| Message ID | 20200624014940.1204448-10-keescook@chromium.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Warn on orphan section placement | expand |
On Tue, Jun 23, 2020 at 06:49:40PM -0700, Kees Cook wrote: > We don't want to depend on the linker's orphan section placement > heuristics as these can vary between linkers, and may change between > versions. All sections need to be explicitly named in the linker > script. > > Avoid .eh_frame* by making sure both -fno-asychronous-unwind-tables and > -fno-unwind-tables are present in both CFLAGS and AFLAGS. Remove one > last instance of .eh_frame by removing the needless Call Frame Information > annotations from arch/arm64/kernel/smccc-call.S. > > Add .plt, .data.rel.ro, .igot.*, and .iplt to discards as they are not > actually used. While .got.plt is also not used, it must be included > otherwise ld.bfd will fail to link with the error: > > aarch64-linux-gnu-ld: discarded output section: `.got.plt' > > However, as it'd be better to validate that it stays effectively empty, > add an assert. > > Explicitly include debug sections when they're present. > > Fix a case of needless quotes in __section(), which Clang doesn't like. > > Finally, enable orphan section warnings. > > Thanks to Ard Biesheuvel for many hints on correct ways to handle > mysterious sections. :) Sorry to be a pain, but this patch is doing 3 or 4 independent things at once. Please could you split it up a bit? e.g. - Removal of cfi directives from smccc macro - Removal of quotes around section name for clang - Avoid generating .eh_frame - Ensure all sections are accounted for in linker script and warn on orphans That way it's a bit easier to manage, we can revert/backport bits later if necessary and you get more patches in the kernel ;) You can also add my Ack on all the patches: Acked-by: Will Deacon <will@kernel.org> Will
On Wed, Jun 24, 2020 at 08:57:12AM +0100, Will Deacon wrote: > On Tue, Jun 23, 2020 at 06:49:40PM -0700, Kees Cook wrote: > > We don't want to depend on the linker's orphan section placement > > heuristics as these can vary between linkers, and may change between > > versions. All sections need to be explicitly named in the linker > > script. > > > > Avoid .eh_frame* by making sure both -fno-asychronous-unwind-tables and > > -fno-unwind-tables are present in both CFLAGS and AFLAGS. Remove one > > last instance of .eh_frame by removing the needless Call Frame Information > > annotations from arch/arm64/kernel/smccc-call.S. > > > > Add .plt, .data.rel.ro, .igot.*, and .iplt to discards as they are not > > actually used. While .got.plt is also not used, it must be included > > otherwise ld.bfd will fail to link with the error: > > > > aarch64-linux-gnu-ld: discarded output section: `.got.plt' > > > > However, as it'd be better to validate that it stays effectively empty, > > add an assert. > > > > Explicitly include debug sections when they're present. > > > > Fix a case of needless quotes in __section(), which Clang doesn't like. > > > > Finally, enable orphan section warnings. > > > > Thanks to Ard Biesheuvel for many hints on correct ways to handle > > mysterious sections. :) > > Sorry to be a pain, but this patch is doing 3 or 4 independent things at > once. Please could you split it up a bit? > e.g. > > - Removal of cfi directives from smccc macro > - Removal of quotes around section name for clang > - Avoid generating .eh_frame > - Ensure all sections are accounted for in linker script and warn on orphans > > That way it's a bit easier to manage, we can revert/backport bits later if > necessary and you get more patches in the kernel ;) Yeah, this one patch did grow a bit. ;) I've split it up now. > You can also add my Ack on all the patches: > > Acked-by: Will Deacon <will@kernel.org> Thanks!
diff --git a/arch/arm64/Makefile b/arch/arm64/Makefile index a0d94d063fa8..fb3aa2d7de4d 100644 --- a/arch/arm64/Makefile +++ b/arch/arm64/Makefile @@ -29,6 +29,10 @@ LDFLAGS_vmlinux += --fix-cortex-a53-843419 endif endif +# We never want expected sections to be placed heuristically by the +# linker. All sections should be explicitly named in the linker script. +LDFLAGS_vmlinux += --orphan-handling=warn + ifeq ($(CONFIG_ARM64_USE_LSE_ATOMICS), y) ifneq ($(CONFIG_ARM64_LSE_ATOMICS), y) $(warning LSE atomics not supported by binutils) @@ -47,13 +51,16 @@ endif KBUILD_CFLAGS += -mgeneral-regs-only \ $(compat_vdso) $(cc_has_k_constraint) -KBUILD_CFLAGS += -fno-asynchronous-unwind-tables KBUILD_CFLAGS += $(call cc-disable-warning, psabi) KBUILD_AFLAGS += $(compat_vdso) KBUILD_CFLAGS += $(call cc-option,-mabi=lp64) KBUILD_AFLAGS += $(call cc-option,-mabi=lp64) +# Avoid generating .eh_frame* sections. +KBUILD_CFLAGS += -fno-asynchronous-unwind-tables -fno-unwind-tables +KBUILD_AFLAGS += -fno-asynchronous-unwind-tables -fno-unwind-tables + ifeq ($(CONFIG_STACKPROTECTOR_PER_TASK),y) prepare: stack_protector_prepare stack_protector_prepare: prepare0 diff --git a/arch/arm64/kernel/smccc-call.S b/arch/arm64/kernel/smccc-call.S index 1f93809528a4..d62447964ed9 100644 --- a/arch/arm64/kernel/smccc-call.S +++ b/arch/arm64/kernel/smccc-call.S @@ -9,7 +9,6 @@ #include <asm/assembler.h> .macro SMCCC instr - .cfi_startproc \instr #0 ldr x4, [sp] stp x0, x1, [x4, #ARM_SMCCC_RES_X0_OFFS] @@ -21,7 +20,6 @@ b.ne 1f str x6, [x4, ARM_SMCCC_QUIRK_STATE_OFFS] 1: ret - .cfi_endproc .endm /* diff --git a/arch/arm64/kernel/vmlinux.lds.S b/arch/arm64/kernel/vmlinux.lds.S index 5427f502c3a6..f6c781768f83 100644 --- a/arch/arm64/kernel/vmlinux.lds.S +++ b/arch/arm64/kernel/vmlinux.lds.S @@ -94,7 +94,8 @@ SECTIONS /DISCARD/ : { *(.interp .dynamic) *(.dynsym .dynstr .hash .gnu.hash) - *(.eh_frame) + *(.plt) *(.data.rel.ro) + *(.igot.*) *(.iplt) } . = KIMAGE_VADDR + TEXT_OFFSET; @@ -244,8 +245,16 @@ SECTIONS _end = .; STABS_DEBUG + DWARF_DEBUG HEAD_SYMBOLS + + /* + * Make sure that the .got.plt is either completely empty or it + * contains only the lazy dispatch entries. + */ + .got.plt (INFO) : { *(.got.plt) } + ASSERT(SIZEOF(.got.plt) == 0 || SIZEOF(.got.plt) == 0x18, ".got.plt not empty") } #include "image-vars.h" diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index 1df25f26571d..dce024ea6084 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -42,7 +42,7 @@ u64 idmap_t0sz = TCR_T0SZ(VA_BITS); u64 idmap_ptrs_per_pgd = PTRS_PER_PGD; -u64 __section(".mmuoff.data.write") vabits_actual; +u64 __section(.mmuoff.data.write) vabits_actual; EXPORT_SYMBOL(vabits_actual); u64 kimage_voffset __ro_after_init;
We don't want to depend on the linker's orphan section placement heuristics as these can vary between linkers, and may change between versions. All sections need to be explicitly named in the linker script. Avoid .eh_frame* by making sure both -fno-asychronous-unwind-tables and -fno-unwind-tables are present in both CFLAGS and AFLAGS. Remove one last instance of .eh_frame by removing the needless Call Frame Information annotations from arch/arm64/kernel/smccc-call.S. Add .plt, .data.rel.ro, .igot.*, and .iplt to discards as they are not actually used. While .got.plt is also not used, it must be included otherwise ld.bfd will fail to link with the error: aarch64-linux-gnu-ld: discarded output section: `.got.plt' However, as it'd be better to validate that it stays effectively empty, add an assert. Explicitly include debug sections when they're present. Fix a case of needless quotes in __section(), which Clang doesn't like. Finally, enable orphan section warnings. Thanks to Ard Biesheuvel for many hints on correct ways to handle mysterious sections. :) Signed-off-by: Kees Cook <keescook@chromium.org> --- arch/arm64/Makefile | 9 ++++++++- arch/arm64/kernel/smccc-call.S | 2 -- arch/arm64/kernel/vmlinux.lds.S | 11 ++++++++++- arch/arm64/mm/mmu.c | 2 +- 4 files changed, 19 insertions(+), 5 deletions(-)