support hooks and tables in virtual packages
cover support hooks and tables in virtual packages
From: Carlos Santos <[email protected]>
Virtual packages are not built but it's useful to allow them to have permission/device/user tables and target-finalize/rootfs-pre-cmd hooks.
With this change we don't need to replicate date as currently done in eudev and systemd user tables or test if the package is selected to add a target-finalize hook as in luainterpreter.
Carlos Santos (3):
package/pkg-generic.mk: support hooks and tables in virtual packages packages/luainterpreter: remove target-finalize hack packages/{eudev,systemd}: move commoun users to package/udev
package/eudev/eudev.mk | 7 ------- package/luainterpreter/luainterpreter.mk | 7 +------ package/pkg-generic.mk | 17 +++++++++++++++++ package/systemd/systemd.mk | 3 --- package/udev/udev.mk | 7 +++++++ 5 files changed, 25 insertions(+), 16 deletions(-)
Signature · 1 line
2.18.1
RFC PATCH 1/3 package/pkg-generic.mk: support hooks and tables in virtual packages
From: Carlos Santos <[email protected]>
Virtual packages are not built but it's useful to allow them to have permission/device/user tables and target-finalize/rootfs-pre-cmd hooks.
With this change we don't need to replicate date as currently done in eudev and systemd user tables or test if the package is selected to add a target-finalize hook as in luainterpreter.
--- package/pkg-generic.mk | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)
package/pkg-generic.mk +17 -0
@@ -930,6 +930,7 @@ $$($(2)_TARGET_DIRCLEAN): PKG=$(2) # Compute the name of the Kconfig option that correspond to the # package being enabled. We handle three cases: the special Linux # kernel case, the bootloaders case, and the normal packages case. +# Virtual packages are handled separately (see below). ifeq ($(1),linux) $(2)_KCONFIG_VAR = BR2_LINUX_KERNEL else ifneq ($$(filter boot/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/boot/%),$(pkgdir)),)
@@ -1118,6 +1119,22 @@ ifneq ($$($(2)_HELP_CMDS),) HELP_PACKAGES += $(2) endif +# Virtual packages are not built but it's useful to allow them to have +# permission/device/user tables and target-finalize/rootfs-pre-cmd hooks. +else ifeq ($$(BR2_PACKAGE_HAS_$(2)),y) # $(2)_KCONFIG_VAR + +ifneq ($$($(2)_PERMISSIONS),) +PACKAGES_PERMISSIONS_TABLE += $$($(2)_PERMISSIONS)$$(sep) +endif +ifneq ($$($(2)_DEVICES),) +PACKAGES_DEVICES_TABLE += $$($(2)_DEVICES)$$(sep) +endif +ifneq ($$($(2)_USERS),) +PACKAGES_USERS += $$($(2)_USERS)$$(sep) +endif +TARGET_FINALIZE_HOOKS += $$($(2)_TARGET_FINALIZE_HOOKS) +ROOTFS_PRE_CMD_HOOKS += $$($(2)_ROOTFS_PRE_CMD_HOOKS) + endif # $(2)_KCONFIG_VAR endef # inner-generic-package
Signature · 1 line
2.18.1
RFC PATCH 1/3 package/pkg-generic.mk: support hooks and tables in virtual packages
Carlos, All,
On 2019-10-30 10:27 -0300, unixmania at gmail.com spake thusly:
Quoted 4 lines
From: Carlos Santos <[email protected]>
Virtual packages are not built but it's useful to allow them to have permission/device/user tables and target-finalize/rootfs-pre-cmd hooks.
Agreed, that would be nice to deduplicate such things.
Quoted 1 line
With this change we don't need to replicate date as currently done in
What do you mean by "date"?
Quoted 21 lines
eudev and systemd user tables or test if the package is selected to add a target-finalize hook as in luainterpreter.
