just curious

Peter da Silva peter-dragonfly at taronga.com
Mon Jul 21 10:13:05 PDT 2003


Matthew Dillon wrote:
    Yes.  Compare it against the Mach 'I'm going to map the data into your
    address space' model.  There will be some cases where having the kernel
    map the address space is more efficient (saving two system calls),
    but most of the time there is no need or reason to map the data and in
    those cases passing a descriptor is far more efficient.
Indeed. Most of the layering cases that Terry Lambert was presenting as
potential problems can be avoided by passing these kinds of handles 
around. Plus, they could easily be exposed over the network via NFS or 
sockets to non-dragonfly operating systems and applications. For a 
webserver that would allow for a heck of an optimization: just open a 
file readonly and then attach the client's socket to it... with smart 
enough hardware you could practically DMA straight from the SCSI card to 
the ethernet card. Or you could move most of NFS to user mode and still 
get kernel NFS performance...

	copyfile(int sfd, int dfd, off_t soffset, off_t doffset, off_t bytes)
I assume this operation would end up being implemented by changing the 
state of VM buffers, yes?






More information about the Kernel mailing list