Re: [PATCH 1/3] tools: memory-model: Expand definition of barrier
From: Paul E. McKenney
Date: Thu Jun 20 2019 - 12:38:30 EST
On Thu, Jun 20, 2019 at 11:55:36AM -0400, Alan Stern wrote:
> Commit 66be4e66a7f4 ("rcu: locking and unlocking need to always be at
> least barriers") added compiler barriers back into rcu_read_lock() and
> rcu_read_unlock(). Furthermore, srcu_read_lock() and
> srcu_read_unlock() have always contained compiler barriers.
> The Linux Kernel Memory Model ought to know about these barriers.
> This patch adds them into the memory model.
> Signed-off-by: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
And yes, much easier to understand this way, thank you!
I have queued them and they both diff equal to your previous patch and
give the expected results on the litmus-tests and github tests having
> tools/memory-model/linux-kernel.cat | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
> Index: usb-devel/tools/memory-model/linux-kernel.cat
> --- usb-devel.orig/tools/memory-model/linux-kernel.cat
> +++ usb-devel/tools/memory-model/linux-kernel.cat
> @@ -47,7 +47,8 @@ let strong-fence = mb | gp
> let nonrw-fence = strong-fence | po-rel | acq-po
> let fence = nonrw-fence | wmb | rmb
> let barrier = fencerel(Barrier | Rmb | Wmb | Mb | Sync-rcu | Sync-srcu |
> - Before-atomic | After-atomic | Acquire | Release) |
> + Before-atomic | After-atomic | Acquire | Release |
> + Rcu-lock | Rcu-unlock | Srcu-lock | Srcu-unlock) |
> (po ; [Release]) | ([Acquire] ; po)