From nobody Wed May 03 10:25:56 2023 X-Original-To: dev-commits-src-all@mlmmj.nyi.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) by mlmmj.nyi.freebsd.org (Postfix) with ESMTP id 4QBClb0vtlz49BlF; Wed, 3 May 2023 10:26:03 +0000 (UTC) (envelope-from git@FreeBSD.org) Received: from mxrelay.nyi.freebsd.org (mxrelay.nyi.freebsd.org [IPv6:2610:1c1:1:606c::19:3]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256 client-signature RSA-PSS (4096 bits) client-digest SHA256) (Client CN "mxrelay.nyi.freebsd.org", Issuer "R3" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 4QBClb0c6Fz3Npb; Wed, 3 May 2023 10:26:03 +0000 (UTC) (envelope-from git@FreeBSD.org) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=freebsd.org; s=dkim; t=1683109563; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding; bh=Z0inLDJVFF1fbJi4sDxdN/Xd+ZGfui/h/MGMh2wA0Ls=; b=hA85Mo6xLkfCS1Cdl30owlHMLbgk1pPuUYiAw+Wwuvda/Bl7ZGCAkOk1K301zEHAvXKugp 8ez/RAcywJhVtnZBpWb7+1c5SL0vsHYR04lYA4fF+F4NI2Uk9Y7daRNjbjkFrUr1zeK1SU xdspQqm3ty2VPlWOMmtJIpJddCiQrU97jGBJERDgZG3zwH6p7thgrURQX7RdT9qVnb9De0 YdlAOMBRxsxt6I8pSNG63nc4LKrF73hn3DQefGTsih20II/4yt1cr3bmGXBBPdF8vOtI/r BKbmoLs09zrEhUzJNJNMSANMNod/lZFcfmZi/A9Hlew1BDZILYNq1falTs/51A== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=freebsd.org; s=dkim; t=1683109563; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding; bh=Z0inLDJVFF1fbJi4sDxdN/Xd+ZGfui/h/MGMh2wA0Ls=; b=bNx6ytIbIPeNZ0ZqIenT52N+eMb//u4cU2iHpkHWzqWEMWw3JczUqCtbqomcSWRIIn5zNE 4/JTzO6OXF1COOef+C/GJ43z2ulMzR0somzFiL/Ivd2FUdwxb4voC9NBmOEsH8csaFZL9N OJ2KLWwudp82+lfLe58b9VhASURFxQfG3aHsHU/7Sw4H4GNjyzMDvldmPBQbt0okpOvXeG yQKHis8y5iCczFak8ke8phYcT+RUrcbsax4+EFVVqK2LeIXUIZxr4+n/+iPdY3OD3m2J9h VM57et8rZDTA/u1AqB41kL95jK9/eC94oWQCYx0E97CzInpRc+B4irb1hcq4lA== ARC-Authentication-Results: i=1; mx1.freebsd.org; none ARC-Seal: i=1; s=dkim; d=freebsd.org; t=1683109563; a=rsa-sha256; cv=none; b=FuHly2Q+yAifyB7N3ikzRhyNukkeAs8hclBTWjPHahsXoAUCVXL08TlhYCQ73PC3FiLvnX 2R/DuriDIvm7Pff6ct84gZc10+LjGyc3hTVCpdArowOHgWrMMn0lNaJdlGvqL3mecsHll+ CPYndvJT2kURyLJPRecOJKwyXzM6Cg36GxUIsGcOXRPikPq83ImKmAg0OVCwoYlA/BRMhT CDtZ8getRdq0zBNdkr/GQYSNyvGg3HTAMJx8vdCYJbyKDQjLaDsgLJDvybkPmD/KcYxKWG uFQnCK5JR1K7P15DTBlg8ZQ7dEagHKCRgGLvlCFgsHP06tyt1kxrGtiBtAM9mg== Received: from gitrepo.freebsd.org (gitrepo.freebsd.org [IPv6:2610:1c1:1:6068::e6a:5]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (Client did not present a certificate) by mxrelay.nyi.freebsd.org (Postfix) with ESMTPS id 4QBClZ6m1mzRRb; Wed, 3 May 2023 10:26:02 +0000 (UTC) (envelope-from git@FreeBSD.org) Received: from gitrepo.freebsd.org ([127.0.1.44]) by gitrepo.freebsd.org (8.16.1/8.16.1) with ESMTP id 343AQ2QX063823; Wed, 3 May 2023 10:26:02 GMT (envelope-from git@gitrepo.freebsd.org) Received: (from git@localhost) by gitrepo.freebsd.org (8.16.1/8.16.1/Submit) id 343APu5b063807; Wed, 3 May 2023 10:25:56 GMT (envelope-from git) Date: Wed, 3 May 2023 10:25:56 GMT Message-Id: <202305031025.343APu5b063807@gitrepo.freebsd.org> To: src-committers@FreeBSD.org, dev-commits-src-all@FreeBSD.org, dev-commits-src-main@FreeBSD.org From: Martin Matuska Subject: git: d411c1d696ef - main - zfs: merge openzfs/zfs@d96e29576 List-Id: Commit messages for all branches of the src repository List-Archive: https://lists.freebsd.org/archives/dev-commits-src-all List-Help: List-Post: List-Subscribe: List-Unsubscribe: Sender: owner-dev-commits-src-all@freebsd.org X-BeenThere: dev-commits-src-all@freebsd.org MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit X-Git-Committer: mm X-Git-Repository: src X-Git-Refname: refs/heads/main X-Git-Reftype: branch X-Git-Commit: d411c1d696ef35d60f8c3564e5eef7aeafa2fece Auto-Submitted: auto-generated X-ThisMailContainsUnwantedMimeParts: N The branch main has been updated by mm: URL: https://cgit.FreeBSD.org/src/commit/?id=d411c1d696ef35d60f8c3564e5eef7aeafa2fece commit d411c1d696ef35d60f8c3564e5eef7aeafa2fece Merge: 16303d2ba6b0 d96e29576c89 Author: Martin Matuska AuthorDate: 2023-05-03 10:04:55 +0000 Commit: Martin Matuska CommitDate: 2023-05-03 10:04:55 +0000 zfs: merge openzfs/zfs@d96e29576 Notable upstream pull request merges: #11680 Add support for zpool user properties #14145 Storage device expansion "silently" fails on degraded vdev #14405 Create zap for