From 4f33cd2350340e9a8c0321b90d2de042d739166c Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Fri, 5 Sep 2025 09:08:15 -0700
Subject: [PATCH 01/23] CI: Increase 'Setup QEMU' timeout to 15 minutes

We've seen Fedora 42 still setting up after 10 min.  Change the timeout
to 15 min.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #17697
---
 .github/workflows/zfs-qemu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/zfs-qemu.yml b/.github/workflows/zfs-qemu.yml
index 4ebb80af1f03..a071c26a09b5 100644
--- a/.github/workflows/zfs-qemu.yml
+++ b/.github/workflows/zfs-qemu.yml
@@ -78,7 +78,7 @@ jobs:
         ref: ${{ github.event.pull_request.head.sha }}
 
     - name: Setup QEMU
-      timeout-minutes: 10
+      timeout-minutes: 15
       run: .github/workflows/scripts/qemu-1-setup.sh
 
     - name: Start build machine

From 0c9302b04209afb89d679bcd6d3d2466156e1eb0 Mon Sep 17 00:00:00 2001
From: Shengqi Chen <harry-chen@outlook.com>
Date: Thu, 4 Sep 2025 11:09:58 +0800
Subject: [PATCH 02/23] ci: use real head sha instead of GITHUB_SHA when
 generating CI type

Because GitHub creates a merge commit on top of real head, so the check
on HEAD will fail regardlessly.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Shengqi Chen <harry-chen@outlook.com>
Closes #17695
---
 .github/workflows/scripts/generate-ci-type.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/scripts/generate-ci-type.py b/.github/workflows/scripts/generate-ci-type.py
index b49255e8381d..08021aabcb61 100755
--- a/.github/workflows/scripts/generate-ci-type.py
+++ b/.github/workflows/scripts/generate-ci-type.py
@@ -65,7 +65,7 @@ def output_type(type, reason):
 
     # check last (HEAD) commit message
     last_commit_message_raw = subprocess.run([
-        'git', 'show', '-s', '--format=%B', 'HEAD'
+        'git', 'show', '-s', '--format=%B', head
     ], check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 
     for line in last_commit_message_raw.stdout.decode().splitlines():

From f4a85d6f2c9edf4ee77c464a7bc5855e6204514c Mon Sep 17 00:00:00 2001
From: Shengqi Chen <harry-chen@outlook.com>
Date: Thu, 4 Sep 2025 11:18:01 +0800
Subject: [PATCH 03/23] ci: fix syntax issues in zfs-qemu.yml

Otherwise it might become `if [ == "" ]` which is ill-formed.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Shengqi Chen <harry-chen@outlook.com>
Closes #17695
---
 .github/workflows/zfs-qemu.yml | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/.github/workflows/zfs-qemu.yml b/.github/workflows/zfs-qemu.yml
index a071c26a09b5..a5dbfc099c90 100644
--- a/.github/workflows/zfs-qemu.yml
+++ b/.github/workflows/zfs-qemu.yml
@@ -44,7 +44,7 @@ jobs:
             os_selection="$FULL_OS"
           fi
 
-          if [ ${{ github.event.inputs.fedora_kernel_ver }} != "" ] ; then
+          if ${{ github.event.inputs.fedora_kernel_ver != '' }}; then
               # They specified a custom kernel version for Fedora.  Use only
               # Fedora runners.
               os_json=$(echo ${os_selection} | jq -c '[.[] | select(startswith("fedora"))]')
@@ -53,9 +53,8 @@ jobs:
               os_json=$(echo ${os_selection} | jq -c)
           fi
 
-          echo $os_json
-          echo "os=$os_json" >> $GITHUB_OUTPUT
-          echo "ci_type=$ci_type" >> $GITHUB_OUTPUT
+          echo "os=$os_json" | tee -a $GITHUB_OUTPUT
+          echo "ci_type=$ci_type" | tee -a $GITHUB_OUTPUT
 
   qemu-vm:
     name: qemu-x86

From cb66796f8741cec35c9bba1a816ccd408b799a74 Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Wed, 10 Sep 2025 10:25:58 -0700
Subject: [PATCH 04/23] CI: Increase setup timeout to 20min, add timestamps

- Increase qemu-1-setup.sh timeout to 20min since it sometimes
  fails to complete after 15min.

- Timestamp all qemu-1-setup.sh lines to look for hangs.

- Add a 'watchdog' process to print out the top running process every
  30sec to help with debugging.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #17714
---
 .github/workflows/scripts/qemu-1-setup.sh | 10 ++++++++++
 .github/workflows/zfs-qemu.yml            |  8 ++++++--
 2 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/scripts/qemu-1-setup.sh b/.github/workflows/scripts/qemu-1-setup.sh
index de29ad1f57b6..0278264d9279 100755
--- a/.github/workflows/scripts/qemu-1-setup.sh
+++ b/.github/workflows/scripts/qemu-1-setup.sh
@@ -6,6 +6,13 @@
 
 set -eu
 
+# We've been seeing this script take over 15min to run.  This may or
+# may not be normal.  Just to get a little more insight, print out
+# a message to stdout with the top running process, and do this every
+# 30 seconds.  We can delete this watchdog later once we get a better
+# handle on what the timeout value should be.
+(while [ 1 ] ; do sleep 30 && echo "[watchdog: $(ps -eo cmd --sort=-pcpu  | head -n 2 | tail -n 1)}')]"; done) &
+
 # install needed packages
 export DEBIAN_FRONTEND="noninteractive"
 sudo apt-get -y update
@@ -65,3 +72,6 @@ sudo zpool create -f -o ashift=12 zpool $SSD1 $SSD2 -O relatime=off \
 for i in /sys/block/s*/queue/scheduler; do
   echo "none" | sudo tee $i
 done
+
+# Kill off our watchdog
+kill $(jobs -p)
diff --git a/.github/workflows/zfs-qemu.yml b/.github/workflows/zfs-qemu.yml
index a5dbfc099c90..69349678d84c 100644
--- a/.github/workflows/zfs-qemu.yml
+++ b/.github/workflows/zfs-qemu.yml
@@ -77,8 +77,12 @@ jobs:
         ref: ${{ github.event.pull_request.head.sha }}
 
     - name: Setup QEMU
-      timeout-minutes: 15
-      run: .github/workflows/scripts/qemu-1-setup.sh
+      timeout-minutes: 20
+      run: |
+        # Add a timestamp to each line to debug timeouts
+        while IFS=$'\n' read -r line; do
+            echo "$(date +'%H:%M:%S') $line"
+        done < <(.github/workflows/scripts/qemu-1-setup.sh)
 
     - name: Start build machine
       timeout-minutes: 10

From a83eefee1ec5ab8d0e62bca7feba55c2163fe2fd Mon Sep 17 00:00:00 2001
From: Alexander Motin <alexander.motin@TrueNAS.com>
Date: Mon, 15 Sep 2025 15:15:31 -0400
Subject: [PATCH 05/23] CI: Switch FreeBSD 15 to 15.0-ALPHA2

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Alexander Motin <alexander.motin@TrueNAS.com>
Closes #17749
---
 .github/workflows/scripts/qemu-2-start.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/scripts/qemu-2-start.sh b/.github/workflows/scripts/qemu-2-start.sh
index 62e06926e268..8439942c5a41 100755
--- a/.github/workflows/scripts/qemu-2-start.sh
+++ b/.github/workflows/scripts/qemu-2-start.sh
@@ -121,7 +121,7 @@ case "$OS" in
     KSRC="$FREEBSD_SNAP/../amd64/$FreeBSD/src.txz"
     ;;
   freebsd15-0c)
-    FreeBSD="15.0-PRERELEASE"
+    FreeBSD="15.0-ALPHA2"
     OSNAME="FreeBSD $FreeBSD"
     OSv="freebsd14.0"
     URLxz="$FREEBSD_SNAP/$FreeBSD/amd64/Latest/FreeBSD-$FreeBSD-amd64-BASIC-CI-ufs.raw.xz"

From 78b21d8d1953f65c164025588f15b366ad06069c Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Thu, 25 Sep 2025 17:47:32 -0700
Subject: [PATCH 06/23] CI: update perf and bpftools with the kernel packages

When updating a Fedora instance to an experimental kernel make sure
to include the matching versioned perf and bpftool packages.  This
helps ensure there are no unexpected conflicts which would prevent
the new packages from being installed.

Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #17791
---
 .github/workflows/scripts/qemu-3-deps-vm.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/scripts/qemu-3-deps-vm.sh b/.github/workflows/scripts/qemu-3-deps-vm.sh
index ee058b488088..4a7e724586f0 100755
--- a/.github/workflows/scripts/qemu-3-deps-vm.sh
+++ b/.github/workflows/scripts/qemu-3-deps-vm.sh
@@ -104,7 +104,7 @@ function install_fedora_experimental_kernel {
   our_version="$1"
   sudo dnf -y copr enable @kernel-vanilla/stable
   sudo dnf -y copr enable @kernel-vanilla/mainline
-  all="$(sudo dnf list --showduplicates kernel-*)"
+  all="$(sudo dnf list --showduplicates kernel-* python3-perf* perf* bpftool*)"
   echo "Available versions:"
   echo "$all"
 

From 395f0993239822c706c6e8397561e90d6d927573 Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Fri, 26 Sep 2025 15:32:41 -0700
Subject: [PATCH 07/23] CI: Remove Buildbot references

The Buildbot CI infrastructure has been fully replaced by GitHub
Actions.  Remove any lingering references from the repository.

Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #17794
---
 .github/PULL_REQUEST_TEMPLATE.md                   | 5 -----
 contrib/pyzfs/libzfs_core/test/test_libzfs_core.py | 2 +-
 2 files changed, 1 insertion(+), 6 deletions(-)

diff --git a/.github/PULL_REQUEST_TEMPLATE.md b/.github/PULL_REQUEST_TEMPLATE.md
index 79809179cf13..47edc8174603 100644
--- a/.github/PULL_REQUEST_TEMPLATE.md
+++ b/.github/PULL_REQUEST_TEMPLATE.md
@@ -2,11 +2,6 @@
 
 <!--- Provide a general summary of your changes in the Title above -->
 
-<!---
-Documentation on ZFS Buildbot options can be found at
-https://openzfs.github.io/openzfs-docs/Developer%20Resources/Buildbot%20Options.html
--->
-
 ### Motivation and Context
 <!--- Why is this change required? What problem does it solve? -->
 <!--- If it fixes an open issue, please link to the issue here. -->
diff --git a/contrib/pyzfs/libzfs_core/test/test_libzfs_core.py b/contrib/pyzfs/libzfs_core/test/test_libzfs_core.py
index 971aa1d0d493..bad1af2d1671 100644
--- a/contrib/pyzfs/libzfs_core/test/test_libzfs_core.py
+++ b/contrib/pyzfs/libzfs_core/test/test_libzfs_core.py
@@ -4223,7 +4223,7 @@ def reset(self):
             self.getRoot().reset()
             return
 
-        # On the Buildbot builders this may fail with "pool is busy"
+        # On the CI builders this may fail with "pool is busy"
         # Retry 5 times before raising an error
         retry = 0
         while True:

From 39b9b62a96ac62f3e159af35b98ab7da6682b1e8 Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Fri, 26 Sep 2025 17:52:57 -0700
Subject: [PATCH 08/23] CI: Switch FreeBSD 15 to 15.0-ALPHA3

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Closes #17795
---
 .github/workflows/scripts/qemu-2-start.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/scripts/qemu-2-start.sh b/.github/workflows/scripts/qemu-2-start.sh
index 8439942c5a41..1c608348ffcd 100755
--- a/.github/workflows/scripts/qemu-2-start.sh
+++ b/.github/workflows/scripts/qemu-2-start.sh
@@ -121,7 +121,7 @@ case "$OS" in
     KSRC="$FREEBSD_SNAP/../amd64/$FreeBSD/src.txz"
     ;;
   freebsd15-0c)
-    FreeBSD="15.0-ALPHA2"
+    FreeBSD="15.0-ALPHA3"
     OSNAME="FreeBSD $FreeBSD"
     OSv="freebsd14.0"
     URLxz="$FREEBSD_SNAP/$FreeBSD/amd64/Latest/FreeBSD-$FreeBSD-amd64-BASIC-CI-ufs.raw.xz"

From 2a331e411d9df15c7d851a9861c32d12569b8ec1 Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Mon, 29 Sep 2025 16:32:05 -0700
Subject: [PATCH 09/23] CI: Add ZTS -O option, log Setup Testing Machines step

Add a -O option to zfs-test.sh to dump debug information on test
timeout.  The debug info includes:

- 30 lines from 'top'
- /proc/<PID>/stack output of process with highest CPU usage
- Last lines strace-ing process with highest CPU usage
- /proc/sysrq-trigger kernel stack traces

All debug information gets dumped to /dev/kmsg (Linux only).

