mirror of https://github.com/dynup/kpatch
kmod: remove redundant -mcmodel=large on ppc64le
In PR #1205, Kamalesh reports: ... I see that the -mcmodel=large flag is being passed twice with KBUILD_CFLAGS_MODULE set: gcc -Wp,-MMD,/root/.kpatch/tmp/patch/.livepatch-meminfo.mod.o.d ............ -mcmodel=medium .... -I/root/kpatch/kmod/patch -mcmodel=large -fplugin=/root/kpatch/kpatch-build/gcc-plugins/ppc64le-plugin.so ... -DMODULE -mno-save-toc-indirect -mcmodel=large -mcmodel=large -DKBUILD_BASENAME='"livepatch_meminfo.mod"' -DKBUILD_MODNAME='"livepatch_meminfo"' -D__KBUILD_MODNAME=kmod_livepatch_meminfo -c -o /root/.kpatch/tmp/patch/livepatch-meminfo.mod.o /root/.kpatch/tmp/patch/livepatch-meminfo.mod.c. I loaded the module built without the KBUILD_CFLAGS_MODULE += -mcmodel=large flag and seems to okay. I guess, we can remove the arch specific flag from the Makefile. Suggested-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com> Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
This commit is contained in:
parent
624e5e3b82
commit
a02668a0f7
|
@ -2,14 +2,6 @@ KPATCH_BUILD ?= /lib/modules/$(shell uname -r)/build
|
|||
KPATCH_MAKE = $(MAKE) -C $(KPATCH_BUILD) M=$(PWD) CFLAGS_MODULE='$(CFLAGS_MODULE)'
|
||||
LDFLAGS += $(KPATCH_LDFLAGS)
|
||||
|
||||
# ppc64le kernel modules are expected to compile with the
|
||||
# -mcmodel=large flag. This enables 64-bit relocations
|
||||
# instead of a 32-bit offset from the TOC pointer.
|
||||
PROCESSOR = $(shell uname -m)
|
||||
ifeq ($(PROCESSOR), ppc64le)
|
||||
CFLAGS_MODULE += -mcmodel=large
|
||||
endif
|
||||
|
||||
obj-m += $(KPATCH_NAME).o
|
||||
ldflags-y += -T $(src)/kpatch.lds
|
||||
targets += kpatch.lds
|
||||
|
|
Loading…
Reference in New Issue