root vdev #14659 Allow MMP to bypass waiting for other threads #14674 Miscellaneous FreBSD compilation bugfixes #14692 Fix some signedness issues in arc_evict() #14702 Fix typo in check_clones() #14715 module: small fixes for FreeBSD/aarch64 #14716 Trim needless zeroes from checksum events #14719 vdev: expose zfs_vdev_max_ms_shift as a module parameter #14722 Fix "Detach spare vdev in case if resilvering does not happen" #14723 freebsd clone range fixes #14728 Fix BLAKE3 aarch64 assembly for FreeBSD and macOS #14735 Fix in check_filesystem() #14739 Fix data corruption when cloning embedded blocks #14758 Fix VERIFY(!zil_replaying(zilog, tx)) panic #14761 Revert "ZFS_IOC_COUNT_FILLED does unnecessary txg_wait_synced()" #14774 FreeBSD .zfs fixups #14776 FreeBSD: make zfs_vfs_held() definition consistent with declaration #14779 powerpc64: Support ELFv2 asm on Big Endian #14788 FreeBSD: add missing vop_fplookup assignments #14789 PAM: support the authentication facility #14790 Revert "Fix data race between zil_commit() and zil_suspend()" #14795 Fix positive ABD size assertion in abd_verify() #14798 Mark TX_COMMIT transaction with TXG_NOTHROTTLE #14804 Correct ABD size for split block ZIOs #14806 Use correct block pointer in block cloning case. #14808 blake3: fix up bogus checksums in face of cpu migration Obtained from: OpenZFS OpenZFS commit: d96e29576c89e6e547cb82b477651d2b85ea0fed lib/libbe/be.c | 8 +- lib/libbe/be_access.c | 6 +- lib/libbe/be_info.c | 4 +- sys/contrib/openzfs/META | 2 +- sys/contrib/openzfs/cmd/zdb/zdb.c | 39 ++++++++- sys/contrib/openzfs/cmd/zfs/zfs_iter.c | 6 +- sys/contrib/openzfs/cmd/zfs/zfs_main.c | 47 ++++++---- sys/contrib/openzfs/cmd/zpool/zpool_main.c | 99 ++++++++++++++++------ sys/contrib/openzfs/config/kernel-acl.m4 | 34 ++++++-- .../openzfs/config/kernel-generic_fillattr.m4 | 33 ++++++-- sys/contrib/openzfs/config/kernel-inode-create.m4 | 41 +++++++-- sys/contrib/openzfs/config/kernel-inode-getattr.m4 | 63 ++++++++++---- .../openzfs/config/kernel-inode-permission.m4 | 35 ++++++-- sys/contrib/openzfs/config/kernel-inode-setattr.m4 | 87 +++++++++++++++++++ .../openzfs/config/kernel-is_owner_or_cap.m4 | 25 +++++- sys/contrib/openzfs/config/kernel-mkdir.m4 | 55 +++++++++--- sys/contrib/openzfs/config/kernel-mknod.m4 | 34 +++++++- sys/contrib/openzfs/config/kernel-rename.m4 | 56 ++++++++---- .../openzfs/config/kernel-setattr-prepare.m4 | 44 +++++++--- sys/contrib/openzfs/config/kernel-symlink.m4 | 33 ++++++-- sys/contrib/openzfs/config/kernel-tmpfile.m4 | 33 ++++++-- sys/contrib/openzfs/config/kernel-writepage_t.m4 | 26 ++++++ sys/contrib/openzfs/config/kernel-xattr-handler.m4 | 91 ++++++++++++-------- sys/contrib/openzfs/config/kernel.m4 | 8 +- .../openzfs/contrib/debian/control.modules.in | 4 +- .../contrib/dracut/90zfs/module-setup.sh.in | 3 + .../contrib/dracut/90zfs/zfs-env-bootfs.service.in | 15 +++- .../contrib/dracut/90zfs/zfs-generator.sh.in | 67 ++------------- .../openzfs/contrib/dracut/90zfs/zfs-lib.sh.in | 2 +- .../90zfs/zfs-nonroot-necessities.service.in | 20 +++++ .../dracut/90zfs/zfs-rollback-bootfs.service.in | 3 +- .../dracut/90zfs/zfs-snapshot-bootfs.service.in | 3 +- sys/contrib/openzfs/contrib/dracut/Makefile.am | 1 + sys/contrib/openzfs/contrib/initramfs/README.md | 25 +++++- sys/contrib/openzfs/contrib/initramfs/hooks/zfs.in | 3 + sys/contrib/openzfs/contrib/initramfs/scripts/zfs | 10 +++ .../openzfs/contrib/pam_zfs_key/pam_zfs_key.c | 67 +++++++++++---- sys/contrib/openzfs/include/libzfs.h | 26 ++++-- sys/contrib/openzfs/include/libzutil.h | 1 + .../include/os/freebsd/spl/sys/simd_aarch64.h | 6 +- .../openzfs/include/os/freebsd/spl/sys/types.h | 2 +- .../include/os/freebsd/zfs/sys/zfs_vnops_os.h | 10 +-- .../include/os/linux/kernel/linux/blkdev_compat.h | 2 +- .../include/os/linux/kernel/linux/vfs_compat.h | 21 ++++- .../include/os/linux/kernel/linux/xattr_compat.h | 17 +++- .../openzfs/include/os/linux/spl/sys/cred.h | 30 ++++--- .../openzfs/include/os/linux/spl/sys/isa_defs.h | 18 +++- .../openzfs/include/os/linux/spl/sys/types.h | 15 +++- .../openzfs/include/os/linux/zfs/sys/policy.h | 6 +- .../openzfs/include/os/linux/zfs/sys/trace_zil.h | 8 ++ .../include/os/linux/zfs/sys/zfs_vnops_os.h | 15 ++-- sys/contrib/openzfs/include/os/linux/zfs/sys/zpl.h | 13 ++- sys/contrib/openzfs/include/sys/dmu_objset.h | 7 ++ sys/contrib/openzfs/include/sys/dnode.h | 10 ++- sys/contrib/openzfs/include/sys/fs/zfs.h | 1 + sys/contrib/openzfs/include/sys/spa.h | 3 + sys/contrib/openzfs/include/sys/vdev_impl.h | 1 + sys/contrib/openzfs/include/sys/zfs_acl.h | 10 +-- sys/contrib/openzfs/include/zfeature_common.h | 1 + .../openzfs/lib/libspl/include/sys/isa_defs.h | 18 +++- sys/contrib/openzfs/lib/libzfs/libzfs.abi | 81 +++++++++++++++--- sys/contrib/openzfs/lib/libzfs/libzfs_changelist.c | 7 +- sys/contrib/openzfs/lib/libzfs/libzfs_crypto.c | 2 +- sys/contrib/openzfs/lib/libzfs/libzfs_dataset.c | 18 ++-- sys/contrib/openzfs/lib/libzfs/libzfs_iter.c | 75 +++++++++++++--- sys/contrib/openzfs/lib/libzfs/libzfs_mount.c | 4 +- sys/contrib/openzfs/lib/libzfs/libzfs_pool.c | 1 + sys/contrib/openzfs/lib/libzfs/libzfs_sendrecv.c | 21 ++--- sys/contrib/openzfs/lib/libzfs/libzfs_util.c | 2 +- sys/contrib/openzfs/lib/libzutil/zutil_import.c | 5 +- sys/contrib/openzfs/man/man4/zfs.4 | 10 ++- sys/contrib/openzfs/man/man7/dracut.zfs.7 | 16 ++-- sys/contrib/openzfs/man/man7/zfsprops.7 | 12 ++- sys/contrib/openzfs/man/man7/zpool-features.7 | 16 ++++ sys/contrib/openzfs/man/man7/zpoolconcepts.7 | 64 +++++++------- sys/contrib/openzfs/module/Makefile.bsd | 10 +++ sys/contrib/openzfs/module/lua/ldo.c | 2 + sys/contrib/openzfs/module/lua/setjmp/setjmp.S | 2 + .../openzfs/module/lua/setjmp/setjmp_loongarch64.S | 82 ++++++++++++++++++ .../openzfs/module/os/freebsd/zfs/sysctl_os.c | 8 -- .../openzfs/module/os/freebsd/zfs/zfs_acl.c | 10 +-- .../openzfs/module/os/freebsd/zfs/zfs_vnops_os.c | 15 ++-- sys/contrib/openzfs/module/os/linux/spl/spl-cred.c | 12 +++ sys/contrib/openzfs/module/os/linux/zfs/policy.c | 13 +-- sys/contrib/openzfs/module/os/linux/zfs/zfs_acl.c | 36 ++++---- sys/contrib/openzfs/module/os/linux/zfs/zfs_dir.c | 4 +- .../openzfs/module/os/linux/zfs/zfs_ioctl_os.c | 4 + .../openzfs/module/os/linux/zfs/zfs_vfsops.c | 4 +- .../openzfs/module/os/linux/zfs/zfs_vnops_os.c | 35 ++++---- .../openzfs/module/os/linux/zfs/zfs_znode.c | 2 +- .../openzfs/module/os/linux/zfs/zpl_ctldir.c | 53 +++++++++--- sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c | 38 +++++++-- .../openzfs/module/os/linux/zfs/zpl_inode.c | 77 ++++++++++++----- .../openzfs/module/os/linux/zfs/zpl_xattr.c | 25 +++--- .../openzfs/module/zcommon/zfeature_common.c | 6 ++ sys/contrib/openzfs/module/zfs/arc.c | 8 +- sys/contrib/openzfs/module/zfs/dbuf.c | 3 +- sys/contrib/openzfs/module/zfs/mmp.c | 2 +- sys/contrib/openzfs/module/zfs/spa.c | 16 +++- sys/contrib/openzfs/module/zfs/spa_errlog.c | 12 +-- sys/contrib/openzfs/module/zfs/spa_misc.c | 29 ++++++- sys/contrib/openzfs/module/zfs/vdev.c | 50 +++++++++-- sys/contrib/openzfs/module/zfs/vdev_indirect.c | 7 +- sys/contrib/openzfs/module/zfs/vdev_label.c | 6 ++ sys/contrib/openzfs/module/zfs/vdev_queue.c | 3 + sys/contrib/openzfs/module/zfs/zfs_fm.c | 10 +-- sys/contrib/openzfs/module/zfs/zfs_replay.c | 14 +-- sys/contrib/openzfs/module/zfs/zfs_vnops.c | 4 +- sys/contrib/openzfs/module/zfs/zil.c | 9 +- sys/contrib/openzfs/tests/runfiles/common.run | 19 +++-- sys/contrib/openzfs/tests/runfiles/linux.run | 6 +- .../tests/test-runner/bin/test-runner.py.in | 34 ++++---- .../openzfs/tests/zfs-tests/tests/Makefile.am | 4 + .../functional/cli_root/zdb/zdb_encrypted.ksh | 4 +- .../functional/cli_root/zpool_get/vdev_get.cfg | 73 ++++++++++++++++ .../cli_root/zpool_get/vdev_get_001_pos.ksh | 62 ++++++++++++++ .../functional/cli_root/zpool_get/zpool_get.cfg | 1 + .../cli_root/zpool_set/vdev_set_001_pos.ksh | 52 ++++++++++++ .../tests/functional/rsend/send-c_volume.ksh | 9 +- .../tests/functional/vdev_zaps/vdev_zaps.kshlib | 13 +++ .../functional/vdev_zaps/vdev_zaps_001_pos.ksh | 3 +- .../functional/vdev_zaps/vdev_zaps_002_pos.ksh | 1 + .../functional/vdev_zaps/vdev_zaps_003_pos.ksh | 1 + .../functional/vdev_zaps/vdev_zaps_004_pos.ksh | 1 + .../functional/vdev_zaps/vdev_zaps_005_pos.ksh | 1 + .../functional/vdev_zaps/vdev_zaps_006_pos.ksh | 1 + .../functional/vdev_zaps/vdev_zaps_007_pos.ksh | 1 + .../functional/zvol/zvol_misc/zvol_misc_trim.ksh | 4 +- sys/modules/zfs/zfs_config.h | 64 ++++++++++++-- sys/modules/zfs/zfs_gitrev.h | 2 +- 130 files