Messages in this thread | | | Date | Thu, 27 Aug 2009 07:20:09 -0700 (PDT) | From | Davide Libenzi <> | Subject | Re: [PATCH 0/2] eventfd: new EFD_STATE flag |
| |
On Thu, 27 Aug 2009, Michael S. Tsirkin wrote:
> On Thu, Aug 27, 2009 at 07:13:32AM +0300, Avi Kivity wrote: > > On 08/27/2009 02:30 AM, Davide Libenzi wrote: > >> On Wed, 26 Aug 2009, Davide Libenzi wrote: > >> > >> > >>>> I see no kernel equivalent to read(), but that's easily done. > >>>> > >>> Adding an in-kernel read based on "ctx", that is no problem at all. > >>> > >> Something like the untested below. > >> I had thought you said the eventfd readers where in userspace, but I might > >> have misunderstood you. > >> > > > > No, they're all over the place. > > Further, with Davide's proposal you must be a reader to signal events.
To signal events, you must have an instance of "ctx" (with the new exposed eventfd_ctx_read). How would you do otherwise?
- Davide
| |