btrfs: remove EXTENT_BUFFER_IN_TREE flag
PATCH 1/1 btrfs: remove EXTENT_BUFFER_IN_TREE flag
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
--- fs/btrfs/extent_io.c | 18 ++++++------------ fs/btrfs/extent_io.h | 1 - 2 files changed, 6 insertions(+), 13 deletions(-)
fs/btrfs/extent_io.c +6 -12
@@ -3048,7 +3048,6 @@ struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, goto again; }check_buffer_tree_ref (eb); -set_bit (EXTENT_BUFFER_IN_TREE , &eb->bflags);return eb; free_eb:
@@ -3368,7 +3367,6 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, }/* add one reference for the tree */ check_buffer_tree_ref (eb); -set_bit (EXTENT_BUFFER_IN_TREE , &eb->bflags);/* * Now it's safe to unlock the pages because any calls to
@@ -3432,18 +3430,14 @@ static int release_extent_buffer(struct extent_buffer *eb)WARN_ON (atomic_read (&eb->refs) ==0 );if (atomic_dec_and_test (&eb->refs)) { -if (test_and_clear_bit (EXTENT_BUFFER_IN_TREE , &eb->bflags)) { -struct btrfs_fs_info *fs_info = eb->fs_info; +struct btrfs_fs_info *fs_info = eb->fs_info; -spin_unlock (&eb->refs_lock); +spin_unlock (&eb->refs_lock); -spin_lock (&fs_info->buffer_lock); -radix_tree_delete (&fs_info->buffer_radix, - eb->start >> fs_info->sectorsize_bits); -spin_unlock (&fs_info->buffer_lock); - }else { -spin_unlock (&eb->refs_lock); - } +spin_lock (&fs_info->buffer_lock); +radix_tree_delete_item (&fs_info->buffer_radix, + eb->start >> fs_info->sectorsize_bits, eb); +spin_unlock (&fs_info->buffer_lock);btrfs_leak_debug_del_eb (eb);/* Should be safe to release folios at this point. */
fs/btrfs/extent_io.h +0 -1
@@ -47,7 +47,6 @@ enum {/* read IO error */ EXTENT_BUFFER_READ_ERR ,EXTENT_BUFFER_UNMAPPED , -EXTENT_BUFFER_IN_TREE ,/* write IO error */ EXTENT_BUFFER_WRITE_ERR,/* Indicate the extent buffer is written zeroed out (for zoned) */
Signature · 1 line
2.47.2
Quoted 2 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
Would be good to add the reference to commit that added the bit, 34b41acec1ccc0 ("Btrfs: use a bit to track if we're in the radix tree") and wanted to make use of it, faa2dbf004e89e ("Btrfs: add sanity tests for new qgroup accounting code"). And both are 10+ years old.
Quoted 1 line
Signed-off-by Daniel Vacek <[email protected]>
Quoted 9 lines · David Sterba
Quoted 2 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
Would be good to add the reference to commit that added the bit,
34b41acec1ccc0("Btrfs: use a bit to track if we're in the radix tree") and wanted to make use of it,faa2dbf004e89e("Btrfs: add sanity tests for new qgroup accounting code"). And both are 10+ years old.
Right, I could have checked the history.
Though honestly from the diff of these two commits I don't see any valid usage of this flag either. Must have been somewhere in the context or I'm missing something.
Quoted 3 lines
Quoted 1 line
Signed-off-by Daniel Vacek <[email protected]>Reviewed-by David Sterba <[email protected]>
Quoted 11 lines · Daniel Vacek
Quoted 9 lines · David Sterba
Quoted 2 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
Would be good to add the reference to commit that added the bit,
34b41acec1ccc0("Btrfs: use a bit to track if we're in the radix tree") and wanted to make use of it,faa2dbf004e89e("Btrfs: add sanity tests for new qgroup accounting code"). And both are 10+ years old.
Quoted 5 lines
Right, I could have checked the history.
Though honestly from the diff of these two commits I don't see any valid usage of this flag either. Must have been somewhere in the context or I'm missing something.
Yeah, from the diff it can be seen if the code has any effects. It's recommended to analyze the code also from the historical context because it can be a leftover from a cleanup (and most of the time it is), but there's still a chance the initial intentions of the code are still valid and it was the cleanup that broke it.
Quoted 2 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
I na similar way the flag EXTENT_BUFFER_READ_ERR is unused (was removed in eb/io rework in 046b562b20a5cf ("btrfs: use a separate end_io handler for read_extent_buffer").
Quoted 7 lines · David Sterba
Quoted 2 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
I na similar way the flag
EXTENT_BUFFER_READ_ERRis unused (was removed in eb/io rework in046b562b20a5cf("btrfs: use a separateend_iohandler forread_extent_buffer").
And EXTENT_BUFFER_READAHEAD, removed by f26c9238602856 ("btrfs: remove reada infrastructure").
Quoted 6 lines · David Sterba
Quoted 4 lines · David Sterba
Quoted 2 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
Quoted 6 lines
Quoted 3 lines
I na similar way the flag
EXTENT_BUFFER_READ_ERRis unused (was removed in eb/io rework in046b562b20a5cf("btrfs: use a separateend_iohandler forread_extent_buffer").And
EXTENT_BUFFER_READAHEAD, removed byf26c9238602856("btrfs: remove reada infrastructure").
I see. I can have a look into further cleanup later. I was not checking any other flags yet as that's unrelated to my issue.
The full story is I'm chasing an eb refcount issue on downstream v6.4 running on aarch64 with 64k pages. First I suspected it's due to this flag racing (only setting the flag after inserting a newly allocated eb into the tree but without holding the eb->refs_lock) but after a bit of hammering I managed to reproduce the soft lockup even with this patch applied.
That will need to be solved downstream. Still this looks like a worthy cleanup.
Quoted 4 lines · Daniel Vacek
This flag is set after inserting the eb to the buffer tree and cleared on it's removal. But it does not bring any added value. Just kill it for good.
Signed-off-by Daniel Vacek <[email protected]>
Added to for-next with the updated changelog, thanks.