changed, 2034 insertions(+), 640 deletions(-) diff --cc lib/libbe/be.c index a5e053cb82df,000000000000..a27025684305 mode 100644,000000..100644 --- a/lib/libbe/be.c +++ b/lib/libbe/be.c @@@ -1,1309 -1,0 +1,1309 @@@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * Copyright (c) 2017 Kyle J. Kneitinger + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "be.h" +#include "be_impl.h" + +struct promote_entry { + char name[BE_MAXPATHLEN]; + SLIST_ENTRY(promote_entry) link; +}; + +struct be_destroy_data { + libbe_handle_t *lbh; + char target_name[BE_MAXPATHLEN]; + char *snapname; + SLIST_HEAD(, promote_entry) promotelist; +}; + +#if SOON +static int be_create_child_noent(libbe_handle_t *lbh, const char *active, + const char *child_path); +static int be_create_child_cloned(libbe_handle_t *lbh, const char *active); +#endif + +/* Arbitrary... should tune */ +#define BE_SNAP_SERIAL_MAX 1024 + +/* + * Iterator function for locating the rootfs amongst the children of the + * zfs_be_root set by loader(8). data is expected to be a libbe_handle_t *. + */ +static int +be_locate_rootfs(libbe_handle_t *lbh) +{ + struct statfs sfs; + struct mnttab entry; + zfs_handle_t *zfs; + + /* + * Check first if root is ZFS; if not, we'll bail on rootfs capture. + * Unfortunately needed because zfs_path_to_zhandle will emit to + * stderr if / isn't actually a ZFS filesystem, which we'd like + * to avoid. + */ + if (statfs("/", &sfs) == 0) { + statfs2mnttab(&sfs, &entry); + if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) + return (1); + } else + return (1); + zfs = zfs_path_to_zhandle(lbh->lzh, "/", ZFS_TYPE_FILESYSTEM); + if (zfs == NULL) + return (1); + + strlcpy(lbh->rootfs, zfs_get_name(zfs), sizeof(lbh->rootfs)); + zfs_close(zfs); + return (0); +} + +/* + * Initializes the libbe context to operate in the root boot environment + * dataset, for example, zroot/ROOT. + */ +libbe_handle_t * +libbe_init(const char *root) +{ + char altroot[MAXPATHLEN]; + libbe_handle_t *lbh; + char *poolname, *pos; + int pnamelen; + + lbh = NULL; + poolname = pos = NULL; + + if ((lbh = calloc(1, sizeof(libbe_handle_t))) == NULL) + goto err; + + if ((lbh->lzh = libzfs_init()) == NULL) + goto err; + + /* + * Grab rootfs, we'll work backwards from there if an optional BE root + * has not been passed in. + */ + if (be_locate_rootfs(lbh) != 0) { + if (root == NULL) + goto err; + *lbh->rootfs = '\0'; + } + if (root == NULL) { + /* Strip off the final slash from rootfs to get the be root */ + strlcpy(lbh->root, lbh->rootfs, sizeof(lbh->root)); + pos = strrchr(lbh->root, '/'); + if (pos == NULL) + goto err; + *pos = '\0'; + } else + strlcpy(lbh->root, root, sizeof(lbh->root)); + + if ((pos = strchr(lbh->root, '/')) == NULL) + goto err; + + pnamelen = pos - lbh->root; + poolname = malloc(pnamelen + 1); + if (poolname == NULL) + goto err; + + strlcpy(poolname, lbh->root, pnamelen + 1); + if ((lbh->active_phandle = zpool_open(lbh->lzh, poolname)) == NULL) + goto err; + free(poolname); + poolname = NULL; + + if (zpool_get_prop(lbh->active_phandle, ZPOOL_PROP_BOOTFS, lbh->bootfs, + sizeof(lbh->bootfs), NULL, true) != 0) + goto err; + + if (zpool_get_prop(lbh->active_phandle, ZPOOL_PROP_ALTROOT, + altroot, sizeof(altroot), NULL, true) == 0 && + strcmp(altroot, "-") != 0) + lbh->altroot_len = strlen(altroot); + + return (lbh); +err: + if (lbh != NULL) { + if (lbh->active_phandle != NULL) + zpool_close(lbh->active_phandle); + if (lbh->lzh != NULL) + libzfs_fini(lbh->lzh); + free(lbh); + } + free(poolname); + return (NULL); +} + + +/* + * Free memory allocated by libbe_init() + */ +void +libbe_close(libbe_handle_t *lbh) +{ + + if (lbh->active_phandle != NULL) + zpool_close(lbh->active_phandle); + libzfs_fini(lbh->lzh); + free(lbh); +} + +/* + * Proxy through to libzfs for the moment. + */ +void +be_nicenum(uint64_t num, char *buf, size_t buflen) +{ + + zfs_nicenum(num, buf, buflen); +} + +static bool +be_should_promote_clones(zfs_handle_t *zfs_hdl, struct be_destroy_data *bdd) +{ + char *atpos; + + if (zfs_get_type(zfs_hdl) != ZFS_TYPE_SNAPSHOT) + return (false); + + /* + * If we're deleting a snapshot, we need to make sure we only promote + * clones that are derived from one of the snapshots we're deleting, + * rather than that of a snapshot