Hello Richard, Richard Levitte> However, resolving all those concealed devices is still a good thing, so I suggest this strategy:
$ ROOT = F$PARSE("sys$disk:[-]A.;0",,,"SYNTAX_ONLY,NO_CONCEAL") - "A.;0" $ ROOT_DEV = F$PARSE(ROOT,,,"DEVICE","SYNTAX_ONLY") $ ROOT_DIR = F$PARSE(ROOT,,,"DIRECTORY","SYNTAX_ONLY") - - ".][000000" - "[000000." - "][" - "[" - "]" $ ROOT = ROOT_DEV + "[" + ROOT_DIR $ DEFINE/NOLOG SSLROOT 'ROOT'.APPS.] /TRANS=CONC Sounds perfect, except that you missed a nasty coma sign. The first line should look like this: $ ROOT = F$PARSE("sys$disk:[-]A.;0",,,,"SYNTAX_ONLY,NO_CONCEAL") - "A.;0" Richard Levitte> Does this happen when you made that SSLROOT definition change given above? You see, openssl-vms.cnf defined RANDFILE on its own You are right... when the SSLROOT is well defined there is no problem with the RANDFILE during tests. Zoltan.Arpadffy> ...I am dreaming of the possibility to choose between the 64 and the 32 bit version build and reuse this "not-used option" in MAKEVMS.COM Zoltan.Arpadffy> During the summer I have submitted a patch for this "feature"... wonder is there any chance for this code to be merged in? Richard Levitte> I recall discussing it, because it seemed to be a no-op... but I can't recall getting to an answer (and I'll admit not being a good participant in that discussion). This would be very useful as there are still users that link against 32 bit libraries for some reasons. >From other side the current 64 bits build stays as a default and you would not >see any difference if the 32 bit parameter is not explicitly used. I'll prepare a patch against STABLE-SNAP-20101124 and when I'm ready I'll submit it again. Thank you. Regards, Z -----Original Message----- From: Richard Levitte [mailto:rich...@levitte.org] Sent: den 29 november 2010 21:59 To: openssl-dev@openssl.org; Arpadffy Zoltan Subject: Re: buid STABLE-SNAP-20101124 on platform: VMS IA64 V8.3 In message <fa849bbcfec13e42a34ae60ebc2d724720d4ad5...@sgmail1> on Thu, 25 Nov 2010 16:55:21 +0100, Arpadffy Zoltan <zoltan.arpad...@scientificgames.se> said: Zoltan.Arpadffy> Solution: Zoltan.Arpadffy> Zoltan.Arpadffy> In tests.com replace Zoltan.Arpadffy> $ sslroot = f$parse("sys$disk:[-.apps];",,,,"syntax_only") - "].;"+ ".]" Zoltan.Arpadffy> $ define /translation_attributes = concealed sslroot 'sslroot' Zoltan.Arpadffy> Zoltan.Arpadffy> with Zoltan.Arpadffy> $ ROOT = F$PARSE(__here,"[]A.;0",,,"SYNTAX_ONLY,NO_CONCEAL") - "A.;0" Zoltan.Arpadffy> $ ROOT_DEV = F$PARSE(ROOT,,,"DEVICE","SYNTAX_ONLY") Zoltan.Arpadffy> $ ROOT_DIR = F$PARSE(ROOT,,,"DIRECTORY","SYNTAX_ONLY") - Zoltan.Arpadffy> - ".][000000" - "[000000." - "][" - "[" - "]" - ".TEST" Zoltan.Arpadffy> $ ROOT = ROOT_DEV + "[" + ROOT_DIR Zoltan.Arpadffy> $ DEFINE/NOLOG SSLROOT 'ROOT'.APPS.] /TRANS=CONC I partly agree. If you have placed the openssl source under, say, FOO:[FOSS.TEST.OPENSSL-1_0_1-DEV], it will fail. However, resolving all those concealed devices is still a good thing, so I suggest this strategy: $ ROOT = F$PARSE("sys$disk:[-]A.;0",,,"SYNTAX_ONLY,NO_CONCEAL") - "A.;0" $ ROOT_DEV = F$PARSE(ROOT,,,"DEVICE","SYNTAX_ONLY") $ ROOT_DIR = F$PARSE(ROOT,,,"DIRECTORY","SYNTAX_ONLY") - - ".][000000" - "[000000." - "][" - "[" - "]" $ ROOT = ROOT_DEV + "[" + ROOT_DIR $ DEFINE/NOLOG SSLROOT 'ROOT'.APPS.] /TRANS=CONC The rationale being that no subdirectory of the OpenSSL source tree will be on a different device, or at the top of any device. Zoltan.Arpadffy> This is a bit more complicated, but works with complicated concealed paths as well. Yup. Zoltan.Arpadffy> I noticed that Zoltan.Arpadffy> ---> TEST_AES Zoltan.Arpadffy> ... is empty as the commented code shows: Zoltan.Arpadffy> Zoltan.Arpadffy> $ test_aes: Zoltan.Arpadffy> $! write sys$output "test AES" Zoltan.Arpadffy> $! !mcr 'texe_dir''aestest' Zoltan.Arpadffy> Zoltan.Arpadffy> ...seems there is no aestest at all. Is there any purpose that this is here? None, other than to mimic this from Makefile: test_aes: #$(AESTEST) # @echo "test Rijndael" # ../util/shlib_wrap.sh ./$(AESTEST) Zoltan.Arpadffy> The test dies with Zoltan.Arpadffy> Zoltan.Arpadffy> Sign the certificate? [y/n]:2071080376:error:24064064:random number generator:SSLEAY_RAND_BYTES:PRNG not seeded:USRDSK Zoltan.Arpadffy> :[ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.CRYPTO.RAND]MD_RAND.C;1:522:You need to read the OpenSSL FAQ, http://www.o Zoltan.Arpadffy> penssl.org/support/faq.html Zoltan.Arpadffy> 2071080376:error:04088003:rsa routines:RSA_setup_blinding:BN lib:USRDSK:[ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.CRY Zoltan.Arpadffy> PTO.RSA]RSA_LIB.C;1:426: Zoltan.Arpadffy> 2071080376:error:04066044:rsa routines:RSA_EAY_PRIVATE_ENCRYPT:internal error:USRDSK:[ZAY.WORK.OPENSSL-101-STABLE-SNAP Zoltan.Arpadffy> -20101124.CRYPTO.RSA]RSA_EAY.C;1:403: Zoltan.Arpadffy> 2071080376:error:0D0C3006:asn1 encoding routines:ASN1_item_sign:EVP lib:USRDSK:[ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101 Zoltan.Arpadffy> 124.CRYPTO.ASN1]A_SIGN.C;1:279: Zoltan.Arpadffy> OpenSSL 1.0.1-dev xx XXX xxxx Zoltan.Arpadffy> built on: 25-NOV-2010 12:23:57.59 Zoltan.Arpadffy> platform: VMS IA64 V8.3 Zoltan.Arpadffy> options: bn(64,64) md2(int) rc4(ptr,int) des(ptr,risc1,16,long) idea(int) blowfish(idx) Zoltan.Arpadffy> compiler: information not available Zoltan.Arpadffy> OPENSSLDIR: N/A Zoltan.Arpadffy> Zoltan.Arpadffy> Solution: define the RANDFILE logical Zoltan.Arpadffy> Zoltan.Arpadffy> TITAN2_ZAY $ diff <.test>TESTGEN.COM Zoltan.Arpadffy> ************ Zoltan.Arpadffy> File USRDSK:<ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.TEST>TESTGEN.COM;2 Zoltan.Arpadffy> 1 $! TESTGEN.COM Zoltan.Arpadffy> 2 $ Zoltan.Arpadffy> ****** Zoltan.Arpadffy> File USRDSK:<ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.TEST>TESTGEN.COM;1 Zoltan.Arpadffy> 1 $! TETSGEN.COM Zoltan.Arpadffy> 2 $ Zoltan.Arpadffy> ************ Zoltan.Arpadffy> ************ Zoltan.Arpadffy> File USRDSK:<ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.TEST>TESTGEN.COM;2 Zoltan.Arpadffy> 20 $ define/nolog randfile <>.rnd Zoltan.Arpadffy> 21 $ append/new nl: .rnd Zoltan.Arpadffy> ****** Zoltan.Arpadffy> File USRDSK:<ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.TEST>TESTGEN.COM;1 Zoltan.Arpadffy> 20 $ Zoltan.Arpadffy> 21 $ append/new nl: .rnd Zoltan.Arpadffy> ************ Does this happen when you made that SSLROOT definition change given above? You see, openssl-vms.cnf defined RANDFILE on its own, and that's whay testgen.com should be able to count on, I believe. Zoltan.Arpadffy> The test fails with: Zoltan.Arpadffy> Zoltan.Arpadffy> ---> TEST_EVP Zoltan.Arpadffy> evptests.txt: no such file or directory Zoltan.Arpadffy> OpenSSL 1.0.1-dev xx XXX xxxx Zoltan.Arpadffy> built on: 25-NOV-2010 12:23:57.59 Zoltan.Arpadffy> platform: VMS IA64 V8.3 Zoltan.Arpadffy> options: bn(64,64) md2(int) rc4(ptr,int) des(ptr,risc1,16,long) idea(int) blowfish(idx) Zoltan.Arpadffy> compiler: information not available Zoltan.Arpadffy> OPENSSLDIR: N/A Zoltan.Arpadffy> Zoltan.Arpadffy> Solution: provide the right path to the evptests.txt file. In the tests.com file the test_evp should look like this: Zoltan.Arpadffy> Zoltan.Arpadffy> $ test_evp: Zoltan.Arpadffy> $ mcr 'texe_dir''evptest' 'ROOT'.CRYPTO.EVP]evptests.txt Zoltan.Arpadffy> $ return Thanks, change applied. Zoltan.Arpadffy> The install went fine Zoltan.Arpadffy> Zoltan.Arpadffy> Zoltan.Arpadffy> Installation done! Zoltan.Arpadffy> Zoltan.Arpadffy> You might want to purge DSA104:[USERS.ZAY.WORK.OPENSSL-101-STABLE-SNAP-20101124.INSTALL...] Zoltan.Arpadffy> Zoltan.Arpadffy> Thank you Richard. Zoltan.Arpadffy> You really did a good job. Zoltan.Arpadffy> Zoltan.Arpadffy> ...except the patches that are written above, I am reaming of the possibility to choose between the 64 and the 32 bit version build and reuse this "not-used option" in MAKEVMS.COM Zoltan.Arpadffy> Zoltan.Arpadffy> TITAN2_ZAY $ @MAKEVMS.COM Zoltan.Arpadffy> USAGE: @MAKEVMS.COM [Target] [not-used option] [Debug option] <Compiler> Zoltan.Arpadffy> Zoltan.Arpadffy> During the summer I have submitted a patch for this "feature"... wonder is there any change for this code to be merged in? I recall discussing it, because it seemed to be a no-op... but I can't recall getting to an answer (and I'll admit not being a good participant in that discussion). Cheers, Richard -- Richard Levitte rich...@levitte.org http://richard.levitte.org/ "Life is a tremendous celebration - and I'm invited!" -- from a friend's blog, translated from Swedish ______________________________________________________________________ OpenSSL Project http://www.openssl.org Development Mailing List openssl-dev@openssl.org Automated List Manager majord...@openssl.org