Minimum compiler version?

[Eric Roshan-Eisner]

I'm looking into contributing my RISCV Vector implementation of poly1305 [1] to openssl. It would need to be rewritten to conform to openssl's internal asm API. I see that most of the simd implementations are in perlasm. Could I contribute a RISCV implementation in C using the newish vector compiler intrinsics [2]? These are supported by gcc 14 and clang 17. I can't find any documented minimum compiler versions for the project as a whole, or for RISCV in particular.

Can openssl use features from newer compilers, or otherwise gate this implementation at build time?

-Eric

[1] https://github.com/edre/rvv-chacha-poly

[2] https://github.com/riscv-non-isa/rvv-intrinsic-doc

[rsbe...@nexbridge.com]

Hi Eric,

 

My requirement is a minimum c99 version so that OpenSSL can be built on older systems. We are planning to move to c11 around the October 2025 timeframe. Unfortunately, this is not a choice I can control, given compiler availability on the HPE NonStop platform – which is a very active environment.

 

Regards,

Randall

 

--
You received this message because you are subscribed to the Google Groups "openssl-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openssl-user...@openssl.org.
To view this discussion visit https://groups.google.com/a/openssl.org/d/msgid/openssl-users/75a8a7f1-0547-4858-b96d-f4a0e0d3c019n%40openssl.org.

[Neil Horman]

you'll find our supported platform list here:

https://mirror.openssl-library.org/policies/general-supplemental/platforms/

Generally speaking our supported platform use gcc 9 as a minimum, but we don't officially support RISCV (note they are considered unadopted on that list).  For those platforms the compilers that have been used by anyone who has tried are a myriad of older compilers (for instance the hpux targets claim gcc, and the latest version that I know of on that OS is gcc 4), so you may be likely to experience pushback from any implementation that uses the newer intrinsics.

Neil

--

[Eric Roshan-Eisner]

Thanks for the replies; that makes sense. I'll fix up my code and send a PR with a new asm file. Perlasm seems like overkill.

[Tomas Mraz]

perlasm is definitely not an overkill for CPUs that are used on
multiple platforms such as Windows.

If you're targetting RISC-V CPU which means AFAIK only POSIX platforms
with fairly new versions of gcc and/or clang, then this might be
acceptable in form of C intrinsics. However please note there is
already a lot of RISC-V assembler written in perlasm so doing it in
perlasm would still be preferred.

Tomas Mraz, Public Support and Security Manager, OpenSSL Foundation

Tomáš Mráz, Public Support and Security Manager, OpenSSL Foundation
Join the Code Protectors or support us on Github Sponsors
https://openssl-foundation.org/donate/