[pve-devel] [PATCH zfsonlinux] pick bug-fixes staged for 2.2.1
Stoiko Ivanov
s.ivanov at proxmox.com
Fri Nov 17 15:03:02 CET 2023
ZFS 2.2.1 is currently being prepared, but the 3 patches added here
seem quite relevant, as the might cause dataloss/panics on setups
which run `zpool upgrade`.
See upstreams discussion for 2.2.1:
https://github.com/openzfs/zfs/pull/15498/
and the most critical issue:
https://github.com/openzfs/zfs/pull/15529
finally:
https://github.com/openzfs/zfs/commit/459c99ff2339a4a514abcf2255f9b3e5324ef09e
should not hurt either
the change to the UBSAN patch (0013) is unrelate, cosmetic only and
happened by running export-patchqueue.
Signed-off-by: Stoiko Ivanov <s.ivanov at proxmox.com>
---
minimally tested by building our current kernel with this and booting it in
2 VMs - the tunable (module parameter) is present and set to 0
...und-UBSAN-errors-for-variable-arrays.patch | 5 +-
...g-between-unencrypted-and-encrypted-.patch | 44 ++++
...Add-a-tunable-to-disable-BRT-support.patch | 201 ++++++++++++++++++
...2.1-Disable-block-cloning-by-default.patch | 42 ++++
debian/patches/series | 3 +
5 files changed, 291 insertions(+), 4 deletions(-)
create mode 100644 debian/patches/0015-Fix-block-cloning-between-unencrypted-and-encrypted-.patch
create mode 100644 debian/patches/0016-Add-a-tunable-to-disable-BRT-support.patch
create mode 100644 debian/patches/0017-zfs-2.2.1-Disable-block-cloning-by-default.patch
diff --git a/debian/patches/0013-Workaround-UBSAN-errors-for-variable-arrays.patch b/debian/patches/0013-Workaround-UBSAN-errors-for-variable-arrays.patch
index 02815311..0b98c42a 100644
--- a/debian/patches/0013-Workaround-UBSAN-errors-for-variable-arrays.patch
+++ b/debian/patches/0013-Workaround-UBSAN-errors-for-variable-arrays.patch
@@ -1,4 +1,4 @@
-From 28be24aefc13b11e4c96e172cf2685994e03150d Mon Sep 17 00:00:00 2001
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2 at llnl.gov>
Date: Thu, 9 Nov 2023 16:43:35 -0800
Subject: [PATCH] Workaround UBSAN errors for variable arrays
@@ -70,6 +70,3 @@ index c13217159..b9c284a24 100644
# Suppress incorrect warnings from versions of objtool which are not
# aware of x86 EVEX prefix instructions used for AVX512.
OBJECT_FILES_NON_STANDARD_vdev_raidz_math_avx512bw.o := y
---
-2.39.2
-
diff --git a/debian/patches/0015-Fix-block-cloning-between-unencrypted-and-encrypted-.patch b/debian/patches/0015-Fix-block-cloning-between-unencrypted-and-encrypted-.patch
new file mode 100644
index 00000000..c2fc506e
--- /dev/null
+++ b/debian/patches/0015-Fix-block-cloning-between-unencrypted-and-encrypted-.patch
@@ -0,0 +1,44 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Martin=20Matu=C5=A1ka?= <mm at FreeBSD.org>
+Date: Tue, 31 Oct 2023 21:49:41 +0100
+Subject: [PATCH] Fix block cloning between unencrypted and encrypted datasets
+
+Block cloning from an encrypted dataset into an unencrypted dataset
+and vice versa is not possible. The current code did allow cloning
+unencrypted files into an encrypted dataset causing a panic when
+these were accessed. Block cloning between encrypted and encrypted
+is currently supported on the same filesystem only.
