This commit was generated by cvs2git to track changes on a CVS vendor

branch.

2 files changed, 280 insertions, 0 deletions

diff --git a/src/lib/libcrypto/rc4/Makefile b/src/lib/libcrypto/rc4/Makefile
new file mode 100644
index 0000000000..64e06924f4
--- /dev/null
+++ b/src/lib/libcrypto/rc4/Makefile
@@ -0,0 +1,123 @@
+#
+# SSLeay/crypto/rc4/Makefile
+#
+
+DIR=    rc4
+TOP=    ../..
+CC=     cc
+CPP=    $(CC) -E
+INCLUDES=
+CFLAG=-g
+INSTALL_PREFIX=
+OPENSSLDIR=     /usr/local/ssl
+INSTALLTOP=/usr/local/ssl
+MAKEDEPPROG=    makedepend
+MAKEDEPEND=     $(TOP)/util/domd $(TOP) -MD $(MAKEDEPPROG)
+MAKEFILE=       Makefile
+AR=             ar r
+
+RC4_ENC=rc4_enc.o
+# or use
+#RC4_ENC=asm/rx86-elf.o
+#RC4_ENC=asm/rx86-out.o
+#RC4_ENC=asm/rx86-sol.o
+#RC4_ENC=asm/rx86bdsi.o
+
+CFLAGS= $(INCLUDES) $(CFLAG)
+ASFLAGS= $(INCLUDES) $(ASFLAG)
+
+GENERAL=Makefile
+TEST=rc4test.c
+APPS=
+
+LIB=$(TOP)/libcrypto.a
+LIBSRC=rc4_skey.c rc4_enc.c
+LIBOBJ=rc4_skey.o $(RC4_ENC)
+
+SRC= $(LIBSRC)
+
+EXHEADER= rc4.h
+HEADER= $(EXHEADER) rc4_locl.h
+
+ALL=    $(GENERAL) $(SRC) $(HEADER)
+
+top:
+        (cd ../..; $(MAKE) DIRS=crypto SDIRS=$(DIR) sub_all)
+
+all:    lib
+
+lib:    $(LIBOBJ)
+        $(AR) $(LIB) $(LIBOBJ)
+        $(RANLIB) $(LIB) || echo Never mind.
+        @touch lib
+
+# elf
+asm/rx86-elf.s: asm/rc4-586.pl ../perlasm/x86asm.pl
+        (cd asm; $(PERL) rc4-586.pl elf $(CFLAGS) > rx86-elf.s)
+
+# a.out
+asm/rx86-out.o: asm/rx86unix.cpp
+        $(CPP) -DOUT asm/rx86unix.cpp | as -o asm/rx86-out.o
+
+# bsdi
+asm/rx86bsdi.o: asm/rx86unix.cpp
+        $(CPP) -DBSDI asm/rx86unix.cpp | sed 's/ :/:/' | as -o asm/rx86bsdi.o
+
+asm/rx86unix.cpp: asm/rc4-586.pl ../perlasm/x86asm.pl
+        (cd asm; $(PERL) rc4-586.pl cpp >rx86unix.cpp)
+
+asm/rc4-amd64.s: asm/rc4-amd64.pl;      $(PERL) asm/rc4-amd64.pl $@
+
+asm/rc4-ia64.s: asm/rc4-ia64.S
+        $(CC) $(CFLAGS) -E asm/rc4-ia64.S > $@
+
+files:
+        $(PERL) $(TOP)/util/files.pl Makefile >> $(TOP)/MINFO
+
+links:
+        @$(PERL) $(TOP)/util/mklink.pl ../../include/openssl $(EXHEADER)
+        @$(PERL) $(TOP)/util/mklink.pl ../../test $(TEST)
+        @$(PERL) $(TOP)/util/mklink.pl ../../apps $(APPS)
+
+install:
+        @headerlist="$(EXHEADER)"; for i in $$headerlist ; \
+        do  \
+        (cp $$i $(INSTALL_PREFIX)$(INSTALLTOP)/include/openssl/$$i; \
+        chmod 644 $(INSTALL_PREFIX)$(INSTALLTOP)/include/openssl/$$i ); \
+        done;
+
+tags:
+        ctags $(SRC)
+
+tests:
+
+lint:
+        lint -DLINT $(INCLUDES) $(SRC)>fluff
+
+depend:
+        $(MAKEDEPEND) -- $(CFLAG) $(INCLUDES) $(DEPFLAG) -- $(PROGS) $(LIBSRC)
+
+dclean:
+        $(PERL) -pe 'if (/^# DO NOT DELETE THIS LINE/) {print; exit(0);}' $(MAKEFILE) >Makefile.new
+        mv -f Makefile.new $(MAKEFILE)
+
+clean:
+        rm -f asm/rx86unix.cpp asm/*-elf.* *.o *.obj lib tags core .pure .nfs* *.old *.bak fluff asm/*.o
+
+# DO NOT DELETE THIS LINE -- make depend depends on it.
