On Sat, Aug 22, 2020 at 12:27 AM Hao Luo <haoluo@xxxxxxxxxx> wrote:
>
> On Fri, Aug 21, 2020 at 4:03 PM Andrii Nakryiko
> <andrii.nakryiko@xxxxxxxxx> wrote:
> >
> > On Thu, Aug 20, 2020 at 10:32 AM Yonghong Song <yhs@xxxxxx> wrote:
> > >
> > >
> > >
> > > On 8/19/20 3:40 PM, Hao Luo wrote:
> > > > Selftests for typed ksyms. Tests two types of ksyms: one is a struct,
> > > > the other is a plain int. This tests two paths in the kernel. Struct
> > > > ksyms will be converted into PTR_TO_BTF_ID by the verifier while int
> > > > typed ksyms will be converted into PTR_TO_MEM.
> > > >
> > > > Signed-off-by: Hao Luo <haoluo@xxxxxxxxxx>
> > > > ---
> > > > .../selftests/bpf/prog_tests/ksyms_btf.c | 77 +++++++++++++++++++
> > > > .../selftests/bpf/progs/test_ksyms_btf.c | 23 ++++++
> > > > 2 files changed, 100 insertions(+)
> > > > create mode 100644 tools/testing/selftests/bpf/prog_tests/ksyms_btf.c
> > > > create mode 100644 tools/testing/selftests/bpf/progs/test_ksyms_btf.c
> > > >
> > > > diff --git a/tools/testing/selftests/bpf/prog_tests/ksyms_btf.c b/tools/testing/selftests/bpf/prog_tests/ksyms_btf.c
> > > > new file mode 100644
> > > > index 000000000000..1dad61ba7e99
> > > > --- /dev/null
> > > > +++ b/tools/testing/selftests/bpf/prog_tests/ksyms_btf.c
> > > > @@ -0,0 +1,77 @@
> > > > +// SPDX-License-Identifier: GPL-2.0
> > > > +/* Copyright (c) 2020 Google */
> > > > +
> > > > +#include <test_progs.h>
> > > > +#include <bpf/libbpf.h>
> > > > +#include <bpf/btf.h>
> > > > +#include "test_ksyms_btf.skel.h"
> > > > +
> > > > +static int duration;
> > > > +
> > > > +static __u64 kallsyms_find(const char *sym)
> > > > +{
> > > > + char type, name[500];
> > > > + __u64 addr, res = 0;
> > > > + FILE *f;
> > > > +
> > > > + f = fopen("/proc/kallsyms", "r");
> > > > + if (CHECK(!f, "kallsyms_fopen", "failed to open: %d\n", errno))
> > > > + return 0;
> > >
> > > could you check whether libbpf API can provide this functionality for
> > > you? As far as I know, libbpf does parse /proc/kallsyms.
> >
> > No need to use libbpf's implementation. We already have
> > kallsyms_find() in prog_tests/ksyms.c and a combination of
> > load_kallsyms() + ksym_get_addr() in trace_helpers.c. It would be good
> > to switch to one implementation for both prog_tests/ksyms.c and this
> > one.
> >
> Ack. I can do some refactoring. The least thing that I can do is
> moving kallsyms_find() to a header for both prog_tests/ksyms.c and
> this test to use.
Please no extra headers. Just put kallsyms_find() in trace_helpers.c,
along other kallsyms-related functions.
>
> >
> > > > diff --git a/tools/testing/selftests/bpf/progs/test_ksyms_btf.c b/tools/testing/selftests/bpf/progs/test_ksyms_btf.c
> > > > new file mode 100644
> > > > index 000000000000..e04e31117f84
> > > > --- /dev/null
> > > > +++ b/tools/testing/selftests/bpf/progs/test_ksyms_btf.c
> > > > @@ -0,0 +1,23 @@
> > > > +// SPDX-License-Identifier: GPL-2.0
> > > > +/* Copyright (c) 2020 Google */
> > > > +
> > > > +#include "vmlinux.h"
> > > > +
> > > > +#include <bpf/bpf_helpers.h>
> > > > +
> > > > +__u64 out__runqueues = -1;
> > > > +__u64 out__bpf_prog_active = -1;
> > > > +
> > > > +extern const struct rq runqueues __ksym; /* struct type global var. */
> > > > +extern const int bpf_prog_active __ksym; /* int type global var. */
> > > > +
> > > > +SEC("raw_tp/sys_enter")
> > > > +int handler(const void *ctx)
> > > > +{
> > > > + out__runqueues = (__u64)&runqueues;
> > > > + out__bpf_prog_active = (__u64)&bpf_prog_active;
> > > > +
> >
> > You didn't test accessing any of the members of runqueues, because BTF
> > only has per-CPU variables, right? Adding global/static variables was
> > adding too much data to BTF or something like that, is that right?
> >
>
> Right. With some experiments, I found the address of a percpu variable
> doesn't necessarily point to a valid structure. So it doesn't make
> sense to dereference runqueues and access its members. However, right
> now there are only percpu variables encoded in BTF, so I can't test
> accessing members of general global/static variables unfortunately.
>
> Hao