+
+Reviewed-by: Alexander Motin <mav at FreeBSD.org>
+Reviewed-by: Kay Pedersen <mail at mkwg.de>
+Reviewed-by: Rob N <robn at despairlabs.com>
+Reviewed-by: Brian Behlendorf <behlendorf1 at llnl.gov>
+Signed-off-by: Martin Matuska <mm at FreeBSD.org>
+Closes #15464
+Closes #15465
+(cherry picked from commit 459c99ff2339a4a514abcf2255f9b3e5324ef09e)
+Signed-off-by: Stoiko Ivanov <s.ivanov at proxmox.com>
+---
+ module/zfs/zfs_vnops.c | 9 +++++++++
+ 1 file changed, 9 insertions(+)
+
+diff --git a/module/zfs/zfs_vnops.c b/module/zfs/zfs_vnops.c
+index 40d6c87a7..84e6b10ef 100644
+--- a/module/zfs/zfs_vnops.c
++++ b/module/zfs/zfs_vnops.c
+@@ -1094,6 +1094,15 @@ zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp,
+
+ ASSERT(!outzfsvfs->z_replay);
+
++ /*
++ * Block cloning from an unencrypted dataset into an encrypted
++ * dataset and vice versa is not supported.
++ */
++ if (inos->os_encrypted != outos->os_encrypted) {
++ zfs_exit_two(inzfsvfs, outzfsvfs, FTAG);
++ return (SET_ERROR(EXDEV));
++ }
++
+ error = zfs_verify_zp(inzp);
+ if (error == 0)
+ error = zfs_verify_zp(outzp);
diff --git a/debian/patches/0016-Add-a-tunable-to-disable-BRT-support.patch b/debian/patches/0016-Add-a-tunable-to-disable-BRT-support.patch
new file mode 100644
index 00000000..53977479
--- /dev/null
+++ b/debian/patches/0016-Add-a-tunable-to-disable-BRT-support.patch
@@ -0,0 +1,201 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Rich Ercolani <214141+rincebrain at users.noreply.github.com>
+Date: Thu, 16 Nov 2023 14:35:22 -0500
+Subject: [PATCH] Add a tunable to disable BRT support.
+
+Copy the disable parameter that FreeBSD implemented, and extend it to
+work on Linux as well, until we're sure this is stable.
+
+Reviewed-by: Alexander Motin <mav at FreeBSD.org>
+Reviewed-by: Brian Behlendorf <behlendorf1 at llnl.gov>
+Signed-off-by: Rich Ercolani <rincebrain at gmail.com>
+Closes #15529
+(cherry picked from commit 87e9e828655c250ce064874ff5df16f870c0a52e)
+Signed-off-by: Stoiko Ivanov <s.ivanov at proxmox.com>
+---
+ include/os/freebsd/zfs/sys/zfs_vfsops_os.h | 1 +
+ include/os/linux/zfs/sys/zfs_vfsops_os.h | 2 ++
+ man/man4/zfs.4 | 5 +++++
+ module/os/freebsd/zfs/zfs_vfsops.c | 4 ++++
+ module/os/freebsd/zfs/zfs_vnops_os.c | 5 +++++
+ module/os/linux/zfs/zfs_vnops_os.c | 4 ++++
+ module/os/linux/zfs/zpl_file_range.c | 5 +++++
+ tests/zfs-tests/include/libtest.shlib | 15 +++++++++++++++
+ tests/zfs-tests/include/tunables.cfg | 1 +
+ .../tests/functional/block_cloning/cleanup.ksh | 4 ++++
+ .../tests/functional/block_cloning/setup.ksh | 5 +++++
+ 11 files changed, 51 insertions(+)
+
+diff --git a/include/os/freebsd/zfs/sys/zfs_vfsops_os.h b/include/os/freebsd/zfs/sys/zfs_vfsops_os.h
+index 24bb03575..56a0ac96a 100644
+--- a/include/os/freebsd/zfs/sys/zfs_vfsops_os.h
++++ b/include/os/freebsd/zfs/sys/zfs_vfsops_os.h
+@@ -286,6 +286,7 @@ typedef struct zfid_long {
+
+ extern uint_t zfs_fsyncer_key;
+ extern int zfs_super_owner;
++extern int zfs_bclone_enabled;
+
+ extern void zfs_init(void);
+ extern void zfs_fini(void);
+diff --git a/include/os/linux/zfs/sys/zfs_vfsops_os.h b/include/os/linux/zfs/sys/zfs_vfsops_os.h
+index b4d5db21f..220466550 100644
+--- a/include/os/linux/zfs/sys/zfs_vfsops_os.h
++++ b/include/os/linux/zfs/sys/zfs_vfsops_os.h
+@@ -45,6 +45,8 @@ extern "C" {
+ typedef struct zfsvfs zfsvfs_t;
+ struct znode;
+
++extern int zfs_bclone_enabled;
++
+ /*
+ * This structure emulates the vfs_t from other platforms. It's purpose
+ * is to facilitate the handling of mount options and minimize structural
+diff --git a/man/man4/zfs.4 b/man/man4/zfs.4
+index cfadd79d8..32f1765a5 100644
+--- a/man/man4/zfs.4
++++ b/man/man4/zfs.4
+@@ -1137,6 +1137,11 @@ Selecting any option other than
+ results in vector instructions
+ from the respective CPU instruction set being used.
+ .
++.It Sy zfs_bclone_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
++Enable the experimental block cloning feature.
++If this setting is 0, then even if feature at block_cloning is enabled,
++attempts to clone blocks will act as though the feature is disabled.
++.
+ .It Sy zfs_blake3_impl Ns = Ns Sy fastest Pq string
+ Select a BLAKE3 implementation.
+ .Pp
+diff --git a/module/os/freebsd/zfs/zfs_vfsops.c b/module/os/freebsd/zfs/zfs_vfsops.c
+index e8b9ada13..09e18de81 100644
+--- a/module/os/freebsd/zfs/zfs_vfsops.c
++++ b/module/os/freebsd/zfs/zfs_vfsops.c
+@@ -89,6 +89,10 @@ int zfs_debug_level;
+ SYSCTL_INT(_vfs_zfs, OID_AUTO, debug, CTLFLAG_RWTUN, &zfs_debug_level, 0,
+ "Debug level");
+
++int zfs_bclone_enabled = 1;
++SYSCTL_INT(_vfs_zfs, OID_AUTO, bclone_enabled, CTLFLAG_RWTUN,
++ &zfs_bclone_enabled, 0, "Enable block cloning");
++
+ struct zfs_jailparam {
+ int mount_snapshot;
+ };
+diff --git a/module/os/freebsd/zfs/zfs_vnops_os.c b/module/os/freebsd/zfs/zfs_vnops_os.c
+index c498a1328..f672deed3 100644
+--- a/module/os/freebsd/zfs/zfs_vnops_os.c
++++ b/module/os/freebsd/zfs/zfs_vnops_os.c
+@@ -6243,6 +6243,11 @@ zfs_freebsd_copy_file_range(struct vop_copy_file_range_args *ap)
+ int error;
+ uint64_t len = *ap->a_lenp;
+
++ if (!zfs_bclone_enabled) {
++ mp = NULL;
++ goto bad_write_fallback;
++ }
++
+ /*
+ * TODO: If offset/length is not aligned to recordsize, use
+ * vn_generic_copy_file_range() on this fragment.
+diff --git a/module/os/linux/zfs/zfs_vnops_os.c b/module/os/linux/zfs/zfs_vnops_os.c
+index 33baac9db..76fac3a02 100644
+--- a/module/os/linux/zfs/zfs_vnops_os.c
++++ b/module/os/linux/zfs/zfs_vnops_os.c
+@@ -4229,4 +4229,8 @@ EXPORT_SYMBOL(zfs_map);
+ module_param(zfs_delete_blocks, ulong, 0644);
+ MODULE_PARM_DESC(zfs_delete_blocks, "Delete files larger than N blocks async");
+
++/* CSTYLED */
++module_param(zfs_bclone_enabled, uint, 0644);
++MODULE_PARM_DESC(zfs_bclone_enabled, "Enable block cloning");
++
+ #endif
+diff --git a/module/os/linux/zfs/zpl_file_range.c b/module/os/linux/zfs/zpl_file_range.c
+index c47fe99da..73476ff40 100644
+--- a/module/os/linux/zfs/zpl_file_range.c
++++ b/module/os/linux/zfs/zpl_file_range.c
+@@ -31,6 +31,8 @@
+ #include <sys/zfs_vnops.h>
+ #include <sys/zfeature.h>
+
++int zfs_bclone_enabled = 1;
++
+ /*
+ * Clone part of a file via block cloning.
