package/faifa: fix incorrect library symlink

As spotted in
http://autobuild.buildroot.net/results/a61/a612cb7a85927d8cfe55c95c34d2901e7694fab0//diffoscope-results.txt,
faifa installs a library symlink with an incorrect target, which was
detected by the reproducible build logic, but is in fact wrong in any
case:

-lrwxrwxrwx   0        0        0        0 2019-11-07 19:38:04.000000 ./usr/lib/libfaifa.so -> /home/naourr/work/instance-3/output-1/target/usr/lib/libfaifa.so.0
+lrwxrwxrwx   0        0        0        0 2019-11-07 19:38:04.000000 ./usr/lib/libfaifa.so -> /home/naourr/work/instance-3/output-2/target/usr/lib/libfaifa.so.0

In practice, this is not a problem at runtime, as the .so symlink is
not used: the library soname is libfaifa.so.0. However, it still makes
sense to fix.

It is fixed by backporting an upstream commit. We considering bumping
to a newer version, but the latest version requires a new dependency
(libevent), so we preferred the backporting approach.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
 ...-fix-asbolute-symlink-of-libfaifa.so.patch | 32 +++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 package/faifa/0003-Makefile.in-fix-asbolute-symlink-of-libfaifa.so.patch

Thomas, All,

On 2019-11-18 22:10 +0100, Thomas Petazzoni spake thusly:
> As spotted in
> http://autobuild.buildroot.net/results/a61/a612cb7a85927d8cfe55c95c34d2901e7694fab0//diffoscope-results.txt,
> faifa installs a library symlink with an incorrect target, which was
> detected by the reproducible build logic, but is in fact wrong in any
> case:
> 
> -lrwxrwxrwx   0        0        0        0 2019-11-07 19:38:04.000000 ./usr/lib/libfaifa.so -> /home/naourr/work/instance-3/output-1/target/usr/lib/libfaifa.so.0
> +lrwxrwxrwx   0        0        0        0 2019-11-07 19:38:04.000000 ./usr/lib/libfaifa.so -> /home/naourr/work/instance-3/output-2/target/usr/lib/libfaifa.so.0
> 
> In practice, this is not a problem at runtime, as the .so symlink is
> not used: the library soname is libfaifa.so.0. However, it still makes
> sense to fix.
> 
> It is fixed by backporting an upstream commit. We considering bumping

*considered

Applied to master with this fixed, thanks!

Regards,
Yann E. MORIN.

> to a newer version, but the latest version requires a new dependency
> (libevent), so we preferred the backporting approach.
> 
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> ---
>  ...-fix-asbolute-symlink-of-libfaifa.so.patch | 32 +++++++++++++++++++
>  1 file changed, 32 insertions(+)
>  create mode 100644 package/faifa/0003-Makefile.in-fix-asbolute-symlink-of-libfaifa.so.patch
> 
> diff --git a/package/faifa/0003-Makefile.in-fix-asbolute-symlink-of-libfaifa.so.patch b/package/faifa/0003-Makefile.in-fix-asbolute-symlink-of-libfaifa.so.patch
> new file mode 100644
> index 0000000000..876d2a60f4
> --- /dev/null
> +++ b/package/faifa/0003-Makefile.in-fix-asbolute-symlink-of-libfaifa.so.patch
> @@ -0,0 +1,32 @@
> +From d16abf61049947c451a6411a13fc15dda32feb47 Mon Sep 17 00:00:00 2001
> +From: Florian Fainelli <f.fainelli@gmail.com>
> +Date: Sat, 6 Sep 2014 11:23:02 -0700
> +Subject: [PATCH] Makefile.in: fix asbolute symlink of libfaifa.so
> +
> +Fixes #7
> +
> +Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
> +[Upstream: da2e279fd736d05bcd0ee6e4609c44fece017ba8]
> +Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> +---
> + Makefile.in | 2 +-
> + 1 file changed, 1 insertion(+), 1 deletion(-)
> +
> +diff --git a/Makefile.in b/Makefile.in
> +index ba65bee..9dec7ee 100644
> +--- a/Makefile.in
> ++++ b/Makefile.in
> +@@ -98,9 +98,9 @@ install: installman strip
> + 	$(INSTALL) -m0755 hpav_cfg $(DESTDIR)$(sbindir)
> + 	$(INSTALL) -d $(DESTDIR)$(libdir)
> + 	$(INSTALL) -m0644 $(LIB_SONAME) $(DESTDIR)$(libdir)
> +-	ln -sf $(DESTDIR)$(libdir)/$(LIB_SONAME) $(DESTDIR)$(libdir)/$(LIB_SHARED_SO)
> + 	$(INSTALL) -d $(DESTDIR)$(includedir)/faifa
> + 	cp $(HEADERS) $(DESTDIR)$(includedir)/faifa
> ++	cd $(DESTDIR)$(libdir) && ln -sf $(LIB_SONAME) $(LIB_SHARED_SO)
> + 
> + strip:
> + 	$(STRIP) $(APP)
> +-- 
> +2.23.0
> +
> -- 
> 2.23.0
> 
> _______________________________________________
> buildroot mailing list
> buildroot@busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot

>>>>> "Thomas" == Thomas Petazzoni <thomas.petazzoni@bootlin.com> writes:

 > As spotted in
 > http://autobuild.buildroot.net/results/a61/a612cb7a85927d8cfe55c95c34d2901e7694fab0//diffoscope-results.txt,
 > faifa installs a library symlink with an incorrect target, which was
 > detected by the reproducible build logic, but is in fact wrong in any
 > case:

 > -lrwxrwxrwx 0 0 0 0 2019-11-07 19:38:04.000000 ./usr/lib/libfaifa.so
 > -> /home/naourr/work/instance-3/output-1/target/usr/lib/libfaifa.so.0
 > +lrwxrwxrwx 0 0 0 0 2019-11-07 19:38:04.000000 ./usr/lib/libfaifa.so
 > -> /home/naourr/work/instance-3/output-2/target/usr/lib/libfaifa.so.0

 > In practice, this is not a problem at runtime, as the .so symlink is
 > not used: the library soname is libfaifa.so.0. However, it still makes
 > sense to fix.

 > It is fixed by backporting an upstream commit. We considering bumping
 > to a newer version, but the latest version requires a new dependency
 > (libevent), so we preferred the backporting approach.

 > Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Committed to 2019.02.x and 2019.08.x, thanks.