Appeal No. 1999-2252
Application 08/780,790
"a writer adapted to permit the application to direct
that the data of the memory object associated with the fast
buffer be transferred to a sink device along a first data path
without copying the data into main memory along a second data
path unless the application attempts to access via a memory
fetch/store, any portion of the transferring data."
We agree with Appellants that neither Druschel or Krieger
teach this limitation.
Druschel teaches that when a PDU arrives from the18
network "An fbuf is allocated in the kernel, filled, and then
transferred . . ." (emphasis added). In addition, Druschel
teaches "The optimization integrates buffer management and19
cross-domain data transfer facility by placing the entire
aggregate object into fbufs" (emphasis added). Druschel
therefore requires copying to fbuf for data transfer without
determining whether or not the data has been modified.
Krieger teaches that for the interface modules the I/O20
read algorithm requires "Read first calls salloc to . . .,
then copies the data from the allocated region to the user
specified buffer" (emphasis added). In addition, Krieger
18Page 194, section 3.2.2.
19Page 195, section 3.2.3.
20Page 80, left column.
10
Page: Previous 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Next
Last modified: November 3, 2007