Signed-off-by Carlos Santos <[email protected]>--- package/pkg-generic.mk | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)package/pkg-generic.mk +1
-0@@ -930,6 +930,7 @@ $$($(2)_TARGET_DIRCLEAN): PKG=$(2) # Compute the name of the Kconfig option that correspond to the # package being enabled. We handle three cases: the special Linux # kernel case, the bootloaders case, and the normal packages case. +# Virtual packages are handled separately (see below). ifeq ($(1),linux) $(2)_KCONFIG_VAR = BR2_LINUX_KERNEL else ifneq ($$(filter boot/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/boot/%),$(pkgdir)),)
package/pkg-generic.mk ifneq ($$($(2)_HELP_CMDS),) · line 1124
Quoted 6 lines
package/pkg-generic.mk +3
-0@@ -1118,6 +1119,22 @@ ifneq ($$($(2)_HELP_CMDS),) HELP_PACKAGES += $(2) endif +# Virtual packages are not built but it's useful to allow them to have +# permission/device/user tables and target-finalize/rootfs-pre-cmd hooks. +else ifeq ($$(BR2_PACKAGE_HAS_$(2)),y) # $(2)_KCONFIG_VAR
Actually, I think that virtual packages should instead also be "nornal" packages, so that they use the standard code path. By "normal" package, I mean they should also have a BR2_PACKAGE_FOO option in addition to the BR2_PACKAGE_HAS_FOO option.
For example, we already have one virtual package that is both normal and
virtual: the skeleton package has both BR2_PACKAGE_HAS_SKELETON and BR2_PACKAGE_SKELETON.
pkg-virtual.mk is probably the place to make that systematic, something like:
BR2_PACKAGE_$(2) = $$(BR2_PACKAGE_HAS_$(2))
... so that virtual packages do not have to all repeat the variable...
Unless there is a big road-block that prevents virtual package to be treated as normal packages...
Regards,
Yann E. MORIN.
Quoted 22 lines
package/pkg-generic.mk +12
-0+ifneq ($$($(2)_PERMISSIONS),) +PACKAGES_PERMISSIONS_TABLE += $$($(2)_PERMISSIONS)$$(sep) +endif +ifneq ($$($(2)_DEVICES),) +PACKAGES_DEVICES_TABLE += $$($(2)_DEVICES)$$(sep) +endif +ifneq ($$($(2)_USERS),) +PACKAGES_USERS += $$($(2)_USERS)$$(sep) +endif +TARGET_FINALIZE_HOOKS += $$($(2)_TARGET_FINALIZE_HOOKS) +ROOTFS_PRE_CMD_HOOKS += $$($(2)_ROOTFS_PRE_CMD_HOOKS) + endif # $(2)_KCONFIG_VAR endef # inner-generic-packageSignature · 6 lines
2.18.1 _______________________________________________ buildroot mailing list buildroot at busybox.net http://lists.busybox.net/mailman/listinfo/buildroot
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
RFC PATCH 1/3 package/pkg-generic.mk: support hooks and tables in virtual packages
Quoted 6 lines · Yann E. MORIN
Carlos, All,
On 2019-10-30 10:27 -0300, unixmania at gmail.com spake thusly:
Quoted 2 lines
From: Carlos Santos <[email protected]>
Quoted 8 lines
Quoted 2 lines
Virtual packages are not built but it's useful to allow them to have permission/device/user tables and target-finalize/rootfs-pre-cmd hooks.
Agreed, that would be nice to deduplicate such things.
Quoted 1 line
With this change we don't need to replicate date as currently done in
What do you mean by "date"?
It means "typo", should be "data".
Quoted 24 lines
Quoted 24 lines
eudev and systemd user tables or test if the package is selected to add a target-finalize hook as in luainterpreter.
Signed-off-by Carlos Santos <[email protected]>--- package/pkg-generic.mk | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)package/pkg-generic.mk +1
-0@@ -930,6 +930,7 @@ $$($(2)_TARGET_DIRCLEAN): PKG=$(2) # Compute the name of the Kconfig option that correspond to the # package being enabled. We handle three cases: the special Linux # kernel case, the bootloaders case, and the normal packages case. +# Virtual packages are handled separately (see below). ifeq ($(1),linux) $(2)_KCONFIG_VAR = BR2_LINUX_KERNEL else ifneq ($$(filter boot/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/boot/%),$(pkgdir)),)@@ -1118,6 +1119,22 @@ ifneq ($$($(2)_HELP_CMDS),) HELP_PACKAGES += $(2) endif
Quoted 8 lines
package/pkg-generic.mk ifneq ($$($(2)_HELP_CMDS),) · line 1124
Quoted 3 lines
package/pkg-generic.mk +3
-0+# Virtual packages are not built but it's useful to allow them to have +# permission/device/user tables and target-finalize/rootfs-pre-cmd hooks. +else ifeq ($$(BR2_PACKAGE_HAS_$(2)),y) # $(2)_KCONFIG_VARActually, I think that virtual packages should instead also be "nornal" packages, so that they use the standard code path. By "normal" package, I mean they should also have a
BR2_PACKAGE_FOOoption in addition to theBR2_PACKAGE_HAS_FOOoption.
