[Buildroot] [PATCH RFC 0/2] Add swtpm to host packages

[Buildroot] [PATCH RFC 0/2] Add swtpm to host packages

[Jarkko Sakkinen]

This work originates from my test sandbox for linux-tpmdd korg tree.
I'm not sure how to refine this as mainline BuildRoot patch set so
I just dump what I've used for long time with RFC stamp :-) Please
instruct me and I'll try to follow the steps.

Cc: Stefan Berger <stefanb@linux.ibm.com>
Cc: Peter Korsgaard <peter@korsgaard.com>

References:
- https://gitlab.com/jarkkojs/linux-tpmdd-test/-/blob/main/board/qemu/run-qemu.sh.in?ref_type=heads
- https://git.kernel.org/pub/scm/linux/kernel/git/jarkko/linux-tpmdd.git

Jarkko Sakkinen (2):
  package/json-glib: add host build
  package/swtpm: add host package

 package/json-glib/json-glib.mk                |  1 +
 package/libtpms/Config.in                     |  9 ++++
 package/libtpms/libtpms.hash                  |  1 +
 package/libtpms/libtpms.mk                    | 15 ++++++
 .../0001-comment-out-expect-and-socat.patch   | 46 +++++++++++++++++++
 package/swtpm/Config.host.in                  |  8 ++++
 package/swtpm/swtpm.hash                      |  1 +
 package/swtpm/swtpm.mk                        | 17 +++++++
 8 files changed, 98 insertions(+)
 create mode 100644 package/libtpms/Config.in
 create mode 100644 package/libtpms/libtpms.hash
 create mode 100644 package/libtpms/libtpms.mk
 create mode 100644 package/swtpm/0001-comment-out-expect-and-socat.patch
 create mode 100644 package/swtpm/Config.host.in
 create mode 100644 package/swtpm/swtpm.hash
 create mode 100644 package/swtpm/swtpm.mk

--
2.43.0

_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

[Buildroot] [PATCH RFC 1/2] package/json-glib: add host build

[Jarkko Sakkinen]

Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
---
 package/json-glib/json-glib.mk | 1 +
 1 file changed, 1 insertion(+)

diff --git a/package/json-glib/json-glib.mk b/package/json-glib/json-glib.mk
index cd53f24cee..c42506993b 100644
--- a/package/json-glib/json-glib.mk
+++ b/package/json-glib/json-glib.mk
@@ -34,3 +34,4 @@ endif
 JSON_GLIB_LDFLAGS = $(TARGET_LDFLAGS) $(TARGET_NLS_LIBS)
 
 $(eval $(meson-package))
+$(eval $(host-meson-package))
-- 
2.43.0

_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

[Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

Add swtpm and its dependency libtpms to host packages. These are useful
for emulating TPM in QEMU environment.

Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
---
 package/libtpms/Config.in                     |  9 ++++
 package/libtpms/libtpms.hash                  |  1 +
 package/libtpms/libtpms.mk                    | 15 ++++++
 .../0001-comment-out-expect-and-socat.patch   | 46 +++++++++++++++++++
 package/swtpm/Config.host.in                  |  8 ++++
 package/swtpm/swtpm.hash                      |  1 +
 package/swtpm/swtpm.mk                        | 17 +++++++
 7 files changed, 97 insertions(+)
 create mode 100644 package/libtpms/Config.in
 create mode 100644 package/libtpms/libtpms.hash
 create mode 100644 package/libtpms/libtpms.mk
 create mode 100644 package/swtpm/0001-comment-out-expect-and-socat.patch
 create mode 100644 package/swtpm/Config.host.in
 create mode 100644 package/swtpm/swtpm.hash
 create mode 100644 package/swtpm/swtpm.mk

