diff options
Diffstat (limited to 'sys/contrib/openzfs')
40 files changed, 969 insertions, 376 deletions
diff --git a/sys/contrib/openzfs/.github/workflows/scripts/qemu-2-start.sh b/sys/contrib/openzfs/.github/workflows/scripts/qemu-2-start.sh index 1c608348ffcd..422b3e9df388 100755 --- a/sys/contrib/openzfs/.github/workflows/scripts/qemu-2-start.sh +++ b/sys/contrib/openzfs/.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" @@ -287,7 +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"' diff --git a/sys/contrib/openzfs/.github/workflows/zfs-qemu.yml b/sys/contrib/openzfs/.github/workflows/zfs-qemu.yml index 69349678d84c..3b164548f9be 100644 --- a/sys/contrib/openzfs/.github/workflows/zfs-qemu.yml +++ b/sys/contrib/openzfs/.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: diff --git a/sys/contrib/openzfs/.mailmap b/sys/contrib/openzfs/.mailmap index e6f09c6c9d43..3397fbc3745d 100644 --- a/sys/contrib/openzfs/.mailmap +++ b/sys/contrib/openzfs/.mailmap @@ -53,6 +53,7 @@ Jason Harmening <jason.harmening@gmail.com> Jeremy Faulkner <gldisater@gmail.com> Jinshan Xiong <jinshan.xiong@gmail.com> John Poduska <jpoduska@datto.com> +Jo Zzsi <jozzsicsataban@gmail.com> Justin Scholz <git@justinscholz.de> Ka Ho Ng <khng300@gmail.com> Kash Pande <github@tripleback.net> @@ -67,6 +68,7 @@ Michael Gmelin <grembo@FreeBSD.org> Olivier Mazouffre <olivier.mazouffre@ims-bordeaux.fr> Piotr Kubaj <pkubaj@anongoth.pl> Quentin Zdanis <zdanisq@gmail.com> +Roberto Ricci <io@r-ricci.it> Roberto Ricci <ricci@disroot.org> Rob Norris <robn@despairlabs.com> Rob Norris <rob.norris@klarasystems.com> @@ -83,7 +85,10 @@ Youzhong Yang <youzhong@gmail.com> # Signed-off-by: overriding Author: Alexander Ziaee <ziaee@FreeBSD.org> <concussious@runbox.com> Felix Schmidt <felixschmidt20@aol.com> <f.sch.prototype@gmail.com> +Jean-Sébastien Pédron <dumbbell@FreeBSD.org> <jean-sebastien.pedron@dumbbell.fr> +Konstantin Belousov <kib@FreeBSD.org> <kib@kib.kiev.ua> Olivier Certner <olce@FreeBSD.org> <olce.freebsd@certner.fr> +Patrick Xia <patrickx@google.com> <octalc0de@aim.com> Phil Sutter <phil@nwl.cc> <p.github@nwl.cc> poscat <poscat@poscat.moe> <poscat0x04@outlook.com> Qiuhao Chen <chenqiuhao1997@gmail.com> <haohao0924@126.com> @@ -125,6 +130,7 @@ buzzingwires <buzzingwires@outlook.com> <131118055+buzzingwires@users.noreply.gi Cedric Maunoury <cedric.maunoury@gmail.com> <38213715+cedricmaunoury@users.noreply.github.com> Charles Suh <charles.suh@gmail.com> <charlessuh@users.noreply.github.com> Chris Peredun <chris.peredun@ixsystems.com> <126915832+chrisperedun@users.noreply.github.com> +classabbyamp <dev@placeviolette.net> <5366828+classabbyamp@users.noreply.github.com> Dacian Reece-Stremtan <dacianstremtan@gmail.com> <35844628+dacianstremtan@users.noreply.github.com> Damian Szuberski <szuberskidamian@gmail.com> <30863496+szubersk@users.noreply.github.com> Daniel Hiepler <d-git@coderdu.de> <32984777+heeplr@users.noreply.github.com> @@ -185,6 +191,7 @@ Michael Niewöhner <foss@mniewoehner.de> <c0d3z3r0@users.noreply.github.com> Michael Zhivich <mzhivich@akamai.com> <33133421+mzhivich@users.noreply.github.com> MigeljanImeri <ImeriMigel@gmail.com> <78048439+MigeljanImeri@users.noreply.github.com> Mo Zhou <cdluminate@gmail.com> <5723047+cdluminate@users.noreply.github.com> +nav1s <nav1s@proton.me> <42621369+nav1s@users.noreply.github.com> Nick Mattis <nickm970@gmail.com> <nmattis@users.noreply.github.com> omni <omni+vagant@hack.org> <79493359+omnivagant@users.noreply.github.com> Pablo Correa Gómez <ablocorrea@hotmail.com> <32678034+pablofsf@users.noreply.github.com> @@ -206,6 +213,7 @@ Samuel Wycliffe <samuelwycliffe@gmail.com> <50765275+npc203@users.noreply.github Savyasachee Jha <hi@savyasacheejha.com> <savyajha@users.noreply.github.com> Scott Colby <scott@scolby.com> <scolby33@users.noreply.github.com> Sean Eric Fagan <kithrup@mac.com> <kithrup@users.noreply.github.com> +Shreshth Srivastava <shreshthsrivastava2@gmail.com> <66148173+Shreshth3@users.noreply.github.com> Spencer Kinny <spencerkinny1995@gmail.com> <30333052+Spencer-Kinny@users.noreply.github.com> Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com> <75025422+nssrikanth@users.noreply.github.com> Stefan Lendl <s.lendl@proxmox.com> <1321542+stfl@users.noreply.github.com> diff --git a/sys/contrib/openzfs/AUTHORS b/sys/contrib/openzfs/AUTHORS index 6c34c07f39ef..e496c0e8a807 100644 --- a/sys/contrib/openzfs/AUTHORS +++ b/sys/contrib/openzfs/AUTHORS @@ -154,6 +154,7 @@ CONTRIBUTORS: Chris Zubrzycki <github@mid-earth.net> Chuck Tuffli <ctuffli@gmail.com> Chunwei Chen <david.chen@nutanix.com> + classabbyamp <dev@placeviolette.net> Clemens Fruhwirth <clemens@endorphin.org> Clemens Lang <cl@clang.name> Clint Armstrong <clint@clintarmstrong.net> @@ -161,6 +162,7 @@ CONTRIBUTORS: Colin Ian King <colin.king@canonical.com> Colin Percival <cperciva@tarsnap.com> Colm Buckley <colm@tuatha.org> + Cong Zhang <congzhangzh@users.noreply.github.com> Crag Wang <crag0715@gmail.com> Craig Loomis <cloomis@astro.princeton.edu> Craig Sanders <github@taz.net.au> @@ -217,6 +219,7 @@ CONTRIBUTORS: Eitan Adler <lists@eitanadler.com> Eli Rosenthal <eli.rosenthal@delphix.com> Eli Schwartz <eschwartz93@gmail.com> + Eric A. Borisch <eborisch@gmail.com> Eric Desrochers <eric.desrochers@canonical.com> Eric Dillmann <eric@jave.fr> Eric Schrock <Eric.Schrock@delphix.com> @@ -288,6 +291,7 @@ CONTRIBUTORS: Henrik Riomar <henrik.riomar@gmail.com> Herb Wartens <wartens2@llnl.gov> Hiếu Lê <leorize+oss@disroot.org> + hoshinomori <hoshinomorimorimo@gmail.com> Huang Liu <liu.huang@zte.com.cn> Håkan Johansson <f96hajo@chalmers.se> Igor K <igor@dilos.org> @@ -300,6 +304,7 @@ CONTRIBUTORS: ilovezfs <ilovezfs@icloud.com> InsanePrawn <Insane.Prawny@gmail.com> Isaac Huang <he.huang@intel.com> + Ivan Shapovalov <intelfx@intelfx.name> Ivan Volosyuk <Ivan.Volosyuk@gmail.com> Jacek Fefliński <feflik@gmail.com> Jacob Adams <tookmund@gmail.com> @@ -322,6 +327,7 @@ CONTRIBUTORS: Javen Wu <wu.javen@gmail.com> Jaydeep Kshirsagar <jkshirsagar@maxlinear.com> Jean-Baptiste Lallement <jean-baptiste@ubuntu.com> + Jean-Sébastien Pédron <dumbbell@FreeBSD.org> Jeff Dike <jdike@akamai.com> Jeremy Faulkner <gldisater@gmail.com> Jeremy Gill <jgill@parallax-innovations.com> @@ -355,7 +361,9 @@ CONTRIBUTORS: Josh Soref <jsoref@users.noreply.github.com> Joshua M. Clulow <josh@sysmgr.org> José Luis Salvador Rufo <salvador.joseluis@gmail.com> + Jo Zzsi <jozzsicsataban@gmail.com> João Carlos Mendes Luís <jonny@jonny.eng.br> + JT Pennington <jt.pennington@klarasystems.com> Julian Brunner <julian.brunner@gmail.com> Julian Heuking <JulianH@beckhoff.com> jumbi77 <jumbi77@users.noreply.github.com> @@ -388,6 +396,7 @@ CONTRIBUTORS: Kleber Tarcísio <klebertarcisio@yahoo.com.br> Kody A Kantor <kody.kantor@gmail.com> Kohsuke Kawaguchi <kk@kohsuke.org> + Konstantin Belousov <kib@FreeBSD.org> Konstantin Khorenko <khorenko@virtuozzo.com> KORN Andras <korn@elan.rulez.org> kotauskas <v.toncharov@gmail.com> @@ -416,6 +425,7 @@ CONTRIBUTORS: luozhengzheng <luo.zhengzheng@zte.com.cn> Luís Henriques <henrix@camandro.org> Madhav Suresh <madhav.suresh@delphix.com> + Maksym Shkolnyi <maksym.shkolnyi@workato.com> manfromafar <jonsonb10@gmail.com> Manoj Joseph <manoj.joseph@delphix.com> Manuel Amador (Rudd-O) <rudd-o@rudd-o.com> @@ -482,6 +492,7 @@ CONTRIBUTORS: Nathaniel Clark <Nathaniel.Clark@misrule.us> Nathaniel Wesley Filardo <nwf@cs.jhu.edu> Nathan Lewis <linux.robotdude@gmail.com> + nav1s <nav1s@proton.me> Nav Ravindranath <nav@delphix.com> Neal Gompa (ニール・ゴンパ) <ngompa13@gmail.com> Ned Bass <bass6@llnl.gov> @@ -506,6 +517,7 @@ CONTRIBUTORS: Palash Gandhi <pbg4930@rit.edu> Patrick Fasano <patrick@patrickfasano.com> Patrick Mooney <pmooney@pfmooney.com> + Patrick Xia <patrickx@google.com> Patrik Greco <sikevux@sikevux.se> Paul B. Henson <henson@acm.org> Paul Dagnelie <pcd@delphix.com> @@ -605,6 +617,7 @@ CONTRIBUTORS: Shengqi Chen <harry-chen@outlook.com> SHENGYI HONG <aokblast@FreeBSD.org> Shen Yan <shenyanxxxy@qq.com> + Shreshth Srivastava <shreshthsrivastava2@gmail.com> Sietse <sietse@wizdom.nu> Simon Guest <simon.guest@tesujimath.org> Simon Howard <fraggle@soulsphere.org> @@ -665,6 +678,7 @@ CONTRIBUTORS: Toyam Cox <aviator45003@gmail.com> Trevor Bautista <trevrb@trevrb.net> Trey Dockendorf <treydock@gmail.com> + trick2011 <trick2011@users.noreply.github.com> Troels Nørgaard <tnn@tradeshift.com> tstabrawa <tstabrawa@users.noreply.github.com> Tulsi Jain <tulsi.jain@delphix.com> diff --git a/sys/contrib/openzfs/cmd/zdb/zdb.c b/sys/contrib/openzfs/cmd/zdb/zdb.c index 70a4ed46f263..2560ad045db3 100644 --- a/sys/contrib/openzfs/cmd/zdb/zdb.c +++ b/sys/contrib/openzfs/cmd/zdb/zdb.c @@ -106,11 +106,15 @@ extern boolean_t spa_mode_readable_spacemaps; extern uint_t zfs_reconstruct_indirect_combinations_max; extern uint_t zfs_btree_verify_intensity; +enum { + ARG_ALLOCATED = 256, + ARG_BLOCK_BIN_MODE, + ARG_BLOCK_CLASSES, +}; + static const char cmdname[] = "zdb"; uint8_t dump_opt[512]; -#define ALLOCATED_OPT 256 - typedef void object_viewer_t(objset_t *, uint64_t, void *data, size_t size); static uint64_t *zopt_metaslab = NULL; @@ -131,6 +135,20 @@ static objset_t *os; static boolean_t kernel_init_done; static boolean_t corruption_found = B_FALSE; +static enum { + BIN_AUTO = 0, + BIN_PSIZE, + BIN_LSIZE, + BIN_ASIZE, +} block_bin_mode = BIN_AUTO; + +static enum { + CLASS_NORMAL = 1 << 1, + CLASS_SPECIAL = 1 << 2, + CLASS_DEDUP = 1 << 3, + CLASS_OTHER = 1 << 4, +} block_classes = 0; + static void snprintf_blkptr_compact(char *, size_t, const blkptr_t *, boolean_t); static void mos_obj_refd(uint64_t); @@ -749,6 +767,12 @@ usage(void) (void) fprintf(stderr, " Options to control amount of output:\n"); (void) fprintf(stderr, " -b --block-stats " "block statistics\n"); + (void) fprintf(stderr, " --bin=(lsize|psize|asize) " + "bin blocks based on this size in all three columns\n"); + (void) fprintf(stderr, + " --class=(normal|special|dedup|other)[,...]