pinentry-curses can be forced to loop endlessly

NIIBE Yutaka gniibe at
Tue Jun 16 04:43:11 CEST 2015

Thank you for your patience.  I think that I finally understand
the problem.

On 06/15/2015 11:23 PM, Steffen Nurpmeso wrote:
> Oh it was fine (was it even coloured?) except that it entered this
> endless loop when i tested the interruption case.

The problem is: Typing control-C, you expect killing pinentry along
with gpg, but pinentry is still there.

With old version of pinentry, it looped endlessly.  With new version,
although it doesn't loop endlessly, the process of pinentry remains.

Configure gpg-agent with pinentry-curses, minimum test case is:

   $ gpg --sign some.txt
   [Type Control-C, when you see curses dialog]

The pinentry process won't be killed by Control-C.

(This is internal of gpg)
Expected behavior is when gpg cancels its operation, it should notify
gpg-agent so that gpg-agent can cancel pinentry.

I created:

More information about the Gnupg-devel mailing list