Re: [PATCH v3 09/10] tracing/probes: Add fprobe events for tracing function entry and exit.

[Google]

On Wed, 29 Mar 2023 08:14:54 -0400
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:

> On Thu,  2 Feb 2023 00:57:13 +0900
> "Masami Hiramatsu (Google)" <mhiramat@xxxxxxxxxx> wrote:
> 
> > From: Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx>
> > 
> > Add fprobe events for tracing function entry and exit instead of kprobe
> > events. With this change, we can continue to trace function entry/exit
> > even if the CONFIG_KPROBES_ON_FTRACE is not available. Since
> > CONFIG_KPROBES_ON_FTRACE requires the CONFIG_DYNAMIC_FTRACE_WITH_REGS,
> > it is not available if the architecture only supports
> > CONFIG_DYNAMIC_FTRACE_WITH_ARGS. And that means kprobe events can not
> > probe function entry/exit effectively on such architecture.
> > But this can be solved if the dynamic events supports fprobe events.
> > 
> > The fprobe event is a new dynamic events which is only for the function
> > (symbol) entry and exit. This event accepts non register fetch arguments
> > so that user can trace the function arguments and return values.
> > 
> 
> Hi Masami,
> 
> After applying this patch I get a bunch of these:
> 
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c:117:1: error: redefinition of ‘fetch_store_strlen_user’
>   117 | fetch_store_strlen_user(unsigned long addr)
>       | ^~~~~~~~~~~~~~~~~~~~~~~
> In file included from /work/git/linux-trace.git/kernel/trace/trace_fprobe.c:16:
> /work/git/linux-trace.git/kernel/trace/trace_probe_kernel.h:15:1: note: previous definition of ‘fetch_store_strlen_user’ with type ‘int(long unsigned int)’
>    15 | fetch_store_strlen_user(unsigned long addr)
>       | ^~~~~~~~~~~~~~~~~~~~~~~
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c: In function ‘fetch_store_strlen_user’:
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c:119:16: error: implicit declaration of function ‘kern_fetch_store_strlen_user’; did you mean ‘fetch_store_strlen_user’? [-Werror=implicit-function-declaration]
>   119 |         return kern_fetch_store_strlen_user(addr);
>       |                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
>       |                fetch_store_strlen_user
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c: At top level:
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c:124:1: error: redefinition of ‘fetch_store_strlen’
>   124 | fetch_store_strlen(unsigned long addr)
>       | ^~~~~~~~~~~~~~~~~~
> /work/git/linux-trace.git/kernel/trace/trace_probe_kernel.h:32:1: note: previous definition of ‘fetch_store_strlen’ with type ‘int(long unsigned int)’
>    32 | fetch_store_strlen(unsigned long addr)
>       | ^~~~~~~~~~~~~~~~~~
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c: In function ‘fetch_store_strlen’:
> /work/git/linux-trace.git/kernel/trace/trace_fprobe.c:126:16: error: implicit declaration of function ‘kern_fetch_store_strlen’; did you mean ‘fetch_store_strlen’? [-Werror=implicit-function-declaration]
>   126 |         return kern_fetch_store_strlen(addr);
>       |                ^~~~~~~~~~~~~~~~~~~~~~~
>       |                fetch_store_strlen
> 
> 
> Can you rebase it on the latest changes (on top of trace/for-next)?

OK, let me update it.

> 
> BTW, I've applied patches 1-8 and I'm currently running them through my
> tests. So if you do rebase, just send patches 9 and 10. I'm hoping to
> post a for-next series later today, that will include those other
> patches.

Thanks!

> 
> -- Steve


-- 
Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx>