I just learned that WSAStartup can be called multiple times. So, it doesn't cause any erroneous behavior which I had been afraid of.

Given the situation we have call of WSAStartup in assuan_sock_init (for Windows), the solution would be:

• Removal of call of WSAStartup in _init_common_subsystems
• Even though it is not needed for POSIX system and it is only needed to call WAStartup on Windows, calling assuan_sock_init from each application (including gpg, gpgsm, dirmngr/dirmngr-client, and tools/* which uses libassuan), would be the solution (not perfect one, though, because it allocates sock_ctx)

I am happy that your use case will be supported, and the bug was fixed before the release.
It's me who say "thank you" to you!

Thanks a lot.
I pushed a fix as rG46d185f60397: scd: PC/SC: Don't release the context when it's in use..

Ah, I identified an issue.
While it's in a loop of trying readers (in select_application in scd/app.c), it should not deallocate resources to access readers, even if reference count == 0.
I'll fix.

Thanks for your testing.

Thanks for your report.
Major reason was multiple card readers/tokens were not supported by PC/SC handling of scdaemon, only a single reader was assumed, so, user had to specify one if it's not the first one.
Multiple reader by PC/SC support was added in master (to be 2.3), so, I think the problem is solved in master.

Sorry, I was confused by assuan_socket_ API and assuan_sock_ API.

Here are the fixes:

diff --git a/common/init.c b/common/init.c
index 073c5cd8a..dbdf40527 100644
--- a/common/init.c
+++ b/common/init.c
@@ -161,17 +161,6 @@ _init_common_subsystems (gpg_err_source_t errsource, int *argcp, char ***argvp)
   /* Try to auto set the character set.  */
   set_native_charset (NULL);

Call of WSAStartup in dirmngr/http.c is no problem, as we define HTTP_NO_WSASTARTUP.

This fix reveals the problem of: T4994: Windows: assuan_sock_init or WSAStartup by main/_init_common_subsystem

A reference might help:
https://blogs.itemis.com/en/openpgp-on-the-job-part-8-ssh-with-openpgp-and-yubikey

@mbrinkers : I think that it was fixed in GnuPG 2.2.21 by T4908: ECDH with AES-128 decryption failure when fully padded.
It was unfortunate that this bug report didn't work to solve problem, with malformed data and discussion went to unrelated thing.

So, where does "ssh-add" command come from? IIUC, it is from OpenSSH.

You mean running OpenSSH (and its tool ssh-add) on Windows, right?
It is not supported. PuTTY is supported.

• compressed representation of EC point can be used in:
  • public key
  • (exporting) private key
  • signature
  • ECDH ephemeral key
• Accepting compressed representation,for the initial implementation, I'd like to limit our effort for curves of NIST and Brainpool, except NIST P-224, which p = 3 mod 4.

Pushed fix to master and STABLE-BRANCH-2-2.

Thanks for your log.

(3) _gcry_ecc_os2ec in libgcrypt/cipher/ecc-misc.c should be modified to support parsing compressed representation.

While I see that it's not the matter of actual use case (but how gpg can be immune to fuzzing), code clean up would be good here.

Thanks for the patch.
I see your point in T4975: undefined-shift in block_filter.
You are right that we have a problem of possible overflow (which could be kicked by fuzzing) here.
(The actual impact would be small, though).

What kind of API should we offer?
(1) offering something like q@comp name for gcry_mpi_ec_get_mpi
But...
If the intended use case will be in create_request function in gpg/sm/certreqgen.c, the 'q' is already generated in the form of SEXP.
It is up to an application (gpgsm), to convert non-compressed point representation to compressed point representation, here.

Or this (don't allow anon keys for different usage):

diff --git a/g10/pubkey-enc.c b/g10/pubkey-enc.c
index 14cbdbb0f..b8d4059cd 100644
--- a/g10/pubkey-enc.c
+++ b/g10/pubkey-enc.c
@@ -91,9 +91,6 @@ get_session_key (ctrl_t ctrl, struct pubkey_enc_list *list, DEK *dek)
       if (err)
         break;

Do you mean something like this?

Fixed in rM1b840a151ad7: python: Fix how to generate documentation..

It's in master (to be gnupg 2.3).
Enjoy.

It seems that nl_langinfo(CODESET) returns US-ASCII on your system.

Yes, it will fix the problem on x32, I suppose.
If it's difficult for dpkg, for some reason for now, workaround for gpgme packaging is disabling pie hardening for x32 until pie will be its compiler default.
For gpgme, it is only test binaries which matter (pie or not), so, the impact (for x32) is minimum.

Some information of Qt5 about -fpic:

• https://lists.qt-project.org/pipermail/development/2015-May/021557.html
• https://bugzilla.redhat.com/show_bug.cgi?id=1700435

Debian's GCC build for PIE default: https://salsa.debian.org/toolchain-team/gcc/-/blob/master/debian/rules.defs#L1400

Here is my understanding. My point is it's not problem of gpgme. To fix it correctly, I think that dpkg should be fixed and it would be needed to fix Qt too.

I think that it is the problem of dpkg to override the compiler flag by the spec file. When compiler default is -fPIE, it works well. If not (for the case of x32), it fails.
In the past, hurd-i386 had same issue, but compiler default seems to be now -fPIE, thus no problem.

Thanks for your report.

When I test it on Debian, disabling by,

Please get log of dirmngr, by putting

log-file /run/user/<YOURNUMBER-LIKE-1000>/dirmngr.log

I think the feature is not (yet) supported on Windows.
Please see: T3883: Add Win32-OpenSSH support to gpg-agent's ssh-agent

Pushed to master as rGa763bb2580b0: gpg,agent: Support Ed448 signing..

Update to [rGc94eea15d}.
Hash defaults to SHA512.

(1) Has no (flags eddsa) in key in SEXP.
(2) Has no (flags eddsa) and no (hash-algo shake256) in data to be signed in SEXP.
(3) Has no (flags eddsa) and no (hash-algo shake256) in data to be verified in SEXP.
(4) Uses SHA256 for hashing of OpenPGP data

Update to rG4bdade5b0bea: agent: Use get_pk_algo_from_key.

The changes just follow the existing practice of Ed25519, which does: