[OE-core] systemtap build fails

[Andreas Müller]

On Mon, Jun 4, 2012 at 4:20 PM, Tom Zanussi <tom.zanussi at intel.com> wrote:
> On Fri, 2012-06-01 at 01:18 +0200, Andreas Müller wrote:
>> Hi,
>>
>> I pulled today, started a build from scratch and ran into trouble with
>> systemtap:
>> ...
>> | In file included from csclient.h:12:0,
>> |                  from main.cxx:24:
>> | cscommon.h:8:17: fatal error: ssl.h: No such file or directory
>> | compilation terminated.
>> | In file included from csclient.h:12:0,
>> |                  from session.cxx:18:
>> | cscommon.h:8:17: fatal error: ssl.h: No such file or directory
>> | compilation terminated.
>> | make[2]: *** [stap-session.o] Error 1
>> | make[2]: *** Waiting for unfinished jobs....
>> | make[2]: *** [stap-main.o] Error 1
>> | make[2]: Leaving directory
>> `/home/Superandy/tmp/oe-core-eglibc/work/armv7a-angstrom-linux-gnueabi/systemtap-1.7+git1+83bd2699d8cff2f2d6b9eaf5ea254e4cb6b33e81-r4/git'
>> | make[1]: *** [all-recursive] Error 1
>> | make[1]: Leaving directory
>> `/home/Superandy/tmp/oe-core-eglibc/work/armv7a-angstrom-linux-gnueabi/systemtap-1.7+git1+83bd2699d8cff2f2d6b9eaf5ea254e4cb6b33e81-r4/git'
>> | make: *** [all] Error 2
>> | ERROR: oe_runmake failed
>> ...
>>
>> I added --without-nss to EXTRA_OECONF and rebuild. Now stap-main.o and
>> stap-session.o were build but now I get
>> ...
>> | modverify.c:28:21: fatal error: pk11pub.h: No such file or directory
>> | compilation terminated.
>> | make[3]: *** [staprun-modverify.o] Error 1
>> | make[3]: *** Waiting for unfinished jobs....
>> | ../../nsscommon.cxx:37:17: fatal error: ssl.h: No such file or directory
>> | compilation terminated.
>> | make[3]: *** [staprun-nsscommon.o] Error 1
>> | make[3]: Leaving directory
>> `/home/Superandy/tmp/oe-core-eglibc/work/armv7a-angstrom-linux-gnueabi/systemtap-1.7+git1+83bd2699d8cff2f2d6b9eaf5ea254e4cb6b33e81-r5/git/runtime/staprun'
>> | make[2]: *** [all] Error 2
>> | make[2]: Leaving directory
>> `/home/Superandy/tmp/oe-core-eglibc/work/armv7a-angstrom-linux-gnueabi/systemtap-1.7+git1+83bd2699d8cff2f2d6b9eaf5ea254e4cb6b33e81-r5/git/runtime/staprun'
>> | make[1]: *** [all-recursive] Error 1
>> | make[1]: Leaving directory
>> `/home/Superandy/tmp/oe-core-eglibc/work/armv7a-angstrom-linux-gnueabi/systemtap-1.7+git1+83bd2699d8cff2f2d6b9eaf5ea254e4cb6b33e81-r5/git'
>> | make: *** [all] Error 2
>> | ERROR: oe_runmake failed
>> ...
>>
>> In config.log I find
>> ...
>> HAVE_NSS_TRUE='#'
>> ...
>>
>> This is strange: In runtime/staprun/Makefile.am the failed files
>> should be excluded from build
>> ...
>> if HAVE_NSS
>> staprun_SOURCES += modverify.c ../../nsscommon.cxx
>> staprun_CFLAGS += $(nss_CFLAGS)
>> staprun_CXXFLAGS += $(nss_CFLAGS)
>> staprun_LDADD += $(nss_LIBS)
>> endif
>> ...
>>
>> Seems HAVE_NSS_TRUE='#' has not reached runtime/staprun
>>
>> Any thoughts?
>>
>
> Looks like systemtap changed some configuration that the recipe was
> relying on.  Could you try this patch and see if it fixes things?
>
> Thanks,
>
> Tom
>
> [PATCH] systemtap: disable some optional features
>
> Version 1.7 changed the way some of the configuration works,
> specifically, it removed some of the header checks for nss and avahi
> that the recipe was relying on and replaced them with a couple new
> config options.
>
> We'll also disable the server, which was the component making use of
> nss in the first place, and the grapher, which isn't core
> functionality either.
>
> Signed-off-by: Tom Zanussi <tom.zanussi at intel.com>
> ---
>  meta/recipes-kernel/systemtap/systemtap_git.bb |   10 +++-------
>  1 files changed, 3 insertions(+), 7 deletions(-)
>
> diff --git a/meta/recipes-kernel/systemtap/systemtap_git.bb b/meta/recipes-kernel/systemtap/systemtap_git.bb
> index 4d340fc..3b727b8 100644
> --- a/meta/recipes-kernel/systemtap/systemtap_git.bb
> +++ b/meta/recipes-kernel/systemtap/systemtap_git.bb
> @@ -6,19 +6,15 @@ DEPENDS = "elfutils sqlite3 systemtap-native"
>  DEPENDS_virtclass-native = "elfutils-native sqlite3-native gettext-native"
>  DEPENDS_virtclass-nativesdk = "elfutils-nativesdk sqlite3-nativesdk gettext-nativesdk"
>
> -PR = "r4"
> +PR = "r5"
>
>  export CC_FOR_BUILD = "${BUILD_CC}"
>  export CFLAGS_FOR_BUILD = "${BUILD_CFLAGS}"
>  export LDFLAGS_FOR_BUILD = "${BUILD_LDFLAGS}"
>
>  EXTRA_OECONF += "--with-libelf=${STAGING_DIR_TARGET} --without-rpm \
> -            ac_cv_file__usr_include_nss=no \
> -            ac_cv_file__usr_include_nss3=no \
> -            ac_cv_file__usr_include_nspr=no \
> -            ac_cv_file__usr_include_nspr4=no \
> -            ac_cv_file__usr_include_avahi_client=no \
> -            ac_cv_file__usr_include_avahi_common=no "
> +            --without-nss --without-avahi \
> +            --disable-server --disable-grapher "
>
>  STAP_DOCS ?= "--disable-docs --disable-publican --disable-refdocs"
>
> --
> 1.7.0.4
Thanks for taking care but since your patch does basically the same as
I written, the error situation remains the same in the second part of
first mail.

I see two solutions:

1. In runtime/staprun/Makefile.am remove
...
if HAVE_NSS
staprun_SOURCES += modverify.c ../../nsscommon.cxx
staprun_CFLAGS += $(nss_CFLAGS)
staprun_CXXFLAGS += $(nss_CFLAGS)
staprun_LDADD += $(nss_LIBS)
endif
...

2. Change some autotool magic so that
...
HAVE_NSS_TRUE='#'
...
from configure finds its way into runtime/staprun (see also
runtime/staprun/Makefile.in).

Since 1. is a hack, I would prefer 2. but did not yet find where
HAVE_NSS - properly detected by configure - gets lost...

Andreas