On 8/31/20 10:16 PM, Alexei Starovoitov wrote:
From: Alexei Starovoitov <ast@xxxxxxxxxx>
Technically the bpf programs can sleep while attached to bpf_lsm_file_mprotect,
but such programs need to access user memory. So they're in might_fault()
category. Which means they cannot be called from file_mprotect lsm hook that
takes write lock on mm->mmap_lock.
Adjust the test accordingly.
Also add might_fault() to __bpf_prog_enter_sleepable() to catch such deadlocks early.
Reported-by: Yonghong Song <yhs@xxxxxx>
Fixes: 1e6c62a88215 ("bpf: Introduce sleepable BPF programs")
Fixes: e68a144547fc ("selftests/bpf: Add sleepable tests")
Signed-off-by: Alexei Starovoitov <ast@xxxxxxxxxx>
---
kernel/bpf/trampoline.c | 1 +
kernel/bpf/verifier.c | 1 -
tools/testing/selftests/bpf/progs/lsm.c | 34 ++++++++++++-------------
3 files changed, 18 insertions(+), 18 deletions(-)
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index c2b76545153c..7dd523a7e32d 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -409,6 +409,7 @@ void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start)
void notrace __bpf_prog_enter_sleepable(void)
{
rcu_read_lock_trace();
+ might_fault();
Makes sense, was wondering about a __might_sleep() but that will cover it internally
too. Applied, thanks!
}
void notrace __bpf_prog_exit_sleepable(void)
