lkml.org 
[lkml]   [2021]   [Jun]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC PATCH 2/6] powerpc/trace: Add support for stack tracer
Naveen N. Rao wrote:
> +
> +unsigned long ftrace_get_traced_func_if_no_stackframe(unsigned long ip, unsigned long *stack)
> +{
> + if (!is_ftrace_entry(ip))
> + return 0;
> +
> + if (IS_ENABLED(CONFIG_PPC32))
> + return stack[11]; /* see MCOUNT_SAVE_FRAME */
> +
> + if (!IS_ENABLED(CONFIG_MPROFILE_KERNEL))
> + return 0;
> +
> + return stack[(STACK_FRAME_OVERHEAD + offsetof(struct pt_regs, nip)) / sizeof(unsigned long)];

Looking at Daniel's patch to address KASAN errors with our stack walk
code in show_stack() [*], I realized that I am not validating the stack
pointer here for the above accesses...

[*] http://lkml.kernel.org/r/20210528074806.1311297-1-dja@axtens.net

> +}
> +
> +#ifdef CONFIG_STACK_TRACER
> +void stack_get_trace(unsigned long traced_ip,
> + unsigned long *stack_ref __maybe_unused,
> + unsigned long stack_size __maybe_unused,
> + int *tracer_frame)
> +{
> + unsigned long sp, newsp, top, ip;
> + int ftrace_call_found = 0;
> + unsigned long *stack;
> + int i = 0;
> +
> + sp = current_stack_frame();
> + top = (unsigned long)task_stack_page(current) + THREAD_SIZE;
> +
> + while (validate_sp(sp, current, STACK_FRAME_OVERHEAD) && i < STACK_TRACE_ENTRIES) {
> + stack = (unsigned long *) sp;
> + newsp = stack[0];
> + ip = stack[STACK_FRAME_LR_SAVE];
> +
> + if (ftrace_call_found) {
> + stack_dump_trace[i] = ip;
> + stack_trace_index[i++] = top - sp;
> + }

And I need to make the above accesses bypass KASAN as well.


- Naveen

\
 
 \ /
  Last update: 2021-06-01 15:53    [W:1.033 / U:0.440 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site