At any rate this isn't an issue. I have simply unlocked and destroyed
the rule after it's been removed from the classifier.
Ethan
On Tue, Aug 13, 2013 at 7:48 AM, Ben Pfaff wrote:
> On Mon, Aug 12, 2013 at 01:23:38PM +0900, YAMAMOTO Takashi wrote:
>> > This patch uses a read-write lock to preve
On Mon, Aug 12, 2013 at 01:23:38PM +0900, YAMAMOTO Takashi wrote:
> > This patch uses a read-write lock to prevent rules from being evicted
> > while they're used by child threads. It also changes the prototypes
> > of the various rule lookup functions so that the thread safety
> > analysis can be
> This patch uses a read-write lock to prevent rules from being evicted
> while they're used by child threads. It also changes the prototypes
> of the various rule lookup functions so that the thread safety
> analysis can be used to ensure that the locking is handled properly.
destroying a rwlock
> I'm a little worried about deadlocks. There are lots of corner cases
> here (eviction, in-band control, fail-open) and I'm concerned that we
> don't hit those much in testing.
Agreed. Not much to do but jump in the deep end.
Thanks for the reviews.
Ethan
X-CudaMail-Whitelist-To: dev@openvswi
On Fri, Aug 09, 2013 at 02:05:59PM -0700, Ethan Jackson wrote:
> This patch uses a read-write lock to prevent rules from being evicted
> while they're used by child threads. It also changes the prototypes
> of the various rule lookup functions so that the thread safety
> analysis can be used to en