+
+rc4_enc.o: ../../e_os.h ../../include/openssl/bio.h
+rc4_enc.o: ../../include/openssl/buffer.h ../../include/openssl/crypto.h
+rc4_enc.o: ../../include/openssl/e_os2.h ../../include/openssl/err.h
+rc4_enc.o: ../../include/openssl/lhash.h ../../include/openssl/opensslconf.h
+rc4_enc.o: ../../include/openssl/opensslv.h ../../include/openssl/rc4.h
+rc4_enc.o: ../../include/openssl/safestack.h ../../include/openssl/stack.h
+rc4_enc.o: ../../include/openssl/symhacks.h ../cryptlib.h rc4_enc.c rc4_locl.h
+rc4_skey.o: ../../e_os.h ../../include/openssl/bio.h
+rc4_skey.o: ../../include/openssl/buffer.h ../../include/openssl/crypto.h
+rc4_skey.o: ../../include/openssl/e_os2.h ../../include/openssl/err.h
+rc4_skey.o: ../../include/openssl/lhash.h ../../include/openssl/opensslconf.h
+rc4_skey.o: ../../include/openssl/opensslv.h ../../include/openssl/rc4.h
+rc4_skey.o: ../../include/openssl/safestack.h ../../include/openssl/stack.h
+rc4_skey.o: ../../include/openssl/symhacks.h ../cryptlib.h rc4_locl.h
+rc4_skey.o: rc4_skey.c
diff --git a/src/lib/libcrypto/rc4/asm/rc4-ia64.S b/src/lib/libcrypto/rc4/asm/rc4-ia64.S
new file mode 100644
index 0000000000..b517d2e88f
--- /dev/null
+++ b/src/lib/libcrypto/rc4/asm/rc4-ia64.S
@@ -0,0 +1,157 @@
+// ====================================================================
+// Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
+// project.
+//
+// Rights for redistribution and usage in source and binary forms are
+// granted according to the OpenSSL license. Warranty of any kind is
+// disclaimed.
+// ====================================================================
+
+.ident  "rc4-ia64.S, Version 1.1"
+.ident  "IA-64 ISA artwork by Andy Polyakov <appro@fy.chalmers.se>"
+
+// What's wrong with compiler generated code? Because of the nature of
+// C language, compiler doesn't [dare to] reorder load and stores. But
+// being memory-bound, RC4 should benefit from reorder [on in-order-
+// execution core such as IA-64]. But what can we reorder? At the very
+// least we can safely reorder references to key schedule in respect
+// to input and output streams. Secondly, from the first [close] glance
+// it appeared that it's possible to pull up some references to
+// elements of the key schedule itself. Original rationale ["prior
+// loads are not safe only for "degenerated" key schedule, when some
+// elements equal to the same value"] was kind of sloppy. I should have
+// formulated as it really was: if we assume that pulling up reference
+// to key[x+1] is not safe, then it would mean that key schedule would
+// "degenerate," which is never the case. The problem is that this
+// holds true in respect to references to key[x], but not to key[y].
+// Legitimate "collisions" do occur within every 256^2 bytes window.
+// Fortunately there're enough free instruction slots to keep prior
+// reference to key[x+1], detect "collision" and compensate for it.
+// All this without sacrificing a single clock cycle:-)
+// Furthermore. In order to compress loop body to the minimum, I chose
+// to deploy deposit instruction, which substitutes for the whole
+// key->data+((x&255)<<log2(sizeof(key->data[0]))). This unfortunately
+// requires key->data to be aligned at sizeof(key->data) boundary.
+// This is why you'll find "RC4_INT pad[512-256-2];" addenum to RC4_KEY
+// and "d=(RC4_INT *)(((size_t)(d+255))&~(sizeof(key->data)-1));" in
+// rc4_skey.c [and rc4_enc.c, where it's retained for debugging
+// purposes]. Throughput is ~210MBps on 900MHz CPU, which is is >3x
+// faster than gcc generated code and +30% - if compared to HP-UX C.
+// Unrolling loop below should give >30% on top of that...
+
+.text
+.explicit
+
+#if defined(_HPUX_SOURCE) && !defined(_LP64)
+# define ADDP   addp4
+#else
+# define ADDP   add
+#endif
+
+#define SZ      4       // this is set to sizeof(RC4_INT)
+// SZ==4 seems to be optimal. At least SZ==8 is not any faster, not for
+// assembler implementation, while SZ==1 code is ~30% slower.
