From c2d6fd9d6f35079f1669f0100f05b46708c74b7f Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Tue, 6 Jun 2023 21:59:23 +0800 Subject: [PATCH 1/9] jbd2: recheck chechpointing non-dirty buffer There is a long-standing metadata corruption issue that happens from time to time, but it's very difficult to reproduce and analyse, benefit from the JBD2_CYCLE_RECORD option, we found out that the problem is the checkpointing process miss to write out some buffers which are raced by another do_get_write_access(). Looks below for detail. jbd2_log_do_checkpoint() //transaction X //buffer A is dirty and not belones to any transaction __buffer_relink_io() //move it to the IO list __flush_batch() write_dirty_buffer() do_get_write_access() clear_buffer_dirty __jbd2_journal_file_buffer() //add buffer A to a new transaction Y lock_buffer(bh) //doesn't write out __jbd2_journal_remove_checkpoint() //finish checkpoint except buffer A //filesystem corrupt if the new transaction Y isn't fully write out. Due to the t_checkpoint_list walking loop in jbd2_log_do_checkpoint() have already handles waiting for buffers under IO and re-added new transaction to complete commit, and it also removing cleaned buffers, this makes sure the list will eventually get empty. So it's fine to leave buffers on the t_checkpoint_list while flushing out and completely stop using the t_checkpoint_io_list. Cc: stable@vger.kernel.org Suggested-by: Jan Kara Signed-off-by: Zhang Yi Tested-by: Zhihao Cheng Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20230606135928.434610-2-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o --- fs/jbd2/checkpoint.c | 102 ++++++++++++------------------------------- 1 file changed, 29 insertions(+), 73 deletions(-) diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c index 51bd38da21cd..25e3c20eb19f 100644 --- a/fs/jbd2/checkpoint.c +++ b/fs/jbd2/checkpoint.c @@ -57,28 +57,6 @@ static inline void __buffer_unlink(struct journal_head *jh) } } -/* - * Move a buffer from the checkpoint list to the checkpoint io list - * - * Called with j_list_lock held - */ -static inline void __buffer_relink_io(struct journal_head *jh) -{ - transaction_t *transaction = jh->b_cp_transaction; - - __buffer_unlink_first(jh); - - if (!transaction->t_checkpoint_io_list) { - jh->b_cpnext = jh->b_cpprev = jh; - } else { - jh->b_cpnext = transaction->t_checkpoint_io_list; - jh->b_cpprev = transaction->t_checkpoint_io_list->b_cpprev; - jh->b_cpprev->b_cpnext = jh; - jh->b_cpnext->b_cpprev = jh; - } - transaction->t_checkpoint_io_list = jh; -} - /* * Check a checkpoint buffer could be release or not. * @@ -183,6 +161,7 @@ __flush_batch(journal_t *journal, int *batch_count) struct buffer_head *bh = journal->j_chkpt_bhs[i]; BUFFER_TRACE(bh, "brelse"); __brelse(bh); + journal->j_chkpt_bhs[i] = NULL; } *batch_count = 0; } @@ -242,6 +221,11 @@ restart: jh = transaction->t_checkpoint_list; bh = jh2bh(jh); + /* + * The buffer may be writing back, or flushing out in the + * last couple of cycles, or re-adding into a new transaction, + * need to check it again until it's unlocked. + */ if (buffer_locked(bh)) { get_bh(bh); spin_unlock(&journal->j_list_lock); @@ -287,28 +271,32 @@ restart: } if (!buffer_dirty(bh)) { BUFFER_TRACE(bh, "remove from checkpoint"); - if (__jbd2_journal_remove_checkpoint(jh)) - /* The transaction was released; we're done */ + /* + * If the transaction was released or the checkpoint + * list was empty, we're done. + */ + if (__jbd2_journal_remove_checkpoint(jh) || + !transaction->t_checkpoint_list) goto out; - continue; + } else { + /* + * We are about to write the buffer, it could be + * raced by some other transaction shrink or buffer + * re-log logic once we release the j_list_lock, + * leave it on the checkpoint list and check status + * again to make sure it's clean. + */ + BUFFER_TRACE(bh, "queue"); + get_bh(bh); + J_ASSERT_BH(bh, !buffer_jwrite(bh)); + journal->j_chkpt_bhs[batch_count++] = bh; + transaction->t_chp_stats.cs_written++; + transaction->t_checkpoint_list = jh->b_cpnext; } - /* - * Important: we are about to write the buffer, and - * possibly block, while still holding the journal - * lock. We cannot afford to let the transaction - * logic start messing around with this buffer before - * we write it to disk, as that would break - * recoverability. - */ - BUFFER_TRACE(bh, "queue"); - get_bh(bh); - J_ASSERT_BH(bh, !buffer_jwrite(bh)); - journal->j_chkpt_bhs[batch_count++] = bh; - __buffer_relink_io(jh); - transaction->t_chp_stats.cs_written++; + if ((batch_count == JBD2_NR_BATCH) || - need_resched() || - spin_needbreak(&journal->j_list_lock)) + need_resched() || spin_needbreak(&journal->j_list_lock) || + jh2bh(transaction->t_checkpoint_list) == journal->j_chkpt_bhs[0]) goto unlock_and_flush; } @@ -322,38 +310,6 @@ restart: goto restart; } - /* - * Now we issued all of the transaction's buffers, let's deal - * with the buffers that are out for I/O. - */ -restart2: - /* Did somebody clean