\n" + " only consider blocks from " + "these allocation classes\n"); (void) fprintf(stderr, " -B --backup " "backup stream\n"); (void) fprintf(stderr, " -c --checksum " @@ -1694,7 +1718,7 @@ dump_metaslab(metaslab_t *msp) (u_longlong_t)msp->ms_id, (u_longlong_t)msp->ms_start, (u_longlong_t)space_map_object(sm), freebuf); - if (dump_opt[ALLOCATED_OPT] || + if (dump_opt[ARG_ALLOCATED] || (dump_opt['m'] > 2 && !dump_opt['L'])) { mutex_enter(&msp->ms_lock); VERIFY0(metaslab_load(msp)); @@ -1705,7 +1729,7 @@ dump_metaslab(metaslab_t *msp) dump_metaslab_stats(msp); } - if (dump_opt[ALLOCATED_OPT]) { + if (dump_opt[ARG_ALLOCATED]) { uint64_t off = msp->ms_start; zfs_range_tree_walk(msp->ms_allocatable, dump_allocated, &off); @@ -1726,7 +1750,7 @@ dump_metaslab(metaslab_t *msp) SPACE_MAP_HISTOGRAM_SIZE, sm->sm_shift); } - if (dump_opt[ALLOCATED_OPT] || + if (dump_opt[ARG_ALLOCATED] || (dump_opt['m'] > 2 && !dump_opt['L'])) { metaslab_unload(msp); mutex_exit(&msp->ms_lock); @@ -5814,6 +5838,34 @@ dump_size_histograms(zdb_cb_t *zcb) (void) printf("\nBlock Size Histogram\n"); + switch (block_bin_mode) { + case BIN_PSIZE: + printf("(note: all categories are binned by %s)\n", "psize"); + break; + case BIN_LSIZE: + printf("(note: all categories are binned by %s)\n", "lsize"); + break; + case BIN_ASIZE: + printf("(note: all categories are binned by %s)\n", "asize"); + break; + default: + printf("(note: all categories are binned separately)\n"); + break; + } + if (block_classes != 0) { + char buf[256] = ""; + if (block_classes & CLASS_NORMAL) + strlcat(buf, "\"normal\", ", sizeof (buf)); + if (block_classes & CLASS_SPECIAL) + strlcat(buf, "\"special\", ", sizeof (buf)); + if (block_classes & CLASS_DEDUP) + strlcat(buf, "\"dedup\", ", sizeof (buf)); + if (block_classes & CLASS_OTHER) + strlcat(buf, "\"other\", ", sizeof (buf)); + buf[strlen(buf)-2] = '\0'; + printf("(note: only blocks in these classes are counted: %s)\n", + buf); + } /* * Print the first line titles */ @@ -6162,29 +6214,85 @@ skipped: [BPE_GET_PSIZE(bp)]++; return; } + + if (block_classes != 0) { + spa_config_enter(zcb->zcb_spa, SCL_CONFIG, FTAG, RW_READER); + + uint64_t vdev = DVA_GET_VDEV(&bp->blk_dva[0]); + uint64_t offset = DVA_GET_OFFSET(&bp->blk_dva[0]); + vdev_t *vd = vdev_lookup_top(zcb->zcb_spa, vdev); + ASSERT(vd != NULL); + metaslab_t *ms = vd->vdev_ms[offset >> vd->vdev_ms_shift]; + ASSERT(ms != NULL); + metaslab_group_t *mg = ms->ms_group; + ASSERT(mg != NULL); + metaslab_class_t *mc = mg->mg_class; + ASSERT(mc != NULL); + + spa_config_exit(zcb->zcb_spa, SCL_CONFIG, FTAG); + + int class; + if (mc == spa_normal_class(zcb->zcb_spa)) { + class = CLASS_NORMAL; + } else if (mc == spa_special_class(zcb->zcb_spa)) { + class = CLASS_SPECIAL; + } else if (mc == spa_dedup_class(zcb->zcb_spa)) { + class = CLASS_DEDUP; + } else { + class = CLASS_OTHER; + } + + if (!(block_classes & class)) { + goto hist_skipped; + } + } + /* * The binning histogram bins by powers of two up to * SPA_MAXBLOCKSIZE rather than creating bins for * every possible blocksize found in the pool. */ - int bin = highbit64(BP_GET_PSIZE(bp)) - 1; + int bin; + + /* + * Binning strategy: each bin includes blocks up to and including + * the given size (excluding blocks that fit into the previous bin). + * This way, the "4K" bin includes blocks within the (2K; 4K] range. + */ +#define BIN(size) (highbit64((size) - 1)) + + switch (block_bin_mode) { + case BIN_PSIZE: bin = BIN(BP_GET_PSIZE(bp)); break; + case BIN_LSIZE: bin = BIN(BP_GET_LSIZE(bp)); break; + case BIN_ASIZE: bin = BIN(BP_GET_ASIZE(bp)); break; + case BIN_AUTO: break; + default: PANIC("bad block_bin_mode"); abort(); + } + + if (block_bin_mode == BIN_AUTO) + bin = BIN(BP_GET_PSIZE(bp)); zcb->zcb_psize_count[bin]++; zcb->zcb_psize_len[bin] += BP_GET_PSIZE(bp); zcb->zcb_psize_total += BP_GET_PSIZE(bp); - bin = highbit64(BP_GET_LSIZE(bp)) - 1; + if (block_bin_mode == BIN_AUTO) + bin = BIN(BP_GET_LSIZE(bp)); zcb->zcb_lsize_count[bin]++; zcb->zcb_lsize_len[bin] += BP_GET_LSIZE(bp); zcb->zcb_lsize_total += BP_GET_LSIZE(bp); - bin = highbit64(BP_GET_ASIZE(bp)) - 1; + if (block_bin_mode == BIN_AUTO) + bin = BIN(BP_GET_ASIZE(bp)); zcb->zcb_asize_count[bin]++; zcb->zcb_asize_len[bin] += BP_GET_ASIZE(bp); zcb->zcb_asize_total += BP_GET_ASIZE(bp); +#undef BIN + +hist_skipped: if (!do_claim) return; @@ -9426,7 +9534,11 @@ main(int argc, char **argv) {"livelist", no_argument, NULL, 'y'}, {"zstd-headers", no_argument, NULL, 'Z'}, {"allocated-map", no_argument, NULL, - ALLOCATED_OPT}, + ARG_ALLOCATED}, + {"bin", required_argument, NULL, + ARG_BLOCK_BIN_MODE}, + {"class", required_argument, NULL, + ARG_BLOCK_CLASSES}, {0, 0, 0, 0} }; @@ -9457,7 +9569,7 @@ main(int argc, char **argv) case 'u': case 'y': case 'Z': - case ALLOCATED_OPT: + case ARG_ALLOCATED: dump_opt[c]++; dump_all = 0; break; @@ -9540,6 +9652,59 @@ main(int argc, char **argv) case 'x': vn_dumpdir = optarg; break; + case ARG_BLOCK_BIN_MODE: + if (strcmp(optarg, "lsize") == 0) { + block_bin_mode = BIN_LSIZE; + } else if (strcmp(optarg, "psize") == 0) { + block_bin_mode = BIN_PSIZE; + } else if (strcmp(optarg, "asize") == 0) { + block_bin_mode = BIN_ASIZE; + } else { + (void) fprintf(stderr, + "--bin=\"%s\" must be one of \"lsize\", " + "\"psize\" or \"asize\"\n", optarg); + usage(); + } + break; + + case ARG_BLOCK_CLASSES: { + char *buf = strdup(optarg), *tok = buf, *next, + *save = NULL; + + while ((next = strtok_r(tok, ",", &save)) != NULL) { + tok = NULL; + + if (strcmp(next, "normal") == 0) { + block_classes |= CLASS_NORMAL; + } else if (strcmp(next, "special") == 0) { + block_classes |= CLASS_SPECIAL; + } else if (strcmp(next, "dedup") == 0) { + block_classes |= CLASS_DEDUP; + } else if (strcmp(next, "other") == 0) { + block_classes |= CLASS_OTHER; + } else { + (void) fprintf(stderr, + "--class=\"%s\" must be a " + "comma-separated list of either " + "\"normal\", \"special\", " + "\"asize\" or \"other\"; " + "got \"%s\"\n", + optarg, next); + usage(); + } + } + + if (block_classes == 0) { + (void) fprintf(stderr, + "--class= must be a comma-separated " + "list of either \"normal\", \"special\", " + "\"asize\" or \"other\"; got empty\n"); + usage(); + } + + free(buf); + break; + } default: usage(); break; @@ -9582,6 +9747,9 @@ main(int argc, char **argv) */ spa_mode_readable_spacemaps = B_TRUE; + libspl_set_assert_ok((dump_opt['A'] == 1) || (dump_opt['A'] > 2)); + zfs_recover = (dump_opt['A'] > 1); + if (dump_all) verbose = MAX(verbose, 1); @@ -9592,9 +9760,6 @@ main(int argc, char **argv) dump_opt[c] += verbose; } - libspl_set_assert_ok((dump_opt['A'] == 1) || (dump_opt['A'] > 2)); - zfs_recover = (dump_opt['A'] > 1); - argc -= optind; argv += optind; if (argc < 2 && dump_opt['R']) diff --git a/sys/contrib/openzfs/cmd/zpool/zpool_main.c b/sys/contrib/openzfs/cmd/zpool/zpool_main.c index 1feec55c0e8b..a6658a9c2800 100644 --- a/sys/contrib/openzfs/cmd/zpool/zpool_main.c +++ b/sys/contrib/openzfs/cmd/zpool/zpool_main.c @@ -6975,7 +6975,6 @@ collect_vdev_prop(zpool_prop_t prop, uint64_t value, const char *str, /* * print static default line per vdev - * not compatible with '-o' <proplist> option */ static void collect_list_stats(zpool_handle_t *zhp, const char *name, nvlist_t *nv, @@ -7031,48 +7030,98 @@ collect_list_stats(zpool_handle_t *zhp, const char *name, nvlist_t *nv, * 'toplevel' boolean value is passed to the print_one_column() * to indicate that the value is valid. */ - if (VDEV_STAT_VALID(vs_pspace, c) && vs->vs_pspace) { - collect_vdev_prop(ZPOOL_PROP_SIZE, vs->vs_pspace, NULL, - scripted, B_TRUE, format, cb->cb_json, props, - cb->cb_json_as_int); - } else { - collect_vdev_prop(ZPOOL_PROP_SIZE, vs->vs_space, NULL, - scripted, toplevel, format, cb->cb_json, props, - cb->cb_json_as_int); + for (zprop_list_t *pl = cb->cb_proplist; pl != NULL; + pl = pl->pl_next) { + switch (pl->pl_prop) { + case ZPOOL_PROP_SIZE: + if (VDEV_STAT_VALID(vs_pspace, c) && + vs->vs_pspace) { + collect_vdev_prop( + ZPOOL_PROP_SIZE, vs->vs_pspace, + NULL, scripted, B_TRUE, format, + cb->cb_json, props, + cb->cb_json_as_int); + } else { + collect_vdev_prop( + ZPOOL_PROP_SIZE, vs->vs_space, NULL, + scripted, toplevel, format, + cb->cb_json, props, + cb->cb_json_as_int); + } + break; + case ZPOOL_PROP_ALLOCATED: + collect_vdev_prop(ZPOOL_PROP_ALLOCATED, + vs->vs_alloc, NULL, scripted, toplevel, + format, cb->cb_json, props, + cb->cb_json_as_int); + break; + + case ZPOOL_PROP_FREE: + collect_vdev_prop(ZPOOL_PROP_FREE, + vs->vs_space - vs->vs_alloc, NULL, scripted, + toplevel, format, cb->cb_json, props, + cb->cb_json_as_int); + break; + + case ZPOOL_PROP_CHECKPOINT: + collect_vdev_prop(ZPOOL_PROP_CHECKPOINT, + vs->vs_checkpoint_space, NULL, scripted, + toplevel, format, cb->cb_json, props, + cb->cb_json_as_int); + break; + + case ZPOOL_PROP_EXPANDSZ: + collect_vdev_prop(ZPOOL_PROP_EXPANDSZ, + vs->vs_esize, NULL, scripted, B_TRUE, + format, cb->cb_json, props, + cb->cb_json_as_int); + break; + + case ZPOOL_PROP_FRAGMENTATION: + collect_vdev_prop( + ZPOOL_PROP_FRAGMENTATION, + vs->vs_fragmentation, NULL, scripted, + (vs->vs_fragmentation != ZFS_FRAG_INVALID && + toplevel), + format, cb->cb_json, props, + cb->cb_json_as_int); + break; + + case ZPOOL_PROP_CAPACITY: + cap = (vs->vs_space == 0) ? + 0 : (vs->vs_alloc * 10000 / vs->vs_space); + collect_vdev_prop(ZPOOL_PROP_CAPACITY, cap, + NULL, scripted, toplevel, format, + cb->cb_json, props, cb->cb_json_as_int); + break; + + case ZPOOL_PROP_HEALTH: + state = zpool_state_to_name(vs->vs_state, + vs->vs_aux); + if (isspare) { + if (vs->vs_aux == VDEV_AUX_SPARED) + state = "INUSE"; + else if (vs->vs_state == + VDEV_STATE_HEALTHY) + state = "AVAIL"; + } + collect_vdev_prop(ZPOOL_PROP_HEALTH, 0, state, + scripted, B_TRUE, format, cb->cb_json, + props, cb->cb_json_as_int); + break; + + case ZPOOL_PROP_NAME: + break; + + default: + collect_vdev_prop(pl->pl_prop, 0, + NULL, scripted, B_FALSE, format, + cb->cb_json, props, cb->cb_json_as_int); + + } + + } - collect_vdev_prop(ZPOOL_PROP_ALLOCATED, vs->vs_alloc, NULL, - scripted, toplevel, format, cb->cb_json, props, - cb->cb_json_as_int); - collect_vdev_prop(ZPOOL_PROP_FREE, vs->vs_space - vs->vs_alloc, - NULL, scripted, toplevel, format, cb->cb_json, props, - cb->cb_json_as_int); - collect_vdev_prop(ZPOOL_PROP_CHECKPOINT, - vs->vs_checkpoint_space, NULL, scripted, toplevel, format, - cb->cb_json, props, cb->cb_json_as_int); - collect_vdev_prop(ZPOOL_PROP_EXPANDSZ, vs->vs_esize, NULL, - scripted, B_TRUE, format, cb->cb_json, props, - cb->cb_json_as_int); - collect_vdev_prop(ZPOOL_PROP_FRAGMENTATION, - vs->vs_fragmentation, NULL, scripted, - (vs->vs_fragmentation != ZFS_FRAG_INVALID && toplevel), - format, cb->cb_json, props, cb->cb_json_as_int); - cap = (vs->vs_space == 0) ? 