diff --git a/package/libtpms/Config.in b/package/libtpms/Config.in
new file mode 100644
index 0000000000..7ef61cf53c
--- /dev/null
+++ b/package/libtpms/Config.in
@@ -0,0 +1,9 @@
+config BR2_PACKAGE_LIBTPMS
+	bool "libtpms"
+        depends on BR2_USE_WCHAR # glib2
+        depends on BR2_TOOLCHAIN_HAS_THREADS # glib2
+        depends on BR2_USE_MMU # glib2
+	help
+	  TPM emulation library
+
+	  https://github.com/stefanberger/libtpms
diff --git a/package/libtpms/libtpms.hash b/package/libtpms/libtpms.hash
new file mode 100644
index 0000000000..c31d824af6
--- /dev/null
+++ b/package/libtpms/libtpms.hash
@@ -0,0 +1 @@
+sha256  2807466f1563ebe45fdd12dd26e501e8a0c4fbb99c7c428fbb508789efd221c0  v0.9.6.tar.gz
diff --git a/package/libtpms/libtpms.mk b/package/libtpms/libtpms.mk
new file mode 100644
index 0000000000..5b1151baff
--- /dev/null
+++ b/package/libtpms/libtpms.mk
@@ -0,0 +1,15 @@
+################################################################################
+#
+# libtpms
+#
+################################################################################
+
+LIBTPMS_VERSION = v0.9.6
+LIBTPMS_SOURCE = $(LIBTPMS_VERSION).tar.gz
+LIBTPMS_SITE = $(call github,stefanberger,libtpms,$(LIBTPMS_VERSION))
+LIBTPMS_LICENSE = BSD-3-Clause
+LIBTPMS_INSTALL_STAGING = YES
+LIBTPMS_AUTORECONF = YES
+
+$(eval $(autotools-package))
+$(eval $(host-autotools-package))
diff --git a/package/swtpm/0001-comment-out-expect-and-socat.patch b/package/swtpm/0001-comment-out-expect-and-socat.patch
new file mode 100644
index 0000000000..09dcc49a7b
--- /dev/null
+++ b/package/swtpm/0001-comment-out-expect-and-socat.patch
@@ -0,0 +1,46 @@
+From 067c32ba93774b273de9af872b5587798dcabb15 Mon Sep 17 00:00:00 2001
+From: Jarkko Sakkinen <jarkko@kernel.org>
+Date: Tue, 19 Dec 2023 05:21:20 +0200
+Subject: [PATCH] configure.ac: comment out "expect" and "socat"
+
+Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
+---
+ configure.ac | 16 ++++++++--------
+ 1 file changed, 8 insertions(+), 8 deletions(-)
+
+diff --git a/configure.ac b/configure.ac
+index 49caf96..4acc763 100644
+--- a/configure.ac
++++ b/configure.ac
+@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks" != "x"],[
+ pcr_bank_checks
+ AC_SUBST([DEFAULT_PCR_BANKS])
+ 
+-AC_PATH_PROG([EXPECT], expect)
+-if test "x$EXPECT" = "x"; then
+-	AC_MSG_ERROR([expect is required: expect package])
+-fi
++# AC_PATH_PROG([EXPECT], expect)
++# if test "x$EXPECT" = "x"; then
++# 	AC_MSG_ERROR([expect is required: expect package])
++# fi
+ 
+ AC_PATH_PROG([GAWK], gawk)
+ if test "x$GAWK" = "x"; then
+ 	AC_MSG_ERROR([gawk is required: gawk package])
+ fi
+ 
+-AC_PATH_PROG([SOCAT], socat)
+-if test "x$SOCAT" = "x"; then
+-	AC_MSG_ERROR([socat is required: socat package])
+-fi
++# AC_PATH_PROG([SOCAT], socat)
++# if test "x$SOCAT" = "x"; then
++# 	AC_MSG_ERROR([socat is required: socat package])
++# fi
+ 
+ AC_PATH_PROG([BASE64], base64)
+ if test "x$BASE64" = "x"; then
+-- 
+2.40.1
+
diff --git a/package/swtpm/Config.host.in b/package/swtpm/Config.host.in
new file mode 100644
index 0000000000..e77eea2aa5
--- /dev/null
+++ b/package/swtpm/Config.host.in
@@ -0,0 +1,8 @@
+config BR2_PACKAGE_HOST_SWTPM
+	bool "swtpm-host"
+	depends on BR2_PACKAGE_GOBJECT_INTROSPECTION_ARCH_SUPPORTS # gobject-introspection
+	select BR2_PACKAGE_GOBJECT_INTROSPECTION
+	help
+	  Compiles SWTPM software TPM emulator for the host.
+
+	  https://github.com/stefanberger/swtpm
diff --git a/package/swtpm/swtpm.hash b/package/swtpm/swtpm.hash
new file mode 100644
index 0000000000..882f06d7a5
--- /dev/null
+++ b/package/swtpm/swtpm.hash
@@ -0,0 +1 @@
+sha256 7bba52aa41090f75087034fac5fe8daed10c3e7e7234df7c9558849318927f41  v0.8.1.tar.gz
diff --git a/package/swtpm/swtpm.mk b/package/swtpm/swtpm.mk
new file mode 100644
index 0000000000..79fbf1f420
--- /dev/null
+++ b/package/swtpm/swtpm.mk
@@ -0,0 +1,17 @@
+################################################################################
+#
+# swtpm
+#
+################################################################################
+
+SWTPM_VERSION = v0.8.1
+SWTPM_SOURCE = $(SWTPM_VERSION).tar.gz
+SWTPM_SITE = $(call github,stefanberger,swtpm,$(SWTPM_VERSION))
+SWTPM_LICENSE = BSD-3-Clause
+SWTPM_AUTORECONF = YES
+
+HOST_SWTPM_DEPENDENCIES = host-libtasn1 host-openssl host-pkgconf host-json-glib host-libtpms
+HOST_SWTPM_CONF_ENV = PKG_CONFIG="$(PKG_CONFIG_HOST_BINARY)"
+HOST_SWTPM_CONF_OPTS = --without-seccomp
+
+$(eval $(host-autotools-package))
-- 
2.43.0

_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Thu, 2024-03-21 at 16:30 -0400, Stefan Berger wrote:
> 
> 
> On 3/21/24 14:21, Jarkko Sakkinen wrote:
> > Add swtpm and its dependency libtpms to host packages. These are
> > useful
> > for emulating TPM in QEMU environment.
> > 
> > Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> > ---
> 
> > +---
> > + configure.ac | 16 ++++++++--------
> > + 1 file changed, 8 insertions(+), 8 deletions(-)
> > +
> > +diff --git a/configure.ac b/configure.ac
> > +index 49caf96..4acc763 100644
> > +--- a/configure.ac
> > ++++ b/configure.ac
> > +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks" !=
> > "x"],[
> > + pcr_bank_checks
> > + AC_SUBST([DEFAULT_PCR_BANKS])
> > +
> > +-AC_PATH_PROG([EXPECT], expect)
> > +-if test "x$EXPECT" = "x"; then
> > +-	AC_MSG_ERROR([expect is required: expect package])
> > +-fi
> > ++# AC_PATH_PROG([EXPECT], expect)
> > ++# if test "x$EXPECT" = "x"; then
> > ++# 	AC_MSG_ERROR([expect is required: expect package])
> > ++# fi
> > +
> > + AC_PATH_PROG([GAWK], gawk)
> > + if test "x$GAWK" = "x"; then
> > + 	AC_MSG_ERROR([gawk is required: gawk package])
> > + fi
> > +
> > +-AC_PATH_PROG([SOCAT], socat)
> > +-if test "x$SOCAT" = "x"; then
> > +-	AC_MSG_ERROR([socat is required: socat package])
> > +-fi
> > ++# AC_PATH_PROG([SOCAT], socat)
> > ++# if test "x$SOCAT" = "x"; then
> > ++# 	AC_MSG_ERROR([socat is required: socat package])
> > ++# fi
> 
> 
> socat and expect are used for running the test cases. When not
> running 
> the tests then an alternative to commenting the above out may be to
> do 
> the following before running autogen.sh:
> 
> tmpdir=$(mktemp -d) || exit 1
> cp /usr/bin/true "${tmpdir}/socat"
> cp /usr/bin/true "${tmpdir}/expect"
> 
> PATH=$PATH:${tmpdir}
> ./autogen --prefix=...
> rm -rf "${tmpdir}"
> 
>      Stefan

... which would be more tedious than simply commenting them out
because then you would have to tweak the build itself with manual
steps instead of using autotools build command.

Up until this is fixed in the upstream project, we are better off
simply commenting them out, as they are test dependency, not a build
dependency for building e.g. initramfs image or some other type of
OS image.

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Fri, 2024-03-22 at 10:22 +0200, Jarkko Sakkinen wrote:
> On Thu, 2024-03-21 at 16:30 -0400, Stefan Berger wrote:
> > 
> > 
> > On 3/21/24 14:21, Jarkko Sakkinen wrote:
> > > Add swtpm and its dependency libtpms to host packages. These are
> > > useful
> > > for emulating TPM in QEMU environment.
> > > 
> > > Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> > > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> > > ---
> > 
> > > +---
> > > + configure.ac | 16 ++++++++--------
> > > + 1 file changed, 8 insertions(+), 8 deletions(-)
> > > +
> > > +diff --git a/configure.ac b/configure.ac
> > > +index 49caf96..4acc763 100644
> > > +--- a/configure.ac
> > > ++++ b/configure.ac
> > > +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks"
> > > !=
> > > "x"],[
> > > + pcr_bank_checks
> > > + AC_SUBST([DEFAULT_PCR_BANKS])
> > > +
> > > +-AC_PATH_PROG([EXPECT], expect)
> > > +-if test "x$EXPECT" = "x"; then
> > > +-	AC_MSG_ERROR([expect is required: expect package])
> > > +-fi
> > > ++# AC_PATH_PROG([EXPECT], expect)
> > > ++# if test "x$EXPECT" = "x"; then
> > > ++# 	AC_MSG_ERROR([expect is required: expect package])
> > > ++# fi
> > > +
> > > + AC_PATH_PROG([GAWK], gawk)
> > > + if test "x$GAWK" = "x"; then
> > > + 	AC_MSG_ERROR([gawk is required: gawk package])
> > > + fi
> > > +
> > > +-AC_PATH_PROG([SOCAT], socat)
> > > +-if test "x$SOCAT" = "x"; then
> > > +-	AC_MSG_ERROR([socat is required: socat package])
> > > +-fi
> > > ++# AC_PATH_PROG([SOCAT], socat)
> > > ++# if test "x$SOCAT" = "x"; then
> > > ++# 	AC_MSG_ERROR([socat is required: socat package])
> > > ++# fi
> > 
> > 
> > cat and expect are used for running the test cases. When not
> > running 
> > the tests then an alternative to commenting the above out may be to
> > do 
> > the following before running autogen.sh:
> > 
> > tmpdir=$(mktemp -d) || exit 1
> > cp /usr/bin/true "${tmpdir}/socat"
> > cp /usr/bin/true "${tmpdir}/expect"
> > 
> > PATH=$PATH:${tmpdir}
> > ./autogen --prefix=...
> > rm -rf "${tmpdir}"
> > 
> >      Stefan
> 
> ... which would be more tedious than simply commenting them out
> because then you would have to tweak the build itself with manual
> steps instead of using autotools build command.
> 
> Up until this is fixed in the upstream project, we are better off
> simply commenting them out, as they are test dependency, not a build
> dependency for building e.g. initramfs image or some other type of
> OS image.

And there's a lot of shenanigans related to the cross-compilation
environment, which render out the suggestion anyway in that form.

It's a bug because neither socat nor expect are actua build
dependencies, and the patch works around the bug.

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Fri Mar 22, 2024 at 10:35 AM EET, Jarkko Sakkinen wrote:
> On Fri, 2024-03-22 at 10:22 +0200, Jarkko Sakkinen wrote:
> > On Thu, 2024-03-21 at 16:30 -0400, Stefan Berger wrote:
> > > 
> > > 
> > > On 3/21/24 14:21, Jarkko Sakkinen wrote:
> > > > Add swtpm and its dependency libtpms to host packages. These are
> > > > useful
> > > > for emulating TPM in QEMU environment.
> > > > 
> > > > Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> > > > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> > > > ---
> > > 
> > > > +---
> > > > + configure.ac | 16 ++++++++--------
> > > > + 1 file changed, 8 insertions(+), 8 deletions(-)
> > > > +
> > > > +diff --git a/configure.ac b/configure.ac
> > > > +index 49caf96..4acc763 100644
> > > > +--- a/configure.ac
> > > > ++++ b/configure.ac
> > > > +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks"
> > > > !=
> > > > "x"],[
> > > > + pcr_bank_checks
> > > > + AC_SUBST([DEFAULT_PCR_BANKS])
> > > > +
> > > > +-AC_PATH_PROG([EXPECT], expect)
> > > > +-if test "x$EXPECT" = "x"; then
> > > > +-	AC_MSG_ERROR([expect is required: expect package])
> > > > +-fi
> > > > ++# AC_PATH_PROG([EXPECT], expect)
> > > > ++# if test "x$EXPECT" = "x"; then
> > > > ++# 	AC_MSG_ERROR([expect is required: expect package])
> > > > ++# fi
> > > > +
> > > > + AC_PATH_PROG([GAWK], gawk)
> > > > + if test "x$GAWK" = "x"; then
> > > > + 	AC_MSG_ERROR([gawk is required: gawk package])
> > > > + fi
> > > > +
> > > > +-AC_PATH_PROG([SOCAT], socat)
> > > > +-if test "x$SOCAT" = "x"; then
> > > > +-	AC_MSG_ERROR([socat is required: socat package])
> > > > +-fi
> > > > ++# AC_PATH_PROG([SOCAT], socat)
> > > > ++# if test "x$SOCAT" = "x"; then
> > > > ++# 	AC_MSG_ERROR([socat is required: socat package])
> > > > ++# fi
> > > 
> > > 
> > > cat and expect are used for running the test cases. When not
> > > running 
> > > the tests then an alternative to commenting the above out may be to
> > > do 
> > > the following before running autogen.sh:
> > > 
> > > tmpdir=$(mktemp -d) || exit 1
> > > cp /usr/bin/true "${tmpdir}/socat"
> > > cp /usr/bin/true "${tmpdir}/expect"
> > > 
> > > PATH=$PATH:${tmpdir}
> > > ./autogen --prefix=...
> > > rm -rf "${tmpdir}"
> > > 
> > >      Stefan
> > 
> > ... which would be more tedious than simply commenting them out
> > because then you would have to tweak the build itself with manual
> > steps instead of using autotools build command.
> > 
> > Up until this is fixed in the upstream project, we are better off
> > simply commenting them out, as they are test dependency, not a build
> > dependency for building e.g. initramfs image or some other type of
> > OS image.
>
> And there's a lot of shenanigans related to the cross-compilation
> environment, which render out the suggestion anyway in that form.
>
> It's a bug because neither socat nor expect are actua build
> dependencies, and the patch works around the bug.

I put out an issue:

https://github.com/stefanberger/swtpm/issues/843

It's not a huge problem for BuildRoot as the build version is always
fixed. We can update the version in BuildRoot and remove the patch
when the time comes.

This is a bigger issue than swtpm in autotools projects. Most of them
have test suites that are not compatible with cross-compilation but
instead assume implicitly that the machine which builds is also the
one that runs the tests, which does not hold when building full OS
images.

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Fri Mar 22, 2024 at 4:04 PM EET, Stefan Berger wrote:
>
>
> On 3/22/24 05:00, Jarkko Sakkinen wrote:
> > On Fri Mar 22, 2024 at 10:35 AM EET, Jarkko Sakkinen wrote:
> >> On Fri, 2024-03-22 at 10:22 +0200, Jarkko Sakkinen wrote:
> >>> On Thu, 2024-03-21 at 16:30 -0400, Stefan Berger wrote:
> >>>>
> >>>>
> >>>> On 3/21/24 14:21, Jarkko Sakkinen wrote:
> >>>>> Add swtpm and its dependency libtpms to host packages. These are
> >>>>> useful
> >>>>> for emulating TPM in QEMU environment.
> >>>>>
> >>>>> Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> >>>>> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> >>>>> ---
> >>>>
> >>>>> +---
> >>>>> + configure.ac | 16 ++++++++--------
> >>>>> + 1 file changed, 8 insertions(+), 8 deletions(-)
> >>>>> +
> >>>>> +diff --git a/configure.ac b/configure.ac
> >>>>> +index 49caf96..4acc763 100644
> >>>>> +--- a/configure.ac
> >>>>> ++++ b/configure.ac
> >>>>> +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks"
> >>>>> !=
> >>>>> "x"],[
> >>>>> + pcr_bank_checks
> >>>>> + AC_SUBST([DEFAULT_PCR_BANKS])
> >>>>> +
> >>>>> +-AC_PATH_PROG([EXPECT], expect)
> >>>>> +-if test "x$EXPECT" = "x"; then
> >>>>> +-	AC_MSG_ERROR([expect is required: expect package])
> >>>>> +-fi
> >>>>> ++# AC_PATH_PROG([EXPECT], expect)
> >>>>> ++# if test "x$EXPECT" = "x"; then
> >>>>> ++# 	AC_MSG_ERROR([expect is required: expect package])
> >>>>> ++# fi
> >>>>> +
> >>>>> + AC_PATH_PROG([GAWK], gawk)
> >>>>> + if test "x$GAWK" = "x"; then
> >>>>> + 	AC_MSG_ERROR([gawk is required: gawk package])
> >>>>> + fi
> >>>>> +
> >>>>> +-AC_PATH_PROG([SOCAT], socat)
> >>>>> +-if test "x$SOCAT" = "x"; then
> >>>>> +-	AC_MSG_ERROR([socat is required: socat package])
> >>>>> +-fi
> >>>>> ++# AC_PATH_PROG([SOCAT], socat)
> >>>>> ++# if test "x$SOCAT" = "x"; then
> >>>>> ++# 	AC_MSG_ERROR([socat is required: socat package])
> >>>>> ++# fi
> >>>>
> >>>>
> >>>> cat and expect are used for running the test cases. When not
> >>>> running
> >>>> the tests then an alternative to commenting the above out may be to
> >>>> do
> >>>> the following before running autogen.sh:
> >>>>
> >>>> tmpdir=$(mktemp -d) || exit 1
> >>>> cp /usr/bin/true "${tmpdir}/socat"
> >>>> cp /usr/bin/true "${tmpdir}/expect"
> >>>>
> >>>> PATH=$PATH:${tmpdir}
> >>>> ./autogen --prefix=...
> >>>> rm -rf "${tmpdir}"
> >>>>
> >>>>       Stefan
> >>>
> >>> ... which would be more tedious than simply commenting them out
> >>> because then you would have to tweak the build itself with manual
> >>> steps instead of using autotools build command.
> >>>
> >>> Up until this is fixed in the upstream project, we are better off
> >>> simply commenting them out, as they are test dependency, not a build
> >>> dependency for building e.g. initramfs image or some other type of
> >>> OS image.
> >>
> >> And there's a lot of shenanigans related to the cross-compilation
> >> environment, which render out the suggestion anyway in that form.
> >>
> >> It's a bug because neither socat nor expect are actua build
> >> dependencies, and the patch works around the bug.
> > 
> > I put out an issue:
> > 
> > https://github.com/stefanberger/swtpm/issues/843
> This PR should help resolve the issue:
>
> https://github.com/stefanberger/swtpm/pull/844

It does thanks!

As commented to the PR I was successfully able to run full TPM2
kselftest:

https://gitlab.com/jarkkojs/linux-tpmdd-test/-/commits/swtpm-fix

Possible to tag a version with the fix? Alternatively I can export
your patch from your tree and include it to the next revision of
this patch set.

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Fri Mar 22, 2024 at 5:48 PM EET, Stefan Berger wrote:
>
>
> On 3/22/24 11:11, Jarkko Sakkinen wrote:
> > On Fri Mar 22, 2024 at 4:04 PM EET, Stefan Berger wrote:
> >>
> >>
> >> On 3/22/24 05:00, Jarkko Sakkinen wrote:
> >>> On Fri Mar 22, 2024 at 10:35 AM EET, Jarkko Sakkinen wrote:
> >>>> On Fri, 2024-03-22 at 10:22 +0200, Jarkko Sakkinen wrote:
> >>>>> On Thu, 2024-03-21 at 16:30 -0400, Stefan Berger wrote:
> >>>>>>
> >>>>>>
> >>>>>> On 3/21/24 14:21, Jarkko Sakkinen wrote:
> >>>>>>> Add swtpm and its dependency libtpms to host packages. These are
> >>>>>>> useful
> >>>>>>> for emulating TPM in QEMU environment.
> >>>>>>>
> >>>>>>> Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> >>>>>>> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> >>>>>>> ---
> >>>>>>
> >>>>>>> +---
> >>>>>>> + configure.ac | 16 ++++++++--------
> >>>>>>> + 1 file changed, 8 insertions(+), 8 deletions(-)
> >>>>>>> +
> >>>>>>> +diff --git a/configure.ac b/configure.ac
> >>>>>>> +index 49caf96..4acc763 100644
> >>>>>>> +--- a/configure.ac
> >>>>>>> ++++ b/configure.ac
> >>>>>>> +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks"
> >>>>>>> !=
> >>>>>>> "x"],[
> >>>>>>> + pcr_bank_checks
> >>>>>>> + AC_SUBST([DEFAULT_PCR_BANKS])
> >>>>>>> +
> >>>>>>> +-AC_PATH_PROG([EXPECT], expect)
> >>>>>>> +-if test "x$EXPECT" = "x"; then
> >>>>>>> +-	AC_MSG_ERROR([expect is required: expect package])
> >>>>>>> +-fi
> >>>>>>> ++# AC_PATH_PROG([EXPECT], expect)
> >>>>>>> ++# if test "x$EXPECT" = "x"; then
> >>>>>>> ++# 	AC_MSG_ERROR([expect is required: expect package])
> >>>>>>> ++# fi
> >>>>>>> +
> >>>>>>> + AC_PATH_PROG([GAWK], gawk)
> >>>>>>> + if test "x$GAWK" = "x"; then
> >>>>>>> + 	AC_MSG_ERROR([gawk is required: gawk package])
> >>>>>>> + fi
> >>>>>>> +
> >>>>>>> +-AC_PATH_PROG([SOCAT], socat)
> >>>>>>> +-if test "x$SOCAT" = "x"; then
> >>>>>>> +-	AC_MSG_ERROR([socat is required: socat package])
> >>>>>>> +-fi
> >>>>>>> ++# AC_PATH_PROG([SOCAT], socat)
> >>>>>>> ++# if test "x$SOCAT" = "x"; then
> >>>>>>> ++# 	AC_MSG_ERROR([socat is required: socat package])
> >>>>>>> ++# fi
> >>>>>>
> >>>>>>
> >>>>>> cat and expect are used for running the test cases. When not
> >>>>>> running
> >>>>>> the tests then an alternative to commenting the above out may be to
> >>>>>> do
> >>>>>> the following before running autogen.sh:
> >>>>>>
> >>>>>> tmpdir=$(mktemp -d) || exit 1
> >>>>>> cp /usr/bin/true "${tmpdir}/socat"
> >>>>>> cp /usr/bin/true "${tmpdir}/expect"
> >>>>>>
> >>>>>> PATH=$PATH:${tmpdir}
> >>>>>> ./autogen --prefix=...
> >>>>>> rm -rf "${tmpdir}"
> >>>>>>
> >>>>>>        Stefan
> >>>>>
> >>>>> ... which would be more tedious than simply commenting them out
> >>>>> because then you would have to tweak the build itself with manual
> >>>>> steps instead of using autotools build command.
> >>>>>
> >>>>> Up until this is fixed in the upstream project, we are better off
> >>>>> simply commenting them out, as they are test dependency, not a build
> >>>>> dependency for building e.g. initramfs image or some other type of
> >>>>> OS image.
> >>>>
> >>>> And there's a lot of shenanigans related to the cross-compilation
> >>>> environment, which render out the suggestion anyway in that form.
> >>>>
> >>>> It's a bug because neither socat nor expect are actua build
> >>>> dependencies, and the patch works around the bug.
> >>>
> >>> I put out an issue:
> >>>
> >>> https://github.com/stefanberger/swtpm/issues/843
> >> This PR should help resolve the issue:
> >>
> >> https://github.com/stefanberger/swtpm/pull/844
> > 
> > It does thanks!
> > 
> > As commented to the PR I was successfully able to run full TPM2
> > kselftest:
> > 
> > https://gitlab.com/jarkkojs/linux-tpmdd-test/-/commits/swtpm-fix
> > 
> > Possible to tag a version with the fix? Alternatively I can export
>
> I'll create v0.8.2 with only this patch.

