Re: [PATCH] PUFFS backend allocation (round 3)

[Thor Lancelot Simon <tls%panix.com@localhost>]

On Wed, Oct 29, 2014 at 12:11:41AM +0000, Emmanuel Dreyfus wrote:
> On Tue, Oct 28, 2014 at 06:07:29PM +0100, J. Hannken-Illjes wrote:
> > Confused.  If write and/or fsync are synchronous (VOP_PUTPAGES with
> > flag PGO_SYNCIO) no write error will be forgotten. 
> 
> puffs_vnop_stratgy() contains this:
>                 /*
>                  * XXXXXXXX: wrong, but kernel can't survive strategy
>                  * failure currently.  Here, have one more X: X.    
>                  */
>                 if (error != ENOMEM) 
>                         error = 0;
> 
> This is where we want to store the error so that it can be 
> recovered by upper layer.
> 
> > If one of these VOP_PUTPAGES is async the write error will be delayed 
> > and if the next (and last) operation is VOP_CLOSE this is the last chance to
> > report the error back.  If any intermediate operation returns a delayed
> > error to the application it is cleared and close will no longer see it.
> (...)
> > There is no close after close.  Once a file is closed it may return
> > a delayed error but has to close anyway.  This is how NFS works and
> > what applications expect.
> 
> Let me rephrase: I call close(), get a error because of a failed
> asynchronous write. The file is not closed, right? Now if I call

Wrong.  It's still closed.