0 : - (vs->vs_alloc * 10000 / vs->vs_space); - collect_vdev_prop(ZPOOL_PROP_CAPACITY, cap, NULL, - scripted, toplevel, format, cb->cb_json, props, - cb->cb_json_as_int); - collect_vdev_prop(ZPOOL_PROP_DEDUPRATIO, 0, NULL, - scripted, toplevel, format, cb->cb_json, props, - cb->cb_json_as_int); - state = zpool_state_to_name(vs->vs_state, vs->vs_aux); - if (isspare) { - if (vs->vs_aux == VDEV_AUX_SPARED) - state = "INUSE"; - else if (vs->vs_state == VDEV_STATE_HEALTHY) - state = "AVAIL"; - } - collect_vdev_prop(ZPOOL_PROP_HEALTH, 0, state, scripted, - B_TRUE, format, cb->cb_json, props, cb->cb_json_as_int); if (cb->cb_json) { fnvlist_add_nvlist(ent, "properties", props); diff --git a/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c b/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c index 088c0108e911..222b5524669e 100644 --- a/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c +++ b/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c @@ -270,14 +270,13 @@ is_spare(nvlist_t *config, const char *path) * draid* Virtual dRAID spare */ static nvlist_t * -make_leaf_vdev(nvlist_t *props, const char *arg, boolean_t is_primary) +make_leaf_vdev(const char *arg, boolean_t is_primary, uint64_t ashift) { char path[MAXPATHLEN]; struct stat64 statbuf; nvlist_t *vdev = NULL; const char *type = NULL; boolean_t wholedisk = B_FALSE; - uint64_t ashift = 0; int err; /* @@ -382,31 +381,6 @@ make_leaf_vdev(nvlist_t *props, const char *arg, boolean_t is_primary) (uint64_t)wholedisk) == 0); /* - * Override defaults if custom properties are provided. - */ - if (props != NULL) { - const char *value = NULL; - - if (nvlist_lookup_string(props, - zpool_prop_to_name(ZPOOL_PROP_ASHIFT), &value) == 0) { - if (zfs_nicestrtonum(NULL, value, &ashift) != 0) { - (void) fprintf(stderr, - gettext("ashift must be a number.\n")); - return (NULL); - } - if (ashift != 0 && - (ashift < ASHIFT_MIN || ashift > ASHIFT_MAX)) { - (void) fprintf(stderr, - gettext("invalid 'ashift=%" PRIu64 "' " - "property: only values between %" PRId32 " " - "and %" PRId32 " are allowed.\n"), - ashift, ASHIFT_MIN, ASHIFT_MAX); - return (NULL); - } - } - } - - /* * If the device is known to incorrectly report its physical sector * size explicitly provide the known correct value. */ @@ -1513,6 +1487,29 @@ construct_spec(nvlist_t *props, int argc, char **argv) const char *type, *fulltype; boolean_t is_log, is_special, is_dedup, is_spare; boolean_t seen_logs; + uint64_t ashift = 0; + + if (props != NULL) { + const char *value = NULL; + + if (nvlist_lookup_string(props, + zpool_prop_to_name(ZPOOL_PROP_ASHIFT), &value) == 0) { + if (zfs_nicestrtonum(NULL, value, &ashift) != 0) { + (void) fprintf(stderr, + gettext("ashift must be a number.\n")); + return (NULL); + } + if (ashift != 0 && + (ashift < ASHIFT_MIN || ashift > ASHIFT_MAX)) { + (void) fprintf(stderr, + gettext("invalid 'ashift=%" PRIu64 "' " + "property: only values between %" PRId32 " " + "and %" PRId32 " are allowed.\n"), + ashift, ASHIFT_MIN, ASHIFT_MAX); + return (NULL); + } + } + } top = NULL; toplevels = 0; @@ -1618,9 +1615,9 @@ construct_spec(nvlist_t *props, int argc, char **argv) children * sizeof (nvlist_t *)); if (child == NULL) zpool_no_memory(); - if ((nv = make_leaf_vdev(props, argv[c], + if ((nv = make_leaf_vdev(argv[c], !(is_log || is_special || is_dedup || - is_spare))) == NULL) { + is_spare), ashift)) == NULL) { for (c = 0; c < children - 1; c++) nvlist_free(child[c]); free(child); @@ -1684,6 +1681,10 @@ construct_spec(nvlist_t *props, int argc, char **argv) ZPOOL_CONFIG_ALLOCATION_BIAS, VDEV_ALLOC_BIAS_DEDUP) == 0); } + if (ashift > 0) { + fnvlist_add_uint64(nv, + ZPOOL_CONFIG_ASHIFT, ashift); + } if (strcmp(type, VDEV_TYPE_RAIDZ) == 0) { verify(nvlist_add_uint64(nv, ZPOOL_CONFIG_NPARITY, @@ -1711,8 +1712,9 @@ construct_spec(nvlist_t *props, int argc, char **argv) * We have a device. Pass off to make_leaf_vdev() to * construct the appropriate nvlist describing the vdev. */ - if ((nv = make_leaf_vdev(props, argv[0], !(is_log || - is_special || is_dedup || is_spare))) == NULL) + if ((nv = make_leaf_vdev(argv[0], !(is_log || + is_special || is_dedup || is_spare), + ashift)) == NULL) goto spec_out; verify(nvlist_add_uint64(nv, diff --git a/sys/contrib/openzfs/config/kernel-block-device-operations.m4 b/sys/contrib/openzfs/config/kernel-block-device-operations.m4 index 4ff20b9c413d..1905340a9c7d 100644 --- a/sys/contrib/openzfs/config/kernel-block-device-operations.m4 +++ b/sys/contrib/openzfs/config/kernel-block-device-operations.m4 @@ -119,15 +119,49 @@ AC_DEFUN([ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK], [ ]) ]) +dnl # +dnl # 6.18 API change +dnl # block_device_operation->getgeo takes struct gendisk* as first arg +dnl # +AC_DEFUN([ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS_GETGEO_GENDISK], [ + ZFS_LINUX_TEST_SRC([block_device_operations_getgeo_gendisk], [ + #include <linux/blkdev.h> + + static int blk_getgeo(struct gendisk *disk, struct hd_geometry *geo) + { + (void) disk, (void) geo; + return (0); + } + + static const struct block_device_operations + bops __attribute__ ((unused)) = { + .getgeo = blk_getgeo, + }; + ], [], []) +]) + +AC_DEFUN([ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS_GETGEO_GENDISK], [ + AC_MSG_CHECKING([whether bops->getgeo() takes gendisk as first arg]) + ZFS_LINUX_TEST_RESULT([block_device_operations_getgeo_gendisk], [ + AC_MSG_RESULT(yes) + AC_DEFINE([HAVE_BLOCK_DEVICE_OPERATIONS_GETGEO_GENDISK], [1], + [Define if getgeo() in block_device_operations takes struct gendisk * as its first arg]) + ],[ + AC_MSG_RESULT(no) + ]) +]) + AC_DEFUN([ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS], [ ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS_CHECK_EVENTS ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS_RELEASE_VOID ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS_RELEASE_1ARG ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK + ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS_GETGEO_GENDISK ]) AC_DEFUN([ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS], [ ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS_CHECK_EVENTS ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS_RELEASE_VOID ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK + ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS_GETGEO_GENDISK ]) diff --git a/sys/contrib/openzfs/config/kernel-drop-inode.m4 b/sys/contrib/openzfs/config/kernel-drop-inode.m4 new file mode 100644 index 000000000000..6f2b12cadc02 --- /dev/null +++ b/sys/contrib/openzfs/config/kernel-drop-inode.m4 @@ -0,0 +1,24 @@ +dnl # +dnl # 6.18 API change +dnl # - generic_drop_inode() renamed to inode_generic_drop() +dnl # - generic_delete_inode() renamed to inode_just_drop() +dnl # +AC_DEFUN([ZFS_AC_KERNEL_SRC_INODE_GENERIC_DROP], [ + ZFS_LINUX_TEST_SRC([inode_generic_drop], [ + #include <linux/fs.h> + ],[ + struct inode *ip = NULL; + inode_generic_drop(ip); + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_INODE_GENERIC_DROP], [ + AC_MSG_CHECKING([whether inode_generic_drop() exists]) + ZFS_LINUX_TEST_RESULT([inode_generic_drop], [ + AC_MSG_RESULT(yes) + AC_DEFINE(HAVE_INODE_GENERIC_DROP, 1, + [inode_generic_drop() exists]) + ],[ + AC_MSG_RESULT(no) + ]) +]) diff --git a/sys/contrib/openzfs/config/kernel-namespace.m4 b/sys/contrib/openzfs/config/kernel-namespace.m4 new file mode 100644 index 000000000000..9b0b12e4eab4 --- /dev/null +++ b/sys/contrib/openzfs/config/kernel-namespace.m4 @@ -0,0 +1,31 @@ +dnl # +dnl # 6.18 API change +dnl # ns->ops->type was moved to ns->ns.ns_type (struct ns_common) +dnl # +AC_DEFUN([ZFS_AC_KERNEL_SRC_NS_COMMON_TYPE], [ + ZFS_LINUX_TEST_SRC([ns_common_type], [ + #include <linux/user_namespace.h> + ],[ + struct user_namespace ns; + ns.ns.ns_type = 0; + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_NS_COMMON_TYPE], [ + AC_MSG_CHECKING([whether ns_type is accessible through ns_common]) + ZFS_LINUX_TEST_RESULT([ns_common_type], [ + AC_MSG_RESULT(yes) + AC_DEFINE([HAVE_NS_COMMON_TYPE], 1, + [Define if ns_type is accessible through ns_common]) + ],[ + AC_MSG_RESULT(no) + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_SRC_NAMESPACE], [ + ZFS_AC_KERNEL_SRC_NS_COMMON_TYPE +]) + +AC_DEFUN([ZFS_AC_KERNEL_NAMESPACE], [ + ZFS_AC_KERNEL_NS_COMMON_TYPE +]) diff --git a/sys/contrib/openzfs/config/kernel-userns-capabilities.m4 b/sys/contrib/openzfs/config/kernel-userns-capabilities.m4 deleted file mode 100644 index f4e24fb1606a..000000000000 --- a/sys/contrib/openzfs/config/kernel-userns-capabilities.m4 +++ /dev/null @@ -1,79 +0,0 @@ -dnl # -dnl # 2.6.38 API change -dnl # ns_capable() was introduced -dnl # -AC_DEFUN([ZFS_AC_KERNEL_SRC_NS_CAPABLE], [ - ZFS_LINUX_TEST_SRC([ns_capable], [ - #include <linux/capability.h> - ],[ - ns_capable((struct user_namespace *)NULL, CAP_SYS_ADMIN); - ]) -]) - -AC_DEFUN([ZFS_AC_KERNEL_NS_CAPABLE], [ - AC_MSG_CHECKING([whether ns_capable exists]) - ZFS_LINUX_TEST_RESULT([ns_capable], [ - AC_MSG_RESULT(yes) - ],[ - ZFS_LINUX_TEST_ERROR([ns_capable()]) - ]) -]) - -dnl # -dnl # 2.6.39 API change -dnl # struct user_namespace was added to struct cred_t as cred->user_ns member -dnl # -AC_DEFUN([ZFS_AC_KERNEL_SRC_CRED_USER_NS], [ - ZFS_LINUX_TEST_SRC([cred_user_ns], [ - #include <linux/cred.h> - ],[ - struct cred cr; - cr.user_ns = (struct user_namespace *)NULL; - ]) -]) - -AC_DEFUN([ZFS_AC_KERNEL_CRED_USER_NS], [ - AC_MSG_CHECKING([whether cred_t->user_ns exists]) - ZFS_LINUX_TEST_RESULT([cred_user_ns], [ - AC_MSG_RESULT(yes) - ],[ - ZFS_LINUX_TEST_ERROR([cred_t->user_ns()]) - ]) -]) - -dnl # -dnl # 3.4 API change -dnl # kuid_has_mapping() and kgid_has_mapping() were added to distinguish -dnl # between internal kernel uids/gids and user namespace uids/gids. -dnl # -AC_DEFUN([ZFS_AC_KERNEL_SRC_KUID_HAS_MAPPING], [ - ZFS_LINUX_TEST_SRC([kuid_has_mapping], [ - #include <linux/uidgid.h> - ],[ - kuid_has_mapping((struct user_namespace *)NULL, KUIDT_INIT(0)); - kgid_has_mapping((struct user_namespace *)NULL, KGIDT_INIT(0)); - ]) -]) - -AC_DEFUN([ZFS_AC_KERNEL_KUID_HAS_MAPPING], [ - AC_MSG_CHECKING([whether kuid_has_mapping/kgid_has_mapping exist]) - ZFS_LINUX_TEST_RESULT([kuid_has_mapping], [ - AC_MSG_RESULT(yes) - ],[ - ZFS_LINUX_TEST_ERROR([kuid_has_mapping()]) - ]) -]) - -AC_DEFUN([ZFS_AC_KERNEL_SRC_USERNS_CAPABILITIES], [ - ZFS_AC_KERNEL_SRC_NS_CAPABLE - ZFS_AC_KERNEL_SRC_HAS_CAPABILITY - ZFS_AC_KERNEL_SRC_CRED_USER_NS - ZFS_AC_KERNEL_SRC_KUID_HAS_MAPPING -]) - -AC_DEFUN([ZFS_AC_KERNEL_USERNS_CAPABILITIES], [ - ZFS_AC_KERNEL_NS_CAPABLE - ZFS_AC_KERNEL_HAS_CAPABILITY - ZFS_AC_KERNEL_CRED_USER_NS - ZFS_AC_KERNEL_KUID_HAS_MAPPING -]) diff --git a/sys/contrib/openzfs/config/kernel-writeback.m4 b/sys/contrib/openzfs/config/kernel-writeback.m4 new file mode 100644 index 000000000000..334d65ef84b6 --- /dev/null +++ b/sys/contrib/openzfs/config/kernel-writeback.m4 @@ -0,0 +1,58 @@ +AC_DEFUN([ZFS_AC_KERNEL_SRC_WRITEPAGE_T], [ + dnl # + dnl # 6.3 API change + dnl # The writepage_t function type now has its first argument as + dnl # struct folio* instead of struct page* + dnl # + ZFS_LINUX_TEST_SRC([writepage_t_folio], [ + #include <linux/writeback.h> + static int putpage(struct folio *folio, + struct writeback_control *wbc, void *data) + { return 0; } + writepage_t func = putpage; + ],[]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_WRITEPAGE_T], [ + AC_MSG_CHECKING([whether int (*writepage_t)() takes struct folio*]) + ZFS_LINUX_TEST_RESULT([writepage_t_folio], [ + AC_MSG_RESULT(yes) + AC_DEFINE(HAVE_WRITEPAGE_T_FOLIO, 1, + [int (*writepage_t)() takes struct folio*]) + ],[ + AC_MSG_RESULT(no) + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_SRC_WRITE_CACHE_PAGES], [ + dnl # + dnl # 6.18 API change + dnl # write_cache_pages() has been removed. + dnl # + ZFS_LINUX_TEST_SRC([write_cache_pages], [ + #include <linux/writeback.h> + ], [ + (void) write_cache_pages(NULL, NULL, NULL, NULL); + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_WRITE_CACHE_PAGES], [ + AC_MSG_CHECKING([whether write_cache_pages() is available]) + ZFS_LINUX_TEST_RESULT([write_cache_pages], [ + AC_MSG_RESULT(yes) + AC_DEFINE(HAVE_WRITE_CACHE_PAGES, 1, + [write_cache_pages() is available]) + ],[ + AC_MSG_RESULT(no) + ]) +]) + +AC_DEFUN([ZFS_AC_KERNEL_SRC_WRITEBACK], [ + ZFS_AC_KERNEL_SRC_WRITEPAGE_T + ZFS_AC_KERNEL_SRC_WRITE_CACHE_PAGES +]) + +AC_DEFUN([ZFS_AC_KERNEL_WRITEBACK], [ + ZFS_AC_KERNEL_WRITEPAGE_T + ZFS_AC_KERNEL_WRITE_CACHE_PAGES +]) diff --git a/sys/contrib/openzfs/config/kernel-writepage_t.m4 b/sys/contrib/openzfs/config/kernel-writepage_t.m4 deleted file mode 100644 index a82cf370c9d4..000000000000 --- a/sys/contrib/openzfs/config/kernel-writepage_t.m4 +++ /dev/null @@ -1,26 +0,0 @@ -AC_DEFUN([ZFS_AC_KERNEL_SRC_WRITEPAGE_T], [ - dnl # - dnl # 6.3 API change - dnl # The writepage_t function type now has its first argument as - dnl # struct folio* instead of struct page* - dnl # - ZFS_LINUX_TEST_SRC([writepage_t_folio], [ - #include <linux/writeback.h> - static int putpage(struct folio *folio, - struct writeback_control *wbc, void *data) - { return 0; } - writepage_t func = putpage; - ],[]) -]) - -AC_DEFUN([ZFS_AC_KERNEL_WRITEPAGE_T], [ - AC_MSG_CHECKING([whether int (*writepage_t)() takes struct folio*]) - ZFS_LINUX_TEST_RESULT([writepage_t_folio], [ - AC_MSG_RESULT(yes) - AC_DEFINE(HAVE_WRITEPAGE_T_FOLIO, 1, - [int (*writepage_t)() takes struct folio*]) - ],[ - AC_MSG_RESULT(no) - ]) -]) - diff --git a/sys/contrib/openzfs/config/kernel.m4 b/sys/contrib/openzfs/config/kernel.m4 index 35819e4d68c5..40b7de739882 100644 --- a/sys/contrib/openzfs/config/kernel.m4 +++ b/sys/contrib/openzfs/config/kernel.m4 @@ -121,7 +121,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [ ZFS_AC_KERNEL_SRC_IDMAP_MNT_API ZFS_AC_KERNEL_SRC_IDMAP_NO_USERNS ZFS_AC_KERNEL_SRC_IATTR_VFSID - ZFS_AC_KERNEL_SRC_WRITEPAGE_T + ZFS_AC_KERNEL_SRC_WRITEBACK ZFS_AC_KERNEL_SRC_RECLAIMED ZFS_AC_KERNEL_SRC_REGISTER_SYSCTL_TABLE ZFS_AC_KERNEL_SRC_REGISTER_SYSCTL_SZ @@ -136,6 +136,8 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [ ZFS_AC_KERNEL_SRC_TIMER ZFS_AC_KERNEL_SRC_SUPER_BLOCK_S_WB_ERR ZFS_AC_KERNEL_SRC_SOPS_FREE_INODE + ZFS_AC_KERNEL_SRC_NAMESPACE + ZFS_AC_KERNEL_SRC_INODE_GENERIC_DROP case "$host_cpu" in powerpc*) ZFS_AC_KERNEL_SRC_CPU_HAS_FEATURE @@ -240,7 +242,7 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [ ZFS_AC_KERNEL_IDMAP_MNT_API ZFS_AC_KERNEL_IDMAP_NO_USERNS ZFS_AC_KERNEL_IATTR_VFSID - ZFS_AC_KERNEL_WRITEPAGE_T + ZFS_AC_KERNEL_WRITEBACK ZFS_AC_KERNEL_RECLAIMED ZFS_AC_KERNEL_REGISTER_SYSCTL_TABLE ZFS_AC_KERNEL_REGISTER_SYSCTL_SZ @@ -256,6 +258,8 @@ AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [ ZFS_AC_KERNEL_TIMER ZFS_AC_KERNEL_SUPER_BLOCK_S_WB_ERR ZFS_AC_KERNEL_SOPS_FREE_INODE + ZFS_AC_KERNEL_NAMESPACE + ZFS_AC_KERNEL_INODE_GENERIC_DROP case "$host_cpu" in powerpc*) ZFS_AC_KERNEL_CPU_HAS_FEATURE diff --git a/sys/contrib/openzfs/include/os/linux/kernel/linux/vfs_compat.h b/sys/contrib/openzfs/include/os/linux/kernel/linux/vfs_compat.h index cbf14e28371f..d9dc904bc322 100644 --- a/sys/contrib/openzfs/include/os/linux/kernel/linux/vfs_compat.h +++ b/sys/contrib/openzfs/include/os/linux/kernel/linux/vfs_compat.h @@ -23,6 +23,7 @@ /* * Copyright (C) 2011 Lawrence Livermore National Security, LLC. * Copyright (C) 2015 Jörg Thalheim. + * Copyright (c) 2025, Rob Norris <robn@despairlabs.com> */ #ifndef _ZFS_VFS_H @@ -262,4 +263,10 @@ zpl_is_32bit_api(void) #define zpl_generic_fillattr(user_ns, ip, sp) generic_fillattr(ip, sp) #endif +#ifdef HAVE_INODE_GENERIC_DROP +/* 6.18 API change. These were renamed, alias the old names to the new. */ +#define generic_delete_inode(ip) inode_just_drop(ip) +#define generic_drop_inode(ip) inode_generic_drop(ip) +#endif + #endif /* _ZFS_VFS_H */ diff --git a/sys/contrib/openzfs/lib/libspl/include/sys/uio.h b/sys/contrib/openzfs/lib/libspl/include/sys/uio.h index 93aa4984d734..9ada482be000 100644 --- a/sys/contrib/openzfs/lib/libspl/include/sys/uio.h +++ b/sys/contrib/openzfs/lib/libspl/include/sys/uio.h @@ -41,6 +41,7 @@ #ifndef _LIBSPL_SYS_UIO_H #define _LIBSPL_SYS_UIO_H +#include <sys/sysmacros.h> #include <sys/types.h> #include_next <sys/uio.h> diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_status.c b/sys/contrib/openzfs/lib/libzfs/libzfs_status.c index bdddefb92165..a589ca6896f0 100644 --- a/sys/contrib/openzfs/lib/libzfs/libzfs_status.c +++ b/sys/contrib/openzfs/lib/libzfs/libzfs_status.c @@ -98,57 +98,57 @@ static const char *const zfs_msgid_table[] = { #define NMSGID (sizeof (zfs_msgid_table) / sizeof (zfs_msgid_table[0])) static int -vdev_missing(vdev_stat_t *vs, uint_t vsc) +vdev_missing(vdev_stat_t *vs, uint_t vsc, void *arg) { - (void) vsc; + (void) vsc, (void) arg; return (vs->vs_state == VDEV_STATE_CANT_OPEN && vs->vs_aux == VDEV_AUX_OPEN_FAILED); } static int -vdev_faulted(vdev_stat_t *vs, uint_t vsc) +vdev_faulted(vdev_stat_t *vs, uint_t vsc, void *arg) { - (void) vsc; + (void) vsc, (void) arg; return (vs->vs_state == VDEV_STATE_FAULTED); } static int -vdev_errors(vdev_stat_t *vs, uint_t vsc) +vdev_errors(vdev_stat_t *vs, uint_t vsc, void *arg) { - (void) vsc; + (void) vsc, (void) arg; return (vs->vs_state == VDEV_STATE_DEGRADED || vs->vs_read_errors != 0 || vs->vs_write_errors != 0 || vs->vs_checksum_errors != 0); } static int -vdev_broken(vdev_stat_t *vs, uint_t vsc) +vdev_broken(vdev_stat_t *vs, uint_t vsc, void *arg) { - (void) vsc; + (void) vsc, (void) arg; return (vs->vs_state == VDEV_STATE_CANT_OPEN); } static int -vdev_offlined(vdev_stat_t *vs, uint_t vsc) +vdev_offlined(vdev_stat_t *vs, uint_t vsc, void *arg) { - (void) vsc; + (void) vsc, (void) arg; return (vs->vs_state == VDEV_STATE_OFFLINE); } static int -vdev_removed(vdev_stat_t *vs, uint_t vsc) +vdev_removed(vdev_stat_t *vs, uint_t vsc, void *arg) { - (void) vsc; + (void) vsc, (void) arg; return (vs->vs_state == VDEV_STATE_REMOVED); } static int -vdev_non_native_ashift(vdev_stat_t *vs, uint_t vsc) +vdev_non_native_ashift(vdev_stat_t *vs, uint_t vsc, void *arg) { - if (getenv("ZPOOL_STATUS_NON_NATIVE_ASHIFT_IGNORE") != NULL) - return (0); + uint64_t ashift = *(uint64_t *)arg; return (VDEV_STAT_VALID(vs_physical_ashift, vsc) && + (ashift == 0 || vs->vs_configured_ashift < ashift) && vs->vs_configured_ashift < vs->vs_physical_ashift); } @@ -156,8 +156,8 @@ vdev_non_native_ashift(vdev_stat_t *vs, uint_t vsc) * Detect if any leaf devices that have seen errors or could not be opened. */ static boolean_t -find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t), - boolean_t ignore_replacing) +find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t, void *), + void *arg, boolean_t ignore_replacing) { nvlist_t **child; uint_t c, children; @@ -177,14 +177,16 @@ find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t), if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, &child, &children) == 0) { - for (c = 0; c < children; c++) - if (find_vdev_problem(child[c], func, ignore_replacing)) + for (c = 0; c < children; c++) { + if (find_vdev_problem(child[c], func, arg, + ignore_replacing)) return (B_TRUE); + } } else { uint_t vsc; vdev_stat_t *vs = (vdev_stat_t *)fnvlist_lookup_uint64_array( vdev, ZPOOL_CONFIG_VDEV_STATS, &vsc); - if (func(vs, vsc) != 0) + if (func(vs, vsc, arg) != 0) return (B_TRUE); } @@ -193,9 +195,11 @@ find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t), */ if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_L2CACHE, &child, &children) == 0) { - for (c = 0; c < children; c++) - if (find_vdev_problem(child[c], func, ignore_replacing)) + for (c = 0; c < children; c++) { + if (find_vdev_problem(child[c], func, arg, + ignore_replacing)) return (B_TRUE); + } } return (B_FALSE); @@ -220,7 +224,7 @@ find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t), */ static zpool_status_t check_status(nvlist_t *config, boolean_t isimport, - zpool_errata_t *erratap, const char *compat) + zpool_errata_t *erratap, const char *compat, uint64_t ashift) { pool_scan_stat_t *ps = NULL; uint_t vsc, psc; @@ -371,15 +375,15 @@ check_status(nvlist_t *config, boolean_t isimport, * Bad devices in non-replicated config. */ if (vs->vs_state == VDEV_STATE_CANT_OPEN && - find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) + find_vdev_problem(nvroot, vdev_faulted, NULL, B_TRUE)) return (ZPOOL_STATUS_FAULTED_DEV_NR); if (vs->vs_state == VDEV_STATE_CANT_OPEN && - find_vdev_problem(nvroot, vdev_missing, B_TRUE)) + find_vdev_problem(nvroot, vdev_missing, NULL, B_TRUE)) return (ZPOOL_STATUS_MISSING_DEV_NR); if (vs->vs_state == VDEV_STATE_CANT_OPEN && - find_vdev_problem(nvroot, vdev_broken, B_TRUE)) + find_vdev_problem(nvroot, vdev_broken, NULL, B_TRUE)) return (ZPOOL_STATUS_CORRUPT_LABEL_NR); /* @@ -402,35 +406,37 @@ check_status(nvlist_t *config, boolean_t isimport, /* * Missing devices in a replicated config. */ - if (find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) + if (find_vdev_problem(nvroot, vdev_faulted, NULL, B_TRUE)) return (ZPOOL_STATUS_FAULTED_DEV_R); - if (find_vdev_problem(nvroot, vdev_missing, B_TRUE)) + if (find_vdev_problem(nvroot, vdev_missing, NULL, B_TRUE)) return (ZPOOL_STATUS_MISSING_DEV_R); - if (find_vdev_problem(nvroot, vdev_broken, B_TRUE)) + if (find_vdev_problem(nvroot, vdev_broken, NULL, B_TRUE)) return (ZPOOL_STATUS_CORRUPT_LABEL_R); /* * Devices with errors */ - if (!isimport && find_vdev_problem(nvroot, vdev_errors, B_TRUE)) + if (!isimport && find_vdev_problem(nvroot, vdev_errors, NULL, B_TRUE)) return (ZPOOL_STATUS_FAILING_DEV); /* * Offlined devices */ - if (find_vdev_problem(nvroot, vdev_offlined, B_TRUE)) + if (find_vdev_problem(nvroot, vdev_offlined, NULL, B_TRUE)) return (ZPOOL_STATUS_OFFLINE_DEV); /* * Removed device */ - if (find_vdev_problem(nvroot, vdev_removed, B_TRUE)) + if (find_vdev_problem(nvroot, vdev_removed, NULL, B_TRUE)) return (ZPOOL_STATUS_REMOVED_DEV); /* * Suboptimal, but usable, ashift configuration. */ - if (find_vdev_problem(nvroot, vdev_non_native_ashift, B_FALSE)) + if (!isimport && + getenv("ZPOOL_STATUS_NON_NATIVE_ASHIFT_IGNORE") == NULL && + find_vdev_problem(nvroot, vdev_non_native_ashift, &ashift, B_FALSE)) return (ZPOOL_STATUS_NON_NATIVE_ASHIFT); /* @@ -510,8 +516,10 @@ zpool_get_status(zpool_handle_t *zhp, const char **msgid, ZFS_MAXPROPLEN, NULL, B_FALSE) != 0) compatibility[0] = '\0'; + uint64_t ashift = zpool_get_prop_int(zhp, ZPOOL_PROP_ASHIFT, NULL); + zpool_status_t ret = check_status(zhp->zpool_config, B_FALSE, errata, - compatibility); + compatibility, ashift); if (msgid != NULL) { if (ret >= NMSGID) @@ -526,7 +534,7 @@ zpool_status_t zpool_import_status(nvlist_t *config, const char **msgid, zpool_errata_t *errata) { - zpool_status_t ret = check_status(config, B_TRUE, errata, NULL); + zpool_status_t ret = check_status(config, B_TRUE, errata, NULL, 0); if (ret >= NMSGID) *msgid = NULL; diff --git a/sys/contrib/openzfs/lib/libzpool/kernel.c b/sys/contrib/openzfs/lib/libzpool/kernel.c index 8ed374627264..70eba5099119 100644 --- a/sys/contrib/openzfs/lib/libzpool/kernel.c +++ b/sys/contrib/openzfs/lib/libzpool/kernel.c @@ -23,6 +23,7 @@ * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2018 by Delphix. All rights reserved. * Copyright (c) 2016 Actifio, Inc. All rights reserved. + * Copyright (c) 2025, Klara, Inc. */ #include <assert.h> @@ -645,39 +646,60 @@ __dprintf(boolean_t dprint, const char *file, const char *func, * cmn_err() and panic() * ========================================================================= */ -static char ce_prefix[CE_IGNORE][10] = { "", "NOTICE: ", "WARNING: ", "" }; -static char ce_suffix[CE_IGNORE][2] = { "", "\n", "\n", "" }; -__attribute__((noreturn)) void -vpanic(const char *fmt, va_list adx) +static __attribute__((noreturn)) void +panic_stop_or_abort(void) { - (void) fprintf(stderr, "error: "); - (void) vfprintf(stderr, fmt, adx); - (void) fprintf(stderr, "\n"); + const char *stopenv = getenv("LIBZPOOL_PANIC_STOP"); + if (stopenv != NULL && atoi(stopenv)) { + fputs("libzpool: LIBZPOOL_PANIC_STOP is set, sending " + "SIGSTOP to process group\n", stderr); + fflush(stderr); + + kill(0, SIGSTOP); + + fputs("libzpool: continued after panic stop, " + "aborting\n", stderr); + } abort(); /* think of it as a "user-level crash dump" */ } -__attribute__((noreturn)) void -panic(const char *fmt, ...) +static void +vcmn_msg(int ce, const char *fmt, va_list adx) { - va_list adx; + switch (ce) { + case CE_IGNORE: + return; + case CE_CONT: + break; + case CE_NOTE: + fputs("libzpool: NOTICE: ", stderr); + break; + case CE_WARN: + fputs("libzpool: WARNING: ", stderr); + break; + case CE_PANIC: + fputs("libzpool: PANIC: ", stderr); + break; + default: + fputs("libzpool: [unknown severity %d]: ", stderr); + break; + } - va_start(adx, fmt); - vpanic(fmt, adx); - va_end(adx); + vfprintf(stderr, fmt, adx); + if (ce != CE_CONT) + fputc('\n', stderr); + fflush(stderr); } void vcmn_err(int ce, const char *fmt, va_list adx) { + vcmn_msg(ce, fmt, adx); + if (ce == CE_PANIC) - vpanic(fmt, adx); - if (ce != CE_NOTE) { /* suppress noise in userland stress testing */ - (void) fprintf(stderr, "%s", ce_prefix[ce]); - (void) vfprintf(stderr, fmt, adx); - (void) fprintf(stderr, "%s", ce_suffix[ce]); - } + panic_stop_or_abort(); } void @@ -690,6 +712,25 @@ cmn_err(int ce, const char *fmt, ...) va_end(adx); } +__attribute__((noreturn)) void +panic(const char *fmt, ...) +{ + va_list adx; + + va_start(adx, fmt); + vcmn_msg(CE_PANIC, fmt, adx); + va_end(adx); + + panic_stop_or_abort(); +} + +__attribute__((noreturn)) void +vpanic(const char *fmt, va_list adx) +{ + vcmn_msg(CE_PANIC, fmt, adx); + panic_stop_or_abort(); +} + /* * ========================================================================= * misc routines diff --git a/sys/contrib/openzfs/man/man8/zdb.8 b/sys/contrib/openzfs/man/man8/zdb.8 index c3290ea14769..f51e24fa849c 100644 --- a/sys/contrib/openzfs/man/man8/zdb.8 +++ b/sys/contrib/openzfs/man/man8/zdb.8 @@ -144,6 +144,20 @@ subcommand. Display statistics regarding the number, size .Pq logical, physical and allocated and deduplication of blocks. +.It Fl -bin Ns = Ns ( Li lsize Ns | Ns Li psize Ns | Ns Li asize ) +When used with +.Fl bb , +sort blocks into all three bins according to the given size (instead of binning +a block for each size separately). +.Pp +For instance, with +.Fl -bin Ns = Ns Li lsize , +a block with lsize of 16K and psize of 4K will be added to the 16K bin +in all three columns. +.It Fl -class Ns = Ns ( Li normal Ns | Ns Li special Ns | Ns Li dedup Ns | Ns Li other ) Ns Op , Ns … +When used with +.Fl bb , +only consider blocks from these allocation classes. .It Fl B , -backup Generate a backup stream, similar to .Nm zfs Cm send , diff --git a/sys/contrib/openzfs/man/man8/zpool-remove.8 b/sys/contrib/openzfs/man/man8/zpool-remove.8 index 4d5fc431d332..483d885f10fd 100644 --- a/sys/contrib/openzfs/man/man8/zpool-remove.8 +++ b/sys/contrib/openzfs/man/man8/zpool-remove.8 @@ -58,8 +58,8 @@ This command supports removing hot spare, cache, log, and both mirrored and non-redundant primary top-level vdevs, including dedup and special vdevs. .Pp Top-level vdevs can only be removed if the primary pool storage does not contain -a top-level raidz vdev, all top-level vdevs have the same sector size, and the -keys for all encrypted datasets are loaded. +a top-level raidz or draid vdev, all top-level vdevs have the same ashift size, +and the keys for all encrypted datasets are loaded. .Pp Removing a top-level vdev reduces the total amount of space in the storage pool. The specified device will be evacuated by copying all allocated space from