efi/libstub/arm64: link stub lib.a conditionally
PATCH v2 1/2 efi/libstub/arm64: link stub lib.a conditionally
Since commit 799c43415442 ("kbuild: thin archives make default for all archs"), core-y is passed to the linker with --whole-archive. Hence, the whole of stub library is linked to vmlinux.
Use libs-y so that lib.a is passed after --no-whole-archive for conditional linking.
The unused drivers/firmware/efi/libstub/relocate.o will be dropped for ARCH=arm64.
---
This patch touches under arch/arm64/, but
this is more related to efi.
I am sending this to Ard.
arch/arm64/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm64/Makefile +1 -1
@@ -145,7 +145,7 @@ export TEXT_OFFSET core-y += arch/arm64/ libs-y := arch/arm64/lib/ $(libs-y) -core-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a +libs-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a # Default target when executing plain make boot := arch/arm64/boot
Signature · 7 lines
2.25.1 _______________________________________________ linux-arm-kernel mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH v2 2/2 efi/libstub: refactor Makefile to not use lib-y syntax
Documentation/kbuild/makefiles.rst says:
Use of lib-y is normally restricted to lib/ and arch/*/lib.
This is because lib-y is inteded to be hooked to KBUILD_VMLINUX_LIBS, which is passed down to scripts/link-vmlinux.sh.
Besides, lib-y is not so interesting because objects from lib-y are mostly linked in normal usecases. For example, lib-y only saves 364 bytes for x86_64_defconfig. You can see the details in commit 7273ad2b08f8 ("kbuild: link lib-y objects to vmlinux forcibly when CONFIG_MODULES=y").
I think we should consider to deprecate lib-y syntax at some point because we should aim for better solution like dead code elimination or LTO.
Other than lib/ and arch/*/lib, this Makefile is the only user of lib-y. Replace lib-y with a custom rule.
---
Changes in v2:
- Add more description
drivers/firmware/efi/libstub/Makefile | 49 +++++++++++++++------------ 1 file changed, 27 insertions(+), 22 deletions(-)
drivers/firmware/efi/libstub/Makefile +27 -22
@@ -44,7 +44,7 @@ OBJECT_FILES_NON_STANDARD := y # Prevents link failures: __sanitizer_cov_trace_pc() is not linked in. KCOV_INSTRUMENT := n -lib-y := efi-stub-helper.o gop.o secureboot.o tpm.o \ +stub-obj-y := efi-stub-helper.o gop.o secureboot.o tpm.o \ file.o mem.o random.o randomalloc.o pci.o \ skip_spaces.o lib-cmdline.o lib-ctype.o \ alignedmem.o relocate.o vsprintf.o
@@ -55,15 +55,19 @@ efi-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c $(obj)/lib-%.o: $(srctree)/lib/%.c FORCE $(call if_changed_rule,cc_o_c) -lib-$(CONFIG_EFI_GENERIC_STUB) += efi-stub.o fdt.o string.o \ +stub-obj-$(CONFIG_EFI_GENERIC_STUB) += efi-stub.o fdt.o string.o \ $(patsubst %.c,lib-%.o,$(efi-deps-y)) -lib-$(CONFIG_ARM) += arm32-stub.o -lib-$(CONFIG_ARM64) += arm64-stub.o -lib-$(CONFIG_X86) += x86-stub.o +stub-obj-$(CONFIG_ARM) += arm32-stub.o +stub-obj-$(CONFIG_ARM64) += arm64-stub.o +stub-obj-$(CONFIG_X86) += x86-stub.o CFLAGS_arm32-stub.o := -DTEXT_OFFSET=$(TEXT_OFFSET) CFLAGS_arm64-stub.o := -DTEXT_OFFSET=$(TEXT_OFFSET) +targets += $(stub-obj-y) +stub-obj-y := $(patsubst %.o,%.stub.o, $(stub-obj-y)) +targets += $(stub-obj-y) + # # For x86, bootloaders like systemd-boot or grub-efi do not zero-initialize the # .bss section, so the .bss section of the EFI stub needs to be included in the
@@ -83,23 +87,6 @@ STUBCOPY_FLAGS-$(CONFIG_ARM) += --rename-section .data=.data.efistub \ --rename-section .bss=.bss.efistub,load,alloc STUBCOPY_RELOC-$(CONFIG_ARM) := R_ARM_ABS -# -# arm64 puts the stub in the kernel proper, which will unnecessarily retain all -# code indefinitely unless it is annotated as __init/__initdata/__initconst etc. -# So let's apply the __init annotations at the section level, by prefixing -# the section names directly. This will ensure that even all the inline string -# literals are covered. -# The fact that the stub and the kernel proper are essentially the same binary -# also means that we need to be extra careful to make sure that the stub does -# not rely on any absolute symbol references, considering that the virtual -# kernel mapping that the linker uses is not active yet when the stub is -# executing. So build all C dependencies of the EFI stub into libstub, and do -# a verification pass to see if any absolute relocations exist in any of the -# object files. -# -extra-y := $(lib-y) -lib-y := $(patsubst %.o,%.stub.o,$(lib-y)) - STUBCOPY_FLAGS-$(CONFIG_ARM64) += --prefix-alloc-sections=.init \ --prefix-symbols=__efistub_ STUBCOPY_RELOC-$(CONFIG_ARM64) := R_AARCH64_ABS
