Hi Rahul,

We don’t have Solaris in our CI (because it is not a tier1 platform) so we 
can’t see these kind of issues… thanks for reaching out.

It looks like the makefile is attempting to build a vectorized version of 
chachapoly on an intel platform supporting AVX2 but that for some reason the 
file libintvector.h which maps intrinsics to C types does not take into account 
some special case for Solaris. 

I could suggest a few things:
0. Check you are building for Intel AVX2, if not the Makefile probably needs to 
be fixed
1. Could you make sure NSS builds if you turn vectorization off; there should 
be some flag like NSS_DISABLE_AVX2 (or something like this)
2. look at libintvector.h (should probably be somewhere in freebl/verified/) 
and make sure that the Solaris case is handled for the 128 and 256 vectorized 
cases

I think 2. is likely the problem and needs to be updated but I haven’t looked 
into the details here. Could you have a look and if you manage to fix the 
problem and to submit a patch in bugzilla? Anna or myself can take care of 
landing that upstream in HACL* and NSS.

Best,
Benjamin




> From: Rahul S
> Sent: Tuesday, March 22, 2022 4:21 PM
> To: dev-tech-crypto@mozilla.org <dev-tech-crypto@mozilla.org>
> Subject: Build NSS 3.68.1 on Solaris 11 x86 with Solarisstudio 12.5
>  
> Hello All,
> 
> I am trying to build NSS version 3.68.1 (ESR release) on Solaris 11x86 
> machine using Solarisstudio 12.5 . I am facing the following compilation issue
> 
> 
> cc -o SunOS5.11_i86pc_64_DBG.OBJ/SunOS_SINGLE_SHLIB/Hacl_Poly1305_256.o -c 
> -std=c99 -g -KPIC -DSVR4 -DSYSV -D__svr4 -D__svr4__ -DSOLARIS -D_REENTRANT 
> -xprefetch=no -m64 -xs  -DXP_UNIX -DSHLIB_SUFFIX=\"so\" 
> -DSHLIB_PREFIX=\"lib\" -DSHLIB_VERSION=\"3\" -DSOFTOKEN_SHLIB_VERSION=\"3\" 
> -DRIJNDAEL_INCLUDE_TABLES -DDEBUG -UNDEBUG -DNSS_NO_INIT_SUPPORT 
> -DUSE_UTIL_DIRECTLY -DNO_NSPR_10_SUPPORT 
> -DSSL_DISABLE_DEPRECATED_CIPHER_SUITE_NAMES -DNSS_USE_64 -DNSS_X86_OR_X64 
> -DNSS_X64 -DINTEL_GCM -DNSS_BEVAND_ARCFOUR -DMPI_AMD64 -DMP_ASSEMBLY_MULTIPLY 
> -DNSS_USE_COMBA -DMP_IS_LITTLE_ENDIAN -DUSE_HW_AES -DKRML_VERIFIED_UINT128 
> -DMP_API_COMPATIBLE 
> -I../../../../../built/nss/SunOS5.11_i86pc_64_DBG.OBJ/include 
> -I../../../../../built/nss/SunOS5.11_i86pc_64_DBG.OBJ/include 
> -I../../../../../built/nss/private/nss -Impi -Iecl -Iverified 
> -Iverified/kremlin/include -Iverified/kremlin/kremlib/dist/minimal 
> -Ideprecated  -mssse3 -msse4.1 -msse4.2 -mavx -mavx2 -maes -mpclmul 
> verified/Hacl_Poly1305_256.c
> cc: Warning: Option -msse4.1 passed to ld, if ld is invoked, ignored otherwise
> cc: Warning: GCC option -mpclmul ignored. Suggest using -xarch=aes
> "verified/Hacl_Poly1305_256.c", line 90: warning: implicit function 
> declaration: _mm256_insert_epi64
> "verified/Hacl_Poly1305_256.c", line 90: warning: implicit function 
> declaration: _mm256_extract_epi64
> "verified/Hacl_Poly1305_256.c", line 90: assignment type mismatch:
>         vector [4] of unsigned long long "=" int
> "verified/Hacl_Poly1305_256.c", line 95: assignment type mismatch:
>         vector [4] of unsigned long long "=" int
> "verified/Hacl_Poly1305_256.c", line 100: assignment type mismatch:
>         vector [4] of unsigned long long "=" int
> "verified/Hacl_Poly1305_256.c", line 105: assignment type mismatch:
>         vector [4] of unsigned long long "=" int
> "verified/Hacl_Poly1305_256.c", line 110: assignment type mismatch:
>         vector [4] of unsigned long long "=" int
> cc: acomp failed for verified/Hacl_Poly1305_256.c
> gmake[6]: *** 
> [SunOS5.11_i86pc_64_DBG.OBJ/SunOS_SINGLE_SHLIB/Hacl_Poly1305_256.o] Error 2
> 
> Any help to resolve this issue would be really helpful.
> 
> Best Regards,
> Rahul.S
> -- 
> You received this message because you are subscribed to the Google Groups 
> "dev-tech-crypto@mozilla.org" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to dev-tech-crypto+unsubscr...@mozilla.org.
> To view this discussion on the web visit 
> https://groups.google.com/a/mozilla.org/d/msgid/dev-tech-crypto/BY5PR10MB39230AD5183416E73100797A9CAE9%40BY5PR10MB3923.namprd10.prod.outlook.com.

-- 
You received this message because you are subscribed to the Google Groups 
"dev-tech-crypto@mozilla.org" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dev-tech-crypto+unsubscr...@mozilla.org.
To view this discussion on the web visit 
https://groups.google.com/a/mozilla.org/d/msgid/dev-tech-crypto/5BC462EC-FD45-4BED-83D0-47FA2A7CF432%40mozilla.com.

Reply via email to