Re: Some -serious- BPF-related litmus tests

[Boqun Feng <boqun.feng@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 5386 bytes --]

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!
> > > > > 
> > > > > 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:
> > > > > 
> > > > > https://lore.kernel.org/bpf/20200517195727.279322-3-andriin@fb.com/
> > > > > 
> > > > > Thoughts?
> > > > 
> > > > I find:
> > > > 
> > > > 	smp_wmb()
> > > > 	smp_store_release()
> > > > 
> > > > a _very_ weird construct. What is that supposed to even do?
> > > 
> > > Indeed, it looks like one or the other of those is redundant (depending
> > > on the context).
> > 
> > Probably.  Peter instead asked what it was supposed to even do.  ;-)
> 
> 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. Maybe
> there was some reason, but might be that I was just over-cautious. See reply
> on patch thread as well ([0]).
> 
>   [0] https://lore.kernel.org/bpf/CAEf4Bza26AbRMtWcoD5+TFhnmnU6p5YJ8zO+SoAJCDtp1jVhcQ@mail.gmail.com/
> 

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 = 1;
 		} else if (rLen == 1) {
 			rCx = 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 = 1;
 		} else if (rLen == 1) {
 			rCx = 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 = 1;
 		} else if (rLen == 1) {
 			rCx = rCx + 1;
-			smp_store_release(cx, rCx);
+			WRITE_ONCE(*cx, rCx);
 		}
 	}
 
@@ -68,7 +68,7 @@ P0(int *len1, int *cx, int *px)
 			rFail = 1;
 		} else if (rLen == 1) {
 			rCx = 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 = 1;
 		} else if (rLen == 1) {
 			rCx = rCx + 1;
-			smp_store_release(cx, rCx);
+			WRITE_ONCE(*cx, rCx);
 		}
 	}
 
@@ -65,7 +65,7 @@ P0(int *len1, int *len2, int *cx, int *px)
 			rFail = 1;
 		} else if (rLen == 1) {
 			rCx = rCx + 1;
-			smp_store_release(cx, rCx);
+			WRITE_ONCE(*cx, rCx);
 		}
 	}
 }

> 
> > 
> > > Also, what use is a spinlock that is accessed in only one thread?
> > 
> > Multiple writers synchronize via the spinlock in this case.  I am
> > guessing that his larger 16-hour test contended this spinlock.
> 
> 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.
> 
> > 
> > > 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?
> > 
> > Agreed, and I responded to that effect to his original patch:
> > 
> > https://lore.kernel.org/bpf/20200522003433.GG2869@paulmck-ThinkPad-P72/
> 
> Yep, makes sense, I'll will move.
> 
> > 
> > 							Thanx, Paul
> > 
> 

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]