So we could use this, instead, instead of the else below :
# Compute the name of the Kconfig option that correspond to the
# package being enabled. We handle three cases: the special Linux # kernel case, the bootloaders case, and the normal packages case. # Virtual packages are not built but it's useful to allow them to have # permission/device/user tables and target-finalize/rootfs-pre-cmd hooks. ifeq ($(1),linux) $(2)_KCONFIG_VAR = BR2_LINUX_KERNEL else ifneq ($$(filter boot/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/boot/%),$(pkgdir)),) $(2)_KCONFIG_VAR = BR2_TARGET_$(2) else ifneq ($$(filter toolchain/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/toolchain/%),$(pkgdir)),) $(2)_KCONFIG_VAR = BR2_$(2) else ifeq ($(2)_IS_VIRTUAL,YES) $(2)_KCONFIG_VAR = BR2_PACKAGE_HAS_$(2) else $(2)_KCONFIG_VAR = BR2_PACKAGE_$(2) endif
Quoted 3 lines
For example, we already have one virtual package that is both normal and
virtual: the skeleton package has bothBR2_PACKAGE_HAS_SKELETONandBR2_PACKAGE_SKELETON.
Right, but BR2_PACKAGE_SKELETON does not seem to be used anywhere. Perhaps it should can be removed.
Quoted 9 lines
pkg-virtual.mk is probably the place to make that systematic, something like:
BR2_PACKAGE_$(2) = $$(BR2_PACKAGE_HAS_$(2))... so that virtual packages do not have to all repeat the variable...
Unless there is a big road-block that prevents virtual package to be treated as normal packages...
It creates a conflict with BR2_PACKAGE_UDEV in Config.in.legacy.
Quoted 33 lines
Regards,
Yann E. MORIN.Quoted 18 lines
package/pkg-generic.mk +12
-1+ifneq ($$($(2)_PERMISSIONS),) +PACKAGES_PERMISSIONS_TABLE += $$($(2)_PERMISSIONS)$$(sep) +endif +ifneq ($$($(2)_DEVICES),) +PACKAGES_DEVICES_TABLE += $$($(2)_DEVICES)$$(sep) +endif +ifneq ($$($(2)_USERS),) +PACKAGES_USERS += $$($(2)_USERS)$$(sep) +endif +TARGET_FINALIZE_HOOKS += $$($(2)_TARGET_FINALIZE_HOOKS) +ROOTFS_PRE_CMD_HOOKS += $$($(2)_ROOTFS_PRE_CMD_HOOKS) + endif # $(2)_KCONFIG_VAR endef # inner-generic-package --2.18.1
Quoted 4 lines
_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot--
.-----------------.--------------------.------------------.--------------------.| Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. |'------------------------------^-------^------------------^--------------------'
Signature · 1 line
Carlos Santos <[email protected]>
RFC PATCH 1/3 package/pkg-generic.mk: support hooks and tables in virtual packages
Quoted 65 lines · Carlos Santos
Quoted 6 lines · Yann E. MORIN
Carlos, All,
On 2019-10-30 10:27 -0300, unixmania at gmail.com spake thusly:
Quoted 2 lines
From: Carlos Santos <[email protected]>
Quoted 8 lines
Quoted 2 lines
Virtual packages are not built but it's useful to allow them to have permission/device/user tables and target-finalize/rootfs-pre-cmd hooks.
Agreed, that would be nice to deduplicate such things.
Quoted 1 line
With this change we don't need to replicate date as currently done in
What do you mean by "date"?
It means "typo", should be "data".
Quoted 24 lines
Quoted 24 lines
eudev and systemd user tables or test if the package is selected to add a target-finalize hook as in luainterpreter.
