Re: Re: [PATCH v3 1/3] kasan: switch kunit tests to console tracepoints

[Paul Heidekrüger]

On 12.12.2023 10:32, Marco Elver wrote:
> On Tue, 12 Dec 2023 at 10:19, Paul Heidekrüger <paul.heidekrueger@xxxxxx> wrote:
> >
> > On 12.12.2023 00:37, Andrey Konovalov wrote:
> > > On Tue, Dec 12, 2023 at 12:35 AM Paul Heidekrüger
> > > <paul.heidekrueger@xxxxxx> wrote:
> > > >
> > > > Using CONFIG_FTRACE=y instead of CONFIG_TRACEPOINTS=y produces the same error
> > > > for me.
> > > >
> > > > So
> > > >
> > > >         CONFIG_KUNIT=y
> > > >         CONFIG_KUNIT_ALL_TESTS=n
> > > >         CONFIG_FTRACE=y
> > > >         CONFIG_KASAN=y
> > > >         CONFIG_KASAN_GENERIC=y
> > > >         CONFIG_KASAN_KUNIT_TEST=y
> > > >
> > > > produces
> > > >
> > > >         ➜   ./tools/testing/kunit/kunit.py run --kunitconfig=mm/kasan/.kunitconfig --arch=arm64
> > > >         Configuring KUnit Kernel ...
> > > >         Regenerating .config ...
> > > >         Populating config with:
> > > >         $ make ARCH=arm64 O=.kunit olddefconfig CC=clang
> > > >         ERROR:root:Not all Kconfig options selected in kunitconfig were in the generated .config.
> > > >         This is probably due to unsatisfied dependencies.
> > > >         Missing: CONFIG_KASAN_KUNIT_TEST=y
> > > >
> > > > By that error message, CONFIG_FTRACE appears to be present in the generated
> > > > config, but CONFIG_KASAN_KUNIT_TEST still isn't. Presumably,
> > > > CONFIG_KASAN_KUNIT_TEST is missing because of an unsatisfied dependency, which
> > > > must be CONFIG_TRACEPOINTS, unless I'm missing something ...
> > > >
> > > > If I just generate an arm64 defconfig and select CONFIG_FTRACE=y,
> > > > CONFIG_TRACEPOINTS=y shows up in my .config. So, maybe this is kunit.py-related
> > > > then?
> > > >
> > > > Andrey, you said that the tests have been working for you; are you running them
> > > > with kunit.py?
> > >
> > > No, I just run the kernel built with a config file that I put together
> > > based on defconfig.
> >
> > Ah. I believe I've figured it out.
> >
> > When I add CONFIG_STACK_TRACER=y in addition to CONFIG_FTRACE=y, it works.
> 
> CONFIG_FTRACE should be enough - maybe also check x86 vs. arm64 to debug more.

See below.

> > CONFIG_STACK_TRACER selects CONFIG_FUNCTION_TRACER, CONFIG_FUNCTION_TRACER
> > selects CONFIG_GENERIC_TRACER, CONFIG_GENERIC_TRACER selects CONFIG_TRACING, and
> > CONFIG_TRACING selects CONFIG_TRACEPOINTS.
> >
> > CONFIG_BLK_DEV_IO_TRACE=y also works instead of CONFIG_STACK_TRACER=y, as it
> > directly selects CONFIG_TRACEPOINTS.
> >
> > CONFIG_FTRACE=y on its own does not appear suffice for kunit.py on arm64.
> 
> When you build manually with just CONFIG_FTRACE, is CONFIG_TRACEPOINTS enabled?

When I add CONFIG_FTRACE and enter-key my way through the FTRACE prompts - I 
believe because CONFIG_FTRACE is a menuconfig? - at the beginning of a build, 
CONFIG_TRACEPOINTS does get set on arm64, yes.

On X86, the defconfig already includes CONIFG_TRACEPOINTS.

I also had a closer look at how kunit.py builds its configs.
I believe it does something along the following lines:

	cp <path_to_kunitconfig> .kunit/.config
	make ARCH=arm64 O=.kunit olddefconfig

On arm64, that isn't enough to set CONFIG_TRACEPOINTS; same behaviour when run 
outside of kunit.py.