OK, cool, thanks!

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Sat Mar 23, 2024 at 4:51 PM EET, Stefan Berger wrote:
>
>
> On 3/22/24 12:46, Jarkko Sakkinen wrote:
> > On Fri Mar 22, 2024 at 5:48 PM EET, Stefan Berger wrote:
>
> >>
> >> I'll create v0.8.2 with only this patch.
> > 
> > OK, cool, thanks!
>
> I tagged it now.

I'll address the remarks from Arnout and bump the version, thanks.

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Arnout Vandecappelle via buildroot]

  Hi Jarkko.

On 21/03/2024 19:21, Jarkko Sakkinen wrote:
> Add swtpm and its dependency libtpms to host packages. These are useful
> for emulating TPM in QEMU environment.

  I don't understand... Does it mean that you run host-swtpm next to host-qemu 
and you somehow connect them so it gets exposed as a TPM2 device inside the qemu VM?

> 
> Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> ---
>   package/libtpms/Config.in                     |  9 ++++
>   package/libtpms/libtpms.hash                  |  1 +
>   package/libtpms/libtpms.mk                    | 15 ++++++

  Please split this in a separate patch for libtpms (so 3 patches in the series).

  You also need package/Config.in to source package/libtpms/Config.in

  Also, please run `make check-package`. There will undoubtedly be some coding 
style issues.

  Finally, please add yourself to the DEVELOPERS file for this package. This 
way, you'll get an e-mail if the package fails in the autobuilders, or when a 
new version is released if the package is registered on release-monitoring.org. 
See https://nightly.buildroot.org/#DEVELOPERS

>   .../0001-comment-out-expect-and-socat.patch   | 46 +++++++++++++++++++
>   package/swtpm/Config.host.in                  |  8 ++++

  You also need to add this to package/Config.in.host

>   package/swtpm/swtpm.hash                      |  1 +
>   package/swtpm/swtpm.mk                        | 17 +++++++
>   7 files changed, 97 insertions(+)
>   create mode 100644 package/libtpms/Config.in
>   create mode 100644 package/libtpms/libtpms.hash
>   create mode 100644 package/libtpms/libtpms.mk
>   create mode 100644 package/swtpm/0001-comment-out-expect-and-socat.patch
>   create mode 100644 package/swtpm/Config.host.in
>   create mode 100644 package/swtpm/swtpm.hash
>   create mode 100644 package/swtpm/swtpm.mk
> 
> diff --git a/package/libtpms/Config.in b/package/libtpms/Config.in
> new file mode 100644
> index 0000000000..7ef61cf53c
> --- /dev/null
> +++ b/package/libtpms/Config.in
> @@ -0,0 +1,9 @@
> +config BR2_PACKAGE_LIBTPMS
> +	bool "libtpms"
> +        depends on BR2_USE_WCHAR # glib2
> +        depends on BR2_TOOLCHAIN_HAS_THREADS # glib2
> +        depends on BR2_USE_MMU # glib2

  If you have those dependencies, I'd expect a corresponding `select 
BR2_PACKAGE_GLIB2`. However, there isn't any dependency at all in the .mk file, 
so I guess this is in fact not needed.

> +	help
> +	  TPM emulation library
> +
> +	  https://github.com/stefanberger/libtpms
> diff --git a/package/libtpms/libtpms.hash b/package/libtpms/libtpms.hash
> new file mode 100644
> index 0000000000..c31d824af6
> --- /dev/null
> +++ b/package/libtpms/libtpms.hash
> @@ -0,0 +1 @@
> +sha256  2807466f1563ebe45fdd12dd26e501e8a0c4fbb99c7c428fbb508789efd221c0  v0.9.6.tar.gz

  Please make sure that the license file is also in the .hash file. You can 
check this with `make legal-info`.

> diff --git a/package/libtpms/libtpms.mk b/package/libtpms/libtpms.mk
> new file mode 100644
> index 0000000000..5b1151baff
> --- /dev/null
> +++ b/package/libtpms/libtpms.mk
> @@ -0,0 +1,15 @@
> +################################################################################
> +#
> +# libtpms
> +#
> +################################################################################
> +
> +LIBTPMS_VERSION = v0.9.6

  Drop the v from the version, otherwise release-monitoring and CPE/CVE checks 
don't work. You can add the v below.

> +LIBTPMS_SOURCE = $(LIBTPMS_VERSION).tar.gz

  Don't override LIBTPMS_SOURCE, there's no need for that, the default 
(libtpms-0.9.6.tar.gz) is better. The github URL will still work. Note that the 
hash will change if you change the filename.

> +LIBTPMS_SITE = $(call github,stefanberger,libtpms,$(LIBTPMS_VERSION))

  This is where the v should be added:

LIBTPMS_SITE = $(call github,stefanberger,libtpms,v$(LIBTPMS_VERSION))

> +LIBTPMS_LICENSE = BSD-3-Clause

  It's actually BSD-4-Clause. And unfortunately, it also contains file which 
seem to be covered with a modified BSD-2-Clause instead, but let's ignore that :-)

  Please add the license file as well:

LIBTPMS_LICENSE_FILES = LICENSE

> +LIBTPMS_INSTALL_STAGING = YES
> +LIBTPMS_AUTORECONF = YES

  You should add a comment explaining why autoreconf is needed - in this case, 
because we get the source from git. It's also good to mention in the commit 
message that upstream doesn't create release tarballs that include the configure 
script.

> +
> +$(eval $(autotools-package))
> +$(eval $(host-autotools-package))
> diff --git a/package/swtpm/0001-comment-out-expect-and-socat.patch b/package/swtpm/0001-comment-out-expect-and-socat.patch
> new file mode 100644
> index 0000000000..09dcc49a7b
> --- /dev/null
> +++ b/package/swtpm/0001-comment-out-expect-and-socat.patch
> @@ -0,0 +1,46 @@
> +From 067c32ba93774b273de9af872b5587798dcabb15 Mon Sep 17 00:00:00 2001
> +From: Jarkko Sakkinen <jarkko@kernel.org>
> +Date: Tue, 19 Dec 2023 05:21:20 +0200
> +Subject: [PATCH] configure.ac: comment out "expect" and "socat"

  Please replace this with the patch from PR 844 (and add --disable-tests). Or 
wait until Stefan releases v0.8.2 (probably very soon).

> +
> +Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> +---
> + configure.ac | 16 ++++++++--------
> + 1 file changed, 8 insertions(+), 8 deletions(-)
> +
> +diff --git a/configure.ac b/configure.ac
> +index 49caf96..4acc763 100644
> +--- a/configure.ac
> ++++ b/configure.ac
> +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks" != "x"],[
> + pcr_bank_checks
> + AC_SUBST([DEFAULT_PCR_BANKS])
> +
> +-AC_PATH_PROG([EXPECT], expect)
> +-if test "x$EXPECT" = "x"; then
> +-	AC_MSG_ERROR([expect is required: expect package])
> +-fi
> ++# AC_PATH_PROG([EXPECT], expect)
> ++# if test "x$EXPECT" = "x"; then
> ++# 	AC_MSG_ERROR([expect is required: expect package])
> ++# fi
> +
> + AC_PATH_PROG([GAWK], gawk)
> + if test "x$GAWK" = "x"; then
> + 	AC_MSG_ERROR([gawk is required: gawk package])
> + fi
> +
> +-AC_PATH_PROG([SOCAT], socat)
> +-if test "x$SOCAT" = "x"; then
> +-	AC_MSG_ERROR([socat is required: socat package])
> +-fi
> ++# AC_PATH_PROG([SOCAT], socat)
> ++# if test "x$SOCAT" = "x"; then
> ++# 	AC_MSG_ERROR([socat is required: socat package])
> ++# fi
> +
> + AC_PATH_PROG([BASE64], base64)
> + if test "x$BASE64" = "x"; then
> +--
> +2.40.1
> +
> diff --git a/package/swtpm/Config.host.in b/package/swtpm/Config.host.in
> new file mode 100644
> index 0000000000..e77eea2aa5
> --- /dev/null
> +++ b/package/swtpm/Config.host.in
> @@ -0,0 +1,8 @@
> +config BR2_PACKAGE_HOST_SWTPM
> +	bool "swtpm-host"

  Should be "host swtpm"

> +	depends on BR2_PACKAGE_GOBJECT_INTROSPECTION_ARCH_SUPPORTS # gobject-introspection
> +	select BR2_PACKAGE_GOBJECT_INTROSPECTION

  This is selecting the _target_ gobject-introspection, which makes no sense for 
a host package.

> +	help
> +	  Compiles SWTPM software TPM emulator for the host.
> +
> +	  https://github.com/stefanberger/swtpm

  We want to point to something like documentation, which in this case is the 
wiki: https://github.com/stefanberger/swtpm/wiki

> diff --git a/package/swtpm/swtpm.hash b/package/swtpm/swtpm.hash
> new file mode 100644
> index 0000000000..882f06d7a5
> --- /dev/null
> +++ b/package/swtpm/swtpm.hash
> @@ -0,0 +1 @@
> +sha256 7bba52aa41090f75087034fac5fe8daed10c3e7e7234df7c9558849318927f41  v0.8.1.tar.gz
> diff --git a/package/swtpm/swtpm.mk b/package/swtpm/swtpm.mk
> new file mode 100644
> index 0000000000..79fbf1f420
> --- /dev/null
> +++ b/package/swtpm/swtpm.mk
> @@ -0,0 +1,17 @@
> +################################################################################
> +#
> +# swtpm
> +#
> +################################################################################
> +
> +SWTPM_VERSION = v0.8.1
> +SWTPM_SOURCE = $(SWTPM_VERSION).tar.gz
> +SWTPM_SITE = $(call github,stefanberger,swtpm,$(SWTPM_VERSION))
> +SWTPM_LICENSE = BSD-3-Clause
> +SWTPM_AUTORECONF = YES

  Same comments as for libtpms for the above 5 lines.

> +
> +HOST_SWTPM_DEPENDENCIES = host-libtasn1 host-openssl host-pkgconf host-json-glib host-libtpms

  Can you try inside a container (e.g. using utils/docker-run) if this is really 
sufficient?

> +HOST_SWTPM_CONF_ENV = PKG_CONFIG="$(PKG_CONFIG_HOST_BINARY)"

  This should already be part of the default configure environment, are you sure 
it is needed?

  Regards,
  Arnout

> +HOST_SWTPM_CONF_OPTS = --without-seccomp
> +
> +$(eval $(host-autotools-package))
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Fri Mar 22, 2024 at 10:47 PM EET, Arnout Vandecappelle wrote:
>   Hi Jarkko.
>
> On 21/03/2024 19:21, Jarkko Sakkinen wrote:
> > Add swtpm and its dependency libtpms to host packages. These are useful
> > for emulating TPM in QEMU environment.
>
>   I don't understand... Does it mean that you run host-swtpm next to host-qemu 
> and you somehow connect them so it gets exposed as a TPM2 device inside the qemu VM?