Signed-off-by Carlos Santos <[email protected]>--- package/pkg-generic.mk | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)package/pkg-generic.mk +1
-0@@ -930,6 +930,7 @@ $$($(2)_TARGET_DIRCLEAN): PKG=$(2) # Compute the name of the Kconfig option that correspond to the # package being enabled. We handle three cases: the special Linux # kernel case, the bootloaders case, and the normal packages case. +# Virtual packages are handled separately (see below). ifeq ($(1),linux) $(2)_KCONFIG_VAR = BR2_LINUX_KERNEL else ifneq ($$(filter boot/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/boot/%),$(pkgdir)),)@@ -1118,6 +1119,22 @@ ifneq ($$($(2)_HELP_CMDS),) HELP_PACKAGES += $(2) endifQuoted 8 lines
package/pkg-generic.mk ifneq ($$($(2)_HELP_CMDS),) · line 1124
Quoted 3 lines
package/pkg-generic.mk +3
-0+# Virtual packages are not built but it's useful to allow them to have +# permission/device/user tables and target-finalize/rootfs-pre-cmd hooks. +else ifeq ($$(BR2_PACKAGE_HAS_$(2)),y) # $(2)_KCONFIG_VARActually, I think that virtual packages should instead also be "nornal" packages, so that they use the standard code path. By "normal" package, I mean they should also have a
BR2_PACKAGE_FOOoption in addition to theBR2_PACKAGE_HAS_FOOoption.So we could use this, instead, instead of the else below :
# Compute the name of the
Kconfigoption that correspond to the# package being enabled. We handle three cases: the special Linux # kernel case, the bootloaders case, and the normal packages case. # Virtual packages are not built but it's useful to allow them to have # permission/device/user tables and target-finalize/rootfs-pre-cmd hooks. ifeq ($(1),linux) $(2)_KCONFIG_VAR = BR2_LINUX_KERNEL else ifneq ($$(filter boot/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/boot/%),$(pkgdir)),) $(2)_KCONFIG_VAR = BR2_TARGET_$(2)
Quoted 8 lines
else ifneq ($$(filter toolchain/% $$(foreach dir,$$(BR2_EXTERNAL_DIRS),$$(dir)/toolchain/%),$(pkgdir)),) $(2)_KCONFIG_VAR = BR2_$(2) else ifeq ($(2)_IS_VIRTUAL,YES) $(2)_KCONFIG_VAR = BR2_PACKAGE_HAS_$(2) else $(2)_KCONFIG_VAR = BR2_PACKAGE_$(2) endif
I attempted this solution but it adds virtual packages to the global package list unless I add something like
ifeq ($$($(2)_IS_VIRTUAL),)
PACKAGES += $(1)
endif
Also it does not seem to make sense to check FOO_PROVIDES and process reverse dependencies.
Signature · 1 line
Carlos Santos <[email protected]>
RFC PATCH 1/3 package/pkg-generic.mk: support hooks and tables in virtual packages
In reply to unixmania at gmail.comQuoted 10 lines · unixmania at gmail.com
From: Carlos Santos <[email protected]>
Virtual packages are not built but it's useful to allow them to have permission/device/user tables and target-finalize/rootfs-pre-cmd hooks.
With this change we don't need to replicate date as currently done in eudev and systemd user tables or test if the package is selected to add a target-finalize hook as in luainterpreter.
Signed-off-by Carlos Santos <[email protected]>
Since Yann's suggestion didn't work, I've applied the original (with typo fixes
in commit message) to next, thanks.
Regards,
Arnout
RFC PATCH 2/3 packages/luainterpreter: remove target-finalize hack
In reply to unixmania at gmail.comFrom: Carlos Santos <[email protected]>
Use LUAINTERPRETER_TARGET_FINALIZE_HOOKS, now that it's suported by virtual packages.
