[PATCH 1/6] multilib-script: fix ALTERNATIVE_${PN} override issue

[PATCH 1/6] multilib-script: fix ALTERNATIVE_${PN} override issue

[Zhixiong Chi]

If multilib scripts are more than one file, the variable
ALTERNATIVE_${PN} will be overrode. Finally, there will be only
one symbol link file.

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
 meta/classes/multilib_script.bbclass | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta/classes/multilib_script.bbclass b/meta/classes/multilib_script.bbclass
index a5a552b716..51c9fcc12c 100644
--- a/meta/classes/multilib_script.bbclass
+++ b/meta/classes/multilib_script.bbclass
@@ -25,7 +25,7 @@ python () {
     for entry in (d.getVar("MULTILIB_SCRIPTS", False) or "").split():
        pkg, script = entry.split(":")
        scriptname = os.path.basename(script)
-       d.setVar("ALTERNATIVE_" + pkg, scriptname)
+       d.appendVar("ALTERNATIVE_" + pkg, scriptname + " ")
        d.setVarFlag("ALTERNATIVE_LINK_NAME", scriptname, script)
        d.setVarFlag("ALTERNATIVE_TARGET", scriptname, script + "-${MULTILIB_SUFFIX}")
        d.appendVar("multilibscript_rename",  "\n	mv ${PKGD}" + script + " ${PKGD}" + script + "-${MULTILIB_SUFFIX}")
-- 
2.17.1

[PATCH 2/6] libcheck: fix the multilib header conflict

[Zhixiong Chi]

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
 meta/recipes-support/libcheck/libcheck_0.12.0.bb | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/meta/recipes-support/libcheck/libcheck_0.12.0.bb b/meta/recipes-support/libcheck/libcheck_0.12.0.bb
index e646d43968..cb97201580 100644
--- a/meta/recipes-support/libcheck/libcheck_0.12.0.bb
+++ b/meta/recipes-support/libcheck/libcheck_0.12.0.bb
@@ -12,7 +12,7 @@ UPSTREAM_CHECK_URI = "https://github.com/libcheck/check/releases/"
 
 S = "${WORKDIR}/check-${PV}"
 
-inherit autotools pkgconfig texinfo
+inherit autotools pkgconfig texinfo multilib_header
 
 CACHED_CONFIGUREVARS += "ac_cv_path_AWK_PATH=${bindir}/gawk"
 
@@ -20,4 +20,8 @@ RREPLACES_${PN} = "check (<= 0.9.5)"
 RDEPENDS_${PN} += "gawk"
 RDEPENDS_${PN}_class-native = ""
 
+do_install_append() {
+        oe_multilib_header check_stdint.h
+}
+
 BBCLASSEXTEND = "native nativesdk"
-- 
2.17.1

[PATCH 3/6] nspr: fix script conflict for multilib

[Zhixiong Chi]

Stop the nspr-config scripts conflicting in a multilib case.

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
 meta/recipes-support/nspr/nspr_4.19.bb | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/meta/recipes-support/nspr/nspr_4.19.bb b/meta/recipes-support/nspr/nspr_4.19.bb
index b43c78081d..b5264be2a0 100644
--- a/meta/recipes-support/nspr/nspr_4.19.bb
+++ b/meta/recipes-support/nspr/nspr_4.19.bb
@@ -149,7 +149,9 @@ TESTS = " \
     xnotify \
     zerolen"
 
-inherit autotools
+inherit autotools multilib_script
+
+MULTILIB_SCRIPTS = "${PN}-dev:${bindir}/nspr-config"
 
 PACKAGECONFIG ??= "${@bb.utils.filter('DISTRO_FEATURES', 'ipv6', d)}"
 PACKAGECONFIG[ipv6] = "--enable-ipv6,--disable-ipv6,"
-- 
2.17.1

[PATCH 4/6] libevent: fix the multilib header conflict

[Zhixiong Chi]

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
 meta/recipes-support/libevent/libevent_2.1.8.bb | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/meta/recipes-support/libevent/libevent_2.1.8.bb b/meta/recipes-support/libevent/libevent_2.1.8.bb
index 83ce4644cf..0f91e05a6c 100644
--- a/meta/recipes-support/libevent/libevent_2.1.8.bb
+++ b/meta/recipes-support/libevent/libevent_2.1.8.bb
@@ -28,12 +28,16 @@ inherit autotools
 # Needed for Debian packaging
 LEAD_SONAME = "libevent-2.1.so"
 
-inherit ptest
+inherit ptest multilib_header
 
 DEPENDS = "zlib"
 
 BBCLASSEXTEND = "native nativesdk"
 
+do_install_append() {
+        oe_multilib_header event2/event-config.h
+}
+
 do_install_ptest() {
 	install -d ${D}${PTEST_PATH}/test
 	for file in ${B}/test/.libs/regress ${B}/test/.libs/test*
-- 
2.17.1

[PATCH 5/6] mobile-broadband-provider-info: fix the script multilib conflict

[Zhixiong Chi]

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
 .../mobile-broadband-provider-info_git.bb                     | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/meta/recipes-connectivity/mobile-broadband-provider-info/mobile-broadband-provider-info_git.bb b/meta/recipes-connectivity/mobile-broadband-provider-info/mobile-broadband-provider-info_git.bb
index dbc578e2d8..1a39932ffc 100644
--- a/meta/recipes-connectivity/mobile-broadband-provider-info/mobile-broadband-provider-info_git.bb
+++ b/meta/recipes-connectivity/mobile-broadband-provider-info/mobile-broadband-provider-info_git.bb
@@ -10,4 +10,6 @@ PE = "1"
 SRC_URI = "git://git.gnome.org/mobile-broadband-provider-info"
 S = "${WORKDIR}/git"
 
-inherit autotools
+inherit autotools multilib_script
+
+MULTILIB_SCRIPTS = "${PN}-dev:${datadir}/pkgconfig/mobile-broadband-provider-info.pc"
-- 
2.17.1

[PATCH 6/6] nss: fix multilib file conflict

[Zhixiong Chi]

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
---
 meta/recipes-support/nss/nss_3.38.bb | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/meta/recipes-support/nss/nss_3.38.bb b/meta/recipes-support/nss/nss_3.38.bb
index f3e5170a89..4f53b55f05 100644
--- a/meta/recipes-support/nss/nss_3.38.bb
+++ b/meta/recipes-support/nss/nss_3.38.bb
@@ -33,7 +33,9 @@ SRC_URI[sha256sum] = "2c643d3c08d6935f4d325f40743719b6990aa25a79ec2f8f712c99d086
 UPSTREAM_CHECK_URI = "https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/NSS_Releases"
 UPSTREAM_CHECK_REGEX = "NSS_(?P<pver>.+)_release_notes"
 
-inherit siteinfo
+inherit siteinfo multilib_script
+
+MULTILIB_SCRIPTS = "${PN}:${sysconfdir}/pki/nssdb/secmod.db ${PN}:${sysconfdir}/pki/nssdb/key3.db"
 
 DEPENDS = "sqlite3 nspr zlib nss-native"
 DEPENDS_class-native = "sqlite3-native nspr-native zlib-native"
-- 
2.17.1

Re: [PATCH 2/6] libcheck: fix the multilib header conflict

[Richard Purdie]

On Fri, 2018-08-10 at 00:31 -0700, Zhixiong Chi wrote:
> Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
> ---
>  meta/recipes-support/libcheck/libcheck_0.12.0.bb | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/recipes-support/libcheck/libcheck_0.12.0.bb
> b/meta/recipes-support/libcheck/libcheck_0.12.0.bb
> index e646d43968..cb97201580 100644
> --- a/meta/recipes-support/libcheck/libcheck_0.12.0.bb
> +++ b/meta/recipes-support/libcheck/libcheck_0.12.0.bb
> @@ -12,7 +12,7 @@ UPSTREAM_CHECK_URI = "https://github.com/libcheck/c
> heck/releases/"
>  
>  S = "${WORKDIR}/check-${PV}"
>  
> -inherit autotools pkgconfig texinfo
> +inherit autotools pkgconfig texinfo multilib_header
>  
>  CACHED_CONFIGUREVARS += "ac_cv_path_AWK_PATH=${bindir}/gawk"
>  
> @@ -20,4 +20,8 @@ RREPLACES_${PN} = "check (<= 0.9.5)"
>  RDEPENDS_${PN} += "gawk"
>  RDEPENDS_${PN}_class-native = ""
>  
> +do_install_append() {
> +        oe_multilib_header check_stdint.h
> +}
> +
>  BBCLASSEXTEND = "native nativesdk"

I'd like to understand the difference between the different headers, do
you have an example diff? 

Is there no way we can make them the same for the different multilibs?
oe_multilib_header should only be used when there is no other way...

The same applies to your other patches, an example of the differences
in the headers would be useful for review please.

Cheers,

Richard

Re: [PATCH 5/6] mobile-broadband-provider-info: fix the script multilib conflict

[Richard Purdie]

On Fri, 2018-08-10 at 00:31 -0700, Zhixiong Chi wrote:
> Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
> ---
>  .../mobile-broadband-provider-info_git.bb                     | 4
> +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/recipes-connectivity/mobile-broadband-provider-
> info/mobile-broadband-provider-info_git.bb b/meta/recipes-
> connectivity/mobile-broadband-provider-info/mobile-broadband-
> provider-info_git.bb
> index dbc578e2d8..1a39932ffc 100644
> --- a/meta/recipes-connectivity/mobile-broadband-provider-
> info/mobile-broadband-provider-info_git.bb
> +++ b/meta/recipes-connectivity/mobile-broadband-provider-
> info/mobile-broadband-provider-info_git.bb
> @@ -10,4 +10,6 @@ PE = "1"
>  SRC_URI = "git://git.gnome.org/mobile-broadband-provider-info"
>  S = "${WORKDIR}/git"
>  
> -inherit autotools
> +inherit autotools multilib_script
> +
> +MULTILIB_SCRIPTS = "${PN}-dev:${datadir}/pkgconfig/mobile-broadband-
> provider-info.pc"

This is not correct, pkgconfig files installed into datadir are meant
to be architecture independent. The ones in libdir can be target
specific. What is the difference causing problems here?

Cheers,

Richard

Re: [PATCH 6/6] nss: fix multilib file conflict

[Richard Purdie]

On Fri, 2018-08-10 at 00:31 -0700, Zhixiong Chi wrote:
> Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
> ---
>  meta/recipes-support/nss/nss_3.38.bb | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/recipes-support/nss/nss_3.38.bb b/meta/recipes-
> support/nss/nss_3.38.bb
> index f3e5170a89..4f53b55f05 100644
> --- a/meta/recipes-support/nss/nss_3.38.bb
> +++ b/meta/recipes-support/nss/nss_3.38.bb
> @@ -33,7 +33,9 @@ SRC_URI[sha256sum] = "2c643d3c08d6935f4d325f40743719b6990aa25a79ec2f8f712c99d086
>  UPSTREAM_CHECK_URI = "https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/NSS_Releases"
>  UPSTREAM_CHECK_REGEX = "NSS_(?P<pver>.+)_release_notes"
>  
> -inherit siteinfo
> +inherit siteinfo multilib_script
> +
> +MULTILIB_SCRIPTS = "${PN}:${sysconfdir}/pki/nssdb/secmod.db ${PN}:${sysconfdir}/pki/nssdb/key3.db"

Shouldn't those files be architecture independent given the files are
in sysconfdir? Isn't this really a determinism problem and we should
fix the builds to be deterministic instead?

Cheers,

Richard

Re: [PATCH 6/6] nss: fix multilib file conflict

[Alexander Kanavin]

2018-08-10 11:03 GMT+02:00 Richard Purdie <richard.purdie@linuxfoundation.org>:
> Shouldn't those files be architecture independent given the files are
> in sysconfdir? Isn't this really a determinism problem and we should
> fix the builds to be deterministic instead?

Yes, absolutely. This 'multilib_script' facility should be used only
when there's a genuine reason for the files to be different for
different multilib configs. Otherwise it does look like someone took a
shortcut without looking at the problem properly.

Alex