Yes:

https://gitlab.com/jarkkojs/linux-tpmdd-test/-/blob/main/board/qemu/run-qemu.sh.in?ref_type=heads

>
> > 
> > Link: https://gitlab.com/jarkkojs/linux-tpmdd-test
> > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> > ---
> >   package/libtpms/Config.in                     |  9 ++++
> >   package/libtpms/libtpms.hash                  |  1 +
> >   package/libtpms/libtpms.mk                    | 15 ++++++
>
>   Please split this in a separate patch for libtpms (so 3 patches in the series).
>
>   You also need package/Config.in to source package/libtpms/Config.in
>
>   Also, please run `make check-package`. There will undoubtedly be some coding 
> style issues.
>
>   Finally, please add yourself to the DEVELOPERS file for this package. This 
> way, you'll get an e-mail if the package fails in the autobuilders, or when a 
> new version is released if the package is registered on release-monitoring.org. 
> See https://nightly.buildroot.org/#DEVELOPERS

OK, got it. I'll follow the steps.

Yeah, these used to be lying in BR2_EXTERNAL and I was not exactly sure
how to proceed so I thought that better not to over-engineer.

>
> >   .../0001-comment-out-expect-and-socat.patch   | 46 +++++++++++++++++++
> >   package/swtpm/Config.host.in                  |  8 ++++
>
>   You also need to add this to package/Config.in.host

+1

> >   package/swtpm/swtpm.hash                      |  1 +
> >   package/swtpm/swtpm.mk                        | 17 +++++++
> >   7 files changed, 97 insertions(+)
> >   create mode 100644 package/libtpms/Config.in
> >   create mode 100644 package/libtpms/libtpms.hash
> >   create mode 100644 package/libtpms/libtpms.mk
> >   create mode 100644 package/swtpm/0001-comment-out-expect-and-socat.patch
> >   create mode 100644 package/swtpm/Config.host.in
> >   create mode 100644 package/swtpm/swtpm.hash
> >   create mode 100644 package/swtpm/swtpm.mk
> > 
> > diff --git a/package/libtpms/Config.in b/package/libtpms/Config.in
> > new file mode 100644
> > index 0000000000..7ef61cf53c
> > --- /dev/null
> > +++ b/package/libtpms/Config.in
> > @@ -0,0 +1,9 @@
> > +config BR2_PACKAGE_LIBTPMS
> > +	bool "libtpms"
> > +        depends on BR2_USE_WCHAR # glib2
> > +        depends on BR2_TOOLCHAIN_HAS_THREADS # glib2
> > +        depends on BR2_USE_MMU # glib2
>
>   If you have those dependencies, I'd expect a corresponding `select 
> BR2_PACKAGE_GLIB2`. However, there isn't any dependency at all in the .mk file, 
> so I guess this is in fact not needed.

+1

>
> > +	help
> > +	  TPM emulation library
> > +
> > +	  https://github.com/stefanberger/libtpms
> > diff --git a/package/libtpms/libtpms.hash b/package/libtpms/libtpms.hash
> > new file mode 100644
> > index 0000000000..c31d824af6
> > --- /dev/null
> > +++ b/package/libtpms/libtpms.hash
> > @@ -0,0 +1 @@
> > +sha256  2807466f1563ebe45fdd12dd26e501e8a0c4fbb99c7c428fbb508789efd221c0  v0.9.6.tar.gz
>
>   Please make sure that the license file is also in the .hash file. You can 
> check this with `make legal-info`.

+1

>
> > diff --git a/package/libtpms/libtpms.mk b/package/libtpms/libtpms.mk
> > new file mode 100644
> > index 0000000000..5b1151baff
> > --- /dev/null
> > +++ b/package/libtpms/libtpms.mk
> > @@ -0,0 +1,15 @@
> > +################################################################################
> > +#
> > +# libtpms
> > +#
> > +################################################################################
> > +
> > +LIBTPMS_VERSION = v0.9.6
>
>   Drop the v from the version, otherwise release-monitoring and CPE/CVE checks 
> don't work. You can add the v below.

+1

>
> > +LIBTPMS_SOURCE = $(LIBTPMS_VERSION).tar.gz
>
>   Don't override LIBTPMS_SOURCE, there's no need for that, the default 
> (libtpms-0.9.6.tar.gz) is better. The github URL will still work. Note that the 
> hash will change if you change the filename.

+1

>
> > +LIBTPMS_SITE = $(call github,stefanberger,libtpms,$(LIBTPMS_VERSION))
>
>   This is where the v should be added:
>
> LIBTPMS_SITE = $(call github,stefanberger,libtpms,v$(LIBTPMS_VERSION))

+1

>
> > +LIBTPMS_LICENSE = BSD-3-Clause
>
>   It's actually BSD-4-Clause. And unfortunately, it also contains file which 
> seem to be covered with a modified BSD-2-Clause instead, but let's ignore that :-)
>
>   Please add the license file as well:
>
> LIBTPMS_LICENSE_FILES = LICENSE

+1

>
> > +LIBTPMS_INSTALL_STAGING = YES
> > +LIBTPMS_AUTORECONF = YES
>
>   You should add a comment explaining why autoreconf is needed - in this case, 
> because we get the source from git. It's also good to mention in the commit 
> message that upstream doesn't create release tarballs that include the configure 
> script.

+1

>
> > +
> > +$(eval $(autotools-package))
> > +$(eval $(host-autotools-package))
> > diff --git a/package/swtpm/0001-comment-out-expect-and-socat.patch b/package/swtpm/0001-comment-out-expect-and-socat.patch
> > new file mode 100644
> > index 0000000000..09dcc49a7b
> > --- /dev/null
> > +++ b/package/swtpm/0001-comment-out-expect-and-socat.patch
> > @@ -0,0 +1,46 @@
> > +From 067c32ba93774b273de9af872b5587798dcabb15 Mon Sep 17 00:00:00 2001
> > +From: Jarkko Sakkinen <jarkko@kernel.org>
> > +Date: Tue, 19 Dec 2023 05:21:20 +0200
> > +Subject: [PATCH] configure.ac: comment out "expect" and "socat"
>
>   Please replace this with the patch from PR 844 (and add --disable-tests). Or 
> wait until Stefan releases v0.8.2 (probably very soon).

Yeah, this happened after I sent this (had a short discussion and I
tested and ack'd the fix).

