Re: [PATCH v8 07/11] btrfs: zoned: allow zoned RAID

From: Johannes Thumshirn
Date: Wed Sep 13 2023 - 01:42:01 EST


On 12.09.23 22:49, David Sterba wrote:
> On Mon, Sep 11, 2023 at 05:52:08AM -0700, Johannes Thumshirn wrote:
>> When we have a raid-stripe-tree, we can do RAID0/1/10 on zoned devices for
>> data block-groups. For meta-data block-groups, we don't actually need
>> anything special, as all meta-data I/O is protected by the
>> btrfs_zoned_meta_io_lock() already.
>>
>> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@xxxxxxx>
>> ---
>> fs/btrfs/raid-stripe-tree.h | 7 ++-
>> fs/btrfs/volumes.c | 2 +
>> fs/btrfs/zoned.c | 113 +++++++++++++++++++++++++++++++++++++++++++-
>> 3 files changed, 119 insertions(+), 3 deletions(-)
>>
>> diff --git a/fs/btrfs/raid-stripe-tree.h b/fs/btrfs/raid-stripe-tree.h
>> index 40aa553ae8aa..30c7d5981890 100644
>> --- a/fs/btrfs/raid-stripe-tree.h
>> +++ b/fs/btrfs/raid-stripe-tree.h
>> @@ -8,6 +8,11 @@
>>
>> #include "disk-io.h"
>>
>> +#define BTRFS_RST_SUPP_BLOCK_GROUP_MASK (BTRFS_BLOCK_GROUP_DUP |\
>> + BTRFS_BLOCK_GROUP_RAID1_MASK |\
>> + BTRFS_BLOCK_GROUP_RAID0 |\
>> + BTRFS_BLOCK_GROUP_RAID10)
>> +
>> struct btrfs_io_context;
>> struct btrfs_io_stripe;
>>
>> @@ -32,7 +37,7 @@ static inline bool btrfs_need_stripe_tree_update(struct btrfs_fs_info *fs_info,
>> if (type != BTRFS_BLOCK_GROUP_DATA)
>> return false;
>>
>> - if (profile & BTRFS_BLOCK_GROUP_RAID1_MASK)
>> + if (profile & BTRFS_RST_SUPP_BLOCK_GROUP_MASK)
>> return true;
>>
>> return false;
>> diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
>> index 7c25f5c77788..9f17e5f290f4 100644
>> --- a/fs/btrfs/volumes.c
>> +++ b/fs/btrfs/volumes.c
>> @@ -6438,6 +6438,8 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
>> * I/O context structure.
>> */
>> if (smap && num_alloc_stripes == 1 &&
>> + !(btrfs_need_stripe_tree_update(fs_info, map->type) &&
>> + op != BTRFS_MAP_READ) &&
>> !((map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) && mirror_num > 1)) {
>> ret = set_io_stripe(fs_info, op, logical, length, smap, map,
>> stripe_index, stripe_offset, stripe_nr);
>> diff --git a/fs/btrfs/zoned.c b/fs/btrfs/zoned.c
>> index c6eedf4bfba9..4ca36875058c 100644
>> --- a/fs/btrfs/zoned.c
>> +++ b/fs/btrfs/zoned.c
>> @@ -1481,8 +1481,9 @@ int btrfs_load_block_group_zone_info(struct btrfs_block_group *cache, bool new)
>> set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE, &cache->runtime_flags);
>> break;
>> case BTRFS_BLOCK_GROUP_DUP:
>> - if (map->type & BTRFS_BLOCK_GROUP_DATA) {
>> - btrfs_err(fs_info, "zoned: profile DUP not yet supported on data bg");
>> + if (map->type & BTRFS_BLOCK_GROUP_DATA &&
>> + !btrfs_stripe_tree_root(fs_info)) {
>> + btrfs_err(fs_info, "zoned: data DUP profile needs stripe_root");
>> ret = -EINVAL;
>> goto out;
>> }
>> @@ -1520,8 +1521,116 @@ int btrfs_load_block_group_zone_info(struct btrfs_block_group *cache, bool new)
>> cache->zone_capacity = min(caps[0], caps[1]);
>> break;
>> case BTRFS_BLOCK_GROUP_RAID1:
>> + case BTRFS_BLOCK_GROUP_RAID1C3:
>> + case BTRFS_BLOCK_GROUP_RAID1C4:
>
> This
>
>> + if (map->type & BTRFS_BLOCK_GROUP_DATA &&
>> + !btrfs_stripe_tree_root(fs_info)) {
>> + btrfs_err(fs_info,
