cause the kernel's ideal size to be used.
+This parameter only applies on Linux.
+This parameter is ignored if
+.Sy zfs_vdev_disk_classic Ns = Ns Sy 1 .
+.
+.It Sy zfs_vdev_disk_classic Ns = Ns 0 Ns | Ns Sy 1 Pq uint
+Controls the method used to submit IO to the Linux block layer
+(default
+.Sy 1 "classic" Ns
+)
+.Pp
+If set to 1, the "classic" method is used.
+This is the method that has been in use since the earliest versions of
+ZFS-on-Linux.
+It has known issues with highly fragmented IO requests and is less efficient on
+many workloads, but it well known and well understood.
+.Pp
+If set to 0, the "new" method is used.
+This method is available since 2.2.4 and should resolve all known issues and be
+far more efficient, but has not had as much testing.
+In the 2.2.x series, this parameter defaults to 1, to use the "classic" method.
+.Pp
+It is not recommended that you change it except on advice from the OpenZFS
+developers.
+If you do change it, please also open a bug report describing why you did so,
+including the workload involved and any error messages.
+.Pp
+This parameter and the "classic" submission method will be removed in a future
+release of OpenZFS once we have total confidence in the new method.
+.Pp
+This parameter only applies on Linux, and can only be set at module load time.
+.
 .It Sy zfs_expire_snapshot Ns = Ns Sy 300 Ns s Pq int
 Time before expiring
 .Pa .zfs/snapshot .
@@ -2169,13 +2230,6 @@ This sets the maximum number of write bytes logged via WR_COPIED.
 It tunes a tradeoff between additional memory copy and possibly worse log
 space efficiency vs additional range lock/unlock.
 .
-.It Sy zil_min_commit_timeout Ns = Ns Sy 5000 Pq u64
-This sets the minimum delay in nanoseconds ZIL care to delay block commit,
-waiting for more records.
-If ZIL writes are too fast, kernel may not be able sleep for so short interval,
-increasing log latency above allowed by
-.Sy zfs_commit_timeout_pct .
-.
 .It Sy zil_nocacheflush Ns = Ns Sy 0 Ns | Ns 1 Pq int
 Disable the cache flush commands that are normally sent to disk by
 the ZIL after an LWB write has completed.
@@ -2333,6 +2387,13 @@ The number of requests which can be handled concurrently is controlled by
 is ignored when running on a kernel that supports block multiqueue
 .Pq Li blk-mq .
 .
+.It Sy zvol_num_taskqs Ns = Ns Sy 0 Pq uint
+Number of zvol taskqs.
+If
+.Sy 0
+(the default) then scaling is done internally to prefer 6 threads per taskq.
+This only applies on Linux.
+.
 .It Sy zvol_threads Ns = Ns Sy 0 Pq uint
 The number of system wide threads to use for processing zvol block IOs.
 If
@@ -44,7 +44,7 @@ section, below.
 Every vdev has a set of properties that export statistics about the vdev
 as well as control various behaviors.
 Properties are not inherited from top-level vdevs, with the exception of
-checksum_n, checksum_t, io_n, and io_t.
+checksum_n, checksum_t, io_n, io_t, slow_io_n, and slow_io_t.
 .Pp
 The values of numeric properties can be specified using human-readable suffixes
 .Po for example,
@@ -117,7 +117,7 @@ If this device is currently being removed from the pool
 .Pp
 The following native properties can be used to change the behavior of a vdev.
 .Bl -tag -width "allocating"
-.It Sy checksum_n , checksum_t , io_n , io_t
+.It Sy checksum_n , checksum_t , io_n , io_t , slow_io_n , slow_io_t
 Tune the fault management daemon by specifying checksum/io thresholds of <N>
 errors in <T> seconds, respectively.
 These properties can be set on leaf and top-level vdevs.
@@ -127,7 +127,13 @@ If the property is only set on the top-level vdev, this value will be used.
 The value of these properties do not persist across vdev replacement.
 For this reason, it is advisable to set the property on the top-level vdev -
 not on the leaf vdev itself.
-The default values are 10 errors in 600 seconds.
+The default values for
+.Sy OpenZFS on Linux
+are 10 errors in 600 seconds.
+For
+.Sy OpenZFS on FreeBSD
+defaults see
+.Xr zfsd 8 .
 .It Sy comment
 A text comment up to 8192 characters long
 .It Sy bootsize
@@ -1613,6 +1613,23 @@ If this property is set to
 then only metadata is cached.
 The default value is
 .Sy all .
+.It Sy prefetch Ns = Ns Sy all Ns | Ns Sy none Ns | Ns Sy metadata
+Controls what speculative prefetch does.
+If this property is set to
+.Sy all ,
+then both user data and metadata are prefetched.
+If this property is set to
+.Sy none ,
+then neither user data nor metadata are prefetched.
+If this property is set to
+.Sy metadata ,
+then only metadata are prefetched.
+The default value is
+.Sy all .
+.Pp
+Please note that the module parameter zfs_disable_prefetch=1 can
+be used to totally disable speculative prefetch, bypassing anything
+this property does.
 .It Sy setuid Ns = Ns Sy on Ns | Ns Sy off
 Controls whether the setuid bit is respected for the file system.
 The default value is
@@ -260,8 +260,8 @@ sufficient replicas exist to continue functioning.
 The underlying conditions are as follows:
 .Bl -bullet -compact
 .It
-The number of checksum errors exceeds acceptable levels and the device is
-degraded as an indication that something may be wrong.
+The number of checksum errors or slow I/Os exceeds acceptable levels and the
+device is degraded as an indication that something may be wrong.
 ZFS continues to use the device as necessary.
 .It
 The number of I/O errors exceeds acceptable levels.
@@ -43,7 +43,7 @@
 .Cm mount
 .Op Fl Oflv
 .Op Fl o Ar options
-.Fl a Ns | Ns Ar filesystem
+.Fl a Ns | Ns Fl R Ar filesystem Ns | Ns Ar filesystem
 .Nm zfs
 .Cm unmount
 .Op Fl fu
@@ -61,7 +61,7 @@ Displays all ZFS file systems currently mounted.
 .Cm mount
 .Op Fl Oflv
 .Op Fl o Ar options
-.Fl a Ns | Ns Ar filesystem
+.Fl a Ns | Ns Fl R Ar filesystem Ns | Ns Ar filesystem
 .Xc
 Mount ZFS filesystem on a path described by its
 .Sy mountpoint
@@ -83,6 +83,8 @@ for more information.
 .It Fl a
 Mount all available ZFS file systems.
 Invoked automatically as part of the boot process if configured.
+.It Fl R
+Mount the specified filesystems along with all their children.
 .It Ar filesystem
 Mount the specified filesystem.
 .It Fl o Ar options
@@ -69,6 +69,7 @@ Force a vdev into the DEGRADED or FAULTED state.
 .Nm zinject
 .Fl d Ar vdev
 .Fl D Ar latency : Ns Ar lanes
+.Op Fl T Ar read|write
 .Ar pool
 .Xc
 Add an artificial delay to I/O requests on a particular
@@ -24,8 +24,9 @@
 .\" Copyright (c) 2018 George Melikov. All Rights Reserved.
 .\" Copyright 2017 Nexenta Systems, Inc.
 .\" Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
+.\" Copyright (c) 2024 by Delphix. All Rights Reserved.
 .\"
-.Dd March 16, 2022
+.Dd March 8, 2024
 .Dt ZPOOL-ADD 8
 .Os
 .
@@ -36,6 +37,7 @@
 .Nm zpool
 .Cm add
 .Op Fl fgLnP
+.Op Fl -allow-in-use -allow-replication-mismatch -allow-ashift-mismatch
 .Oo Fl o Ar property Ns = Ns Ar value Oc
 .Ar pool vdev Ns …
 .
@@ -56,7 +58,8 @@ subcommand.
 .It Fl f
 Forces use of
 .Ar vdev Ns s ,
-even if they appear in use or specify a conflicting replication level.
+even if they appear in use, have conflicting ashift values, or specify
+a conflicting replication level.
 Not all devices can be overridden in this manner.
 .It Fl g
 Display
@@ -91,6 +94,17 @@ See the
 manual page for a list of valid properties that can be set.
 The only property supported at the moment is
 .Sy ashift .
+.It Fl -allow-ashift-mismatch
+Disable the ashift validation which allows mismatched ashift values in the
+pool.
+Adding top-level
+.Ar vdev Ns s
+with different sector sizes will prohibit future device removal operations, see
+.Xr zpool-remove 8 .
+.It Fl -allow-in-use
+Allow vdevs to be added even if they might be in use in another pool.
+.It Fl -allow-replication-mismatch
+Allow vdevs with conflicting replication levels to be added to the pool.
 .El
 .
 .Sh EXAMPLES
@@ -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
@@ -36,7 +36,7 @@
 .Sh SYNOPSIS
 .Nm zpool
 .Cm status
-.Op Fl DeigLpPstvx
+.Op Fl DegiLpPstvx
 .Op Fl T Sy u Ns | Ns Sy d
 .Op Fl c Op Ar SCRIPT1 Ns Oo , Ns Ar SCRIPT2 Oc Ns …
 .Oo Ar pool Oc Ns …
@@ -69,14 +69,20 @@ See the
 option of
 .Nm zpool Cm iostat
 for complete details.
+.It Fl D
+Display a histogram of deduplication statistics, showing the allocated
+.Pq physically present on disk
+and referenced
+.Pq logically referenced in the pool
+block counts and sizes by reference count.
 .It Fl e
 Only show unhealthy vdevs (not-ONLINE or with errors).
-.It Fl i
-Display vdev initialization status.
 .It Fl g
 Display vdev GUIDs instead of the normal device names
 These GUIDs can be used in place of device names for the zpool
 detach/offline/remove/replace commands.
+.It Fl i
+Display vdev initialization status.
 .It Fl L
 Display real paths for vdevs resolving all symbolic links.
 This can be used to look up the current block device name regardless of the
@@ -90,12 +96,6 @@ the path.
 This can be used in conjunction with the
 .Fl L
 flag.
-.It Fl D
-Display a histogram of deduplication statistics, showing the allocated
-.Pq physically present on disk
-and referenced
-.Pq logically referenced in the pool
-block counts and sizes by reference count.
 .It Fl s
 Display the number of leaf vdev slow I/O operations.
 This is the number of I/O operations that didn't complete in
@@ -32,6 +32,14 @@
 */

