Received-SPF: pass (google.com: domain of linux-kernel+bounces-62890-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
References: <20240122062535.8265-1-khuey@kylehuey.com> <CAP045Apecy=G_Wmcw6TMjSDfa3TbkMfFVkzGDJ9xTVksCLkZ0w@mail.gmail.com>
In-Reply-To: <CAP045Apecy=G_Wmcw6TMjSDfa3TbkMfFVkzGDJ9xTVksCLkZ0w@mail.gmail.com>
From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date: Mon, 12 Feb 2024 18:42:05 -0800
Message-ID: <CAADnVQ+tRwMZiPa9Zrf6nD22dfF9MAiqv-1ML5Z2pELFNKa9KQ@mail.gmail.com>
Subject: Re: [PATCH v5 0/4] Combine perf and bpf for fast eval of hw
 breakpoint conditions
To: Kyle Huey <me@kylehuey.com>
Cc: Kyle Huey <khuey@kylehuey.com>, LKML <linux-kernel@vger.kernel.org>, 
	Andrii Nakryiko <andrii.nakryiko@gmail.com>, Jiri Olsa <jolsa@kernel.org>, 
	Namhyung Kim <namhyung@kernel.org>, Marco Elver <elver@google.com>, 
	Yonghong Song <yonghong.song@linux.dev>, "Robert O'Callahan" <robert@ocallahan.org>, 
	bpf <bpf@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Mon, Feb 12, 2024 at 8:37=E2=80=AFAM Kyle Huey <me@kylehuey.com> wrote:
>
> On Sun, Jan 21, 2024 at 10:25=E2=80=AFPM Kyle Huey <me@kylehuey.com> wrot=
e:
> >
> > rr, a userspace record and replay debugger[0], replays asynchronous eve=
nts
> > such as signals and context switches by essentially[1] setting a breakp=
oint
> > at the address where the asynchronous event was delivered during record=
ing
> > with a condition that the program state matches the state when the even=
t
> > was delivered.
> >
> > Currently, rr uses software breakpoints that trap (via ptrace) to the
> > supervisor, and evaluates the condition from the supervisor. If the
> > asynchronous event is delivered in a tight loop (thus requiring the
> > breakpoint condition to be repeatedly evaluated) the overhead can be
> > immense. A patch to rr that uses hardware breakpoints via perf events w=
ith
> > an attached BPF program to reject breakpoint hits where the condition i=
s
> > not satisfied reduces rr's replay overhead by 94% on a pathological (bu=
t a
> > real customer-provided, not contrived) rr trace.
> >
> > The only obstacle to this approach is that while the kernel allows a BP=
F
> > program to suppress sample output when a perf event overflows it does n=
ot
> > suppress signalling the perf event fd or sending the perf event's SIGTR=
AP.
> > This patch set redesigns __perf_overflow_handler() and
> > bpf_overflow_handler() so that the former invokes the latter directly w=
hen
> > appropriate rather than through the generic overflow handler machinery,
> > passes the return code of the BPF program back to __perf_overflow_handl=
er()
> > to allow it to decide whether to execute the regular overflow handler,
> > reorders bpf_overflow_handler() and the side effects of perf event
> > overflow, changes __perf_overflow_handler() to suppress those side effe=
cts
> > if the BPF program returns zero, and adds a selftest.
> >
> > The previous version of this patchset can be found at
> > https://lore.kernel.org/linux-kernel/20240119001352.9396-1-khuey@kylehu=
ey.com/
> >
> > Changes since v4:
> >
> > Patches 1, 2, 3, 4 added various Acked-by.
> >
> > Patch 4 addresses additional nits from Song.
> >
> > v3 of this patchset can be found at
> > https://lore.kernel.org/linux-kernel/20231211045543.31741-1-khuey@kyleh=
uey.com/
> >
> > Changes since v3:
> >
> > Patches 1, 2, 3 added various Acked-by.
> >
> > Patch 4 addresses Song's review comments by dropping signals_expected a=
nd the
> > corresponding ASSERT_OKs, handling errors from signal(), and fixing mul=
tiline
> > comment formatting.
> >
> > v2 of this patchset can be found at
> > https://lore.kernel.org/linux-kernel/20231207163458.5554-1-khuey@kylehu=
ey.com/
> >
> > Changes since v2:
> >
> > Patches 1 and 2 were added from a suggestion by Namhyung Kim to refacto=
r
> > this code to implement this feature in a cleaner way. Patch 2 is separa=
ted
> > for the benefit of the ARM arch maintainers.
> >
> > Patch 3 conceptually supercedes v2's patches 1 and 2, now with a cleane=
r
> > implementation thanks to the earlier refactoring.
> >
> > Patch 4 is v2's patch 3, and addresses review comments about C++ style
> > comments, getting a TRAP_PERF definition into the test, and unnecessary
> > NULL checks.
> >
> > [0] https://rr-project.org/
> > [1] Various optimizations exist to skip as much as execution as possibl=
e
> > before setting a breakpoint, and to determine a set of program state th=
at
> > is practical to check and verify.
>
> Since everyone seems to be satisfied with this now, can we get it into
> bpf-next (or wherever) for 6.9?

The changes look fine, but since they change perf side we need
perf maintainer's ack-s before we can land the patches.
And none of them were cc-ed.
So please resend the whole set and cc
PERFORMANCE EVENTS SUBSYSTEM
M:      Peter Zijlstra <peterz@infradead.org>
M:      Ingo Molnar <mingo@redhat.com>
M:      Arnaldo Carvalho de Melo <acme@kernel.org>
M:      Namhyung Kim <namhyung@kernel.org>