In addition, print out the VM console lines from the "Setup Testing
Machines" step.  We have often see VMs timeout at this step and don't
know why.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #17753
---
 .github/workflows/scripts/qemu-3-deps-vm.sh |  9 +--
 .github/workflows/scripts/qemu-5-setup.sh   | 25 ++++++---
 .github/workflows/scripts/qemu-6-tests.sh   |  2 +-
 scripts/zfs-tests.sh                        |  9 ++-
 tests/test-runner/bin/test-runner.py.in     | 61 +++++++++++++++++++++
 5 files changed, 93 insertions(+), 13 deletions(-)

diff --git a/.github/workflows/scripts/qemu-3-deps-vm.sh b/.github/workflows/scripts/qemu-3-deps-vm.sh
index 4a7e724586f0..f67bb2f68e94 100755
--- a/.github/workflows/scripts/qemu-3-deps-vm.sh
+++ b/.github/workflows/scripts/qemu-3-deps-vm.sh
@@ -20,7 +20,7 @@ function archlinux() {
   sudo pacman -Sy --noconfirm base-devel bc cpio cryptsetup dhclient dkms \
     fakeroot fio gdb inetutils jq less linux linux-headers lsscsi nfs-utils \
     parted pax perf python-packaging python-setuptools qemu-guest-agent ksh \
-    samba sysstat rng-tools rsync wget xxhash
+    samba strace sysstat rng-tools rsync wget xxhash
   echo "##[endgroup]"
 }
 
@@ -43,7 +43,8 @@ function debian() {
     lsscsi nfs-kernel-server pamtester parted python3 python3-all-dev \
     python3-cffi python3-dev python3-distlib python3-packaging libtirpc-dev \
     python3-setuptools python3-sphinx qemu-guest-agent rng-tools rpm2cpio \
-    rsync samba sysstat uuid-dev watchdog wget xfslibs-dev  xxhash zlib1g-dev
+    rsync samba strace sysstat uuid-dev watchdog wget xfslibs-dev xxhash \
+    zlib1g-dev
   echo "##[endgroup]"
 }
 
@@ -87,8 +88,8 @@ function rhel() {
     libuuid-devel lsscsi mdadm nfs-utils openssl-devel pam-devel pamtester \
     parted perf python3 python3-cffi python3-devel python3-packaging \
     kernel-devel python3-setuptools qemu-guest-agent rng-tools rpcgen \
-    rpm-build rsync samba sysstat systemd watchdog wget xfsprogs-devel xxhash \
-    zlib-devel
+    rpm-build rsync samba strace sysstat systemd watchdog wget xfsprogs-devel \
+    xxhash zlib-devel
   echo "##[endgroup]"
 }
 
diff --git a/.github/workflows/scripts/qemu-5-setup.sh b/.github/workflows/scripts/qemu-5-setup.sh
index 0adcad2a99bc..4869c1003e48 100755
--- a/.github/workflows/scripts/qemu-5-setup.sh
+++ b/.github/workflows/scripts/qemu-5-setup.sh
@@ -108,19 +108,30 @@ echo '*/5 * * * *  /root/cronjob.sh' > crontab.txt
 sudo crontab crontab.txt
 rm crontab.txt
 
-# check if the machines are okay
-echo "Waiting for vm's to come up...  (${VMs}x CPU=$CPU RAM=$RAM)"
-for ((i=1; i<=VMs; i++)); do
-  .github/workflows/scripts/qemu-wait-for-vm.sh vm$i
-done
-echo "All $VMs VMs are up now."
-
 # Save the VM's serial output (ttyS0) to /var/tmp/console.txt
 # - ttyS0 on the VM corresponds to a local /dev/pty/N entry
 # - use 'virsh ttyconsole' to lookup the /dev/pty/N entry
 for ((i=1; i<=VMs; i++)); do
   mkdir -p $RESPATH/vm$i
   read "pty" <<< $(sudo virsh ttyconsole vm$i)
+
+  # Create the file so we can tail it, even if there's no output.
+  touch $RESPATH/vm$i/console.txt
+
   sudo nohup bash -c "cat $pty > $RESPATH/vm$i/console.txt" &
+
+  # Write all VM boot lines to the console to aid in debugging failed boots.
+  # The boot lines from all the VMs will be munged together, so prepend each
+  # line with the vm hostname (like 'vm1:').
+  (while IFS=$'\n' read -r line; do echo "vm$i: $line" ; done < <(sudo tail -f $RESPATH/vm$i/console.txt)) &
+
 done
 echo "Console logging for ${VMs}x $OS started."
+
+
+# check if the machines are okay
+echo "Waiting for vm's to come up...  (${VMs}x CPU=$CPU RAM=$RAM)"
+for ((i=1; i<=VMs; i++)); do
+  .github/workflows/scripts/qemu-wait-for-vm.sh vm$i
+done
+echo "All $VMs VMs are up now."
diff --git a/.github/workflows/scripts/qemu-6-tests.sh b/.github/workflows/scripts/qemu-6-tests.sh
index 5ab822f4f076..ca6ac77f146d 100755
--- a/.github/workflows/scripts/qemu-6-tests.sh
+++ b/.github/workflows/scripts/qemu-6-tests.sh
@@ -111,7 +111,7 @@ fi
 sudo dmesg -c > dmesg-prerun.txt
 mount > mount.txt
 df -h > df-prerun.txt
-$TDIR/zfs-tests.sh -vK -s 3GB -T $TAGS
+$TDIR/zfs-tests.sh -vKO -s 3GB -T $TAGS
 RV=$?
 df -h > df-postrun.txt
 echo $RV > tests-exitcode.txt
diff --git a/scripts/zfs-tests.sh b/scripts/zfs-tests.sh
index 04f3b6f32cb8..5a0a1a609448 100755
--- a/scripts/zfs-tests.sh
+++ b/scripts/zfs-tests.sh
@@ -38,6 +38,7 @@ DEBUG=""
 CLEANUP="yes"
 CLEANUPALL="no"
 KMSG=""
+TIMEOUT_DEBUG=""
 LOOPBACK="yes"
 STACK_TRACER="no"
 FILESIZE="4G"
@@ -364,6 +365,7 @@ OPTIONS:
 	-k          Disable cleanup after test failure
 	-K          Log test names to /dev/kmsg
 	-f          Use files only, disables block device tests
+	-O          Dump debugging info to /dev/kmsg on test timeout
 	-S          Enable stack tracer (negative performance impact)
 	-c          Only create and populate constrained path
 	-R          Automatically rerun failing tests
@@ -402,7 +404,7 @@ $0 -x
 EOF
 }
 
-while getopts 'hvqxkKfScRmn:d:Ds:r:?t:T:u:I:' OPTION; do
+while getopts 'hvqxkKfScRmOn:d:Ds:r:?t:T:u:I:' OPTION; do
 	case $OPTION in
 	h)
 		usage
@@ -445,6 +447,9 @@ while getopts 'hvqxkKfScRmn:d:Ds:r:?t:T:u:I:' OPTION; do
 		export NFS=1
 		. "$nfsfile"
 		;;
+	O)
+		TIMEOUT_DEBUG="yes"
+		;;
 	d)
 		FILEDIR="$OPTARG"
 		;;
@@ -773,6 +778,7 @@ msg "${TEST_RUNNER}" \
     "${DEBUG:+-D}" \
     "${KMEMLEAK:+-m}" \
     "${KMSG:+-K}" \
+    "${TIMEOUT_DEBUG:+-O}" \
     "-c \"${RUNFILES}\"" \
     "-T \"${TAGS}\"" \
     "-i \"${STF_SUITE}\"" \
@@ -783,6 +789,7 @@ msg "${TEST_RUNNER}" \
     ${DEBUG:+-D} \
     ${KMEMLEAK:+-m} \
     ${KMSG:+-K} \
+    ${TIMEOUT_DEBUG:+-O} \
     -c "${RUNFILES}" \
     -T "${TAGS}" \
     -i "${STF_SUITE}" \
diff --git a/tests/test-runner/bin/test-runner.py.in b/tests/test-runner/bin/test-runner.py.in
index 2158208be6e5..d2c1185e4a94 100755
--- a/tests/test-runner/bin/test-runner.py.in
+++ b/tests/test-runner/bin/test-runner.py.in
@@ -34,6 +34,7 @@ from select import select
 from subprocess import PIPE
 from subprocess import Popen
 from subprocess import check_output
+from subprocess import run
 from threading import Timer
 from time import time, CLOCK_MONOTONIC
 from os.path import exists
@@ -187,6 +188,63 @@ User: %s
 ''' % (self.pathname, self.identifier, self.outputdir, self.timeout, self.user)
 
     def kill_cmd(self, proc, options, kmemleak, keyboard_interrupt=False):
+
+        """
+        We're about to kill a command due to a timeout.
+        If we're running with the -O option, then dump debug info about the
+        process with the highest CPU usage to /dev/kmsg (Linux only).  This can
+        help debug the timeout.
+
+        Debug info includes:
+        - 30 lines from 'top'
+        - /proc/<PID>/stack output of process with highest CPU usage
+        - Last lines strace-ing process with highest CPU usage
+        """
+        if exists("/dev/kmsg"):
+            c = """
+TOP_OUT="$(COLUMNS=160 top -b -n 1 | head -n 30)"
+read -r PID CMD <<< $(echo "$TOP_OUT" | /usr/bin/awk \
+"/COMMAND/{
+    print_next=1
+    next
+}
+{
+    if (print_next == 1) {
+        print \\$1\\" \\"\\$12
+        exit
+    }
+}")
+echo "##### ZTS timeout debug #####"
+echo "----- top -----"
+echo "$TOP_OUT"
+echo "----- /proc/$PID/stack ($CMD)) -----"
+cat /proc/$PID/stack
+echo "----- strace ($CMD) -----"
+TMPFILE="$(mktemp --suffix=ZTS)"
+/usr/bin/strace -k --stack-traces -p $PID &> "$TMPFILE" &
+sleep 0.1
+killall strace
+tail -n 30 $TMPFILE
+rm "$TMPFILE"
+echo "##### /proc/sysrq-trigger stack #####"
+"""
+            c = "sudo bash -c '" + c + "'"
+            data = run(c, capture_output=True, shell=True, text=True)
+            out = data.stdout
+            try:
+                kp = Popen([SUDO, "sh", "-c",
+                            "echo '" + out + "' > /dev/kmsg"])
+                kp.wait()
+
+                """
+                Trigger kernel stack traces
+                """
+                kp = Popen([SUDO, "sh", "-c",
+                            "echo l > /proc/sysrq-trigger"])
+                kp.wait()
+            except Exception:
+                pass
+
         """
         Kill a running command due to timeout, or ^C from the keyboard. If
         sudo is required, this user was verified previously.
@@ -1129,6 +1187,9 @@ def parse_args():
     parser.add_option('-o', action='callback', callback=options_cb,
                       default=BASEDIR, dest='outputdir', type='string',
                       metavar='outputdir', help='Specify an output directory.')
+    parser.add_option('-O', action='store_true', default=False,
+                      dest='timeout_debug',
+                      help='Dump debugging info to /dev/kmsg on test timeout')
     parser.add_option('-i', action='callback', callback=options_cb,
                       default=TESTDIR, dest='testdir', type='string',
                       metavar='testdir', help='Specify a test directory.')

From 03c956e806ad1db519d9f0958c013ccbf206f595 Mon Sep 17 00:00:00 2001
From: Shreshth3 <66148173+Shreshth3@users.noreply.github.com>
Date: Wed, 1 Oct 2025 10:15:46 -0700
Subject: [PATCH 10/23] docs: fix a few small typos (#17804)

Signed-off-by: Shreshth Srivastava <shreshthsrivastava2@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
---
 .github/ISSUE_TEMPLATE/feature_request.md | 2 +-
 contrib/intel_qat/readme.md               | 2 +-
 etc/init.d/README.md                      | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/ISSUE_TEMPLATE/feature_request.md b/.github/ISSUE_TEMPLATE/feature_request.md
index 9b50a4a3d96e..f3d4316f6f67 100644
--- a/.github/ISSUE_TEMPLATE/feature_request.md
+++ b/.github/ISSUE_TEMPLATE/feature_request.md
@@ -14,7 +14,7 @@ Please check our issue tracker before opening a new feature request.
 Filling out the following template will help other contributors better understand your proposed feature.
 -->
 
-### Describe the feature would like to see added to OpenZFS
+### Describe the feature you would like to see added to OpenZFS
 
 <!--
 Provide a clear and concise description of the feature.
diff --git a/contrib/intel_qat/readme.md b/contrib/intel_qat/readme.md
index 7e45d395bb80..04c299b6404c 100644
--- a/contrib/intel_qat/readme.md
+++ b/contrib/intel_qat/readme.md
@@ -8,7 +8,7 @@ This contrib contains community compatibility patches to get Intel QAT working o
 These patches are based on the following Intel QAT version:
 [1.7.l.4.10.0-00014](https://01.org/sites/default/files/downloads/qat1.7.l.4.10.0-00014.tar.gz)
 
-When using QAT with above kernels versions, the following patches needs to be applied using:
+When using QAT with the above kernel versions, the following patches need to be applied using:
 patch -p1 < _$PATCH_
 _Where $PATCH refers to the path of the patch in question_
 
diff --git a/etc/init.d/README.md b/etc/init.d/README.md
index da780fdc1222..3852dd9a6b2e 100644
--- a/etc/init.d/README.md
+++ b/etc/init.d/README.md
@@ -1,5 +1,5 @@
 DESCRIPTION
-  These script were written with the primary intention of being portable and
+  These scripts were written with the primary intention of being portable and
   usable on as many systems as possible.
 