 #if defined(__aarch64__)
+
+/* make gcc <= 9 happy */
+#if LD_VERSION >= 233010000
+#define CFI_NEGATE_RA_STATE .cfi_negate_ra_state
+#else
+#define CFI_NEGATE_RA_STATE
+#endif
+
 	.text
 	.section	.note.gnu.property,"a",@note
 	.p2align	3
@@ -51,7 +59,7 @@
 zfs_blake3_compress_in_place_sse2:
 	.cfi_startproc
 	hint	#25
-	.cfi_negate_ra_state
+	CFI_NEGATE_RA_STATE
 	sub	sp, sp, #96
 	stp	x29, x30, [sp, #64]
 	add	x29, sp, #64
@@ -555,7 +563,7 @@ compress_pre:
 zfs_blake3_compress_xof_sse2:
 	.cfi_startproc
 	hint	#25
-	.cfi_negate_ra_state
+	CFI_NEGATE_RA_STATE
 	sub	sp, sp, #96
 	stp	x29, x30, [sp, #64]
 	add	x29, sp, #64
@@ -608,7 +616,7 @@ zfs_blake3_compress_xof_sse2:
 zfs_blake3_hash_many_sse2:
 	.cfi_startproc
 	hint	#25
-	.cfi_negate_ra_state
+	CFI_NEGATE_RA_STATE
 	stp	d15, d14, [sp, #-160]!
 	stp	d13, d12, [sp, #16]
 	stp	d11, d10, [sp, #32]
@@ -32,6 +32,14 @@
 */

 #if defined(__aarch64__)
+
+/* make gcc <= 9 happy */
+#if LD_VERSION >= 233010000
+#define CFI_NEGATE_RA_STATE .cfi_negate_ra_state
+#else
+#define CFI_NEGATE_RA_STATE
+#endif
+
 	.text
 	.section	.note.gnu.property,"a",@note
 	.p2align	3
@@ -51,7 +59,7 @@
 zfs_blake3_compress_in_place_sse41:
 	.cfi_startproc
 	hint	#25
-	.cfi_negate_ra_state
+	CFI_NEGATE_RA_STATE
 	sub	sp, sp, #96
 	stp	x29, x30, [sp, #64]
 	add	x29, sp, #64
@@ -565,7 +573,7 @@ compress_pre:
 zfs_blake3_compress_xof_sse41:
 	.cfi_startproc
 	hint	#25
-	.cfi_negate_ra_state
+	CFI_NEGATE_RA_STATE
 	sub	sp, sp, #96
 	stp	x29, x30, [sp, #64]
 	add	x29, sp, #64
@@ -21,6 +21,16 @@

 #if defined(__aarch64__)

+	.section	.note.gnu.property,"a",@note
+	.p2align	3
+	.word	4
+	.word	16
+	.word	5
+	.asciz	"GNU"
+	.word	3221225472
+	.word	4
+	.word	3
+	.word	0
 .text

 .align	6
@@ -21,6 +21,16 @@

 #if defined(__aarch64__)

+	.section	.note.gnu.property,"a",@note
+	.p2align	3
+	.word	4
+	.word	16
+	.word	5
+	.asciz	"GNU"
+	.word	3221225472
+	.word	4
+	.word	3
+	.word	0
 .text

 .align	6
@@ -41,6 +41,7 @@
 #include <sys/types.h>
 #include <sys/param.h>
 #include <sys/string.h>
+#include <rpc/types.h>
 #include <rpc/xdr.h>
 #include <sys/mod.h>

@@ -417,10 +417,8 @@ abd_iter_init(struct abd_iter *aiter, abd_t *abd)
 {
 	ASSERT(!abd_is_gang(abd));
 	abd_verify(abd);
+	memset(aiter, 0, sizeof (struct abd_iter));
 	aiter->iter_abd = abd;
-	aiter->iter_pos = 0;
-	aiter->iter_mapaddr = NULL;
-	aiter->iter_mapsize = 0;
 }

 /*
@@ -1869,10 +1869,8 @@ zfs_readdir(vnode_t *vp, zfs_uio_t *uio, cred_t *cr, int *eofp,

 		ASSERT3S(outcount, <=, bufsize);

-		/* Prefetch znode */
 		if (prefetch)
-			dmu_prefetch(os, objnum, 0, 0, 0,
-			    ZIO_PRIORITY_SYNC_READ);
+			dmu_prefetch_dnode(os, objnum, ZIO_PRIORITY_SYNC_READ);

 		/*
 		 * Move to the next entry, fill in the previous offset.
@@ -1273,7 +1273,7 @@ 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));

@@ -36,12 +36,12 @@ static int spl_taskq_thread_bind = 0;
 module_param(spl_taskq_thread_bind, int, 0644);
 MODULE_PARM_DESC(spl_taskq_thread_bind, "Bind taskq thread to CPU by default");

-static uint_t spl_taskq_thread_timeout_ms = 10000;
+static uint_t spl_taskq_thread_timeout_ms = 5000;
 /* BEGIN CSTYLED */
 module_param(spl_taskq_thread_timeout_ms, uint, 0644);
 /* END CSTYLED */
 MODULE_PARM_DESC(spl_taskq_thread_timeout_ms,
-	"Time to require a dynamic thread be idle before it gets cleaned up");
+	"Minimum idle threads exit interval for dynamic taskqs");

 static int spl_taskq_thread_dynamic = 1;
 module_param(spl_taskq_thread_dynamic, int, 0444);
@@ -594,8 +594,7 @@ taskq_dispatch(taskq_t *tq, task_func_t func, void *arg, uint_t flags)
 	ASSERT(tq->tq_nactive <= tq->tq_nthreads);
 	if ((flags & TQ_NOQUEUE) && (tq->tq_nactive == tq->tq_nthreads)) {
 		/* Dynamic taskq may be able to spawn another thread */
-		if (!(tq->tq_flags & TASKQ_DYNAMIC) ||
-		    taskq_thread_spawn(tq) == 0)
+		if (taskq_thread_spawn(tq) == 0)
 			goto out;
 	}

