| Message ID | 20230215143109.2721722-7-johannes.thumshirn@wdc.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | btrfs-progs: add support for RAID stripe tree | expand |
On 15/02/2023 22:31, Johannes Thumshirn wrote: > Allow for RAID levels 0, 1 and 10 on zoned devices if the RAID stripe tree > is used. > > Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> > --- > common/fsfeatures.c | 8 +++++ > kernel-shared/ctree.h | 3 +- > kernel-shared/zoned.c | 35 ++++++++++++++++-- > kernel-shared/zoned.h | 4 +-- > mkfs/main.c | 82 ++++++++++++++++++++++++++++++++++++++++--- > 5 files changed, 122 insertions(+), 10 deletions(-) > > diff --git a/common/fsfeatures.c b/common/fsfeatures.c > index 169e47e92582..d18226f43d7c 100644 > --- a/common/fsfeatures.c > +++ b/common/fsfeatures.c > @@ -195,6 +195,14 @@ static const struct btrfs_feature mkfs_features[] = { > VERSION_NULL(safe), > VERSION_NULL(default), > .desc = "new extent tree format" > + } , { > + .name = "raid-stripe-tree", > + .incompat_flag = BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE, > + .sysfs_name = NULL, > + VERSION_NULL(compat), > + VERSION_NULL(safe), > + VERSION_NULL(default), > + .desc = "raid stripe tree" > }, > #endif > /* Keep this one last */ > diff --git a/kernel-shared/ctree.h b/kernel-shared/ctree.h > index a9bb6eb39752..0fb5014d0696 100644 > --- a/kernel-shared/ctree.h > +++ b/kernel-shared/ctree.h > @@ -560,7 +560,8 @@ BUILD_ASSERT(sizeof(struct btrfs_super_block) == BTRFS_SUPER_INFO_SIZE); > BTRFS_FEATURE_INCOMPAT_RAID1C34 | \ > BTRFS_FEATURE_INCOMPAT_METADATA_UUID | \ > BTRFS_FEATURE_INCOMPAT_ZONED | \ > - BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2) > + BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2 | \ > + BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE) > #else > #define BTRFS_FEATURE_INCOMPAT_SUPP \ > (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \ > diff --git a/kernel-shared/zoned.c b/kernel-shared/zoned.c > index a79fc6a5dbc3..4f07384817aa 100644 > --- a/kernel-shared/zoned.c > +++ b/kernel-shared/zoned.c > @@ -795,7 +795,7 @@ out: > return ret; > } > > -bool zoned_profile_supported(u64 map_type) > +bool zoned_profile_supported(u64 map_type, bool rst) > { > bool data = (map_type & BTRFS_BLOCK_GROUP_DATA); > u64 flags = (map_type & BTRFS_BLOCK_GROUP_PROFILE_MASK); > @@ -804,9 +804,37 @@ bool zoned_profile_supported(u64 map_type) > if (flags == 0) > return true; > > - /* We can support DUP on metadata */ > +#if EXPERIMENTAL > + if (data) { > + if ((flags & BTRFS_BLOCK_GROUP_DUP) && rst) > + return true; > + /* Data RAID1 needs a raid-stripe-tree */ > + if ((flags & BTRFS_BLOCK_GROUP_RAID1_MASK) && rst) > + return true; > + /* Data RAID0 needs a raid-stripe-tree */ > + if ((flags & BTRFS_BLOCK_GROUP_RAID0) && rst) > + return true; > + /* Data RAID10 needs a raid-stripe-tree */ > + if ((flags & BTRFS_BLOCK_GROUP_RAID10) && rst) > + return true; > + } else { > + /* We can support DUP on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_DUP) > + return true; > + /* We can support RAID1 on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_RAID1_MASK) > + return true; > + /* We can support RAID0 on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_RAID0) > + return true; > + /* We can support RAID10 on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_RAID10) > + return true; Do we really require RST to support