Skip to contentSkip to conversation

glib: set glib-specific cross properties

PATCH In review Ross Burton 3 participants 5 messages Mar 26, 2019 – Mar 28, 2019 Arm/Amlogic Meson SoC support

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 glib: set glib-specific cross properties

GLib wants to execute code to identify several properties of the system, for example is vsprintf() broken and what direction does the stack grow. In cross-compiled environments these fallback to the default values which are not always correct (our vsprintf() is fine, but the test assumes it isn't) so set these properties explicitly.

Ideally this happens magically by meson.bbclass using the same logic as siteinfo.bbclass, but we need to fix glib now.

---
 meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross | 6 ++++++
 meta/recipes-core/glib-2.0/glib-2.0_2.58.3.bb        | 1 +
 meta/recipes-core/glib-2.0/glib.inc                  | 1 +
 3 files changed, 8 insertions(+)
 create mode 100644 meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross
meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross new file +6 -0
new file mode 100644
index 00000000000..8420f9874e1
@@ -0,0 +1,6 @@
+[properties]
+have_c99_vsnprintf = true
+have_c99_snprintf = true
+have_unix98_printf = true
+va_val_copy = true
+growing_stack = false
meta/recipes-core/glib-2.0/glib-2.0_2.58.3.bb +1 -0
index 2de2affa51a..c872c749202 100644
@@ -16,6 +16,7 @@ SRC_URI = "${GNOME_MIRROR}/glib/${SHRT_VER}/glib-${PV}.tar.xz \
            file://0001-Set-host_machine-correctly-when-building-with-mingw3.patch \
            file://0001-Do-not-write-bindir-into-pkg-config-files.patch \
            file://0001-meson.build-do-not-hardcode-linux-as-the-host-system.patch \
+           file://glib-meson.cross \
            "
 
 SRC_URI_append_class-native = " file://relocate-modules.patch"
meta/recipes-core/glib-2.0/glib.inc +1 -0
index 6f0ec1ad2e9..574bb38557b 100644
@@ -53,6 +53,7 @@ PACKAGECONFIG[libelf] = ",,elfutils"
 PACKAGECONFIG[tests] = "-Dinstalled_tests=true,-Dinstalled_tests=false,dbus"
 
 EXTRA_OEMESON = "-Ddtrace=false -Dfam=false -Dsystemtap=false -Dselinux=false"
+EXTRA_OEMESON_append_class-target = " --cross-file ${WORKDIR}/glib-meson.cross"
 
 do_configure_prepend() {
 	sed -i -e '1s,#!.*,#!${USRBINPATH}/env python3,' ${S}/gio/gdbus-2.0/codegen/gdbus-codegen.in
Signature · 1 line
2.11.0
#1 lore raw
Quoted 24 lines · Ross Burton

GLib wants to execute code to identify several properties of the system, for example is vsprintf() broken and what direction does the stack grow. In cross-compiled environments these fallback to the default values which are not always correct (our vsprintf() is fine, but the test assumes it isn't) so set these properties explicitly.

Ideally this happens magically by meson.bbclass using the same logic as siteinfo.bbclass, but we need to fix glib now.

---
 meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross | 6 ++++++
 meta/recipes-core/glib-2.0/glib-2.0_2.58.3.bb        | 1 +
 meta/recipes-core/glib-2.0/glib.inc                  | 1 +
 3 files changed, 8 insertions(+)
 create mode 100644 meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross
meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross new file +0 -0
new file mode 100644
index 00000000000..8420f9874e1

meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross line 6

Quoted 6 lines
meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross new file +6 -0
@@ -0,0 +1,6 @@
+[properties]
+have_c99_vsnprintf = true
+have_c99_snprintf = true
+have_unix98_printf = true
+va_val_copy = true
+growing_stack = false

this is mostly true for arches we support but maybe a comment to check it when doing explicit arch ports ( e.g. arm can grown both ways)

Quoted 32 lines
meta/recipes-core/glib-2.0/glib-2.0_2.58.3.bb +1 -0
index 2de2affa51a..c872c749202 100644
@@ -16,6 +16,7 @@ SRC_URI = "${GNOME_MIRROR}/glib/${SHRT_VER}/glib-${PV}.tar.xz \
            file://0001-Set-host_machine-correctly-when-building-with-mingw3.patch \
            file://0001-Do-not-write-bindir-into-pkg-config-files.patch \
            file://0001-meson.build-do-not-hardcode-linux-as-the-host-system.patch \
+           file://glib-meson.cross \
            "
 
 SRC_URI_append_class-native = " file://relocate-modules.patch"
meta/recipes-core/glib-2.0/glib.inc +1 -1
index 6f0ec1ad2e9..574bb38557b 100644
@@ -53,6 +53,7 @@ PACKAGECONFIG[libelf] = ",,elfutils"
 PACKAGECONFIG[tests] = "-Dinstalled_tests=true,-Dinstalled_tests=false,dbus"
 
 EXTRA_OEMESON = "-Ddtrace=false -Dfam=false -Dsystemtap=false -Dselinux=false"
+EXTRA_OEMESON_append_class-target = " --cross-file ${WORKDIR}/glib-meson.cross"
 
 do_configure_prepend() {
        sed -i -e '1s,#!.*,#!${USRBINPATH}/env python3,' ${S}/gio/gdbus-2.0/codegen/gdbus-codegen.in
--

2.11.0

--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

#2 lore raw
Quoted 19 lines · Khem Raj
Quoted 18 lines · Ross Burton

GLib wants to execute code to identify several properties of the system, for example is vsprintf() broken and what direction does the stack grow. In cross-compiled environments these fallback to the default values which are not always correct (our vsprintf() is fine, but the test assumes it isn't) so set these properties explicitly.

Ideally this happens magically by meson.bbclass using the same logic as siteinfo.bbclass, but we need to fix glib now.

---
 meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross | 6 ++++++
 meta/recipes-core/glib-2.0/glib-2.0_2.58.3.bb        | 1 +
 meta/recipes-core/glib-2.0/glib.inc                  | 1 +
 3 files changed, 8 insertions(+)
 create mode 100644 meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross
Quoted 15 lines

meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross line 6

Quoted 12 lines
meta/recipes-core/glib-2.0/glib-2.0/glib-meson.cross new file +6 -0
new file mode 100644
index 00000000000..8420f9874e1
@@ -0,0 +1,6 @@
+[properties]
+have_c99_vsnprintf = true
+have_c99_snprintf = true
+have_unix98_printf = true
+va_val_copy = true
+growing_stack = false

this is mostly true for arches we support but maybe a comment to check it when doing explicit arch ports ( e.g. arm can grown both ways)

Not on EABI or any other ARM ABI which could possibly be targeted by OE... unless you have evidence to prove otherwise?

Quoted 37 lines
Quoted 27 lines
meta/recipes-core/glib-2.0/glib-2.0_2.58.3.bb +1 -0
index 2de2affa51a..c872c749202 100644
@@ -16,6 +16,7 @@ SRC_URI = "${GNOME_MIRROR}/glib/${SHRT_VER}/glib-${PV}.tar.xz \
            file://0001-Set-host_machine-correctly-when-building-with-mingw3.patch \
            file://0001-Do-not-write-bindir-into-pkg-config-files.patch \
            file://0001-meson.build-do-not-hardcode-linux-as-the-host-system.patch \
+           file://glib-meson.cross \
            "
 
 SRC_URI_append_class-native = " file://relocate-modules.patch"
meta/recipes-core/glib-2.0/glib.inc +1 -1
index 6f0ec1ad2e9..574bb38557b 100644
@@ -53,6 +53,7 @@ PACKAGECONFIG[libelf] = ",,elfutils"
 PACKAGECONFIG[tests] = "-Dinstalled_tests=true,-Dinstalled_tests=false,dbus"
 
 EXTRA_OEMESON = "-Ddtrace=false -Dfam=false -Dsystemtap=false -Dselinux=false"
+EXTRA_OEMESON_append_class-target = " --cross-file ${WORKDIR}/glib-meson.cross"
 
 do_configure_prepend() {
        sed -i -e '1s,#!.*,#!${USRBINPATH}/env python3,' ${S}/gio/gdbus-2.0/codegen/gdbus-codegen.in
--

2.11.0

Quoted 5 lines

--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

#3 lore raw
Quoted 7 lines · Andre McCurdy
Quoted 4 lines
Quoted 1 line

+growing_stack = false

this is mostly true for arches we support but maybe a comment to check it when doing explicit arch ports ( e.g. arm can grown both ways)

Not on EABI or any other ARM ABI which could possibly be targeted by OE... unless you have evidence to prove otherwise?

Evidence, no. Anecdotally:

$ grep glib_cv_stack_grows *
arc-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
arm-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
arm-darwin:glib_cv_stack_grows=${glib_cv_stack_grows=no}
arm-darwin8:glib_cv_stack_grows=${glib_cv_stack_grows=no}
ix86-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
microblaze-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsel-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsel-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
nios2-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
powerpc-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
riscv32-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
riscv64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
sh-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
sparc-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
x86_64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}

Yes, this technique needs to be extended and generalised so it can be overridden on a per-platform basis, but for now it appears that everything we care about has the same values.

Ross

#4 lore raw
Quoted 43 lines · Ross
Quoted 4 lines · Andre McCurdy
Quoted 4 lines
Quoted 1 line

+growing_stack = false

this is mostly true for arches we support but maybe a comment to check it when doing explicit arch ports ( e.g. arm can grown both

ways)

Quoted 3 lines

Not on EABI or any other ARM ABI which could possibly be targeted by OE... unless you have evidence to prove otherwise?

Evidence, no. Anecdotally:

$ grep glib_cv_stack_grows *
arc-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
arm-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
arm-darwin:glib_cv_stack_grows=${glib_cv_stack_grows=no}
arm-darwin8:glib_cv_stack_grows=${glib_cv_stack_grows=no}
ix86-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
microblaze-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsel-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsel-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa32r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6el-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mipsisa64r6-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
mips-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
nios2-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
powerpc-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
riscv32-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
riscv64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
sh-common:glib_cv_stack_grows=${glib_cv_stack_grows=no}
sparc-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
x86_64-linux:glib_cv_stack_grows=${glib_cv_stack_grows=no}
Quoted 3 lines

Yes, this technique needs to be extended and generalised so it can be overridden on a per-platform basis, but for now it appears that everything we care about has the same values.

All I asked was to add a comment about it
So if someone is porting OE to new architecture then this will be a great hint to check for

Quoted 3 lines

Ross