--- package/luainterpreter/luainterpreter.mk | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)
package/luainterpreter/luainterpreter.mk +1 -6
@@ -6,16 +6,11 @@ LUAINTERPRETER_ABIVER = $(call qstrip,$(BR2_PACKAGE_LUAINTERPRETER_ABI_VERSION)) -# Lua packages often install documentation, clean that up globally -# Since luainterpreter is a virtual package, we can't use -# LUAINTERPRETER_TARGET_FINALIZE_HOOKS -ifeq ($(BR2_PACKAGE_HAS_LUAINTERPRETER),y) define LUAINTERPRETER_REMOVE_DOC rm -rf $(TARGET_DIR)/usr/share/lua/$(LUAINTERPRETER_ABIVER)/doc endef -TARGET_FINALIZE_HOOKS += LUAINTERPRETER_REMOVE_DOC -endif +LUAINTERPRETER_TARGET_FINALIZE_HOOKS += LUAINTERPRETER_REMOVE_DOC $(eval $(virtual-package)) $(eval $(host-virtual-package))
Signature · 1 line
2.18.1
RFC PATCH 2/3 packages/luainterpreter: remove target-finalize hack
Quoted 15 lines · unixmania at gmail.com
From: Carlos Santos <[email protected]>
Use
LUAINTERPRETER_TARGET_FINALIZE_HOOKS, now that it's suported by virtual packages.Signed-off-by Carlos Santos <[email protected]>--- package/luainterpreter/luainterpreter.mk | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)package/luainterpreter/luainterpreter.mk +0
-0
package/luainterpreter/luainterpreter.mk line 9
Quoted 4 lines
package/luainterpreter/luainterpreter.mk +0
-1@@ -6,16 +6,11 @@ LUAINTERPRETER_ABIVER = $(call qstrip,$(BR2_PACKAGE_LUAINTERPRETER_ABI_VERSION)) -# Lua packages often install documentation, clean that up globally
This comment was still relevant, so I kept that and applied to next, thanks.
Regards,
Arnout
Quoted 14 lines
package/luainterpreter/luainterpreter.mk +1
-5-# Since luainterpreter is a virtual package, we can't use -# LUAINTERPRETER_TARGET_FINALIZE_HOOKS -ifeq ($(BR2_PACKAGE_HAS_LUAINTERPRETER),y) define LUAINTERPRETER_REMOVE_DOC rm -rf $(TARGET_DIR)/usr/share/lua/$(LUAINTERPRETER_ABIVER)/doc endef -TARGET_FINALIZE_HOOKS += LUAINTERPRETER_REMOVE_DOC -endif +LUAINTERPRETER_TARGET_FINALIZE_HOOKS += LUAINTERPRETER_REMOVE_DOC $(eval $(virtual-package)) $(eval $(host-virtual-package))
RFC PATCH 3/3 packages/{eudev, systemd}: move commoun users to package/udev
In reply to unixmania at gmail.comFrom: Carlos Santos <[email protected]>
Use UDEV_USERS, now that it's suported by virtual packages.
--- package/eudev/eudev.mk | 7 ------- package/systemd/systemd.mk | 3 --- package/udev/udev.mk | 7 +++++++ 3 files changed, 7 insertions(+), 10 deletions(-)
package/eudev/eudev.mk +0 -7
@@ -48,11 +48,4 @@ define EUDEV_INSTALL_INIT_SYSV $(INSTALL) -D -m 0755 package/eudev/S10udev $(TARGET_DIR)/etc/init.d/S10udev endef -# Required by default rules for input devices -define EUDEV_USERS - - - input -1 * - - - Input device group - - - render -1 * - - - DRI rendering nodes - - - kvm -1 * - - - kvm nodes -endef - $(eval $(autotools-package))
package/systemd/systemd.mk +0 -3
@@ -477,10 +477,7 @@ define SYSTEMD_INSTALL_IMAGES_CMDS endef define SYSTEMD_USERS - - - input -1 * - - - Input device group - - systemd-journal -1 * - - - Journal - - - render -1 * - - - DRI rendering nodes - - - kvm -1 * - - - kvm nodes systemd-bus-proxy -1 systemd-bus-proxy -1 * - - - Proxy D-Bus messages to/from a bus systemd-journal-gateway -1 systemd-journal-gateway -1 * /var/log/journal - - Journal Gateway systemd-journal-remote -1 systemd-journal-remote -1 * /var/log/journal/remote - - Journal Remote
package/udev/udev.mk +7 -0
@@ -4,4 +4,11 @@ # ################################################################################ +# Required by default rules for input devices +define UDEV_USERS + - - input -1 * - - - Input device group + - - render -1 * - - - DRI rendering nodes + - - kvm -1 * - - - kvm nodes +endef + $(eval $(virtual-package))
Signature · 1 line
2.18.1
Arnout Vandecappelle
AppliedRFC PATCH 3/3 packages/{eudev, systemd}: move commoun users to package/udev
Quoted 5 lines · unixmania at gmail.com
From: Carlos Santos <[email protected]>
Use
UDEV_USERS, now that it's suported by virtual packages.Signed-off-by Carlos Santos <[email protected]>
Applied to next, thanks.
Regards,
Arnout