we're not touching. This keeps stuff + * in a consistent state, making sure that we don't error out unless + * we really need to. + */ + if (bdd->snapname == NULL) + return (true); + + atpos = strchr(zfs_get_name(zfs_hdl), '@'); + return (strcmp(atpos + 1, bdd->snapname) == 0); +} + +/* + * This is executed from be_promote_dependent_clones via zfs_iter_dependents, + * It checks if the dependent type is a snapshot then attempts to find any + * clones associated with it. Any clones not related to the destroy target are + * added to the promote list. + */ +static int +be_dependent_clone_cb(zfs_handle_t *zfs_hdl, void *data) +{ + int err; + bool found; + const char *name; + struct nvlist *nvl; + struct nvpair *nvp; + struct be_destroy_data *bdd; + struct promote_entry *entry, *newentry; + + nvp = NULL; + err = 0; + bdd = (struct be_destroy_data *)data; + + if (be_should_promote_clones(zfs_hdl, bdd) && + (nvl = zfs_get_clones_nvl(zfs_hdl)) != NULL) { + while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) { + name = nvpair_name(nvp); + + /* + * Skip if the clone is equal to, or a child of, the + * destroy target. + */ + if (strncmp(name, bdd->target_name, + strlen(bdd->target_name)) == 0 || + strstr(name, bdd->target_name) == name) { + continue; + } + + found = false; + SLIST_FOREACH(entry, &bdd->promotelist, link) { + if (strcmp(entry->name, name) == 0) { + found = true; + break; + } + } + + if (found) + continue; + + newentry = malloc(sizeof(struct promote_entry)); + if (newentry == NULL) { + err = ENOMEM; + break; + } + +#define BE_COPY_NAME(entry, src) \ + strlcpy((entry)->name, (src), sizeof((entry)->name)) + if (BE_COPY_NAME(newentry, name) >= + sizeof(newentry->name)) { + /* Shouldn't happen. */ + free(newentry); + err = ENAMETOOLONG; + break; + } +#undef BE_COPY_NAME + + /* + * We're building up a SLIST here to make sure both that + * we get the order right and so that we don't + * inadvertently observe the wrong state by promoting + * datasets while we're still walking the tree. The + * latter can lead to situations where we promote a BE + * then effectively demote it again. + */ + SLIST_INSERT_HEAD(&bdd->promotelist, newentry, link); + } + nvlist_free(nvl); + } + zfs_close(zfs_hdl); + return (err); +} + +/* + * This is called before a destroy, so that any datasets(environments) that are + * dependent on this one get promoted before destroying the target. + */ +static int +be_promote_dependent_clones(zfs_handle_t *zfs_hdl, struct be_destroy_data *bdd) +{ + int err; + zfs_handle_t *clone; + struct promote_entry *entry; + + snprintf(bdd->target_name, BE_MAXPATHLEN, "%s/", zfs_get_name(zfs_hdl)); - err = zfs_iter_dependents(zfs_hdl, 0, true, be_dependent_clone_cb, bdd); ++ err = zfs_iter_dependents(zfs_hdl, true, be_dependent_clone_cb, bdd); + + /* + * Drain the list and walk away from it if we're only deleting a + * snapshot. + */ + if (bdd->snapname != NULL && !SLIST_EMPTY(&bdd->promotelist)) + err = BE_ERR_HASCLONES; + while (!SLIST_EMPTY(&bdd->promotelist)) { + entry = SLIST_FIRST(&bdd->promotelist); + SLIST_REMOVE_HEAD(&bdd->promotelist, link); + +#define ZFS_GRAB_CLONE() \ + zfs_open(bdd->lbh->lzh, entry->name, ZFS_TYPE_FILESYSTEM) + /* + * Just skip this part on error, we still want to clean up the + * promotion list after the first error. We'll then preserve it + * all the way back. + */ + if (err == 0 && (clone = ZFS_GRAB_CLONE()) != NULL) { + err = zfs_promote(clone); + if (err != 0) + err = BE_ERR_DESTROYMNT; + zfs_close(clone); + } +#undef ZFS_GRAB_CLONE + free(entry); + } + + return (err); +} + +static int +be_destroy_cb(zfs_handle_t *zfs_hdl, void *data) +{ + char path[BE_MAXPATHLEN]; + struct be_destroy_data *bdd; + zfs_handle_t *snap; + int err; + + bdd = (struct be_destroy_data *)data; + if (bdd->snapname == NULL) { - err = zfs_iter_children(zfs_hdl, 0, be_destroy_cb, data); ++ err = zfs_iter_children(zfs_hdl, be_destroy_cb, data); + if (err != 0) + return (err); + return (zfs_destroy(zfs_hdl, false)); + } + /* If we're dealing with snapshots instead, delete that one alone */ - err = zfs_iter_filesystems(zfs_hdl, 0, be_destroy_cb, data); ++ err = zfs_iter_filesystems(zfs_hdl, be_destroy_cb, data); + if (err != 0) + return (err); + /* + * This part is intentionally