it to diff --git a/sys/contrib/openzfs/module/icp/algs/sha2/sha2_generic.c b/sys/contrib/openzfs/module/icp/algs/sha2/sha2_generic.c index d0fcca798fa9..ad707341eec7 100644 --- a/sys/contrib/openzfs/module/icp/algs/sha2/sha2_generic.c +++ b/sys/contrib/openzfs/module/icp/algs/sha2/sha2_generic.c @@ -77,7 +77,8 @@ static const uint32_t SHA256_K[64] = { h = g, g = f, f = e, e = d + T1; \ d = c, c = b, b = a, a = T1 + T2; -static void sha256_generic(uint32_t state[8], const void *data, size_t num_blks) +static void +icp_sha256_generic(uint32_t state[8], const void *data, size_t num_blks) { uint64_t blk; @@ -173,7 +174,8 @@ static const uint64_t SHA512_K[80] = { 0x5fcb6fab3ad6faec, 0x6c44198c4a475817 }; -static void sha512_generic(uint64_t state[8], const void *data, size_t num_blks) +static void +icp_sha512_generic(uint64_t state[8], const void *data, size_t num_blks) { uint64_t blk; @@ -226,7 +228,8 @@ static void sha512_generic(uint64_t state[8], const void *data, size_t num_blks) } } -static void sha256_update(sha256_ctx *ctx, const uint8_t *data, size_t len) +static void +icp_sha256_update(sha256_ctx *ctx, const uint8_t *data, size_t len) { uint64_t pos = ctx->count[0]; uint64_t total = ctx->count[1]; @@ -258,7 +261,8 @@ static void sha256_update(sha256_ctx *ctx, const uint8_t *data, size_t len) ctx->count[1] = total; } -static void sha512_update(sha512_ctx *ctx, const uint8_t *data, size_t len) +static void +icp_sha512_update(sha512_ctx *ctx, const uint8_t *data, size_t len) { uint64_t pos = ctx->count[0]; uint64_t total = ctx->count[1]; @@ -290,7 +294,8 @@ static void sha512_update(sha512_ctx *ctx, const uint8_t *data, size_t len) ctx->count[1] = total; } -static void sha256_final(sha256_ctx *ctx, uint8_t *result, int bits) +static void +icp_sha256_final(sha256_ctx *ctx, uint8_t *result, int bits) { uint64_t mlen, pos = ctx->count[0]; uint8_t *m = ctx->wbuf; @@ -334,7 +339,8 @@ static void sha256_final(sha256_ctx *ctx, uint8_t *result, int bits) memset(ctx, 0, sizeof (*ctx)); } -static void sha512_final(sha512_ctx *ctx, uint8_t *result, int bits) +static void +icp_sha512_final(sha512_ctx *ctx, uint8_t *result, int bits) { uint64_t mlen, pos = ctx->count[0]; uint8_t *m = ctx->wbuf, *r; @@ -461,14 +467,14 @@ SHA2Update(SHA2_CTX *ctx, const void *data, size_t len) switch (ctx->algotype) { case SHA256: - sha256_update(&ctx->sha256, data, len); + icp_sha256_update(&ctx->sha256, data, len); break; case SHA512: case SHA512_HMAC_MECH_INFO_TYPE: - sha512_update(&ctx->sha512, data, len); + icp_sha512_update(&ctx->sha512, data, len); break; case SHA512_256: - sha512_update(&ctx->sha512, data, len); + icp_sha512_update(&ctx->sha512, data, len); break; } } @@ -479,32 +485,33 @@ SHA2Final(void *digest, SHA2_CTX *ctx) { switch (ctx->algotype) { case SHA256: - sha256_final(&ctx->sha256, digest, 256); + icp_sha256_final(&ctx->sha256, digest, 256); break; case SHA512: case SHA512_HMAC_MECH_INFO_TYPE: - sha512_final(&ctx->sha512, digest, 512); + icp_sha512_final(&ctx->sha512, digest, 512); break; case SHA512_256: - sha512_final(&ctx->sha512, digest, 256); + icp_sha512_final(&ctx->sha512, digest, 256); break; } } /* the generic implementation is always okay */ -static boolean_t sha2_is_supported(void) +static boolean_t +icp_sha2_is_supported(void) { return (B_TRUE); } const sha256_ops_t sha256_generic_impl = { .name = "generic", - .transform = sha256_generic, - .is_supported = sha2_is_supported + .transform = icp_sha256_generic, + .is_supported = icp_sha2_is_supported }; const sha512_ops_t sha512_generic_impl = { .name = "generic", - .transform = sha512_generic, - .is_supported = sha2_is_supported + .transform = icp_sha512_generic, + .is_supported = icp_sha2_is_supported }; diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c index 4de48e013ec4..d0a9c662e6f0 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c @@ -762,8 +762,7 @@ zfsctl_common_pathconf(struct vop_pathconf_args *ap) return (0); case _PC_MIN_HOLE_SIZE: - *ap->a_retval = (int)SPA_MINBLOCKSIZE; - return (0); + return (EINVAL); case _PC_ACL_EXTENDED: *ap->a_retval = 0; diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c index 411225786089..f34a2fd37a77 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c @@ -4116,6 +4116,7 @@ zfs_pathconf(vnode_t *vp, int cmd, ulong_t *valp, cred_t *cr, { znode_t *zp; zfsvfs_t *zfsvfs; + uint_t blksize, iosize; int error; switch (cmd) { @@ -4127,8 +4128,20 @@ zfs_pathconf(vnode_t *vp, int cmd, ulong_t *valp, cred_t *cr, *valp = 64; return (0); case _PC_MIN_HOLE_SIZE: - *valp = (int)SPA_MINBLOCKSIZE; - return (0); + iosize = vp->v_mount->mnt_stat.f_iosize; + if (vp->v_type == VREG) { + zp = VTOZ(vp); + blksize = zp->z_blksz; + if (zp->z_size <= blksize) + blksize = MAX(blksize, iosize); + *valp = (int)blksize; + return (0); + } + if (vp->v_type == VDIR) { + *valp = (int)iosize; + return (0); + } + return (EINVAL); case _PC_ACL_EXTENDED: #if 0 /* POSIX ACLs are not implemented for ZFS on FreeBSD yet. */ zp = VTOZ(vp); @@ -4210,8 +4223,20 @@ zfs_getpages(struct vnode *vp, vm_page_t *ma, int count, int *rbehind, zfs_vmobject_wlock(object); (void) vm_page_grab_pages(object, OFF_TO_IDX(start), - VM_ALLOC_NORMAL | VM_ALLOC_WAITOK | VM_ALLOC_ZERO, + VM_ALLOC_NORMAL | VM_ALLOC_WAITOK, ma, count); + if (!vm_page_all_valid(ma[count - 1])) { + /* + * Later in this function, we copy DMU data to + * invalid pages only. The last page may not be + * entirely filled though, if the file does not + * end on a page boundary. Therefore, we zero + * that last page here to make sure it does not + * contain garbage after the end of file. + */ + ASSERT(vm_page_none_valid(ma[count - 1])); + vm_page_zero_invalid(ma[count - 1], FALSE); + } zfs_vmobject_wunlock(object); } if (blksz == zp->z_blksz) diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zio_crypt.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zio_crypt.c index 91cf38016e00..8562c42b3220 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/zio_crypt.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zio_crypt.c @@ -437,6 +437,7 @@ zio_crypt_key_wrap(crypto_key_t *cwkey, zio_crypt_key_t *key, uint8_t *iv, ASSERT3U(crypt, <, ZIO_CRYPT_FUNCTIONS); + memset(&cuio_s, 0, sizeof (cuio_s)); zfs_uio_init(&cuio, &cuio_s); keydata_len = zio_crypt_table[crypt].ci_keylen; @@ -519,6 +520,7 @@ zio_crypt_key_unwrap(crypto_key_t *cwkey, uint64_t crypt, uint64_t version, keydata_len = zio_crypt_table[crypt].ci_keylen; rw_init(&key->zk_salt_lock, NULL, RW_DEFAULT, NULL); + memset(&cuio_s, 0, sizeof (cuio_s)); zfs_uio_init(&cuio, &cuio_s); /* diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c index 0dd2ecd7fd8d..3ddbfcb97184 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c @@ -183,6 +183,7 @@ static struct filterops zvol_filterops_vnode = { .f_isfd = 1, .f_detach = zvol_filter_detach, .f_event = zvol_filter_vnode, + .f_copy = knote_triv_copy, }; extern uint_t zfs_geom_probe_vdev_key; diff --git a/sys/contrib/openzfs/module/os/linux/spl/spl-zone.c b/sys/contrib/openzfs/module/os/linux/spl/spl-zone.c index 45c2999a4bb1..b2eae5d00b10 100644 --- a/sys/contrib/openzfs/module/os/linux/spl/spl-zone.c +++ b/sys/contrib/openzfs/module/os/linux/spl/spl-zone.c @@ -25,6 +25,10 @@ * SUCH DAMAGE. */ +/* + * Copyright (c) 2025, Rob Norris <robn@despairlabs.com> + */ + #include <sys/types.h> #include <sys/sysmacros.h> #include <sys/kmem.h> @@ -56,6 +60,19 @@ typedef struct zone_dataset { } zone_dataset_t; #ifdef CONFIG_USER_NS + +/* + * Linux 6.18 moved the generic namespace type away from ns->ops->type onto + * ns_common itself. + */ +#ifdef HAVE_NS_COMMON_TYPE +#define ns_is_newuser(ns) \ + ((ns)->ns_type == CLONE_NEWUSER) +#else +#define ns_is_newuser(ns) \ + ((ns)->ops != NULL && (ns)->ops->type == CLONE_NEWUSER) +#endif + /* * Returns: * - 0 on success @@ -84,7 +101,7 @@ user_ns_get(int fd, struct user_namespace **userns) goto done; } ns = get_proc_ns(file_inode(nsfile)); - if (ns->ops->type != CLONE_NEWUSER) { + if (!ns_is_newuser(ns)) { error = ENOTTY; goto done; } diff --git a/sys/contrib/openzfs/module/os/linux/zfs/abd_os.c b/sys/contrib/openzfs/module/os/linux/zfs/abd_os.c index 8a8316f63c48..18f2426fbbfc 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/abd_os.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/abd_os.c @@ -23,6 +23,7 @@ * Copyright (c) 2014 by Chunwei Chen. All rights reserved. * Copyright (c) 2019 by Delphix. All rights reserved. * Copyright (c) 2023, 2024, Klara Inc. + * Copyright (c) 2025, Rob Norris <robn@despairlabs.com> */ /* @@ -1109,6 +1110,14 @@ abd_return_buf_copy(abd_t *abd, void *buf, size_t n) #define ABD_ITER_PAGE_SIZE(page) (PAGESIZE) #endif +#ifndef nth_page +/* + * Since 6.18 nth_page() no longer exists, and is no longer required to iterate + * within a single SG entry, so we replace it with a simple addition. + */ +#define nth_page(p, n) ((p)+(n)) +#endif + void abd_iter_page(struct abd_iter *aiter) { diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zfs_acl.c b/sys/contrib/openzfs/module/os/linux/zfs/zfs_acl.c index daa4b5776837..934d74a112fd 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/zfs_acl.