+ *
+@@ -50,6 +52,9 @@ __zpl_clone_file_range(struct file *src_file, loff_t src_off,
+ fstrans_cookie_t cookie;
+ int err;
+
++ if (!zfs_bclone_enabled)
++ return (-EOPNOTSUPP);
++
+ if (!spa_feature_is_enabled(
+ dmu_objset_spa(ITOZSB(dst_i)->z_os), SPA_FEATURE_BLOCK_CLONING))
+ return (-EOPNOTSUPP);
+diff --git a/tests/zfs-tests/include/libtest.shlib b/tests/zfs-tests/include/libtest.shlib
+index 844caa17d..d5d7bb6c8 100644
+--- a/tests/zfs-tests/include/libtest.shlib
++++ b/tests/zfs-tests/include/libtest.shlib
+@@ -3334,6 +3334,21 @@ function set_tunable_impl
+ esac
+ }
+
++function save_tunable
++{
++ [[ ! -d $TEST_BASE_DIR ]] && return 1
++ [[ -e $TEST_BASE_DIR/tunable-$1 ]] && return 2
++ echo "$(get_tunable """$1""")" > "$TEST_BASE_DIR"/tunable-"$1"
++}
++
++function restore_tunable
++{
++ [[ ! -e $TEST_BASE_DIR/tunable-$1 ]] && return 1
++ val="$(cat $TEST_BASE_DIR/tunable-"""$1""")"
++ set_tunable64 "$1" "$val"
++ rm $TEST_BASE_DIR/tunable-$1
++}
++
+ #
+ # Get a global system tunable
+ #
+diff --git a/tests/zfs-tests/include/tunables.cfg b/tests/zfs-tests/include/tunables.cfg
+index 80e7bcb3b..a0edad14d 100644
+--- a/tests/zfs-tests/include/tunables.cfg
++++ b/tests/zfs-tests/include/tunables.cfg
+@@ -90,6 +90,7 @@ VOL_INHIBIT_DEV UNSUPPORTED zvol_inhibit_dev
+ VOL_MODE vol.mode zvol_volmode
+ VOL_RECURSIVE vol.recursive UNSUPPORTED
+ VOL_USE_BLK_MQ UNSUPPORTED zvol_use_blk_mq
++BCLONE_ENABLED zfs_bclone_enabled zfs_bclone_enabled
+ XATTR_COMPAT xattr_compat zfs_xattr_compat
+ ZEVENT_LEN_MAX zevent.len_max zfs_zevent_len_max
+ ZEVENT_RETAIN_MAX zevent.retain_max zfs_zevent_retain_max
+diff --git a/tests/zfs-tests/tests/functional/block_cloning/cleanup.ksh b/tests/zfs-tests/tests/functional/block_cloning/cleanup.ksh
+index 7ac13adb6..b985445a5 100755
+--- a/tests/zfs-tests/tests/functional/block_cloning/cleanup.ksh
++++ b/tests/zfs-tests/tests/functional/block_cloning/cleanup.ksh
+@@ -31,4 +31,8 @@ verify_runnable "global"
+
+ default_cleanup_noexit
+
++if tunable_exists BCLONE_ENABLED ; then
++ log_must restore_tunable BCLONE_ENABLED
++fi
++
+ log_pass
+diff --git a/tests/zfs-tests/tests/functional/block_cloning/setup.ksh b/tests/zfs-tests/tests/functional/block_cloning/setup.ksh
+index 512f5a064..58441bf8f 100755
+--- a/tests/zfs-tests/tests/functional/block_cloning/setup.ksh
++++ b/tests/zfs-tests/tests/functional/block_cloning/setup.ksh
+@@ -33,4 +33,9 @@ fi
+
+ verify_runnable "global"
+
++if tunable_exists BCLONE_ENABLED ; then