@@ -121,3 +108,21 @@ quiet_cmd_stubcopy = STUBCPY $@ /bin/false; \ fi; \ $(OBJCOPY) $(STUBCOPY_FLAGS-y) $< $@ + +# arm64 puts the stub in the kernel proper, which will unnecessarily retain all +# code indefinitely unless it is annotated as __init/__initdata/__initconst etc. +# So let's apply the __init annotations at the section level, by prefixing +# the section names directly. This will ensure that even all the inline string +# literals are covered. +# The fact that the stub and the kernel proper are essentially the same binary +# also means that we need to be extra careful to make sure that the stub does +# not rely on any absolute symbol references, considering that the virtual +# kernel mapping that the linker uses is not active yet when the stub is +# executing. So build all C dependencies of the EFI stub into libstub, and do +# a verification pass to see if any absolute relocations exist in any of the +# object files. +# +extra-y += lib.a + +$(obj)/lib.a: $(addprefix $(obj)/, $(stub-obj-y)) FORCE + $(call if_changed,ar)
Signature · 1 line
2.25.1
Will Deacon
maintainerQuoted 10 lines · Masahiro Yamada
Since commit
799c43415442("kbuild: thin archives make default for all archs"), core-y is passed to the linker with --whole-archive. Hence, the whole of stub library is linked to vmlinux.Use libs-y so that lib.a is passed after --no-whole-archive for conditional linking.
The unused
drivers/firmware/efi/libstub/relocate.owill be dropped for ARCH=arm64.
Quoted 6 lines
Signed-off-by Masahiro Yamada <[email protected]>---
This patch touches under
arch/arm64/, but
this is more related to efi.
I am sending this to Ard.
Ok, I'll ignore this then. Ard -- please yell if you want me to do anything else with it.
Will
Quoted 16 lines
arch/arm64/Makefile +1
-1@@ -145,7 +145,7 @@ export TEXT_OFFSET core-y += arch/arm64/ libs-y := arch/arm64/lib/ $(libs-y) -core-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a +libs-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a # Default target when executing plain make boot := arch/arm64/bootSignature · 1 line
2.25.1
Ard Biesheuvel
maintainerQuoted 15 lines · Will Deacon
Quoted 13 lines · Masahiro Yamada
Since commit
799c43415442("kbuild: thin archives make default for all archs"), core-y is passed to the linker with --whole-archive. Hence, the whole of stub library is linked to vmlinux.Use libs-y so that lib.a is passed after --no-whole-archive for conditional linking.
The unused
drivers/firmware/efi/libstub/relocate.owill be dropped for ARCH=arm64.Signed-off-by Masahiro Yamada <[email protected]>---
Quoted 7 lines
Quoted 3 lines
This patch touches under
arch/arm64/, but
this is more related to efi.
I am sending this to Ard.Ok, I'll ignore this then. Ard -- please yell if you want me to do anything else with it.
I am going to send a batch of EFI fixes early next week, so I can take this as well, or alternatively, you can apply it directly.
Quoted 17 lines
Quoted 17 lines
arch/arm64/Makefile +1
-2@@ -145,7 +145,7 @@ export TEXT_OFFSET core-y += arch/arm64/ libs-y := arch/arm64/lib/ $(libs-y) -core-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a +libs-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a # Default target when executing plain make boot := arch/arm64/boot --2.25.1
Will Deacon
maintainerQuoted 20 lines · Ard Biesheuvel
Quoted 19 lines · Will Deacon
Quoted 16 lines · Masahiro Yamada
Since commit
799c43415442("kbuild: thin archives make default for all archs"), core-y is passed to the linker with --whole-archive. Hence, the whole of stub library is linked to vmlinux.Use libs-y so that lib.a is passed after --no-whole-archive for conditional linking.
