[Bug sanitizer/94910] New: detect_stack_use_after_return=1 is much slower than clang's

[rafael at espindo dot la]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94910

            Bug ID: 94910
           Summary: detect_stack_use_after_return=1 is much slower than
                    clang's
           Product: gcc
           Version: 9.3.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: sanitizer
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rafael at espindo dot la
                CC: dodji at gcc dot gnu.org, dvyukov at gcc dot gnu.org,
                    jakub at gcc dot gnu.org, kcc at gcc dot gnu.org, marxin at 
gcc dot gnu.org
  Target Milestone: ---

The test I am using is from https://github.com/scylladb/seastar/. It can be
build with

$ cmake -DCMAKE_BUILD_TYPE=Debug -GNinja <src>
$ ninja tests/unit/chunked_fifo_test

And can be run with:

Clang:

$ time ASAN_OPTIONS=detect_stack_use_after_return=1
./tests/unit/chunked_fifo_test -t chunked_fifo_big
...
1.80s user 0.02s system 99% cpu 1.826 total
$ time ASAN_OPTIONS=detect_stack_use_after_return=0
./tests/unit/chunked_fifo_test -t chunked_fifo_big
...
1.67s user 0.01s system 99% cpu 1.691 total

GCC:

$ time ASAN_OPTIONS=detect_stack_use_after_return=1
./tests/unit/chunked_fifo_test -t chunked_fifo_big
89.12s user 0.03s system 99% cpu 1:29.34 total
$ time ASAN_OPTIONS=detect_stack_use_after_return=0
./tests/unit/chunked_fifo_test -t chunked_fifo_big
1.32s user 0.02s system 99% cpu 1.350 total


So while plain asan is faster with gcc, enabling detect_stack_use_after_return
makes it much slower.