This was reported to me by a customer:
gpgsm: DBG: chan_4 -> GET_PASSPHRASE --data -- X X X Please+enter+the+passphrase+to+unprotect+the+PKCS#12+object. gpgsm: DBG: chan_4 <- INQUIRE PINENTRY_LAUNCHED 31933 qt 1.2.1 /dev/pts/3 xterm-256color :1 20600/1000/5 1000/1000 0 gpgsm: DBG: chan_4 -> END gpgsm: DBG: chan_4 <- D foobar,123456 gpgsm: DBG: chan_4 <- OK
Usually we do not log passphrases but for PKCS#12 we apparently do.
I guess that it's a case of specifying static passphrase. If so, here is the patch:
diff --git a/g10/call-agent.c b/g10/call-agent.c index cb7053396..c44c1cddb 100644 --- a/g10/call-agent.c +++ b/g10/call-agent.c @@ -161,6 +161,7 @@ default_inq_cb (void *opaque, const char *line) || has_leading_keyword (line, "NEW_PASSPHRASE")) && opt.pinentry_mode == PINENTRY_MODE_LOOPBACK) { + assuan_begin_confidential (parm->ctx); if (have_static_passphrase ()) { s = get_static_passphrase (); @@ -187,6 +188,7 @@ default_inq_cb (void *opaque, const char *line) err = assuan_send_data (parm->ctx, pw, strlen (pw)); xfree (pw); } + assuan_end_confidential (parm->ctx); } else if ((s = has_leading_keyword (line, "CONFIRM")) && opt.pinentry_mode == PINENTRY_MODE_LOOPBACK diff --git a/sm/call-agent.c b/sm/call-agent.c index 883c0c644..7f7205f26 100644 --- a/sm/call-agent.c +++ b/sm/call-agent.c @@ -222,7 +222,9 @@ default_inq_cb (void *opaque, const char *line) && have_static_passphrase ()) { const char *s = get_static_passphrase (); + assuan_begin_confidential (parm->ctx); err = assuan_send_data (parm->ctx, s, strlen (s)); + assuan_end_confidential (parm->ctx); } else log_error ("ignoring gpg-agent inquiry '%s'\n", line);
(I also found similar case for gpg as well as gpgsm.)
Thanks, I will test this and if it works as expected I would also put it in 2.2. since it was pointed out to me from a customer at our approval institution and I think they will be glad if they see that this is gone in the next release and I don't see any regression risk associated with that change.
Sorry @ebo tested this on Windows with 2.2. I myself should have tested it since the test is trivial and only took me about 30 seconds to type. Similar to T6701 this should have never reached the QA stage. I am including myself now that we have someone for QA that I test my own changes less. We need to talk / think about that in our whole team. We developers should test more before sending an issue into QA.
My test:
GNUPGHOME=$(mktemp -d) gpgsm --debug-level guru --import berta.boss@demo.gnupg.com.p12 ..... gpgsm: DBG: chan_4 -> GETINFO version gpgsm: DBG: chan_4 <- D 2.4.4-beta63 gpgsm: DBG: chan_4 <- OK gpgsm: DBG: chan_4 -> OPTION allow-pinentry-notify gpgsm: DBG: chan_4 <- OK gpgsm: DBG: chan_4 -> GET_PASSPHRASE --data -- X X X Please+enter+the+passphrase+to+unprotect+the+PKCS#12+object. gpgsm: DBG: chan_4 <- INQUIRE PINENTRY_LAUNCHED 23356 qt 1.2.1-beta21 /dev/pts/6 xterm-256color :1 20600/1000/5 1000/1000 0 gpgsm: DBG: chan_4 -> END gpgsm: DBG: chan_4 <- D abc,123456 gpgsm: DBG: chan_4 <- OK