Re: [PATCH 3/4] ftrace: abstract DYNAMIC_FTRACE_WITH_ARGS accesses

[Mark Rutland]

On Wed, Oct 26, 2022 at 12:17:54AM +0900, Masami Hiramatsu wrote:
> On Tue, 25 Oct 2022 11:30:38 +0100
> Mark Rutland <mark.rutland@xxxxxxx> wrote:
> 
> > On Tue, Oct 25, 2022 at 05:40:01PM +0900, Masami Hiramatsu wrote:
> > > Hi Mark,
> > > 
> > > On Mon, 24 Oct 2022 15:08:45 +0100
> > > Mark Rutland <mark.rutland@xxxxxxx> wrote:
> > > 
> > > > In subsequent patches we'll arrange for architectures to have an
> > > > ftrace_regs which is entirely distinct from pt_regs. In preparation for
> > > > this, we need to minimize the use of pt_regs to where strictly necessary
> > > > in the core ftrace code.
> > > > 
> > > > This patch adds new ftrace_regs_{get,set}_*() helpers which can be used
> > > > to manipulate ftrace_regs. When CONFIG_HAVE_DYNAMIC_FTRACE_WITH_ARGS=y,
> > > > these can always be used on any ftrace_regs, and when
> > > > CONFIG_HAVE_DYNAMIC_FTRACE_WITH_ARGS=n these can be used when regs are
> > > > available. A new ftrace_regs_has_args(fregs) helper is added which code
> > > > can use to check when these are usable.
> > > 
> > > Can you also add the ftrace_regs_query_register_offset() as a wrapper of
> > > regs_query_register_offset()? I would like to use it for fprobe_events.
> > 
> > Sure!
> > 
> > Just to check, with FTRACE_WITH_REGS, does fprobe always sample the full
> > pt_regs, or do callers also need to check ftrace_regs_has_args(fregs)?
> 
> No, please return -ENOENT or any error value if the given register
> is not saved on arm64.

Sure, that's what I intend to implement for arm64. I'll use -EINVAL to match
the existing regs_query_register_offset() logic.

> Others will just return regs_query_register_offset(&fregs->regs, name). That
> is enough at this moment. Later we can improve it.

Sorry, what I was trying to ask was whether fprobe currently always set
FTRACE_OPS_FL_SAVE_REGS (which AFAICT it does); so I now agree that's
sufficient -- sorry for the noise!

Thanks,
Mark.