Messages in this thread | | | Date | Thu, 27 Aug 2009 07:38:46 -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 07:21:49AM -0700, Davide Libenzi wrote: > > On Thu, 27 Aug 2009, Paolo Bonzini wrote: > > > > > > Ok, so why not using the eventfd counter as state? > > > > On the device side: > > > > > > > > void write_state(int sfd, int state) { > > > > u64 cnt; > > > > > > > > /* Clear the current state, sfd is in non-blocking mode */ > > > > read(sfd,&cnt, sizeof(cnt)); > > > > /* Writes new state */ > > > > cnt = 1 + !!state; > > > > write(sfd,&cnt, sizeof(cnt)); > > > > } > > > > > > It's interesting [no sarcasm intended, mind] that EFD_SEMAPHORE was > > > added exactly to avoid a read+write combination for the case of > > > decrementing a value. > > > > Like I repeated 25 times already, EFD_SEMAPHORE was added, because a > > *semaphore* is a pretty widely known and used abstraction. > > what about an atomic variable, btw? does it make sense to implement > write that does compare and exchange?
It is surprising to me, that is front of a workable solution w/out any use-once additions, yet you want to try to add optimizations and new ad-hoc abstractions to user visible interfaces. Now, you tell me what an atomic variable has to do with an eventfd.
- Davide
| |