@@ -629,11 +628,11 @@ taskq_dispatch(taskq_t *tq, task_func_t func, void *arg, uint_t flags)
 	spin_unlock(&t->tqent_lock);

 	wake_up(&tq->tq_work_waitq);
-out:
+
 	/* Spawn additional taskq threads if required. */
 	if (!(flags & TQ_NOQUEUE) && tq->tq_nactive == tq->tq_nthreads)
 		(void) taskq_thread_spawn(tq);
-
+out:
 	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
 	return (rc);
 }
@@ -676,10 +675,11 @@ taskq_dispatch_delay(taskq_t *tq, task_func_t func, void *arg,
 	ASSERT(!(t->tqent_flags & TQENT_FLAG_PREALLOC));

 	spin_unlock(&t->tqent_lock);
-out:
+
 	/* Spawn additional taskq threads if required. */
 	if (tq->tq_nactive == tq->tq_nthreads)
 		(void) taskq_thread_spawn(tq);
+out:
 	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
 	return (rc);
 }
@@ -704,9 +704,8 @@ taskq_dispatch_ent(taskq_t *tq, task_func_t func, void *arg, uint_t flags,

 	if ((flags & TQ_NOQUEUE) && (tq->tq_nactive == tq->tq_nthreads)) {
 		/* Dynamic taskq may be able to spawn another thread */
-		if (!(tq->tq_flags & TASKQ_DYNAMIC) ||
-		    taskq_thread_spawn(tq) == 0)
-			goto out2;
+		if (taskq_thread_spawn(tq) == 0)
+			goto out;
 		flags |= TQ_FRONT;
 	}

@@ -742,11 +741,11 @@ taskq_dispatch_ent(taskq_t *tq, task_func_t func, void *arg, uint_t flags,
 	spin_unlock(&t->tqent_lock);

 	wake_up(&tq->tq_work_waitq);
-out:
+
 	/* Spawn additional taskq threads if required. */
 	if (tq->tq_nactive == tq->tq_nthreads)
 		(void) taskq_thread_spawn(tq);
-out2:
+out:
 	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
 }
 EXPORT_SYMBOL(taskq_dispatch_ent);
@@ -825,6 +824,7 @@ taskq_thread_spawn(taskq_t *tq)
 	if (!(tq->tq_flags & TASKQ_DYNAMIC))
 		return (0);

+	tq->lastspawnstop = jiffies;
 	if ((tq->tq_nthreads + tq->tq_nspawn < tq->tq_maxthreads) &&
 	    (tq->tq_flags & TASKQ_ACTIVE)) {
 		spawning = (++tq->tq_nspawn);
@@ -836,9 +836,9 @@ taskq_thread_spawn(taskq_t *tq)
 }

 /*
- * Threads in a dynamic taskq should only exit once it has been completely
- * drained and no other threads are actively servicing tasks.  This prevents
- * threads from being created and destroyed more than is required.
+ * Threads in a dynamic taskq may exit once there is no more work to do.
+ * To prevent threads from being created and destroyed too often limit
+ * the exit rate to one per spl_taskq_thread_timeout_ms.
 *
 * The first thread is the thread list is treated as the primary thread.
 * There is nothing special about the primary thread but in order to avoid
@@ -847,44 +847,22 @@ taskq_thread_spawn(taskq_t *tq)
 static int
 taskq_thread_should_stop(taskq_t *tq, taskq_thread_t *tqt)
 {
-	if (!(tq->tq_flags & TASKQ_DYNAMIC))
+	ASSERT(!taskq_next_ent(tq));
+	if (!(tq->tq_flags & TASKQ_DYNAMIC) || !spl_taskq_thread_dynamic)
 		return (0);
-
+	if (!(tq->tq_flags & TASKQ_ACTIVE))
+		return (1);
 	if (list_first_entry(&(tq->tq_thread_list), taskq_thread_t,
 	    tqt_thread_list) == tqt)
 		return (0);
-
-	int no_work =
-	    ((tq->tq_nspawn == 0) &&	/* No threads are being spawned */
-	    (tq->tq_nactive == 0) &&	/* No threads are handling tasks */
-	    (tq->tq_nthreads > 1) &&	/* More than 1 thread is running */
-	    (!taskq_next_ent(tq)) &&	/* There are no pending tasks */
-	    (spl_taskq_thread_dynamic)); /* Dynamic taskqs are allowed */
-
-	/*
-	 * If we would have said stop before, let's instead wait a bit, maybe
-	 * we'll see more work come our way soon...
-	 */
-	if (no_work) {
-		/* if it's 0, we want the old behavior. */
-		/* if the taskq is being torn down, we also want to go away. */
-		if (spl_taskq_thread_timeout_ms == 0 ||
-		    !(tq->tq_flags & TASKQ_ACTIVE))
-			return (1);
-		unsigned long lasttime = tq->lastshouldstop;
-		if (lasttime > 0) {
-			if (time_after(jiffies, lasttime +
-			    msecs_to_jiffies(spl_taskq_thread_timeout_ms)))
-				return (1);
-			else
-				return (0);
-		} else {
-			tq->lastshouldstop = jiffies;
-		}
-	} else {
-		tq->lastshouldstop = 0;
-	}
-	return (0);
+	ASSERT3U(tq->tq_nthreads, >, 1);
+	if (tq->tq_nspawn != 0)
+		return (0);
+	if (time_before(jiffies, tq->lastspawnstop +
+	    msecs_to_jiffies(spl_taskq_thread_timeout_ms)))
+		return (0);
+	tq->lastspawnstop = jiffies;
+	return (1);
 }

 static int
@@ -935,10 +913,8 @@ taskq_thread(void *args)
 		if (list_empty(&tq->tq_pend_list) &&
 		    list_empty(&tq->tq_prio_list)) {

-			if (taskq_thread_should_stop(tq, tqt)) {
-				wake_up_all(&tq->tq_wait_waitq);
+			if (taskq_thread_should_stop(tq, tqt))
 				break;
-			}

 			add_wait_queue_exclusive(&tq->tq_work_waitq, &wait);
 			spin_unlock_irqrestore(&tq->tq_lock, flags);
@@ -1013,9 +989,6 @@ taskq_thread(void *args)
 			tqt->tqt_id = TASKQID_INVALID;
 			tqt->tqt_flags = 0;
 			wake_up_all(&tq->tq_wait_waitq);
-		} else {
-			if (taskq_thread_should_stop(tq, tqt))
-				break;
 		}

 		set_current_state(TASK_INTERRUPTIBLE);
@@ -1122,7 +1095,7 @@ taskq_create(const char *name, int threads_arg, pri_t pri,
 	tq->tq_flags = (flags | TASKQ_ACTIVE);
 	tq->tq_next_id = TASKQID_INITIAL;
 	tq->tq_lowest_id = TASKQID_INITIAL;
-	tq->lastshouldstop = 0;
+	tq->lastspawnstop = jiffies;
 	INIT_LIST_HEAD(&tq->tq_free_list);
 	INIT_LIST_HEAD(&tq->tq_pend_list);
 	INIT_LIST_HEAD(&tq->tq_prio_list);
@@ -25,6 +25,7 @@
 #include <sys/debug.h>
 #include <sys/types.h>
 #include <sys/sysmacros.h>
+#include <rpc/types.h>
 #include <rpc/xdr.h>

 /*
@@ -21,6 +21,7 @@
 /*
 * Copyright (c) 2014 by Chunwei Chen. All rights reserved.
 * Copyright (c) 2019 by Delphix. All rights reserved.
+ * Copyright (c) 2023, 2024, Klara Inc.
 */

 /*
@@ -59,7 +60,9 @@
 #include <sys/zfs_znode.h>
 #ifdef _KERNEL
 #include <linux/kmap_compat.h>
+#include <linux/mm_compat.h>
 #include <linux/scatterlist.h>
+#include <linux/version.h>
 #endif

