Re: [PATCH] perf tests: Add install prefix tests

From: Josh Boyer
Date: Wed Jan 22 2014 - 09:46:42 EST


On Wed, Jan 22, 2014 at 9:44 AM, Jiri Olsa <jolsa@xxxxxxxxxx> wrote:
> On Wed, Jan 22, 2014 at 09:03:21AM -0500, Josh Boyer wrote:
>> On Wed, Jan 22, 2014 at 8:02 AM, Jiri Olsa <jolsa@xxxxxxxxxx> wrote:
>> > On Tue, Jan 21, 2014 at 07:58:41PM -0500, Josh Boyer wrote:
>> >> On Tue, Jan 21, 2014 at 3:02 PM, Josh Boyer <jwboyer@xxxxxxxxxxxxxxxxx> wrote:
>> >> > Hi All,
>> >> >
>> >> > I went to build Linux v3.13-737-g7fe67a1 in Fedora this morning and it
>> >> > resulted in RPM complaining that the perf and perf.so binaries had
>> >>
>> >
>> > SNIP
>> >
>> >> I'm not sure why DESTDIR is included in the -DPLUGIN_DIR definition in
>> >> tools/lib/traceevent/Makefile, nor why $(prefix) is in plugindir_SQ in
>> >> tools/perf/config/Makefile.
>> >>
>> >> josh
>> >>
>> >> diff --git a/tools/lib/traceevent/Makefile b/tools/lib/traceevent/Makefile
>> >> index 56d52a3..005c9cc 100644
>> >> --- a/tools/lib/traceevent/Makefile
>> >> +++ b/tools/lib/traceevent/Makefile
>> >> @@ -63,7 +63,7 @@ endif
>> >> endif
>> >>
>> >> ifeq ($(set_plugin_dir),1)
>> >> -PLUGIN_DIR = -DPLUGIN_DIR="$(DESTDIR)/$(plugin_dir)"
>> >> +PLUGIN_DIR = -DPLUGIN_DIR="$(plugin_dir)"
>> >> PLUGIN_DIR_SQ = '$(subst ','\'',$(PLUGIN_DIR))'
>> >
>> > hi,
>> > this one needs to stay, otherwise perf have wrong path
>> > to load plugins from..
>>
>> Hm. Are you sure? I believe I tested with just the $(prefix) removed
>> below and it still failed the RPM build. If DESTDIR is included in
>> what is passed as DPLUGIN_DIR, it will include the buildroot path in
>> the binaries, but when the RPM is installed it does not install to the
>> buildroot path. E.g. you will have:
>>
>> -DPLUGIN_DIR="/home/jwboyer/rpmbuild/BUILDROOT/kernel-3.14.0-0.rc0.git1.1.fc21.x86_64//usr/lib64/traceevent/plugins"
>>
>> but the plugins will be installed to /usr/lib64/traceevent/plugins by
>> RPM. I'll try again with just the libdir fix, but I'm fairly sure
>> what I've said above is correct. At least in terms of how RPMs are
>> built.
>
> ok, now I'm woken up... you're right, both hunks are needed,
> got confused with the DESTDIR real meaning, sry

OK, good. FWIW, I did just retest with the DESTDIR hunk removed and
it does fail as I said above. I'll come up with an actual patch with
a S-o-B and send it out shortly.

Thanks!

josh
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/