up the transaction in the meanwhile? */ - if (journal->j_checkpoint_transactions != transaction || - transaction->t_tid != this_tid) - goto out; - - while (transaction->t_checkpoint_io_list) { - jh = transaction->t_checkpoint_io_list; - bh = jh2bh(jh); - if (buffer_locked(bh)) { - get_bh(bh); - spin_unlock(&journal->j_list_lock); - wait_on_buffer(bh); - /* the journal_head may have gone by now */ - BUFFER_TRACE(bh, "brelse"); - __brelse(bh); - spin_lock(&journal->j_list_lock); - goto restart2; - } - - /* - * Now in whatever state the buffer currently is, we - * know that it has been written out and so we can - * drop it from the list - */ - if (__jbd2_journal_remove_checkpoint(jh)) - break; - } out: spin_unlock(&journal->j_list_lock); result = jbd2_cleanup_journal_tail(journal); From be22255360f80d3af789daad00025171a65424a5 Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Tue, 6 Jun 2023 21:59:24 +0800 Subject: [PATCH 2/9] jbd2: remove t_checkpoint_io_list Since t_checkpoint_io_list was stop using in jbd2_log_do_checkpoint() now, it's time to remove the whole t_checkpoint_io_list logic. Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20230606135928.434610-3-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o --- fs/jbd2/checkpoint.c | 42 ++---------------------------------------- fs/jbd2/commit.c | 3 +-- include/linux/jbd2.h | 6 ------ 3 files changed, 3 insertions(+), 48 deletions(-) diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c index 25e3c20eb19f..55d6efdbea64 100644 --- a/fs/jbd2/checkpoint.c +++ b/fs/jbd2/checkpoint.c @@ -27,7 +27,7 @@ * * Called with j_list_lock held. */ -static inline void __buffer_unlink_first(struct journal_head *jh) +static inline void __buffer_unlink(struct journal_head *jh) { transaction_t *transaction = jh->b_cp_transaction; @@ -40,23 +40,6 @@ static inline void __buffer_unlink_first(struct journal_head *jh) } } -/* - * Unlink a buffer from a transaction checkpoint(io) list. - * - * Called with j_list_lock held. - */ -static inline void __buffer_unlink(struct journal_head *jh) -{ - transaction_t *transaction = jh->b_cp_transaction; - - __buffer_unlink_first(jh); - if (transaction->t_checkpoint_io_list == jh) { - transaction->t_checkpoint_io_list = jh->b_cpnext; - if (transaction->t_checkpoint_io_list == jh) - transaction->t_checkpoint_io_list = NULL; - } -} - /* * Check a checkpoint buffer could be release or not. * @@ -503,15 +486,6 @@ again: break; if (need_resched() || spin_needbreak(&journal->j_list_lock)) break; - if (released) - continue; - - nr_freed += journal_shrink_one_cp_list(transaction->t_checkpoint_io_list, - nr_to_scan, &released); - if (*nr_to_scan == 0) - break; - if (need_resched() || spin_needbreak(&journal->j_list_lock)) - break; } while (transaction != last_transaction); if (transaction != last_transaction) { @@ -566,17 +540,6 @@ void __jbd2_journal_clean_checkpoint_list(journal_t *journal, bool destroy) */ if (need_resched()) return; - if (ret) - continue; - /* - * It is essential that we are as careful as in the case of - * t_checkpoint_list with removing the buffer from the list as - * we can possibly see not yet submitted buffers on io_list - */ - ret = journal_clean_one_cp_list(transaction-> - t_checkpoint_io_list, destroy); - if (need_resched()) - return; /* * Stop scanning if we couldn't free the transaction. This * avoids pointless scanning of transactions which still @@ -661,7 +624,7 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh) jbd2_journal_put_journal_head(jh); /* Is this transaction empty? */ - if (transaction->t_checkpoint_list || transaction->t_checkpoint_io_list) + if (transaction->t_checkpoint_list) return 0; /* @@ -753,7 +716,6 @@ void __jbd2_journal_drop_transaction(journal_t *journal, transaction_t *transact J_ASSERT(transaction->t_forget == NULL); J_ASSERT(transaction->t_shadow_list == NULL); J_ASSERT(transaction->t_checkpoint_list == NULL); - J_ASSERT(transaction->t_checkpoint_io_list == NULL); J_ASSERT(atomic_read(&transaction->t_updates) == 0); J_ASSERT(journal->j_committing_transaction != transaction); J_ASSERT(journal->j_running_transaction != transaction); diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index b33155dd7001..1073259902a6 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -1141,8 +1141,7 @@ restart_loop: spin_lock(&journal->j_list_lock); commit_transaction->t_state = T_FINISHED; /* Check if the transaction can be dropped now that we are finished */ - if (commit_transaction->t_checkpoint_list == NULL && - commit_transaction->t_checkpoint_io_list == NULL) { + if (commit_transaction->t_checkpoint_list == NULL) { __jbd2_journal_drop_transaction(journal, commit_transaction); jbd2_journal_free_transaction(commit_transaction); } diff --git a/include/linux/jbd2.h b/include/linux/jbd2.h index d860499e15e4..bd660aac8e07 100644 --- a/include/linux/jbd2.h +++ b/include/linux/jbd2.h @@ -613,12 +613,6 @@ struct transaction_s */ struct journal_head *t_checkpoint_list; - /* - * Doubly-linked circular list of all buffers submitted for IO while - * checkpointing. [j_list_lock] - */ - struct journal_head *t_checkpoint_io_list; - /* * Doubly-linked circular list of metadata buffers being * shadowed by log IO. The IO buffers on the iobuf list and From b98dba273a0e47dbfade89c9af73c5b012a4eabb Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Tue, 6 Jun 2023 21:59:25 +0800 Subject: [PATCH 3/9] jbd2: remove journal_clean_one_cp_list() journal_clean_one_cp_list() and journal_shrink_one_cp_list() are almost the same, so merge them into journal_shrink_one_cp_list(), remove the nr_to_scan parameter, always scan and try to free the whole checkpoint list. Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20230606135928.434610-4-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o --- fs/jbd2/checkpoint.c | 85 ++++++++++--------------------------- include/trace/events/jbd2.h | 12 ++---- 2 files changed, 26 insertions(+), 71 deletions(-) diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c index 55d6efdbea64..b94f847960c2 100644 --- a/fs/jbd2/checkpoint.c +++ b/fs/jbd2/checkpoint.c @@ -348,19 +348,24 @@ int jbd2_cleanup_journal_tail(journal_t *journal) /* Checkpoint list management */ /* - * journal_clean_one_cp_list + * journal_shrink_one_cp_list * - * Find all the written-back checkpoint buffers in the given list and - * release them. If 'destroy' is set, clean all buffers unconditionally. + * Find all the written-back checkpoint buffers in the given list + * and try to release them. If the whole transaction is released, set + * the 'released' parameter. Return the number of released checkpointed + * buffers. * * Called with j_list_lock held. - * Returns 1 if we freed the transaction, 0 otherwise. */ -static int journal_clean_one_cp_list(struct journal_head *jh, bool destroy) +static unsigned long journal_shrink_one_cp_list(struct journal_head *jh, + bool destroy, bool *released) { struct journal_head *last_jh; struct journal_head *next_jh = jh; + unsigned long nr_freed = 0; + int ret; + *released = false; if (!jh) return 0; @@ -370,52 +375,6 @@ static int journal_clean_one_cp_list(struct journal_head *jh, bool destroy) next_jh = jh->b_cpnext; if (!destroy && __cp_buffer_busy(jh)) - return 0; - - if (__jbd2_journal_remove_checkpoint(jh)) - return 1; - /* - * This function only frees up some memory - * if possible so we dont have an obligation - * to finish processing. Bail out if preemption - * requested: - */ - if (need_resched()) - return 0; - } while (jh != last_jh); - - return 0; -} - -/* - * journal_shrink_one_cp_list - * - * Find 'nr_to_scan' written-back checkpoint buffers in the given list - * and try to release them. If the whole transaction is released, set - * the 'released' parameter. Return the number of released checkpointed - * buffers. - * - * Called with j_list_lock held. - */ -static unsigned long journal_shrink_one_cp_list(struct journal_head *jh, - unsigned long *nr_to_scan, - bool *released) -{ - struct journal_head *last_jh; - struct journal_head *next_jh = jh; - unsigned long nr_freed = 0; - int ret; - - if (!jh || *nr_to_scan == 0) - return 0; - - last_jh = jh->b_cpprev; - do { - jh = next_jh; - next_jh = jh->b_cpnext; - - (*nr_to_scan)--; - if (__cp_buffer_busy(jh)) continue; nr_freed++; @@ -427,7 +386,7 @@ static unsigned long journal_shrink_one_cp_list(struct journal_head *jh, if (need_resched()) break; - } while (jh != last_jh && *nr_to_scan); + } while (jh != last_jh); return nr_freed; } @@ -445,11 +404,11 @@ unsigned long jbd2_journal_shrink_checkpoint_list(journal_t *journal, unsigned long *nr_to_scan) { transaction_t *transaction, *last_transaction, *next_transaction; - bool released; + bool __maybe_unused released; tid_t first_tid = 0, last_tid = 0, next_tid = 0; tid_t tid = 0; unsigned long nr_freed = 0; - unsigned long nr_scanned = *nr_to_scan; + unsigned long freed; again: spin_lock(&journal->j_list_lock); @@ -478,10 +437,11 @@ again: transaction = next_transaction; next_transaction = transaction->t_cpnext; tid = transaction->t_tid; - released = false; - nr_freed += journal_shrink_one_cp_list(transaction->t_checkpoint_list, - nr_to_scan, &released); + freed = journal_shrink_one_cp_list(transaction->t_checkpoint_list, + false, &released); + nr_freed += freed; + (*nr_to_scan) -= min(*nr_to_scan, freed); if (*nr_to_scan == 0) break; if (need_resched() || spin_needbreak(&journal->j_list_lock)) @@ -502,9 +462,8 @@ again: if (*nr_to_scan && next_tid) goto again; out: - nr_scanned -= *nr_to_scan; trace_jbd2_shrink_checkpoint_list(journal, first_tid, tid, last_tid, - nr_freed, nr_scanned, next_tid); + nr_freed, next_tid); return nr_freed; } @@ -520,7 +479,7 @@ out: void __jbd2_journal_clean_checkpoint_list(journal_t *journal, bool destroy) { transaction_t *transaction, *last_transaction, *next_transaction; - int ret; + bool released; transaction = journal->j_checkpoint_transactions; if (!transaction) @@ -531,8 +490,8 @@ void __jbd2_journal_clean_checkpoint_list(journal_t *journal, bool destroy) do { transaction = next_transaction; next_transaction = transaction->t_cpnext; - ret = journal_clean_one_cp_list(transaction->t_checkpoint_list, - destroy); + journal_shrink_one_cp_list(transaction->t_checkpoint_list, + destroy, &released); /* * This function only frees up some memory if possible so we * dont have an obligation to finish processing. Bail out if @@ -545,7 +504,7 @@ void __jbd2_journal_clean_checkpoint_list(journal_t *journal, bool destroy) * avoids pointless scanning of transactions which still * weren't checkpointed. */ - if (!ret) + if (!released) return; } while (transaction != last_transaction); } diff --git a/include/trace/events/jbd2.h b/include/trace/events/jbd2.h index 8f5ee380d309..5646ae15a957 100644 --- a/include/trace/events/jbd2.h +++ b/include/trace/events/jbd2.h @@ -462,11 +462,9 @@ TRACE_EVENT(jbd2_shrink_scan_exit, TRACE_EVENT(jbd2_shrink_checkpoint_list, TP_PROTO(journal_t *journal, tid_t first_tid, tid_t tid, tid_t last_tid, - unsigned long nr_freed, unsigned long nr_scanned, - tid_t next_tid), + unsigned long nr_freed, tid_t next_tid), - TP_ARGS(journal, first_tid, tid, last_tid, nr_freed, - nr_scanned, next_tid), + TP_ARGS(journal, first_tid, tid, last_tid, nr_freed, next_tid), TP_STRUCT__entry( __field(dev_t, dev) @@ -474,7 +472,6 @@ TRACE_EVENT(jbd2_shrink_checkpoint_list, __field(tid_t, tid) __field(tid_t, last_tid) __field(unsigned long, nr_freed) - __field(unsigned long, nr_scanned) __field(tid_t, next_tid) ), @@ -484,15 +481,14 @@ TRACE_EVENT(jbd2_shrink_checkpoint_list, __entry->tid = tid; __entry->last_tid = last_tid; __entry->nr_freed = nr_freed; - __entry->nr_scanned = nr_scanned; __entry->next_tid = next_tid; ), TP_printk("dev %d,%d shrink transaction %u-%u(%u) freed %lu " - "scanned %lu next transaction %u", + "next transaction %u", MAJOR(__entry->dev), MINOR(__entry->dev), __entry->first_tid, __entry->tid, __entry->last_tid, - __entry->nr_freed, __entry->nr_scanned, __entry->next_tid) + __entry->nr_freed, __entry->next_tid) ); #endif /* _TRACE_JBD2_H */ From e34c8dd238d0c9368b746480f313055f5bab5040 Mon Sep 17 00:00:00 2001 From: Zhihao Cheng Date: Tue, 6 Jun 2023 21:59:26 +0800 Subject: [PATCH 4/9] jbd2: Fix wrongly judgement for buffer head removing while doing checkpoint Following process, jbd2_journal_commit_transaction // there are several dirty buffer heads in transaction->t_checkpoint_list P1 wb_workfn jbd2_log_do_checkpoint if (buffer_locked(bh)) // false __block_write_full_page trylock_buffer(bh) test_clear_buffer_dirty(bh) if (!buffer_dirty(bh)) __jbd2_journal_remove_checkpoint(jh) if (buffer_write_io_error(bh)) // false >> bh IO error occurs << jbd2_cleanup_journal_tail __jbd2_update_log_tail jbd2_write_superblock // The bh won't be replayed in next mount. , which could corrupt the ext4 image, fetch a reproducer in [Link]. Since writeback process clears buffer dirty after locking buffer head, we can fix it by try locking buffer and check dirtiness while buffer is locked, the buffer head can be removed if it is neither dirty nor locked. Link: https://bugzilla.kernel.org/show_bug.cgi?id=217490 Fixes: 470decc613ab ("[PATCH] jbd2: initial copy of files from jbd") Signed-off-by: Zhihao Cheng Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20230606135928.434610-5-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o --- fs/jbd2/checkpoint.c | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-) diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c index b94f847960c2..42b34cab64fb 100644 --- a/fs/jbd2/checkpoint.c +++ b/fs/jbd2/checkpoint.c @@ -204,20 +204,6 @@ restart: jh = transaction->t_checkpoint_list; bh = jh2bh(jh); - /* - * The buffer may be writing back, or flushing out in the - * last couple of cycles, or re-adding into a new transaction, - * need to check it again until it's unlocked. - */ - if (buffer_locked(bh)) { - get_bh(bh); - spin_unlock(&journal->j_list_lock); - wait_on_buffer(bh); - /* the journal_head may have gone by now */ - BUFFER_TRACE(bh, "brelse"); - __brelse(bh); - goto retry; - } if (jh->b_transaction != NULL) { transaction_t *t = jh->b_transaction; tid_t tid = t->t_tid; @@ -252,7 +238,22 @@ restart: spin_lock(&journal->j_list_lock); goto restart; } - if (!buffer_dirty(bh)) { + if (!trylock_buffer(bh)) { + /* + * The buffer is locked, it may be writing back, or + * flushing out in the last couple of cycles, or + * re-adding into a new transaction, need to check + * it again until it's unlocked. + */ + get_bh(bh); + spin_unlock(&journal->j_list_lock); + wait_on_buffer(bh); + /* the journal_head may have gone by now */ + BUFFER_TRACE(bh, "brelse"); + __brelse(bh); + goto retry; + } else if (!buffer_dirty(bh)) { + unlock_buffer(bh); BUFFER_TRACE(bh, "remove from checkpoint"); /* * If the transaction was released or the checkpoint @@ -262,6 +263,7 @@ restart: !transaction->t_checkpoint_list) goto out; } else { + unlock_buffer(bh); /* * We are about to write the buffer, it could be * raced by some other transaction shrink or buffer From 46f881b5b1758dc4a35fba4a643c10717d0cf427 Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Tue, 6 Jun 2023 21:59:27 +0800 Subject: [PATCH 5/9] jbd2: fix a race when checking checkpoint buffer busy Before removing checkpoint buffer from the t_checkpoint_list, we have to check both BH_Dirty and BH_Lock bits together to distinguish buffers have not been or were being written back. But __cp_buffer_busy() checks them separately, it first check lock state and then check dirty, the window between these two checks could be raced by writing back procedure, which locks buffer and clears buffer dirty before I/O completes. So it cannot guarantee checkpointing buffers been written back to disk if some error happens later. Finally, it may clean checkpoint transactions and lead to inconsistent filesystem. jbd2_journal_forget() and __journal_try_to_free_buffer() also have the same problem (journal_unmap_buffer() escape from this issue since it's running under the buffer lock), so fix them through introducing a new helper to try holding the buffer lock and remove really clean buffer. Link: https://bugzilla.kernel.org/show_bug.cgi?id=217490 Cc: stable@vger.kernel.org Suggested-by: Jan Kara Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20230606135928.434610-6-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o --- fs/jbd2/checkpoint.c | 38 +++++++++++++++++++++++++++++++++++--- fs/jbd2/transaction.c | 17 +++++------------ include/linux/jbd2.h | 1 + 3 files changed, 41 insertions(+), 15 deletions(-) diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c index 42b34cab64fb..9ec91017a7f3 100644 --- a/fs/jbd2/checkpoint.c +++ b/fs/jbd2/checkpoint.c @@ -376,11 +376,15 @@ static unsigned long journal_shrink_one_cp_list(struct journal_head *jh, jh = next_jh; next_jh = jh->b_cpnext; - if (!destroy && __cp_buffer_busy(jh)) - continue; + if (destroy) { + ret = __jbd2_journal_remove_checkpoint(jh); + } else { + ret = jbd2_journal_try_remove_checkpoint(jh); + if (ret < 0) + continue; + } nr_freed++; - ret = __jbd2_journal_remove_checkpoint(jh); if (ret) { *released = true; break; @@ -616,6 +620,34 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh) return 1; } +/* + * Check the checkpoint buffer and try to remove it from the checkpoint + * list if it's clean. Returns -EBUSY if it is not clean, returns 1 if + * it frees the transaction, 0 otherwise. + * + * This function is called with j_list_lock held. + */ +int jbd2_journal_try_remove_checkpoint(struct journal_head *jh) +{ + struct buffer_head *bh = jh2bh(jh); + + if (!trylock_buffer(bh)) + return -EBUSY; + if (buffer_dirty(bh)) { + unlock_buffer(bh); + return -EBUSY; + } + unlock_buffer(bh); + + /* + * Buffer is clean and the IO has finished (we held the buffer + * lock) so the checkpoint is done. We can safely remove the + * buffer from this transaction. + */ + JBUFFER_TRACE(jh, "remove from checkpoint list"); + return __jbd2_journal_remove_checkpoint(jh); +} + /* * journal_insert_checkpoint: put a committed buffer onto a checkpoint * list so that we know when it is safe to clean the transaction out of diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c index 18611241f451..6ef5022949c4 100644 --- a/fs/jbd2/transaction.c +++ b/fs/jbd2/transaction.c @@ -1784,8 +1784,7 @@ int jbd2_journal_forget(handle_t *handle, struct buffer_head *bh) * Otherwise, if the buffer has been written to disk, * it is safe to remove the checkpoint and drop it. */ - if (!buffer_dirty(bh)) { - __jbd2_journal_remove_checkpoint(jh); + if (jbd2_journal_try_remove_checkpoint(jh) >= 0) { spin_unlock(&journal->j_list_lock); goto drop; } @@ -2112,20 +2111,14 @@ __journal_try_to_free_buffer(journal_t *journal, struct buffer_head *bh) jh = bh2jh(bh); - if (buffer_locked(bh) || buffer_dirty(bh)) - goto out; - if (jh->b_next_transaction != NULL || jh->b_transaction != NULL) - goto out; + return; spin_lock(&journal->j_list_lock); - if (jh->b_cp_transaction != NULL) { - /* written-back checkpointed metadata buffer */ - JBUFFER_TRACE(jh, "remove from checkpoint list"); - __jbd2_journal_remove_checkpoint(jh); - } + /* Remove written-back checkpointed metadata buffer */ + if (jh->b_cp_transaction != NULL) + jbd2_journal_try_remove_checkpoint(jh); spin_unlock(&journal->j_list_lock); -out: return; } diff --git a/include/linux/jbd2.h b/include/linux/jbd2.h index bd660aac8e07..44c298aa58d4 100644 --- a/include/linux/jbd2.h +++ b/include/linux/jbd2.h @@ -1443,6 +1443,7 @@ extern void jbd2_journal_commit_transaction(journal_t *); void __jbd2_journal_clean_checkpoint_list(journal_t *journal, bool destroy); unsigned long jbd2_journal_shrink_checkpoint_list(journal_t *journal, unsigned long *nr_to_scan); int __jbd2_journal_remove_checkpoint(struct journal_head *); +int jbd2_journal_try_remove_checkpoint(struct journal_head *jh); void jbd2_journal_destroy_checkpoint(journal_t *journal); void __jbd2_journal_insert_checkpoint(struct journal_head *, transaction_t *); From 3c55097c553c49deab60ac62c83ef17565004a97 Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Tue, 6 Jun 2023 21:59:28 +0800 Subject: [PATCH 6/9] jbd2: remove __journal_try_to_free_buffer() __journal_try_to_free_buffer() has only one caller and it's logic is much simple now, so just remove it and open code in jbd2_journal_try_to_free_buffers(). Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20230606135928.434610-7-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o --- fs/jbd2/transaction.c | 31 +++++++------------------------ 1 file changed, 7 insertions(+), 24 deletions(-) diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c index 6ef5022949c4..4d1fda1f7143 100644 --- a/fs/jbd2/transaction.c +++ b/fs/jbd2/transaction.c @@ -2099,29 +2099,6 @@ void jbd2_journal_unfile_buffer(journal_t *journal, struct journal_head *jh) __brelse(bh); } -/* - * Called from jbd2_journal_try_to_free_buffers(). - * - * Called under jh->b_state_lock - */ -static void -__journal_try_to_free_buffer(journal_t *journal, struct buffer_head *bh) -{ - struct journal_head *jh; - - jh = bh2jh(bh); - - if (jh->b_next_transaction != NULL || jh->b_transaction != NULL) - return; - - spin_lock(&journal->j_list_lock); - /* Remove written-back checkpointed metadata buffer */ - if (jh->b_cp_transaction != NULL) - jbd2_journal_try_remove_checkpoint(jh); - spin_unlock(&journal->j_list_lock); - return; -} - /** * jbd2_journal_try_to_free_buffers() - try to free page buffers. * @journal: journal for operation @@ -2179,7 +2156,13 @@ bool jbd2_journal_try_to_free_buffers(journal_t *journal, struct folio *folio) continue; spin_lock(&jh->b_state_lock); - __journal_try_to_free_buffer(journal, bh); + if (!jh->b_transaction && !jh->b_next_transaction) { + spin_lock(&journal->j_list_lock); + /* Remove written-back checkpointed metadata buffer */ + if (jh->b_cp_transaction != NULL) + jbd2_journal_try_remove_checkpoint(jh); + spin_unlock(&journal->j_list_lock); + } spin_unlock(&jh->b_state_lock); jbd2_journal_put_journal_head(jh); if (buffer_jbd(bh)) From 6909cf5c4101214f4305a62d582a5b93c7e1eb9a Mon Sep 17 00:00:00 2001 From: Eric Whitney Date: Mon, 22 May 2023 14:15:20 -0400 Subject: [PATCH 7/9] ext4: correct inline offset when handling xattrs in inode body When run on a file system where the inline_data feature has been enabled, xfstests generic/269, generic/270, and generic/476 cause ext4 to emit error messages indicating that inline directory entries are corrupted. This occurs because the inline offset used to locate inline directory entries in the inode body is not updated when an xattr in that shared region is deleted and the region is shifted in memory to recover the space it occupied. If the deleted xattr precedes the system.data attribute, which points to the inline directory entries, that attribute will be moved further up in the region. The inline offset continues to point to whatever is located in system.data's former location, with unfortunate effects when used to access directory entries or (presumably) inline data in the inode body. Cc: stable@kernel.org Signed-off-by: Eric Whitney Link: https://lore.kernel.org/r/20230522181520.1570360-1-enwlinux@gmail.com Signed-off-by: Theodore Ts'o --- fs/ext4/xattr.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index 321e3a888c20..05151d61b00b 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -1782,6 +1782,20 @@ static int ext4_xattr_set_entry(struct ext4_xattr_info *i, memmove(here, (void *)here + size, (void *)last - (void *)here + sizeof(__u32)); memset(last, 0, size); + + /* + * Update i_inline_off - moved ibody region might contain + * system.data attribute. Handling a failure here won't + * cause other complications for setting an xattr. + */ + if (!is_block && ext4_has_inline_data(inode)) { + ret = ext4_find_inline_data_nolock(inode); + if (ret) { + ext4_warning_inode(inode, + "unable to update i_inline_off"); + goto out; + } + } } else if (s->not_found) { /* Insert new name. */ size_t size = EXT4_XATTR_LEN(name_len); From 5d5460fa7932bed3a9082a6a8852cfbdb46acbe8 Mon Sep 17 00:00:00 2001 From: Ojaswin Mujoo Date: Fri, 9 Jun 2023 16:04:03 +0530 Subject: [PATCH 8/9] ext4: fix off by one issue in ext4_mb_choose_next_group_best_avail() In ext4_mb_choose_next_group_best_avail(), we want the start order to be 1 less than goal length and the min_order to be, at max, 1 more than the original length. This commit fixes an off by one issue that arose due to the fact that 1 << fls(n) > (n). After all the processing: order = 1 order below goal len min_order = maximum of the three:- - order - trim_order - 1 order below B2C(s_stripe) - 1 order above original len Cc: stable@kernel.org Fixes: 33122aa930 ("ext4: Add allocation criteria 1.5 (CR1_5)") Signed-off-by: Ojaswin Mujoo Link: https://lore.kernel.org/r/20230609103403.112807-1-ojaswin@linux.ibm.com Signed-off-by: Theodore Ts'o --- fs/ext4/mballoc.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-) diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index a2475b8c9fb5..456150ef6111 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -1006,14 +1006,11 @@ static void ext4_mb_choose_next_group_best_avail(struct ext4_allocation_context * fls() instead since we need to know the actual length while modifying * goal length. */ - order = fls(ac->ac_g_ex.fe_len); + order = fls(ac->ac_g_ex.fe_len) - 1; min_order = order - sbi->s_mb_best_avail_max_trim_order; if (min_order < 0) min_order = 0; - if (1 << min_order < ac->ac_o_ex.fe_len) - min_order = fls(ac->ac_o_ex.fe_len) + 1; - if (sbi->s_stripe > 0) { /* * We are assuming that stripe size is always a multiple of @@ -1021,9 +1018,16 @@ static void ext4_mb_choose_next_group_best_avail(struct ext4_allocation_context */ num_stripe_clusters = EXT4_NUM_B2C(sbi, sbi->s_stripe); if (1 << min_order < num_stripe_clusters) - min_order = fls(num_stripe_clusters); + /* + * We consider 1 order less because later we round + * up the goal len to num_stripe_clusters + */ + min_order = fls(num_stripe_clusters) - 1; } + if (1 << min_order < ac->ac_o_ex.fe_len) + min_order = fls(ac->ac_o_ex.fe_len); + for (i = order; i >= min_order; i--) { int frag_order; /* From 9d3de7ee192a6a253f475197fe4d2e2af10a731f Mon Sep 17 00:00:00 2001 From: Ojaswin Mujoo Date: Sat, 22 Jul 2023 22:45:24 +0530 Subject: [PATCH 9/9] ext4: fix rbtree traversal bug in ext4_mb_use_preallocated During allocations, while looking for preallocations(PA) in the per inode rbtree, we can't do a direct traversal of the tree because ext4_mb_discard_group_preallocation() can paralelly mark the pa deleted and that can cause direct traversal to skip some entries. This was leading to a BUG_ON() being hit [1] when we missed a PA that could satisfy our request and ultimately tried to create a new PA that would overlap with the missed one. To makes sure we handle that case while still keeping the performance of the rbtree, we make use of the fact that the only pa that could possibly overlap the original goal start is the one that satisfies the below conditions: 1. It must have it's logical start immediately to the left of (ie less than) original logical start. 2. It must not be deleted To find this pa we use the following traversal method: 1. Descend into the rbtree normally to find the immediate neighboring PA. Here we keep descending irrespective of if the PA is deleted or if it overlaps with our request etc. The goal is to find an immediately adjacent PA. 2. If the found PA is on right of original goal, use rb_prev() to find the left adjacent PA. 3. Check if this PA is deleted and keep moving left with rb_prev() until a non deleted PA is found. 4. This is the PA we are looking for. Now we can check if it can satisfy the original request and proceed accordingly. This approach also takes care of having deleted PAs in the tree. (While we are at it, also fix a possible overflow bug in calculating the end of a PA) [1] https://lore.kernel.org/linux-ext4/CA+G9fYv2FRpLqBZf34ZinR8bU2_ZRAUOjKAD3+tKRFaEQHtt8Q@mail.gmail.com/ Cc: stable@kernel.org # 6.4 Fixes: 3872778664e3 ("ext4: Use rbtrees to manage PAs instead of inode i_prealloc_list") Signed-off-by: Ojaswin Mujoo Reported-by: Naresh Kamboju Reviewed-by: Ritesh Harjani (IBM) ritesh.list@gmail.com Tested-by: Ritesh Harjani (IBM) ritesh.list@gmail.com Link: https://lore.kernel.org/r/edd2efda6a83e6343c5ace9deea44813e71dbe20.1690045963.git.ojaswin@linux.ibm.com Signed-off-by: Theodore Ts'o --- fs/ext4/mballoc.c | 164 +++++++++++++++++++++++++++++++++++++--------- 1 file changed, 134 insertions(+), 30 deletions(-) diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 456150ef6111..21b903fe546e 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -4765,8 +4765,8 @@ ext4_mb_use_preallocated(struct ext4_allocation_context *ac) int order, i; struct ext4_inode_info *ei = EXT4_I(ac->ac_inode); struct ext4_locality_group *lg; - struct ext4_prealloc_space *tmp_pa, *cpa = NULL; - ext4_lblk_t tmp_pa_start, tmp_pa_end; + struct ext4_prealloc_space *tmp_pa = NULL, *cpa = NULL; + loff_t tmp_pa_end; struct rb_node *iter; ext4_fsblk_t goal_block; @@ -4774,47 +4774,151 @@ ext4_mb_use_preallocated(struct ext4_allocation_context *ac) if (!