RAID1x, RAID0, and RAID10? If not, we can potentially enable it separately from the RST patchset to test its functionality. > + } > +#else > if (!data && (flags & BTRFS_BLOCK_GROUP_DUP)) > return true; > +#endif > > /* All other profiles are not supported yet */ > return false; > @@ -923,7 +951,8 @@ int btrfs_load_block_group_zone_info(struct btrfs_fs_info *fs_info, > } > } > > - if (!zoned_profile_supported(map->type)) { > + if (!zoned_profile_supported(map->type, > + btrfs_fs_incompat(fs_info, RAID_STRIPE_TREE))) { > error("zoned: profile %s not yet supported", > btrfs_group_profile_str(map->type)); > ret = -EINVAL; > diff --git a/kernel-shared/zoned.h b/kernel-shared/zoned.h > index cc0d6b6f166d..c56788bcf07b 100644 > --- a/kernel-shared/zoned.h > +++ b/kernel-shared/zoned.h > @@ -132,7 +132,7 @@ static inline bool btrfs_dev_is_empty_zone(struct btrfs_device *device, u64 pos) > return zinfo->zones[zno].cond == BLK_ZONE_COND_EMPTY; > } > > -bool zoned_profile_supported(u64 map_type); > +bool zoned_profile_supported(u64 map_type, bool rst); > int btrfs_reset_dev_zone(int fd, struct blk_zone *zone); > u64 btrfs_find_allocatable_zones(struct btrfs_device *device, u64 hole_start, > u64 hole_end, u64 num_bytes); > @@ -213,7 +213,7 @@ static inline int btrfs_wipe_temporary_sb(struct btrfs_fs_devices *fs_devices) > return 0; > } > > -static inline bool zoned_profile_supported(u64 map_type) > +static inline bool zoned_profile_supported(u64 map_type, bool rst) > { > return false; > } > diff --git a/mkfs/main.c b/mkfs/main.c > index f4ff2c58a81f..a8a370d1c4df 100644 > --- a/mkfs/main.c > +++ b/mkfs/main.c > @@ -958,6 +958,38 @@ fail: > return ret; > } > > +static int setup_raid_stripe_tree_root(struct btrfs_fs_info *fs_info) > +{ > + struct btrfs_trans_handle *trans; > + struct btrfs_root *stripe_root; > + struct btrfs_key key = { > + .objectid = BTRFS_RAID_STRIPE_TREE_OBJECTID, > + .type = BTRFS_ROOT_ITEM_KEY, > + }; > + int ret; > + > + trans = btrfs_start_transaction(fs_info->tree_root, 0); > + if (IS_ERR(trans)) > + return PTR_ERR(trans); > + > + stripe_root = btrfs_create_tree(trans, fs_info, &key); > + if (IS_ERR(stripe_root)) { > + ret = PTR_ERR(stripe_root); > + goto abort; > + } > + fs_info->stripe_root = stripe_root; > + add_root_to_dirty_list(stripe_root); > + > + ret = btrfs_commit_transaction(trans, fs_info->tree_root); > + if (ret) > + return ret; > + > + return 0; > +abort: > + btrfs_abort_transaction(trans, ret); > + return ret; > +} > + > /* Thread callback for device preparation */ > static void *prepare_one_device(void *ctx) > { > @@ -1356,10 +1388,12 @@ int BOX_MAIN(mkfs)(int argc, char **argv) > exit(1); > } > > +#if !defined(EXPERIMENTAL) > if (features.incompat_flags & BTRFS_FEATURE_INCOMPAT_RAID56) { > error("cannot enable RAID5/6 in zoned mode"); > exit(1); > } > +#endif From my understanding of zoned_profile_supported() changes above, it seems that RST does not currently support RAID56? If so, I am curious as to why we are permitting RAID56 in zoned mode here if EXPERIMENTAL is defined. > } > > if (btrfs_check_nodesize(nodesize, sectorsize, &features)) > @@ -1464,10 +1498,40 @@ int BOX_MAIN(mkfs)(int argc, char **argv) > if (ret) > goto error; > > - if (opt_zoned && (!zoned_profile_supported(BTRFS_BLOCK_GROUP_METADATA | metadata_profile) || > - !zoned_profile_supported(BTRFS_BLOCK_GROUP_DATA | data_profile))) { > - error("zoned mode