The unused
drivers/firmware/efi/libstub/relocate.owill be dropped for ARCH=arm64.Signed-off-by Masahiro Yamada <[email protected]>---
This patch touches under
arch/arm64/, but
this is more related to efi.
I am sending this to Ard.
Quoted 6 lines
Quoted 3 lines
Ok, I'll ignore this then. Ard -- please yell if you want me to do anything else with it.
I am going to send a batch of EFI fixes early next week, so I can take this as well, or alternatively, you can apply it directly.
No, please take it. Just wanted to make sure it didn't slip through the cracks, that's all.
Will
Ard Biesheuvel
maintainerRe: [PATCH v2 2/2] efi/libstub: refactor Makefile to not use lib-y syntax
In reply to Masahiro YamadaQuoted 14 lines · Masahiro Yamada
Documentation/kbuild/makefiles.rstsays:Use of lib-y is normally restricted to
lib/andarch/*/lib.This is because lib-y is inteded to be hooked to
KBUILD_VMLINUX_LIBS, which is passed down toscripts/link-vmlinux.sh.Besides, lib-y is not so interesting because objects from lib-y are mostly linked in normal usecases. For example, lib-y only saves 364 bytes for
x86_64_defconfig. You can see the details in commit7273ad2b08f8("kbuild: link lib-y objects to vmlinux forcibly whenCONFIG_MODULES=y").
Quoted 8 lines
I think we should consider to deprecate lib-y syntax at some point because we should aim for better solution like dead code elimination or LTO.
Other than
lib/andarch/*/lib, thisMakefileis the only user of lib-y. Replace lib-y with a custom rule.Signed-off-by Masahiro Yamada <[email protected]>
Series queued in efi/urgent.
Thanks
Quoted 94 lines
---
Changes in v2:
- Add more descriptiondrivers/firmware/efi/libstub/Makefile | 49 +++++++++++++++------------ 1 file changed, 27 insertions(+), 22 deletions(-)drivers/firmware/efi/libstub/Makefile +27
-23@@ -44,7 +44,7 @@ OBJECT_FILES_NON_STANDARD := y # Prevents link failures: __sanitizer_cov_trace_pc() is not linked in. KCOV_INSTRUMENT := n -lib-y := efi-stub-helper.o gop.o secureboot.o tpm.o \ +stub-obj-y := efi-stub-helper.o gop.o secureboot.o tpm.o \ file.o mem.o random.o randomalloc.o pci.o \ skip_spaces.o lib-cmdline.o lib-ctype.o \ alignedmem.o relocate.o vsprintf.o@@ -55,15 +55,19 @@ efi-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c $(obj)/lib-%.o: $(srctree)/lib/%.c FORCE $(call if_changed_rule,cc_o_c) -lib-$(CONFIG_EFI_GENERIC_STUB) += efi-stub.o fdt.o string.o \ +stub-obj-$(CONFIG_EFI_GENERIC_STUB) += efi-stub.o fdt.o string.o \ $(patsubst %.c,lib-%.o,$(efi-deps-y)) -lib-$(CONFIG_ARM) += arm32-stub.o -lib-$(CONFIG_ARM64) += arm64-stub.o -lib-$(CONFIG_X86) += x86-stub.o +stub-obj-$(CONFIG_ARM) += arm32-stub.o +stub-obj-$(CONFIG_ARM64) += arm64-stub.o +stub-obj-$(CONFIG_X86) += x86-stub.o CFLAGS_arm32-stub.o := -DTEXT_OFFSET=$(TEXT_OFFSET) CFLAGS_arm64-stub.o := -DTEXT_OFFSET=$(TEXT_OFFSET) +targets += $(stub-obj-y) +stub-obj-y := $(patsubst %.o,%.stub.o, $(stub-obj-y)) +targets += $(stub-obj-y) + # # For x86, bootloaders like systemd-boot or grub-efi do not zero-initialize the # .bss section, so the .bss section of the EFI stub needs to be included in the@@ -83,23 +87,6 @@ STUBCOPY_FLAGS-$(CONFIG_ARM) += --rename-section .data=.data.efistub \ --rename-section .bss=.bss.efistub,load,alloc STUBCOPY_RELOC-$(CONFIG_ARM) := R_ARM_ABS -# -# arm64 puts the stub in the kernel proper, which will unnecessarily retain all -# code indefinitely unless it is annotated as __init/__initdata/__initconst etc. -# So let's apply the __init annotations at the section level, by prefixing -# the section names directly. This will ensure that even all the inline string -# literals are covered. -# The fact that the stub and the kernel proper are essentially the same binary -# also means that we need to be extra