Bug 596504

Summary:	sys-apps/iproute2 with net-libs/libmnl-1.0.3-r1 built with LLVM/CLANG - bearer.c:(.text+0xfd): undefined reference to `mnl_attr_nest_start'
Product:	Gentoo Linux	Reporter:	Aleksander <spider.ch>
Component:	Current packages	Assignee:	Gentoo's Team for Core System packages <base-system>
Status:	RESOLVED FIXED
Severity:	normal	CC:	spider.ch, wizardedit
Priority:	Normal
Version:	unspecified
Hardware:	AMD64
OS:	Linux
See Also:	https://bugs.gentoo.org/show_bug.cgi?id=557874
Whiteboard:
Package list:		Runtime testing required:	---
Bug Depends on:
Bug Blocks:	408963
Attachments:	build.log emerge --info

Description Aleksander 2016-10-08 04:14:53 UTC

emerge --info: https://paste.pound-python.org/show/uNgqWnyfssvN1BUncdlJ/
build.log: https://paste.pound-python.org/show/7whMQoY02FhZ3diLLHiS/

Reproducible: Always

Steps to Reproduce:
1. Set clang as system compiler
2. Try to build iproute2
3. ...
4. Profit!

Comment 1 Aleksander 2016-10-08 04:20:59 UTC

Ooops.
Now it fails to build event with gcc with same error.

I setup llvm/clang as system compiler and try to rebuild system:
emerge -e system

Comment 2 Aleksander 2016-10-08 04:25:42 UTC

Problem that libmnl build with clang.
If i rebuild it with gcc, iproute2 builds fine even with llvm/clang

Comment 3 Aleksander 2016-10-08 04:28:42 UTC

Problem with stable libmnl from portage(1.0.3-r1)
If i unmask 1.0.4 everything builds fine with LLVM/CLANG

Sorry for all this flood.

Comment 4 Jeroen Roovers (RETIRED) gentoo-dev

2016-10-15 12:08:04 UTC

(In reply to Aleksander from comment #0)
> emerge --info: https://paste.pound-python.org/show/uNgqWnyfssvN1BUncdlJ/
> build.log: https://paste.pound-python.org/show/7whMQoY02FhZ3diLLHiS/

Attach those, please.

Comment 5 Aleksander 2016-10-16 02:19:07 UTC

Created attachment 450334 [details]
build.log

Comment 6 Aleksander 2016-10-16 02:19:24 UTC

Created attachment 450336 [details]
emerge --info

Comment 7 SpanKY gentoo-dev

2016-11-11 19:57:28 UTC

if it's fixed already with 1.0.4, then there's nothing to do here

for new experimental work, we rarely care about the state of the stable tree.  make sure things are working in unstable, and eventually that state converges into the stable tree.