On Wed, 12 Jun 2024 12:51:30 -0400 Steven Rostedt rostedt@goodmis.org wrote:
[ 100.600222] Hardware name: Radxa ROCK Pi 4B (DT) [ 100.600229] pstate: 800003c5 (Nzcv DAIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--) [ 100.600239] pc : ftrace_ops_test+0x34/0x138
Hmm, could you show the exact line of the above code? Specifically we have:
rcu_assign_pointer(hash.filter_hash, ops->func_hash->filter_hash); rcu_assign_pointer(hash.notrace_hash, ops->func_hash->notrace_hash);
Hmm, it's a NULL pointer dereference at 0x8, so ops is likely not NULL, as func_hash is much farther down. But if func_hash is NULL, filter_hash is at the 0x8 offset.
So now the question is, how did func_hash become NULL. It should always be pointing at something. May have to do with the subops. Will investigate.
[ 100.600258] lr : function_graph_enter+0x144/0x208
I wonder if we need the following patch:
diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c index 8317d1a7f43a..fc205ad167a9 100644 --- a/kernel/trace/fgraph.c +++ b/kernel/trace/fgraph.c @@ -641,7 +641,7 @@ int function_graph_enter(unsigned long ret, unsigned long func, { for_each_set_bit(i, &fgraph_array_bitmask, sizeof(fgraph_array_bitmask) * BITS_PER_BYTE) { - struct fgraph_ops *gops = fgraph_array[i]; + struct fgraph_ops *gops = READ_ONCE(fgraph_array[i]); int save_curr_ret_stack;
if (gops == &fgraph_stub)
Because if the compiler decides to re-read gops from fgraph_array[i] after the above check for the following line that does:
save_curr_ret_stack = current->curr_ret_stack; if (ftrace_ops_test(&gops->ops, func, NULL) && gops->entryfunc(&trace, gops)) bitmap |= BIT(i);
and gops now points to fgraph_stub, it will trigger this bug.
Can you apply the above change and see if the bug goes away?
Thanks,
-- Steve