Re: perf test BPF failing on 4.15.0-rc6

From: Arnaldo Carvalho de Melo
Date: Wed Jan 03 2018 - 13:33:16 EST


Em Wed, Jan 03, 2018 at 03:27:01PM -0300, Arnaldo Carvalho de Melo escreveu:
> > Continuing investigation...
>
> After applying the fallback patch to allow new tools to work with older
> kernels:
>
> [root@felicio ~]# perf test bpf
> 39: BPF filter :
> 39.1: Basic BPF filtering : Ok
> 39.2: BPF pinning : Ok
> 39.3: BPF prologue generation : Ok
> 39.4: BPF relocation checker : Ok
> [root@felicio ~]# uname -a
> Linux felicio.ghostprotocols.net 4.13.0-rc7+ #1 SMP Mon Sep 11 13:56:18 -03 2017 x86_64 x86_64 x86_64 GNU/Linux
> [root@felicio ~]# rpm -q glibc
> glibc-2.17-157.el7_3.2.x86_64
> [root@felicio ~]#
>
> After applying the patch below I get to, which is what I am trying to
> fix now:
>
> [root@jouet ~]# perf test bpf
> 39: BPF filter :
> 39.1: Basic BPF filtering : Ok
> 39.2: BPF pinning : Ok
> 39.3: BPF prologue generation : FAILED!
> 39.4: BPF relocation checker : Skip
> [root@jouet ~]#

Update the patch to the one at the end of this message to make it work
with older glibcs, so that we ask for epoll_pwait() and hook into that
as well().

Now checking why 39.3 fails...

- Arnaldo

diff --git a/tools/perf/tests/bpf-script-example.c b/tools/perf/tests/bpf-script-example.c
index 268e5f8e4aa2..d1e67e271675 100644
--- a/tools/perf/tests/bpf-script-example.c
+++ b/tools/perf/tests/bpf-script-example.c
@@ -31,7 +31,7 @@ struct bpf_map_def SEC("maps") flip_table = {
.max_entries = 1,
};

-SEC("func=SyS_epoll_wait")
+SEC("func=SyS_epoll_pwait")
int bpf_func__SyS_epoll_wait(void *ctx)
{
int ind =0;
diff --git a/tools/perf/tests/bpf.c b/tools/perf/tests/bpf.c
index 0512f1b5bfdb..7c04e2d5b60b 100644
--- a/tools/perf/tests/bpf.c
+++ b/tools/perf/tests/bpf.c
@@ -25,7 +25,7 @@ static int epoll_wait_loop(void)

/* Should fail NR_ITERS times */
for (i = 0; i < NR_ITERS; i++)
- epoll_wait(-(i + 1), NULL, 0, 0);
+ epoll_pwait(-(i + 1), NULL, 0, 0, NULL);
return 0;
}