Sorry, I only now noticed that you used the --export-secret-ssh-key. Unfortunately commit
rGafe5fcda52e88438c7a7278117b2e03f510a9c1c states in the comment:
"Due to time constraints the code is not yet ready." Let's turn this into a feature request.

assuan_sock_accept approach is taken in gnupg master.

gniibe/t6606 patches are all pushed into master.

Applied: rGb8c5d99406c9: gpg,agent,kbx,sm,scd,tpm2d,g13: Use assuan_control.

In T6606#173044, @gniibe wrote:

More care is needed to be perfect; There are places in GnuPG where assuan_sock_connect may be used before syscall clamp set up (after the first assuan_sock_bind failure).

Applied to master.

Applied the changes for libassuan T6487 into gniibe/t6606.

Pushed the change in gniibe/t6606 branch.

Another approach would be:

• Use assuan_sock_accept which has consistent API with gnupg_fd_t

On 64-bit Windows, the situation now is:

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>

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.

Changes are pushed.

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).

Just to clarify this change for readers not accustomed to Windows internals: This function was used to translate the file descriptor as passed to gpg (which is a HANDLE) to the libc file descriptor as used by stdio. Obviously we won't anymore work with stdio file descriptors in the future but use the Windows32 API (ReadFile et al). libc fds 0,1,2 are handled in a special way on Windows.

Fixed in master.

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

With the fix of T6523, make check goes all well (on Wine emulation and on Windows, for i686 and for x86_64).

Fixed in master.

I modified ffi.c, to have renamed process-spawn-io function doing I/O by C.

Calling assuan_release before kbx_client_data_release is the best (and we join the thread).

tests/openpgp/import.scm hangs with 4096*4.

Test with Wine (i686) emulation, I encountered another hang at: Checking armored_key_8192

It looks like having the datastream_thread may be not worth.
One possibility is to implement synchronous read from pipe in kbx_client_data_wait, instead of datastream_thread.

See also commit rG6fcc263c18 from 2020 where I switched to D-lines.

Let's schedule that for 2.6

You are right, it is a bad habit not to check this. Thanks for your patch.