On Wed, 13 May 2020 at 02:00, Kevin Brott <kevin.brott at gmail.com> wrote: [...]> Latest snapshot is openssh-SNAP-20200502.tar.gz 02-May-2020 00:30 - is this the 8.3p1 candidate?No, it looks like the snapshot updater stopped updating. Damien's onto it. [...]> Using git-clone: > > *Linux Mint 19.3 / gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0 ** > fail tests here:* > ssh-add -l with different agent did not fail with exit code 1 (exit code 0) > failed simple agent test > Makefile:213: recipe for target 't-exec' failed > make[1]: *** [t-exec] Error 1 > make[1]: Leaving directory '/home/brottk/sandbox/openssh-portable/regress' > Makefile:673: recipe for target 't-exec' failed > make: *** [t-exec] Error 2This one I'm not sure about. [...]> *AIX 7200-04-01-1939 / gcc (GCC) 8.3.0 > fails out here ...* > ... > key option expiry-time invalid > key option expiry-time expired > key option expiry-time valid > failed key optionsThe individual test that failed should be a further back in the log. The debug logs for the failing invocations of ssh and sshd should be in regress/failed-ssh*.log, could you please send those (off-list if you prefer).> *Using the xlc_r compiler fails to compile enough to get to the tests ...*[...] The Makefile doesn't know how to make shared libraries with xlc so it can't build the sk-dummy library. We don't actually use it if you're not building with security key support, I'll see if I can make it skip building it too. Thanks for the thorough testing! -- Darren Tucker (dtucker at dtucker.net) GPG key 11EAA6FA / A86E 3E07 5B19 5880 E860 37F4 9357 ECEF 11EA A6FA (new) Good judgement comes with experience. Unfortunately, the experience usually comes from bad judgement.
On Wed, May 13, 2020 at 12:30:29PM +1000, Darren Tucker wrote: [...]> The Makefile doesn't know how to make shared libraries with xlc so it > can't build the sk-dummy library. We don't actually use it if you're > not building with security key support, I'll see if I can make it skip > building it too.This is a bit ugly but should do it. Can anyone think of a neater way? diff --git a/Makefile.in b/Makefile.in index 7c5ed0bd..c9e4294d 100644 --- a/Makefile.in +++ b/Makefile.in @@ -635,6 +635,8 @@ SK_DUMMY_OBJS=\ regress/misc/sk-dummy/fatal.lo \ ed25519.lo hash.lo ge25519.lo fe25519.lo sc25519.lo verify.lo +SK_DUMMY_LIBRARY=@SK_DUMMY_LIBRARY@ + .c.lo: Makefile.in config.h $(CC) $(CFLAGS_NOPIE) $(PICFLAG) $(CPPFLAGS) -c $< -o $@ @@ -648,7 +650,7 @@ regress-binaries: regress-prep $(LIBCOMPAT) \ regress/netcat$(EXEEXT) \ regress/check-perm$(EXEEXT) \ regress/mkdtemp$(EXEEXT) \ - regress/misc/sk-dummy/sk-dummy.so + $(SK_DUMMY_LIBRARY) regress-unit-binaries: regress-prep $(REGRESSLIBS) \ regress/unittests/sshbuf/test_sshbuf$(EXEEXT) \ diff --git a/configure.ac b/configure.ac index 003b235e..46038375 100644 --- a/configure.ac +++ b/configure.ac @@ -3137,6 +3137,10 @@ AC_MSG_RESULT([$enable_pkcs11]) AC_MSG_CHECKING([whether to enable U2F]) if test "x$enable_sk" = "xyes" ; then AC_DEFINE([ENABLE_SK], [], [Enable for U2F/FIDO support]) + AC_SUBST(SK_DUMMY_LIBRARY, [regress/misc/sk-dummy/sk-dummy.so]) +else + # Do not try to build sk-dummy library. + AC_SUBST(SK_DUMMY_LIBRARY, [""]) fi AC_MSG_RESULT([$enable_sk]) -- Darren Tucker (dtucker at dtucker.net) GPG key 11EAA6FA / A86E 3E07 5B19 5880 E860 37F4 9357 ECEF 11EA A6FA (new) Good judgement comes with experience. Unfortunately, the experience usually comes from bad judgement.
On Wed, 13 May 2020, Darren Tucker wrote:> On Wed, May 13, 2020 at 12:30:29PM +1000, Darren Tucker wrote: > [...] > > The Makefile doesn't know how to make shared libraries with xlc so it > > can't build the sk-dummy library. We don't actually use it if you're > > not building with security key support, I'll see if I can make it skip > > building it too. > > This is a bit ugly but should do it. Can anyone think of a neater way?No, and this should do the right thing via regress/test-exec,sh as it skips testing sk-* key types if no sk-dummy.so is found. ok djm> diff --git a/Makefile.in b/Makefile.in > index 7c5ed0bd..c9e4294d 100644 > --- a/Makefile.in > +++ b/Makefile.in > @@ -635,6 +635,8 @@ SK_DUMMY_OBJS=\ > regress/misc/sk-dummy/fatal.lo \ > ed25519.lo hash.lo ge25519.lo fe25519.lo sc25519.lo verify.lo > > +SK_DUMMY_LIBRARY=@SK_DUMMY_LIBRARY@ > + > .c.lo: Makefile.in config.h > $(CC) $(CFLAGS_NOPIE) $(PICFLAG) $(CPPFLAGS) -c $< -o $@ > > @@ -648,7 +650,7 @@ regress-binaries: regress-prep $(LIBCOMPAT) \ > regress/netcat$(EXEEXT) \ > regress/check-perm$(EXEEXT) \ > regress/mkdtemp$(EXEEXT) \ > - regress/misc/sk-dummy/sk-dummy.so > + $(SK_DUMMY_LIBRARY) > > regress-unit-binaries: regress-prep $(REGRESSLIBS) \ > regress/unittests/sshbuf/test_sshbuf$(EXEEXT) \ > diff --git a/configure.ac b/configure.ac > index 003b235e..46038375 100644 > --- a/configure.ac > +++ b/configure.ac > @@ -3137,6 +3137,10 @@ AC_MSG_RESULT([$enable_pkcs11]) > AC_MSG_CHECKING([whether to enable U2F]) > if test "x$enable_sk" = "xyes" ; then > AC_DEFINE([ENABLE_SK], [], [Enable for U2F/FIDO support]) > + AC_SUBST(SK_DUMMY_LIBRARY, [regress/misc/sk-dummy/sk-dummy.so]) > +else > + # Do not try to build sk-dummy library. > + AC_SUBST(SK_DUMMY_LIBRARY, [""]) > fi > AC_MSG_RESULT([$enable_sk]) > > -- > Darren Tucker (dtucker at dtucker.net) > GPG key 11EAA6FA / A86E 3E07 5B19 5880 E860 37F4 9357 ECEF 11EA A6FA (new) > Good judgement comes with experience. Unfortunately, the experience > usually comes from bad judgement. >
On Wed, 13 May 2020, Darren Tucker wrote:> On Wed, 13 May 2020 at 02:00, Kevin Brott <kevin.brott at gmail.com> wrote: > [...] > > Latest snapshot is openssh-SNAP-20200502.tar.gz 02-May-2020 00:30 - is this the 8.3p1 candidate? > > No, it looks like the snapshot updater stopped updating. Damien's onto it.Should be good now - sorry for the inconvenience. -d