perf's builtin-sched.c triggers extremly long build times on some
architectures due to gcc 4.7+ var-tracking functionality.
To fix this, we can cherry pick the 3.12 commit:
f36f83f94 [perf sched: Move struct perf_sched definition out of cmd_sched()]
With this change build times are reduced from 15 to 20 minutes for qemuarm to:
real 2m19.940s
user 1m35.438s
sys 0m11.165s
For kernel's that are not carrying this patch, the following can be added
to the perf recipe to also fix the issue:
+++ b/meta/recipes-kernel/perf/perf.bb
@@ -68,6 +68,7 @@ EXTRA_OEMAKE = \
CC="${CC}" \
AR="${AR}" \
perfexecdir=${libexecdir} \
+ EXTRA_CFLAGS="-fno-var-tracking" \
NO_GTK2=1 ${TUI_DEFINES} NO_DWARF=1 ${SCRIPTING_DEFINES} \
(From OE-Core rev: 82ad5305381c2f541ef051a8fc28243cd91776fe)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
4ed858bc76