Re: [PATCH 4.14 09/69] x86: vdso: Use $LD instead of $CC to link

[Nathan Chancellor <natechancellor@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 6409 bytes --]

On Fri, Apr 26, 2019 at 11:41:30AM +0000, Rantala, Tommi T. (Nokia - FI/Espoo) wrote:
> On Mon, 2019-04-15 at 20:58 +0200, Greg Kroah-Hartman wrote:
> > commit 379d98ddf41344273d9718556f761420f4dc80b3 upstream.
> > 
> 
> Hi,
> 
> With this patch in 4.14.112 build-id is now missing in vdso32.so:
> 
> $ file arch/x86/entry/vdso/vdso*so*
> arch/x86/entry/vdso/vdso32.so:     ELF 32-bit LSB pie executable, Intel
> 80386, version 1 (SYSV), dynamically linked, stripped
> arch/x86/entry/vdso/vdso32.so.dbg: ELF 32-bit LSB pie executable, Intel
> 80386, version 1 (SYSV), dynamically linked, with debug_info, not
> stripped
> arch/x86/entry/vdso/vdso64.so:     ELF 64-bit LSB pie executable, x86-
> 64, version 1 (SYSV), dynamically linked,
> BuildID[sha1]=d80730a5b561a3161e488a369d1c76c250b584b4, stripped
> arch/x86/entry/vdso/vdso64.so.dbg: ELF 64-bit LSB pie executable, x86-
> 64, version 1 (SYSV), dynamically linked,
> BuildID[sha1]=d80730a5b561a3161e488a369d1c76c250b584b4, with
> debug_info, not stripped
> 
> 
> Based on quick check, "$(call ld-option, --build-id)" fails due to some
> 32/64 bit mismatch, so the --build-id linker flag is not used when
> linking vdso32.so
> 
> Perhaps scripts/Kbuild.include is missing some change in 4.14.y to make
> this work properly.
> 

Hi Tommi,

This appears to be fixed by commit 0294e6f4a000 ("kbuild: simplify
ld-option implementation") upstream. Could you test the attached
backport and make sure everything works on your end? Assuming that it
does, I will test the other stable releases and see if this is needed
and send those backports along.