   This is, in practice, usually not possible. But the intention is there.

From 007f325e1b96b5fca37a5f2824d0e6681116c364 Mon Sep 17 00:00:00 2001
From: Tino Reichardt <milky-zfs@mcmilk.de>
Date: Tue, 7 Oct 2025 01:41:01 +0200
Subject: [PATCH 11/23] CI: Switch FreeBSD 15 to 15.0-ALPHA4 and add FreeBSD 16

Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tino Reichardt <milky-zfs@mcmilk.de>
Closes #17815
---
 .github/workflows/scripts/qemu-2-start.sh | 9 ++++++++-
 .github/workflows/zfs-qemu.yml            | 2 +-
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/scripts/qemu-2-start.sh b/.github/workflows/scripts/qemu-2-start.sh
index 1c608348ffcd..9b1c5e084926 100755
--- a/.github/workflows/scripts/qemu-2-start.sh
+++ b/.github/workflows/scripts/qemu-2-start.sh
@@ -121,7 +121,14 @@ case "$OS" in
     KSRC="$FREEBSD_SNAP/../amd64/$FreeBSD/src.txz"
     ;;
   freebsd15-0c)
-    FreeBSD="15.0-ALPHA3"
+    FreeBSD="15.0-ALPHA4"
+    OSNAME="FreeBSD $FreeBSD"
+    OSv="freebsd14.0"
+    URLxz="$FREEBSD_SNAP/$FreeBSD/amd64/Latest/FreeBSD-$FreeBSD-amd64-BASIC-CI-ufs.raw.xz"
+    KSRC="$FREEBSD_SNAP/../amd64/$FreeBSD/src.txz"
+    ;;
+  freebsd16-0c)
+    FreeBSD="16.0-CURRENT"
     OSNAME="FreeBSD $FreeBSD"
     OSv="freebsd14.0"
     URLxz="$FREEBSD_SNAP/$FreeBSD/amd64/Latest/FreeBSD-$FreeBSD-amd64-BASIC-CI-ufs.raw.xz"
diff --git a/.github/workflows/zfs-qemu.yml b/.github/workflows/zfs-qemu.yml
index 69349678d84c..3b164548f9be 100644
--- a/.github/workflows/zfs-qemu.yml
+++ b/.github/workflows/zfs-qemu.yml
@@ -68,7 +68,7 @@ jobs:
         # FreeBSD variants of 2025-06:
         # FreeBSD Release: freebsd13-5r, freebsd14-2r, freebsd14-3r
         # FreeBSD Stable:  freebsd13-5s, freebsd14-3s
-        # FreeBSD Current: freebsd15-0c
+        # FreeBSD Current: freebsd15-0c, freebsd16-0c
         os: ${{ fromJson(needs.test-config.outputs.test_os) }}
     runs-on: ubuntu-24.04
     steps:

From 4b24cbba80a3f2c1ba4b2fb073a53094ccbe1502 Mon Sep 17 00:00:00 2001
From: Ameer Hamza <ahamza@ixsystems.com>
Date: Tue, 14 Oct 2025 06:19:47 +0500
Subject: [PATCH 12/23] CI: Fix FreeBSD 15.0 by staying on ALPHA4 due to broken
 ALPHA5 image

FreeBSD 15.0-ALPHA5 image fails to boot on cloud VMs due to missing
/boot/efi mount point, causing the system to drop to single user mode
where SSH cannot start. Work around this by staying on ALPHA4 and
setting IGNORE_OSVERSION=yes to bypass pkg's kernel version mismatch
prompt during bootstrap. This allows CI to proceed with ALPHA4 until we
have a stable FreeBSD 15.0 image.

Signed-off-by: Ameer Hamza <ahamza@ixsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Closes #17846
---
 .github/workflows/scripts/qemu-2-start.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/scripts/qemu-2-start.sh b/.github/workflows/scripts/qemu-2-start.sh
index 9b1c5e084926..422b3e9df388 100755
--- a/.github/workflows/scripts/qemu-2-start.sh
+++ b/.github/workflows/scripts/qemu-2-start.sh
@@ -294,7 +294,7 @@ else
   while pidof /usr/bin/qemu-system-x86_64 >/dev/null; do
     ssh 2>/dev/null root@vm0 "uname -a" && break
   done
-  ssh root@vm0 "pkg install -y bash ca_root_nss git qemu-guest-agent python3 py311-cloud-init"
+  ssh root@vm0 "env IGNORE_OSVERSION=yes pkg install -y bash ca_root_nss git qemu-guest-agent python3 py311-cloud-init"
   ssh root@vm0 "chsh -s $BASH root"
   ssh root@vm0 'sysrc qemu_guest_agent_enable="YES"'
   ssh root@vm0 'sysrc cloudinit_enable="YES"'

From 365926a37c06de23f6fecb7f538e3c56189402c6 Mon Sep 17 00:00:00 2001
From: Rob Norris <robn@despairlabs.com>
Date: Thu, 31 Jul 2025 13:12:43 +1000
Subject: [PATCH 13/23] Linux 6.17: d_set_d_op() is no longer available

We only have extremely narrow uses, so move it all into a single
function that does only what we need, with and without d_set_d_op().

Sponsored-by: https://despairlabs.com/sponsor/
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Rob Norris <robn@despairlabs.com>
Closes #17621
---
 config/kernel-dentry-operations.m4            | 10 +++-
 include/os/linux/kernel/linux/dcache_compat.h | 26 ---------
 module/os/linux/zfs/zpl_ctldir.c              | 55 ++++++++++++++++---
 3 files changed, 55 insertions(+), 36 deletions(-)

diff --git a/config/kernel-dentry-operations.m4 b/config/kernel-dentry-operations.m4
index aa5a9f2aff39..9d1bb3a74b1d 100644
--- a/config/kernel-dentry-operations.m4
+++ b/config/kernel-dentry-operations.m4
@@ -24,6 +24,9 @@ dnl #
 dnl # 2.6.38 API change
 dnl # Added d_set_d_op() helper function.
 dnl #
+dnl # 6.17 API change
+dnl # d_set_d_op() removed. No direct replacement.
+dnl #
 AC_DEFUN([ZFS_AC_KERNEL_SRC_D_SET_D_OP], [
 	ZFS_LINUX_TEST_SRC([d_set_d_op], [
 		#include <linux/dcache.h>
@@ -34,11 +37,12 @@ AC_DEFUN([ZFS_AC_KERNEL_SRC_D_SET_D_OP], [
 
 AC_DEFUN([ZFS_AC_KERNEL_D_SET_D_OP], [
 	AC_MSG_CHECKING([whether d_set_d_op() is available])
-	ZFS_LINUX_TEST_RESULT_SYMBOL([d_set_d_op],
-	    [d_set_d_op], [fs/dcache.c], [
+	ZFS_LINUX_TEST_RESULT([d_set_d_op], [
 		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_D_SET_D_OP, 1,
+		    [Define if d_set_d_op() is available])
 	], [
-		ZFS_LINUX_TEST_ERROR([d_set_d_op])
+		AC_MSG_RESULT(no)
 	])
 ])
 
diff --git a/include/os/linux/kernel/linux/dcache_compat.h b/include/os/linux/kernel/linux/dcache_compat.h
index 16e8a319a5f8..152e5a606f0e 100644
--- a/include/os/linux/kernel/linux/dcache_compat.h
+++ b/include/os/linux/kernel/linux/dcache_compat.h
@@ -60,32 +60,6 @@
 	} while (0)
 #endif
 
-/*
- * 2.6.30 API change,
- * The const keyword was added to the 'struct dentry_operations' in
- * the dentry structure.  To handle this we define an appropriate
- * dentry_operations_t typedef which can be used.
- */
-typedef const struct dentry_operations	dentry_operations_t;
-
-/*
- * 2.6.38 API addition,
- * Added d_clear_d_op() helper function which clears some flags and the
- * registered dentry->d_op table.  This is required because d_set_d_op()
- * issues a warning when the dentry operations table is already set.
- * For the .zfs control directory to work properly we must be able to
- * override the default operations table and register custom .d_automount
- * and .d_revalidate callbacks.
- */
-static inline void
-d_clear_d_op(struct dentry *dentry)
-{
-	dentry->d_op = NULL;
-	dentry->d_flags &= ~(
-	    DCACHE_OP_HASH | DCACHE_OP_COMPARE |
-	    DCACHE_OP_REVALIDATE | DCACHE_OP_DELETE);
-}
-
 /*
  * Walk and invalidate all dentry aliases of an inode
  * unless it's a mountpoint
diff --git a/module/os/linux/zfs/zpl_ctldir.c b/module/os/linux/zfs/zpl_ctldir.c
index 48dae79a2373..81ac26cb0c93 100644
--- a/module/os/linux/zfs/zpl_ctldir.c
+++ b/module/os/linux/zfs/zpl_ctldir.c
@@ -202,7 +202,7 @@ zpl_snapdir_revalidate(struct dentry *dentry, unsigned int flags)
 	return (!!dentry->d_inode);
 }
 
-static dentry_operations_t zpl_dops_snapdirs = {
+static const struct dentry_operations zpl_dops_snapdirs = {
 /*
  * Auto mounting of snapshots is only supported for 2.6.37 and
  * newer kernels.  Prior to this kernel the ops->follow_link()
@@ -215,6 +215,51 @@ static dentry_operations_t zpl_dops_snapdirs = {
 	.d_revalidate	= zpl_snapdir_revalidate,
 };
 
+/*
+ * For the .zfs control directory to work properly we must be able to override
+ * the default operations table and register custom .d_automount and
+ * .d_revalidate callbacks.
+ */
+static void
+set_snapdir_dentry_ops(struct dentry *dentry, unsigned int extraflags) {
+	static const unsigned int op_flags =
+	    DCACHE_OP_HASH | DCACHE_OP_COMPARE |
+	    DCACHE_OP_REVALIDATE | DCACHE_OP_DELETE |
+	    DCACHE_OP_PRUNE | DCACHE_OP_WEAK_REVALIDATE | DCACHE_OP_REAL;
+
+#ifdef HAVE_D_SET_D_OP
+	/*
+	 * d_set_d_op() will set the DCACHE_OP_ flags according to what it
+	 * finds in the passed dentry_operations, so we don't have to.
+	 *
+	 * We clear the flags and the old op table before calling d_set_d_op()
+	 * because issues a warning when the dentry operations table is already
+	 * set.
+	 */
+	dentry->d_op = NULL;
+	dentry->d_flags &= ~op_flags;
+	d_set_d_op(dentry, &zpl_dops_snapdirs);
+	dentry->d_flags |= extraflags;
+#else
+	/*
+	 * Since 6.17 there's no exported way to modify dentry ops, so we have
+	 * to reach in and do it ourselves. This should be safe for our very
+	 * narrow use case, which is to create or splice in an entry to give
+	 * access to a snapshot.
+	 *
+	 * We need to set the op flags directly. We hardcode
+	 * DCACHE_OP_REVALIDATE because that's the only operation we have; if
+	 * we ever extend zpl_dops_snapdirs we will need to update the op flags
+	 * to match.
+	 */
+	spin_lock(&dentry->d_lock);
+	dentry->d_op = &zpl_dops_snapdirs;
+	dentry->d_flags &= ~op_flags;
+	dentry->d_flags |= DCACHE_OP_REVALIDATE | extraflags;
+	spin_unlock(&dentry->d_lock);
+#endif
+}
+
 static struct dentry *
 zpl_snapdir_lookup(struct inode *dip, struct dentry *dentry,
     unsigned int flags)
@@ -236,10 +281,7 @@ zpl_snapdir_lookup(struct inode *dip, struct dentry *dentry,
 		return (ERR_PTR(error));
 
 	ASSERT(error == 0 || ip == NULL);
-	d_clear_d_op(dentry);
-	d_set_d_op(dentry, &zpl_dops_snapdirs);
-	dentry->d_flags |= DCACHE_NEED_AUTOMOUNT;
-
+	set_snapdir_dentry_ops(dentry, DCACHE_NEED_AUTOMOUNT);
 	return (d_splice_alias(ip, dentry));
 }
 
@@ -373,8 +415,7 @@ zpl_snapdir_mkdir(struct inode *dip, struct dentry *dentry, umode_t mode)
 
 	error = -zfsctl_snapdir_mkdir(dip, dname(dentry), vap, &ip, cr, 0);
 	if (error == 0) {
-		d_clear_d_op(dentry);
-		d_set_d_op(dentry, &zpl_dops_snapdirs);
+		set_snapdir_dentry_ops(dentry, 0);
 		d_instantiate(dentry, ip);
 	}
 

From e5ca94f21e457cc225afadc9f4ee9a550ac4d35a Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Thu, 25 Sep 2025 09:32:59 -0700
Subject: [PATCH 14/23] Fix 'zpool add' safety check corner cases

Three cases were discovered where 'zpool add' would fail to
warn when adding vdevs to a pool with a mismatched replication
level.  These are:

  1. When a pool contains mixed file and disk vdevs.
  2. When a pool contains an active dRAID distributed spare
  3. When a pool contains an active hot spare

The lack of warnings are caused by get_replication() assessing
the current pool configuration an inconsistent and disabling
the mismatched replication check for the new pool configuration
after 'zpool add'.  This change updates get_replication() to
be slightly more tolerant in the non-fatal case.

