linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Florent Revest <revest@chromium.org>
To: Mark Rutland <mark.rutland@arm.com>
Cc: linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org, linux-trace-kernel@vger.kernel.org,
	bpf@vger.kernel.org, catalin.marinas@arm.com, will@kernel.org,
	rostedt@goodmis.org, mhiramat@kernel.org, ast@kernel.org,
	daniel@iogearbox.net, andrii@kernel.org, kpsingh@kernel.org,
	jolsa@kernel.org, xukuohai@huaweicloud.com
Subject: Re: [PATCH 4/8] ftrace: Store direct called addresses in their ops
Date: Thu, 2 Feb 2023 18:41:45 +0100	[thread overview]
Message-ID: <CABRcYmJQiyZE__JD=xyt-PhAwj0Jpo2kkgmLcm-cD58cryD0ZQ@mail.gmail.com> (raw)
In-Reply-To: <Y9vW99htjOphDXqY@FVFF77S0Q05N.cambridge.arm.com>

On Thu, Feb 2, 2023 at 4:30 PM Mark Rutland <mark.rutland@arm.com> wrote:
>
> On Wed, Feb 01, 2023 at 05:34:16PM +0100, Florent Revest wrote:
> > All direct calls are now registered using the register_ftrace_direct API
> > so each ops can jump to only one direct-called trampoline.
> >
> > By storing the direct called trampoline address directly in the ops we
> > can save one hashmap lookup in the direct call ops and implement arm64
> > direct calls on top of call ops.
> >
> > Signed-off-by: Florent Revest <revest@chromium.org>
> > ---
> >  include/linux/ftrace.h | 3 +++
> >  kernel/trace/ftrace.c  | 6 ++++--
> >  2 files changed, 7 insertions(+), 2 deletions(-)
> >
> > diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
> > index a7dbd307c3a4..84f717f8959e 100644
> > --- a/include/linux/ftrace.h
> > +++ b/include/linux/ftrace.h
> > @@ -321,6 +321,9 @@ struct ftrace_ops {
> >       unsigned long                   trampoline_size;
> >       struct list_head                list;
> >       ftrace_ops_func_t               ops_func;
> > +#ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
> > +     unsigned long                   direct_call;
> > +#endif
> >  #endif
> >  };
> >
> > diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
> > index cb77a0a208c7..b0426de11c45 100644
> > --- a/kernel/trace/ftrace.c
> > +++ b/kernel/trace/ftrace.c
> > @@ -2577,9 +2577,8 @@ ftrace_add_rec_direct(unsigned long ip, unsigned long addr,
> >  static void call_direct_funcs(unsigned long ip, unsigned long pip,
> >                             struct ftrace_ops *ops, struct ftrace_regs *fregs)
> >  {
> > -     unsigned long addr;
> > +     unsigned long addr = ops->direct_call;
> >
> > -     addr = ftrace_find_rec_direct(ip);
> >       if (!addr)
> >               return;
> >
> > @@ -5375,6 +5374,7 @@ int register_ftrace_direct(struct ftrace_ops *ops, unsigned long addr)
> >       ops->func = call_direct_funcs;
> >       ops->flags = MULTI_FLAGS;
> >       ops->trampoline = FTRACE_REGS_ADDR;
> > +     ops->direct_call = addr;
> >
> >       err = register_ftrace_function_nolock(ops);
> >
> > @@ -5445,6 +5445,7 @@ __modify_ftrace_direct(struct ftrace_ops *ops, unsigned long addr)
> >       /* Enable the tmp_ops to have the same functions as the direct ops */
> >       ftrace_ops_init(&tmp_ops);
> >       tmp_ops.func_hash = ops->func_hash;
> > +     tmp_ops.direct_call = addr;
> >
> >       err = register_ftrace_function_nolock(&tmp_ops);
> >       if (err)
> > @@ -5466,6 +5467,7 @@ __modify_ftrace_direct(struct ftrace_ops *ops, unsigned long addr)
> >                       entry->direct = addr;
> >               }
> >       }
> > +     ops->direct_call = addr;
>
> AFAICT we don't synchronize threads when installing the tmp_ops, so IIUC on
> arches with call_ops, there could be a a thread in the middle of ftrace_caller
> which has loaded the ops pointer from the patch-site, but hasn't yet loaded the
> ops::direct_func pointer, and could race with this assignment.
>
> Given that, I think this needs to be:
>
>         WRITE_ONCE(ops->direct_call, addr);
>
> ... in order to avoid the risk of the store being torn, and the ftrace_caller
> trampoline loading a corrupted pointer.

