Re: [PATCH 1/1] lib: kunit: Suppress a compilation warning of frame size

From: Andrew Morton
Date: Tue May 18 2021 - 17:38:21 EST


On Tue, 18 May 2021 17:45:33 +0800 Zhen Lei <thunder.leizhen@xxxxxxxxxx> wrote:

> lib/bitfield_kunit.c: In function ‘test_bitfields_constants’:
> lib/bitfield_kunit.c:93:1: warning: the frame size of 7456 bytes is larger than 2048 bytes [-Wframe-larger-than=]
> }
> ^
>
> As the description of BITFIELD_KUNIT in lib/Kconfig.debug, it "Only useful
> for kernel devs running the KUnit test harness, and not intended for
> inclusion into a production build". Therefore, it is not worth modifying
> variable 'test_bitfields_constants' to clear this warning. Just suppress
> it.

Well, it would be better to fix this rather than hiding it. The
warning is there for a reason!

Firstly, why is this happening? Do those macros end up generating a
vast number of `kunit_assert' instances and gcc dumbly fails to reuse
the same stack slots?

It would be trivial to split test_bitfields_constants() into four
functions. Probably those should use noinline_for_stack to prevent gcc
from just inlining everything into the caller.

Also, what's up with this?

/*
* NOTE
* This whole function compiles (or at least should, if everything
* is going according to plan) to nothing after optimisation.
*/

If that's the case then why did the function use all this stack?

> diff --git a/lib/Makefile b/lib/Makefile
> index e11cfc18b6c0826..2cc359ec1fdd3e1 100644
> --- a/lib/Makefile
> +++ b/lib/Makefile
> @@ -348,6 +348,7 @@ obj-$(CONFIG_OBJAGG) += objagg.o
> obj-$(CONFIG_PLDMFW) += pldmfw/
>
> # KUnit tests
> +CFLAGS_bitfield_kunit.o := $(call cc-option,-Wframe-larger-than=10240)
> obj-$(CONFIG_BITFIELD_KUNIT) += bitfield_kunit.o
> obj-$(CONFIG_LIST_KUNIT_TEST) += list-test.o
> obj-$(CONFIG_LINEAR_RANGES_TEST) += test_linear_ranges.o