From mboxrd@z Thu Jan 1 00:00:00 1970 From: Boqun Feng Subject: Re: Some -serious- BPF-related litmus tests Date: Mon, 25 May 2020 22:53:25 +0800 Message-ID: <20200525145325.GB2066@tardis> References: <20200522003850.GA32698@paulmck-ThinkPad-P72> <20200522094407.GK325280@hirez.programming.kicks-ass.net> <20200522143201.GB32434@rowland.harvard.edu> <20200522174352.GJ2869@paulmck-ThinkPad-P72> <006e2bc6-7516-1584-3d8c-e253211c157e@fb.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="XF85m9dhOBO43t/C" Return-path: Content-Disposition: inline In-Reply-To: <006e2bc6-7516-1584-3d8c-e253211c157e@fb.com> Sender: linux-kernel-owner@vger.kernel.org To: Andrii Nakryiko Cc: paulmck@kernel.org, Alan Stern , Peter Zijlstra , parri.andrea@gmail.com, will@kernel.org, npiggin@gmail.com, dhowells@redhat.com, j.alglave@ucl.ac.uk, luc.maranget@inria.fr, akiyks@gmail.com, dlustig@nvidia.com, joel@joelfernandes.org, linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org, "andrii.nakryiko@gmail.com" List-Id: linux-arch.vger.kernel.org --XF85m9dhOBO43t/C Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Hi Andrii, On Fri, May 22, 2020 at 12:38:21PM -0700, Andrii Nakryiko wrote: > On 5/22/20 10:43 AM, Paul E. McKenney wrote: > > On Fri, May 22, 2020 at 10:32:01AM -0400, Alan Stern wrote: > > > On Fri, May 22, 2020 at 11:44:07AM +0200, Peter Zijlstra wrote: > > > > On Thu, May 21, 2020 at 05:38:50PM -0700, Paul E. McKenney wrote: > > > > > Hello! > > > > >=20 > > > > > Just wanted to call your attention to some pretty cool and pretty= serious > > > > > litmus tests that Andrii did as part of his BPF ring-buffer work: > > > > >=20 > > > > > https://lore.kernel.org/bpf/20200517195727.279322-3-andriin@fb.co= m/ > > > > >=20 > > > > > Thoughts? > > > >=20 > > > > I find: > > > >=20 > > > > smp_wmb() > > > > smp_store_release() > > > >=20 > > > > a _very_ weird construct. What is that supposed to even do? > > >=20 > > > Indeed, it looks like one or the other of those is redundant (dependi= ng > > > on the context). > >=20 > > Probably. Peter instead asked what it was supposed to even do. ;-) >=20 > I agree, I think smp_wmb() is redundant here. Can't remember why I thought > that it's necessary, this algorithm went through a bunch of iterations, > starting as completely lockless, also using READ_ONCE/WRITE_ONCE at some > point, and settling on smp_read_acquire/smp_store_release, eventually. Ma= ybe > there was some reason, but might be that I was just over-cautious. See re= ply > on patch thread as well ([0]). >=20 > [0] https://lore.kernel.org/bpf/CAEf4Bza26AbRMtWcoD5+TFhnmnU6p5YJ8zO+So= AJCDtp1jVhcQ@mail.gmail.com/ >=20 While we are at it, could you explain a bit on why you use smp_store_release() on consumer_pos? I ask because IIUC, consumer_pos is only updated at consumer side, and there is no other write at consumer side that we want to order with the write to consumer_pos. So I fail to find why smp_store_release() is necessary. I did the following modification on litmus tests, and I didn't see different results (on States) between two versions of litmus tests. Regards, Boqun ---------------------->8 diff --git a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus index cafd17afe11e..255b23be7fa9 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus @@ -46,7 +46,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } diff --git a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus index 84f660598015..5eecf14f87d1 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus @@ -44,7 +44,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } diff --git a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus index 900104c4933b..54da1e5d7ec0 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus @@ -50,7 +50,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } =20 @@ -68,7 +68,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } diff --git a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus index 83372e9eb079..fd19433f4d9b 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus @@ -47,7 +47,7 @@ P0(int *len1, int *len2, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } =20 @@ -65,7 +65,7 @@ P0(int *len1, int *len2, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } >=20 > >=20 > > > Also, what use is a spinlock that is accessed in only one thread? > >=20 > > Multiple writers synchronize via the spinlock in this case. I am > > guessing that his larger 16-hour test contended this spinlock. >=20 > Yes, spinlock is for coordinating multiple producers. 2p1c cases (bounded > and unbounded) rely on this already. 1p1c cases are sort of subsets (but > very fast to verify) checking only consumer/producer interaction. >=20 > >=20 > > > Finally, I doubt that these tests belong under tools/memory-model. > > > Shouldn't they go under the new Documentation/ directory for litmus > > > tests? And shouldn't the patch update a README file? > >=20 > > Agreed, and I responded to that effect to his original patch: > >=20 > > https://lore.kernel.org/bpf/20200522003433.GG2869@paulmck-ThinkPad-P72/ >=20 > Yep, makes sense, I'll will move. >=20 > >=20 > > Thanx, Paul > >=20 >=20 --XF85m9dhOBO43t/C Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQEzBAABCAAdFiEEj5IosQTPz8XU1wRHSXnow7UH+rgFAl7L2+EACgkQSXnow7UH +riErgf/Wzb00Sg8hmGkgBVyjxygRexYtRXtEokGPsNEdDv76PThrk72IDghZJfW lBbGr8OwiUApz1WipXNWj68D+zakm1jELPtlCw/6zVn7wkpyRCFPeGUrtXjs3YIa VWMrwcQv4NZMFsCZ5gql3ORE6sYsRHRy2NuEwdNqhn87xpfbsUEJU4B+EjwsYL2t Jx+4VTnlDLs2CEWvMX3RVvcD248guxEyqz8dP9KDAZuIqu7+WU8z54iZ3i5xPusl ytaDcK2kji9+AlsPNHmVmrcWOS6jgK9dDGoZ/HMxbFT64x7DGOCWcfXdzk3rmL1W GWueVqzLuTDAeQdGMl64XqLmcJPDwQ== =Q0lL -----END PGP SIGNATURE----- --XF85m9dhOBO43t/C-- From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from lindbergh.monkeyblade.net ([23.128.96.19]:50228 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S2404003AbgEYOxn (ORCPT ); Mon, 25 May 2020 10:53:43 -0400 Date: Mon, 25 May 2020 22:53:25 +0800 From: Boqun Feng Subject: Re: Some -serious- BPF-related litmus tests Message-ID: <20200525145325.GB2066@tardis> References: <20200522003850.GA32698@paulmck-ThinkPad-P72> <20200522094407.GK325280@hirez.programming.kicks-ass.net> <20200522143201.GB32434@rowland.harvard.edu> <20200522174352.GJ2869@paulmck-ThinkPad-P72> <006e2bc6-7516-1584-3d8c-e253211c157e@fb.com> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="XF85m9dhOBO43t/C" Content-Disposition: inline In-Reply-To: <006e2bc6-7516-1584-3d8c-e253211c157e@fb.com> Sender: linux-arch-owner@vger.kernel.org List-ID: To: Andrii Nakryiko Cc: paulmck@kernel.org, Alan Stern , Peter Zijlstra , parri.andrea@gmail.com, will@kernel.org, npiggin@gmail.com, dhowells@redhat.com, j.alglave@ucl.ac.uk, luc.maranget@inria.fr, akiyks@gmail.com, dlustig@nvidia.com, joel@joelfernandes.org, linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org, "andrii.nakryiko@gmail.com" Message-ID: <20200525145325.Kig-EbCANkD1ENLywJwCLa-l3gRoSTihhhHQ4DZPx0U@z> --XF85m9dhOBO43t/C Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Hi Andrii, On Fri, May 22, 2020 at 12:38:21PM -0700, Andrii Nakryiko wrote: > On 5/22/20 10:43 AM, Paul E. McKenney wrote: > > On Fri, May 22, 2020 at 10:32:01AM -0400, Alan Stern wrote: > > > On Fri, May 22, 2020 at 11:44:07AM +0200, Peter Zijlstra wrote: > > > > On Thu, May 21, 2020 at 05:38:50PM -0700, Paul E. McKenney wrote: > > > > > Hello! > > > > >=20 > > > > > Just wanted to call your attention to some pretty cool and pretty= serious > > > > > litmus tests that Andrii did as part of his BPF ring-buffer work: > > > > >=20 > > > > > https://lore.kernel.org/bpf/20200517195727.279322-3-andriin@fb.co= m/ > > > > >=20 > > > > > Thoughts? > > > >=20 > > > > I find: > > > >=20 > > > > smp_wmb() > > > > smp_store_release() > > > >=20 > > > > a _very_ weird construct. What is that supposed to even do? > > >=20 > > > Indeed, it looks like one or the other of those is redundant (dependi= ng > > > on the context). > >=20 > > Probably. Peter instead asked what it was supposed to even do. ;-) >=20 > I agree, I think smp_wmb() is redundant here. Can't remember why I thought > that it's necessary, this algorithm went through a bunch of iterations, > starting as completely lockless, also using READ_ONCE/WRITE_ONCE at some > point, and settling on smp_read_acquire/smp_store_release, eventually. Ma= ybe > there was some reason, but might be that I was just over-cautious. See re= ply > on patch thread as well ([0]). >=20 > [0] https://lore.kernel.org/bpf/CAEf4Bza26AbRMtWcoD5+TFhnmnU6p5YJ8zO+So= AJCDtp1jVhcQ@mail.gmail.com/ >=20 While we are at it, could you explain a bit on why you use smp_store_release() on consumer_pos? I ask because IIUC, consumer_pos is only updated at consumer side, and there is no other write at consumer side that we want to order with the write to consumer_pos. So I fail to find why smp_store_release() is necessary. I did the following modification on litmus tests, and I didn't see different results (on States) between two versions of litmus tests. Regards, Boqun ---------------------->8 diff --git a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus index cafd17afe11e..255b23be7fa9 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+1p1c+bounded.litmus @@ -46,7 +46,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } diff --git a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus index 84f660598015..5eecf14f87d1 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+1p1c+unbound.litmus @@ -44,7 +44,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } diff --git a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus index 900104c4933b..54da1e5d7ec0 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+2p1c+bounded.litmus @@ -50,7 +50,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } =20 @@ -68,7 +68,7 @@ P0(int *len1, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } diff --git a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus b/= tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus index 83372e9eb079..fd19433f4d9b 100644 --- a/tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus +++ b/tools/memory-model/litmus-tests/mpsc-rb+2p1c+unbound.litmus @@ -47,7 +47,7 @@ P0(int *len1, int *len2, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } =20 @@ -65,7 +65,7 @@ P0(int *len1, int *len2, int *cx, int *px) rFail =3D 1; } else if (rLen =3D=3D 1) { rCx =3D rCx + 1; - smp_store_release(cx, rCx); + WRITE_ONCE(*cx, rCx); } } } >=20 > >=20 > > > Also, what use is a spinlock that is accessed in only one thread? > >=20 > > Multiple writers synchronize via the spinlock in this case. I am > > guessing that his larger 16-hour test contended this spinlock. >=20 > Yes, spinlock is for coordinating multiple producers. 2p1c cases (bounded > and unbounded) rely on this already. 1p1c cases are sort of subsets (but > very fast to verify) checking only consumer/producer interaction. >=20 > >=20 > > > Finally, I doubt that these tests belong under tools/memory-model. > > > Shouldn't they go under the new Documentation/ directory for litmus > > > tests? And shouldn't the patch update a README file? > >=20 > > Agreed, and I responded to that effect to his original patch: > >=20 > > https://lore.kernel.org/bpf/20200522003433.GG2869@paulmck-ThinkPad-P72/ >=20 > Yep, makes sense, I'll will move. >=20 > >=20 > > Thanx, Paul > >=20 >=20 --XF85m9dhOBO43t/C Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQEzBAABCAAdFiEEj5IosQTPz8XU1wRHSXnow7UH+rgFAl7L2+EACgkQSXnow7UH +riErgf/Wzb00Sg8hmGkgBVyjxygRexYtRXtEokGPsNEdDv76PThrk72IDghZJfW lBbGr8OwiUApz1WipXNWj68D+zakm1jELPtlCw/6zVn7wkpyRCFPeGUrtXjs3YIa VWMrwcQv4NZMFsCZ5gql3ORE6sYsRHRy2NuEwdNqhn87xpfbsUEJU4B+EjwsYL2t Jx+4VTnlDLs2CEWvMX3RVvcD248guxEyqz8dP9KDAZuIqu7+WU8z54iZ3i5xPusl ytaDcK2kji9+AlsPNHmVmrcWOS6jgK9dDGoZ/HMxbFT64x7DGOCWcfXdzk3rmL1W GWueVqzLuTDAeQdGMl64XqLmcJPDwQ== =Q0lL -----END PGP SIGNATURE----- --XF85m9dhOBO43t/C--