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

From: Google
Date: Tue Oct 25 2022 - 11:18:10 EST

Next message: Chao Peng: "[PATCH v9 0/8] KVM: mm: fd-based approach for supporting KVM"
Previous message: Petr Mladek: "Re: [PATCH printk v2 27/38] printk: console_flush_all: use srcu console list iterator"
In reply to: Google: "Re: [PATCH 3/4] ftrace: abstract DYNAMIC_FTRACE_WITH_ARGS accesses"
Next in thread: Mark Rutland: "Re: [PATCH 3/4] ftrace: abstract DYNAMIC_FTRACE_WITH_ARGS accesses"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

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

> I ask because if neither of those are the case, with FTRACE_WITH_REGS,
> ftrace_regs_query_register_offset() would accept names of registers which might
> not have been sampled, and could give offsets to uninitialized memory.

Currently fprobe depends on CONFIG_HAVE_DYNAMIC_FTRACE_WITH_REGS, but
in the future, I will move it on WITH_ARGS.

> Atop that, I'm not exactly sure what to implement for powerpc/s390/x86 here. If
> those might be used without a full pt_regs, I think
> ftrace_regs_query_register_offset() should also take the fregs as a parameter
> and use that to check which registers are available.
>
> ... does that make sense to you?

Yeah, that is OK. I think only arm64 changes the ftrace_regs not wraps
pt_regs. So there is no problem even if we access the empty register.
Only arm64 implementation is different, so it should have different
implementation.