careful to make sure that the stub does -# not rely on any absolute symbol references, considering that the virtual -# kernel mapping that the linker uses is not active yet when the stub is -# executing. So build all C dependencies of the EFI stub into libstub, and do -# a verification pass to see if any absolute relocations exist in any of the -# object files. -# -extra-y := $(lib-y) -lib-y := $(patsubst %.o,%.stub.o,$(lib-y)) - STUBCOPY_FLAGS-$(CONFIG_ARM64) += --prefix-alloc-sections=.init \ --prefix-symbols=__efistub_ STUBCOPY_RELOC-$(CONFIG_ARM64) := R_AARCH64_ABS@@ -121,3 +108,21 @@ quiet_cmd_stubcopy = STUBCPY $@ /bin/false; \ fi; \ $(OBJCOPY) $(STUBCOPY_FLAGS-y) $< $@ + +# arm64 puts the stub in the kernel proper, which will unnecessarily retain all +# code indefinitely unless it is annotated as __init/__initdata/__initconst etc. +# So let's apply the __init annotations at the section level, by prefixing +# the section names directly. This will ensure that even all the inline string +# literals are covered. +# The fact that the stub and the kernel proper are essentially the same binary +# also means that we need to be extra careful to make sure that the stub does +# not rely on any absolute symbol references, considering that the virtual +# kernel mapping that the linker uses is not active yet when the stub is +# executing. So build all C dependencies of the EFI stub into libstub, and do +# a verification pass to see if any absolute relocations exist in any of the +# object files. +# +extra-y += lib.a + +$(obj)/lib.a: $(addprefix $(obj)/, $(stub-obj-y)) FORCE + $(call if_changed,ar) --2.25.1
Ard Biesheuvel
maintainerRe: [PATCH v2 2/2] efi/libstub: refactor Makefile to not use lib-y syntax
Quoted 20 lines · Ard Biesheuvel
Quoted 18 lines · Masahiro Yamada
Documentation/kbuild/makefiles.rstsays:Use of lib-y is normally restricted to
lib/andarch/*/lib.This is because lib-y is inteded to be hooked to
KBUILD_VMLINUX_LIBS, which is passed down toscripts/link-vmlinux.sh.Besides, lib-y is not so interesting because objects from lib-y are mostly linked in normal usecases. For example, lib-y only saves 364 bytes for
x86_64_defconfig. You can see the details in commit7273ad2b08f8("kbuild: link lib-y objects to vmlinux forcibly whenCONFIG_MODULES=y").I think we should consider to deprecate lib-y syntax at some point because we should aim for better solution like dead code elimination or LTO.
Quoted 7 lines
Quoted 4 lines
Other than
lib/andarch/*/lib, thisMakefileis the only user of lib-y. Replace lib-y with a custom rule.Signed-off-by Masahiro Yamada <[email protected]>Series queued in efi/urgent.
Actually, I am seeing this now
/home/ard/linux/scripts/Makefile.build:421: warning: overriding recipe for target 'drivers/firmware/efi/libstub/lib.a' /home/ard/linux/drivers/firmware/efi/libstub/Makefile:131: warning:
ignoring old recipe for target 'drivers/firmware/efi/libstub/lib.a'
Re: [PATCH v2 2/2] efi/libstub: refactor Makefile to not use lib-y syntax
Quoted 30 lines · Ard Biesheuvel
Quoted 20 lines · Ard Biesheuvel
Quoted 18 lines · Masahiro Yamada
Documentation/kbuild/makefiles.rstsays:Use of lib-y is normally restricted to
lib/andarch/*/lib.This is because lib-y is inteded to be hooked to
KBUILD_VMLINUX_LIBS, which is passed down toscripts/link-vmlinux.sh.Besides, lib-y is not so interesting because objects from lib-y are mostly linked in normal usecases. For example, lib-y only saves 364 bytes for
x86_64_defconfig. You can see the details in commit7273ad2b08f8("kbuild: link lib-y objects to vmlinux forcibly whenCONFIG_MODULES=y").I think we should consider to deprecate lib-y syntax at some point because we should aim for better solution like dead code elimination or LTO.
