Richard L. Hamilton wrote: >> Roland Mainz wrote: >>> Darren: >>> 1. Are the changes Ok for you ? If "yes" I'll post >> a diff for the ARC >>> case+manpages... >> Yes that is fine. >> >>> 2. Are there any CDDL or Sun-owned versions of the >> optimized MD5/SHA* >>> functions (I'm asking since the ksh93-integration >> project has the >>> permission to contribute any Sun/CDDL code back to >> AT&T upstream under >>> their license and AFAIK some optimized versions of >> these functions may >>> be "nice to have") ? >> See $SRC/uts/common/crypto/ >> -- >> Darren J Moffat > > I'm not so sure that's where it is...I went looking for the MD5 software > implementation (assuming that the kernel has hooks for hardware > assisted implementations as well), and eventually found it in > http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/common/crypto/ > > e.g. under common one level up, not under uts/common. > > Note: I just track stuff down when I get curious; but I don't take the time > to understand any more than the immediate {question, problem, bit of > curiosity} requires, so that's all I can offer at this time.
$SRC/common/crypto is what it should have been my typo. It is in $SRC/common because the same algorithm implementation is used in libmd and in kernel/crypto/md5, same goes for all the other algorithms under $SRC/common/crypto (and also for $SRC/common/bignum which is used to build rsa,dsa,dh). The userland crypto framework is in $SRC/lib/pkcs11/libpkcs11 with the plugins being pkcs11_softtoken and pkcs11_kernel at the same dir level. The kernel crypto framework is in $SRC/uts/common/crypto/core. The io dirs at the same level are for the plugins and some linkage stuff like /dev/cryptoadm. -- Darren J Moffat
