Ran configure thus : $ ./configure --libdir=/usr/local/lib64 --disable-silent-rules \ > --enable-dependency-tracking --enable-static=yes --enable-shared=yes \ > --enable-large-data-tests --enable-hmac-binary-check \ > --disable-O-flag-munging --disable-optimization --with-gnu-ld \ > --with-sysroot=/usr/local --with-libgpg-error-prefix=/usr/local That resulted in what seems to be reasonable : checking for a BSD-compatible install... /bin/install -c
[... configure output stripped -wk ...]
Libgcrypt v1.8.1 has been configured as follows:
Platform: GNU/Linux (x86_64-pc-linux-gnu)
Hardware detection module: hwf-x86
Enabled cipher algorithms: arcfour blowfish cast5 des aes twofish
serpent rfc2268 seed camellia idea salsa20
gost28147 chacha20
Enabled digest algorithms: crc gostr3411-94 md4 md5 rmd160 sha1
sha256 sha512 sha3 tiger whirlpool stribog
blake2
Enabled kdf algorithms: s2k pkdf2 scrypt
Enabled pubkey algorithms: dsa elgamal rsa ecc
Random number generator: default
Try using jitter entropy: yes
Using linux capabilities: no
Try using Padlock crypto: yes
Try using AES-NI crypto: yes
Try using Intel PCLMUL: yes
Try using Intel SSE4.1: yes
Try using DRNG (RDRAND): yes
Try using Intel AVX: yes
Try using Intel AVX2: yes
Try using ARM NEON: n/a
Try using ARMv8 crypto: n/a
The compile ran fine however there were many strange warnings thus :
libtool: link: warning: library `/usr/local/lib64/libgpg-error.la' was moved.
That makes no sense as libgpg-error.la was not moved and has always been built and installed into /usr/local with the usual libs in /usr/local/lib64 and here is an example :
gcc -DHAVE_CONFIG_H -I. -I.. -I/usr/local/include -I/usr/local/include -m64 -g -Wl,-rpath=/usr/local/lib64,--enable-new-dtags -malign-double -m128bit-long-double -mpc80 -D_LARGEFILE64_SOURCE -fvisibility=hidden -Wall -MT mpicalc-mpicalc.o -MD -MP -MF .deps/mpicalc-mpicalc.Tpo -c -o mpicalc-mpicalc.o `test -f 'mpicalc.c' || echo './'`mpicalc.c
mv -f .deps/mpicalc-mpicalc.Tpo .deps/mpicalc-mpicalc.Po
/bin/sh ../libtool --tag=CC --mode=link gcc -I/usr/local/include -m64 -g -Wl,-rpath=/usr/local/lib64,--enable-new-dtags -malign-double -m128bit-long-double -mpc80 -D_LARGEFILE64_SOURCE -fvisibility=hidden -Wall -o mpicalc mpicalc-mpicalc.o libgcrypt.la -ldl -L/usr/local/lib64 -lgpg-error
libtool: link: warning: library `/usr/local/lib64/libgpg-error.la' was moved.
libtool: link: warning: library `/usr/local/lib64/libgpg-error.la' was moved.
Regardless the compile finished with reasonable output thus :
$ readelf -d ./src/.libs/libgcrypt.so.20.2.1
Dynamic section at offset 0x157dc0 contains 29 entries:
Tag Type Name/Value
0x0000000000000001 (NEEDED) Shared library: [libgpg-error.so.0]
0x0000000000000001 (NEEDED) Shared library: [libc.so.6]
0x000000000000000e (SONAME) Library soname: [libgcrypt.so.20]
0x000000000000001d (RUNPATH) Library runpath: [/usr/local/lib64]
0x000000000000000c (INIT) 0xb710
0x000000000000000d (FINI) 0x11c470
0x0000000000000019 (INIT_ARRAY) 0x3571e0
0x000000000000001b (INIT_ARRAYSZ) 16 (bytes)
0x000000000000001a (FINI_ARRAY) 0x3571f0
0x000000000000001c (FINI_ARRAYSZ) 8 (bytes)
0x000000006ffffef5 (GNU_HASH) 0x1f0
0x0000000000000005 (STRTAB) 0x24c8
0x0000000000000006 (SYMTAB) 0x950
0x000000000000000a (STRSZ) 4544 (bytes)
0x000000000000000b (SYMENT) 24 (bytes)
0x0000000000000003 (PLTGOT) 0x358000
0x0000000000000002 (PLTRELSZ) 1944 (bytes)
0x0000000000000014 (PLTREL) RELA
0x0000000000000017 (JMPREL) 0xaf78
0x0000000000000007 (RELA) 0x3970
0x0000000000000008 (RELASZ) 30216 (bytes)
0x0000000000000009 (RELAENT) 24 (bytes)
0x000000006ffffffc (VERDEF) 0x38d8
0x000000006ffffffd (VERDEFNUM) 2
0x000000006ffffffe (VERNEED) 0x3910
0x000000006fffffff (VERNEEDNUM) 2
0x000000006ffffff0 (VERSYM) 0x3688
0x000000006ffffff9 (RELACOUNT) 1253
0x0000000000000000 (NULL) 0x0
$
A few tests failed where I note the README claims :
--enable-hmac-binary-check
Include support to check the binary at runtime
against a HMAC checksum. This works only in FIPS
mode and on systems providing the dladdr function.
Perhaps that is a problem on Red Hat Enterprise Linux 7.4 but I don't
know. There is no reason that I can see why the non-POSIX Glibc
extension does not exist on this system but perhaps it is an error to
use the extension? I would not know but the test fails:
.