glossing over any potential errors, + * because there's a lot less potential for errors when we're cleaning + * up snapshots rather than a full deep BE. The primary error case + * here being if the snapshot doesn't exist in the first place, which + * the caller will likely deem insignificant as long as it doesn't + * exist after the call. Thus, such a missing snapshot shouldn't jam + * up the destruction. + */ + snprintf(path, sizeof(path), "%s@%s", zfs_get_name(zfs_hdl), + bdd->snapname); + if (!zfs_dataset_exists(bdd->lbh->lzh, path, ZFS_TYPE_SNAPSHOT)) + return (0); + snap = zfs_open(bdd->lbh->lzh, path, ZFS_TYPE_SNAPSHOT); + if (snap != NULL) + zfs_destroy(snap, false); + return (0); +} + +#define BE_DESTROY_WANTORIGIN (BE_DESTROY_ORIGIN | BE_DESTROY_AUTOORIGIN) +/* + * Destroy the boot environment or snapshot specified by the name + * parameter. Options are or'd together with the possible values: + * BE_DESTROY_FORCE : forces operation on mounted datasets + * BE_DESTROY_ORIGIN: destroy the origin snapshot as well + */ +static int +be_destroy_internal(libbe_handle_t *lbh, const char *name, int options, + bool odestroyer) +{ + struct be_destroy_data bdd; + char origin[BE_MAXPATHLEN], path[BE_MAXPATHLEN]; + zfs_handle_t *fs; + char *snapdelim; + int err, force, mounted; + size_t rootlen; + + bdd.lbh = lbh; + bdd.snapname = NULL; + SLIST_INIT(&bdd.promotelist); + force = options & BE_DESTROY_FORCE; + *origin = '\0'; + + be_root_concat(lbh, name, path); + + if ((snapdelim = strchr(path, '@')) == NULL) { + if (!zfs_dataset_exists(lbh->lzh, path, ZFS_TYPE_FILESYSTEM)) + return (set_error(lbh, BE_ERR_NOENT)); + + if (strcmp(path, lbh->rootfs) == 0 || + strcmp(path, lbh->bootfs) == 0) + return (set_error(lbh, BE_ERR_DESTROYACT)); + + fs = zfs_open(lbh->lzh, path, ZFS_TYPE_FILESYSTEM); + if (fs == NULL) + return (set_error(lbh, BE_ERR_ZFSOPEN)); + + /* Don't destroy a mounted dataset unless force is specified */ + if ((mounted = zfs_is_mounted(fs, NULL)) != 0) { + if (force) { + zfs_unmount(fs, NULL, 0); + } else { + free(bdd.snapname); + return (set_error(lbh, BE_ERR_DESTROYMNT)); + } + } + } else { + /* + * If we're initially destroying a snapshot, origin options do + * not make sense. If we're destroying the origin snapshot of + * a BE, we want to maintain the options in case we need to + * fake success after failing to promote. + */ + if (!odestroyer) + options &= ~BE_DESTROY_WANTORIGIN; + if (!zfs_dataset_exists(lbh->lzh, path, ZFS_TYPE_SNAPSHOT)) + return (set_error(lbh, BE_ERR_NOENT)); + + bdd.snapname = strdup(snapdelim + 1); + if (bdd.snapname == NULL) + return (set_error(lbh, BE_ERR_NOMEM)); + *snapdelim = '\0'; + fs = zfs_open(lbh->lzh, path, ZFS_TYPE_DATASET); + if (fs == NULL) { + free(bdd.snapname); + return (set_error(lbh, BE_ERR_ZFSOPEN)); + } + } + + /* + * Whether we're destroying a BE or a single snapshot, we need to walk + * the tree of what we're going to destroy and promote everything in our + * path so that we can make it happen. + */ + if ((err = be_promote_dependent_clones(fs, &bdd)) != 0) { + free(bdd.snapname); + + /* + * If we're just destroying the origin of some other dataset + * we were invoked to destroy, then we just ignore + * BE_ERR_HASCLONES and return success unless the caller wanted + * to force the issue. + */ + if (odestroyer && err == BE_ERR_HASCLONES && + (options & BE_DESTROY_AUTOORIGIN) != 0) + return (0); + return (set_error(lbh, err)); + } + + /* + * This was deferred until after we promote all of the derivatives so + * that we grab the new origin after everything's settled down. + */ + if ((options & BE_DESTROY_WANTORIGIN) != 0 && + zfs_prop_get(fs, ZFS_PROP_ORIGIN, origin, sizeof(origin), + NULL, NULL, 0, 1) != 0 && + (options & BE_DESTROY_ORIGIN) != 0) + return (set_error(lbh, BE_ERR_NOORIGIN)); + + /* + * If the caller wants auto-origin destruction and the origin + * name matches one of our automatically created snapshot names + * (i.e. strftime("%F-%T") with a serial at the end), then + * we'll set the DESTROY_ORIGIN flag and nuke it + * be_is_auto_snapshot_name is exported from libbe(3) so that + * the caller can determine if it needs to warn about the origin + * not being destroyed or not. + */ + if ((options & BE_DESTROY_AUTOORIGIN) != 