tracing: Fix buffer overflow in trace event filter

It was reported that if an trace event was larger than a page
 and was filtered, that it caused memory corruption. The reason
 is that filtered events first go into a buffer to test the filter
 before being written into the ring buffer. Unfortunately,
 this write did not check the size.
 -----BEGIN PGP SIGNATURE-----
 
 iIoEABYIADIWIQRRSw7ePDh/lE+zeZMp5XQQmuv6qgUCYCaSHBQccm9zdGVkdEBn
 b29kbWlzLm9yZwAKCRAp5XQQmuv6qqOpAQCUSlZdBxLzs87zeHgXbkMudWvCYSbA
 mndzddqtxPXlXwEAsRnO8BERyZnasEdXnJ98JJwQaFFYH0dBCA2pTU2onQc=
 =NokV
 -----END PGP SIGNATURE-----

Merge tag 'trace-v5.11-rc7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace

Pull tracing fix from Steven Rostedt:
 "Fix buffer overflow in trace event filter.

  It was reported that if an trace event was larger than a page and was
  filtered, that it caused memory corruption. The reason is that
  filtered events first go into a buffer to test the filter before being
  written into the ring buffer. Unfortunately, this write did not check
  the size"

* tag 'trace-v5.11-rc7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  tracing: Check length before giving out the filter buffer
This commit is contained in:
Linus Torvalds 2021-02-12 11:16:17 -08:00
commit e77a6817d4

View File

@ -2745,7 +2745,7 @@ trace_event_buffer_lock_reserve(struct trace_buffer **current_rb,
(entry = this_cpu_read(trace_buffered_event))) {
/* Try to use the per cpu buffer first */
val = this_cpu_inc_return(trace_buffered_event_cnt);
if (val == 1) {
if ((len < (PAGE_SIZE - sizeof(*entry))) && val == 1) {
trace_event_setup(entry, type, flags, pc);
entry->array[0] = len;
return entry;