The zpool_add_010_pos.ksh test case was split in to separate
tests: zpool_add_warn_create.ksh, pool_add_warn_degraded.ksh,
and zpool_add_warn_removal.  These test were extended to
include coverage for dRAID pools and the three scenarios
described above.

Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #17780
---
 cmd/zpool/zpool_vdev.c                        |  26 ++-
 tests/runfiles/common.run                     |   5 +-
 tests/zfs-tests/tests/Makefile.am             |   4 +-
 .../cli_root/zpool_add/zpool_add.kshlib       |  42 ++++
 ..._010_pos.ksh => zpool_add_warn_create.ksh} | 100 ++-------
 .../zpool_add/zpool_add_warn_degraded.ksh     | 195 ++++++++++++++++++
 .../zpool_add/zpool_add_warn_removal.ksh      | 117 +++++++++++
 7 files changed, 392 insertions(+), 97 deletions(-)
 rename tests/zfs-tests/tests/functional/cli_root/zpool_add/{zpool_add_010_pos.ksh => zpool_add_warn_create.ksh} (57%)
 create mode 100755 tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_degraded.ksh
 create mode 100755 tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_removal.ksh

diff --git a/cmd/zpool/zpool_vdev.c b/cmd/zpool/zpool_vdev.c
index 07868a30d7e7..19ffcf98b595 100644
--- a/cmd/zpool/zpool_vdev.c
+++ b/cmd/zpool/zpool_vdev.c
@@ -609,23 +609,29 @@ get_replication(nvlist_t *nvroot, boolean_t fatal)
 				verify(nvlist_lookup_string(cnv,
 				    ZPOOL_CONFIG_PATH, &path) == 0);
 
+				/*
+				 * Skip active spares they should never cause
+				 * the pool to be evaluated as inconsistent.
+				 */
+				if (is_spare(NULL, path))
+					continue;
+
 				/*
 				 * If we have a raidz/mirror that combines disks
-				 * with files, report it as an error.
+				 * with files, only report it as an error when
+				 * fatal is set to ensure all the replication
+				 * checks aren't skipped in check_replication().
 				 */
-				if (!dontreport && type != NULL &&
+				if (fatal && !dontreport && type != NULL &&
 				    strcmp(type, childtype) != 0) {
 					if (ret != NULL)
 						free(ret);
 					ret = NULL;
-					if (fatal)
-						vdev_error(gettext(
-						    "mismatched replication "
-						    "level: %s contains both "
-						    "files and devices\n"),
-						    rep.zprl_type);
-					else
-						return (NULL);
+					vdev_error(gettext(
+					    "mismatched replication "
+					    "level: %s contains both "
+					    "files and devices\n"),
+					    rep.zprl_type);
 					dontreport = B_TRUE;
 				}
 
diff --git a/tests/runfiles/common.run b/tests/runfiles/common.run
index 6d96ec435841..1681c5fca157 100644
--- a/tests/runfiles/common.run
+++ b/tests/runfiles/common.run
@@ -390,8 +390,9 @@ tags = ['functional', 'cli_root', 'zpool']
 [tests/functional/cli_root/zpool_add]
 tests = ['zpool_add_001_pos', 'zpool_add_002_pos', 'zpool_add_003_pos',
     'zpool_add_004_pos', 'zpool_add_006_pos', 'zpool_add_007_neg',
-    'zpool_add_008_neg', 'zpool_add_009_neg', 'zpool_add_010_pos',
-    'add-o_ashift', 'add_prop_ashift', 'zpool_add_dryrun_output']
+    'zpool_add_008_neg', 'zpool_add_009_neg', 'zpool_add_warn_create',
+    'zpool_add_warn_degraded', 'zpool_add_warn_removal', 'add-o_ashift',
+    'add_prop_ashift', 'zpool_add_dryrun_output']
 tags = ['functional', 'cli_root', 'zpool_add']
 
 [tests/functional/cli_root/zpool_attach]
diff --git a/tests/zfs-tests/tests/Makefile.am b/tests/zfs-tests/tests/Makefile.am
index d9ea50ebe923..3cad700d836f 100644
--- a/tests/zfs-tests/tests/Makefile.am
+++ b/tests/zfs-tests/tests/Makefile.am
@@ -1018,7 +1018,9 @@ nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \
 	functional/cli_root/zpool_add/zpool_add_007_neg.ksh \
 	functional/cli_root/zpool_add/zpool_add_008_neg.ksh \
 	functional/cli_root/zpool_add/zpool_add_009_neg.ksh \
-	functional/cli_root/zpool_add/zpool_add_010_pos.ksh \
+	functional/cli_root/zpool_add/zpool_add_warn_create.ksh \
+	functional/cli_root/zpool_add/zpool_add_warn_degraded.ksh \
+	functional/cli_root/zpool_add/zpool_add_warn_removal.ksh \
 	functional/cli_root/zpool_add/zpool_add_dryrun_output.ksh \
 	functional/cli_root/zpool_attach/attach-o_ashift.ksh \
 	functional/cli_root/zpool_attach/cleanup.ksh \
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add.kshlib b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add.kshlib
index 091d65bb4f33..74780bb02141 100644
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add.kshlib
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add.kshlib
@@ -27,6 +27,7 @@
 
 #
 # Copyright (c) 2012, 2016 by Delphix. All rights reserved.
+# Copyright 2025 by Lawrence Livermore National Security, LLC.
 #
 
 . $STF_SUITE/include/libtest.shlib
@@ -89,3 +90,44 @@ function save_dump_dev
 	fi
 	echo $dumpdev
 }
+
+function zpool_create_add_setup
+{
+	typeset -i i=0
+
+	while ((i < 10)); do
+		log_must truncate -s $MINVDEVSIZE $TEST_BASE_DIR/vdev$i
+
+		eval vdev$i=$TEST_BASE_DIR/vdev$i
+		((i += 1))
+	done
+
+	if is_linux; then
+	        vdev_lo="$(losetup -f "$vdev4" --show)"
+	elif is_freebsd; then
+	        vdev_lo=/dev/"$(mdconfig -a -t vnode -f "$vdev4")"
+	else
+	        vdev_lo="$(lofiadm -a "$vdev4")"
+	fi
+}
+
+function zpool_create_add_cleanup
+{
+	datasetexists $TESTPOOL1 && destroy_pool $TESTPOOL1
+
+	if [[ -e $vdev_lo ]]; then
+		if is_linux; then
+			log_must losetup -d "$vdev_lo"
+		elif is_freebsd; then
+			log_must mdconfig -d -u "$vdev_lo"
+		else
+			log_must lofiadm -d "$vdev_lo"
+		fi
+	fi
+
+	typeset -i i=0
+	while ((i < 10)); do
+		rm -f $TEST_BASE_DIR/vdev$i
+		((i += 1))
+	done
+}
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_010_pos.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_create.ksh
similarity index 57%
rename from tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_010_pos.ksh
rename to tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_create.ksh
index df085a2ec746..6b0f0dd61388 100755
--- a/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_010_pos.ksh
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_create.ksh
@@ -23,67 +23,45 @@
 
 #
 # Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
-# Use is subject to license terms.
-#
-
-#
-# Copyright (c) 2012, 2016 by Delphix. All rights reserved.
+# Copyright 2012, 2016 by Delphix. All rights reserved.
+# Copyright 2025 by Lawrence Livermore National Security, LLC.
 #
 
 . $STF_SUITE/include/libtest.shlib
-. $STF_SUITE/tests/functional/cli_root/zpool_create/zpool_create.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_add/zpool_add.kshlib
 
 #
 # DESCRIPTION:
-#	Verify zpool add succeed when adding vdevs with matching redundancy.
+#	Verify zpool add succeeds when adding vdevs with matching redundancy
+#	and warns with differing redundancy for a healthy pool.
 #
 # STRATEGY:
 #	1. Create several files == $MINVDEVSIZE.
 #	2. Verify 'zpool add' succeeds with matching redundancy.
 #	3. Verify 'zpool add' warns with differing redundancy.
-#	4. Verify 'zpool add' warns with differing redundancy after removal.
 #
 
 verify_runnable "global"
 
-function cleanup
-{
-	datasetexists $TESTPOOL1 && destroy_pool $TESTPOOL1
-
-	typeset -i i=0
-	while ((i < 10)); do
-		rm -f $TEST_BASE_DIR/vdev$i
-		((i += 1))
-	done
-}
-
+log_assert "Verify 'zpool add' warns for differing redundancy."
+log_onexit zpool_create_add_cleanup
 
-log_assert "Verify 'zpool add' succeed with keywords combination."
-log_onexit cleanup
+zpool_create_add_setup
 
-# 1. Create several files == $MINVDEVSIZE.
 typeset -i i=0
-while ((i < 10)); do
-	log_must truncate -s $MINVDEVSIZE $TEST_BASE_DIR/vdev$i
-
-	eval vdev$i=$TEST_BASE_DIR/vdev$i
-	((i += 1))
-done
+typeset -i j=0
 
 set -A redundancy0_create_args \
 	"$vdev0"
 
 set -A redundancy1_create_args \
-	"mirror $vdev0 $vdev1" \
-	"raidz1 $vdev0 $vdev1"
+	"mirror $vdev0 $vdev1"
 
 set -A redundancy2_create_args \
-	"mirror $vdev0 $vdev1 $vdev2" \
-	"raidz2 $vdev0 $vdev1 $vdev2"
+	"mirror $vdev0 $vdev1 $vdev2"
 
 set -A redundancy3_create_args \
-	"mirror $vdev0 $vdev1 $vdev2 $vdev3" \
-	"raidz3 $vdev0 $vdev1 $vdev2 $vdev3"
+	"mirror $vdev0 $vdev1 $vdev2 $vdev3"
 
 set -A redundancy0_add_args \
 	"$vdev5" \
@@ -92,22 +70,13 @@ set -A redundancy0_add_args \
 set -A redundancy1_add_args \
 	"mirror $vdev5 $vdev6" \
 	"raidz1 $vdev5 $vdev6" \
-	"raidz1 $vdev5 $vdev6 mirror $vdev7 $vdev8" \
-	"mirror $vdev5 $vdev6 raidz1 $vdev7 $vdev8"
+	"raidz1 $vdev5 $vdev6 mirror $vdev7 $vdev8"
 
 set -A redundancy2_add_args \
-	"mirror $vdev5 $vdev6 $vdev7" \
-	"raidz2 $vdev5 $vdev6 $vdev7"
+	"mirror $vdev5 $vdev6 $vdev7"
 
 set -A redundancy3_add_args \
-	"mirror $vdev5 $vdev6 $vdev7 $vdev8" \
-	"raidz3 $vdev5 $vdev6 $vdev7 $vdev8"
-
-set -A log_args "log" "$vdev4"
-set -A cache_args "cache" "$vdev4"
-set -A spare_args "spare" "$vdev4"
-
-typeset -i j=0
+	"mirror $vdev5 $vdev6 $vdev7 $vdev8"
 
 function zpool_create_add
 {
@@ -148,30 +117,6 @@ function zpool_create_forced_add
 	done
 }
 
-function zpool_create_rm_add
-{
-	typeset -n create_args=$1
-	typeset -n add_args=$2
-	typeset -n rm_args=$3
-
-	i=0
-	while ((i < ${#create_args[@]})); do
-		j=0
-		while ((j < ${#add_args[@]})); do
-			log_must zpool create $TESTPOOL1 ${create_args[$i]}
-			log_must zpool add $TESTPOOL1 ${rm_args[0]} ${rm_args[1]}
-			log_must zpool add $TESTPOOL1 ${add_args[$j]}
-			log_must zpool remove $TESTPOOL1 ${rm_args[1]}
-			log_mustnot zpool add $TESTPOOL1 ${rm_args[1]}
-			log_must zpool add $TESTPOOL1 ${rm_args[0]} ${rm_args[1]}
-			log_must zpool destroy -f $TESTPOOL1
-
-			((j += 1))
-		done
-		((i += 1))
-	done
-}
-
 # 2. Verify 'zpool add' succeeds with matching redundancy.
 zpool_create_add redundancy0_create_args redundancy0_add_args
 zpool_create_add redundancy1_create_args redundancy1_add_args
@@ -195,17 +140,4 @@ zpool_create_forced_add redundancy3_create_args redundancy0_add_args
 zpool_create_forced_add redundancy3_create_args redundancy1_add_args
 zpool_create_forced_add redundancy3_create_args redundancy2_add_args
 
-# 4. Verify 'zpool add' warns with differing redundancy after removal.
-zpool_create_rm_add redundancy1_create_args redundancy1_add_args log_args
-zpool_create_rm_add redundancy2_create_args redundancy2_add_args log_args
-zpool_create_rm_add redundancy3_create_args redundancy3_add_args log_args
-
-zpool_create_rm_add redundancy1_create_args redundancy1_add_args cache_args
-zpool_create_rm_add redundancy2_create_args redundancy2_add_args cache_args
-zpool_create_rm_add redundancy3_create_args redundancy3_add_args cache_args
-
-zpool_create_rm_add redundancy1_create_args redundancy1_add_args spare_args
-zpool_create_rm_add redundancy2_create_args redundancy2_add_args spare_args
-zpool_create_rm_add redundancy3_create_args redundancy3_add_args spare_args
-
-log_pass "'zpool add' succeed with keywords combination."
+log_pass "Verify 'zpool add' warns for differing redundancy."
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_degraded.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_degraded.ksh
new file mode 100755
index 000000000000..ded9cb3c68f8
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_degraded.ksh
@@ -0,0 +1,195 @@
+#!/bin/ksh -p
+# SPDX-License-Identifier: CDDL-1.0
+#
+# 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 2009 Sun Microsystems, Inc.  All rights reserved.
+# Copyright 2012, 2016 by Delphix. All rights reserved.
+# Copyright 2025 by Lawrence Livermore National Security, LLC.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_add/zpool_add.kshlib
+
+#
+# DESCRIPTION:
+#	Verify zpool add succeeds when adding vdevs with matching redundancy
+#	and warns with differing redundancy for a degraded pool.
+#
+# STRATEGY:
+#	1. Create several files == $MINVDEVSIZE.
+#	2. Verify 'zpool add' succeeds with matching redundancy
+#	3. Verify 'zpool add' warns with differing redundancy when
+#	  a. Degraded pool with replaced mismatch vdev (file vs disk)
+#	  b. Degraded pool dRAID distributed spare active
+#	  c. Degraded pool hot spare active
+#
+
+verify_runnable "global"
+
+log_assert "Verify 'zpool add' warns for differing redundancy."
+log_onexit zpool_create_add_cleanup
+
+zpool_create_add_setup
+
+set -A redundancy1_create_args \
+	"mirror $vdev0 $vdev1" \
+	"raidz1 $vdev0 $vdev1" \
+	"draid1:1s $vdev0 $vdev1 $vdev9"
+
+set -A redundancy2_create_args \
+	"mirror $vdev0 $vdev1 $vdev2" \
+	"raidz2 $vdev0 $vdev1 $vdev2" \
+	"draid2:1s $vdev0 $vdev1 $vdev2 $vdev9"
+
+set -A redundancy3_create_args \
+	"mirror $vdev0 $vdev1 $vdev2 $vdev3" \
+	"raidz3 $vdev0 $vdev1 $vdev2 $vdev3" \
+	"draid3:1s $vdev0 $vdev1 $vdev2 $vdev3 $vdev9"
+
+set -A redundancy1_add_args \
+	"mirror $vdev5 $vdev6"
+
+set -A redundancy2_add_args \
+	"mirror $vdev5 $vdev6 $vdev7"
+
+set -A redundancy3_add_args \
+	"mirror $vdev5 $vdev6 $vdev7 $vdev8"
+
+set -A redundancy1_create_draid_args \
+	"draid1:1s $vdev0 $vdev1 $vdev2"
+
+set -A redundancy2_create_draid_args \
+	"draid2:1s $vdev0 $vdev1 $vdev2 $vdev3"
+
+set -A redundancy3_create_draid_args \
+	"draid3:1s $vdev0 $vdev1 $vdev2 $vdev3 $vdev9"
+
+set -A redundancy1_create_spare_args \
+	"mirror $vdev0 $vdev1 spare $vdev_lo" \
+	"raidz1 $vdev0 $vdev1 spare $vdev_lo" \
+	"draid1 $vdev0 $vdev1 spare $vdev_lo"
+
+set -A redundancy2_create_spare_args \
+	"mirror $vdev0 $vdev1 $vdev2 spare $vdev_lo" \
+	"raidz2 $vdev0 $vdev1 $vdev2 spare $vdev_lo" \
+	"draid2 $vdev0 $vdev1 $vdev2 spare $vdev_lo"
+
+set -A redundancy3_create_spare_args \
+	"mirror $vdev0 $vdev1 $vdev2 $vdev3 spare $vdev_lo" \
+	"raidz3 $vdev0 $vdev1 $vdev2 $vdev3 spare $vdev_lo" \
+	"draid3 $vdev0 $vdev1 $vdev2 $vdev3 spare $vdev_lo"
+
+set -A replace_args "$vdev1" "$vdev_lo"
+set -A draid1_args "$vdev1" "draid1-0-0"
+set -A draid2_args "$vdev1" "draid2-0-0"
+set -A draid3_args "$vdev1" "draid3-0-0"
+
+typeset -i i=0
+typeset -i j=0
+
+function zpool_create_degraded_add
+{
+	typeset -n create_args=$1
+	typeset -n add_args=$2
+	typeset -n rm_args=$3
+
+	i=0
+	while ((i < ${#create_args[@]})); do
+		j=0
+		while ((j < ${#add_args[@]})); do
+			log_must zpool create $TESTPOOL1 ${create_args[$i]}
+			log_must zpool offline -f $TESTPOOL1 ${rm_args[0]}
+			log_must zpool replace -w $TESTPOOL1 ${rm_args[0]} ${rm_args[1]}
+			log_must zpool add $TESTPOOL1 ${add_args[$j]}
+			log_must zpool destroy -f $TESTPOOL1
+			log_must zpool labelclear -f ${rm_args[0]}
+
+			((j += 1))
+		done
+		((i += 1))
+	done
+}
+
+function zpool_create_forced_degraded_add
+{
+	typeset -n create_args=$1
+	typeset -n add_args=$2
+	typeset -n rm_args=$3
+
+	i=0
+	while ((i < ${#create_args[@]})); do
+		j=0
+		while ((j < ${#add_args[@]})); do
+			log_must zpool create $TESTPOOL1 ${create_args[$i]}
+			log_must zpool offline -f $TESTPOOL1 ${rm_args[0]}
+			log_must zpool replace -w $TESTPOOL1 ${rm_args[0]} ${rm_args[1]}
+			log_mustnot zpool add $TESTPOOL1 ${add_args[$j]}
+			log_must zpool add --allow-replication-mismatch $TESTPOOL1 ${add_args[$j]}
+			log_must zpool destroy -f $TESTPOOL1
+			log_must zpool labelclear -f ${rm_args[0]}
+
+			((j += 1))
+		done
+		((i += 1))
+	done
+}
+
+# 2. Verify 'zpool add' succeeds with matching redundancy and a degraded pool.
+zpool_create_degraded_add redundancy1_create_args redundancy1_add_args replace_args
+zpool_create_degraded_add redundancy2_create_args redundancy2_add_args replace_args
+zpool_create_degraded_add redundancy3_create_args redundancy3_add_args replace_args
+
+# 3. Verify 'zpool add' warns with differing redundancy and a degraded pool.
+#
+# a. Degraded pool with replaced mismatch vdev (file vs disk)
+zpool_create_forced_degraded_add redundancy1_create_args redundancy2_add_args replace_args
+zpool_create_forced_degraded_add redundancy1_create_args redundancy3_add_args replace_args
+
+zpool_create_forced_degraded_add redundancy2_create_args redundancy1_add_args replace_args
+zpool_create_forced_degraded_add redundancy2_create_args redundancy3_add_args replace_args
+
+zpool_create_forced_degraded_add redundancy3_create_args redundancy1_add_args replace_args
+zpool_create_forced_degraded_add redundancy3_create_args redundancy2_add_args replace_args
+
+# b. Degraded pool dRAID distributed spare active
+
+zpool_create_forced_degraded_add redundancy1_create_draid_args redundancy2_add_args draid1_args
+zpool_create_forced_degraded_add redundancy1_create_draid_args redundancy3_add_args draid1_args
+
+zpool_create_forced_degraded_add redundancy2_create_draid_args redundancy1_add_args draid2_args
+zpool_create_forced_degraded_add redundancy2_create_draid_args redundancy3_add_args draid2_args
+
+zpool_create_forced_degraded_add redundancy3_create_draid_args redundancy1_add_args draid3_args
+zpool_create_forced_degraded_add redundancy3_create_draid_args redundancy2_add_args draid3_args
+
+# c. Degraded pool hot spare active
+zpool_create_forced_degraded_add redundancy1_create_spare_args redundancy2_add_args replace_args
+zpool_create_forced_degraded_add redundancy1_create_spare_args redundancy3_add_args replace_args
+
+zpool_create_forced_degraded_add redundancy2_create_spare_args redundancy1_add_args replace_args
+zpool_create_forced_degraded_add redundancy2_create_spare_args redundancy3_add_args replace_args
+
+zpool_create_forced_degraded_add redundancy3_create_spare_args redundancy1_add_args replace_args
+zpool_create_forced_degraded_add redundancy3_create_spare_args redundancy2_add_args replace_args
+
+log_pass "Verify 'zpool add' warns for differing redundancy."
diff --git a/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_removal.ksh b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_removal.ksh
new file mode 100755
index 000000000000..56feaf1f9965
--- /dev/null
+++ b/tests/zfs-tests/tests/functional/cli_root/zpool_add/zpool_add_warn_removal.ksh
@@ -0,0 +1,117 @@
+#!/bin/ksh -p
+# SPDX-License-Identifier: CDDL-1.0
+#
+# 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 2009 Sun Microsystems, Inc.  All rights reserved.
+# Copyright 2012, 2016 by Delphix. All rights reserved.
+# Copyright 2025 by Lawrence Livermore National Security, LLC.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/cli_root/zpool_add/zpool_add.kshlib
+
+#
+# DESCRIPTION:
+#	Verify zpool add succeeds when adding vdevs with matching redundancy
+#	and warns with differing redundancy after removal.
+#
+# STRATEGY:
+#	1. Create several files == $MINVDEVSIZE.
+#	2. Verify 'zpool add' warns with differing redundancy after removal.
+#
+
+verify_runnable "global"
+
+log_assert "Verify 'zpool add' warns for differing redundancy."
+log_onexit zpool_create_add_cleanup
+
+zpool_create_add_setup
+
+typeset -i i=0
+typeset -i j=0
+
+set -A redundancy1_create_args \
+	"mirror $vdev0 $vdev1" \
+	"raidz1 $vdev0 $vdev1" \
+	"draid1:1s $vdev0 $vdev1 $vdev9"
+
+set -A redundancy2_create_args \
+	"mirror $vdev0 $vdev1 $vdev2" \
+	"raidz2 $vdev0 $vdev1 $vdev2" \
+	"draid2:1s $vdev0 $vdev1 $vdev2 $vdev9"
+
+set -A redundancy3_create_args \
+	"mirror $vdev0 $vdev1 $vdev2 $vdev3" \
+	"raidz3 $vdev0 $vdev1 $vdev2 $vdev3" \
+	"draid3:1s $vdev0 $vdev1 $vdev2 $vdev3 $vdev9"
+
+set -A redundancy1_add_args \
+	"mirror $vdev5 $vdev6"
+
+set -A redundancy2_add_args \
+	"mirror $vdev5 $vdev6 $vdev7"
+
+set -A redundancy3_add_args \
+	"mirror $vdev5 $vdev6 $vdev7 $vdev8"
+
+set -A log_args "log" "$vdev_lo"
+set -A cache_args "cache" "$vdev_lo"
+set -A spare_args "spare" "$vdev_lo"
+
+
+function zpool_create_rm_add
+{
+	typeset -n create_args=$1
+	typeset -n add_args=$2
+	typeset -n rm_args=$3
+
+	i=0
+	while ((i < ${#create_args[@]})); do
+		j=0
+		while ((j < ${#add_args[@]})); do
+			log_must zpool create $TESTPOOL1 ${create_args[$i]}
+			log_must zpool add $TESTPOOL1 ${rm_args[0]} ${rm_args[1]}
+			log_must zpool add $TESTPOOL1 ${add_args[$j]}
+			log_must zpool remove $TESTPOOL1 ${rm_args[1]}
+			log_mustnot zpool add $TESTPOOL1 ${rm_args[1]}
+			log_must zpool add $TESTPOOL1 ${rm_args[0]} ${rm_args[1]}
+			log_must zpool destroy -f $TESTPOOL1
+
+			((j += 1))
+		done
+		((i += 1))
+	done
+}
+
+# 2. Verify 'zpool add' warns with differing redundancy after removal.
+zpool_create_rm_add redundancy1_create_args redundancy1_add_args log_args
+zpool_create_rm_add redundancy2_create_args redundancy2_add_args log_args
+zpool_create_rm_add redundancy3_create_args redundancy3_add_args log_args
+
+zpool_create_rm_add redundancy1_create_args redundancy1_add_args cache_args
+zpool_create_rm_add redundancy2_create_args redundancy2_add_args cache_args
+zpool_create_rm_add redundancy3_create_args redundancy3_add_args cache_args
+
+zpool_create_rm_add redundancy1_create_args redundancy1_add_args spare_args
+zpool_create_rm_add redundancy2_create_args redundancy2_add_args spare_args
+zpool_create_rm_add redundancy3_create_args redundancy3_add_args spare_args

From 660077ffeeb1a09f354a76dc0f5d44ffabc58065 Mon Sep 17 00:00:00 2001
From: Rob Norris <robn@despairlabs.com>
Date: Wed, 13 Aug 2025 07:09:59 +1000
Subject: [PATCH 15/23] contrib/initramfs/scripts/zfs: shellcheck fixup

I got a newer shellcheck, and it pointed out that read without a target
variable is not POSIXly. The var was removed in c3ef9f7528, so I put it
back, and now shellcheck complains about an unused var. That's actually
correct, but necessary, so I've added a suppression for that, probably
better.

Sponsored-by: https://despairlabs.com/sponsor/
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Rob Norris <robn@despairlabs.com>
Closes #17626
---
 contrib/initramfs/scripts/zfs | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/contrib/initramfs/scripts/zfs b/contrib/initramfs/scripts/zfs
index c569b2528368..67707e9d80f4 100644
--- a/contrib/initramfs/scripts/zfs
+++ b/contrib/initramfs/scripts/zfs
@@ -979,7 +979,8 @@ mountroot()
 
 	touch /run/zfs_unlock_complete
 	if [ -e /run/zfs_unlock_complete_notify ]; then
-		read -r < /run/zfs_unlock_complete_notify
+		# shellcheck disable=SC2034
+		read -r zfs_unlock_complete_notify < /run/zfs_unlock_complete_notify
 	fi
 
 	# ------------

From 2b0502c5781407c78195eadba573499b807e1860 Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Thu, 25 Sep 2025 09:35:35 -0700
Subject: [PATCH 16/23] Add interface to interface
 spa_get_worst_case_min_alloc() function

Provide an interface to retrieve the lowest and highest minimum
allocation size for the normal allocation class.  This can be used
by external consumers of the DMU to estimate potential wasted
capacity when setting the recordsize for an object.

The new "min_alloc" and "max_alloc" keys are added to the pool
configuration and used by default_volblocksize() to warn when
an ineffecient block size is requested.  For older kmods which
don't yet include the new keys fallback to the previous logic.

Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #17758
---
 cmd/zfs/zfs_main.c      | 28 +++++++++++++++++++++-------
 include/sys/fs/zfs.h    |  2 ++
 include/sys/spa.h       |  1 +
 include/sys/spa_impl.h  |  1 +
 module/zfs/spa_config.c |  2 ++
 module/zfs/spa_misc.c   | 15 +++++++++++++++
 module/zfs/vdev.c       | 15 ++++++++-------
 module/zfs/vdev_label.c |  2 ++
 8 files changed, 52 insertions(+), 14 deletions(-)

diff --git a/cmd/zfs/zfs_main.c b/cmd/zfs/zfs_main.c
index a16e9148420a..6f7dd31a57a3 100644
--- a/cmd/zfs/zfs_main.c
+++ b/cmd/zfs/zfs_main.c
@@ -930,19 +930,15 @@ zfs_do_clone(int argc, char **argv)
 }
 
 /*
- * Return a default volblocksize for the pool which always uses more than
- * half of the data sectors.  This primarily applies to dRAID which always
- * writes full stripe widths.
+ * Calculate the minimum allocation size based on the top-level vdevs.
  */
 static uint64_t
-default_volblocksize(zpool_handle_t *zhp, nvlist_t *props)
+calculate_volblocksize(nvlist_t *config)
 {
-	uint64_t volblocksize, asize = SPA_MINBLOCKSIZE;
+	uint64_t asize = SPA_MINBLOCKSIZE;
 	nvlist_t *tree, **vdevs;
 	uint_t nvdevs;
 
-	nvlist_t *config = zpool_get_config(zhp, NULL);
-
 	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &tree) != 0 ||
 	    nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN,
 	    &vdevs, &nvdevs) != 0) {
@@ -973,6 +969,24 @@ default_volblocksize(zpool_handle_t *zhp, nvlist_t *props)
 		}
 	}
 
+	return (asize);
+}
+
+/*
+ * Return a default volblocksize for the pool which always uses more than
+ * half of the data sectors.  This primarily applies to dRAID which always
+ * writes full stripe widths.
+ */
+static uint64_t
+default_volblocksize(zpool_handle_t *zhp, nvlist_t *props)
+{
+	uint64_t volblocksize, asize = SPA_MINBLOCKSIZE;
+
+	nvlist_t *config = zpool_get_config(zhp, NULL);
+
+	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_MAX_ALLOC, &asize) != 0)
+		asize = calculate_volblocksize(config);
+
 	/*
 	 * Calculate the target volblocksize such that more than half
 	 * of the asize is used. The following table is for 4k sectors.
diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h
index bfaea6fb3fde..b6f9202cea5b 100644
--- a/include/sys/fs/zfs.h
+++ b/include/sys/fs/zfs.h
@@ -740,6 +740,8 @@ typedef struct zpool_load_policy {
 #define	ZPOOL_CONFIG_METASLAB_SHIFT	"metaslab_shift"
 #define	ZPOOL_CONFIG_ASHIFT		"ashift"
 #define	ZPOOL_CONFIG_ASIZE		"asize"
+#define	ZPOOL_CONFIG_MIN_ALLOC		"min_alloc"
+#define	ZPOOL_CONFIG_MAX_ALLOC		"max_alloc"
 #define	ZPOOL_CONFIG_DTL		"DTL"
 #define	ZPOOL_CONFIG_SCAN_STATS		"scan_stats"	/* not stored on disk */
 #define	ZPOOL_CONFIG_REMOVAL_STATS	"removal_stats"	/* not stored on disk */
diff --git a/include/sys/spa.h b/include/sys/spa.h
index 880f60471b65..5259d32c165f 100644
--- a/include/sys/spa.h
+++ b/include/sys/spa.h
@@ -1055,6 +1055,7 @@ extern pool_state_t spa_state(spa_t *spa);
 extern spa_load_state_t spa_load_state(spa_t *spa);
 extern uint64_t spa_freeze_txg(spa_t *spa);
 extern uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize);
+extern void spa_get_min_alloc_range(spa_t *spa, uint64_t *min, uint64_t *max);
 extern uint64_t spa_get_dspace(spa_t *spa);
 extern uint64_t spa_get_checkpoint_space(spa_t *spa);
 extern uint64_t spa_get_slop_space(spa_t *spa);
diff --git a/include/sys/spa_impl.h b/include/sys/spa_impl.h
index a3fbf8504050..cabf07a3c313 100644
--- a/include/sys/spa_impl.h
+++ b/include/sys/spa_impl.h
@@ -267,6 +267,7 @@ struct spa {
 	uint64_t	spa_min_ashift;		/* of vdevs in normal class */
 	uint64_t	spa_max_ashift;		/* of vdevs in normal class */
 	uint64_t	spa_min_alloc;		/* of vdevs in normal class */
+	uint64_t	spa_max_alloc;		/* of vdevs in normal class */
 	uint64_t	spa_gcd_alloc;		/* of vdevs in normal class */
 	uint64_t	spa_config_guid;	/* config pool guid */
 	uint64_t	spa_load_guid;		/* spa_load initialized guid */
diff --git a/module/zfs/spa_config.c b/module/zfs/spa_config.c
index 7d4d06659146..a38b1e8730fc 100644
--- a/module/zfs/spa_config.c
+++ b/module/zfs/spa_config.c
@@ -461,6 +461,8 @@ spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats)
 	fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, txg);
 	fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_guid(spa));
 	fnvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata);
+	fnvlist_add_uint64(config, ZPOOL_CONFIG_MIN_ALLOC, spa->spa_min_alloc);
+	fnvlist_add_uint64(config, ZPOOL_CONFIG_MAX_ALLOC, spa->spa_max_alloc);
 	if (spa->spa_comment != NULL)
 		fnvlist_add_string(config, ZPOOL_CONFIG_COMMENT,
 		    spa->spa_comment);
diff --git a/module/zfs/spa_misc.c b/module/zfs/spa_misc.c
index 3d98a0449630..bcf53ab9ab8e 100644
--- a/module/zfs/spa_misc.c
+++ b/module/zfs/spa_misc.c
@@ -814,6 +814,7 @@ spa_add(const char *name, nvlist_t *config, const char *altroot)
 	spa->spa_min_ashift = INT_MAX;
 	spa->spa_max_ashift = 0;
 	spa->spa_min_alloc = INT_MAX;
+	spa->spa_max_alloc = 0;
 	spa->spa_gcd_alloc = INT_MAX;
 
 	/* Reset cached value */
@@ -1876,6 +1877,19 @@ spa_get_worst_case_asize(spa_t *spa, uint64_t lsize)
 	return (MAX(lsize, 1 << spa->spa_max_ashift) * spa_asize_inflation);
 }
 
+/*
+ * Return the range of minimum allocation sizes for the normal allocation
+ * class. This can be used by external consumers of the DMU to estimate
+ * potential wasted capacity when setting the recordsize for an object.
+ * This is mainly for dRAID pools which always pad to a full stripe width.
+ */
+void
+spa_get_min_alloc_range(spa_t *spa, uint64_t *min_alloc, uint64_t *max_alloc)
+{
+	*min_alloc = spa->spa_min_alloc;
+	*max_alloc = spa->spa_max_alloc;
+}
+
 /*
  * Return the amount of slop space in bytes.  It is typically 1/32 of the pool
  * (3.2%), minus the embedded log space.  On very small pools, it may be
@@ -3083,6 +3097,7 @@ EXPORT_SYMBOL(spa_version);
 EXPORT_SYMBOL(spa_state);
 EXPORT_SYMBOL(spa_load_state);
 EXPORT_SYMBOL(spa_freeze_txg);
+EXPORT_SYMBOL(spa_get_min_alloc_range); /* for Lustre */
 EXPORT_SYMBOL(spa_get_dspace);
 EXPORT_SYMBOL(spa_update_dspace);
 EXPORT_SYMBOL(spa_deflate);
diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c
index bada0aada27d..373d3427057e 100644
--- a/module/zfs/vdev.c
+++ b/module/zfs/vdev.c
@@ -1474,12 +1474,14 @@ vdev_spa_set_alloc(spa_t *spa, uint64_t min_alloc)
 {
 	if (min_alloc < spa->spa_min_alloc)
 		spa->spa_min_alloc = min_alloc;
-	if (spa->spa_gcd_alloc == INT_MAX) {
+
+	if (min_alloc > spa->spa_max_alloc)
+		spa->spa_max_alloc = min_alloc;
+
+	if (spa->spa_gcd_alloc == INT_MAX)
 		spa->spa_gcd_alloc = min_alloc;
-	} else {
-		spa->spa_gcd_alloc = vdev_gcd(min_alloc,
-		    spa->spa_gcd_alloc);
-	}
+	else
+		spa->spa_gcd_alloc = vdev_gcd(min_alloc, spa->spa_gcd_alloc);
 }
 
 void
@@ -1533,8 +1535,7 @@ vdev_metaslab_group_create(vdev_t *vd)
 			if (vd->vdev_ashift < spa->spa_min_ashift)
 				spa->spa_min_ashift = vd->vdev_ashift;
 
-			uint64_t min_alloc = vdev_get_min_alloc(vd);
-			vdev_spa_set_alloc(spa, min_alloc);
+			vdev_spa_set_alloc(spa, vdev_get_min_alloc(vd));
 		}
 	}
 }
diff --git a/module/zfs/vdev_label.c b/module/zfs/vdev_label.c
index 6baa6236aac2..13e4624c5469 100644
--- a/module/zfs/vdev_label.c
+++ b/module/zfs/vdev_label.c
@@ -511,6 +511,8 @@ vdev_config_generate(spa_t *spa, vdev_t *vd, boolean_t getstats,
 		fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASHIFT, vd->vdev_ashift);
 		fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASIZE,
 		    vd->vdev_asize);
+		fnvlist_add_uint64(nv, ZPOOL_CONFIG_MIN_ALLOC,
+		    vdev_get_min_alloc(vd));
 		fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_LOG, vd->vdev_islog);
 		if (vd->vdev_noalloc) {
 			fnvlist_add_uint64(nv, ZPOOL_CONFIG_NONALLOCATING,

From 1baecd3a78f3228e3aa304abe8787e21a990688e Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Fri, 26 Sep 2025 10:00:18 -0700
Subject: [PATCH 17/23] Linux 6.17 compat: META

Update the META file to reflect compatibility with the 6.17
kernel.

Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Reviewed-by: Rob Norris <robn@despairlabs.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #17789
---
 META | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/META b/META
index f2c90626528b..c96da9b726b1 100644
--- a/META
+++ b/META
@@ -6,5 +6,5 @@ Release:       1
 Release-Tags:  relext
 License:       CDDL
 Author:        OpenZFS
-Linux-Maximum: 6.16
+Linux-Maximum: 6.17
 Linux-Minimum: 4.18

From 0bb5950e7235795105f59eb090ee7208f956b7be Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Mon, 29 Sep 2025 16:29:20 -0700
Subject: [PATCH 18/23] zvol: Fix blk-mq sync

The zvol blk-mq codepaths would erroneously send FLUSH and TRIM
commands down the read codepath, rather than write.  This fixes
the issue, and updates the zvol_misc_fua test to verify that
sync writes are actually happening.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Reviewed-by: Ameer Hamza <ahamza@ixsystems.com>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #17761
Closes #17765
---
 include/os/linux/kernel/linux/blkdev_compat.h | 18 ---------
 module/os/linux/zfs/zvol_os.c                 | 23 ++++++++++-
 .../zvol/zvol_misc/zvol_misc_fua.ksh          | 40 ++++++++++++++++++-
 3 files changed, 61 insertions(+), 20 deletions(-)

diff --git a/include/os/linux/kernel/linux/blkdev_compat.h b/include/os/linux/kernel/linux/blkdev_compat.h
index 076dab8ba6dc..214f3ea0e787 100644
--- a/include/os/linux/kernel/linux/blkdev_compat.h
+++ b/include/os/linux/kernel/linux/blkdev_compat.h
@@ -542,24 +542,6 @@ blk_generic_alloc_queue(make_request_fn make_request, int node_id)
 }
 #endif /* !HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS */
 
-/*
- * All the io_*() helper functions below can operate on a bio, or a rq, but
- * not both.  The older submit_bio() codepath will pass a bio, and the
- * newer blk-mq codepath will pass a rq.
- */
-static inline int
-io_data_dir(struct bio *bio, struct request *rq)
-{
-	if (rq != NULL) {
-		if (op_is_write(req_op(rq))) {
-			return (WRITE);
-		} else {
-			return (READ);
-		}
-	}
-	return (bio_data_dir(bio));
-}
-
 static inline int
 io_is_flush(struct bio *bio, struct request *rq)
 {
diff --git a/module/os/linux/zfs/zvol_os.c b/module/os/linux/zfs/zvol_os.c
index 702a6481c195..d5b5ea796c90 100644
--- a/module/os/linux/zfs/zvol_os.c
+++ b/module/os/linux/zfs/zvol_os.c
@@ -523,7 +523,28 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 	fstrans_cookie_t cookie = spl_fstrans_mark();
 	uint64_t offset = io_offset(bio, rq);
 	uint64_t size = io_size(bio, rq);
-	int rw = io_data_dir(bio, rq);
+	int rw;
+
+	if (rq != NULL) {
+		/*
+		 * Flush & trim requests go down the zvol_write codepath.  Or
+		 * more specifically:
+		 *
+		 * If request is a write, or if it's op_is_sync() and not a
+		 * read, or if it's a flush, or if it's a discard, then send the
+		 * request down the write path.
+		 */
+		if (op_is_write(rq->cmd_flags) ||
+		    (op_is_sync(rq->cmd_flags) && req_op(rq) != REQ_OP_READ) ||
+		    req_op(rq) == REQ_OP_FLUSH ||
+		    op_is_discard(rq->cmd_flags)) {
+			rw = WRITE;
+		} else {
+			rw = READ;
+		}
+	} else {
+		rw = bio_data_dir(bio);
+	}
 
 	if (unlikely(zv->zv_flags & ZVOL_REMOVING)) {
 		zvol_end_io(bio, rq, -SET_ERROR(ENXIO));
diff --git a/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_fua.ksh b/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_fua.ksh
index 571a698eb63a..502ebada22dc 100755
--- a/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_fua.ksh
+++ b/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_fua.ksh
@@ -50,17 +50,53 @@ fi
 
 typeset datafile1="$(mktemp -t zvol_misc_fua1.XXXXXX)"
 typeset datafile2="$(mktemp -t zvol_misc_fua2.XXXXXX)"
+typeset datafile3="$(mktemp -t zvol_misc_fua3_log.XXXXXX)"
 typeset zvolpath=${ZVOL_DEVDIR}/$TESTPOOL/$TESTVOL
 
+typeset DISK1=${DISKS%% *}
 function cleanup
 {
-       rm "$datafile1" "$datafile2"
+	log_must zpool remove $TESTPOOL $datafile3
+	rm "$datafile1" "$datafile2" "$datafile2"
+}
+
+# Prints the total number of sync writes for a vdev
+# $1: vdev
+function get_sync
+{
+	zpool iostat -p -H -v -r $TESTPOOL $1 | \
+	    awk '/[0-9]+$/{s+=$4+$5} END{print s}'
 }
 
 function do_test {
 	# Wait for udev to create symlinks to our zvol
 	block_device_wait $zvolpath
 
+	# Write using sync (creates FLUSH calls after writes, but not FUA)
+	old_vdev_writes=$(get_sync $DISK1)
+	old_log_writes=$(get_sync $datafile3)
+
+	log_must fio --name=write_iops --size=5M \
+		--ioengine=libaio --verify=0 --bs=4K \
+		--iodepth=1 --rw=randwrite --group_reporting=1 \
+		--filename=$zvolpath --sync=1
+
+	vdev_writes=$(( $(get_sync $DISK1) - $old_vdev_writes))
+	log_writes=$(( $(get_sync $datafile3) - $old_log_writes))
+
+	# When we're doing sync writes, we should see many more writes go to
+	# the log vs the first vdev.  Experiments show anywhere from a 160-320x
+	# ratio of writes to the log vs the first vdev (due to some straggler
+	# writes to the first vdev).
+	#
+	# Check that we have a large ratio (100x) of sync writes going to the
+	# log device
+	ratio=$(($log_writes / $vdev_writes))
+	log_note "Got $log_writes log writes, $vdev_writes vdev writes."
+	if [ $ratio -lt 100 ] ; then
+		log_fail "Expected > 100x more log writes than vdev writes. "
+	fi
+
 	# Create a data file
 	log_must dd if=/dev/urandom of="$datafile1" bs=1M count=5
 
@@ -81,6 +117,8 @@ log_assert "Verify that a ZFS volume can do Force Unit Access (FUA)"
 log_onexit cleanup
 
 log_must zfs set compression=off $TESTPOOL/$TESTVOL
+log_must truncate -s 100M $datafile3
+log_must zpool add $TESTPOOL log $datafile3
 
 log_note "Testing without blk-mq"
 

From 2a5349bf932041e3271d267ead3e5a40f2920b71 Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Mon, 6 Oct 2025 16:54:09 -0700
Subject: [PATCH 19/23] zvol: verify IO type is supported

ZVOLs don't support all block layer IO request types.  Add a check for
the IO types we do support.  Also, remove references to
io_is_secure_erase() since they are not supported on ZVOLs.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #17803
---
 module/os/linux/zfs/zvol_os.c                 | 77 ++++++++++++-------
 .../zvol/zvol_misc/zvol_misc_trim.ksh         |  7 ++
 2 files changed, 56 insertions(+), 28 deletions(-)

diff --git a/module/os/linux/zfs/zvol_os.c b/module/os/linux/zfs/zvol_os.c
index d5b5ea796c90..592ec7bd42f0 100644
--- a/module/os/linux/zfs/zvol_os.c
+++ b/module/os/linux/zfs/zvol_os.c
@@ -377,16 +377,14 @@ zvol_discard(zv_request_t *zvr)
 	}
 
 	/*
-	 * Align the request to volume block boundaries when a secure erase is
-	 * not required.  This will prevent dnode_free_range() from zeroing out
-	 * the unaligned parts which is slow (read-modify-write) and useless
-	 * since we are not freeing any space by doing so.
+	 * Align the request to volume block boundaries. This will prevent
+	 * dnode_free_range() from zeroing out the unaligned parts which is
+	 * slow (read-modify-write) and useless since we are not freeing any
+	 * space by doing so.
 	 */
-	if (!io_is_secure_erase(bio, rq)) {
-		start = P2ROUNDUP(start, zv->zv_volblocksize);
-		end = P2ALIGN_TYPED(end, zv->zv_volblocksize, uint64_t);
-		size = end - start;
-	}
+	start = P2ROUNDUP(start, zv->zv_volblocksize);
+	end = P2ALIGN_TYPED(end, zv->zv_volblocksize, uint64_t);
+	size = end - start;
 
 	if (start >= end)
 		goto unlock;
@@ -506,6 +504,24 @@ zvol_read_task(void *arg)
 	zv_request_task_free(task);
 }
 
+/*
+ * Note:
+ *
+ * The kernel uses different enum names for the IO opcode, depending on the
+ * kernel version ('req_opf', 'req_op').  To sidestep this, use macros rather
+ * than inline functions for these checks.
+ */
+/* Should this IO go down the zvol write path? */
+#define	ZVOL_OP_IS_WRITE(op) \
+	(op == REQ_OP_WRITE || \
+	op == REQ_OP_FLUSH || \
+	op == REQ_OP_DISCARD)
+
+/* Is this IO type supported by zvols? */
+#define	ZVOL_OP_IS_SUPPORTED(op) (op == REQ_OP_READ || ZVOL_OP_IS_WRITE(op))
+
+/* Get the IO opcode */
+#define	ZVOL_OP(bio, rq) (bio != NULL ? bio_op(bio) : req_op(rq))
 
 /*
  * Process a BIO or request
@@ -525,27 +541,32 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 	uint64_t size = io_size(bio, rq);
 	int rw;
 
-	if (rq != NULL) {
-		/*
-		 * Flush & trim requests go down the zvol_write codepath.  Or
-		 * more specifically:
-		 *
-		 * If request is a write, or if it's op_is_sync() and not a
-		 * read, or if it's a flush, or if it's a discard, then send the
-		 * request down the write path.
-		 */
-		if (op_is_write(rq->cmd_flags) ||
-		    (op_is_sync(rq->cmd_flags) && req_op(rq) != REQ_OP_READ) ||
-		    req_op(rq) == REQ_OP_FLUSH ||
-		    op_is_discard(rq->cmd_flags)) {
-			rw = WRITE;
-		} else {
-			rw = READ;
-		}
+	if (unlikely(!ZVOL_OP_IS_SUPPORTED(ZVOL_OP(bio, rq)))) {
+		zfs_dbgmsg("Unsupported zvol %s, op=%d, flags=0x%x",
+		    rq != NULL ? "request" : "BIO",
+		    ZVOL_OP(bio, rq),
+		    rq != NULL ? rq->cmd_flags : bio->bi_opf);
+		ASSERT(ZVOL_OP_IS_SUPPORTED(ZVOL_OP(bio, rq)));
+		zvol_end_io(bio, rq, SET_ERROR(ENOTSUPP));
+		goto out;
+	}
+
+	if (ZVOL_OP_IS_WRITE(ZVOL_OP(bio, rq))) {
+		rw = WRITE;
 	} else {
-		rw = bio_data_dir(bio);
+		rw = READ;
 	}
 
+	/*
+	 * Sanity check
+	 *
+	 * If we're a BIO, check our rw matches the kernel's
+	 * bio_data_dir(bio) rw.  We need to check because we support fewer
+	 * IO operations, and want to verify that what we think are reads and
+	 * writes from those operations match what the kernel thinks.
+	 */
+	ASSERT(rq != NULL || rw == bio_data_dir(bio));
+
 	if (unlikely(zv->zv_flags & ZVOL_REMOVING)) {
 		zvol_end_io(bio, rq, -SET_ERROR(ENXIO));
 		goto out;
@@ -649,7 +670,7 @@ zvol_request_impl(zvol_state_t *zv, struct bio *bio, struct request *rq,
 		 * interfaces lack this functionality (they block waiting for
 		 * the i/o to complete).
 		 */
-		if (io_is_discard(bio, rq) || io_is_secure_erase(bio, rq)) {
+		if (io_is_discard(bio, rq)) {
 			if (force_sync) {
 				zvol_discard(&zvr);
 			} else {
diff --git a/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh b/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh
index bb697b76a9f3..9de308c4a11a 100755
--- a/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh
+++ b/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh
@@ -41,6 +41,7 @@
 # 5. TRIM the first 1MB and last 2MB of the 5MB block of data.
 # 6. Observe 2MB of used space on the zvol
 # 7. Verify the trimmed regions are zero'd on the zvol
+# 8. Verify Secure Erase does not work on zvols (Linux only)
 
 verify_runnable "global"
 
@@ -56,6 +57,7 @@ if is_linux ; then
 	else
 		trimcmd='blkdiscard'
 	fi
+	secure_trimcmd="$trimcmd --secure"
 else
 	# By default, FreeBSD 'trim' always does a dry-run.  '-f' makes
 	# it perform the actual operation.
@@ -114,6 +116,11 @@ function do_test {
 	log_must diff $datafile1 $datafile2
 
 	log_must rm $datafile1 $datafile2
+
+	# Secure erase should not work (Linux check only).
+	if [ -n "$secure_trimcmd" ] ; then
+		log_mustnot $secure_trimcmd $zvolpath
+	fi
 }
 
 log_assert "Verify that a ZFS volume can be TRIMed"

From a60214e792153e6c122a89f6543d9f96af8df153 Mon Sep 17 00:00:00 2001
From: Robert Evans <evansr@google.com>
Date: Thu, 25 Sep 2025 14:06:28 -0400
Subject: [PATCH 20/23] dnode_next_offset: backtrack if lower level does not
 match

This changes the basic search algorithm from a single search up and down
the tree to a full depth-first traversal to handle conditions where the
tree matches at a higher level but not a lower level.

Normally higher level blocks always point to matching blocks, but there
are cases where this does not happen:

1. Racing block pointer updates from dbuf_write_ready.

   Before f664f1ee7f (#8946), both dbuf_write_ready and
   dnode_next_offset held dn_struct_rwlock which protected against
   pointer writes from concurrent syncs.

   This no longer applies, so sync context can f.e. clear or fill all
   L1->L0 BPs before the L2->L1 BP and higher BP's are updated.

   dnode_free_range in particular can reach this case and skip over L1
   blocks that need to be dirtied. Later, sync will panic in
   free_children when trying to clear a non-dirty indirect block.

   This case was found with ztest.

2. txg > 0, non-hole case. This is #11196.

   Freeing blocks/dnodes breaks the assumption that a match at a higher
   level implies a match at a lower level when filtering txg > 0.

   Whenever some but not all L0 blocks are freed, the parent L1 block is
   rewritten. Its updated L2->L1 BP reflects a newer birth txg.

   Later when searching by txg, if the L1 block matches since the txg is
   newer, it is possible that none of the remaining L1->L0 BPs match if
   none have been updated.

   The same behavior is possible with dnode search at L0.

   This is reachable from dsl_destroy_head for synchronous freeing.
   When this happens open context fails to free objects leaving sync
   context stuck freeing potentially many objects.

   This is also reachable from traverse_pool for extreme rewind where it
   is theoretically possible that datasets not dirtied after txg are
   skipped if the MOS has high enough indirection to trigger this case.

In both of these cases, without backtracking the search ends prematurely
as ESRCH result implies no more matches in the entire object.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Akash B <akash-b@hpe.com>
Signed-off-by: Robert Evans <evansr@google.com>
Closes #16025
Closes #11196
---
 module/zfs/dnode.c | 65 ++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 54 insertions(+), 11 deletions(-)

diff --git a/module/zfs/dnode.c b/module/zfs/dnode.c
index 5a067fd59499..352147a440b2 100644
--- a/module/zfs/dnode.c
+++ b/module/zfs/dnode.c
@@ -2694,6 +2694,32 @@ dnode_next_offset_level(dnode_t *dn, int flags, uint64_t *offset,
 	return (error);
 }
 
+/*
+ * Adjust *offset to the next (or previous) block byte offset at lvl.
+ * Returns FALSE if *offset would overflow or underflow.
+ */
+static boolean_t
+dnode_next_block(dnode_t *dn, int flags, uint64_t *offset, int lvl)
+{
+	int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
+	int span = lvl * epbs + dn->dn_datablkshift;
+	uint64_t blkid, maxblkid;
+
+	if (span >= 8 * sizeof (uint64_t))
+		return (B_FALSE);
+
+	blkid = *offset >> span;
+	maxblkid = 1ULL << (8 * sizeof (*offset) - span);
+	if (!(flags & DNODE_FIND_BACKWARDS) && blkid + 1 < maxblkid)
+		*offset = (blkid + 1) << span;
+	else if ((flags & DNODE_FIND_BACKWARDS) && blkid > 0)
+		*offset = (blkid << span) - 1;
+	else
+		return (B_FALSE);
+
+	return (B_TRUE);
+}
+
 /*
  * Find the next hole, data, or sparse region at or after *offset.
  * The value 'blkfill' tells us how many items we expect to find
@@ -2721,7 +2747,7 @@ int
 dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset,
     int minlvl, uint64_t blkfill, uint64_t txg)
 {
-	uint64_t initial_offset = *offset;
+	uint64_t matched = *offset;
 	int lvl, maxlvl;
 	int error = 0;
 
@@ -2745,16 +2771,36 @@ dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset,
 
 	maxlvl = dn->dn_phys->dn_nlevels;
 
-	for (lvl = minlvl; lvl <= maxlvl; lvl++) {
+	for (lvl = minlvl; lvl <= maxlvl; ) {
 		error = dnode_next_offset_level(dn,
 		    flags, offset, lvl, blkfill, txg);
-		if (error != ESRCH)
+		if (error == 0 && lvl > minlvl) {
+			--lvl;
+			matched = *offset;
+		} else if (error == ESRCH && lvl < maxlvl &&
+		    dnode_next_block(dn, flags, &matched, lvl)) {
+			/*
+			 * Continue search at next/prev offset in lvl+1 block.
+			 *
+			 * Usually we only search upwards at the start of the
+			 * search as higher level blocks point at a matching
+			 * minlvl block in most cases, but we backtrack if not.
+			 *
+			 * This can happen for txg > 0 searches if the block
+			 * contains only BPs/dnodes freed at that txg. It also
+			 * happens if we are still syncing out the tree, and
+			 * some BP's at higher levels are not updated yet.
+			 *
+			 * We must adjust offset to avoid coming back to the
+			 * same offset and getting stuck looping forever. This
+			 * also deals with the case where offset is already at
+			 * the beginning or end of the object.
+			 */
+			++lvl;
+			*offset = matched;
+		} else {
 			break;
-	}
-
-	while (error == 0 && --lvl >= minlvl) {
-		error = dnode_next_offset_level(dn,
-		    flags, offset, lvl, blkfill, txg);
+		}
 	}
 
 	/*
@@ -2766,9 +2812,6 @@ dnode_next_offset(dnode_t *dn, int flags, uint64_t *offset,
 		error = 0;
 	}
 
-	if (error == 0 && (flags & DNODE_FIND_BACKWARDS ?
-	    initial_offset < *offset : initial_offset > *offset))
-		error = SET_ERROR(ESRCH);
 out:
 	if (!(flags & DNODE_FIND_HAVELOCK))
 		rw_exit(&dn->dn_struct_rwlock);

From 81ceee0cff137936d92f4a1bedc6a40c2f8c9e4f Mon Sep 17 00:00:00 2001
From: Alexander Motin <alexander.motin@TrueNAS.com>
Date: Sat, 13 Sep 2025 12:58:48 -0400
Subject: [PATCH 21/23] Fix two infinite loops if dmu_prefetch_max set to zero

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <alexander.motin@TrueNAS.com>
Closes #17692
Closes #17729
---
 module/zfs/dmu.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/module/zfs/dmu.c b/module/zfs/dmu.c
index 8216786fb5d9..a761fd8b3ee2 100644
--- a/module/zfs/dmu.c
+++ b/module/zfs/dmu.c
@@ -770,6 +770,8 @@ dmu_prefetch_by_dnode(dnode_t *dn, int64_t level, uint64_t offset,
 		 */
 		uint8_t ibps = ibs - SPA_BLKPTRSHIFT;
 		limit = P2ROUNDUP(dmu_prefetch_max, 1 << ibs) >> ibs;
+		if (limit == 0)
+			end2 = start2;
 		do {
 			level2++;
 			start2 = P2ROUNDUP(start2, 1 << ibps) >> ibps;
@@ -1684,8 +1686,8 @@ dmu_object_cached_size(objset_t *os, uint64_t object,
 
 	dmu_object_info_from_dnode(dn, &doi);
 
-	for (uint64_t off = 0; off < doi.doi_max_offset;
-	    off += dmu_prefetch_max) {
+	for (uint64_t off = 0; off < doi.doi_max_offset &&
+	    dmu_prefetch_max > 0; off += dmu_prefetch_max) {
 		/* dbuf_read doesn't prefetch L1 blocks. */
 		dmu_prefetch_by_dnode(dn, 1, off,
 		    dmu_prefetch_max, ZIO_PRIORITY_SYNC_READ);

From 9f2cbea1dc3ac96da4ef916ba60a6ce43f6bc50a Mon Sep 17 00:00:00 2001
From: Alexander Motin <alexander.motin@TrueNAS.com>
Date: Tue, 23 Sep 2025 19:09:37 -0400
Subject: [PATCH 22/23] zdb: Fix asize overflow in verify_livelist_allocs()

Spacemap entry might be too big to fit into a block pointer ashift.
We hit an assertion trying to run `zdb -bvy` on a large pool.  But
it seems the code does not really need size there, since we only
need to search for a range of offsets, so setting it to zero should
just make btree return position just before the first entry.  I
suspect the previous code could actually miss the first entry
due to this if its size was smaller.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Igor Kozhukhov <igor@dilos.org>
Signed-off-by: Alexander Motin <alexander.motin@TrueNAS.com>
Closes #17764
---
 cmd/zdb/zdb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/cmd/zdb/zdb.c b/cmd/zdb/zdb.c
index 8685109db1c2..843bedf48b9f 100644
--- a/cmd/zdb/zdb.c
+++ b/cmd/zdb/zdb.c
@@ -381,7 +381,7 @@ verify_livelist_allocs(metaslab_verify_t *mv, uint64_t txg,
 	sublivelist_verify_block_t svb = {{{0}}};
 	DVA_SET_VDEV(&svb.svb_dva, mv->mv_vdid);
 	DVA_SET_OFFSET(&svb.svb_dva, offset);
-	DVA_SET_ASIZE(&svb.svb_dva, size);
+	DVA_SET_ASIZE(&svb.svb_dva, 0);
 	zfs_btree_index_t where;
 	uint64_t end_offset = offset + size;
 

From 10c460b74f2bf383dfaa4cd195126478b8075ad7 Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Tue, 21 Oct 2025 10:41:28 -0700
Subject: [PATCH 23/23] Tag zfs-2.3.5

META file and changelog updated.

Signed-off-by: Tony Hutter <hutter2@llnl.gov>
---
 META | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/META b/META
index c96da9b726b1..b2c73bee3283 100644
--- a/META
+++ b/META
@@ -1,7 +1,7 @@
 Meta:          1
 Name:          zfs
 Branch:        1.0
-Version:       2.3.4
+Version:       2.3.5
 Release:       1
 Release-Tags:  relext
 License:       CDDL