On 11.03.2014 10:41, Felix Fietkau wrote:
It may be better to send a patch to make the build fail early with a
descriptive error message.
You mean something like this? It is not pretty but appears to work.
Kernels with CONFIG_MODVERSIONS set are not supported by openwrt, because the
kernel is build after the modules. Instead of producing a running kernel with
unloadable modules, better fail immediately with an error message.
Signed-off-by: Nils Rennebarth <nils.renneba...@bintec-elmeg.com>
---
include/kernel-defaults.mk | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/include/kernel-defaults.mk b/include/kernel-defaults.mk
index 96f2b2c..114c017 100644
--- a/include/kernel-defaults.mk
+++ b/include/kernel-defaults.mk
@@ -106,6 +106,11 @@ define Kernel/Configure/Default
rm -rf $(KERNEL_BUILD_DIR)/modules
[ -d $(LINUX_DIR)/user_headers ] || $(MAKE) $(KERNEL_MAKEOPTS)
INSTALL_HDR_PATH=$(LINUX_DIR)/user_headers headers_install
$(SH_FUNC) grep '=[ym]' $(LINUX_DIR)/.config | LC_ALL=C sort | md5s >
$(LINUX_DIR)/.vermagic
+ if grep -q "^CONFIG_MODVERSIONS=y" $(LINUX_DIR)/.config; then \
+ printf "\nERROR: Module versioning not supported for openwrt
kernels\n" ; \
+ printf "Please change your kernel config so that CONFIG_MODVERSIONS
is not set\n\n" ; \
+ exit 1; \
+ fi
endef
define Kernel/Configure/Initramfs
--
1.8.1.1
Having said that, I agree that openwrt itself does not need CONFIG_MODVERSIONS.
However,
when developing kernel modules (hardware in particular), with the common
workflow:
do {
hack the driver sources inside a kernel checkout
make
transfer module to target, rmmod, insmod
} while (!module works)
CONFIG_MODVERSIONS is still useful imho, so I would still like to support it.
Can't
we simply remove vmlinux only once per openwrt make run, as part of the overall
preparation of the kernel tree, e.g. in the compile: target of BuildKernel
in kernel-build.mk, and then of course calling
make vmlinux modules
?
We could even encapsulate the whole thing into a new developer option
"support versioned modules in kernel", so that "normal" users still see the
error above. Has such a patch a chance to be accepted?
--
Nils Rennebarth
Software developer
bintec elmeg security GmbH
Mönchhaldenstraße 28
D-70191 Stuttgart
Germany
Tel: +49-711-900600-64
Email: nils.renneba...@bintec-elmeg.com
www: www.bintec-elmeg.com
Location: Nuremberg, Local Court Nuremberg, HRB 25481<br>
Managing Director: Bernd Büttner
--------------------------------
The information contained in this e-mail has been carefully researched,
but the possibility of it being inapplicable in individual cases cannot
be ruled out. We therefore regret that we cannot accept responsibility
or liability of any kind whatsoever for the correctness of the
information given. Please notify us if you discover that information is
inapplicable.
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