does not yet support RAID/DUP profiles, please specify '-d single -m single' manually"); > - goto error; > +#if EXPERIMENTAL > + if (opt_zoned && device_count) { > + switch (data_profile & BTRFS_BLOCK_GROUP_PROFILE_MASK) { > + case BTRFS_BLOCK_GROUP_DUP: > + case BTRFS_BLOCK_GROUP_RAID1: > + case BTRFS_BLOCK_GROUP_RAID1C3: > + case BTRFS_BLOCK_GROUP_RAID1C4: > + case BTRFS_BLOCK_GROUP_RAID0: > + case BTRFS_BLOCK_GROUP_RAID10: > + case BTRFS_BLOCK_GROUP_RAID56_MASK: > + features.incompat_flags |= > + BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE; > + break; > + default: > + break; > + } > + } > +#endif This appears to be the primary control to activate RST. It seems at other changes where "#if EXPERIMENTAL" is used, it may be unnecessary. Instead, it would be better to enable the code anyway to test with BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE unset. Where possible. > + > + if (opt_zoned) { > + u64 metadata = BTRFS_BLOCK_GROUP_METADATA | metadata_profile; > + u64 data = BTRFS_BLOCK_GROUP_DATA | data_profile; > +#if EXPERIMENTAL > + bool rst = features.incompat_flags & > + BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE; > +#else > + bool rst = false; > +#endif For example: We can remove the #if EXPERIMENTAL and delete the else part here, that is bool rst = features.incompat_flags & BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE; because BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE will not be set unless EXPERIMENTAL is enabled. Thanks, Anand > + > + if (!zoned_profile_supported(metadata, rst) || > + !zoned_profile_supported(data, rst)) { > + error("zoned mode does not yet support RAID/DUP profiles, please specify '-d single -m single' manually"); > + goto error; > + } > } > > t_prepare = calloc(device_count, sizeof(*t_prepare)); > @@ -1755,6 +1819,16 @@ raid_groups: > goto out; > } > } > + if (features.incompat_flags & BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE) { > + ret = setup_raid_stripe_tree_root(fs_info); > + if (ret < 0) { > + error("failed to initialize raid-stripe-tree: %d (%m)", > + ret); > + goto out; > + } > + } > + > + > if (bconf.verbose) { > char features_buf[BTRFS_FEATURE_STRING_BUF_SIZE]; >
On 09.03.23 02:14, Anand Jain wrote: >> +#if EXPERIMENTAL >> + if (data) { >> + if ((flags & BTRFS_BLOCK_GROUP_DUP) && rst) >> + return true; >> + /* Data RAID1 needs a raid-stripe-tree */ >> + if ((flags & BTRFS_BLOCK_GROUP_RAID1_MASK) && rst) >> + return true; >> + /* Data RAID0 needs a raid-stripe-tree */ >> + if ((flags & BTRFS_BLOCK_GROUP_RAID0) && rst) >> + return true; >> + /* Data RAID10 needs a raid-stripe-tree */ >> + if ((flags & BTRFS_BLOCK_GROUP_RAID10) && rst) >> + return true; >> + } else { >> + /* We can support DUP on metadata/system */ >> + if (flags & BTRFS_BLOCK_GROUP_DUP) >> + return true; >> + /* We can support RAID1 on metadata/system */ >> + if (flags & BTRFS_BLOCK_GROUP_RAID1_MASK) >> + return true; >> + /* We can support RAID0 on metadata/system */ >> + if (flags & BTRFS_BLOCK_GROUP_RAID0) >> + return true; >> + /* We can support RAID10 on metadata/system */ >> + if (flags & BTRFS_BLOCK_GROUP_RAID10) >> + return true; > Do we really require RST to support RAID1x, RAID0, and RAID10? > If not, we can potentially enable it separately from the RST > patchset to test its functionality. > > For data yes, for metadata no. But I don't really want to separate both.
