From: Josef Bacik <josef@toxicpanda.com>
Date: Wed, 28 Apr 2021 13:38:48 -0400
Subject: btrfs: handle preemptive delalloc flushing slightly differently
Git-commit: 385f421f18be653d21ccfd6520fbddf206ad43eb
Patch-mainline: v5.14-rc1
References: bsc#1202528
If we decide to flush delalloc from the preemptive flusher, we really do
not want to wait on ordered extents, as it gains us nothing. However
there was logic to go ahead and wait on ordered extents if there was
more ordered bytes than delalloc bytes. We do not want this behavior,
so pass through whether this flushing is for preemption, and do not wait
for ordered extents if that's the case. Also break out of the shrink
loop after the first flushing, as we just want to one shot shrink
delalloc.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Acked-by: Nikolay Borisov <nborisov@suse.com>
---
fs/btrfs/space-info.c | 132 +++++++-------------------------------------------
1 file changed, 19 insertions(+), 113 deletions(-)
--- a/fs/btrfs/space-info.c
+++ b/fs/btrfs/space-info.c
@@ -373,7 +373,8 @@ static inline u64 calc_reclaim_items_nr(
*/
static void shrink_delalloc(struct btrfs_fs_info *fs_info,
struct btrfs_space_info *space_info,
- u64 to_reclaim, bool wait_ordered)
+ u64 to_reclaim, bool wait_ordered,
+ bool for_preempt)
{
struct btrfs_trans_handle *trans;
u64 delalloc_bytes;
@@ -410,7 +411,7 @@ static void shrink_delalloc(struct btrfs
* ordered extents, otherwise we'll waste time trying to flush delalloc
* that likely won't give us the space back we need.
*/
- if (ordered_bytes > delalloc_bytes)
+ if (ordered_bytes > delalloc_bytes && !for_preempt)
wait_ordered = true;
loops = 0;
@@ -428,6 +429,14 @@ static void shrink_delalloc(struct btrfs
break;
}
+ /*
+ * If we are for preemption we just want a one-shot of delalloc
+ * flushing so we can stop flushing if we decide we don't need
+ * to anymore.
+ */
+ if (for_preempt)
+ break;
+
spin_lock(&space_info->lock);
if (list_empty(&space_info->tickets) &&
list_empty(&space_info->priority_tickets)) {
@@ -443,110 +452,6 @@ static void shrink_delalloc(struct btrfs
}
}
-/**
- * maybe_commit_transaction - possibly commit the transaction if its ok to
- * @root - the root we're allocating for
- * @bytes - the number of bytes we want to reserve
- * @force - force the commit
- *
- * This will check to make sure that committing the transaction will actually
- * get us somewhere and then commit the transaction if it does. Otherwise it
- * will return -ENOSPC.
- */
-static int may_commit_transaction(struct btrfs_fs_info *fs_info,
- struct btrfs_space_info *space_info)
-{
- struct reserve_ticket *ticket = NULL;
- struct btrfs_block_rsv *delayed_rsv = &fs_info->delayed_block_rsv;
- struct btrfs_block_rsv *delayed_refs_rsv = &fs_info->delayed_refs_rsv;
- struct btrfs_block_rsv *trans_rsv = &fs_info->trans_block_rsv;
- struct btrfs_trans_handle *trans;
- u64 reclaim_bytes = 0;
- u64 bytes_needed = 0;
- u64 cur_free_bytes = 0;
-
- trans = (struct btrfs_trans_handle *)current->journal_info;
- if (trans)
- return -EAGAIN;
-
- spin_lock(&space_info->lock);
- cur_free_bytes = btrfs_space_info_used(space_info, true);
- if (cur_free_bytes < space_info->total_bytes)
- cur_free_bytes = space_info->total_bytes - cur_free_bytes;
- else
- cur_free_bytes = 0;
-
- if (!list_empty(&space_info->priority_tickets))
- ticket = list_first_entry(&space_info->priority_tickets,
- struct reserve_ticket, list);
- else if (!list_empty(&space_info->tickets))
- ticket = list_first_entry(&space_info->tickets,
- struct reserve_ticket, list);
- if (ticket)
- bytes_needed = ticket->bytes;
-
- if (bytes_needed > cur_free_bytes)
- bytes_needed -= cur_free_bytes;
- else
- bytes_needed = 0;
- spin_unlock(&space_info->lock);
-
- if (!bytes_needed)
- return 0;
-
- trans = btrfs_join_transaction(fs_info->extent_root);
- if (IS_ERR(trans))
- return PTR_ERR(trans);
-
- /*
- * See if there is enough pinned space to make this reservation, or if
- * we have block groups that are going to be freed, allowing us to
- * possibly do a chunk allocation the next loop through.
