Use of FD2INT for the first argument of select is semantically not good. It's the number of file descriptor. When we use FD2INT here, the type is converted to 64-bit integer, then implicitly demoted to 32-bit integer. We need new macro, say, FD2NUM to convert FD into 32-bit integer.
<--- done in: rGea1935252e28: commond: Introduce FD2NUM to express conversion to number of fds.

Here is a test program for 64-bit Windows to see how cast works:

#include <stdint.h>
#include <stdio.h>

Update the patch to allow --compatibility-flags hashing-in-parallel.

Fixed in master.
Will backport into 2.4, after testing.

Updated the patch:

This is a patch for master (which uses nPth for gpg):

I'm trying to use a thread for computation of hashing. It works, but no performance gain, due to nPth.
I do:

npth_unprotect ();
gcry_md_write (mfx->md, mfx->buf, mfx->written);
npth_protect ();

to expect hashing is done in parallel.

It seemed I was wrong that it is due to buffering.
In the use case of --sign and --encrypt, hashing is done with IOBUF's 64KiB buffer (already).

I observed the benchmark by libgcrypt (Windows emulation 32-bit on Debian):

Thank you for your report.

OK. I'll take the signing part (possible performance improvement).

I looked through the code. What I observed is:

• By jussi's improvements, AEAD code is optimized with AEAD_ENC_BUFFER_SIZE of 64KiB
  • this contributes much for better performance
• If we invoke gpg --sign | gpg --encrypt then we can take advantage of multiple CPUs (but gpg is currently not automatically threaded in that way)
  • signing could be improved likewise, using larger buffer like 64KiB
• CFB+MDC, it uses two functions together; encryption and hashing, and not with larger buffer like 64KiB
  • when signed, it also does hashing for signing, so three functions

The case in check_special_filename is fixed. So, there is no cases in GnuPG where the value of out of range is silently converted to wrong value.

Remaining places are:

• common/sysutil.c debug output to format HANDLE --> T6597
• common/sysutil.c iobuf_get_fd and its use cases --> T6580
• agent, kbx, scdaemon, dirmngr: socket and FD2INT/INT2FD --> T6598

Except a case, all use cases of translate_sys2libc_fd_int is with a result of integer from command line argument.

Add the check of digest algorithm for EdDSA in: rCd15fe6aac10b: cipher:ecc:fips: Only allow defined digest algo for EdDSA.

Changes are pushed.

No, there are use cases in GnuPG, where we specify the hash algo for signing, and our own tests/benchmark.c.

For the first issue, I added a check in: rCf65c30d470f5: cipher:ecc:fips: Reject use of SHAKE when it's ECDSA with RFC6979.

We need to keep the gpgtar part of commit in rG2756147e392c: gpg,sm,tools: Use string for option --*-fd..

The changes are intrusive to other implementations (POSIX and Windows 32-bit).
So, I revert the changes of replacing translate_sys2libc_fd_int.

I don't argue about the technical necessity for the change. I agree the fact it works (without such changes).

Fixed in master.

Pushed a change in master.

Applied.

I found the case of X.509, which also uses fixed length output for RSA-PSS and ECDSA: https://www.rfc-editor.org/rfc/rfc8692.html

The use cases are:

• oPassphraseFD for gpgsm, gpg
• oStatusFD for gpg-auth, gpg-wks-client, gpg-card, gpg-pair-tool, gpgtar, gpgconf, gpgsm, gpg, gpgv
• oLoggerFD for gpgsm, gpg, gpgv
• oAttributeFD for gpg
• oCommandFD for gpg
• oOverrideSessionKeyFD for gpg