summaryrefslogtreecommitdiff
path: root/include/trace
diff options
context:
space:
mode:
authorZhang Yi <yi.zhang@huawei.com>2021-06-10 19:24:37 +0800
committerTheodore Ts'o <tytso@mit.edu>2021-06-24 10:54:49 -0400
commit4ba3fcdde7e36af93610ceb3cc38365b14539865 (patch)
treecfe6b1077ca18cc5614a43446bd015e4f0c93afa /include/trace
parent214eb5a4d8a2032fb9f0711d1b202eb88ee02920 (diff)
jbd2,ext4: add a shrinker to release checkpointed buffers
Current metadata buffer release logic in bdev_try_to_free_page() have a lot of use-after-free issues when umount filesystem concurrently, and it is difficult to fix directly because ext4 is the only user of s_op->bdev_try_to_free_page callback and we may have to add more special refcount or lock that is only used by ext4 into the common vfs layer, which is unacceptable. One better solution is remove the bdev_try_to_free_page callback, but the real problem is we cannot easily release journal_head on the checkpointed buffer, so try_to_free_buffers() cannot release buffers and page under memory pressure, which is more likely to trigger out-of-memory. So we cannot remove the callback directly before we find another way to release journal_head. This patch introduce a shrinker to free journal_head on the checkpointed transaction. After the journal_head got freed, try_to_free_buffers() could free buffer properly. Signed-off-by: Zhang Yi <yi.zhang@huawei.com> Suggested-by: Jan Kara <jack@suse.cz> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20210610112440.3438139-6-yi.zhang@huawei.com Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Diffstat (limited to 'include/trace')
-rw-r--r--include/trace/events/jbd2.h101
1 files changed, 101 insertions, 0 deletions
diff --git a/include/trace/events/jbd2.h b/include/trace/events/jbd2.h
index d16a32867f3a..a4dfe005983d 100644
--- a/include/trace/events/jbd2.h
+++ b/include/trace/events/jbd2.h
@@ -394,6 +394,107 @@ TRACE_EVENT(jbd2_lock_buffer_stall,
__entry->stall_ms)
);
+DECLARE_EVENT_CLASS(jbd2_journal_shrink,
+
+ TP_PROTO(journal_t *journal, unsigned long nr_to_scan,
+ unsigned long count),
+
+ TP_ARGS(journal, nr_to_scan, count),
+
+ TP_STRUCT__entry(
+ __field(dev_t, dev)
+ __field(unsigned long, nr_to_scan)
+ __field(unsigned long, count)
+ ),
+
+ TP_fast_assign(
+ __entry->dev = journal->j_fs_dev->bd_dev;
+ __entry->nr_to_scan = nr_to_scan;
+ __entry->count = count;
+ ),
+
+ TP_printk("dev %d,%d nr_to_scan %lu count %lu",
+ MAJOR(__entry->dev), MINOR(__entry->dev),
+ __entry->nr_to_scan, __entry->count)
+);
+
+DEFINE_EVENT(jbd2_journal_shrink, jbd2_shrink_count,
+
+ TP_PROTO(journal_t *journal, unsigned long nr_to_scan, unsigned long count),
+
+ TP_ARGS(journal, nr_to_scan, count)
+);
+
+DEFINE_EVENT(jbd2_journal_shrink, jbd2_shrink_scan_enter,
+
+ TP_PROTO(journal_t *journal, unsigned long nr_to_scan, unsigned long count),
+
+ TP_ARGS(journal, nr_to_scan, count)
+);
+
+TRACE_EVENT(jbd2_shrink_scan_exit,
+
+ TP_PROTO(journal_t *journal, unsigned long nr_to_scan,
+ unsigned long nr_shrunk, unsigned long count),
+
+ TP_ARGS(journal, nr_to_scan, nr_shrunk, count),
+
+ TP_STRUCT__entry(
+ __field(dev_t, dev)
+ __field(unsigned long, nr_to_scan)
+ __field(unsigned long, nr_shrunk)
+ __field(unsigned long, count)
+ ),
+
+ TP_fast_assign(
+ __entry->dev = journal->j_fs_dev->bd_dev;
+ __entry->nr_to_scan = nr_to_scan;
+ __entry->nr_shrunk = nr_shrunk;
+ __entry->count = count;
+ ),
+
+ TP_printk("dev %d,%d nr_to_scan %lu nr_shrunk %lu count %lu",
+ MAJOR(__entry->dev), MINOR(__entry->dev),
+ __entry->nr_to_scan, __entry->nr_shrunk,
+ __entry->count)
+);
+
+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),
+
+ TP_ARGS(journal, first_tid, tid, last_tid, nr_freed,
+ nr_scanned, next_tid),
+
+ TP_STRUCT__entry(
+ __field(dev_t, dev)
+ __field(tid_t, first_tid)
+ __field(tid_t, tid)
+ __field(tid_t, last_tid)
+ __field(unsigned long, nr_freed)
+ __field(unsigned long, nr_scanned)
+ __field(tid_t, next_tid)
+ ),
+
+ TP_fast_assign(
+ __entry->dev = journal->j_fs_dev->bd_dev;
+ __entry->first_tid = first_tid;
+ __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",
+ MAJOR(__entry->dev), MINOR(__entry->dev),
+ __entry->first_tid, __entry->tid, __entry->last_tid,
+ __entry->nr_freed, __entry->nr_scanned, __entry->next_tid)
+);
+
#endif /* _TRACE_JBD2_H */
/* This part must be outside protection */