linux-kernel - Re: [PATCH 03/13] Makefile: Generate CLANG

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAKwvOd=YV1Ck3hYKEC9035o+yghy_Oh1VWAyeGLQP5B9SR9xLw@mail.gmail.com>
Date:   Wed, 14 Apr 2021 16:46:59 -0700
From:   Nick Desaulniers <ndesaulniers@...gle.com>
To:     Miguel Ojeda <ojeda@...nel.org>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        rust-for-linux@...r.kernel.org,
        Linux Kbuild mailing list <linux-kbuild@...r.kernel.org>,
        Linux Doc Mailing List <linux-doc@...r.kernel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Alex Gaynor <alex.gaynor@...il.com>,
        Geoffrey Thomas <geofft@...reload.com>,
        Finn Behrens <me@...enk.de>,
        Adam Bratschi-Kaye <ark.email@...il.com>,
        Wedson Almeida Filho <wedsonaf@...gle.com>
Subject: Re: [PATCH 03/13] Makefile: Generate CLANG_FLAGS even in GCC builds

On Wed, Apr 14, 2021 at 11:48 AM <ojeda@...nel.org> wrote:
>
> From: Miguel Ojeda <ojeda@...nel.org>
>
> To support Rust under GCC-built kernels, we need to save the flags that
> would have been passed if the kernel was being compiled with Clang.
>
> The reason is that bindgen -- the tool we use to generate Rust bindings
> to the C side of the kernel -- relies on libclang to parse C. Ideally:
>
>   - bindgen would support a GCC backend (requested at [1]),
>
>   - or the Clang driver would be perfectly compatible with GCC,
>     including plugins. Unlikely, of course, but perhaps a big
>     subset of configs may be possible to guarantee to be kept
>     compatible nevertheless.
>
> This is also the reason why GCC builds are very experimental and some
> configurations may not work (e.g. GCC_PLUGIN_RANDSTRUCT). However,
> we keep GCC builds working (for some example configs) in the CI
> to avoid diverging/regressing further, so that we are better prepared
> for the future when a solution might become available.
>
> [1] https://github.com/rust-lang/rust-bindgen/issues/1949
>
> Link: https://github.com/Rust-for-Linux/linux/issues/167
>
> Co-developed-by: Alex Gaynor <alex.gaynor@...il.com>
> Signed-off-by: Alex Gaynor <alex.gaynor@...il.com>
> Co-developed-by: Geoffrey Thomas <geofft@...reload.com>
> Signed-off-by: Geoffrey Thomas <geofft@...reload.com>
> Co-developed-by: Finn Behrens <me@...enk.de>
> Signed-off-by: Finn Behrens <me@...enk.de>
> Co-developed-by: Adam Bratschi-Kaye <ark.email@...il.com>
> Signed-off-by: Adam Bratschi-Kaye <ark.email@...il.com>
> Co-developed-by: Wedson Almeida Filho <wedsonaf@...gle.com>
> Signed-off-by: Wedson Almeida Filho <wedsonaf@...gle.com>
> Signed-off-by: Miguel Ojeda <ojeda@...nel.org>
> ---
>  Makefile | 27 ++++++++++++++++-----------
>  1 file changed, 16 insertions(+), 11 deletions(-)
>
> diff --git a/Makefile b/Makefile
> index d4784d181123..9c75354324ed 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -559,26 +559,31 @@ ifdef building_out_of_srctree
>         { echo "# this is build directory, ignore it"; echo "*"; } > .gitignore
>  endif
>
> -# The expansion should be delayed until arch/$(SRCARCH)/Makefile is included.
> -# Some architectures define CROSS_COMPILE in arch/$(SRCARCH)/Makefile.
> -# CC_VERSION_TEXT is referenced from Kconfig (so it needs export),
> -# and from include/config/auto.conf.cmd to detect the compiler upgrade.
> -CC_VERSION_TEXT = $(shell $(CC) --version 2>/dev/null | head -n 1 | sed 's/\#//g')
> +TENTATIVE_CLANG_FLAGS := -Werror=unknown-warning-option
>
> -ifneq ($(findstring clang,$(CC_VERSION_TEXT)),)
>  ifneq ($(CROSS_COMPILE),)
> -CLANG_FLAGS    += --target=$(notdir $(CROSS_COMPILE:%-=%))
> +TENTATIVE_CLANG_FLAGS  += --target=$(notdir $(CROSS_COMPILE:%-=%))
>  GCC_TOOLCHAIN_DIR := $(dir $(shell which $(CROSS_COMPILE)elfedit))
> -CLANG_FLAGS    += --prefix=$(GCC_TOOLCHAIN_DIR)$(notdir $(CROSS_COMPILE))
> +TENTATIVE_CLANG_FLAGS  += --prefix=$(GCC_TOOLCHAIN_DIR)$(notdir $(CROSS_COMPILE))
>  GCC_TOOLCHAIN  := $(realpath $(GCC_TOOLCHAIN_DIR)/..)
>  endif
>  ifneq ($(GCC_TOOLCHAIN),)
> -CLANG_FLAGS    += --gcc-toolchain=$(GCC_TOOLCHAIN)
> +TENTATIVE_CLANG_FLAGS  += --gcc-toolchain=$(GCC_TOOLCHAIN)
>  endif
>  ifneq ($(LLVM_IAS),1)
> -CLANG_FLAGS    += -no-integrated-as
> +TENTATIVE_CLANG_FLAGS  += -no-integrated-as
>  endif
> -CLANG_FLAGS    += -Werror=unknown-warning-option
> +
> +export TENTATIVE_CLANG_FLAGS

I'm ok with this approach, but I'm curious:
If the user made a copy of the CLANG_FLAGS variable and modified its
copy, would TENTATIVE_CLANG_FLAGS even be necessary? IIUC,
TENTATIVE_CLANG_FLAGS is used to filter out certain flags before
passing them to bindgen?

Or, I'm curious whether you even need to rename this variable (or
create a new variable) at all? Might make for a shorter diff if you
just keep the existing identifier (CLANG_FLAGS), but create them
unconditionally (at least not conditional on CC=clang).


> +
> +# The expansion should be delayed until arch/$(SRCARCH)/Makefile is included.
> +# Some architectures define CROSS_COMPILE in arch/$(SRCARCH)/Makefile.
> +# CC_VERSION_TEXT is referenced from Kconfig (so it needs export),
> +# and from include/config/auto.conf.cmd to detect the compiler upgrade.
> +CC_VERSION_TEXT = $(shell $(CC) --version 2>/dev/null | head -n 1 | sed 's/\#//g')
> +
> +ifneq ($(findstring clang,$(CC_VERSION_TEXT)),)
> +CLANG_FLAGS    += $(TENTATIVE_CLANG_FLAGS)
>  KBUILD_CFLAGS  += $(CLANG_FLAGS)
>  KBUILD_AFLAGS  += $(CLANG_FLAGS)
>  export CLANG_FLAGS
> --
> 2.17.1
>


--
Thanks,
~Nick Desaulniers