by Masami Hiramatsu

[permalink] [raw]

Subject: Re: [PATCH v10 12/12] fprobe: Add a selftest for fprobe

On Fri, 11 Mar 2022 10:35:30 -0800
Andrii Nakryiko <[email protected]> wrote:

> On Thu, Mar 10, 2022 at 7:11 AM Masami Hiramatsu <[email protected]> wrote:
> >
> > On Wed, 9 Mar 2022 16:40:00 -0800
> > Andrii Nakryiko <[email protected]> wrote:
> >
> > > On Wed, Mar 9, 2022 at 4:17 PM Masami Hiramatsu <[email protected]> wrote:
> > > >
> > > > Hi,
> > > >
> > > > On Tue, 8 Mar 2022 20:10:48 +0900
> > > > Masami Hiramatsu <[email protected]> wrote:
> > > >
> > > > > Add a KUnit based selftest for fprobe interface.
> > > > >
> > > > > Signed-off-by: Masami Hiramatsu <[email protected]>
> > > > > ---
> > > > > Changes in v9:
> > > > > - Rename fprobe_target* to fprobe_selftest_target*.
> > > > > - Find the correct expected ip by ftrace_location_range().
> > > > > - Since the ftrace_location_range() is not exposed to module, make
> > > > > this test only for embedded.
> > > > > - Add entry only test.
> > > > > - Reset the fprobe structure before reuse it.
> > > > > ---
> > > > > lib/Kconfig.debug | 12 ++++
> > > > > lib/Makefile | 2 +
> > > > > lib/test_fprobe.c | 174 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > > > 3 files changed, 188 insertions(+)
> > > > > create mode 100644 lib/test_fprobe.c
> > > > >
> > > > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> > > > > index 14b89aa37c5c..ffc469a12afc 100644
> > > > > --- a/lib/Kconfig.debug
> > > > > +++ b/lib/Kconfig.debug
> > > > > @@ -2100,6 +2100,18 @@ config KPROBES_SANITY_TEST
> > > > >
> > > > > Say N if you are unsure.
> > > > >
> > > > > +config FPROBE_SANITY_TEST
> > > > > + bool "Self test for fprobe"
> > > > > + depends on DEBUG_KERNEL
> > > > > + depends on FPROBE
> > > > > + depends on KUNIT
> > > >
> > > > Hmm, this caused a build error with allmodconfig because KUNIT=m but FPROBE_SANITY_TEST=y.
> > > > Let me fix this issue.
> > >
> > > Please base on top of bpf-next and add [PATCH v11 bpf-next] to subject.
> >
> > OK, let me rebase on it.
> > There are master and for-next branch, which one is better to use?
> >
>
> Sorry, missed your reply earlier. Always rebase against master.
>
> You forgot to add "bpf-next" into [PATCH] prefix, so I had to manually
> mark it in patchworks as delegated to bpf queue (this is necessary for
> our CI to properly pick it up). For future submissions to bpf-next,
> please don't forget to add "bpf-next" marker.
>

Oops, sorry, I forgot that. Should I resend with that tag?

Thank you,

> > Thank you,
> >
> > >
> > > >
> > > > Thank you,
> > > >
> > > > > + help
> > > > > + This option will enable testing the fprobe when the system boot.
> > > > > + A series of tests are made to verify that the fprobe is functioning
> > > > > + properly.
> > > > > +
> > > > > + Say N if you are unsure.
> > > > > +
> > > > > config BACKTRACE_SELF_TEST
> > > > > tristate "Self test for the backtrace code"
> > > > > depends on DEBUG_KERNEL
> > > > > diff --git a/lib/Makefile b/lib/Makefile
> > > > > index 300f569c626b..154008764b16 100644
> > > > > --- a/lib/Makefile
> > > > > +++ b/lib/Makefile
> > > > > @@ -103,6 +103,8 @@ obj-$(CONFIG_TEST_HMM) += test_hmm.o
> > > > > obj-$(CONFIG_TEST_FREE_PAGES) += test_free_pages.o
> > > > > obj-$(CONFIG_KPROBES_SANITY_TEST) += test_kprobes.o
> > > > > obj-$(CONFIG_TEST_REF_TRACKER) += test_ref_tracker.o
> > > > > +CFLAGS_test_fprobe.o += $(CC_FLAGS_FTRACE)
> > > > > +obj-$(CONFIG_FPROBE_SANITY_TEST) += test_fprobe.o
> > > > > #
> > > > > # CFLAGS for compiling floating point code inside the kernel. x86/Makefile turns
> > > > > # off the generation of FPU/SSE* instructions for kernel proper but FPU_FLAGS
> > > > > diff --git a/lib/test_fprobe.c b/lib/test_fprobe.c
> > > > > new file mode 100644
> > > > > index 000000000000..ed70637a2ffa
> > > > > --- /dev/null
> > > > > +++ b/lib/test_fprobe.c
> > > > > @@ -0,0 +1,174 @@
> > > > > +// SPDX-License-Identifier: GPL-2.0-or-later
> > > > > +/*
> > > > > + * test_fprobe.c - simple sanity test for fprobe
> > > > > + */
> > > > > +
> > > > > +#include <linux/kernel.h>
> > > > > +#include <linux/fprobe.h>
> > > > > +#include <linux/random.h>
> > > > > +#include <kunit/test.h>
> > > > > +
> > > > > +#define div_factor 3
> > > > > +
> > > > > +static struct kunit *current_test;
> > > > > +
> > > > > +static u32 rand1, entry_val, exit_val;
> > > > > +
> > > > > +/* Use indirect calls to avoid inlining the target functions */
> > > > > +static u32 (*target)(u32 value);
> > > > > +static u32 (*target2)(u32 value);
> > > > > +static unsigned long target_ip;
> > > > > +static unsigned long target2_ip;
> > > > > +
> > > > > +static noinline u32 fprobe_selftest_target(u32 value)
> > > > > +{
> > > > > + return (value / div_factor);
> > > > > +}
> > > > > +
> > > > > +static noinline u32 fprobe_selftest_target2(u32 value)
> > > > > +{
> > > > > + return (value / div_factor) + 1;
> > > > > +}
> > > > > +
> > > > > +static notrace void fp_entry_handler(struct fprobe *fp, unsigned long ip, struct pt_regs *regs)
> > > > > +{
> > > > > + KUNIT_EXPECT_FALSE(current_test, preemptible());
> > > > > + /* This can be called on the fprobe_selftest_target and the fprobe_selftest_target2 */
> > > > > + if (ip != target_ip)
> > > > > + KUNIT_EXPECT_EQ(current_test, ip, target2_ip);
> > > > > + entry_val = (rand1 / div_factor);
> > > > > +}
> > > > > +
> > > > > +static notrace void fp_exit_handler(struct fprobe *fp, unsigned long ip, struct pt_regs *regs)
> > > > > +{
> > > > > + unsigned long ret = regs_return_value(regs);
> > > > > +
> > > > > + KUNIT_EXPECT_FALSE(current_test, preemptible());
> > > > > + if (ip != target_ip) {
> > > > > + KUNIT_EXPECT_EQ(current_test, ip, target2_ip);
> > > > > + KUNIT_EXPECT_EQ(current_test, ret, (rand1 / div_factor) + 1);
> > > > > + } else
> > > > > + KUNIT_EXPECT_EQ(current_test, ret, (rand1 / div_factor));
> > > > > + KUNIT_EXPECT_EQ(current_test, entry_val, (rand1 / div_factor));
> > > > > + exit_val = entry_val + div_factor;
> > > > > +}
> > > > > +
> > > > > +/* Test entry only (no rethook) */
> > > > > +static void test_fprobe_entry(struct kunit *test)
> > > > > +{
> > > > > + struct fprobe fp_entry = {
> > > > > + .entry_handler = fp_entry_handler,
> > > > > + };
> > > > > +
> > > > > + current_test = test;
> > > > > +
> > > > > + /* Before register, unregister should be failed. */
> > > > > + KUNIT_EXPECT_NE(test, 0, unregister_fprobe(&fp_entry));
> > > > > + KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp_entry, "fprobe_selftest_target*", NULL));
> > > > > +
> > > > > + entry_val = 0;
> > > > > + exit_val = 0;
> > > > > + target(rand1);
> > > > > + KUNIT_EXPECT_NE(test, 0, entry_val);
> > > > > + KUNIT_EXPECT_EQ(test, 0, exit_val);
> > > > > +
> > > > > + entry_val = 0;
> > > > > + exit_val = 0;
> > > > > + target2(rand1);
> > > > > + KUNIT_EXPECT_NE(test, 0, entry_val);
> > > > > + KUNIT_EXPECT_EQ(test, 0, exit_val);
> > > > > +
> > > > > + KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp_entry));
> > > > > +}
> > > > > +
> > > > > +static void test_fprobe(struct kunit *test)
> > > > > +{
> > > > > + struct fprobe fp = {
> > > > > + .entry_handler = fp_entry_handler,
> > > > > + .exit_handler = fp_exit_handler,
> > > > > + };
> > > > > +
> > > > > + current_test = test;
> > > > > + KUNIT_EXPECT_EQ(test, 0, register_fprobe(&fp, "fprobe_selftest_target*", NULL));
> > > > > +
> > > > > + entry_val = 0;
> > > > > + exit_val = 0;
> > > > > + target(rand1);
> > > > > + KUNIT_EXPECT_NE(test, 0, entry_val);
> > > > > + KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
> > > > > +
> > > > > + entry_val = 0;
> > > > > + exit_val = 0;
> > > > > + target2(rand1);
> > > > > + KUNIT_EXPECT_NE(test, 0, entry_val);
> > > > > + KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
> > > > > +
> > > > > + KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp));
> > > > > +}
> > > > > +
> > > > > +static void test_fprobe_syms(struct kunit *test)
> > > > > +{
> > > > > + static const char *syms[] = {"fprobe_selftest_target", "fprobe_selftest_target2"};
> > > > > + struct fprobe fp = {
> > > > > + .entry_handler = fp_entry_handler,
> > > > > + .exit_handler = fp_exit_handler,
> > > > > + };
> > > > > +
> > > > > + current_test = test;
> > > > > + KUNIT_EXPECT_EQ(test, 0, register_fprobe_syms(&fp, syms, 2));
> > > > > +
> > > > > + entry_val = 0;
> > > > > + exit_val = 0;
> > > > > + target(rand1);
> > > > > + KUNIT_EXPECT_NE(test, 0, entry_val);
> > > > > + KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
> > > > > +
> > > > > + entry_val = 0;
> > > > > + exit_val = 0;
> > > > > + target2(rand1);
> > > > > + KUNIT_EXPECT_NE(test, 0, entry_val);
> > > > > + KUNIT_EXPECT_EQ(test, entry_val + div_factor, exit_val);
> > > > > +
> > > > > + KUNIT_EXPECT_EQ(test, 0, unregister_fprobe(&fp));
> > > > > +}
> > > > > +
> > > > > +static unsigned long get_ftrace_location(void *func)
> > > > > +{
> > > > > + unsigned long size, addr = (unsigned long)func;
> > > > > +
> > > > > + if (!kallsyms_lookup_size_offset(addr, &size, NULL) || !size)
> > > > > + return 0;
> > > > > +
> > > > > + return ftrace_location_range(addr, addr + size - 1);
> > > > > +}
> > > > > +
> > > > > +static int fprobe_test_init(struct kunit *test)
> > > > > +{
> > > > > + do {
> > > > > + rand1 = prandom_u32();
> > > > > + } while (rand1 <= div_factor);
> > > > > +
> > > > > + target = fprobe_selftest_target;
> > > > > + target2 = fprobe_selftest_target2;
> > > > > + target_ip = get_ftrace_location(target);
> > > > > + target2_ip = get_ftrace_location(target2);
> > > > > +
> > > > > + return 0;
> > > > > +}
> > > > > +
> > > > > +static struct kunit_case fprobe_testcases[] = {
> > > > > + KUNIT_CASE(test_fprobe_entry),
> > > > > + KUNIT_CASE(test_fprobe),
> > > > > + KUNIT_CASE(test_fprobe_syms),
> > > > > + {}
> > > > > +};
> > > > > +
> > > > > +static struct kunit_suite fprobe_test_suite = {
> > > > > + .name = "fprobe_test",
> > > > > + .init = fprobe_test_init,
> > > > > + .test_cases = fprobe_testcases,
> > > > > +};
> > > > > +
> > > > > +kunit_test_suites(&fprobe_test_suite);
> > > > > +
> > > > > +MODULE_LICENSE("GPL");
> > > > >
> > > >
> > > >
> > > > --
> > > > Masami Hiramatsu <[email protected]>
> >
> >
> > --
> > Masami Hiramatsu <[email protected]>

--
Masami Hiramatsu <[email protected]>