On Wed, Aug 07, 2013 at 07:45:17AM +0000, Dilger, Andreas wrote: > The problem still remains that the kernel loop driver eventually depends on > a local block device for the pages/bios to be written. The Lustre lloop > driver bypasses the VFS and block layer to generate RPCs from the submitted > pages to RDMA over the network without a data copy. No, it doesn't. It still consumes bios just like the regular loop driver. Besides missing all kinds of fixes from years of kernel development the only difference is that it takes a lustre-specific shortcut into the direct I/O code instead of going through the pagecache. The patch series I've pointed you to does exactly that in a generic way and thus superceeds the lloop driver fully. In case my previous reference was a bit to vague the series starts at: [PATCH V8 00/33] loop: Issue O_DIRECT aio using bio_vec please take a look and make sure to review it in case you see any shortcomings. _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
