[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZxuZ-wGccb3yhBAD@andrea>
Date: Fri, 25 Oct 2024 16:15:51 +0300
From: Andrea Parri <parri.andrea@...il.com>
To: Hernan Ponce de Leon <hernan.poncedeleon@...weicloud.com>
Cc: puranjay@...nel.org, paulmck@...nel.org, bpf@...r.kernel.org,
lkmm@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: Some observations (results) on BPF acquire and release
> > BPF R+release+fence
> > {
> > 0:r2=x; 0:r4=y;
> > 1:r2=y; 1:r4=x; 1:r6=l;
> > }
> > P0 | P1 ;
> > r1 = 1 | r1 = 2 ;
> > *(u32 *)(r2 + 0) = r1 | *(u32 *)(r2 + 0) = r1 ;
> > r3 = 1 | r5 = atomic_fetch_add((u32 *)(r6 + 0), r5) ;
> > store_release((u32 *)(r4 + 0), r3) | r3 = *(u32 *)(r4 + 0) ;
> > exists ([y]=2 /\ 1:r3=0)
> >
> > This "exists" condition is not satisfiable according to the BPF model;
> > however, if we adopt the "natural"/intended(?) PowerPC implementations
> > of the synchronization primitives above (aka, with store_release() -->
> > LWSYNC and atomic_fetch_add() --> SYNC ; [...] ), then we see that the
> > condition in question becomes (architecturally) satisfiable on PowerPC
> > (although I'm not aware of actual observations on PowerPC hardware).
>
> Are the resulting PPC tests available somewhere?
My data go back to the LKMM paper, cf. e.g. the R+pooncerelease+fencembonceonce
entry at https://diy.inria.fr/linux/hard.html#unseen .
Andrea
Powered by blists - more mailing lists