Thanks and sorry for the trouble!
Nathan

> -Tommi
> 
> > The vdso{32,64}.so can fail to link with CC=clang when clang tries to
> > find
> > a suitable GCC toolchain to link these libraries with.
> > 
> > /usr/bin/ld: arch/x86/entry/vdso/vclock_gettime.o:
> >   access beyond end of merged section (782)
> > 
> > This happens because the host environment leaked into the cross
> > compiler
> > environment due to the way clang searches for suitable GCC
> > toolchains.
> > 
> > Clang is a retargetable compiler, and each invocation of it must
> > provide
> > --target=<something> --gcc-toolchain=<something> to allow it to find
> > the
> > correct binutils for cross compilation. These flags had been added to
> > KBUILD_CFLAGS, but the vdso code uses CC and not KBUILD_CFLAGS (for
> > various
> > reasons) which breaks clang's ability to find the correct linker when
> > cross
> > compiling.
> > 
> > Most of the time this goes unnoticed because the host linker is new
> > enough
> > to work anyway, or is incompatible and skipped, but this cannot be
> > reliably
> > assumed.
> > 
> > This change alters the vdso makefile to just use LD directly, which
> > bypasses clang and thus the searching problem. The makefile will just
> > use
> > ${CROSS_COMPILE}ld instead, which is always what we want. This
> > matches the
> > method used to link vmlinux.
> > 
> > This drops references to DISABLE_LTO; this option doesn't seem to be
> > set
> > anywhere, and not knowing what its possible values are, it's not
> > clear how
> > to convert it from CC to LD flag.
> > 
> > Signed-off-by: Alistair Strachan <astrachan@google.com>
> > Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> > Acked-by: Andy Lutomirski <luto@kernel.org>
> > Cc: "H. Peter Anvin" <hpa@zytor.com>
> > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > Cc: kernel-team@android.com
> > Cc: joel@joelfernandes.org
> > Cc: Andi Kleen <andi.kleen@intel.com>
> > Link: 
> > https://lkml.kernel.org/r/20180803173931.117515-1-astrachan@google.com
> > Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
> > Signed-off-by: Sasha Levin <sashal@kernel.org>
> > ---
> >  arch/x86/entry/vdso/Makefile | 22 +++++++++-------------
> >  1 file changed, 9 insertions(+), 13 deletions(-)
> > 
> > diff --git a/arch/x86/entry/vdso/Makefile
> > b/arch/x86/entry/vdso/Makefile
> > index 0a550dc5c525..0defcc939ab4 100644
> > --- a/arch/x86/entry/vdso/Makefile
> > +++ b/arch/x86/entry/vdso/Makefile
> > @@ -48,10 +48,8 @@ targets += $(vdso_img_sodbg)
> >  
> >  export CPPFLAGS_vdso.lds += -P -C
> >  
> > -VDSO_LDFLAGS_vdso.lds = -m64 -Wl,-soname=linux-vdso.so.1 \
> > -			-Wl,--no-undefined \
> > -			-Wl,-z,max-page-size=4096 -Wl,-z,common-page-
> > size=4096 \
> > -			$(DISABLE_LTO)
> > +VDSO_LDFLAGS_vdso.lds = -m elf_x86_64 -soname linux-vdso.so.1 --no-
> > undefined \
> > +			-z max-page-size=4096 -z common-page-size=4096
> >  
> >  $(obj)/vdso64.so.dbg: $(src)/vdso.lds $(vobjs) FORCE
> >  	$(call if_changed,vdso)
> > @@ -103,10 +101,8 @@ CFLAGS_REMOVE_vvar.o = -pg
> >  #
> >  
> >  CPPFLAGS_vdsox32.lds = $(CPPFLAGS_vdso.lds)
> > -VDSO_LDFLAGS_vdsox32.lds = -Wl,-m,elf32_x86_64 \
> > -			   -Wl,-soname=linux-vdso.so.1 \
> > -			   -Wl,-z,max-page-size=4096 \
> > -			   -Wl,-z,common-page-size=4096
> > +VDSO_LDFLAGS_vdsox32.lds = -m elf32_x86_64 -soname linux-vdso.so.1 \
> > +			   -z max-page-size=4096 -z common-page-
> > size=4096
> >  
> >  # 64-bit objects to re-brand as x32
> >  vobjs64-for-x32 := $(filter-out $(vobjs-nox32),$(vobjs-y))
> > @@ -134,7 +130,7 @@ $(obj)/vdsox32.so.dbg: $(src)/vdsox32.lds
> > $(vobjx32s) FORCE
> >  	$(call if_changed,vdso)
> >  
> >  CPPFLAGS_vdso32.lds = $(CPPFLAGS_vdso.lds)
> > -VDSO_LDFLAGS_vdso32.lds = -m32 -Wl,-m,elf_i386 -Wl,-soname=linux-
> > gate.so.1
> > +VDSO_LDFLAGS_vdso32.lds = -m elf_i386 -soname linux-gate.so.1
> >  
> >  # This makes sure the $(obj) subdirectory exists even though vdso32/
> >  # is not a kbuild sub-make subdirectory.
> > @@ -180,13 +176,13 @@ $(obj)/vdso32.so.dbg: FORCE \
> >  # The DSO images are built using a special linker script.
> >  #
> >  quiet_cmd_vdso = VDSO    $@
> > -      cmd_vdso = $(CC) -nostdlib -o $@ \
> > +      cmd_vdso = $(LD) -nostdlib -o $@ \
> >  		       $(VDSO_LDFLAGS) $(VDSO_LDFLAGS_$(filter
> > %.lds,$(^F))) \
> > -		       -Wl,-T,$(filter %.lds,$^) $(filter %.o,$^) && \
> > +		       -T $(filter %.lds,$^) $(filter %.o,$^) && \
> >  		 sh $(srctree)/$(src)/checkundef.sh '$(NM)' '$@'
> >  
> > -VDSO_LDFLAGS = -fPIC -shared $(call cc-ldoption, -Wl$(comma)--hash-
> > style=both) \
> > -	$(call cc-ldoption, -Wl$(comma)--build-id) -Wl,-Bsymbolic
> > $(LTO_CFLAGS)
> > +VDSO_LDFLAGS = -shared $(call ld-option, --hash-style=both) \
> > +	$(call ld-option, --build-id) -Bsymbolic
> >  GCOV_PROFILE := n
> >  
> >  #
> 

[-- Attachment #2: 0001-kbuild-simplify-ld-option-implementation.patch --]
[-- Type: text/plain, Size: 3376 bytes --]

From 030323b8729be25808a36281f385c0458d232f39 Mon Sep 17 00:00:00 2001
From: Masahiro Yamada <yamada.masahiro@socionext.com>
Date: Fri, 23 Feb 2018 13:56:53 +0900
Subject: [PATCH] kbuild: simplify ld-option implementation

commit 0294e6f4a0006856e1f36b8cd8fa088d9e499e98 upstream.

