Hi,
This patchset fixes an endless loop condition that can occur in
bpf_for_each_hash_elem, causing the core to softlock. My understanding is
that a combination of RCU list deletion and insertion introduces the new
element after the iteration cursor and that there is a chance that an RCU
reader may in fact use this new element in iteration. The patch uses a
_safe variant of the macro which gets the next element to iterate before
executing the loop body for the current element.
I have also added a subtest in the for_each selftest that can trigger this
condition without the fix.
Changes since v2:
- Renaming and additional checks in selftests/bpf/prog_tests/for_each.c
Changes since v1:
- Added missing Signed-off-by lines to both patches
Thanks,
Brandon Kammerdiener
Brandon Kammerdiener (2):
bpf: fix possible endless loop in BPF map iteration
selftests/bpf: add test for softlock when modifying hashmap while
iterating
kernel/bpf/hashtab.c | 2 +-
.../selftests/bpf/prog_tests/for_each.c | 37 +++++++++++++++++++
.../bpf/progs/for_each_hash_modify.c | 30 +++++++++++++++
3 files changed, 68 insertions(+), 1 deletion(-)
create mode 100644 tools/testing/selftests/bpf/progs/for_each_hash_modify.c
--
2.49.0
