diff options
author | Joel Becker <joel.becker@oracle.com> | 2008-09-11 18:35:47 -0400 |
---|---|---|
committer | Mark Fasheh <mfasheh@suse.com> | 2009-01-05 11:40:30 -0500 |
commit | e06c8227fd94ec181849ba206bf032be31c4295c (patch) | |
tree | 67261cd94aa86dc6112d7de74304c2a1af5b64fb /fs | |
parent | 754938c142ae0c28360426c43f965ddc5164b21e (diff) |
jbd2: Add buffer triggers
Filesystems often to do compute intensive operation on some
metadata. If this operation is repeated many times, it can be very
expensive. It would be much nicer if the operation could be performed
once before a buffer goes to disk.
This adds triggers to jbd2 buffer heads. Just before writing a metadata
buffer to the journal, jbd2 will optionally call a commit trigger associated
with the buffer. If the journal is aborted, an abort trigger will be
called on any dirty buffers as they are dropped from pending
transactions.
ocfs2 will use this feature.
Initially I tried to come up with a more generic trigger that could be
used for non-buffer-related events like transaction completion. It
doesn't tie nicely, because the information a buffer trigger needs
(specific to a journal_head) isn't the same as what a transaction
trigger needs (specific to a tranaction_t or perhaps journal_t). So I
implemented a buffer set, with the understanding that
journal/transaction wide triggers should be implemented separately.
There is only one trigger set allowed per buffer. I can't think of any
reason to attach more than one set. Contrast this with a journal or
transaction in which multiple places may want to watch the entire
transaction separately.
The trigger sets are considered static allocation from the jbd2
perspective. ocfs2 will just have one trigger set per block type,
setting the same set on every bh of the same type.
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: <linux-ext4@vger.kernel.org>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Diffstat (limited to 'fs')
-rw-r--r-- | fs/jbd2/commit.c | 9 | ||||
-rw-r--r-- | fs/jbd2/journal.c | 19 | ||||
-rw-r--r-- | fs/jbd2/transaction.c | 47 |
3 files changed, 75 insertions, 0 deletions
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index ebc667bc54a8..c8a1bace685a 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c | |||
@@ -509,6 +509,10 @@ void jbd2_journal_commit_transaction(journal_t *journal) | |||
509 | if (is_journal_aborted(journal)) { | 509 | if (is_journal_aborted(journal)) { |
510 | clear_buffer_jbddirty(jh2bh(jh)); | 510 | clear_buffer_jbddirty(jh2bh(jh)); |
511 | JBUFFER_TRACE(jh, "journal is aborting: refile"); | 511 | JBUFFER_TRACE(jh, "journal is aborting: refile"); |
512 | jbd2_buffer_abort_trigger(jh, | ||
513 | jh->b_frozen_data ? | ||
514 | jh->b_frozen_triggers : | ||
515 | jh->b_triggers); | ||
512 | jbd2_journal_refile_buffer(journal, jh); | 516 | jbd2_journal_refile_buffer(journal, jh); |
513 | /* If that was the last one, we need to clean up | 517 | /* If that was the last one, we need to clean up |
514 | * any descriptor buffers which may have been | 518 | * any descriptor buffers which may have been |
@@ -844,6 +848,9 @@ restart_loop: | |||
844 | * data. | 848 | * data. |
845 | * | 849 | * |
846 | * Otherwise, we can just throw away the frozen data now. | 850 | * Otherwise, we can just throw away the frozen data now. |
851 | * | ||
852 | * We also know that the frozen data has already fired | ||
853 | * its triggers if they exist, so we can clear that too. | ||
847 | */ | 854 | */ |