(ac->ac_flags & EXT4_MB_HINT_DATA)) return false; - /* first, try per-file preallocation */ + /* + * first, try per-file preallocation by searching the inode pa rbtree. + * + * Here, we can't do a direct traversal of the tree because + * ext4_mb_discard_group_preallocation() can paralelly mark the pa + * deleted and that can cause direct traversal to skip some entries. + */ read_lock(&ei->i_prealloc_lock); + + if (RB_EMPTY_ROOT(&ei->i_prealloc_node)) { + goto try_group_pa; + } + + /* + * Step 1: Find a pa with logical start immediately adjacent to the + * original logical start. This could be on the left or right. + * + * (tmp_pa->pa_lstart never changes so we can skip locking for it). + */ for (iter = ei->i_prealloc_node.rb_node; iter; iter = ext4_mb_pa_rb_next_iter(ac->ac_o_ex.fe_logical, - tmp_pa_start, iter)) { + tmp_pa->pa_lstart, iter)) { tmp_pa = rb_entry(iter, struct ext4_prealloc_space, pa_node.inode_node); + } - /* all fields in this condition don't change, - * so we can skip locking for them */ - tmp_pa_start = tmp_pa->pa_lstart; - tmp_pa_end = tmp_pa->pa_lstart + EXT4_C2B(sbi, tmp_pa->pa_len); + /* + * Step 2: The adjacent pa might be to the right of logical start, find + * the left adjacent pa. After this step we'd have a valid tmp_pa whose + * logical start is towards the left of original request's logical start + */ + if (tmp_pa->pa_lstart > ac->ac_o_ex.fe_logical) { + struct rb_node *tmp; + tmp = rb_prev(&tmp_pa->pa_node.inode_node); - /* original request start doesn't lie in this PA */ - if (ac->ac_o_ex.fe_logical < tmp_pa_start || - ac->ac_o_ex.fe_logical >= tmp_pa_end) - continue; - - /* non-extent files can't have physical blocks past 2^32 */ - if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS)) && - (tmp_pa->pa_pstart + EXT4_C2B(sbi, tmp_pa->pa_len) > - EXT4_MAX_BLOCK_FILE_PHYS)) { + if (tmp) { + tmp_pa = rb_entry(tmp, struct ext4_prealloc_space, + pa_node.inode_node); + } else { /* - * Since PAs don't overlap, we won't find any - * other PA to satisfy this. + * If there is no adjacent pa to the left then finding + * an overlapping pa is not possible hence stop searching + * inode pa tree + */ + goto try_group_pa; + } + } + + BUG_ON(!(tmp_pa && tmp_pa->pa_lstart <= ac->ac_o_ex.fe_logical)); + + /* + * Step 3: If the left adjacent pa is deleted, keep moving left to find + * the first non deleted adjacent pa. After this step we should have a + * valid tmp_pa which is guaranteed to be non deleted. + */ + for (iter = &tmp_pa->pa_node.inode_node;; iter = rb_prev(iter)) { + if (!iter) { + /* + * no non deleted left adjacent pa, so stop searching + * inode pa tree + */ + goto try_group_pa; + } + tmp_pa = rb_entry(iter, struct ext4_prealloc_space, + pa_node.inode_node); + spin_lock(&tmp_pa->pa_lock); + if (tmp_pa->pa_deleted == 0) { + /* + * We will keep holding the pa_lock from + * this point on because we don't want group discard + * to delete this pa underneath us. Since group + * discard is anyways an ENOSPC operation it + * should be okay for it to wait a few more cycles. */ break; - } - - /* found preallocated blocks, use them */ - spin_lock(&tmp_pa->pa_lock); - if (tmp_pa->pa_deleted == 0 && tmp_pa->pa_free && - likely(ext4_mb_pa_goal_check(ac, tmp_pa))) { - atomic_inc(&tmp_pa->pa_count); - ext4_mb_use_inode_pa(ac, tmp_pa); + } else { spin_unlock(&tmp_pa->pa_lock); - read_unlock(&ei->i_prealloc_lock); - return true; } - spin_unlock(&tmp_pa->pa_lock); } + + BUG_ON(!(tmp_pa && tmp_pa->pa_lstart <= ac->ac_o_ex.fe_logical)); + BUG_ON(tmp_pa->pa_deleted == 1); + + /* + * Step 4: We now have the non deleted left adjacent pa. Only this + * pa can possibly satisfy the request hence check if it overlaps + * original logical start and stop searching if it doesn't. + */ + tmp_pa_end = (loff_t)tmp_pa->pa_lstart + EXT4_C2B(sbi, tmp_pa->pa_len); + + if (ac->ac_o_ex.fe_logical >= tmp_pa_end) { + spin_unlock(&tmp_pa->pa_lock); + goto try_group_pa; + } + + /* non-extent files can't have physical blocks past 2^32 */ + if (!(ext4_test_inode_flag(ac->ac_inode, EXT4_INODE_EXTENTS)) && + (tmp_pa->pa_pstart + EXT4_C2B(sbi, tmp_pa->pa_len) > + EXT4_MAX_BLOCK_FILE_PHYS)) { + /* + * Since PAs don't overlap, we won't find any other PA to + * satisfy this. + */ + spin_unlock(&tmp_pa->pa_lock); + goto try_group_pa; + } + + if (tmp_pa->pa_free && likely(ext4_mb_pa_goal_check(ac, tmp_pa))) { + atomic_inc(&tmp_pa->pa_count); + ext4_mb_use_inode_pa(ac, tmp_pa); + spin_unlock(&tmp_pa->pa_lock); + read_unlock(&ei->i_prealloc_lock); + return true; + } else { + /* + * We found a valid overlapping pa but couldn't use it because + * it had no free blocks. This should ideally never happen + * because: + * + * 1. When a new inode pa is added to rbtree it must have + * pa_free > 0 since otherwise we won't actually need + * preallocation. + * + * 2. An inode pa that is in the rbtree can only have it's + * pa_free become zero when another thread calls: + * ext4_mb_new_blocks + * ext4_mb_use_preallocated + * ext4_mb_use_inode_pa + * + * 3. Further, after the above calls make pa_free == 0, we will + * immediately remove it from the rbtree in: + * ext4_mb_new_blocks + * ext4_mb_release_context + * ext4_mb_put_pa + * + * 4. Since the pa_free becoming 0 and pa_free getting removed + * from tree both happen in ext4_mb_new_blocks, which is always + * called with i_data_sem held for data allocations, we can be + * sure that another process will never see a pa in rbtree with + * pa_free == 0. + */ + WARN_ON_ONCE(tmp_pa->pa_free == 0); + } + spin_unlock(&tmp_pa->pa_lock); +try_group_pa: read_unlock(&ei->i_prealloc_lock); /* can we use group allocation? */