0 && *origin != '\0' && + be_is_auto_snapshot_name(lbh, origin)) + options |= BE_DESTROY_ORIGIN; + + err = be_destroy_cb(fs, &bdd); + zfs_close(fs); + free(bdd.snapname); + if (err != 0) { + /* Children are still present or the mount is referenced */ + if (err == EBUSY) + return (set_error(lbh, BE_ERR_DESTROYMNT)); + return (set_error(lbh, BE_ERR_UNKNOWN)); + } + + if ((options & BE_DESTROY_ORIGIN) == 0) + return (0); + + /* The origin can't possibly be shorter than the BE root */ + rootlen = strlen(lbh->root); + if (*origin == '\0' || strlen(origin) <= rootlen + 1) + return (set_error(lbh, BE_ERR_INVORIGIN)); + + /* + * We'll be chopping off the BE root and running this back through + * be_destroy, so that we properly handle the origin snapshot whether + * it be that of a deep BE or not. + */ + if (strncmp(origin, lbh->root, rootlen) != 0 || origin[rootlen] != '/') + return (0); + + return (be_destroy_internal(lbh, origin + rootlen + 1, + options & ~BE_DESTROY_ORIGIN, true)); +} + +int +be_destroy(libbe_handle_t *lbh, const char *name, int options) +{ + + /* + * The consumer must not set both BE_DESTROY_AUTOORIGIN and + * BE_DESTROY_ORIGIN. Internally, we'll set the latter from the former. + * The latter should imply that we must succeed at destroying the + * origin, or complain otherwise. + */ + if ((options & BE_DESTROY_WANTORIGIN) == BE_DESTROY_WANTORIGIN) + return (set_error(lbh, BE_ERR_UNKNOWN)); + return (be_destroy_internal(lbh, name, options, false)); +} + +static void +be_setup_snapshot_name(libbe_handle_t *lbh, char *buf, size_t buflen) +{ + time_t rawtime; + int len, serial; + + time(&rawtime); + len = strlen(buf); + len += strftime(buf + len, buflen - len, "@%F-%T", localtime(&rawtime)); + /* No room for serial... caller will do its best */ + if (buflen - len < 2) + return; + + for (serial = 0; serial < BE_SNAP_SERIAL_MAX; ++serial) { + snprintf(buf + len, buflen - len, "-%d", serial); + if (!zfs_dataset_exists(lbh->lzh, buf, ZFS_TYPE_SNAPSHOT)) + return; + } +} + +bool +be_is_auto_snapshot_name(libbe_handle_t *lbh __unused, const char *name) +{ + const char *snap; + int day, hour, minute, month, second, serial, year; + + if ((snap = strchr(name, '@')) == NULL) + return (false); + ++snap; + /* We'll grab the individual components and do some light validation. */ + if (sscanf(snap, "%d-%d-%d-%d:%d:%d-%d", &year, &month, &day, &hour, + &minute, &second, &serial) != 7) + return (false); + return (year >= 1970) && (month >= 1 && month <= 12) && + (day >= 1 && day <= 31) && (hour >= 0 && hour <= 23) && + (minute >= 0 && minute <= 59) && (second >= 0 && second <= 60) && + serial >= 0; +} + +int +be_snapshot(libbe_handle_t *lbh, const char *source, const char *snap_name, + bool recursive, char *result) +{ + char buf[BE_MAXPATHLEN]; + int err; + + be_root_concat(lbh, source, buf); + + if ((err = be_exists(lbh, buf)) != 0) + return (set_error(lbh, err)); + + if (snap_name != NULL) { + if (strlcat(buf, "@", sizeof(buf)) >= sizeof(buf)) + return (set_error(lbh, BE_ERR_INVALIDNAME)); + + if (strlcat(buf, snap_name, sizeof(buf)) >= sizeof(buf)) + return (set_error(lbh, BE_ERR_INVALIDNAME)); + + if (result != NULL) + snprintf(result, BE_MAXPATHLEN, "%s@%s", source, + snap_name); + } else { + be_setup_snapshot_name(lbh, buf, sizeof(buf)); + + if (result != NULL && strlcpy(result, strrchr(buf, '/') + 1, + sizeof(buf)) >= sizeof(buf)) + return (set_error(lbh, BE_ERR_INVALIDNAME)); + } + if ((err = zfs_snapshot(lbh->lzh, buf, recursive, NULL)) != 0) { + switch (err) { + case EZFS_INVALIDNAME: + return (set_error(lbh, BE_ERR_INVALIDNAME)); + + default: + /* + * The other errors that zfs_ioc_snapshot might return + * shouldn't happen if we've set things up properly, so + * we'll gloss over them and call it UNKNOWN as it will + * require further triage. + */ + if (errno == ENOTSUP) + return (set_error(lbh, BE_ERR_NOPOOL)); + return (set_error(lbh, BE_ERR_UNKNOWN)); + } + } + + return (BE_ERR_SUCCESS); +} + + +/* + * Create the boot environment specified by the name parameter + */ +int +be_create(libbe_handle_t *lbh, const char *name) +{ + int err; + + err = be_create_from_existing(lbh, name, be_active_path(lbh)); + + return (set_error(lbh, err)); +} + +static int +be_deep_clone_prop(int