.
.
PASS: prime
libgcrypt selftest: binary (0): No such file or directory (/usr/local/build/libgcrypt-1.8.1_3.10.0-693.11.1.el7.x86_64.001/src/.libs/.libgcrypt.so.20.hmac)
basic: running self-test failed
FAIL: basic
.
.
.
PASS: t-cv25519
now running 'basic' test with all hardware features disabled.
libgcrypt selftest: binary (0): No such file or directory (/usr/local/build/libgcrypt-1.8.1_3.10.0-693.11.1.el7.x86_64.001/src/.libs/.libgcrypt.so.20.hmac)
basic: running self-test failed
FAIL: basic-disable-all-hwf
.
.
.
Perhaps this is related to some "FIPS mode" requirement ?
I did not see anything in the README nor elsewhere to see what this "FIPS mode" was.
All other tests seems to run fine and this includes the really long tests with 256G of data fed
to the various SHA hash algorithms. However I do note that the "cycles/byte" column is always
just a blank and someone spelled megabytes as "mebibytes" somewhere. May want to fix that
in line 581 of tests/bench-slope.c thus :
567 static void
568 bench_print_header (int algo_width, const char *algo_name)
569 {
570 if (csv_mode)
571 {
572 gcry_free (current_algo_name);
573 current_algo_name = gcry_xstrdup (algo_name);
574 }
575 else
576 {
577 if (algo_width < 0)
578 printf (" %-*s | ", -algo_width, algo_name);
579 else
580 printf (" %-*s | ", algo_width, algo_name);
581 printf ("%14s %15s %13s\n", "nanosecs/byte", "mebibytes/sec",
582 "cycles/byte");
583 }
584 }
585
So that output looks like this :
PASS: benchmark
Note: bench-slope running in quick regression test mode.
Hash:
| nanosecs/byte mebibytes/sec cycles/byte
MD5 | 13.96 ns/B 68.32 MiB/s - c/B
SHA1 | 28.70 ns/B 33.23 MiB/s - c/B
RIPEMD160 | 41.97 ns/B 22.72 MiB/s - c/B
TIGER192 | 14.31 ns/B 66.66 MiB/s - c/B
SHA256 | 61.45 ns/B 15.52 MiB/s - c/B
SHA384 | 52.70 ns/B 18.10 MiB/s - c/B
SHA512 | 49.12 ns/B 19.42 MiB/s - c/B
SHA224 | 54.79 ns/B 17.41 MiB/s - c/B
MD4 | 7.36 ns/B 129.6 MiB/s - c/B
CRC32 | 5.99 ns/B 159.3 MiB/s - c/B
CRC32RFC1510 | 5.96 ns/B 160.1 MiB/s - c/B
CRC24RFC2440 | 6.04 ns/B 157.8 MiB/s - c/B
WHIRLPOOL | 8.64 ns/B 110.4 MiB/s - c/B
TIGER | 10.36 ns/B 92.03 MiB/s - c/B
TIGER2 | 10.36 ns/B 92.04 MiB/s - c/B
GOSTR3411_94 | 136.8 ns/B 6.97 MiB/s - c/B
STRIBOG256 | 134.8 ns/B 7.07 MiB/s - c/B
STRIBOG512 | 147.3 ns/B 6.48 MiB/s - c/B
GOSTR3411_CP | 170.6 ns/B 5.59 MiB/s - c/B
SHA3-224 | 15.26 ns/B 62.48 MiB/s - c/B
SHA3-256 | 16.17 ns/B 58.99 MiB/s - c/B
SHA3-384 | 21.07 ns/B 45.27 MiB/s - c/B
SHA3-512 | 30.34 ns/B 31.43 MiB/s - c/B
SHAKE128 | 12.71 ns/B 75.05 MiB/s - c/B
SHAKE256 | 14.90 ns/B 64.01 MiB/s - c/B
BLAKE2B_512 | 24.20 ns/B 39.40 MiB/s - c/B
BLAKE2B_384 | 24.66 ns/B 38.67 MiB/s - c/B
BLAKE2B_256 | 24.69 ns/B 38.62 MiB/s - c/B
BLAKE2B_160 | 22.47 ns/B 42.45 MiB/s - c/B
BLAKE2S_256 | 38.31 ns/B 24.89 MiB/s - c/B
BLAKE2S_224 | 38.68 ns/B 24.65 MiB/s - c/B
BLAKE2S_160 | 38.83 ns/B 24.56 MiB/s - c/B
BLAKE2S_128 | 38.85 ns/B 24.55 MiB/s - c/B
=
Not sure what cycles per byte should be about.
In any case the testsuite runs for hours and then outputs some spooky results :
======================================
2 of 28 tests failed
Please report to http://bugs.gnupg.org
======================================
gmake[2]: *** [Makefile:864: check-TESTS] Error 1
gmake[2]: Leaving directory '/usr/local/build/libgcrypt-1.8.1_3.10.0-693.11.1.el7.x86_64.001/tests'
gmake[1]: *** [Makefile:987: check-am] Error 2
gmake[1]: Leaving directory '/usr/local/build/libgcrypt-1.8.1_3.10.0-693.11.1.el7.x86_64.001/tests'
gmake: *** [Makefile:477: check-recursive] Error 1
$
Perhaps the "--enable-hmac-binary-check" should be checked in the
configure stage for the required mystery "FIPS mode" as well as
the needed Glibc extension dladdr?
In any case the result can not be installed as is with tests failed.
Dennis Clarke