Re: next-20250604: qemu-riscv64 ltp tracing warning kernel trace trace.c ignore_event

[Steven Rostedt <rostedt@goodmis.org>]

On Thu, 5 Jun 2025 18:24:08 +0530
Naresh Kamboju <naresh.kamboju@linaro.org> wrote:

> Regression while running LTP tracing on the Linux next-20250604 and
> next-20250605
> the following kernel warnings found on the qemu-riscv64.
> 
> Regressions found on qemu-riscv64
>  - LTP tracing
> 
> Regression Analysis:
>  - New regression? Yes
>  - Reproducibility? Yes
> 
> First seen on the next-20250604
> Good: next-20250603
> Bad:  next-20250604
> 
> Test regression: qemu-riscv64 ltp tracing warning kernel trace trace.c

Does riscv64 place __func__ in rodata? If not, this may consider the
string unsafe to use in an event. 

> ignore_event
> 
> Reported-by: Linux Kernel Functional Testing <lkft@linaro.org>
> 
> ## Test error
> 
> ------------[ cut here ]------------
> [   28.134981] event 'ma_read' has unsafe pointer field 'fn'

In lib/maple_tree.c we have:

        trace_ma_read(__func__, &mas);

Where it records the address of "__func__" into the ring buffer 'fn' field.

At a much later time (when the ring buffer is read), that pointer is
dereferenced. As rodata doesn't change for the life of the kernel, it
is safe to dereference. But if riscv doesn't place __func__ into
rodata, this will be considered unsafe to use.

-- Steve