Krishna Yenduri wrote:
> Dan Anderson wrote:
> > Here's a review for CR 6665607 Need a SHA256/SHA384/SHA512 implementation
> > optimized for 64-bit x86
> > http://dan.drydog.com/reviews/6665607-sha2/
> usr/src/common/crypto/sha2/amd64/sha512-x86_64.pl
> KY-1 lines 220, 314 T3
> No need for these lines as this is just a digest algorithm.
FIXED; also fixed $SRC/common/crypto/sha1/amd64/sha1-x86_64.pl
> usr/src/lib/libmd/amd64/Makefile
> KY-2 lines 44-59 T4
> Consider making this a single rule using a pattern matching rule to match all
> the file names (I think it is %).
FIXED with: pics/%.o: %.s
> usr/src/lib/pkcs11/pkcs11_softtoken/Makefile.com
> KY-3 lines 135-136
> Does this mean we have been linking sha1 and sha2 objects in to softtoken
> library so far! Good catch.
Thanks. Fortunately, SHA*_COBJECTS and SHA*_PSR_OBJECTS are not defined, so
nothing was linked in. It was still confusing, so I removed it.
> usr/src/uts/intel/sha2/Makefile
> KY-4 lines 105-118 T4
> You don't need these rules if the SHA2_OBJS entry is moved to
> usr/src/uts/intel/Makefile.files.
> If they need to stay here, consider having a single rule for sha512-x86_64.o
> and sha256-x86_64.o by using pattern matching rules (I think it is %).
I get an error (below) if I move SHA2_OBJS to Makefile.files, so I added these
rule patterns instead:
$(OBJS_DIR)/%.o: %.s
$(OBJS_DIR)/%.ln: %.s
/usr/ccs/bin/ld -r -o debug64/sha2 debug64/sha512-x86_64.o
debug64/sha256-x86_64.o debug64/sha2.o debug64/sha2_mod.o
ld: fatal: file debug64/sha512-x86_64.o: open failed: No such file or directory
*** Error code 1
make: Fatal error: Command failed for target `debug64/sha2'
Current working directory /workspace/danx/nv85_sha2/usr/src/uts/intel/sha2
I revised the review and posted it here:
http://dan.drydog.com/reviews/6665607-sha2/
For those who may be interested, a merged ARCFOUR/SHA2/DIGEST/MAC review:
http://dan.drydog.com/reviews/6652716-6665607-6662791-ac4-sha2-mac64/
This message posted from opensolaris.org