prop, void *cb) +{ + int err; + struct libbe_dccb *dccb; + zprop_source_t src; + char pval[BE_MAXPATHLEN]; + char source[BE_MAXPATHLEN]; + char *val; + + dccb = cb; + /* Skip some properties we don't want to touch */ + if (prop == ZFS_PROP_CANMOUNT) + return (ZPROP_CONT); + + /* Don't copy readonly properties */ + if (zfs_prop_readonly(prop)) + return (ZPROP_CONT); + + if ((err = zfs_prop_get(dccb->zhp, prop, (char *)&pval, + sizeof(pval), &src, (char *)&source, sizeof(source), false))) + /* Just continue if we fail to read a property */ + return (ZPROP_CONT); + + /* + * Only copy locally defined or received properties. This continues + * to avoid temporary/default/local properties intentionally without + * breaking received datasets. + */ + if (src != ZPROP_SRC_LOCAL && src != ZPROP_SRC_RECEIVED) + return (ZPROP_CONT); + + /* Augment mountpoint with altroot, if needed */ + val = pval; + if (prop == ZFS_PROP_MOUNTPOINT) + val = be_mountpoint_augmented(dccb->lbh, val); + + nvlist_add_string(dccb->props, zfs_prop_to_name(prop), val); + + return (ZPROP_CONT); +} + +/* + * Return the corresponding boot environment path for a given + * dataset path, the constructed path is placed in 'result'. + * + * example: say our new boot environment name is 'bootenv' and + * the dataset path is 'zroot/ROOT/default/data/set'. + * + * result should produce: 'zroot/ROOT/bootenv/data/set' + */ +static int +be_get_path(struct libbe_deep_clone *ldc, const char *dspath, char *result, int result_size) +{ + char *pos; + char *child_dataset; + + /* match the root path for the boot environments */ + pos = strstr(dspath, ldc->lbh->root); + + /* no match, different pools? */ + if (pos == NULL) + return (BE_ERR_BADPATH); + + /* root path of the new boot environment */ + snprintf(result, result_size, "%s/%s", ldc->lbh->root, ldc->bename); + + /* gets us to the parent dataset, the +1 consumes a trailing slash */ + pos += strlen(ldc->lbh->root) + 1; + + /* skip the parent dataset */ + if ((child_dataset = strchr(pos, '/')) != NULL) + strlcat(result, child_dataset, result_size); + + return (BE_ERR_SUCCESS); +} + +static int +be_clone_cb(zfs_handle_t *ds, void *data) +{ + int err; + char be_path[BE_MAXPATHLEN]; + char snap_path[BE_MAXPATHLEN]; + const char *dspath; + zfs_handle_t *snap_hdl; + nvlist_t *props; + struct libbe_deep_clone *ldc; + struct libbe_dccb dccb; + + ldc = (struct libbe_deep_clone *)data; + dspath = zfs_get_name(ds); + + snprintf(snap_path, sizeof(snap_path), "%s@%s", dspath, ldc->snapname); + + /* construct the boot environment path from the dataset we're cloning */ + if (be_get_path(ldc, dspath, be_path, sizeof(be_path)) != BE_ERR_SUCCESS) + return (BE_ERR_UNKNOWN); + + /* the dataset to be created (i.e. the boot environment) already exists */ + if (zfs_dataset_exists(ldc->lbh->lzh, be_path, ZFS_TYPE_DATASET)) + return (BE_ERR_EXISTS); + + /* no snapshot found for this dataset, silently skip it */ + if (!zfs_dataset_exists(ldc->lbh->lzh, snap_path, ZFS_TYPE_SNAPSHOT)) + return (0); + + if ((snap_hdl = + zfs_open(ldc->lbh->lzh, snap_path, ZFS_TYPE_SNAPSHOT)) == NULL) + return (BE_ERR_ZFSOPEN); + + nvlist_alloc(&props, NV_UNIQUE_NAME, KM_SLEEP); + nvlist_add_string(props, "canmount", "noauto"); + + dccb.lbh = ldc->lbh; + dccb.zhp = ds; + dccb.props = props; + if (zprop_iter(be_deep_clone_prop, &dccb, B_FALSE, B_FALSE, + ZFS_TYPE_FILESYSTEM) == ZPROP_INVAL) + return (-1); + + if ((err = zfs_clone(snap_hdl, be_path, props)) != 0) + return (BE_ERR_ZFSCLONE); + + nvlist_free(props); + zfs_close(snap_hdl); + + if (ldc->depth_limit == -1 || ldc->depth < ldc->depth_limit) { + ldc->depth++; - err = zfs_iter_filesystems(ds, 0, be_clone_cb, ldc); ++ err = zfs_iter_filesystems(ds, be_clone_cb, ldc); + ldc->depth--; + } + + return (err); +} + +/* + * Create a boot environment with a given name from a given snapshot. + * Snapshots can be in the format 'zroot/ROOT/default@snapshot' or + * 'default@snapshot'. In the latter case, 'default@snapshot' will be prepended + * with the root path that libbe was initailized with. +*/ +static int +be_clone(libbe_handle_t *lbh, const char *bename, const char *snapshot, int depth) +{ + int err; + char snap_path[BE_MAXPATHLEN]; + char *parentname, *snapname; *** 2351 LINES SKIPPED ***