848 | if (jh->b_committed_data) { | 855 | if (jh->b_committed_data) { |
849 | jbd2_free(jh->b_committed_data, bh->b_size); | 856 | jbd2_free(jh->b_committed_data, bh->b_size); |
@@ -851,10 +858,12 @@ restart_loop: | |||
851 | if (jh->b_frozen_data) { | 858 | if (jh->b_frozen_data) { |
852 | jh->b_committed_data = jh->b_frozen_data; | 859 | jh->b_committed_data = jh->b_frozen_data; |
853 | jh->b_frozen_data = NULL; | 860 | jh->b_frozen_data = NULL; |
861 | jh->b_frozen_triggers = NULL; | ||
854 | } | 862 | } |
855 | } else if (jh->b_frozen_data) { | 863 | } else if (jh->b_frozen_data) { |
856 | jbd2_free(jh->b_frozen_data, bh->b_size); | 864 | jbd2_free(jh->b_frozen_data, bh->b_size); |
857 | jh->b_frozen_data = NULL; | 865 | jh->b_frozen_data = NULL; |
866 | jh->b_frozen_triggers = NULL; | ||
858 | } | 867 | } |
859 | 868 | ||
860 | spin_lock(&journal->j_list_lock); | 869 | spin_lock(&journal->j_list_lock); |
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c index e70d657a19f8..f6bff9d6f8df 100644 --- a/fs/jbd2/journal.c +++ b/fs/jbd2/journal.c | |||
@@ -50,6 +50,7 @@ EXPORT_SYMBOL(jbd2_journal_unlock_updates); | |||
50 | EXPORT_SYMBOL(jbd2_journal_get_write_access); | 50 | EXPORT_SYMBOL(jbd2_journal_get_write_access); |
51 | EXPORT_SYMBOL(jbd2_journal_get_create_access); | 51 | EXPORT_SYMBOL(jbd2_journal_get_create_access); |
52 | EXPORT_SYMBOL(jbd2_journal_get_undo_access); | 52 | EXPORT_SYMBOL(jbd2_journal_get_undo_access); |
53 | EXPORT_SYMBOL(jbd2_journal_set_triggers); | ||
53 | EXPORT_SYMBOL(jbd2_journal_dirty_metadata); | 54 | EXPORT_SYMBOL(jbd2_journal_dirty_metadata); |
54 | EXPORT_SYMBOL(jbd2_journal_release_buffer); | 55 | EXPORT_SYMBOL(jbd2_journal_release_buffer); |
55 | EXPORT_SYMBOL(jbd2_journal_forget); | 56 | EXPORT_SYMBOL(jbd2_journal_forget); |
@@ -290,6 +291,7 @@ int jbd2_journal_write_metadata_buffer(transaction_t *transaction, | |||
290 | struct page *new_page; | 291 | struct page *new_page; |
291 | unsigned int new_offset; | 292 | unsigned int new_offset; |
292 | struct buffer_head *bh_in = jh2bh(jh_in); | 293 | struct buffer_head *bh_in = jh2bh(jh_in); |
294 | struct jbd2_buffer_trigger_type *triggers; | ||
293 | 295 | ||
294 | /* | 296 | /* |
295 | * The buffer really shouldn't be locked: only the current committing | 297 | * The buffer really shouldn't be locked: only the current committing |
@@ -314,13 +316,23 @@ repeat: | |||
314 | done_copy_out = 1; | 316 | done_copy_out = 1; |
315 | new_page = virt_to_page(jh_in->b_frozen_data); | 317 | new_page = virt_to_page(jh_in->b_frozen_data); |
316 | new_offset = offset_in_page(jh_in->b_frozen_data); | 318 | new_offset = offset_in_page(jh_in->b_frozen_data); |
319 | triggers = jh_in->b_frozen_triggers; | ||
317 | } else { | 320 | } else { |
318 | new_page = jh2bh(jh_in)->b_page; | 321 | new_page = jh2bh(jh_in)->b_page; |
319 | new_offset = offset_in_page(jh2bh(jh_in)->b_data); | 322 | new_offset = offset_in_page(jh2bh(jh_in)->b_data); |
323 | triggers = jh_in->b_triggers; | ||
320 | } | 324 | } |
321 | 325 | ||
322 | mapped_data = kmap_atomic(new_page, KM_USER0); | 326 | mapped_data = kmap_atomic(new_page, KM_USER0); |
323 | /* | 327 | /* |
328 | * Fire any commit trigger. Do this before checking for escaping, | ||
329 | * as the trigger may modify the magic offset. If a copy-out | ||
330 | * happens afterwards, it will have the correct data in the buffer. | ||
331 | */ | ||
332 | jbd2_buffer_commit_trigger(jh_in, mapped_data + new_offset, | ||