++ log_must save_tunable BCLONE_ENABLED
++ log_must set_tunable32 BCLONE_ENABLED 1
++fi
++
+ log_pass
diff --git a/debian/patches/0017-zfs-2.2.1-Disable-block-cloning-by-default.patch b/debian/patches/0017-zfs-2.2.1-Disable-block-cloning-by-default.patch
new file mode 100644
index 00000000..53a088da
--- /dev/null
+++ b/debian/patches/0017-zfs-2.2.1-Disable-block-cloning-by-default.patch
@@ -0,0 +1,42 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Tony Hutter <hutter2 at llnl.gov>
+Date: Thu, 16 Nov 2023 11:42:19 -0800
+Subject: [PATCH] zfs-2.2.1: Disable block cloning by default
+
+Disable block cloning by default to mitigate possible data corruption
+(see #15529 and #15526).
+
+Signed-off-by: Tony Hutter <hutter2 at llnl.gov>
+(cherry picked from commit 479dca51c66a731e637bd2d4f9bba01a05f9ac9f)
+Signed-off-by: Stoiko Ivanov <s.ivanov at proxmox.com>
+---
+ module/os/freebsd/zfs/zfs_vfsops.c | 2 +-
+ module/os/linux/zfs/zpl_file_range.c | 2 +-
+ 2 files changed, 2 insertions(+), 2 deletions(-)
+
+diff --git a/module/os/freebsd/zfs/zfs_vfsops.c b/module/os/freebsd/zfs/zfs_vfsops.c
+index 09e18de81..0ac670ed9 100644
+--- a/module/os/freebsd/zfs/zfs_vfsops.c
++++ b/module/os/freebsd/zfs/zfs_vfsops.c
+@@ -89,7 +89,7 @@ int zfs_debug_level;
+ SYSCTL_INT(_vfs_zfs, OID_AUTO, debug, CTLFLAG_RWTUN, &zfs_debug_level, 0,
+ "Debug level");
+
+-int zfs_bclone_enabled = 1;
++int zfs_bclone_enabled = 0;
+ SYSCTL_INT(_vfs_zfs, OID_AUTO, bclone_enabled, CTLFLAG_RWTUN,
+ &zfs_bclone_enabled, 0, "Enable block cloning");
+
+diff --git a/module/os/linux/zfs/zpl_file_range.c b/module/os/linux/zfs/zpl_file_range.c
+index 73476ff40..139c51cf4 100644
+--- a/module/os/linux/zfs/zpl_file_range.c
++++ b/module/os/linux/zfs/zpl_file_range.c
+@@ -31,7 +31,7 @@
+ #include <sys/zfs_vnops.h>
+ #include <sys/zfeature.h>
+
+-int zfs_bclone_enabled = 1;
++int zfs_bclone_enabled = 0;
+
+ /*
+ * Clone part of a file via block cloning.
diff --git a/debian/patches/series b/debian/patches/series
index 5927d521..d20a6054 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -12,3 +12,6 @@
0012-Fix-nfs_truncate_shares-without-etc-exports.d.patch
0013-Workaround-UBSAN-errors-for-variable-arrays.patch
0014-zpool-status-tighten-bounds-for-noalloc-stat-availab.patch
+0015-Fix-block-cloning-between-unencrypted-and-encrypted-.patch
+0016-Add-a-tunable-to-disable-BRT-support.patch
+0017-zfs-2.2.1-Disable-block-cloning-by-default.patch
--
2.39.2
More information about the pve-devel
mailing list