On Tue, Jan 03, 2017 at 09:23:29AM +0000, Bryan O'Donoghue wrote:
On 02/01/17 16:16, Johan Hovold wrote:
On Tue, Dec 27, 2016 at 01:01:35PM +0000, Bryan O'Donoghue wrote:
Later patches don't want or need to serialize the cancellation of an operation. This patch adds gb_operation_cancel_async() as a simple subset of the existing gb_operation_cancel() sans the synchronous wait on the cancellation queue.
This one is not needed and complicates the interface for no good reason (the async suffix also falsely indicates that this function could be called from atomic context).
Just cancel synchronously in the delayed worker function you add later in the series.
Johan
hmm ... sounds sensible in principle.
Let me test that out
I started reviewing the rest of the series, but got interrupted. Will send more comments on the rest later.
gott nytt år !
God fortsättning!
Johan