 #ifdef _KERNEL
@@ -895,14 +898,9 @@ abd_iter_init(struct abd_iter *aiter, abd_t *abd)
 {
 	ASSERT(!abd_is_gang(abd));
 	abd_verify(abd);
+	memset(aiter, 0, sizeof (struct abd_iter));
 	aiter->iter_abd = abd;
-	aiter->iter_mapaddr = NULL;
-	aiter->iter_mapsize = 0;
-	aiter->iter_pos = 0;
-	if (abd_is_linear(abd)) {
-		aiter->iter_offset = 0;
-		aiter->iter_sg = NULL;
-	} else {
+	if (!abd_is_linear(abd)) {
 		aiter->iter_offset = ABD_SCATTER(abd).abd_offset;
 		aiter->iter_sg = ABD_SCATTER(abd).abd_sgl;
 	}
@@ -915,6 +913,7 @@ abd_iter_init(struct abd_iter *aiter, abd_t *abd)
 boolean_t
 abd_iter_at_end(struct abd_iter *aiter)
 {
+	ASSERT3U(aiter->iter_pos, <=, aiter->iter_abd->abd_size);
 	return (aiter->iter_pos == aiter->iter_abd->abd_size);
 }

@@ -926,8 +925,15 @@ abd_iter_at_end(struct abd_iter *aiter)
 void
 abd_iter_advance(struct abd_iter *aiter, size_t amount)
 {
+	/*
+	 * Ensure that last chunk is not in use. abd_iterate_*() must clear
+	 * this state (directly or abd_iter_unmap()) before advancing.
+	 */
 	ASSERT3P(aiter->iter_mapaddr, ==, NULL);
 	ASSERT0(aiter->iter_mapsize);
+	ASSERT3P(aiter->iter_page, ==, NULL);
+	ASSERT0(aiter->iter_page_doff);
+	ASSERT0(aiter->iter_page_dsize);

 	/* There's nothing left to advance to, so do nothing */
 	if (abd_iter_at_end(aiter))
@@ -1009,6 +1015,106 @@ abd_cache_reap_now(void)
 }

 #if defined(_KERNEL)
+/*
+ * Yield the next page struct and data offset and size within it, without
+ * mapping it into the address space.
+ */
+void
+abd_iter_page(struct abd_iter *aiter)
+{
+	if (abd_iter_at_end(aiter)) {
+		aiter->iter_page = NULL;
+		aiter->iter_page_doff = 0;
+		aiter->iter_page_dsize = 0;
+		return;
+	}
+
+	struct page *page;
+	size_t doff, dsize;
+
+	if (abd_is_linear(aiter->iter_abd)) {
+		ASSERT3U(aiter->iter_pos, ==, aiter->iter_offset);
+
+		/* memory address at iter_pos */
+		void *paddr = ABD_LINEAR_BUF(aiter->iter_abd) + aiter->iter_pos;
+
+		/* struct page for address */
+		page = is_vmalloc_addr(paddr) ?
+		    vmalloc_to_page(paddr) : virt_to_page(paddr);
+
+		/* offset of address within the page */
+		doff = offset_in_page(paddr);
+
+		/* total data remaining in abd from this position */
+		dsize = aiter->iter_abd->abd_size - aiter->iter_offset;
+	} else {
+		ASSERT(!abd_is_gang(aiter->iter_abd));
+
+		/* current scatter page */
+		page = sg_page(aiter->iter_sg);
+
+		/* position within page */
+		doff = aiter->iter_offset;
+
+		/* remaining data in scatterlist */
+		dsize = MIN(aiter->iter_sg->length - aiter->iter_offset,
+		    aiter->iter_abd->abd_size - aiter->iter_pos);
+	}
+	ASSERT(page);
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 5, 0)
+	if (PageTail(page)) {
+		/*
+		 * This page is part of a "compound page", which is a group of
+		 * pages that can be referenced from a single struct page *.
+		 * Its organised as a "head" page, followed by a series of
+		 * "tail" pages.
+		 *
+		 * In OpenZFS, compound pages are allocated using the
+		 * __GFP_COMP flag, which we get from scatter ABDs and SPL
+		 * vmalloc slabs (ie >16K allocations). So a great many of the
+		 * IO buffers we get are going to be of this type.
+		 *
+		 * The tail pages are just regular PAGE_SIZE pages, and can be
+		 * safely used as-is. However, the head page has length
+		 * covering itself and all the tail pages. If this ABD chunk
+		 * spans multiple pages, then we can use the head page and a
+		 * >PAGE_SIZE length, which is far more efficient.
+		 *
+		 * To do this, we need to adjust the offset to be counted from
+		 * the head page. struct page for compound pages are stored
+		 * contiguously, so we can just adjust by a simple offset.
+		 *
+		 * Before kernel 4.5, compound page heads were refcounted
+		 * separately, such that moving back to the head page would
+		 * require us to take a reference to it and releasing it once
+		 * we're completely finished with it. In practice, that means
+		 * when our caller is done with the ABD, which we have no
+		 * insight into from here. Rather than contort this API to
+		 * track head page references on such ancient kernels, we just
+		 * compile this block out and use the tail pages directly. This
+		 * is slightly less efficient, but makes everything far
+		 * simpler.
+		 */
+		struct page *head = compound_head(page);
+		doff += ((page - head) * PAGESIZE);
+		page = head;
+	}
+#endif
+
+	/* final page and position within it */
+	aiter->iter_page = page;
+	aiter->iter_page_doff = doff;
+
+	/* amount of data in the chunk, up to the end of the page */
+	aiter->iter_page_dsize = MIN(dsize, page_size(page) - doff);
+}
+
+/*
+ * Note: ABD BIO functions only needed to support vdev_classic. See comments in
+ * vdev_disk.c.
+ */
+
 /*
 * bio_nr_pages for ABD.
 * @off is the offset in @abd
@@ -1163,4 +1269,5 @@ MODULE_PARM_DESC(zfs_abd_scatter_min_size,
 module_param(zfs_abd_scatter_max_order, uint, 0644);
 MODULE_PARM_DESC(zfs_abd_scatter_max_order,
 	"Maximum order allocation used for a scatter ABD.");
-#endif
+
+#endif /* _KERNEL */
@@ -1610,11 +1610,8 @@ zfs_readdir(struct inode *ip, zpl_dir_context_t *ctx, cred_t *cr)
 		if (done)
 			break;

-		/* Prefetch znode */
-		if (prefetch) {
-			dmu_prefetch(os, objnum, 0, 0, 0,
-			    ZIO_PRIORITY_SYNC_READ);
-		}
+		if (prefetch)
+			dmu_prefetch_dnode(os, objnum, ZIO_PRIORITY_SYNC_READ);

 		/*
 		 * Move to the next entry, fill in the previous offset.
@@ -3792,11 +3789,8 @@ zfs_putpage(struct inode *ip, struct page *pp, struct writeback_control *wbc,
 	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
 	zfs_sa_upgrade_txholds(tx, zp);

-	err = dmu_tx_assign(tx, TXG_NOWAIT);
+	err = dmu_tx_assign(tx, TXG_WAIT);
 	if (err != 0) {
-		if (err == ERESTART)
-			dmu_tx_wait(tx);
-
 		dmu_tx_abort(tx);
 #ifdef HAVE_VFS_FILEMAP_DIRTY_FOLIO
 		filemap_dirty_folio(page_mapping(pp), page_folio(pp));
@@ -720,23 +720,23 @@ zpl_putpage(struct page *pp, struct writeback_control *wbc, void *data)
 {
 	boolean_t *for_sync = data;
 	fstrans_cookie_t cookie;
+	int ret;

 	ASSERT(PageLocked(pp));
 	ASSERT(!PageWriteback(pp));

 	cookie = spl_fstrans_mark();
-	(void) zfs_putpage(pp->mapping->host, pp, wbc, *for_sync);
+	ret = zfs_putpage(pp->mapping->host, pp, wbc, *for_sync);
 	spl_fstrans_unmark(cookie);