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/zfs_acl.c @@ -2524,7 +2524,7 @@ zfs_zaccess_common(znode_t *zp, uint32_t v4_mode, uint32_t *working_mode, * Also note: DOS R/O is ignored for directories. */ if ((v4_mode & WRITE_MASK_DATA) && - S_ISDIR(ZTOI(zp)->i_mode) && + !S_ISDIR(ZTOI(zp)->i_mode) && (zp->z_pflags & ZFS_READONLY)) { return (SET_ERROR(EPERM)); } diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c b/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c index 6106726651a3..e845ad69ad78 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c @@ -2033,10 +2033,7 @@ zfs_setattr(znode_t *zp, vattr_t *vap, int flags, cred_t *cr, zidmap_t *mnt_ns) goto out3; } - if ((mask & ATTR_SIZE) && (zp->z_pflags & ZFS_READONLY)) { - err = SET_ERROR(EPERM); - goto out3; - } + /* ZFS_READONLY will be handled in zfs_zaccess() */ /* * Verify timestamps doesn't overflow 32 bits. diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c b/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c index d07317b0d910..02965ac8cbee 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c @@ -23,6 +23,7 @@ * Copyright (c) 2011, Lawrence Livermore National Security, LLC. * Copyright (c) 2015 by Chunwei Chen. All rights reserved. * Copyright (c) 2025, Klara, Inc. + * Copyright (c) 2025, Rob Norris <robn@despairlabs.com> */ @@ -478,6 +479,7 @@ zpl_putpage(struct page *pp, struct writeback_control *wbc, void *data) return (ret); } +#ifdef HAVE_WRITE_CACHE_PAGES #ifdef HAVE_WRITEPAGE_T_FOLIO static int zpl_putfolio(struct folio *pp, struct writeback_control *wbc, void *data) @@ -499,6 +501,78 @@ zpl_write_cache_pages(struct address_space *mapping, #endif return (result); } +#else +static inline int +zpl_write_cache_pages(struct address_space *mapping, + struct writeback_control *wbc, void *data) +{ + pgoff_t start = wbc->range_start >> PAGE_SHIFT; + pgoff_t end = wbc->range_end >> PAGE_SHIFT; + + struct folio_batch fbatch; + folio_batch_init(&fbatch); + + /* + * This atomically (-ish) tags all DIRTY pages in the range with + * TOWRITE, allowing users to continue dirtying or undirtying pages + * while we get on with writeback, without us treading on each other. + */ + tag_pages_for_writeback(mapping, start, end); + + int err = 0; + unsigned int npages; + + /* + * Grab references to the TOWRITE pages just flagged. This may not get + * all of them, so we do it in a loop until there are none left. + */ + while ((npages = filemap_get_folios_tag(mapping, &start, end, + PAGECACHE_TAG_TOWRITE, &fbatch)) != 0) { + + /* Loop over each page and write it out. */ + struct folio *folio; + while ((folio = folio_batch_next(&fbatch)) != NULL) { + folio_lock(folio); + + /* + * If the folio has been remapped, or is no longer + * dirty, then there's nothing to do. + */ + if (folio->mapping != mapping || + !folio_test_dirty(folio)) { + folio_unlock(folio); + continue; + } + + /* + * If writeback is already in progress, wait for it to + * finish. We continue after this even if the page + * ends up clean; zfs_putpage() will skip it if no + * further work is required. + */ + while (folio_test_writeback(folio)) + folio_wait_bit(folio, PG_writeback); + + /* + * Write it out and collect any error. zfs_putpage() + * will clear the TOWRITE and DIRTY flags, and return + * with the page unlocked. + */ + int ferr = zpl_putpage(&folio->page, wbc, data); + if (err == 0 && ferr != 0) + err = ferr; + + /* Housekeeping for the caller. */ + wbc->nr_to_write -= folio_nr_pages(folio); + } + + /* Release any remaining references on the batch. */ + folio_batch_release(&fbatch); + } + + return (err); +} +#endif static int zpl_writepages(struct address_space *mapping, struct writeback_control *wbc) diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zpl_super.c b/sys/contrib/openzfs/module/os/linux/zfs/zpl_super.c index 444948d03cb3..347b352506e5 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/zpl_super.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/zpl_super.c @@ -23,6 +23,7 @@ * Copyright (c) 2011, Lawrence Livermore National Security, LLC. * Copyright (c) 2023, Datto Inc. All rights reserved. * Copyright (c) 2025, Klara, Inc. + * Copyright (c) 2025, Rob Norris <robn@despairlabs.com> */ @@ -33,6 +34,7 @@ #include <sys/zpl.h> #include <linux/iversion.h> #include <linux/version.h> +#include <linux/vfs_compat.h> /* * What to do when the last reference to an inode is released. If 0, the kernel @@ -104,7 +106,7 @@ zpl_dirty_inode(struct inode *ip, int flags) * reporting memory pressure and requests OpenZFS release some memory (see * zfs_prune()). * - * When set to 1, we call generic_delete_node(), which always returns "destroy + * When set to 1, we call generic_delete_inode(), which always returns "destroy * immediately", resulting in inodes being destroyed immediately, releasing * their associated dnodes and dbufs to the dbuf cached and the ARC to be * evicted as normal. diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c b/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c index 967a018640e1..fe939150b641 100644 --- a/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c +++ b/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c @@ -21,7 +21,7 @@ */ /* * Copyright (c) 2012, 2020 by Delphix. All rights reserved. - * Copyright (c) 2024, Rob Norris <robn@despairlabs.com> + * Copyright (c) 2024, 2025, Rob Norris <robn@despairlabs.com> * Copyright (c) 2024, 2025, Klara, Inc. */ @@ -337,16 +337,14 @@ 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; @@ -467,6 +465,24 @@ 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 @@ -486,27 +502,32 @@ 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; @@ -610,7 +631,7 @@ 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 { @@ -1011,12 +1032,12 @@ zvol_os_update_volsize(zvol_state_t *zv, uint64_t volsize) * tiny devices. For devices over 1 Mib a standard head and sector count * is used to keep the cylinders count reasonable. */ -static int -zvol_getgeo(struct block_device *bdev, struct hd_geometry *geo) +static inline int +zvol_getgeo_impl(struct gendisk *disk, struct hd_geometry *geo) { + zvol_state_t *zv = atomic_load_ptr(&disk->private_data); sector_t sectors; - zvol_state_t *zv = atomic_load_ptr(&bdev->bd_disk->private_data); ASSERT3P(zv, !=, NULL); ASSERT3U(zv->zv_open_count, >, 0); @@ -1036,6 +1057,20 @@ zvol_getgeo(struct block_device *bdev, struct hd_geometry *geo) return (0); } +#ifdef HAVE_BLOCK_DEVICE_OPERATIONS_GETGEO_GENDISK +static int +zvol_getgeo(struct gendisk *disk, struct hd_geometry *geo) +{ + return (zvol_getgeo_impl(disk, geo)); +} +#else +static int +zvol_getgeo(struct block_device *bdev, struct hd_geometry *geo) +{ + return (zvol_getgeo_impl(bdev->bd_disk, geo)); +} +#endif + /* * Why have two separate block_device_operations structs? * @@ -1479,7 +1514,7 @@ zvol_os_remove_minor(zvol_state_t *zv) if (zso->use_blk_mq) blk_mq_free_tag_set(&zso->tag_set); - ida_simple_remove(&zvol_ida, MINOR(zso->zvo_dev) >> ZVOL_MINOR_BITS); + ida_free(&zvol_ida, MINOR(zso->zvo_dev) >> ZVOL_MINOR_BITS); kmem_free(zso, sizeof (struct zvol_state_os)); @@ -1634,7 +1669,7 @@ zvol_os_create_minor(const char *name) if (zvol_inhibit_dev) return (0); - idx = ida_simple_get(&zvol_ida, 0, 0, kmem_flags_convert(KM_SLEEP)); + idx = ida_alloc(&zvol_ida, kmem_flags_convert(KM_SLEEP)); if (idx < 0) return (SET_ERROR(-idx)); minor = idx << ZVOL_MINOR_BITS; @@ -1642,7 +1677,7 @@ zvol_os_create_minor(const char *name) /* too many partitions can cause an overflow */ zfs_dbgmsg("zvol: create minor overflow: %s, minor %u/%u", name, minor, MINOR(minor)); - ida_simple_remove(&zvol_ida, idx); + ida_free(&zvol_ida, idx); return (SET_ERROR(EINVAL)); } @@ -1650,7 +1685,7 @@ zvol_os_create_minor(const char *name) if (zv) { ASSERT(MUTEX_HELD(&zv->zv_state_lock)); mutex_exit(&zv->zv_state_lock); - ida_simple_remove(&zvol_ida, idx); + ida_free(&zvol_ida, idx); return (SET_ERROR(EEXIST)); } @@ -1750,7 +1785,7 @@ out_doi: rw_exit(&zvol_state_lock); error = zvol_os_add_disk(zv->zv_zso->zvo_disk); } else { - ida_simple_remove(&zvol_ida, idx); + ida_free(&zvol_ida, idx); } return (error); diff --git a/sys/contrib/openzfs/module/zfs/arc.c b/sys/contrib/openzfs/module/zfs/arc.c index b677f90280d7..dbb5e942e2e6 100644 --- a/sys/contrib/openzfs/module/zfs/arc.c +++ b/sys/contrib/openzfs/module/zfs/arc.c @@ -1157,7 +1157,7 @@ buf_fini(void) #if defined(_KERNEL) /* * Large allocations which do not require contiguous pages - * should be using vmem_free() in the linux kernel\ + * should be using vmem_free() in the linux kernel. */ vmem_free(buf_hash_table.ht_table, (buf_hash_table.ht_mask + 1) * sizeof (void *)); @@ -4651,10 +4651,10 @@ arc_flush_task(void *arg) arc_flush_impl(spa_guid, B_FALSE); arc_async_flush_remove(spa_guid, af->af_cache_level); - uint64_t elaspsed = NSEC2MSEC(gethrtime() - start_time); - if (elaspsed > 0) { + uint64_t elapsed = NSEC2MSEC(gethrtime() - start_time); + if (elapsed > 0) { zfs_dbgmsg("spa %llu arc flushed in %llu ms", - (u_longlong_t)spa_guid, (u_longlong_t)elaspsed); + (u_longlong_t)spa_guid, (u_longlong_t)elapsed); } } @@ -9152,7 +9152,7 @@ top: if (dev->l2ad_first) { /* * This is the first sweep through the device. There is - * nothing to evict. We have already trimmmed the + * nothing to evict. We have already trimmed the * whole device. */ goto out; @@ -10086,12 +10086,12 @@ l2arc_device_teardown(void *arg) kmem_free(remdev->l2ad_dev_hdr, remdev->l2ad_dev_hdr_asize); vmem_free(remdev, sizeof (l2arc_dev_t)); - uint64_t elaspsed = NSEC2MSEC(gethrtime() - start_time); - if (elaspsed > 0) { + uint64_t elapsed = NSEC2MSEC(gethrtime() - start_time); + if (elapsed > 0) { zfs_dbgmsg("spa %llu, vdev %llu removed in %llu ms", (u_longlong_t)rva->rva_spa_gid, (u_longlong_t)rva->rva_vdev_gid, - (u_longlong_t)elaspsed); + (u_longlong_t)elapsed); } if (rva->rva_async) diff --git a/sys/contrib/openzfs/module/zfs/vdev_removal.c b/sys/contrib/openzfs/module/zfs/vdev_removal.c index 2f7a739da241..abb71543e3ab 100644 --- a/sys/contrib/openzfs/module/zfs/vdev_removal.c +++ b/sys/contrib/openzfs/module/zfs/vdev_removal.c @@ -51,34 +51,70 @@ #include <sys/trace_zfs.h> /* - * This file contains