>> The SecurityPolicy.pdf claims that HP-UX 11i IA64 is a Supported >> Configuration; how can this claim be made when the code does nto even >> compile correctly? > > The FIPS module compiles correctly but there is the duplicated symbol issue > when you use the FIPS capable OpenSSL. > > Although we can't change the FIPS module symbols without a change letter you > can change the symbols on the OpenSSL side. So if you change them to (say) > ossl_AES_Te and ossl_AES_Td and make sure the non-FIPS capable OpenSSL still > compiles it should work.
Presumably simpler option should be to remove .global AES_T[ed]# directives in crypto/aes/asm/aes-ia64.S (see attached). I don't have possibility to actually verify it on HP-UX, but at least Linux linker output look proper (can't execute either, only look at binaries).
diff --git a/crypto/aes/asm/aes-ia64.S b/crypto/aes/asm/aes-ia64.S index 7f6c4c3..0018e02 100644 --- a/crypto/aes/asm/aes-ia64.S +++ b/crypto/aes/asm/aes-ia64.S @@ -791,7 +791,6 @@ AES_decrypt: // leave it in .text segment... .align 64 -.global AES_Te# .type AES_Te#,@object AES_Te: data4 0xc66363a5,0xc66363a5, 0xf87c7c84,0xf87c7c84 data4 0xee777799,0xee777799, 0xf67b7b8d,0xf67b7b8d @@ -957,7 +956,6 @@ AES_Te: data4 0xc66363a5,0xc66363a5, 0xf87c7c84,0xf87c7c84 .size AES_Te#,2048+256 // HP-UX assembler fails to ".-AES_Te#" .align 64 -.global AES_Td# .type AES_Td#,@object AES_Td: data4 0x51f4a750,0x51f4a750, 0x7e416553,0x7e416553 data4 0x1a17a4c3,0x1a17a4c3, 0x3a275e96,0x3a275e96
_______________________________________________ openssl-dev mailing list To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-dev