Quoted 7 lines
Quoted 4 lines
Other than
lib/andarch/*/lib, thisMakefileis the only user of lib-y. Replace lib-y with a custom rule.Signed-off-by Masahiro Yamada <[email protected]>Series queued in efi/urgent.
Quoted 6 lines
Actually, I am seeing this now
/home/ard/linux/scripts/Makefile.build:421: warning: overriding recipe for target 'drivers/firmware/efi/libstub/lib.a'/home/ard/linux/drivers/firmware/efi/libstub/Makefile:131: warning:
ignoring old recipe for target 'drivers/firmware/efi/libstub/lib.a'
Sorry, please drop this patch for now.
(Actually, it was working when I submitted it,
but is broken after e578edc72276280b8fae57f6bf79cb443ceee7a2)
Signature · 2 lines
Best Regards Masahiro Yamada
Ard Biesheuvel
maintainerRe: [PATCH v2 2/2] efi/libstub: refactor Makefile to not use lib-y syntax
Quoted 41 lines · Masahiro Yamada
Quoted 30 lines · Ard Biesheuvel
Quoted 20 lines · Ard Biesheuvel
Quoted 18 lines · Masahiro Yamada
Documentation/kbuild/makefiles.rstsays:Use of lib-y is normally restricted to
lib/andarch/*/lib.This is because lib-y is inteded to be hooked to
KBUILD_VMLINUX_LIBS, which is passed down toscripts/link-vmlinux.sh.Besides, lib-y is not so interesting because objects from lib-y are mostly linked in normal usecases. For example, lib-y only saves 364 bytes for
x86_64_defconfig. You can see the details in commit7273ad2b08f8("kbuild: link lib-y objects to vmlinux forcibly whenCONFIG_MODULES=y").I think we should consider to deprecate lib-y syntax at some point because we should aim for better solution like dead code elimination or LTO.
Quoted 7 lines
Quoted 4 lines
Other than
lib/andarch/*/lib, thisMakefileis the only user of lib-y. Replace lib-y with a custom rule.Signed-off-by Masahiro Yamada <[email protected]>Series queued in efi/urgent.
Quoted 6 lines
Actually, I am seeing this now
/home/ard/linux/scripts/Makefile.build:421: warning: overriding recipe for target 'drivers/firmware/efi/libstub/lib.a'/home/ard/linux/drivers/firmware/efi/libstub/Makefile:131: warning:
ignoring old recipe for target 'drivers/firmware/efi/libstub/lib.a'
Quoted 5 lines
Sorry, please drop this patch for now.
(Actually, it was working when I submitted it,
but is broken aftere578edc72276280b8fae57f6bf79cb443ceee7a2)
OK.
The following commit has been merged into the efi/urgent branch of tip:
The following commit has been merged into the efi/urgent branch of tip:
Commit-ID: c1aac64ddc01112e137121a43645b96c3633c41b Gitweb: https://git.kernel.org/tip/c1aac64ddc01112e137121a43645b96c3633c41b Author: Masahiro Yamada <[email protected]> AuthorDate: Thu, 04 Jun 2020 11:20:30 +09:00 Committer: Ard Biesheuvel <[email protected]> CommitterDate: Thu, 09 Jul 2020 09:45:09 +03:00
efi/libstub/arm64: link stub lib.a conditionally
Since commit 799c43415442 ("kbuild: thin archives make default for all archs"), core-y is passed to the linker with --whole-archive. Hence, the whole of stub library is linked to vmlinux.
Use libs-y so that lib.a is passed after --no-whole-archive for conditional linking.
The unused drivers/firmware/efi/libstub/relocate.o will be dropped for ARCH=arm64.
--- arch/arm64/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm64/Makefile +1 -1
@@ -143,7 +143,7 @@ export TEXT_OFFSET core-y += arch/arm64/ libs-y := arch/arm64/lib/ $(libs-y) -core-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a +libs-$(CONFIG_EFI_STUB) += $(objtree)/drivers/firmware/efi/libstub/lib.a # Default target when executing plain make boot := arch/arm64/boot