-	return (0);
+	return (ret);
 }

 #ifdef HAVE_WRITEPAGE_T_FOLIO
 static int
 zpl_putfolio(struct folio *pp, struct writeback_control *wbc, void *data)
 {
-	(void) zpl_putpage(&pp->page, wbc, data);
-	return (0);
+	return (zpl_putpage(&pp->page, wbc, data));
 }
 #endif

@@ -26,6 +26,9 @@
 #include <linux/compat.h>
 #endif
 #include <linux/fs.h>
+#ifdef HAVE_VFS_SPLICE_COPY_FILE_RANGE
+#include <linux/splice.h>
+#endif
 #include <sys/file.h>
 #include <sys/zfs_znode.h>
 #include <sys/zfs_vnops.h>
@@ -102,7 +105,7 @@ zpl_copy_file_range(struct file *src_file, loff_t src_off,
 	ret = zpl_clone_file_range_impl(src_file, src_off,
 	    dst_file, dst_off, len);

-#ifdef HAVE_VFS_GENERIC_COPY_FILE_RANGE
+#if defined(HAVE_VFS_GENERIC_COPY_FILE_RANGE)
 	/*
 	 * Since Linux 5.3 the filesystem driver is responsible for executing
 	 * an appropriate fallback, and a generic fallback function is provided.
@@ -111,6 +114,15 @@ zpl_copy_file_range(struct file *src_file, loff_t src_off,
 	    ret == -EAGAIN)
 		ret = generic_copy_file_range(src_file, src_off, dst_file,
 		    dst_off, len, flags);
+#elif defined(HAVE_VFS_SPLICE_COPY_FILE_RANGE)
+	/*
+	 * Since 6.8 the fallback function is called splice_copy_file_range
+	 * and has a slightly different signature.
+	 */
+	if (ret == -EOPNOTSUPP || ret == -EINVAL || ret == -EXDEV ||
+	    ret == -EAGAIN)
+		ret = splice_copy_file_range(src_file, src_off, dst_file,
+		    dst_off, len);
 #else
 	/*
 	 * Before Linux 5.3 the filesystem has to return -EOPNOTSUPP to signal
@@ -118,7 +130,7 @@ zpl_copy_file_range(struct file *src_file, loff_t src_off,
 	 */
 	if (ret == -EINVAL || ret == -EXDEV || ret == -EAGAIN)
 		ret = -EOPNOTSUPP;
-#endif /* HAVE_VFS_GENERIC_COPY_FILE_RANGE */
+#endif /* HAVE_VFS_GENERIC_COPY_FILE_RANGE || HAVE_VFS_SPLICE_COPY_FILE_RANGE */

 	return (ret);
 }
@@ -37,6 +37,7 @@
 #include <sys/spa_impl.h>
 #include <sys/zvol.h>
 #include <sys/zvol_impl.h>
+#include <cityhash.h>

 #include <linux/blkdev_compat.h>
 #include <linux/task_io_accounting_ops.h>
@@ -53,6 +54,12 @@ static unsigned int zvol_request_sync = 0;
 static unsigned int zvol_prefetch_bytes = (128 * 1024);
 static unsigned long zvol_max_discard_blocks = 16384;

+/*
+ * Switch taskq at multiple of 512 MB offset. This can be set to a lower value
+ * to utilize more threads for small files but may affect prefetch hits.
+ */
+#define	ZVOL_TASKQ_OFFSET_SHIFT 29
+
 #ifndef HAVE_BLKDEV_GET_ERESTARTSYS
 static unsigned int zvol_open_timeout_ms = 1000;
 #endif
@@ -76,6 +83,8 @@ static boolean_t zvol_use_blk_mq = B_FALSE;
 static unsigned int zvol_blk_mq_blocks_per_thread = 8;
 #endif

+static unsigned int zvol_num_taskqs = 0;
+
 #ifndef	BLKDEV_DEFAULT_RQ
 /* BLKDEV_MAX_RQ was renamed to BLKDEV_DEFAULT_RQ in the 5.16 kernel */
 #define	BLKDEV_DEFAULT_RQ BLKDEV_MAX_RQ
@@ -114,7 +123,11 @@ struct zvol_state_os {
 	boolean_t use_blk_mq;
 };

-static taskq_t *zvol_taskq;
+typedef struct zv_taskq {
+	uint_t tqs_cnt;
+	taskq_t **tqs_taskq;
+} zv_taskq_t;
+static zv_taskq_t zvol_taskqs;
 static struct ida zvol_ida;

 typedef struct zv_request_stack {
@@ -532,6 +545,22 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 	}

 	zv_request_task_t *task;
+	zv_taskq_t *ztqs = &zvol_taskqs;
+	uint_t blk_mq_hw_queue = 0;
+	uint_t tq_idx;
+	uint_t taskq_hash;
+#ifdef HAVE_BLK_MQ
+	if (rq)
+#ifdef HAVE_BLK_MQ_RQ_HCTX
+		blk_mq_hw_queue = rq->mq_hctx->queue_num;
+#else
+		blk_mq_hw_queue =
+		    rq->q->queue_hw_ctx[rq->q->mq_map[rq->cpu]]->queue_num;
+#endif
+#endif
+	taskq_hash = cityhash4((uintptr_t)zv, offset >> ZVOL_TASKQ_OFFSET_SHIFT,
+	    blk_mq_hw_queue, 0);
+	tq_idx = taskq_hash % ztqs->tqs_cnt;

 	if (rw == WRITE) {
 		if (unlikely(zv->zv_flags & ZVOL_RDONLY)) {
@@ -601,7 +630,7 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 				zvol_discard(&zvr);
 			} else {
 				task = zv_request_task_create(zvr);
-				taskq_dispatch_ent(zvol_taskq,
+				taskq_dispatch_ent(ztqs->tqs_taskq[tq_idx],
 				    zvol_discard_task, task, 0, &task->ent);
 			}
 		} else {
@@ -609,7 +638,7 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 				zvol_write(&zvr);
 			} else {
 				task = zv_request_task_create(zvr);
-				taskq_dispatch_ent(zvol_taskq,
+				taskq_dispatch_ent(ztqs->tqs_taskq[tq_idx],
 				    zvol_write_task, task, 0, &task->ent);
 			}
 		}
@@ -631,7 +660,7 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 			zvol_read(&zvr);
 		} else {
 			task = zv_request_task_create(zvr);
-			taskq_dispatch_ent(zvol_taskq,
+			taskq_dispatch_ent(ztqs->tqs_taskq[tq_idx],
 			    zvol_read_task, task, 0, &task->ent);
 		}
 	}
@@ -1053,6 +1082,16 @@ zvol_alloc_non_blk_mq(struct zvol_state_os *zso)
 	if (zso->zvo_disk == NULL)
 		return (1);

+	zso->zvo_disk->minors = ZVOL_MINORS;
+	zso->zvo_queue = zso->zvo_disk->queue;
+#elif defined(HAVE_BLK_ALLOC_DISK_2ARG)
+	struct gendisk *disk = blk_alloc_disk(NULL, NUMA_NO_NODE);
+	if (IS_ERR(disk)) {
+		zso->zvo_disk = NULL;
+		return (1);
+	}
+
+	zso->zvo_disk = disk;
 	zso->zvo_disk->minors = ZVOL_MINORS;
 	zso->zvo_queue = zso->zvo_disk->queue;
 #else
@@ -1103,6 +1142,17 @@ zvol_alloc_blk_mq(zvol_state_t *zv)
 	}
 	zso->zvo_queue = zso->zvo_disk->queue;
 	zso->zvo_disk->minors = ZVOL_MINORS;
+#elif defined(HAVE_BLK_ALLOC_DISK_2ARG)
+	struct gendisk *disk = blk_mq_alloc_disk(&zso->tag_set, NULL, zv);
+	if (IS_ERR(disk)) {
+		zso->zvo_disk = NULL;
+		blk_mq_free_tag_set(&zso->tag_set);
+		return (1);
+	}
+
+	zso->zvo_disk = disk;
+	zso->zvo_queue = zso->zvo_disk->queue;
+	zso->zvo_disk->minors = ZVOL_MINORS;
 #else
 	zso->zvo_disk = alloc_disk(ZVOL_MINORS);
 	if (zso->zvo_disk == NULL) {
@@ -1256,7 +1306,7 @@ zvol_os_free(zvol_state_t *zv)

 	del_gendisk(zv->zv_zso->zvo_disk);
 #if defined(HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS) && \
-	defined(HAVE_BLK_ALLOC_DISK)
+	(defined(HAVE_BLK_ALLOC_DISK) || defined(HAVE_BLK_ALLOC_DISK_2ARG))
 #if defined(HAVE_BLK_CLEANUP_DISK)
 	blk_cleanup_disk(zv->zv_zso->zvo_disk);
 #else
@@ -1313,6 +1363,13 @@ zvol_os_create_minor(const char *name)
 	if (idx < 0)
 		return (SET_ERROR(-idx));
 	minor = idx << ZVOL_MINOR_BITS;
+	if (MINOR(minor) != minor) {
+		/* too many partitions can cause an overflow */
+		zfs_dbgmsg("zvol: create minor overflow: %s, minor %u/%u",
+		    name, minor, MINOR(minor));
+		ida_simple_remove(&zvol_ida, idx);
+		return (SET_ERROR(EINVAL));
+	}

 	zv = zvol_find_by_name_hash(name, hash, RW_NONE);
 	if (zv) {
@@ -1507,7 +1564,7 @@ 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));

