linux-kernel - Re: [RFC PATCH 2/2] kbuild: Rename and use BINUTILS

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]
Date:   Sun, 19 Apr 2020 09:40:15 -0700
From:   Nathan Chancellor <natechancellor@...il.com>
To:     Sedat Dilek <sedat.dilek@...il.com>
Cc:     Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will@...nel.org>,
        Masahiro Yamada <masahiroy@...nel.org>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        "Steven Rostedt (VMware)" <rostedt@...dmis.org>,
        Jessica Yu <jeyu@...nel.org>,
        "Peter Zijlstra (Intel)" <peterz@...radead.org>,
        Mauro Carvalho Chehab <mchehab+samsung@...nel.org>,
        "Joel Fernandes (Google)" <joel@...lfernandes.org>,
        Patrick Bellasi <patrick.bellasi@....com>,
        Krzysztof Kozlowski <krzk@...nel.org>,
        Dan Williams <dan.j.williams@...el.com>,
        "Eric W. Biederman" <ebiederm@...ssion.com>,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 2/2] kbuild: Rename and use BINUTILS_VERSION where
 needed

Hi Sedat,

On Sun, Apr 19, 2020 at 03:21:42PM +0200, Sedat Dilek wrote:
> In the first patch I introduced LD_IS_BINUTILS Kconfig.

This sentence can removed, this offers no context when applied to the
git tree.

> To be consistent in naming convention I renamed from LD_VERSION
> to BINUTILS_VERSION.

Is this the only motivation behind this patch? Is this fixing a bug that
you have notice or is it just because there is a linker aside from GNU
ld that works with the kernel?

> So, we have the double "LD_IS_BINUTILS" and "BINUTILS_VERSION"
> like "CC_IS_GCC" and "GCC_VERSION".
> 
> For the same reason I renamed the shell script to detect the GNU ld
> linker version.
> 
> In Documentation/process/changes.rst we use "binutils" and GNU ld
> binary is part of it (see [3]).

                            ^ Kind of weird to have 3 come before 1 and 2

I also don't think that linking to the documentation like that is
necessary.

> The patches "init/kconfig: Add LD_VERSION Kconfig" (see [1]) and
> "arm64: Kconfig: ptrauth: Add binutils version check to fix mismatch"
> (see [2]) added checks for binutils >=2.23.1 whereas binutils

                                       ^2.33.1

> version 2.23 is minimum supported version (see [3]).

I am not sure what relevance this parapgraph has to the patch?

If it is indeed relevant, it should be reworked to use the standard
kernel way of referring to commits:

    9553d16fa671 ("init/kconfig: Add LD_VERSION Kconfig")
    15cd0e675f3f ("arm64: Kconfig: ptrauth: Add binutils version check to fix mismatch")

instead of the title of the patch with a link to the web interface.

> I have renamed to BINUTILS_VERSION where needed.
> 
> [1] https://git.kernel.org/linus/9553d16fa671b9621c5e2847d08bd90d3be3349c
> [2] https://git.kernel.org/linus/15cd0e675f3f76b4d21c313795fe0c23df0ee20f
> [3] https://git.kernel.org/linus/Documentation/process/changes.rst#n79
> 
> Signed-off-by: Sedat Dilek <sedat.dilek@...il.com>
> ---
>  arch/arm64/Kconfig                             | 2 +-
>  init/Kconfig                                   | 4 ++--
>  scripts/{ld-version.sh => binutils-version.sh} | 0
>  3 files changed, 3 insertions(+), 3 deletions(-)
>  rename scripts/{ld-version.sh => binutils-version.sh} (100%)
> 
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index 40fb05d96c60..274ba9b3ac95 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -1504,7 +1504,7 @@ config ARM64_PTR_AUTH
>  	depends on (CC_HAS_SIGN_RETURN_ADDRESS || CC_HAS_BRANCH_PROT_PAC_RET) && AS_HAS_PAC
>  	# GCC 9.1 and later inserts a .note.gnu.property section note for PAC
>  	# which is only understood by binutils starting with version 2.33.1.
> -	depends on !CC_IS_GCC || GCC_VERSION < 90100 || LD_VERSION >= 233010000
> +	depends on !CC_IS_GCC || GCC_VERSION < 90100 || BINUTILS_VERSION >= 233010000
>  	depends on !CC_IS_CLANG || AS_HAS_CFI_NEGATE_RA_STATE
>  	depends on (!FUNCTION_GRAPH_TRACER || DYNAMIC_FTRACE_WITH_REGS)
>  	help
> diff --git a/init/Kconfig b/init/Kconfig
> index 520116efea0f..946db4786951 100644
> --- a/init/Kconfig
> +++ b/init/Kconfig
> @@ -19,9 +19,9 @@ config GCC_VERSION
>  config LD_IS_BINUTILS
>  	def_bool $(success,$(LD) -v | head -n 1 | grep -q 'GNU ld')
>  
> -config LD_VERSION
> +config BINUTILS_VERSION
>  	int
> -	default $(shell,$(LD) --version | $(srctree)/scripts/ld-version.sh) if LD_IS_BINUTILS
> +	default $(shell,$(LD) --version | $(srctree)/scripts/binutils-version.sh) if LD_IS_BINUTILS

This is not the only place that ld-version.sh is used, all of these
spots need to be updated with the move.

$ rg --no-heading "ld-version.sh"
scripts/coccicheck:SPATCH_VERSION_NUM=$(echo $SPATCH_VERSION | ${DIR}/scripts/ld-version.sh)
scripts/coccicheck:    REQ_NUM=$(echo $REQ | ${DIR}/scripts/ld-version.sh)
scripts/nsdeps:SPATCH_REQ_VERSION_NUM=$(echo $SPATCH_REQ_VERSION | ${DIR}/scripts/ld-version.sh)
scripts/nsdeps:SPATCH_VERSION_NUM=$(echo $SPATCH_VERSION | ${DIR}/scripts/ld-version.sh)
scripts/Kbuild.include:ld-version = $(shell $(LD) --version | $(srctree)/scripts/ld-version.sh)
init/Kconfig:	default $(shell,$(LD) --version | $(srctree)/scripts/ld-version.sh)

>  config CC_IS_CLANG
>  	def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
> diff --git a/scripts/ld-version.sh b/scripts/binutils-version.sh
> similarity index 100%
> rename from scripts/ld-version.sh
> rename to scripts/binutils-version.sh
> -- 
> 2.26.1
> 

For what it's worth, I think the idea behind this patch is fine but I do
not think the first patch in the series is necessary. The only reason
that LD_IS_BINUTILS even exists is to hide BINUTILS_VERSION. However,
the only reason I have seen from you to do this is to match GCC_VERSION.
The reason that CONFIG_GCC_VERSION does not run gcc-version.sh is to
avoid mistaking clang for gcc due to the use of the __GNUC*__ macros.
scripts/ld-version.sh already returns zero for ld.lld so I think that
it is just sufficient to add CONFIG_LD_IS_LLD whenever we are ready for
it (I might have a reason to send it out today, we'll see).

$ ld.lld --version | ./scripts/ld-version.sh
0

Cheers,
Nathan