On Wed, Feb 15, 2023 at 06:31:09AM -0800, Johannes Thumshirn wrote: > Allow for RAID levels 0, 1 and 10 on zoned devices if the RAID stripe tree > is used. > > Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> > --- > +#if EXPERIMENTAL > + if (data) { > + if ((flags & BTRFS_BLOCK_GROUP_DUP) && rst) > + return true; > + /* Data RAID1 needs a raid-stripe-tree */ > + if ((flags & BTRFS_BLOCK_GROUP_RAID1_MASK) && rst) > + return true; > + /* Data RAID0 needs a raid-stripe-tree */ > + if ((flags & BTRFS_BLOCK_GROUP_RAID0) && rst) > + return true; > + /* Data RAID10 needs a raid-stripe-tree */ > + if ((flags & BTRFS_BLOCK_GROUP_RAID10) && rst) > + return true; > + } else { > + /* We can support DUP on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_DUP) > + return true; > + /* We can support RAID1 on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_RAID1_MASK) > + return true; > + /* We can support RAID0 on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_RAID0) > + return true; > + /* We can support RAID10 on metadata/system */ > + if (flags & BTRFS_BLOCK_GROUP_RAID10) > + return true; > + } Please turn the above to a bitmask. > +#else > if (!data && (flags & BTRFS_BLOCK_GROUP_DUP)) > return true; > +#endif > > /* All other profiles are not supported yet */ > return false; > +#if !defined(EXPERIMENTAL) I've noticed that though it's not changed in this patch, such use of the macro is wrong as EXPERIMENTAL is always defined and must be checked for 0 or 1. > if (features.incompat_flags & BTRFS_FEATURE_INCOMPAT_RAID56) { > error("cannot enable RAID5/6 in zoned mode"); > exit(1); > } > +#endif > }
On 09.03.23 19:45, David Sterba wrote: > I've noticed that though it's not changed in this patch, such use of the > macro is wrong as EXPERIMENTAL is always defined and must be checked for > 0 or 1. Ah damn, ok will do
diff --git a/common/fsfeatures.c b/common/fsfeatures.c index 169e47e92582..d18226f43d7c 100644 --- a/common/fsfeatures.c +++ b/common/fsfeatures.c @@ -195,6 +195,14 @@ static const struct btrfs_feature mkfs_features[] = { VERSION_NULL(safe), VERSION_NULL(default), .desc = "new extent tree format" + } , { + .name = "raid-stripe-tree", + .incompat_flag = BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE, + .sysfs_name = NULL, + VERSION_NULL(compat), + VERSION_NULL(safe), + VERSION_NULL(default), + .desc = "raid stripe tree" }, #endif /* Keep this one last */ diff --git a/kernel-shared/ctree.h b/kernel-shared/ctree.h index a9bb6eb39752..0fb5014d0696 100644 --- a/kernel-shared/ctree.h +++ b/kernel-shared/ctree.h @@ -560,7 +560,8 @@ BUILD_ASSERT(sizeof(struct btrfs_super_block) == BTRFS_SUPER_INFO_SIZE); BTRFS_FEATURE_INCOMPAT_RAID1C34 | \ BTRFS_FEATURE_INCOMPAT_METADATA_UUID | \ BTRFS_FEATURE_INCOMPAT_ZONED | \ - BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2) + BTRFS_FEATURE_INCOMPAT_EXTENT_TREE_V2 | \ + BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE) #else #define BTRFS_FEATURE_INCOMPAT_SUPP \ (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \ diff --git