@@ -1563,8 +1620,40 @@ zvol_init(void)
 		zvol_actual_threads = MIN(MAX(zvol_threads, 1), 1024);
 	}

+	/*
+	 * Use atleast 32 zvol_threads but for many core system,
+	 * prefer 6 threads per taskq, but no more taskqs
+	 * than threads in them on large systems.
+	 *
+	 *                 taskq   total
+	 * cpus    taskqs  threads threads
+	 * ------- ------- ------- -------
+	 * 1       1       32       32
+	 * 2       1       32       32
+	 * 4       1       32       32
+	 * 8       2       16       32
+	 * 16      3       11       33
+	 * 32      5       7        35
+	 * 64      8       8        64
+	 * 128     11      12       132
+	 * 256     16      16       256
+	 */
+	zv_taskq_t *ztqs = &zvol_taskqs;
+	uint_t num_tqs = MIN(num_online_cpus(), zvol_num_taskqs);
+	if (num_tqs == 0) {
+		num_tqs = 1 + num_online_cpus() / 6;
+		while (num_tqs * num_tqs > zvol_actual_threads)
+			num_tqs--;
+	}
+	uint_t per_tq_thread = zvol_actual_threads / num_tqs;
+	if (per_tq_thread * num_tqs < zvol_actual_threads)
+		per_tq_thread++;
+	ztqs->tqs_cnt = num_tqs;
+	ztqs->tqs_taskq = kmem_alloc(num_tqs * sizeof (taskq_t *), KM_SLEEP);
 	error = register_blkdev(zvol_major, ZVOL_DRIVER);
 	if (error) {
+		kmem_free(ztqs->tqs_taskq, ztqs->tqs_cnt * sizeof (taskq_t *));
+		ztqs->tqs_taskq = NULL;
 		printk(KERN_INFO "ZFS: register_blkdev() failed %d\n", error);
 		return (error);
 	}
@@ -1584,11 +1673,22 @@ zvol_init(void)
 		    1024);
 	}
 #endif
-	zvol_taskq = taskq_create(ZVOL_DRIVER, zvol_actual_threads, maxclsyspri,
-	    zvol_actual_threads, INT_MAX, TASKQ_PREPOPULATE | TASKQ_DYNAMIC);
-	if (zvol_taskq == NULL) {
-		unregister_blkdev(zvol_major, ZVOL_DRIVER);
-		return (-ENOMEM);
+	for (uint_t i = 0; i < num_tqs; i++) {
+		char name[32];
+		(void) snprintf(name, sizeof (name), "%s_tq-%u",
+		    ZVOL_DRIVER, i);
+		ztqs->tqs_taskq[i] = taskq_create(name, per_tq_thread,
+		    maxclsyspri, per_tq_thread, INT_MAX,
+		    TASKQ_PREPOPULATE | TASKQ_DYNAMIC);
+		if (ztqs->tqs_taskq[i] == NULL) {
+			for (int j = i - 1; j >= 0; j--)
+				taskq_destroy(ztqs->tqs_taskq[j]);
+			unregister_blkdev(zvol_major, ZVOL_DRIVER);
+			kmem_free(ztqs->tqs_taskq, ztqs->tqs_cnt *
+			    sizeof (taskq_t *));
+			ztqs->tqs_taskq = NULL;
+			return (-ENOMEM);
+		}
 	}

 	zvol_init_impl();
@@ -1599,9 +1699,22 @@ zvol_init(void)
 void
 zvol_fini(void)
 {
+	zv_taskq_t *ztqs = &zvol_taskqs;
 	zvol_fini_impl();
 	unregister_blkdev(zvol_major, ZVOL_DRIVER);
-	taskq_destroy(zvol_taskq);
+
+	if (ztqs->tqs_taskq == NULL) {
+		ASSERT3U(ztqs->tqs_cnt, ==, 0);
+	} else {
+		for (uint_t i = 0; i < ztqs->tqs_cnt; i++) {
+			ASSERT3P(ztqs->tqs_taskq[i], !=, NULL);
+			taskq_destroy(ztqs->tqs_taskq[i]);
+		}
+		kmem_free(ztqs->tqs_taskq, ztqs->tqs_cnt *
+		    sizeof (taskq_t *));
+		ztqs->tqs_taskq = NULL;
+	}
+
 	ida_destroy(&zvol_ida);
 }

@@ -1622,6 +1735,9 @@ MODULE_PARM_DESC(zvol_request_sync, "Synchronously handle bio requests");
 module_param(zvol_max_discard_blocks, ulong, 0444);
 MODULE_PARM_DESC(zvol_max_discard_blocks, "Max number of blocks to discard");

+module_param(zvol_num_taskqs, uint, 0444);
+MODULE_PARM_DESC(zvol_num_taskqs, "Number of zvol taskqs");
+
 module_param(zvol_prefetch_bytes, uint, 0644);
 MODULE_PARM_DESC(zvol_prefetch_bytes, "Prefetch N bytes at zvol start+end");

@@ -345,6 +345,13 @@ zfs_prop_init(void)
 		{ NULL }
 	};

+	static const zprop_index_t prefetch_table[] = {
+		{ "none",	ZFS_PREFETCH_NONE },
+		{ "metadata",	ZFS_PREFETCH_METADATA },
+		{ "all",	ZFS_PREFETCH_ALL },
+		{ NULL }
+	};
+
 	static const zprop_index_t sync_table[] = {
 		{ "standard",	ZFS_SYNC_STANDARD },
 		{ "always",	ZFS_SYNC_ALWAYS },
@@ -453,6 +460,10 @@ zfs_prop_init(void)
 	    ZFS_CACHE_ALL, PROP_INHERIT,
 	    ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT | ZFS_TYPE_VOLUME,
 	    "all | none | metadata", "SECONDARYCACHE", cache_table, sfeatures);
+	zprop_register_index(ZFS_PROP_PREFETCH, "prefetch",
+	    ZFS_PREFETCH_ALL, PROP_INHERIT,
+	    ZFS_TYPE_FILESYSTEM | ZFS_TYPE_SNAPSHOT | ZFS_TYPE_VOLUME,
+	    "none | metadata | all", "PREFETCH", prefetch_table, sfeatures);
 	zprop_register_index(ZFS_PROP_LOGBIAS, "logbias", ZFS_LOGBIAS_LATENCY,
 	    PROP_INHERIT, ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME,
 	    "latency | throughput", "LOGBIAS", logbias_table, sfeatures);
@@ -431,6 +431,12 @@ vdev_prop_init(void)
 	zprop_register_number(VDEV_PROP_IO_T, "io_t", UINT64_MAX,
 	    PROP_DEFAULT, ZFS_TYPE_VDEV, "<seconds>", "IO_T", B_FALSE,
 	    sfeatures);
+	zprop_register_number(VDEV_PROP_SLOW_IO_N, "slow_io_n", UINT64_MAX,
+	    PROP_DEFAULT, ZFS_TYPE_VDEV, "<events>", "SLOW_IO_N", B_FALSE,
+	    sfeatures);
+	zprop_register_number(VDEV_PROP_SLOW_IO_T, "slow_io_t", UINT64_MAX,
+	    PROP_DEFAULT, ZFS_TYPE_VDEV, "<seconds>", "SLOW_IO_T", B_FALSE,
+	    sfeatures);

