Re: samples don't build on v4.18

[Joel Fernandes]

On Wed, Aug 15, 2018 at 12:34 PM, Jakub Kicinski
<jakub.kicinski@xxxxxxxxxxxxx> wrote:
> On Tue, 14 Aug 2018 20:01:32 -0700, Joel Fernandes wrote:
>> On Tue, Aug 14, 2018 at 06:22:21PM -0700, Joel Fernandes wrote:
>> > Forgot to add the patch author, doing so now. thanks
>> >
>> > On Tue, Aug 14, 2018 at 6:20 PM, Joel Fernandes <joelaf@xxxxxxxxxx> wrote:
>> > >
>> > > Hi,
>> > > When building BPF samples on v4.18, I get the following errors:
>> > >
>> > > $ cd samples/bpf/
>> > > $ make
>> > >
>> > > Auto-detecting system features:
>> > > ...                        libelf: [ OFF ]
>> > > ...                           bpf: [ OFF ]
>> > >
>> > > No libelf found
>> > > Makefile:213: recipe for target 'elfdep' failed
>> > > -----------
>> > >
>> > > I bissected it down to commit 5f9380572b4bb24f60cd492b1
>> > >
>> > > Author: Jakub Kicinski <jakub.kicinski@xxxxxxxxxxxxx>
>> > > Date:   Thu May 10 10:24:39 2018 -0700
>> > >
>> > >     samples: bpf: compile and link against full libbpf
>> > > ---------
>> > >
>> > > Checking out a kernel before this commit makes the samples build. Also I do
>> > > have libelf on my system.
>> > >
>> > > Any thoughts on this issue?
>>
>> There is some weirdness going on with my kernel tree. If I do a fresh clone
>> of v4.18 and build samples, everything works.
>>
>> However if I take my existing checkout, do a:
>> git clean -f -d
>> make mrproper
>>
>> Then I try to build the samples, I get the "No libelf found".
>>
>> Obviously the existing checked out kernel tree is in some weird state that I
>> am not yet able to fix. But atleast if I blow the whole tree and clone again,
>> I'm able to build...
>>
>> Is this related to the intermittent "No libelf found" issues that were
>> recently discussed?
>
> Can't reproduce, could you provide all exact commands you run to see
> this, including the initial clone?

Not sure if you saw that I replied to my own email. As I was saying,
doing a fresh clone and build of the kernel tree makes things work for
me. The problematic kernel tree which I cloned many months ago was the
one I was using when I reported the issue.

On the problematic tree, the steps I did to reproduce issue were:
git clean -f -d
make mrproper
make x86_64_defconfig
cd samples/bpf/
make

I have since moved onto using the fresh cloned tree since that's working for me.

I will let you know if I run into this again. Thanks for your time!

 - Joel