On Mon, Sep 30, 2024 at 12:57:08PM +0200, Jonas Oberhauser wrote:
> Herd7 transforms successful RMW with Mb tags by inserting smp_mb() fences
> around them. We emulate this by considering imaginary po-edges before the
> RMW read and before the RMW write, and extending the smp_mb() ordering
> rule, which currently only applies to real po edges that would be found
> around a really inserted smp_mb(), also to cases of the only imagined po
> edges.
>
> Reported-by: Viktor Vafeiadis <viktor@mpi-sws.org>
> Suggested-by: Alan Stern <stern@rowland.harvard.edu>
> Signed-off-by: Jonas Oberhauser <jonas.oberhauser@huaweicloud.com>
Reviewed-by: Boqun Feng <boqun.feng@gmail.com>
Regards,
Boqun
> ---
> tools/memory-model/linux-kernel.cat | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/tools/memory-model/linux-kernel.cat b/tools/memory-model/linux-kernel.cat
> index adf3c4f41229..d7e7bf13c831 100644
> --- a/tools/memory-model/linux-kernel.cat
> +++ b/tools/memory-model/linux-kernel.cat
> @@ -34,6 +34,16 @@ let R4rmb = R \ Noreturn (* Reads for which rmb works *)
> let rmb = [R4rmb] ; fencerel(Rmb) ; [R4rmb]
> let wmb = [W] ; fencerel(Wmb) ; [W]
> let mb = ([M] ; fencerel(Mb) ; [M]) |
> + (*
> + * full-barrier RMWs (successful cmpxchg(), xchg(), etc.) act as
> + * though there were enclosed by smp_mb().
> + * The effect of these virtual smp_mb() is formalized by adding
> + * Mb tags to the read and write of the operation, and providing
> + * the same ordering as though there were additional po edges
> + * between the Mb tag and the read resp. write.
> + *)
> + ([M] ; po ; [Mb & R]) |
> + ([Mb & W] ; po ; [M]) |
> ([M] ; fencerel(Before-atomic) ; [RMW] ; po? ; [M]) |
> ([M] ; po? ; [RMW] ; fencerel(After-atomic) ; [M]) |
> ([M] ; po? ; [LKW] ; fencerel(After-spinlock) ; [M]) |
> --
> 2.34.1
>
>