>> + "zoned: data %s needs stripe_root",
>> + btrfs_bg_type_to_raid_name(map->type));
>> + ret = -EIO;
>> + goto out;
>> +
>> + }
>> +
>> + for (i = 0; i < map->num_stripes; i++) {
>> + if (alloc_offsets[i] == WP_MISSING_DEV ||
>> + alloc_offsets[i] == WP_CONVENTIONAL)
>> + continue;
>> +
>> + if ((alloc_offsets[0] != alloc_offsets[i]) &&
>> + !btrfs_test_opt(fs_info, DEGRADED)) {
>> + btrfs_err(fs_info,
>> + "zoned: write pointer offset mismatch of zones in %s profile",
>> + btrfs_bg_type_to_raid_name(map->type));
>> + ret = -EIO;
>> + goto out;
>> + }
>> + if (test_bit(0, active) != test_bit(i, active)) {
>> + if (!btrfs_test_opt(fs_info, DEGRADED) &&
>> + !btrfs_zone_activate(cache)) {
>> + ret = -EIO;
>> + goto out;
>> + }
>> + } else {
>> + if (test_bit(0, active))
>> + set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
>> + &cache->runtime_flags);
>> + }
>> + /*
>> + * In case a device is missing we have a cap of 0, so don't
>> + * use it.
>> + */
>> + cache->zone_capacity = min_not_zero(caps[0], caps[i]);
>> + }
>> +
>> + if (alloc_offsets[0] != WP_MISSING_DEV)
>> + cache->alloc_offset = alloc_offsets[0];
>> + else
>> + cache->alloc_offset = alloc_offsets[i - 1];
>
> whole block
>
>> + break;
>> case BTRFS_BLOCK_GROUP_RAID0:
>
> and
>
>> + if (map->type & BTRFS_BLOCK_GROUP_DATA &&
>> + !btrfs_stripe_tree_root(fs_info)) {
>> + btrfs_err(fs_info,
>> + "zoned: data %s needs stripe_root",
>> + btrfs_bg_type_to_raid_name(map->type));
>> + ret = -EIO;
>> + goto out;
>> +
>> + }
>> + for (i = 0; i < map->num_stripes; i++) {
>> + if (alloc_offsets[i] == WP_MISSING_DEV ||
>> + alloc_offsets[i] == WP_CONVENTIONAL)
>> + continue;
>> +
>> + if (test_bit(0, active) != test_bit(i, active)) {
>> + if (!btrfs_zone_activate(cache)) {
>> + ret = -EIO;
>> + goto out;
>> + }
>> + } else {
>> + if (test_bit(0, active))
>> + set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
>> + &cache->runtime_flags);
>> + }
>> + cache->zone_capacity += caps[i];
>> + cache->alloc_offset += alloc_offsets[i];
>> +
>> + }
>> + break;
>> case BTRFS_BLOCK_GROUP_RAID10:
>> + if (map->type & BTRFS_BLOCK_GROUP_DATA &&
>> + !btrfs_stripe_tree_root(fs_info)) {
>> + btrfs_err(fs_info,
>> + "zoned: data %s needs stripe_root",
>> + btrfs_bg_type_to_raid_name(map->type));
>> + ret = -EIO;
>> + goto out;
>> +
>> + }
>> + for (i = 0; i < map->num_stripes; i++) {
>> + if (alloc_offsets[i] == WP_MISSING_DEV ||
>> + alloc_offsets[i] == WP_CONVENTIONAL)
>> + continue;
>> +
>> + if (test_bit(0, active) != test_bit(i, active)) {
>> + if (!btrfs_zone_activate(cache)) {
>> + ret = -EIO;
>> + goto out;
>> + }
>> + } else {
>> + if (test_bit(0, active))
>> + set_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
>> + &cache->runtime_flags);
>> + }
>> + if ((i % map->sub_stripes) == 0) {
>> + cache->zone_capacity += caps[i];
>> + cache->alloc_offset += alloc_offsets[i];
>> + }
>> +
>> + }
>> + break;
>
> Seem to be quite long and nested for a case, can they be factored to
> helpers?

Sure, but I'd love to have
https://lore.kernel.org/all/20230605085108.580976-1-hch@xxxxxx/
pulled in first. This patchset handles (among other things) the DUP and
single cases as well.