- */
- if (test_bit(BTRFS_TRANS_HAVE_FREE_BGS, &trans->transaction->flags) ||
- __percpu_counter_compare(&space_info->total_bytes_pinned,
- bytes_needed,
- BTRFS_TOTAL_BYTES_PINNED_BATCH) >= 0)
- goto commit;
-
- /*
- * See if there is some space in the delayed insertion reserve for this
- * reservation. If the space_info's don't match (like for DATA or
- * SYSTEM) then just go enospc, reclaiming this space won't recover any
- * space to satisfy those reservations.
- */
- if (space_info != delayed_rsv->space_info)
- goto enospc;
-
- spin_lock(&delayed_rsv->lock);
- reclaim_bytes += delayed_rsv->reserved;
- spin_unlock(&delayed_rsv->lock);
-
- spin_lock(&delayed_refs_rsv->lock);
- reclaim_bytes += delayed_refs_rsv->reserved;
- spin_unlock(&delayed_refs_rsv->lock);
-
- spin_lock(&trans_rsv->lock);
- reclaim_bytes += trans_rsv->reserved;
- spin_unlock(&trans_rsv->lock);
-
- if (reclaim_bytes >= bytes_needed)
- goto commit;
- bytes_needed -= reclaim_bytes;
-
- if (__percpu_counter_compare(&space_info->total_bytes_pinned,
- bytes_needed,
- BTRFS_TOTAL_BYTES_PINNED_BATCH) < 0) {
- goto enospc;
- }
-
-commit:
- return btrfs_commit_transaction(trans);
-enospc:
- btrfs_end_transaction(trans);
- return -ENOSPC;
-}
-
static int flush_space(struct btrfs_fs_info *fs_info,
struct btrfs_space_info *space_info, u64 num_bytes,
enum btrfs_flush_state state, bool for_preempt)
@@ -575,7 +480,7 @@ static int flush_space(struct btrfs_fs_i
case FLUSH_DELALLOC:
case FLUSH_DELALLOC_WAIT:
shrink_delalloc(fs_info, space_info, num_bytes,
- state == FLUSH_DELALLOC_WAIT);
+ state == FLUSH_DELALLOC_WAIT, for_preempt);
break;
case FLUSH_DELAYED_REFS_NR:
case FLUSH_DELAYED_REFS:
@@ -1075,12 +980,13 @@ static void btrfs_preempt_reclaim_metada
* the information it needs to make the right decision.
*
* COMMIT_TRANS
- * This is where we reclaim all of the pinned space generated by the previous
- * two stages. We will not commit the transaction if we don't think we're
- * likely to satisfy our request, which means if our current free space +
- * total_bytes_pinned < reservation we will not commit. This is why the
- * previous states are actually important, to make sure we know for sure
- * whether committing the transaction will allow us to make progress.
+ * This will commit the transaction. Historically we had a lot of logic
+ * surrounding whether or not we'd commit the transaction, but this waits born
+ * out of a pre-tickets era where we could end up committing the transaction
+ * thousands of times in a row without making progress. Now thanks to our
+ * ticketing system we know if we're not making progress and can error
+ * everybody out after a few commits rather than burning the disk hoping for
+ * a different answer.
*
* ALLOC_CHUNK_FORCE
* For data we start with alloc chunk force, however we could have been full