For CONFIG_TRACEPOINTS, `make ARCH=arm64 menuconfig` shows:

	Symbol: TRACEPOINTS [=n]
	Type  : bool
	Defined at init/Kconfig:1920
	Selected by [n]:
		- TRACING [=n]
		- BLK_DEV_IO_TRACE [=n] && FTRACE [=y] && SYSFS [=y] && BLOCK [=y]

So, CONFIG_TRACING or CONFIG_BLK_DEV_IO_TRACE are the two options that prevent 
CONFIG_TRACEPOINTS from being set on arm64.

For CONFIG_TRACING we have:

	Symbol: TRACING [=n]
	Type  : bool
	Defined at kernel/trace/Kconfig:157
	Selects: RING_BUFFER [=n] && STACKTRACE [=y] && TRACEPOINTS [=n] && NOP_TRACER [=n] && BINARY_PRINTF [=n] && EVENT_TRACING [=n] && TRACE_CLOCK [=y] && TASKS_RCU [=n]
	Selected by [n]:
		- DRM_I915_TRACE_GEM [=n] && HAS_IOMEM [=y] && DRM_I915 [=n] && EXPERT [=n] && DRM_I915_DEBUG_GEM [=n]
		- DRM_I915_TRACE_GTT [=n] && HAS_IOMEM [=y] && DRM_I915 [=n] && EXPERT [=n] && DRM_I915_DEBUG_GEM [=n]
		- PREEMPTIRQ_TRACEPOINTS [=n] && (TRACE_PREEMPT_TOGGLE [=n] || TRACE_IRQFLAGS [=n])
		- GENERIC_TRACER [=n]
		- ENABLE_DEFAULT_TRACERS [=n] && FTRACE [=y] && !GENERIC_TRACER [=n]
		- FPROBE_EVENTS [=n] && FTRACE [=y] && FPROBE [=n] && HAVE_REGS_AND_STACK_ACCESS_API [=y]
		- KPROBE_EVENTS [=n] && FTRACE [=y] && KPROBES [=n] && HAVE_REGS_AND_STACK_ACCESS_API [=y]
		- UPROBE_EVENTS [=n] && FTRACE [=y] && ARCH_SUPPORTS_UPROBES [=y] && MMU [=y] && PERF_EVENTS [=n]
		- SYNTH_EVENTS [=n] && FTRACE [=y]
		- USER_EVENTS [=n] && FTRACE [=y]
		- HIST_TRIGGERS [=n] && FTRACE [=y] && ARCH_HAVE_NMI_SAFE_CMPXCHG [=y]

> > I believe the reason my .kunitconfig as well as the existing
> > mm/kfence/.kunitconfig work on X86 is because CONFIG_TRACEPOINTS=y is present in
> > an X86 defconfig.
> >
> > Does this make sense?
> >
> > Would you welcome a patch addressing this for the existing
> > mm/kfence/.kunitconfig?
> >
> > I would also like to submit a patch for an mm/kasan/.kunitconfig. Do you think
> > that would be helpful too?
> >
> > FWICT, kernel/kcsan/.kunitconfig might also be affected since
> > CONFIG_KCSAN_KUNIT_TEST also depends on CONFIG_TRACEPOITNS, but I would have to
> > test that. That could be a third patch.
> 
> I'd support figuring out the minimal config (CONFIG_FTRACE or
> something else?) that satisfies the TRACEPOINTS dependency. I always
> thought CONFIG_FTRACE ought to be the one config option, but maybe
> something changed.

If we want a minimal config, setting CONFIG_BLK_DEV_IO_TRACE, 
CONFIG_SYNTH_EVENTS or CONFIG_USER_EVENTS seem like viable options, for 
instance. But AFAICT, setting them in the context of KASan doesn't really make 
sense, and I might be missing an obvious choice here too.

What do you think?

> Also maybe one of the tracing maintainers can help untangle what's
> going on here.
> 
> Thanks,
> -- Marco

Many thanks,
Paul