Re: [GIT PULL] ftrace: Fixes for v6.13

[Linus Torvalds]

On Sat, 14 Dec 2024 at 19:11, Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
>
> That is, even though:
>
>   TP_fast_assign(
>       __entry->ipv6 = ptr_to_ipv6;
>   )
>
>   TP_printk("ipv6=%pI6", __entry->ipv6)

Again, you're not talking about the code that is being fixed.

I'm complaining about the mess that you try to fix in commit
ff54c8dc3f7a ("tracing: Fix trace output when pointer hash is
disabled"), which is that eldritch horror called
trace_check_vprintf().

You are describing something entirely different: the admittedly
similarly named test_event_printk(), called from the event init code.

I think trace_check_vprintf() should be removed entirely.

And I think test_event_printk() could probably be expanded to handle
at least some cases of '%s' (because, as you say, pointers to various
static allocations are fine for it).

And no, I don't know the code all that well, so maybe I'm overlooking
something, but trace_check_vprintf() really is a horror. And all your
arguments for it seem to be missing the mark and are actually
arguments for test_event_printk().

               Linus