a/kernel-shared/zoned.c b/kernel-shared/zoned.c index a79fc6a5dbc3..4f07384817aa 100644 --- a/kernel-shared/zoned.c +++ b/kernel-shared/zoned.c @@ -795,7 +795,7 @@ out: return ret; } -bool zoned_profile_supported(u64 map_type) +bool zoned_profile_supported(u64 map_type, bool rst) { bool data = (map_type & BTRFS_BLOCK_GROUP_DATA); u64 flags = (map_type & BTRFS_BLOCK_GROUP_PROFILE_MASK); @@ -804,9 +804,37 @@ bool zoned_profile_supported(u64 map_type) if (flags == 0) return true; - /* We can support DUP on metadata */ +#if EXPERIMENTAL + if (data) { + if ((flags & BTRFS_BLOCK_GROUP_DUP) && rst) + return true; + /* Data RAID1 needs a raid-stripe-tree */ + if ((flags & BTRFS_BLOCK_GROUP_RAID1_MASK) && rst) + return true; + /* Data RAID0 needs a raid-stripe-tree */ + if ((flags & BTRFS_BLOCK_GROUP_RAID0) && rst) + return true; + /* Data RAID10 needs a raid-stripe-tree */ + if ((flags & BTRFS_BLOCK_GROUP_RAID10) && rst) + return true; + } else { + /* We can support DUP on metadata/system */ + if (flags & BTRFS_BLOCK_GROUP_DUP) + return true; + /* We can support RAID1 on metadata/system */ + if (flags & BTRFS_BLOCK_GROUP_RAID1_MASK) + return true; + /* We can support RAID0 on metadata/system */ + if (flags & BTRFS_BLOCK_GROUP_RAID0) + return true; + /* We can support RAID10 on metadata/system */ + if (flags & BTRFS_BLOCK_GROUP_RAID10) + return true; + } +#else if (!data && (flags & BTRFS_BLOCK_GROUP_DUP)) return true; +#endif /* All other profiles are not supported yet */ return false; @@ -923,7 +951,8 @@ int btrfs_load_block_group_zone_info(struct btrfs_fs_info *fs_info, } } - if (!zoned_profile_supported(map->type)) { + if (!zoned_profile_supported(map->type, + btrfs_fs_incompat(fs_info, RAID_STRIPE_TREE))) { error("zoned: profile %s not yet supported", btrfs_group_profile_str(map->type)); ret = -EINVAL; diff --git a/kernel-shared/zoned.h b/kernel-shared/zoned.h index cc0d6b6f166d..c56788bcf07b 100644 --- a/kernel-shared/zoned.h +++ b/kernel-shared/zoned.h @@ -132,7 +132,7 @@ static inline bool btrfs_dev_is_empty_zone(struct btrfs_device *device, u64 pos) return zinfo->zones[zno].cond == BLK_ZONE_COND_EMPTY; } -bool zoned_profile_supported(u64 map_type); +bool zoned_profile_supported(u64 map_type, bool rst); int btrfs_reset_dev_zone(int fd, struct blk_zone *zone); u64 btrfs_find_allocatable_zones(struct btrfs_device *device, u64 hole_start, u64 hole_end, u64 num_bytes); @@ -213,7 +213,7 @@ static inline int btrfs_wipe_temporary_sb(struct btrfs_fs_devices *fs_devices) return 0; } -static inline bool zoned_profile_supported(u64 map_type) +static inline bool zoned_profile_supported(u64 map_type, bool rst) { return false; } diff --git a/mkfs/main.c b/mkfs/main.c index f4ff2c58a81f..a8a370d1c4df 100644 --- a/mkfs/main.c +++ b/mkfs/main.c @@ -958,6 +958,38 @@ fail: return ret; } +static int setup_raid_stripe_tree_root(struct btrfs_fs_info *fs_info) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *stripe_root; + struct btrfs_key key = { + .objectid = BTRFS_RAID_STRIPE_TREE_OBJECTID, + .type = BTRFS_ROOT_ITEM_KEY, + }; + int ret; + + trans = btrfs_start_transaction(fs_info->tree_root, 0); + if (IS_ERR(trans)) + return PTR_ERR(trans); + + stripe_root = btrfs_create_tree(trans, fs_info, &key); + if (IS_ERR(stripe_root)) { + ret = PTR_ERR(stripe_root); + goto abort; + } + fs_info->stripe_root = stripe_root; + add_root_to_dirty_list(stripe_root); + + ret = btrfs_commit_transaction(trans, fs_info->tree_root); + if (ret) + return ret; + + return 0; +abort: + btrfs_abort_transaction(trans, ret); + return ret; +} + /* Thread callback for device preparation */ static void *prepare_one_device(void *ctx) { @@ -1356,10 +1388,12 @@ int BOX_MAIN(mkfs)(int argc, char **argv) exit(1); } +#if !defined(EXPERIMENTAL) if (features.incompat_flags & BTRFS_FEATURE_INCOMPAT_RAID56) { error("cannot enable RAID5/6 in zoned mode"); exit(1); } +#endif } if (btrfs_check_nodesize(nodesize, sectorsize, &features)) @@ -1464,10 +1498,40 @@ int BOX_MAIN(mkfs)(int argc, char **argv) if (ret) goto error; - if (opt_zoned && (!zoned_profile_supported(BTRFS_BLOCK_GROUP_METADATA | metadata_profile) || - !zoned_profile_supported(BTRFS_BLOCK_GROUP_DATA | data_profile))) { - error("zoned mode does not yet support RAID/DUP profiles, please specify '-d single -m single' manually"); - goto error; +#if EXPERIMENTAL + if (opt_zoned && device_count) { + switch (data_profile & BTRFS_BLOCK_GROUP_PROFILE_MASK) { + case BTRFS_BLOCK_GROUP_DUP: + case BTRFS_BLOCK_GROUP_RAID1: + case BTRFS_BLOCK_GROUP_RAID1C3: + case BTRFS_BLOCK_GROUP_RAID1C4: + case BTRFS_BLOCK_GROUP_RAID0: + case BTRFS_BLOCK_GROUP_RAID10: + case BTRFS_BLOCK_GROUP_RAID56_MASK: + features.incompat_flags |= + BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE; + break; + default: + break; + } + } +#endif + + if (opt_zoned) { + u64 metadata = BTRFS_BLOCK_GROUP_METADATA | metadata_profile; + u64 data = BTRFS_BLOCK_GROUP_DATA | data_profile; +#if EXPERIMENTAL + bool rst = features.incompat_flags & + BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE; +#else + bool rst = false; +#endif + + if (!zoned_profile_supported(metadata, rst) || + !zoned_profile_supported(data, rst)) { + error("zoned mode does not yet support RAID/DUP profiles, please specify '-d single -m single' manually"); + goto error; + } } t_prepare = calloc(device_count, sizeof(*t_prepare)); @@ -1755,6 +1819,16 @@ raid_groups: goto out; } } + if (features.incompat_flags & BTRFS_FEATURE_INCOMPAT_RAID_STRIPE_TREE) { + ret = setup_raid_stripe_tree_root(fs_info); + if (ret < 0) { + error("failed to initialize raid-stripe-tree: %d (%m)", + ret); + goto out; + } + } + + if (bconf.verbose) { char features_buf[BTRFS_FEATURE_STRING_BUF_SIZE];
Allow for RAID levels 0, 1 and 10 on zoned devices if the RAID stripe tree is used. Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> --- common/fsfeatures.c | 8 +++++ kernel-shared/ctree.h | 3 +- kernel-shared/zoned.c | 35 ++++++++++++++++-- kernel-shared/zoned.h | 4 +-- mkfs/main.c | 82 ++++++++++++++++++++++++++++++++++++++++--- 5 files changed, 122 insertions(+), 10 deletions(-)