Messages in this thread | | | Date | Wed, 24 Apr 2019 11:31:07 -0700 | From | "Paul E. McKenney" <> | Subject | Re: [PATCH 1/3] tools: memory-model: Prepare for data-race detection |
| |
On Wed, Apr 24, 2019 at 10:24:42AM -0400, Alan Stern wrote: > On Wed, 24 Apr 2019, Paul E. McKenney wrote: > > > On Mon, Apr 22, 2019 at 12:17:45PM -0400, Alan Stern wrote: > > > This patch makes some slight alterations to linux-kernel.cat in > > > preparation for adding support for data-race detection to the > > > Linux-Kernel Memory Model. > > > > > > The definitions of relations involved in Acquire, Release, and > > > unlock-lock ordering are moved up earlier in the source file. > > > > > > The rmb relation is factored through the new R4rmb class: the > > > class of reads to which rmb will apply. > > > > > > The definition of the fence relation is moved earlier, and it > > > is split up into read- and write-fences (rmb and wmb) and all > > > the others. > > > > > > This should not make any functional changes. > > > > > > Signed-off-by: Alan Stern <stern@rowland.harvard.edu> > > > > Thank you, Alan, I have queued all three onto -rcu for review and testing. > > FYI, I rebased my smp_mb__{before,after}_atomic() patch on top of yours > > to avoid the conflict. > > > > Which demonstrates non-commutativity of patches. Your patches conflict > > with mine, but mine does not conflict with yours. ;-) > > :-) > > Besides, who knows where we'll end up with the > smp_mb__{before,after}_atomic stuff once Peter is done with it, anyway?
True enough!
Thanx, Paul
| |