Good point, I'll do that in v2.

  reply	other threads:[~2023-02-02 17:42 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-02-01 16:34 [PATCH 0/8] Add ftrace direct call for arm64 Florent Revest
2023-02-01 16:34 ` [PATCH 1/8] ftrace: Replace uses of _ftrace_direct APIs with _ftrace_direct_multi Florent Revest
2023-02-02 15:01   ` Mark Rutland
2023-02-02 17:37     ` Florent Revest
2023-02-07 15:21       ` Florent Revest
2023-02-07 15:35         ` Steven Rostedt
2023-02-07 16:19           ` Florent Revest
2023-02-01 16:34 ` [PATCH 2/8] ftrace: Remove the legacy _ftrace_direct API Florent Revest
2023-02-02 15:11   ` Mark Rutland
2023-02-01 16:34 ` [PATCH 3/8] ftrace: Rename _ftrace_direct_multi APIs to _ftrace_direct APIs Florent Revest
2023-02-02 15:17   ` Mark Rutland
2023-02-01 16:34 ` [PATCH 4/8] ftrace: Store direct called addresses in their ops Florent Revest
2023-02-02 15:29   ` Mark Rutland
2023-02-02 17:41     ` Florent Revest [this message]
2023-02-01 16:34 ` [PATCH 5/8] ftrace: Make DIRECT_CALLS work WITH_ARGS and !WITH_REGS Florent Revest
2023-02-02 15:54   ` Mark Rutland
2023-02-02 16:56     ` Mark Rutland
2023-02-02 18:19       ` Florent Revest
2023-02-03 10:03         ` Mark Rutland
2023-02-03 11:01           ` Florent Revest
2023-02-02 18:18     ` Florent Revest
2023-02-01 16:34 ` [PATCH 6/8] ftrace: Fix dead loop caused by direct call in ftrace selftest Florent Revest
2023-02-02 19:03   ` Mark Rutland
2023-02-03 12:35     ` Florent Revest
2023-02-03 15:37       ` Mark Rutland
2023-02-06 16:25         ` Florent Revest
2023-02-01 16:34 ` [PATCH 7/8] arm64: ftrace: Add direct call support Florent Revest
2023-02-03 15:34   ` Mark Rutland
2023-02-06 16:25     ` Florent Revest
2023-02-01 16:34 ` [PATCH 8/8] arm64: ftrace: Add direct called trampoline samples support Florent Revest
2023-02-02  8:36 ` [PATCH 0/8] Add ftrace direct call for arm64 Xu Kuohai
2023-02-02 10:50   ` Daniel Borkmann
2023-02-02 17:32     ` Florent Revest
2023-02-02 20:06 ` Steven Rostedt
2023-02-03  9:49   ` Mark Rutland

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CABRcYmJQiyZE__JD=xyt-PhAwj0Jpo2kkgmLcm-cD58cryD0ZQ@mail.gmail.com' \
    --to=revest@chromium.org \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=catalin.marinas@arm.com \
    --cc=daniel@iogearbox.net \
    --cc=jolsa@kernel.org \
    --cc=kpsingh@kernel.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-trace-kernel@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=mhiramat@kernel.org \
    --cc=rostedt@goodmis.org \
    --cc=will@kernel.org \
    --cc=xukuohai@huaweicloud.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).