Message ID | 1458901049-3693-1-git-send-email-romain.izard.pro@gmail.com |
---|---|
State | Superseded |
Headers | show |
Romain, All, On 2016-03-25 11:17 +0100, Romain Izard spake thusly: > Using 'yes "" | make oldconfig' to regenerate a .config file from an > existing defconfig does not work reliably in all cases. Specifically, it > does not work well with tristate choice entries. > > The correct way to do it is to use 'make olddefconfig', but this target > is not supported in projects that use an old version of kconfig. As > most projects do not use tristate entries, there is no need for them to > upgrade. > > As a result, add support in the pkg-kconfig infrastructure for both > cases, with a configuration entry to select which mode to use. In fact, we've already discussed this, and we concluded that we would want an automatic behaviour. I.e. we try to use olddefconfig, and fallback to the yes-trick, like so: make olddefconfig || { yes "" |make oldconfig; } Care to adapt your patch and respin, please? Of course, your second patch will no longer make sense after that. In the meantime, I marked your patches as "Changes Requested" in patchwork. Thanks! ;-) Regards, Yann E. MORIN. > Signed-off-by: Romain Izard <romain.izard.pro@gmail.com> > --- > docs/manual/adding-packages-kconfig.txt | 5 +++++ > package/pkg-kconfig.mk | 10 ++++++---- > 2 files changed, 11 insertions(+), 4 deletions(-) > > diff --git a/docs/manual/adding-packages-kconfig.txt b/docs/manual/adding-packages-kconfig.txt > index 3290024..50c4ecb 100644 > --- a/docs/manual/adding-packages-kconfig.txt > +++ b/docs/manual/adding-packages-kconfig.txt > @@ -53,6 +53,11 @@ It is mandatory to provide either, but not both: > * +FOO_KCONFIG_FILE+ specifies the path to a defconfig or full-config file > to be used to configure the package. > > +* +FOO_KCONFIG_OLDDEFCONFIG+ when set specifies that the 'make' rule to > + call with defconfig files is 'olddefconfig'. Only some projects support > + it, but the default rule called 'oldconfig' can misbehave when the > + config file changes tristate 'choice' entries. > + > * +FOO_KCONFIG_DEFCONFIG+ specifies the defconfig 'make' rule to call to > configure the package. > > diff --git a/package/pkg-kconfig.mk b/package/pkg-kconfig.mk > index 90f3f9f..71c6865 100644 > --- a/package/pkg-kconfig.mk > +++ b/package/pkg-kconfig.mk > @@ -60,8 +60,7 @@ $$($(2)_KCONFIG_FILE) $$($(2)_KCONFIG_FRAGMENT_FILES): | $(1)-patch > # The specified source configuration file and any additional configuration file > # fragments are merged together to .config, after the package has been patched. > # Since the file could be a defconfig file it needs to be expanded to a > -# full .config first. We use 'make oldconfig' because this can be safely > -# done even when the package does not support defconfigs. > +# full .config first. > $$($(2)_DIR)/.config: $$($(2)_KCONFIG_FILE) $$($(2)_KCONFIG_FRAGMENT_FILES) > $$(if $$($(2)_KCONFIG_DEFCONFIG), \ > $$($(2)_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ > @@ -69,8 +68,11 @@ $$($(2)_DIR)/.config: $$($(2)_KCONFIG_FILE) $$($(2)_KCONFIG_FRAGMENT_FILES) > cp $$($(2)_KCONFIG_FILE) $$(@)) > support/kconfig/merge_config.sh -m -O $$(@D) \ > $$(@) $$($(2)_KCONFIG_FRAGMENT_FILES) > - $$(Q)yes "" | $$($(2)_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ > - $$($(2)_KCONFIG_OPTS) oldconfig > + $$(if $$($(2)_KCONFIG_OLDDEFCONFIG), \ > + $$($2_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ > + $$($(2)_KCONFIG_OPTS) olddefconfig, \ > + $$(Q)yes "" | $$($(2)_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ > + $$($(2)_KCONFIG_OPTS) oldconfig) > > # If _KCONFIG_FILE or _KCONFIG_FRAGMENT_FILES exists, this dependency is > # already implied, but if we only have a _KCONFIG_DEFCONFIG we have to add > -- > 2.5.0 > > _______________________________________________ > buildroot mailing list > buildroot@busybox.net > http://lists.busybox.net/mailman/listinfo/buildroot
Hello, On Sat, 26 Mar 2016 18:42:03 +0100, Yann E. MORIN wrote: > In fact, we've already discussed this, and we concluded that we would > want an automatic behaviour. > > I.e. we try to use olddefconfig, and fallback to the yes-trick, like so: > > make olddefconfig || { yes "" |make oldconfig; } > > Care to adapt your patch and respin, please? Of course, your second > patch will no longer make sense after that. > > In the meantime, I marked your patches as "Changes Requested" in > patchwork. And here is the reason why we want this automatic behavior rather than the explicit FOO_KCONFIG_OLDDEFCONFIG variable that you proposed Romain: the Linux, Barebox and U-Boot packages, which are some of the main users of kconfig-package infrastructure, have user-configurable versions. Therefore, we don't know if the user will select a recent enough version of the kernel, Barebox or U-Boot with olddefconfig support. For this reason, we cannot state in the Barebox, Linux or U-Boot packages whether they support olddefconfig or not, as it depends on the version. By having an automatic handling, regardless of the package and its version, we will use olddefconfig if available, and fallback to the old method if not. Thanks, Thomas
