Björn Töpel [off-list ref] writes:

Andy Chiu [off-list ref] writes:

quoted

We use an AUIPC+JALR pair to jump into a ftrace trampoline. Since
instruction fetch can break down to 4 byte at a time, it is impossible
to update two instructions without a race. In order to mitigate it, we
initialize the patchable entry to AUIPC + NOP4. Then, the run-time code
patching can change NOP4 to JALR to eable/disable ftrcae from a

                                      enable        ftrace

quoted

function. This limits the reach of each ftrace entry to +-2KB displacing
from ftrace_caller.

Starting from the trampoline, we add a level of indirection for it to
reach ftrace caller target. Now, it loads the target address from a
memory location, then perform the jump. This enable the kernel to update
the target atomically.

The +-2K limit is for direct calls, right?

...and this I would say breaks DIRECT_CALLS (which should be implemented
using call_ops later)?

Thinking a bit more, and re-reading the series.

This series is good work, and it's a big improvement for DYNAMIC_FTRACE,
but

+int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
+{
+	unsigned long distance, orig_addr;
+
+	orig_addr = (unsigned long)&ftrace_caller;
+	distance = addr > orig_addr ? addr - orig_addr : orig_addr - addr;
+	if (distance > JALR_RANGE)
+		return -EINVAL;
+
+	return __ftrace_modify_call(rec->ip, addr, false);
+}
+

breaks WITH_DIRECT_CALLS. The direct trampoline will *never* be within
the JALR_RANGE.

Unless we're happy with a break (I'm not) -- I really think Puranjay's
CALL_OPS patch needs to be baked in in the series!


Björn