 	/* default index (boolean) properties */
 	zprop_register_index(VDEV_PROP_REMOVING, "removing", 0,
@@ -826,6 +826,48 @@ abd_iterate_func(abd_t *abd, size_t off, size_t size,
 	return (ret);
 }

+#if defined(__linux__) && defined(_KERNEL)
+int
+abd_iterate_page_func(abd_t *abd, size_t off, size_t size,
+    abd_iter_page_func_t *func, void *private)
+{
+	struct abd_iter aiter;
+	int ret = 0;
+
+	if (size == 0)
+		return (0);
+
+	abd_verify(abd);
+	ASSERT3U(off + size, <=, abd->abd_size);
+
+	abd_t *c_abd = abd_init_abd_iter(abd, &aiter, off);
+
+	while (size > 0) {
+		IMPLY(abd_is_gang(abd), c_abd != NULL);
+
+		abd_iter_page(&aiter);
+
+		size_t len = MIN(aiter.iter_page_dsize, size);
+		ASSERT3U(len, >, 0);
+
+		ret = func(aiter.iter_page, aiter.iter_page_doff,
+		    len, private);
+
+		aiter.iter_page = NULL;
+		aiter.iter_page_doff = 0;
+		aiter.iter_page_dsize = 0;
+
+		if (ret != 0)
+			break;
+
+		size -= len;
+		c_abd = abd_advance_abd_iter(abd, c_abd, &aiter, len);
+	}
+
+	return (ret);
+}
+#endif
+
 struct buf_arg {
 	void *arg_buf;
 };
@@ -3872,7 +3872,7 @@ arc_evict_state_impl(multilist_t *ml, int idx, arc_buf_hdr_t *marker,

 	ASSERT3P(marker, !=, NULL);

-	mls = multilist_sublist_lock(ml, idx);
+	mls = multilist_sublist_lock_idx(ml, idx);

 	for (hdr = multilist_sublist_prev(mls, marker); likely(hdr != NULL);
 	    hdr = multilist_sublist_prev(mls, marker)) {
@@ -3984,6 +3984,26 @@ arc_evict_state_impl(multilist_t *ml, int idx, arc_buf_hdr_t *marker,
 	return (bytes_evicted);
 }

+static arc_buf_hdr_t *
+arc_state_alloc_marker(void)
+{
+	arc_buf_hdr_t *marker = kmem_cache_alloc(hdr_full_cache, KM_SLEEP);
+
+	/*
+	 * A b_spa of 0 is used to indicate that this header is
+	 * a marker. This fact is used in arc_evict_state_impl().
+	 */
+	marker->b_spa = 0;
+
+	return (marker);
+}
+
+static void
+arc_state_free_marker(arc_buf_hdr_t *marker)
+{
+	kmem_cache_free(hdr_full_cache, marker);
+}
+
 /*
 * Allocate an array of buffer headers used as placeholders during arc state
 * eviction.
@@ -3994,16 +4014,8 @@ arc_state_alloc_markers(int count)
 	arc_buf_hdr_t **markers;

 	markers = kmem_zalloc(sizeof (*markers) * count, KM_SLEEP);
-	for (int i = 0; i < count; i++) {
-		markers[i] = kmem_cache_alloc(hdr_full_cache, KM_SLEEP);
-
-		/*
-		 * A b_spa of 0 is used to indicate that this header is
-		 * a marker. This fact is used in arc_evict_state_impl().
-		 */
-		markers[i]->b_spa = 0;
-
-	}
+	for (int i = 0; i < count; i++)
+		markers[i] = arc_state_alloc_marker();
 	return (markers);
 }

@@ -4011,7 +4023,7 @@ static void
 arc_state_free_markers(arc_buf_hdr_t **markers, int count)
 {
 	for (int i = 0; i < count; i++)
-		kmem_cache_free(hdr_full_cache, markers[i]);
+		arc_state_free_marker(markers[i]);
 	kmem_free(markers, sizeof (*markers) * count);
 }

@@ -4055,7 +4067,7 @@ arc_evict_state(arc_state_t *state, arc_buf_contents_t type, uint64_t spa,
 	for (int i = 0; i < num_sublists; i++) {
 		multilist_sublist_t *mls;

-		mls = multilist_sublist_lock(ml, i);
+		mls = multilist_sublist_lock_idx(ml, i);
 		multilist_sublist_insert_tail(mls, markers[i]);
 		multilist_sublist_unlock(mls);
 	}
@@ -4120,7 +4132,7 @@ arc_evict_state(arc_state_t *state, arc_buf_contents_t type, uint64_t spa,
 	}

 	for (int i = 0; i < num_sublists; i++) {
-		multilist_sublist_t *mls = multilist_sublist_lock(ml, i);
+		multilist_sublist_t *mls = multilist_sublist_lock_idx(ml, i);
 		multilist_sublist_remove(mls, markers[i]);
 		multilist_sublist_unlock(mls);
 	}
@@ -8628,7 +8640,7 @@ l2arc_sublist_lock(int list_num)
 	 * sublists being selected.
 	 */
 	idx = multilist_get_random_index(ml);
-	return (multilist_sublist_lock(ml, idx));
+	return (multilist_sublist_lock_idx(ml, idx));
 }

 /*
@@ -9040,9 +9052,9 @@ l2arc_blk_fetch_done(zio_t *zio)
 static uint64_t
 l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 {
-	arc_buf_hdr_t 		*hdr, *hdr_prev, *head;
-	uint64_t 		write_asize, write_psize, write_lsize, headroom;
-	boolean_t		full;
+	arc_buf_hdr_t 		*hdr, *head, *marker;
+	uint64_t 		write_asize, write_psize, headroom;
+	boolean_t		full, from_head = !arc_warm;
 	l2arc_write_callback_t	*cb = NULL;
 	zio_t 			*pio, *wzio;
 	uint64_t 		guid = spa_load_guid(spa);
@@ -9051,10 +9063,11 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 	ASSERT3P(dev->l2ad_vdev, !=, NULL);

 	pio = NULL;
-	write_lsize = write_asize = write_psize = 0;
+	write_asize = write_psize = 0;
 	full = B_FALSE;
 	head = kmem_cache_alloc(hdr_l2only_cache, KM_PUSHPAGE);
 	arc_hdr_set_flags(head, ARC_FLAG_L2_WRITE_HEAD | ARC_FLAG_HAS_L2HDR);
+	marker = arc_state_alloc_marker();

 	/*
 	 * Copy buffers for L2ARC writing.
@@ -9069,40 +9082,34 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 				continue;
 		}

-		multilist_sublist_t *mls = l2arc_sublist_lock(pass);
 		uint64_t passed_sz = 0;
-
-		VERIFY3P(mls, !=, NULL);
-
-		/*
-		 * L2ARC fast warmup.
-		 *
-		 * Until the ARC is warm and starts to evict, read from the
-		 * head of the ARC lists rather than the tail.
-		 */
-		if (arc_warm == B_FALSE)
-			hdr = multilist_sublist_head(mls);
-		else
-			hdr = multilist_sublist_tail(mls);
-
 		headroom = target_sz * l2arc_headroom;
 		if (zfs_compressed_arc_enabled)
 			headroom = (headroom * l2arc_headroom_boost) / 100;

-		for (; hdr; hdr = hdr_prev) {
+		/*
+		 * Until the ARC is warm and starts to evict, read from the
+		 * head of the ARC lists rather than the tail.
+		 */
+		multilist_sublist_t *mls = l2arc_sublist_lock(pass);
+		ASSERT3P(mls, !=, NULL);
+		if (from_head)
+			hdr = multilist_sublist_head(mls);
+		else
+			hdr = multilist_sublist_tail(mls);
+
+		while (hdr != NULL) {
 			kmutex_t *hash_lock;
 			abd_t *to_write = NULL;

