On Wed, 10 Jun 2020 at 13:24, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Wed, Jun 10, 2020 at 01:10:14PM +0100, Lorenz Bauer wrote:
> > On Wed, 10 Jun 2020 at 12:50, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > On Wed, Jun 10, 2020 at 11:16:16AM +0100, Lorenz Bauer wrote:
> > > > On Fri, 22 May 2020 at 01:09, Sasha Levin <sashal@xxxxxxxxxx> wrote:
> > > > >
> > > > > On Thu, May 21, 2020 at 03:48:41PM +0100, Lorenz Bauer wrote:
> > > > > >commit 634efb750435 ("selftests: bpf: Reset global state between
> > > > > >reuseport test runs") uses a macro RET_IF which doesn't exist in
> > > > > >the v4.19 tree. It is defined as follows:
> > > > > >
> > > > > > #define RET_IF(condition, tag, format...) ({
> > > > > > if (CHECK_FAIL(condition)) {
> > > > > > printf(tag " " format);
> > > > > > return;
> > > > > > }
> > > > > > })
> > > > > >
> > > > > >CHECK_FAIL in turn is defined as:
> > > > > >
> > > > > > #define CHECK_FAIL(condition) ({
> > > > > > int __ret = !!(condition);
> > > > > > int __save_errno = errno;
> > > > > > if (__ret) {
> > > > > > test__fail();
> > > > > > fprintf(stdout, "%s:FAIL:%d\n", __func__, __LINE__);
> > > > > > }
> > > > > > errno = __save_errno;
> > > > > > __ret;
> > > > > > })
> > > > > >
> > > > > >Replace occurences of RET_IF with CHECK. This will abort the test binary
> > > > > >if clearing the intermediate state fails.
> > > > > >
> > > > > >Fixes: 634efb750435 ("selftests: bpf: Reset global state between reuseport test runs")
> > > > > >Reported-by: kernel test robot <rong.a.chen@xxxxxxxxx>
> > > > > >Signed-off-by: Lorenz Bauer <lmb@xxxxxxxxxxxxxx>
> > > > >
> > > > > Thanks for the backport Lorenz. We'll need to wait for it to make it
> > > > > into Linus's tree before queueing up for the stable trees.
> > > >
> > > > Apologies for sending the patch too early (?), I'm still new to this process.
> > > > I've just hit this on 4.19.127. Do you want me to re-submit the patch somewhere?
> > >
> > > Is this patch in Linus's tree yet? If so, just tell us the git commit
> > > id. If not, it needs to go there first before we can take it to any
> > > stable tree.
> >
> > The patch isn't in Linus' tree because the problem doesn't exist
> > there. It fixes a build problem on
> > v4.19 which was introduced by the backport of an earlier fix of mine,
> > commit 634efb750435
> > ("selftests: bpf: Reset global state between reuseport test runs").
> >
> > There is a similar fix from Andrii Nakryiko that went into 5.4 as
> > commit aee43146cc10
> > ("selftest/bpf: fix backported test_select_reuseport selftest
> > changes"), which I hadn't seen
> > at the time.
>
> Ah, ok, that wasn't very obvious, sorry. I'll queue this up after the
> next round of kernels are released in a day or so...
No, it was my bad. What can I do to avoid this next time?
I've tried to follow the -stable FAQ, but that didn't work as you
can see ;)
Lorenz
--
Lorenz Bauer | Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK
www.cloudflare.com