Anyway:

+1

>
> > +
> > +Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
> > +---
> > + configure.ac | 16 ++++++++--------
> > + 1 file changed, 8 insertions(+), 8 deletions(-)
> > +
> > +diff --git a/configure.ac b/configure.ac
> > +index 49caf96..4acc763 100644
> > +--- a/configure.ac
> > ++++ b/configure.ac
> > +@@ -394,20 +394,20 @@ AS_IF([test "x$enable_default_pcr_banks" != "x"],[
> > + pcr_bank_checks
> > + AC_SUBST([DEFAULT_PCR_BANKS])
> > +
> > +-AC_PATH_PROG([EXPECT], expect)
> > +-if test "x$EXPECT" = "x"; then
> > +-	AC_MSG_ERROR([expect is required: expect package])
> > +-fi
> > ++# AC_PATH_PROG([EXPECT], expect)
> > ++# if test "x$EXPECT" = "x"; then
> > ++# 	AC_MSG_ERROR([expect is required: expect package])
> > ++# fi
> > +
> > + AC_PATH_PROG([GAWK], gawk)
> > + if test "x$GAWK" = "x"; then
> > + 	AC_MSG_ERROR([gawk is required: gawk package])
> > + fi
> > +
> > +-AC_PATH_PROG([SOCAT], socat)
> > +-if test "x$SOCAT" = "x"; then
> > +-	AC_MSG_ERROR([socat is required: socat package])
> > +-fi
> > ++# AC_PATH_PROG([SOCAT], socat)
> > ++# if test "x$SOCAT" = "x"; then
> > ++# 	AC_MSG_ERROR([socat is required: socat package])
> > ++# fi
> > +
> > + AC_PATH_PROG([BASE64], base64)
> > + if test "x$BASE64" = "x"; then
> > +--
> > +2.40.1
> > +
> > diff --git a/package/swtpm/Config.host.in b/package/swtpm/Config.host.in
> > new file mode 100644
> > index 0000000000..e77eea2aa5
> > --- /dev/null
> > +++ b/package/swtpm/Config.host.in
> > @@ -0,0 +1,8 @@
> > +config BR2_PACKAGE_HOST_SWTPM
> > +	bool "swtpm-host"
>
>   Should be "host swtpm"

+1

>
> > +	depends on BR2_PACKAGE_GOBJECT_INTROSPECTION_ARCH_SUPPORTS # gobject-introspection
> > +	select BR2_PACKAGE_GOBJECT_INTROSPECTION
>
>   This is selecting the _target_ gobject-introspection, which makes no sense for 
> a host package.

Hmm... do not want to say anything just yet because tbh cannot recall
why it is there but I'll look into this.

>
> > +	help
> > +	  Compiles SWTPM software TPM emulator for the host.
> > +
> > +	  https://github.com/stefanberger/swtpm
>
>   We want to point to something like documentation, which in this case is the 
> wiki: https://github.com/stefanberger/swtpm/wiki

+1

>
> > diff --git a/package/swtpm/swtpm.hash b/package/swtpm/swtpm.hash
> > new file mode 100644
> > index 0000000000..882f06d7a5
> > --- /dev/null
> > +++ b/package/swtpm/swtpm.hash
> > @@ -0,0 +1 @@
> > +sha256 7bba52aa41090f75087034fac5fe8daed10c3e7e7234df7c9558849318927f41  v0.8.1.tar.gz
> > diff --git a/package/swtpm/swtpm.mk b/package/swtpm/swtpm.mk
> > new file mode 100644
> > index 0000000000..79fbf1f420
> > --- /dev/null
> > +++ b/package/swtpm/swtpm.mk
> > @@ -0,0 +1,17 @@
> > +################################################################################
> > +#
> > +# swtpm
> > +#
> > +################################################################################
> > +
> > +SWTPM_VERSION = v0.8.1
> > +SWTPM_SOURCE = $(SWTPM_VERSION).tar.gz
> > +SWTPM_SITE = $(call github,stefanberger,swtpm,$(SWTPM_VERSION))
> > +SWTPM_LICENSE = BSD-3-Clause
> > +SWTPM_AUTORECONF = YES
>
>   Same comments as for libtpms for the above 5 lines.

+1

>
> > +
> > +HOST_SWTPM_DEPENDENCIES = host-libtasn1 host-openssl host-pkgconf host-json-glib host-libtpms
>
>   Can you try inside a container (e.g. using utils/docker-run) if this is really 
> sufficient?

Does it work with podman?

>
> > +HOST_SWTPM_CONF_ENV = PKG_CONFIG="$(PKG_CONFIG_HOST_BINARY)"
>
>   This should already be part of the default configure environment, are you sure 
> it is needed?

Tbh, no I'm not sure :-) I'll try to remove it and see what happens!

>
>   Regards,
>   Arnout
>
> > +HOST_SWTPM_CONF_OPTS = --without-seccomp
> > +
> > +$(eval $(host-autotools-package))

OK, thanks for the throughout and sane remarks!

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Peter Korsgaard]

>>>>> "Arnout" == Arnout Vandecappelle via buildroot <buildroot@buildroot.org> writes:

 >  Hi Jarkko.
 > On 21/03/2024 19:21, Jarkko Sakkinen wrote:
 >> Add swtpm and its dependency libtpms to host packages. These are useful
 >> for emulating TPM in QEMU environment.

 >  I don't understand... Does it mean that you run host-swtpm next to
 >  host-qemu and you somehow connect them so it gets exposed as a TPM2
 > device inside the qemu VM?

yes, swtpm listens on a unix socket and you can configure qemu to expose
that to the VM with the tpm-tis driver:

https://www.qemu.org/docs/master/specs/tpm.html

Quite handy for doing TPM stuff (E.G. CI) without having to mess with
real HW.

-- 
Bye, Peter Korsgaard
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] [PATCH RFC 2/2] package/swtpm: add host package

[Jarkko Sakkinen]

On Tue Mar 26, 2024 at 5:08 PM EET, Peter Korsgaard wrote:
> >>>>> "Arnout" == Arnout Vandecappelle via buildroot <buildroot@buildroot.org> writes:
>
>  >  Hi Jarkko.
>  > On 21/03/2024 19:21, Jarkko Sakkinen wrote:
>  >> Add swtpm and its dependency libtpms to host packages. These are useful
>  >> for emulating TPM in QEMU environment.
>
>  >  I don't understand... Does it mean that you run host-swtpm next to
>  >  host-qemu and you somehow connect them so it gets exposed as a TPM2
>  > device inside the qemu VM?
>
> yes, swtpm listens on a unix socket and you can configure qemu to expose
> that to the VM with the tpm-tis driver:
>
> https://www.qemu.org/docs/master/specs/tpm.html
>
> Quite handy for doing TPM stuff (E.G. CI) without having to mess with
> real HW.

Yeah, for instance Linux based operating systems have started to use
TPM2 chips for encrypted boot. And it is pretty generic passive smart
card like chip so you have bunch of ways to realize that. BuildRoot is
a great tool for prototyping such things.

And I don't see it totally unimaginable that someone would might want
to do similar e.g. in a home router some day.

BR, Jarkko
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot