Skip to content

Commit

Permalink
tuple: don't use offset_slot_cache in vinyl threads
Browse files Browse the repository at this point in the history
`key_part::offset_slot_cache` and `key_part::format_epoch` are used for
speeding up tuple field lookup in `tuple_field_raw_by_part()`. These
structure members are accessed and updated without any locks, assuming
this code is executed exclusively in the tx thread. However, this isn't
necessarily true because we also perform tuple field lookups in vinyl
read threads. Apparently, this can result in unexpected races and bugs,
for example:

```
  #1  0x590be9f7eb6d in crash_collect+256
  #2  0x590be9f7f5a9 in crash_signal_cb+100
  #3  0x72b111642520 in __sigaction+80
  #4  0x590bea385e3c in load_u32+35
  #5  0x590bea231eba in field_map_get_offset+46
  #6  0x590bea23242a in tuple_field_raw_by_path+417
  #7  0x590bea23282b in tuple_field_raw_by_part+203
  #8  0x590bea23288c in tuple_field_by_part+91
  #9  0x590bea24cd2d in unsigned long tuple_hint<(field_type)5, false, false>(tuple*, key_def*)+103
  #10 0x590be9d4fba3 in tuple_hint+40
  #11 0x590be9d50acf in vy_stmt_hint+178
  #12 0x590be9d53531 in vy_page_stmt+168
  #13 0x590be9d535ea in vy_page_find_key+142
  #14 0x590be9d545e6 in vy_page_read_cb+210
  #15 0x590be9f94ef0 in cbus_call_perform+44
  #16 0x590be9f94eae in cmsg_deliver+52
  #17 0x590be9f9583e in cbus_process+100
  #18 0x590be9f958a5 in cbus_loop+28
  #19 0x590be9d512da in vy_run_reader_f+381
  #20 0x590be9cb4147 in fiber_cxx_invoke(int (*)(__va_list_tag*), __va_list_tag*)+34
  #21 0x590be9f8b697 in fiber_loop+219
  #22 0x590bea374bb6 in coro_init+120
```

Fix this by skipping this optimization for threads other than tx.

No test is added because reproducing this race is tricky. Ideally, bugs
like this one should be caught by fuzzing tests or thread sanitizers.

Closes #10123

NO_DOC=bug fix
NO_TEST=tested manually with fuzzer

(cherry picked from commit 19d1f1c)
  • Loading branch information
locker committed Jun 13, 2024
1 parent 1bad1af commit 7d90a94
Show file tree
Hide file tree
Showing 2 changed files with 17 additions and 9 deletions.
4 changes: 4 additions & 0 deletions changelogs/unreleased/gh-10123-vy-tuple-field-lookup-fix.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,4 @@
## bugfix/vinyl

* Fixed a bug when internal optimization algorithm caused a crash while a read
thread tried to look up a tuple field (gh-10123).
22 changes: 13 additions & 9 deletions src/box/tuple.h
Original file line number Diff line number Diff line change
Expand Up @@ -1026,19 +1026,23 @@ tuple_field_raw_by_part(struct tuple_format *format, const char *data,
const uint32_t *field_map,
struct key_part *part, int multikey_idx)
{
if (unlikely(part->format_epoch != format->epoch)) {
assert(format->epoch != 0);
part->format_epoch = format->epoch;
/*
* Clear the offset slot cache, since it's stale.
* The cache will be reset by the lookup.
*/
part->offset_slot_cache = TUPLE_OFFSET_SLOT_NIL;
int32_t *offset_slot_cache = NULL;
if (cord_is_main()) {
offset_slot_cache = &part->offset_slot_cache;
if (unlikely(part->format_epoch != format->epoch)) {
assert(format->epoch != 0);
part->format_epoch = format->epoch;
/*
* Clear the offset slot cache, since it's stale.
* The cache will be reset by the lookup.
*/
*offset_slot_cache = TUPLE_OFFSET_SLOT_NIL;
}
}
return tuple_field_raw_by_path(format, data, field_map, part->fieldno,
part->path, part->path_len,
TUPLE_INDEX_BASE,
&part->offset_slot_cache, multikey_idx);
offset_slot_cache, multikey_idx);
}

/**
Expand Down

0 comments on commit 7d90a94

Please sign in to comment.