333 | triggers); | ||
334 | |||
335 | /* | ||
324 | * Check for escaping | 336 | * Check for escaping |
325 | */ | 337 | */ |
326 | if (*((__be32 *)(mapped_data + new_offset)) == | 338 | if (*((__be32 *)(mapped_data + new_offset)) == |
@@ -352,6 +364,13 @@ repeat: | |||
352 | new_page = virt_to_page(tmp); | 364 | new_page = virt_to_page(tmp); |
353 | new_offset = offset_in_page(tmp); | 365 | new_offset = offset_in_page(tmp); |
354 | done_copy_out = 1; | 366 | done_copy_out = 1; |
367 | |||
368 | /* | ||
369 | * This isn't strictly necessary, as we're using frozen | ||
370 | * data for the escaping, but it keeps consistency with | ||
371 | * b_frozen_data usage. | ||
372 | */ | ||
373 | jh_in->b_frozen_triggers = jh_in->b_triggers; | ||
355 | } | 374 | } |
356 | 375 | ||
357 | /* | 376 | /* |
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c index 39b7805a599a..4f925a4f3d05 100644 --- a/fs/jbd2/transaction.c +++ b/fs/jbd2/transaction.c | |||
@@ -741,6 +741,12 @@ done: | |||
741 | source = kmap_atomic(page, KM_USER0); | 741 | source = kmap_atomic(page, KM_USER0); |
742 | memcpy(jh->b_frozen_data, source+offset, jh2bh(jh)->b_size); | 742 | memcpy(jh->b_frozen_data, source+offset, jh2bh(jh)->b_size); |
743 | kunmap_atomic(source, KM_USER0); | 743 | kunmap_atomic(source, KM_USER0); |
744 | |||
745 | /* | ||
746 | * Now that the frozen data is saved off, we need to store | ||
747 | * any matching triggers. | ||
748 | */ | ||
749 | jh->b_frozen_triggers = jh->b_triggers; | ||
744 | } | 750 | } |
745 | jbd_unlock_bh_state(bh); | 751 | jbd_unlock_bh_state(bh); |
746 | 752 | ||
@@ -944,6 +950,47 @@ out: | |||
944 | } | 950 | } |
945 | 951 | ||
946 | /** | 952 | /** |
953 | * void jbd2_journal_set_triggers() - Add triggers for commit writeout | ||
954 | * @bh: buffer to trigger on | ||
955 | * @type: struct jbd2_buffer_trigger_type containing the trigger(s). | ||
956 | * | ||
957 | * Set any triggers on this journal_head. This is always safe, because | ||
958 | * triggers for a committing buffer will be saved off, and triggers for | ||
959 | * a running transaction will match the buffer in that transaction. | ||
960 | * | ||
961 | * Call with NULL to clear the triggers. | ||
962 | */ | ||
963 | void jbd2_journal_set_triggers(struct buffer_head *bh, | ||
964 | struct jbd2_buffer_trigger_type *type) | ||
965 | { | ||
966 | struct journal_head *jh = bh2jh(bh); | ||
967 | |||
968 | jh->b_triggers = type; | ||
969 | } | ||
970 | |||
971 | void jbd2_buffer_commit_trigger(struct journal_head *jh, void *mapped_data, | ||
972 | struct jbd2_buffer_trigger_type *triggers) | ||
973 | { | ||
974 | struct buffer_head *bh = jh2bh(jh); | ||
975 | |||
976 | if (!triggers || !triggers->t_commit) | ||
977 | return; | ||
978 | |||
979 | triggers->t_commit(triggers, bh, mapped_data, bh->b_size); | ||
980 | } | ||
981 | |||
982 | void jbd2_buffer_abort_trigger(struct journal_head *jh, | ||
983 | struct jbd2_buffer_trigger_type *triggers) | ||
984 | { | ||
985 | if (!triggers || !triggers->t_abort) | ||
986 | return; | ||
987 | |||
988 | triggers->t_abort(triggers, jh2bh(jh)); | ||
989 | } | ||
990 | |||
991 | |||
992 | |||
993 | /** | ||
947 | * int jbd2_journal_dirty_metadata() - mark a buffer as containing dirty metadata | 994 | * int jbd2_journal_dirty_metadata() - mark a buffer as containing dirty metadata |
948 | * @handle: transaction to add buffer to. | 995 | * @handle: transaction to add buffer to. |
949 | * @bh: buffer to mark | 996 | * @bh: buffer to mark |