[PATCH 1/1] build: restore missing libwpa_client.so

[Johannes Berg]

> Commit a41a29192e5d ("build: Pull common fragments into a build.rules
> file") introduced regression into wpa_supplicant build process. Build
> target libwpa_client.so is not built regardless of whether the option
> CONFIG_BUILD_WPA_CLIENT_SO is set or not. This happens because config
> option is used before it is imported from the configuration file.

Hmm, oops.

> Moving its usage after including build.rules does not help: variable
> ALL is processed by build.rules and further changes are not applied.

Right.

> This commit moves imports of hostapd/wpa_s configuration files from
> build.rules back into Makefiles.

That's kind of annoying. What do you think of this instead? It makes it
rely on the "_all" internal, but still feels somewhat better to me.

johannes


--- a/wpa_supplicant/Makefile
+++ b/wpa_supplicant/Makefile
@@ -10,15 +10,18 @@ ALL += systemd/wpa_supplicant at .service
 ALL += systemd/wpa_supplicant-nl80211 at .service
 ALL += systemd/wpa_supplicant-wired at .service
 ALL += dbus/fi.w1.wpa_supplicant1.service
-ifdef CONFIG_BUILD_WPA_CLIENT_SO
-ALL += libwpa_client.so
-endif
 
 EXTRA_TARGETS=dynamic_eap_methods
 
 CONFIG_FILE=.config
 include ../src/build.rules
 
+ifdef CONFIG_BUILD_WPA_CLIENT_SO
+# add the dependency this way to allow CONFIG_BUILD_WPA_CLIENT_SO
+# being set in the config which is read by build.rules
+_all: libwpa_client.so
+endif
+
 ifdef LIBS
 # If LIBS is set with some global build system defaults, clone those for
 # LIBS_c and LIBS_p to cover wpa_passphrase and wpa_cli as well.


johannes