+#if SZ==1       // RC4_INT is unsigned char
+# define        LDKEY   ld1
+# define        STKEY   st1
+# define        OFF     0
+#elif SZ==4     // RC4_INT is unsigned int
+# define        LDKEY   ld4
+# define        STKEY   st4
+# define        OFF     2
+#elif SZ==8     // RC4_INT is unsigned long
+# define        LDKEY   ld8
+# define        STKEY   st8
+# define        OFF     3
+#endif
+
+out=r8;         // [expanded] output pointer
+inp=r9;         // [expanded] output pointer
+prsave=r10;
+key=r28;        // [expanded] pointer to RC4_KEY
+ksch=r29;       // (key->data+255)[&~(sizeof(key->data)-1)]
+xx=r30;
+yy=r31;
+
+// void RC4(RC4_KEY *key,size_t len,const void *inp,void *out);
+.global RC4#
+.proc   RC4#
+.align  32
+.skip   16
+RC4:
+        .prologue
+        .fframe 0
+        .save   ar.pfs,r2
+        .save   ar.lc,r3
+        .save   pr,prsave
+{ .mii; alloc   r2=ar.pfs,4,12,0,16
+        mov     prsave=pr
+        ADDP    key=0,in0               };;
+{ .mib; cmp.eq  p6,p0=0,in1                     // len==0?
+        mov     r3=ar.lc
+(p6)    br.ret.spnt.many        b0      };;     // emergency exit
+
+        .body
+        .rotr   dat[4],key_x[4],tx[2],rnd[2],key_y[2],ty[1];
+
+{ .mib; LDKEY   xx=[key],SZ                     // load key->x
+        add     in1=-1,in1                      // adjust len for loop counter
+        nop.b   0                       }
+{ .mib; ADDP    inp=0,in2
+        ADDP    out=0,in3
+        brp.loop.imp    .Ltop,.Lexit-16 };;
+{ .mmi; LDKEY   yy=[key]                        // load key->y
+        add     ksch=(255+1)*SZ,key             // as ksch will be used with
+                                                // deposit instruction only,
+                                                // I don't have to &~255...
+        mov     ar.lc=in1               }
+{ .mmi; mov     key_y[1]=r0                     // guarantee inequality
+                                                // in first iteration
+        add     xx=1,xx
+        mov     pr.rot=1<<16            };;
+{ .mii; nop.m   0
+        dep     key_x[1]=xx,ksch,OFF,8
+        mov     ar.ec=3                 };;     // note that epilogue counter
+                                                // is off by 1. I compensate
+                                                // for this at exit...
+.Ltop:
+// The loop is scheduled for 3*(n+2) spin-rate on Itanium 2, which
+// theoretically gives asymptotic performance of clock frequency
+// divided by 3 bytes per seconds, or 500MBps on 1.5GHz CPU. Measured
+// performance however is distinctly lower than 1/4:-( The culplrit
+// seems to be *(out++)=dat, which inadvertently splits the bundle,
+// even though there is M-port available... Unrolling is due...
+// Unrolled loop should collect output with variable shift instruction
+// in order to avoid starvation for integer shifter... It should be
+// possible to get pretty close to theoretical peak...
+{ .mmi; (p16)   LDKEY   tx[0]=[key_x[1]]                // tx=key[xx]
+        (p17)   LDKEY   ty[0]=[key_y[1]]                // ty=key[yy]   
+        (p18)   dep     rnd[1]=rnd[1],ksch,OFF,8}       // &key[(tx+ty)&255]
+{ .mmi; (p19)   st1     [out]=dat[3],1                  // *(out++)=dat
+        (p16)   add     xx=1,xx                         // x++
+        (p16)   cmp.ne.unc p20,p21=key_x[1],key_y[1]    };;
+{ .mmi; (p18)   LDKEY   rnd[1]=[rnd[1]]                 // rnd=key[(tx+ty)&255]
+        (p16)   ld1     dat[0]=[inp],1                  // dat=*(inp++)
+        (p16)   dep     key_x[0]=xx,ksch,OFF,8  }       // &key[xx&255]
+.pred.rel       "mutex",p20,p21
+{ .mmi; (p21)   add     yy=yy,tx[1]                     // (p16)
+        (p20)   add     yy=yy,tx[0]                     // (p16) y+=tx
+        (p21)   mov     tx[0]=tx[1]             };;     // (p16)
+{ .mmi; (p17)   STKEY   [key_y[1]]=tx[1]                // key[yy]=tx
+        (p17)   STKEY   [key_x[2]]=ty[0]                // key[xx]=ty
+        (p16)   dep     key_y[0]=yy,ksch,OFF,8  }       // &key[yy&255]
+{ .mmb; (p17)   add     rnd[0]=tx[1],ty[0]              // tx+=ty
+        (p18)   xor     dat[2]=dat[2],rnd[1]            // dat^=rnd
+        br.ctop.sptk    .Ltop                   };;
+.Lexit:
+{ .mib; STKEY   [key]=yy,-SZ                    // save key->y
+        mov     pr=prsave,0x1ffff
+        nop.b   0                       }
+{ .mib; st1     [out]=dat[3],1                  // compensate for truncated
+                                                // epilogue counter
+        add     xx=-1,xx
+        nop.b   0                       };;
+{ .mib; STKEY   [key]=xx                        // save key->x
+        mov     ar.lc=r3
+        br.ret.sptk.many        b0      };;
+.endp   RC4#