the necessary logic to remove vdevs from a - * storage pool. Currently, the only devices that can be removed - * are log, cache, and spare devices; and top level vdevs from a pool - * w/o raidz or mirrors. (Note that members of a mirror can be removed - * by the detach operation.) + * This file contains the necessary logic to remove vdevs from a storage + * pool. Note that members of a mirror can be removed by the detach + * operation. Currently, the only devices that can be removed are: * - * Log vdevs are removed by evacuating them and then turning the vdev - * into a hole vdev while holding spa config locks. + * 1) Traditional hot spare and cache vdevs. Note that draid distributed + * spares are fixed at creation time and cannot be removed. * - * Top level vdevs are removed and converted into an indirect vdev via - * a multi-step process: + * 2) Log vdevs are removed by evacuating them and then turning the vdev + * into a hole vdev while holding spa config locks. * - * - Disable allocations from this device (spa_vdev_remove_top). + * 3) Top-level singleton and mirror vdevs, including dedup and special + * vdevs, are removed and converted into an indirect vdev via a + * multi-step process: * - * - From a new thread (spa_vdev_remove_thread), copy data from - * the removing vdev to a different vdev. The copy happens in open - * context (spa_vdev_copy_impl) and issues a sync task - * (vdev_mapping_sync) so the sync thread can update the partial - * indirect mappings in core and on disk. + * - Disable allocations from this device (spa_vdev_remove_top). * - * - If a free happens during a removal, it is freed from the - * removing vdev, and if it has already been copied, from the new - * location as well (free_from_removing_vdev). + * - From a new thread (spa_vdev_remove_thread), copy data from the + * removing vdev to a different vdev. The copy happens in open context + * (spa_vdev_copy_impl) and issues a sync task (vdev_mapping_sync) so + * the sync thread can update the partial indirect mappings in core + * and on disk. * - * - After the removal is completed, the copy thread converts the vdev - * into an indirect vdev (vdev_remove_complete) before instructing - * the sync thread to destroy the space maps and finish the removal - * (spa_finish_removal). + * - If a free happens during a removal, it is freed from the removing + * vdev, and if it has already been copied, from the new location as + * well (free_from_removing_vdev). + * + * - After the removal is completed, the copy thread converts the vdev + * into an indirect vdev (vdev_remove_complete) before instructing + * the sync thread to destroy the space maps and finish the removal + * (spa_finish_removal). + * + * The following constraints currently apply primary device removal: + * + * - All vdevs must be online, healthy, and not be missing any data + * according to the DTLs. + * + * - When removing a singleton or mirror vdev, regardless of it's a + * special, dedup, or primary device, it must have the same ashift + * as the devices in the normal allocation class. Furthermore, all + * vdevs in the normal allocation class must have the same ashift to + * ensure the new allocations never includes additional padding. + * + * - The normal allocation class cannot contain any raidz or draid + * top-level vdevs since segments are copied without regard for block + * boundaries. This makes it impossible to calculate the required + * parity columns when using these vdev types as the destination. + * + * - The encryption keys must be loaded so the ZIL logs can be reset + * in order to prevent writing to the device being removed. + * + * N.B. ashift and raidz/draid constraints for primary top-level device + * removal could be slightly relaxed if it were possible to request that + * DVAs from a mirror or singleton in the specified allocation class be + * used (metaslab_alloc_dva). + * + * This flexibility would be particularly useful for raidz/draid pools which + * often include a mirrored special device. If a mistakenly added top-level + * singleton were added it could then still be removed at the cost of some + * special device capacity. This may be a worthwhile tradeoff depending on + * the pool capacity and expense (cost, complexity, time) of creating a new + * pool and copying all of the data to correct the configuration. + * + * Furthermore, while not currently supported it should be possible to allow + * vdevs of any type to be removed as long as they've never been written to. */ typedef struct vdev_copy_arg { diff --git a/sys/contrib/openzfs/module/zfs/zvol.c b/sys/contrib/openzfs/module/zfs/zvol.c index faced0db7e9e..00f98168d3d8 100644 --- a/sys/contrib/openzfs/module/zfs/zvol.c +++ b/sys/contrib/openzfs/module/zfs/zvol.c @@ -410,7 +410,7 @@ zvol_set_volthreading(const char *name, boolean_t value) { zvol_state_t *zv = zvol_find_by_name(name, RW_NONE); if (zv == NULL) - return (SET_ERROR(ENOENT)); + return (-1); zv->zv_threading = value; mutex_exit(&zv->zv_state_lock); return (0); diff --git a/sys/contrib/openzfs/scripts/zfs-tests.sh b/sys/contrib/openzfs/scripts/zfs-tests.sh index 5a0a1a609448..09a15bafc27e 100755 --- a/sys/contrib/openzfs/scripts/zfs-tests.sh +++ b/sys/contrib/openzfs/scripts/zfs-tests.sh @@ -797,6 +797,10 @@ msg "${TEST_RUNNER}" \ 2>&1; echo $? >"$REPORT_FILE"; } | tee "$RESULTS_FILE" read -r RUNRESULT <"$REPORT_FILE" +if [[ "$RUNRESULT" -eq "255" ]] ; then + fail "$TEST_RUNNER failed, test aborted." +fi + # # Analyze the results. # diff --git a/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in b/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in index d2c1185e4a94..6688b6c4beb6 100755 --- a/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in +++ b/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in @@ -25,6 +25,7 @@ import sys import ctypes import re import configparser +import traceback from datetime import datetime from optparse import OptionParser @@ -1138,7 +1139,7 @@ def filter_tests(testrun, options): testrun.filter(failed) -def fail(retstr, ret=1): +def fail(retstr, ret=255): print('%s: %s' % (sys.argv[0], retstr)) exit(ret) @@ -1247,23 +1248,27 @@ def parse_args(): def main(): options = parse_args() - testrun = TestRun(options) + try: + testrun = TestRun(options) - if options.runfiles: - testrun.read(options) - else: - find_tests(testrun, options) + if options.runfiles: + testrun.read(options) + else: + find_tests(testrun, options) + + if options.logfile: + filter_tests(testrun, options) - if options.logfile: - filter_tests(testrun, options) + if options.template: + testrun.write(options) + exit(0) - if options.template: - testrun.write(options) - exit(0) + testrun.complete_outputdirs() + testrun.run(options) + exit(testrun.summary()) - testrun.complete_outputdirs() - testrun.run(options) - exit(testrun.summary()) + except Exception: + fail("Uncaught exception in test runner:\n" + traceback.format_exc()) if __name__ == '__main__': diff --git a/sys/contrib/openzfs/tests/zfs-tests/cmd/mmap_seek.c b/sys/contrib/openzfs/tests/zfs-tests/cmd/mmap_seek.c index 45ba17e36c35..f46980cad111 100644 --- a/sys/contrib/openzfs/tests/zfs-tests/cmd/mmap_seek.c +++ b/sys/contrib/openzfs/tests/zfs-tests/cmd/mmap_seek.c @@ -47,25 +47,34 @@ #endif static void +seek_expect(int fd, off_t offset, int whence, off_t expect_offset) +{ + errno = 0; + off_t seek_offset = lseek(fd, offset, whence); + if (seek_offset == expect_offset) + return; + + int err = errno; + fprintf(stderr, "lseek(fd, %ld, SEEK_%s) = %ld (expected %ld)", + offset, (whence == SEEK_DATA ? "DATA" : "HOLE"), + seek_offset, expect_offset); + if (err != 0) + fprintf(stderr, " (errno %d [%s])\n", err, strerror(err)); + else + fputc('\n', stderr); + exit(2); +} + +static inline void seek_data(int fd, off_t offset, off_t expected) { - off_t data_offset = lseek(fd, offset, SEEK_DATA); - if (data_offset != expected) { - fprintf(stderr, "lseek(fd, %d, SEEK_DATA) = %d (expected %d)\n", - (int)offset, (int)data_offset, (int)expected); - exit(2); - } + seek_expect(fd, offset, SEEK_DATA, expected); } -static void +static inline void seek_hole(int fd, off_t offset, off_t expected) { - off_t hole_offset = lseek(fd, offset, SEEK_HOLE); - if (hole_offset != expected) { - fprintf(stderr, "lseek(fd, %d, SEEK_HOLE) = %d (expected %d)\n", - (int)offset, (int)hole_offset, (int)expected); - exit(2); - } + seek_expect(fd, offset, SEEK_HOLE, expected); } int diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh index 2bd9f616170b..1a9774331ba1 100755 --- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh +++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zpool_reopen/zpool_reopen_004_pos.ksh @@ -47,6 +47,8 @@ function cleanup # bring back removed disk online for further tests insert_disk $REMOVED_DISK $scsi_host poolexists $TESTPOOL && destroy_pool $TESTPOOL + # Since the disk was offline during destroy, remove the label + zpool labelclear $DISK2 -f } log_assert "Testing zpool reopen with pool name as argument" diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh index bb697b76a9f3..9de308c4a11a 100755 --- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh +++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/zvol/zvol_misc/zvol_misc_trim.ksh @@ -41,6 +41,7 @@ # 5. TRIM the first 1MB and last 2MB of the 5MB block of data. # 6. Observe 2MB of used space on the zvol # 7. Verify the trimmed regions are zero'd on the zvol +# 8. Verify Secure Erase does not work on zvols (Linux only) verify_runnable "global" @@ -56,6 +57,7 @@ if is_linux ; then else trimcmd='blkdiscard' fi + secure_trimcmd="$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" |