2016-03-26 18:52 GMT+01:00 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>: > Hello, > > On Sat, 26 Mar 2016 18:42:03 +0100, Yann E. MORIN wrote: > >> In fact, we've already discussed this, and we concluded that we would >> want an automatic behaviour. >> >> I.e. we try to use olddefconfig, and fallback to the yes-trick, like >> so: >> >> make olddefconfig || { yes "" |make oldconfig; } >> >> Care to adapt your patch and respin, please? Of course, your second >> patch will no longer make sense after that. >> >> In the meantime, I marked your patches as "Changes Requested" in >> patchwork. > > And here is the reason why we want this automatic behavior rather than > the explicit FOO_KCONFIG_OLDDEFCONFIG variable that you proposed > Romain: the Linux, Barebox and U-Boot packages, which are some of the > main users of kconfig-package infrastructure, have user-configurable > versions. Therefore, we don't know if the user will select a recent > enough version of the kernel, Barebox or U-Boot with olddefconfig > support. > > For this reason, we cannot state in the Barebox, Linux or U-Boot > packages whether they support olddefconfig or not, as it depends on > the version. By having an automatic handling, regardless of the > package and its version, we will use olddefconfig if available, and > fallback to the old method if not. If we want to handle not only the existing case, but also the historic cases, we have to use the 'oldnoconfig' target name, as the kernel target was only available under this name between versions 2.6.36 and 3.6. It is still supported, even if it is a misnomer. I will update the patch to handle this. Best regards,
diff --git a/docs/manual/adding-packages-kconfig.txt b/docs/manual/adding-packages-kconfig.txt index 3290024..50c4ecb 100644 --- a/docs/manual/adding-packages-kconfig.txt +++ b/docs/manual/adding-packages-kconfig.txt @@ -53,6 +53,11 @@ It is mandatory to provide either, but not both: * +FOO_KCONFIG_FILE+ specifies the path to a defconfig or full-config file to be used to configure the package. +* +FOO_KCONFIG_OLDDEFCONFIG+ when set specifies that the 'make' rule to + call with defconfig files is 'olddefconfig'. Only some projects support + it, but the default rule called 'oldconfig' can misbehave when the + config file changes tristate 'choice' entries. + * +FOO_KCONFIG_DEFCONFIG+ specifies the defconfig 'make' rule to call to configure the package. diff --git a/package/pkg-kconfig.mk b/package/pkg-kconfig.mk index 90f3f9f..71c6865 100644 --- a/package/pkg-kconfig.mk +++ b/package/pkg-kconfig.mk @@ -60,8 +60,7 @@ $$($(2)_KCONFIG_FILE) $$($(2)_KCONFIG_FRAGMENT_FILES): | $(1)-patch # The specified source configuration file and any additional configuration file # fragments are merged together to .config, after the package has been patched. # Since the file could be a defconfig file it needs to be expanded to a -# full .config first. We use 'make oldconfig' because this can be safely -# done even when the package does not support defconfigs. +# full .config first. $$($(2)_DIR)/.config: $$($(2)_KCONFIG_FILE) $$($(2)_KCONFIG_FRAGMENT_FILES) $$(if $$($(2)_KCONFIG_DEFCONFIG), \ $$($(2)_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ @@ -69,8 +68,11 @@ $$($(2)_DIR)/.config: $$($(2)_KCONFIG_FILE) $$($(2)_KCONFIG_FRAGMENT_FILES) cp $$($(2)_KCONFIG_FILE) $$(@)) support/kconfig/merge_config.sh -m -O $$(@D) \ $$(@) $$($(2)_KCONFIG_FRAGMENT_FILES) - $$(Q)yes "" | $$($(2)_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ - $$($(2)_KCONFIG_OPTS) oldconfig + $$(if $$($(2)_KCONFIG_OLDDEFCONFIG), \ + $$($2_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ + $$($(2)_KCONFIG_OPTS) olddefconfig, \ + $$(Q)yes "" | $$($(2)_MAKE_ENV) $$(MAKE) -C $$($(2)_DIR) \ + $$($(2)_KCONFIG_OPTS) oldconfig) # If _KCONFIG_FILE or _KCONFIG_FRAGMENT_FILES exists, this dependency is # already implied, but if we only have a _KCONFIG_DEFCONFIG we have to add
Using 'yes "" | make oldconfig' to regenerate a .config file from an existing defconfig does not work reliably in all cases. Specifically, it does not work well with tristate choice entries. The correct way to do it is to use 'make olddefconfig', but this target is not supported in projects that use an old version of kconfig. As most projects do not use tristate entries, there is no need for them to upgrade. As a result, add support in the pkg-kconfig infrastructure for both cases, with a configuration entry to select which mode to use. Signed-off-by: Romain Izard <romain.izard.pro@gmail.com> --- docs/manual/adding-packages-kconfig.txt | 5 +++++ package/pkg-kconfig.mk | 10 ++++++---- 2 files changed, 11 insertions(+), 4 deletions(-)