Currently, linker options are tested by the coordination of $(CC) and
$(LD) because $(LD) needs some object to link.

As commit 86a9df597cdd ("kbuild: fix linker feature test macros when
cross compiling with Clang") addressed, we need to make sure $(CC)
and $(LD) agree the underlying architecture of the passed object.

This could be a bit complex when we combine tools from different groups.
For example, we can use clang for $(CC), but we still need to rely on
GCC toolchain for $(LD).

So, I was searching for a way of standalone testing of linker options.
A trick I found is to use '-v'; this not only prints the version string,
but also tests if the given option is recognized.

If a given option is supported,

  $ aarch64-linux-gnu-ld -v --fix-cortex-a53-843419
  GNU ld (Linaro_Binutils-2017.11) 2.28.2.20170706
  $ echo $?
  0

If unsupported,

  $ aarch64-linux-gnu-ld -v --fix-cortex-a53-843419
  GNU ld (crosstool-NG linaro-1.13.1-4.7-2013.04-20130415 - Linaro GCC 2013.04) 2.23.1
  aarch64-linux-gnu-ld: unrecognized option '--fix-cortex-a53-843419'
  aarch64-linux-gnu-ld: use the --help option for usage information
  $ echo $?
  1

Gold works likewise.

  $ aarch64-linux-gnu-ld.gold -v --fix-cortex-a53-843419
  GNU gold (Linaro_Binutils-2017.11 2.28.2.20170706) 1.14
  masahiro@pug:~/ref/linux$ echo $?
  0
  $ aarch64-linux-gnu-ld.gold -v --fix-cortex-a53-999999
  GNU gold (Linaro_Binutils-2017.11 2.28.2.20170706) 1.14
  aarch64-linux-gnu-ld.gold: --fix-cortex-a53-999999: unknown option
  aarch64-linux-gnu-ld.gold: use the --help option for usage information
  $ echo $?
  1

LLD too.

  $ ld.lld -v --gc-sections
  LLD 7.0.0 (http://llvm.org/git/lld.git 4a0e4190e74cea19f8a8dc625ccaebdf8b5d1585) (compatible with GNU linkers)
  $ echo $?
  0
  $ ld.lld -v --fix-cortex-a53-843419
  LLD 7.0.0 (http://llvm.org/git/lld.git 4a0e4190e74cea19f8a8dc625ccaebdf8b5d1585) (compatible with GNU linkers)
  $ echo $?
  0
  $ ld.lld -v --fix-cortex-a53-999999
  ld.lld: error: unknown argument: --fix-cortex-a53-999999
  LLD 7.0.0 (http://llvm.org/git/lld.git 4a0e4190e74cea19f8a8dc625ccaebdf8b5d1585) (compatible with GNU linkers)
  $ echo $?
  1

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Tested-by: Nick Desaulniers <ndesaulniers@google.com>
[nc: try-run-cached was added later, just use try-run, which is the
     current mainline state]
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
---
 scripts/Kbuild.include | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/scripts/Kbuild.include b/scripts/Kbuild.include
index a0ad87e869f9..a33fa1a91873 100644
--- a/scripts/Kbuild.include
+++ b/scripts/Kbuild.include
@@ -165,9 +165,7 @@ cc-ldoption = $(call try-run,\
 
 # ld-option
 # Usage: LDFLAGS += $(call ld-option, -X)
-ld-option = $(call try-run,\
-	$(CC) $(KBUILD_CPPFLAGS) $(CC_OPTION_CFLAGS) -x c /dev/null -c -o "$$TMPO"; \
-	$(LD) $(LDFLAGS) $(1) "$$TMPO" -o "$$TMP",$(1),$(2))
+ld-option = $(call try-run, $(LD) $(LDFLAGS) $(1) -v,$(1),$(2))
 
 # ar-option
 # Usage: KBUILD_ARFLAGS := $(call ar-option,D)
-- 
2.21.0