| Message ID | 20181203195805.28835-1-ard.biesheuvel@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | arm64: relocatable: fix inconsistencies in linker script and options | expand |
Hey Ard, On Mon, Dec 03, 2018 at 08:58:05PM +0100, Ard Biesheuvel wrote: > readelf complains about the section layout of vmlinux when building > with CONFIG_RELOCATABLE=y (for KASLR): > > readelf: Warning: [21]: Link field (0) should index a symtab section. > readelf: Warning: [21]: Info field (0) should index a relocatable section. > > Also, it seems that our use of '-pie -shared' is contradictory, and > thus ambiguous. In general, the way KASLR is wired up at the moment > is highly tailored to how ld.bfd happens to implement (and conflate) > PIE executables and shared libraries, so given the current effort to > support other toolchains, let's fix some of these issues as well. > > - Drop the -pie linker argument and just leave -shared. In ld.bfd, > the differences between them are unclear (except for the ELF type > of the produced image [0]) but lld chokes on seeing both at the > same time. > > - Rename the .rela output section to .rela.dyn, as is customary for > shared libraries and PIE executables, so that it is not misidentified > by readelf as a static relocation section (producing the warnings > above). > > - Pass the -z notext and -z norelro options to explicitly instruct the > linker to permit text relocations, and to omit the RELRO program > header (which requires a certain section layout that we don't adhere > to in the kernel). These are the defaults for current versions of > ld.bfd. > > - Discard .eh_frame and .gnu.hash sections to avoid them from being > emitted between .head.text and .text, screwing up the section layout. > > These changes only affect the ELF image, and produce the same binary > image. Thanks, I've queued this up on for-next/core. > [0] b9dce7f1ba01 ("arm64: kernel: force ET_DYN ELF type for ...") > > Cc: Nick Desaulniers <ndesaulniers@google.com> > Cc: Peter Smith <peter.smith@linaro.org> > Tested-by: Nick Desaulniers <ndesaulniers@google.com> > Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> > --- > This supersedes > > [RFT PATCH] arm64: relocatable: build the kernel as a proper shared library > [RFT PATCH] arm64: add support for building the KASLR kernel with LLVM lld Also, this ^^ is /really/ helpful :) Thanks, Will
diff --git a/arch/arm64/Makefile b/arch/arm64/Makefile index 6cb9fc7e9382..8978f60779c4 100644 --- a/arch/arm64/Makefile +++ b/arch/arm64/Makefile @@ -18,7 +18,7 @@ ifeq ($(CONFIG_RELOCATABLE), y) # Pass --no-apply-dynamic-relocs to restore pre-binutils-2.27 behaviour # for relative relocs, since this leads to better Image compression # with the relocation offsets always being zero. -LDFLAGS_vmlinux += -pie -shared -Bsymbolic \ +LDFLAGS_vmlinux += -shared -Bsymbolic -z notext -z norelro \ $(call ld-option, --no-apply-dynamic-relocs) endif diff --git a/arch/arm64/kernel/vmlinux.lds.S b/arch/arm64/kernel/vmlinux.lds.S index 03b00007553d..7fa008374907 100644 --- a/arch/arm64/kernel/vmlinux.lds.S +++ b/arch/arm64/kernel/vmlinux.lds.S @@ -99,7 +99,8 @@ SECTIONS *(.discard) *(.discard.*) *(.interp .dynamic) - *(.dynsym .dynstr .hash) + *(.dynsym .dynstr .hash .gnu.hash) + *(.eh_frame) } . = KIMAGE_VADDR + TEXT_OFFSET; @@ -192,12 +193,12 @@ SECTIONS PERCPU_SECTION(L1_CACHE_BYTES) - .rela : ALIGN(8) { + .rela.dyn : ALIGN(8) { *(.rela .rela*) } - __rela_offset = ABSOLUTE(ADDR(.rela) - KIMAGE_VADDR); - __rela_size = SIZEOF(.rela); + __rela_offset = ABSOLUTE(ADDR(.rela.dyn) - KIMAGE_VADDR); + __rela_size = SIZEOF(.rela.dyn); . = ALIGN(SEGMENT_ALIGN); __initdata_end = .;
readelf complains about the section layout of vmlinux when building with CONFIG_RELOCATABLE=y (for KASLR): readelf: Warning: [21]: Link field (0) should index a symtab section. readelf: Warning: [21]: Info field (0) should index a relocatable section. Also, it seems that our use of '-pie -shared' is contradictory, and thus ambiguous. In general, the way KASLR is wired up at the moment is highly tailored to how ld.bfd happens to implement (and conflate) PIE executables and shared libraries, so given the current effort to support other toolchains, let's fix some of these issues as well. - Drop the -pie linker argument and just leave -shared. In ld.bfd, the differences between them are unclear (except for the ELF type of the produced image [0]) but lld chokes on seeing both at the same time. - Rename the .rela output section to .rela.dyn, as is customary for shared libraries and PIE executables, so that it is not misidentified by readelf as a static relocation section (producing the warnings above). - Pass the -z notext and -z norelro options to explicitly instruct the linker to permit text relocations, and to omit the RELRO program header (which requires a certain section layout that we don't adhere to in the kernel). These are the defaults for current versions of ld.bfd. - Discard .eh_frame and .gnu.hash sections to avoid them from being emitted between .head.text and .text, screwing up the section layout. These changes only affect the ELF image, and produce the same binary image. [0] b9dce7f1ba01 ("arm64: kernel: force ET_DYN ELF type for ...") Cc: Nick Desaulniers <ndesaulniers@google.com> Cc: Peter Smith <peter.smith@linaro.org> Tested-by: Nick Desaulniers <ndesaulniers@google.com> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> --- This supersedes [RFT PATCH] arm64: relocatable: build the kernel as a proper shared library [RFT PATCH] arm64: add support for building the KASLR kernel with LLVM lld arch/arm64/Makefile | 2 +- arch/arm64/kernel/vmlinux.lds.S | 9 +++++---- 2 files changed, 6 insertions(+), 5 deletions(-)