From: Gal R. <GA...@il...> - 2010-10-07 13:33:33
|
See comments below. Miklos Szeredi <mi...@sz...> wrote on 10/07/2010 11:32:56 AM: > From: > > Miklos Szeredi <mi...@sz...> > > To: > > Gal Rosen/Haifa/IBM@IBMIL > > Cc: > > fus...@li... > > Date: > > 10/07/2010 11:33 AM > > Subject: > > Re: [fuse-devel] FUSE kernel module blocking/non-blocking > > On Wed, 6 Oct 2010, Gal Rosen wrote: > > Hi all, > > > > By reading the code I notice that all syscalls except release at the end > > put the FUSE request by fuse_request_send() routine, which is blocking and > > waiting for response. > > Even async read/write are at the end reading/writing pages using > > fuse_request_send() and not by using fuse_request_send_background(). > > Readahead is asynchronous (see fuse_send_readpages()). > But when this function is called ? And one more question. The FUSE library read one request at a time, and each request in kernel pass maximum of 4KB data. Isn't lead to a very low performance ? is there a patch fixing it ? Thanks, Gal. > > If this is the case what is the point of implementing asynchronous > > operations using the low level API, if at the end the client process is > > waiting on a waitq in FUSE device ? > > The purpose of being an asynchronous API on the low level interface is > that a separate thread is not necessary for each request. > > Thanks, > Miklos > > ------------------------------------------------------------------------------ > Beautiful is writing same markup. Internet Explorer 9 supports > standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. > Spend less time writing and rewriting code and more time creating great > experiences on the web. Be a part of the beta today. > http://p.sf.net/sfu/beautyoftheweb > _______________________________________________ > fuse-devel mailing list > fus...@li... > https://lists.sourceforge.net/lists/listinfo/fuse-devel |