fwiw, ensuring that overflow for either field results in ULONG_MAX (rather than wrapping around) would go a long way toward this problem being something that we can reasonably put off for another 50 years.

I found a solution for master and 2.1.19 which minimizes the risk of regressions:

In case you use gpgme we have a flag which can be queried to see whether a redraw is required:

@gniibe Thank you!

In 2.2.18, this fix is not included. (partial fix was reverted)

Applied and pushed.

The last fix was in 3681ee7dc1e9d8c94fdb046d7be0bbcfeba1cfe9, on 2017-07-05.
And it is included from the release of 2.1.22.

allow-loopback-pinentry in gpg-agent.conf is actually the default. This options advises gpg-agent to accept a request for a loopback-pinentry. If you would configure no-allow-loopback-pinentry, requests from gpg to use a loopback pinentry are rejected.

I think this is now resolved.

@gniibe - Thanks for your explanation. Is --pinentry-mode=loopback the same as specifying in ~/.gnupg/gpg-agent.conf:

My analysis is that it's not a race condition but... it's about secure memory.
It is true that we have a race condition between putting an entry to cache after pinentry interaction _and_ next examining cache to invoke pinentry. But for this test case, the gpg process of unlock the key (and cache the passphrase) is finished before running the run-threaded command.

I believe the problem was fixed in the master of pinentry with newer gpg-error-config and libassuan-config which support cross build better.

Confirmed that the support of --no-global-grab doesn't work well.

My message above is: The reported issue of ^C was fixed in pinentry-tty and GnuPG in master branch. Please test that fixes.

Please note that pinentry-tty/curses is a kind of emulation of CLI user interface, it's not the real one (I'm going to explain in the next paragraph).
It is, by any means, not robust, as users would expect, from the implementation's view. It only works specific simple use cases (while I do my best to stabilize it in master branch of GnuPG).

That is actually a GnuPG thing. We originally did it this way to help people remember their passphrase before they start using the key. I agree it is annoying and I would like to remove it too. At the same time we should really think about making no-passphrase the default and require it only with certain compliance settings.

The most plausible fix to the Y2K38 problem on 32-bit machines is to simply move to a 64-bit time_t at the same time as any other major system-wide ABI break. However, if that ABI break doesn't also change the size of long to more than 32 bits, GPGME will remain unfixed in spite of any architectural correction.

Fixed for 2.2.19 and master