Thank you,

>
> Thanks,
> Mark.
>
> >
> > Thank you,
> >
> > >
> > > Co-developed-by: Florent Revest <revest@xxxxxxxxxxxx>
> > > Signed-off-by: Florent Revest <revest@xxxxxxxxxxxx>
> > > Signed-off-by: Mark Rutland <mark.rutland@xxxxxxx>
> > > Cc: Masami Hiramatsu <mhiramat@xxxxxxxxxx>
> > > Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
> > > ---
> > > arch/powerpc/include/asm/ftrace.h | 17 +++++++++++++++++
> > > arch/s390/include/asm/ftrace.h | 17 +++++++++++++++++
> > > arch/x86/include/asm/ftrace.h | 14 ++++++++++++++
> > > include/linux/ftrace.h | 27 +++++++++++++++++++++++++++
> > > kernel/trace/Kconfig | 6 +++---
> > > 5 files changed, 78 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/arch/powerpc/include/asm/ftrace.h b/arch/powerpc/include/asm/ftrace.h
> > > index c3eb48f67566..faecb20d78bf 100644
> > > --- a/arch/powerpc/include/asm/ftrace.h
> > > +++ b/arch/powerpc/include/asm/ftrace.h
> > > @@ -44,6 +44,23 @@ ftrace_regs_set_instruction_pointer(struct ftrace_regs *fregs,
> > > regs_set_return_ip(&fregs->regs, ip);
> > > }
> > >
> > > +static __always_inline unsigned long
> > > +ftrace_regs_get_instruction_pointer(const struct ftrace_regs *fregs)
> > > +{
> > > + return instruction_pointer(&fregs->regs)
> > > +}
> > > +
> > > +#define ftrace_regs_get_argument(fregs, n) \
> > > + regs_get_kernel_argument(&(fregs)->regs, n)
> > > +#define ftrace_regs_get_stack_pointer(fregs) \
> > > + kernel_stack_pointer(&(fregs)->regs)
> > > +#define ftrace_regs_return_value(fregs) \
> > > + regs_return_value(&(fregs)->regs)
> > > +#define ftrace_regs_set_return_value(fregs, ret) \
> > > + regs_set_return_value(&(fregs)->regs, ret)
> > > +#define ftrace_override_function_with_return(fregs) \
> > > + override_function_with_return(&(fregs)->regs)
> > > +
> > > struct ftrace_ops;
> > >
> > > #define ftrace_graph_func ftrace_graph_func
> > > diff --git a/arch/s390/include/asm/ftrace.h b/arch/s390/include/asm/ftrace.h
> > > index b8957882404f..5fdc806458aa 100644
> > > --- a/arch/s390/include/asm/ftrace.h
> > > +++ b/arch/s390/include/asm/ftrace.h
> > > @@ -54,6 +54,12 @@ static __always_inline struct pt_regs *arch_ftrace_get_regs(struct ftrace_regs *
> > > return NULL;
> > > }
> > >
> > > +static __always_inline unsigned long
> > > +ftrace_regs_get_instruction_pointer(const struct ftrace_regs *fregs)
> > > +{
> > > + return fregs->regs.psw.addr;
> > > +}
> > > +
> > > static __always_inline void
> > > ftrace_regs_set_instruction_pointer(struct ftrace_regs *fregs,
> > > unsigned long ip)
> > > @@ -61,6 +67,17 @@ ftrace_regs_set_instruction_pointer(struct ftrace_regs *fregs,
> > > fregs->regs.psw.addr = ip;
> > > }
> > >
> > > +#define ftrace_regs_get_argument(fregs, n) \
> > > + regs_get_kernel_argument(&(fregs)->regs, n)
> > > +#define ftrace_regs_get_stack_pointer(fregs) \
> > > + kernel_stack_pointer(&(fregs)->regs)
> > > +#define ftrace_regs_return_value(fregs) \
> > > + regs_return_value(&(fregs)->regs)
> > > +#define ftrace_regs_set_return_value(fregs, ret) \
> > > + regs_set_return_value(&(fregs)->regs, ret)
> > > +#define ftrace_override_function_with_return(fregs) \
> > > + override_function_with_return(&(fregs)->regs)
> > > +
> > > /*
> > > * When an ftrace registered caller is tracing a function that is
> > > * also set by a register_ftrace_direct() call, it needs to be
> > > diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h
> > > index b73e858bd96f..b3737b42e8a1 100644
> > > --- a/arch/x86/include/asm/ftrace.h
> > > +++ b/arch/x86/include/asm/ftrace.h
> > > @@ -51,6 +51,20 @@ arch_ftrace_get_regs(struct ftrace_regs *fregs)
> > > #define ftrace_regs_set_instruction_pointer(fregs, _ip) \
> > > do { (fregs)->regs.ip = (_ip); } while (0)
> > >
> > > +#define ftrace_regs_get_instruction_pointer(fregs) \
> > > + ((fregs)->regs.ip)
> > > +
> > > +#define ftrace_regs_get_argument(fregs, n) \
> > > + regs_get_kernel_argument(&(fregs)->regs, n)
> > > +#define ftrace_regs_get_stack_pointer(fregs) \
> > > + kernel_stack_pointer(&(fregs)->regs)
> > > +#define ftrace_regs_return_value(fregs) \
> > > + regs_return_value(&(fregs)->regs)
> > > +#define ftrace_regs_set_return_value(fregs, ret) \
> > > + regs_set_return_value(&(fregs)->regs, ret)
> > > +#define ftrace_override_function_with_return(fregs) \
> > > + override_function_with_return(&(fregs)->regs)
> > > +
> > > struct ftrace_ops;
> > > #define ftrace_graph_func ftrace_graph_func
> > > void ftrace_graph_func(unsigned long ip, unsigned long parent_ip,
> > > diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
> > > index e9905f741916..3b13e3c21438 100644
> > > --- a/include/linux/ftrace.h
> > > +++ b/include/linux/ftrace.h
> > > @@ -125,6 +125,33 @@ static __always_inline struct pt_regs *ftrace_get_regs(struct ftrace_regs *fregs
> > > return arch_ftrace_get_regs(fregs);
> > > }
> > >
> > > +/*
> > > + * When true, the ftrace_regs_{get,set}_*() functions may be used on fregs.
> > > + * Note: this can be true even when ftrace_get_regs() cannot provide a pt_regs.
> > > + */
> > > +static __always_inline bool ftrace_regs_has_args(struct ftrace_regs *fregs)
> > > +{
> > > + if (IS_ENABLED(CONFIG_HAVE_DYNAMIC_FTRACE_WITH_ARGS))
> > > + return true;
> > > +
> > > + return ftrace_get_regs(fregs) != NULL;
> > > +}
> > > +
> > > +#ifndef CONFIG_HAVE_DYNAMIC_FTRACE_WITH_ARGS
> > > +#define ftrace_regs_get_instruction_pointer(fregs) \
> > > + instruction_pointer(ftrace_get_regs(fregs))
> > > +#define ftrace_regs_get_argument(fregs, n) \
> > > + regs_get_kernel_argument(ftrace_get_regs(fregs), n)
> > > +#define ftrace_regs_get_stack_pointer(fregs) \
> > > + kernel_stack_pointer(ftrace_get_regs(fregs))
> > > +#define ftrace_regs_return_value(fregs) \
> > > + regs_return_value(ftrace_get_regs(fregs))
> > > +#define ftrace_regs_set_return_value(fregs, ret) \
> > > + regs_set_return_value(ftrace_get_regs(fregs), ret)
> > > +#define ftrace_override_function_with_return(fregs) \
> > > + override_function_with_return(ftrace_get_regs(fregs))
> > > +#endif
> > > +
> > > typedef void (*ftrace_func_t)(unsigned long ip, unsigned long parent_ip,
> > > struct ftrace_ops *op, struct ftrace_regs *fregs);
> > >
> > > diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
> > > index e9e95c790b8e..2c6611c13f99 100644
> > > --- a/kernel/trace/Kconfig
> > > +++ b/kernel/trace/Kconfig
> > > @@ -46,10 +46,10 @@ config HAVE_DYNAMIC_FTRACE_WITH_ARGS
> > > bool
> > > help
> > > If this is set, then arguments and stack can be found from
> > > - the pt_regs passed into the function callback regs parameter
> > > + the ftrace_regs passed into the function callback regs parameter
> > > by default, even without setting the REGS flag in the ftrace_ops.
> > > - This allows for use of regs_get_kernel_argument() and
> > > - kernel_stack_pointer().
> > > + This allows for use of ftrace_regs_get_argument() and
> > > + ftrace_regs_get_stack_pointer().
> > >
> > > config HAVE_DYNAMIC_FTRACE_NO_PATCHABLE
> > > bool
> > > --
> > > 2.30.2
> > >
> >
> >
> > --
> > Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx>

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

Next message: Chao Peng: "[PATCH v9 0/8] KVM: mm: fd-based approach for supporting KVM"
Previous message: Petr Mladek: "Re: [PATCH printk v2 27/38] printk: console_flush_all: use srcu console list iterator"
In reply to: Google: "Re: [PATCH 3/4] ftrace: abstract DYNAMIC_FTRACE_WITH_ARGS accesses"
Next in thread: Mark Rutland: "Re: [PATCH 3/4] ftrace: abstract DYNAMIC_FTRACE_WITH_ARGS accesses"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]