[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YyEU70K1aY8b/EXZ@dev-arch.thelio-3990X>
Date: Tue, 13 Sep 2022 16:40:31 -0700
From: Nathan Chancellor <nathan@...nel.org>
To: Petr Pavlu <petr.pavlu@...e.com>
Cc: tglx@...utronix.de, mingo@...hat.com, bp@...en8.de,
dave.hansen@...ux.intel.com, x86@...nel.org, hpa@...or.com,
nicolas@...sle.eu, masahiroy@...nel.org,
kirill.shutemov@...ux.intel.com, tony.luck@...el.com,
michael.roth@....com, ndesaulniers@...gle.com,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] x86: Avoid relocation information in final vmlinux
Hi Petr,
On Tue, Sep 13, 2022 at 03:29:11PM +0200, Petr Pavlu wrote:
> The Linux build process on x86 roughly consists of compiling all input
> files, statically linking them into a vmlinux ELF file, and then taking
> and turning this file into an actual bzImage bootable file.
>
> vmlinux has in this process two main purposes:
> 1) It is an intermediate build target on the way to produce the final
> bootable image.
> 2) It is a file that is expected to be used by debuggers and standard
> ELF tooling to work with the built kernel.
>
> For the second purpose, a vmlinux file is typically collected by various
> package build recipes, such as distribution spec files, including the
> kernel's own binrpm-pkg target.
>
> When building a kernel supporting KASLR with CONFIG_X86_NEED_RELOCS,
> vmlinux contains also relocation information produced by using the
> --emit-relocs linker option. This is utilized by subsequent build steps
> to create vmlinux.relocs and produce a relocatable image. However, the
> information is not needed by debuggers and other standard ELF tooling.
>
> The issue is then that the collected vmlinux file and hence distribution
> packages end up unnecessarily large because of this extra data. The
> following is a size comparison of vmlinux v6.0-rc5 with and without the
> relocation information:
> | Configuration | With relocs | Stripped relocs |
> | x86_64_defconfig | 70 MB | 43 MB |
> | +CONFIG_DEBUG_INFO | 818 MB | 367 MB |
>
> The patch optimizes a resulting vmlinux by adding a postlink step that
> splits the relocation information into vmlinux.relocs and then strips it
> from the vmlinux binary.
>
> Signed-off-by: Petr Pavlu <petr.pavlu@...e.com>
> ---
>
> Notes:
> The implemented approach has two shortcomings which is worth mentioning:
> * The vmlinux file is rewritten in place. This creates a problem if the
> build is interrupted when ld produced vmlinux but before the postlink
> is started. A subsequent restart of the build would use the
> incompletely processed vmlinux.
>
> The current build logic already appears to have this problem as
> scripts/link-vmlinux.sh rewrites vmlinux in place, for example, when
> handling CONFIG_BUILDTIME_TABLE_SORT. It could be solved by using an
> intermediate target and renaming it to vmlinux only once the file is
> final.
>
> * vmlinux.relocs is hidden from the Makefile workflow and Make is
> additionally told about the file in arch/x86/boot/compressed/Makefile.
>
> .gitignore | 1 +
> arch/x86/Makefile.postlink | 41 +++++++++++++++++++++++++++++++
> arch/x86/boot/compressed/Makefile | 10 +++-----
> 3 files changed, 46 insertions(+), 6 deletions(-)
> create mode 100644 arch/x86/Makefile.postlink
>
> diff --git a/.gitignore b/.gitignore
> index 265959544978..cd4ef88584ea 100644
> --- a/.gitignore
> +++ b/.gitignore
> @@ -37,6 +37,7 @@
> *.o
> *.o.*
> *.patch
> +*.relocs
> *.s
> *.so
> *.so.dbg
> diff --git a/arch/x86/Makefile.postlink b/arch/x86/Makefile.postlink
> new file mode 100644
> index 000000000000..4650aaf6d8b3
> --- /dev/null
> +++ b/arch/x86/Makefile.postlink
> @@ -0,0 +1,41 @@
> +# SPDX-License-Identifier: GPL-2.0
> +# ===========================================================================
> +# Post-link x86 pass
> +# ===========================================================================
> +#
> +# 1. Separate relocations from vmlinux into vmlinux.relocs.
> +# 2. Strip relocations from vmlinux.
> +
> +PHONY := __archpost
> +__archpost:
> +
> +-include include/config/auto.conf
> +include scripts/Kbuild.include
> +
> +CMD_RELOCS = arch/x86/tools/relocs
> +quiet_cmd_relocs = RELOCS $@...locs
> + cmd_relocs = $(CMD_RELOCS) $@ > $@...locs;$(CMD_RELOCS) --abs-relocs $@
> +
> +quiet_cmd_strip_relocs = RSTRIP $@
> + cmd_strip_relocs = objcopy --remove-relocations='*' $@
Just a small drive by comment, prefer $(OBJCOPY) over objcopy so that
the user's choice of objcopy is respected (such as llvm-objcopy).
Unfortunately, llvm-objcopy does not appear to support
'--remove-relocations'. We can certainly file a feature request for this
upstream but is there a way to accomplish this in a different way? Or
perhaps this could be something that is controlled via Kconfig so it
is only selectable with GNU objcopy??
Cheers,
Nathan
> +
> +# `@...e` prevents complaint when there is nothing to be done
> +
> +vmlinux: FORCE
> + @true
> +ifeq ($(CONFIG_X86_NEED_RELOCS),y)
> + $(call cmd,relocs)
> + $(call cmd,strip_relocs)
> +endif
> +
> +%.ko: FORCE
> + @true
> +
> +clean:
> + @rm -f vmlinux.relocs
> +
> +PHONY += FORCE clean
> +
> +FORCE:
> +
> +.PHONY: $(PHONY)
> diff --git a/arch/x86/boot/compressed/Makefile b/arch/x86/boot/compressed/Makefile
> index 35ce1a64068b..eba7709d75ae 100644
> --- a/arch/x86/boot/compressed/Makefile
> +++ b/arch/x86/boot/compressed/Makefile
> @@ -120,14 +120,12 @@ $(obj)/vmlinux.bin: vmlinux FORCE
>
> targets += $(patsubst $(obj)/%,%,$(vmlinux-objs-y)) vmlinux.bin.all vmlinux.relocs
>
> -CMD_RELOCS = arch/x86/tools/relocs
> -quiet_cmd_relocs = RELOCS $@
> - cmd_relocs = $(CMD_RELOCS) $< > $@;$(CMD_RELOCS) --abs-relocs $<
> -$(obj)/vmlinux.relocs: vmlinux FORCE
> - $(call if_changed,relocs)
> +# vmlinux.relocs is created by the vmlinux postlink step.
> +vmlinux.relocs: vmlinux
> + @true
>
> vmlinux.bin.all-y := $(obj)/vmlinux.bin
> -vmlinux.bin.all-$(CONFIG_X86_NEED_RELOCS) += $(obj)/vmlinux.relocs
> +vmlinux.bin.all-$(CONFIG_X86_NEED_RELOCS) += vmlinux.relocs
>
> $(obj)/vmlinux.bin.gz: $(vmlinux.bin.all-y) FORCE
> $(call if_changed,gzip)
> --
> 2.35.3
>
Powered by blists - more mailing lists