-			if (arc_warm == B_FALSE)
-				hdr_prev = multilist_sublist_next(mls, hdr);
-			else
-				hdr_prev = multilist_sublist_prev(mls, hdr);
-
 			hash_lock = HDR_LOCK(hdr);
 			if (!mutex_tryenter(hash_lock)) {
-				/*
-				 * Skip this buffer rather than waiting.
-				 */
+skip:
+				/* Skip this buffer rather than waiting. */
+				if (from_head)
+					hdr = multilist_sublist_next(mls, hdr);
+				else
+					hdr = multilist_sublist_prev(mls, hdr);
 				continue;
 			}

@@ -9117,11 +9124,10 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)

 			if (!l2arc_write_eligible(guid, hdr)) {
 				mutex_exit(hash_lock);
-				continue;
+				goto skip;
 			}

 			ASSERT(HDR_HAS_L1HDR(hdr));
-
 			ASSERT3U(HDR_GET_PSIZE(hdr), >, 0);
 			ASSERT3U(arc_hdr_size(hdr), >, 0);
 			ASSERT(hdr->b_l1hdr.b_pabd != NULL ||
@@ -9143,12 +9149,18 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 			}

 			/*
-			 * We rely on the L1 portion of the header below, so
-			 * it's invalid for this header to have been evicted out
-			 * of the ghost cache, prior to being written out. The
-			 * ARC_FLAG_L2_WRITING bit ensures this won't happen.
+			 * We should not sleep with sublist lock held or it
+			 * may block ARC eviction.  Insert a marker to save
+			 * the position and drop the lock.
 			 */
-			arc_hdr_set_flags(hdr, ARC_FLAG_L2_WRITING);
+			if (from_head) {
+				multilist_sublist_insert_after(mls, hdr,
+				    marker);
+			} else {
+				multilist_sublist_insert_before(mls, hdr,
+				    marker);
+			}
+			multilist_sublist_unlock(mls);

 			/*
 			 * If this header has b_rabd, we can use this since it
@@ -9179,32 +9191,45 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 				    &to_write);
 				if (ret != 0) {
 					arc_hdr_clear_flags(hdr,
-					    ARC_FLAG_L2_WRITING);
+					    ARC_FLAG_L2CACHE);
 					mutex_exit(hash_lock);
-					continue;
+					goto next;
 				}

 				l2arc_free_abd_on_write(to_write, asize, type);
 			}

+			hdr->b_l2hdr.b_dev = dev;
+			hdr->b_l2hdr.b_daddr = dev->l2ad_hand;
+			hdr->b_l2hdr.b_hits = 0;
+			hdr->b_l2hdr.b_arcs_state =
+			    hdr->b_l1hdr.b_state->arcs_state;
+			mutex_enter(&dev->l2ad_mtx);
 			if (pio == NULL) {
 				/*
 				 * Insert a dummy header on the buflist so
 				 * l2arc_write_done() can find where the
 				 * write buffers begin without searching.
 				 */
-				mutex_enter(&dev->l2ad_mtx);
 				list_insert_head(&dev->l2ad_buflist, head);
-				mutex_exit(&dev->l2ad_mtx);
+			}
+			list_insert_head(&dev->l2ad_buflist, hdr);
+			mutex_exit(&dev->l2ad_mtx);
+			arc_hdr_set_flags(hdr, ARC_FLAG_HAS_L2HDR |
+			    ARC_FLAG_L2_WRITING);

+			(void) zfs_refcount_add_many(&dev->l2ad_alloc,
+			    arc_hdr_size(hdr), hdr);
+			l2arc_hdr_arcstats_increment(hdr);
+
+			boolean_t commit = l2arc_log_blk_insert(dev, hdr);
+			mutex_exit(hash_lock);
+
+			if (pio == NULL) {
 				cb = kmem_alloc(
 				    sizeof (l2arc_write_callback_t), KM_SLEEP);
 				cb->l2wcb_dev = dev;
 				cb->l2wcb_head = head;
-				/*
-				 * Create a list to save allocated abd buffers
-				 * for l2arc_log_blk_commit().
-				 */
 				list_create(&cb->l2wcb_abd_list,
 				    sizeof (l2arc_lb_abd_buf_t),
 				    offsetof(l2arc_lb_abd_buf_t, node));
@@ -9212,54 +9237,34 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 				    ZIO_FLAG_CANFAIL);
 			}

-			hdr->b_l2hdr.b_dev = dev;
-			hdr->b_l2hdr.b_hits = 0;
-
-			hdr->b_l2hdr.b_daddr = dev->l2ad_hand;
-			hdr->b_l2hdr.b_arcs_state =
-			    hdr->b_l1hdr.b_state->arcs_state;
-			arc_hdr_set_flags(hdr, ARC_FLAG_HAS_L2HDR);
-
-			mutex_enter(&dev->l2ad_mtx);
-			list_insert_head(&dev->l2ad_buflist, hdr);
-			mutex_exit(&dev->l2ad_mtx);
-
-			(void) zfs_refcount_add_many(&dev->l2ad_alloc,
-			    arc_hdr_size(hdr), hdr);
-
 			wzio = zio_write_phys(pio, dev->l2ad_vdev,
-			    hdr->b_l2hdr.b_daddr, asize, to_write,
+			    dev->l2ad_hand, asize, to_write,
 			    ZIO_CHECKSUM_OFF, NULL, hdr,
 			    ZIO_PRIORITY_ASYNC_WRITE,
 			    ZIO_FLAG_CANFAIL, B_FALSE);

-			write_lsize += HDR_GET_LSIZE(hdr);
 			DTRACE_PROBE2(l2arc__write, vdev_t *, dev->l2ad_vdev,
 			    zio_t *, wzio);
+			zio_nowait(wzio);

 			write_psize += psize;
 			write_asize += asize;
 			dev->l2ad_hand += asize;
-			l2arc_hdr_arcstats_increment(hdr);
 			vdev_space_update(dev->l2ad_vdev, asize, 0, 0);

-			mutex_exit(hash_lock);
-
-			/*
-			 * Append buf info to current log and commit if full.
-			 * arcstat_l2_{size,asize} kstats are updated
-			 * internally.
-			 */
-			if (l2arc_log_blk_insert(dev, hdr)) {
-				/*
-				 * l2ad_hand will be adjusted in
-				 * l2arc_log_blk_commit().
-				 */
+			if (commit) {
+				/* l2ad_hand will be adjusted inside. */
 				write_asize +=
 				    l2arc_log_blk_commit(dev, pio, cb);
 			}

-			zio_nowait(wzio);
+next:
+			multilist_sublist_lock(mls);
+			if (from_head)
+				hdr = multilist_sublist_next(mls, marker);
+			else
+				hdr = multilist_sublist_prev(mls, marker);
+			multilist_sublist_remove(mls, marker);
 		}

 		multilist_sublist_unlock(mls);
@@ -9268,9 +9273,11 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
 			break;
 	}

+	arc_state_free_marker(marker);
+
 	/* No buffers selected for writing? */
 	if (pio == NULL) {
-		ASSERT0(write_lsize);
+		ASSERT0(write_psize);
 		ASSERT(!HDR_HAS_L1HDR(head));
 		kmem_cache_free(hdr_l2only_cache, head);

@@ -10598,7 +10605,7 @@ l2arc_log_blk_insert(l2arc_dev_t *dev, const arc_buf_hdr_t *hdr)
 	L2BLK_SET_TYPE((le)->le_prop, hdr->b_type);
 	L2BLK_SET_PROTECTED((le)->le_prop, !!(HDR_PROTECTED(hdr)));
 	L2BLK_SET_PREFETCH((le)->le_prop, !!(HDR_PREFETCH(hdr)));
-	L2BLK_SET_STATE((le)->le_prop, hdr->b_l1hdr.b_state->arcs_state);
+	L2BLK_SET_STATE((le)->le_prop, hdr->b_l2hdr.b_arcs_state);

 	dev->l2ad_log_blk_payload_asize += vdev_psize_to_asize(dev->l2ad_vdev,
 	    HDR_GET_PSIZE(hdr));
--- a/Show More
+++ b/Show More