Trouble building WireGuard since December releases

Trouble building WireGuard since December releases

[Kevin]

Hi all,

I've been having trouble building WireGuard on Solus since its December
snapshots. Earlier snapshots (e.g. November) still build without a
problem, even on new kernels. Does this have to do with Solus kernel
builds, or am I overlooking something?

user@t450s-solus ~/bin/WireGuard-0.0.20180118/src $ make debug
test -e include/generated/autoconf.h -a -e include/config/auto.conf ||
(	\
echo >&2;							\
echo >&2 "  ERROR: Kernel configuration is invalid.";		\
echo >&2 "         include/generated/autoconf.h or
include/config/auto.conf are missing.";\
echo >&2 "         Run 'make oldconfig && make prepare' on kernel src
to fix it.";	\
echo >&2 ;							\
/bin/false)
mkdir -p /home/user/bin/WireGuard-0.0.20180118/src/.tmp_versions ; rm
-f /home/user/bin/WireGuard-0.0.20180118/src/.tmp_versions/*
make -f ./scripts/Makefile.build obj=/home/user/bin/WireGuard-
0.0.20180118/src
make[2]: *** No rule to make target '/home/user/bin/WireGuard-
0.0.20180118/src/main.o', needed by '/home/user/bin/WireGuard-
0.0.20180118/src/wireguard.o'.  Stop.
make[1]: *** [Makefile:1507: _module_/home/user/bin/WireGuard-
0.0.20180118/src] Error 2
make: *** [Makefile:39: module-debug] Error 2

Re: Trouble building WireGuard since December releases

[Greg KH]

On Mon, Jan 22, 2018 at 03:42:25PM +0100, Kevin wrote:
> Hi all,
> 
> I've been having trouble building WireGuard on Solus since its December
> snapshots. Earlier snapshots (e.g. November) still build without a
> problem, even on new kernels. Does this have to do with Solus kernel
> builds, or am I overlooking something?
> 
> user@t450s-solus ~/bin/WireGuard-0.0.20180118/src $ make debug
> test -e include/generated/autoconf.h -a -e include/config/auto.conf ||
> (	\
> echo >&2;							\
> echo >&2 "  ERROR: Kernel configuration is invalid.";		\
> echo >&2 "         include/generated/autoconf.h or
> include/config/auto.conf are missing.";\
> echo >&2 "         Run 'make oldconfig && make prepare' on kernel src
> to fix it.";	\

Have you run 'make oldconfig' and 'make prepare' on the kernel source
tree you are wanting to build the wireguard code against?

thanks,

greg k-h

Re: Trouble building WireGuard since December releases

[Jason A. Donenfeld]

Strange. I'm getting Solus installed in a VM now, so I should have an
answer for you not before too long.

Re: Trouble building WireGuard since December releases

[Jason A. Donenfeld]

Worked exactly as intended for me.

1. I downloaded the latest ISO from the solus website.
2. I installed it using the GUI installer.
3. I booted into the new installation.
4. I updated all packages using the GUI updater.
5. I ran these commands in the built-in terminal:
$ sudo eopkg install git
$ sudo eopkg install -c kernel.devel
$ sudo eopkg install -c system.devel
$ sudo eopkg install libmnl-devel
$ git clone https://git.zx2c4.com/WireGuard
$ cd WireGuard/src && make debug -j$(nproc) && sudo make install
$ sudo modprobe wireguard && dmesg | grep wireguard
[   31.554056] wireguard: loading out-of-tree module taints kernel.
[   31.555061] wireguard: allowedips self-tests: pass
[   31.555463] wireguard: nonce counter self-tests: pass
[   31.555761] wireguard: curve25519 self-tests: pass
[   31.555798] wireguard: chacha20poly1305 self-tests: pass
[   31.555808] wireguard: poly1305 self-tests: pass
[   31.556129] wireguard: blake2s self-tests: pass
[   31.755518] wireguard: ratelimiter self-tests: pass
[   31.755531] wireguard: WireGuard 0.0.20180118-2-gc80cbfa loaded.
See www.wireguard.com for information.
[   31.755532] wireguard: Copyright (C) 2015-2018 Jason A. Donenfeld
<Jason@zx2c4.com>. All Rights Reserved.

Re: Trouble building WireGuard since December releases

[Jason A. Donenfeld]

Ahh, small update. When building this, I noticed:

Makefile:944: "Cannot use CONFIG_STACK_VALIDATION=y, please install
libelf-dev, libelf-devel or elfutils-libelf-devel"

After installing libelf-devel, the exact problem you reported
appeared. I then tried this hack:

$ sudo ln -s /bin/true /lib/modules/$(uname -r)/build/tools/objtool/objtool

And after, the build worked again. It therefore appears that Solus is
not distributing objtool with their kernel headers, when they should
be.

I opened a bug for your issue, but as I've never used Solus until
today, I won't be following up on it. So, please direct your
distribution to the bug at: https://dev.solus-project.com/T5607 and
see what you can do about helping to get this fixed.

Regards,
Jason

Re: Trouble building WireGuard since December releases

[Daniel Kahn Gillmor]

On Mon 2018-01-22 17:07:38 +0100, Jason A. Donenfeld wrote:
> Ahh, small update. When building this, I noticed:
>
> Makefile:944: "Cannot use CONFIG_STACK_VALIDATION=y, please install
> libelf-dev, libelf-devel or elfutils-libelf-devel"

fwiw, debian had the same issue around libelf-dev with their kernel
source packaging, and treated it as a bug there, which was resolved in
the kernel package itself, rather than requiring additional build
dependencies of all kernel modules: https://bugs.debian.org/886474

(this message is just about libelf-dev; i haven't followed the details
about objtool)

             --dkg

Re: Trouble building WireGuard since December releases

[Jason A. Donenfeld]

On Wed, Jan 24, 2018 at 7:49 PM, Daniel Kahn Gillmor
<dkg@fifthhorseman.net> wrote:
>
> On Mon 2018-01-22 17:07:38 +0100, Jason A. Donenfeld wrote:
> > Ahh, small update. When building this, I noticed:
> >
> > Makefile:944: "Cannot use CONFIG_STACK_VALIDATION=y, please install
> > libelf-dev, libelf-devel or elfutils-libelf-devel"
>
> fwiw, debian had the same issue around libelf-dev with their kernel
> source packaging, and treated it as a bug there, which was resolved in
> the kernel package itself, rather than requiring additional build
> dependencies of all kernel modules: https://bugs.debian.org/886474

Seems like a sane resolution.

>
> (this message is just about libelf-dev; i haven't followed the details
> about objtool)

The detail about objtool is just that they forgot to include it in
their headers, so when the build system tries to use it, it can't find
it and fails. Simple problem with a simple solution, thankfully.