[PATCH] arm64/efi: move arm64 specific stub C code to libstub

[PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Ard Biesheuvel]

Now that we added special handling to the C files in libstub, move
the one remaining arm64 specific EFI stub C file to libstub as
well, so that it gets the same treatment. This should prevent future
changes from resulting in binaries that may execute incorrectly in
UEFI context.

With efi-entry.S the only remaining EFI stub source file under
arch/arm64, we can also simplify the Makefile logic somewhat.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---

I would like to suggest that this be taken on top of the stuff that is
queued for 4.4 at the moment (if it is not too late already).

 arch/arm64/kernel/Makefile                                                | 9 ++-------
 drivers/firmware/efi/libstub/Makefile                                     | 3 +++
 arch/arm64/kernel/efi-stub.c => drivers/firmware/efi/libstub/arm64-stub.c | 0
 3 files changed, 5 insertions(+), 7 deletions(-)

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index 1b6bda2ff102..474691f8b13a 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -4,11 +4,8 @@
 
 CPPFLAGS_vmlinux.lds	:= -DTEXT_OFFSET=$(TEXT_OFFSET)
 AFLAGS_head.o		:= -DTEXT_OFFSET=$(TEXT_OFFSET)
-CFLAGS_efi-stub.o 	:= -DTEXT_OFFSET=$(TEXT_OFFSET)
 CFLAGS_armv8_deprecated.o := -I$(src)
 
-KASAN_SANITIZE_efi-stub.o	:= n
-
 CFLAGS_REMOVE_ftrace.o = -pg
 CFLAGS_REMOVE_insn.o = -pg
 CFLAGS_REMOVE_return_address.o = -pg
@@ -22,9 +19,7 @@ arm64-obj-y		:= debug-monitors.o entry.o irq.o fpsimd.o		\
 			   cpufeature.o alternative.o cacheinfo.o		\
 			   smp.o smp_spin_table.o topology.o
 
-stub-obj				:= efi-stub.o efi-entry.o
-extra-y					:= $(stub-obj)
-stub-obj				:= $(patsubst %.o,%.stub.o,$(stub-obj))
+extra-$(CONFIG_EFI)			:= efi-entry.o
 
 OBJCOPYFLAGS := --prefix-symbols=__efistub_
 $(obj)/%.stub.o: $(obj)/%.o FORCE
@@ -42,7 +37,7 @@ arm64-obj-$(CONFIG_CPU_PM)		+= sleep.o suspend.o
 arm64-obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
 arm64-obj-$(CONFIG_JUMP_LABEL)		+= jump_label.o
 arm64-obj-$(CONFIG_KGDB)		+= kgdb.o
-arm64-obj-$(CONFIG_EFI)			+= efi.o $(stub-obj)
+arm64-obj-$(CONFIG_EFI)			+= efi.o efi-entry.stub.o
 arm64-obj-$(CONFIG_PCI)			+= pci.o
 arm64-obj-$(CONFIG_ARMV8_DEPRECATED)	+= armv8_deprecated.o
 arm64-obj-$(CONFIG_ACPI)		+= acpi.o
diff --git a/drivers/firmware/efi/libstub/Makefile b/drivers/firmware/efi/libstub/Makefile
index bca9a76cbd33..92ae557abbbc 100644
--- a/drivers/firmware/efi/libstub/Makefile
+++ b/drivers/firmware/efi/libstub/Makefile
@@ -34,6 +34,9 @@ $(obj)/lib-%.o: $(srctree)/lib/%.c FORCE
 lib-$(CONFIG_EFI_ARMSTUB)	+= arm-stub.o fdt.o string.o \
 				   $(patsubst %.c,lib-%.o,$(arm-deps))
 
+lib-$(CONFIG_ARM64)		+= arm64-stub.o
+CFLAGS_arm64-stub.o 		:= -DTEXT_OFFSET=$(TEXT_OFFSET)
+
 #
 # arm64 puts the stub in the kernel proper, which will unnecessarily retain all
 # code indefinitely unless it is annotated as __init/__initdata/__initconst etc.
diff --git a/arch/arm64/kernel/efi-stub.c b/drivers/firmware/efi/libstub/arm64-stub.c
similarity index 100%
rename from arch/arm64/kernel/efi-stub.c
rename to drivers/firmware/efi/libstub/arm64-stub.c
-- 
2.1.4

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Matt Fleming]

On Fri, 23 Oct, at 04:48:14PM, Ard Biesheuvel wrote:
> Now that we added special handling to the C files in libstub, move
> the one remaining arm64 specific EFI stub C file to libstub as
> well, so that it gets the same treatment. This should prevent future
> changes from resulting in binaries that may execute incorrectly in
> UEFI context.
 
Sorry, I'm coming at this cache-cold: What special handling was added
to libstub that you want to make use of here?

> With efi-entry.S the only remaining EFI stub source file under
> arch/arm64, we can also simplify the Makefile logic somewhat.
> 
> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> ---
> 
> I would like to suggest that this be taken on top of the stuff that is
> queued for 4.4 at the moment (if it is not too late already).
 
That's upto the tip folks (Cc'd). Ingo, would you consider applying
a minimal build cleanup patch like this for v4.4?

>  arch/arm64/kernel/Makefile                                                | 9 ++-------
>  drivers/firmware/efi/libstub/Makefile                                     | 3 +++
>  arch/arm64/kernel/efi-stub.c => drivers/firmware/efi/libstub/arm64-stub.c | 0
>  3 files changed, 5 insertions(+), 7 deletions(-)
> 
> diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
> index 1b6bda2ff102..474691f8b13a 100644
> --- a/arch/arm64/kernel/Makefile
> +++ b/arch/arm64/kernel/Makefile
> @@ -4,11 +4,8 @@
>  
>  CPPFLAGS_vmlinux.lds	:= -DTEXT_OFFSET=$(TEXT_OFFSET)
>  AFLAGS_head.o		:= -DTEXT_OFFSET=$(TEXT_OFFSET)
> -CFLAGS_efi-stub.o 	:= -DTEXT_OFFSET=$(TEXT_OFFSET)
>  CFLAGS_armv8_deprecated.o := -I$(src)
>  
> -KASAN_SANITIZE_efi-stub.o	:= n
> -
>  CFLAGS_REMOVE_ftrace.o = -pg
>  CFLAGS_REMOVE_insn.o = -pg
>  CFLAGS_REMOVE_return_address.o = -pg
> @@ -22,9 +19,7 @@ arm64-obj-y		:= debug-monitors.o entry.o irq.o fpsimd.o		\
>  			   cpufeature.o alternative.o cacheinfo.o		\
>  			   smp.o smp_spin_table.o topology.o
>  
> -stub-obj				:= efi-stub.o efi-entry.o
> -extra-y					:= $(stub-obj)
> -stub-obj				:= $(patsubst %.o,%.stub.o,$(stub-obj))
> +extra-$(CONFIG_EFI)			:= efi-entry.o
>  
>  OBJCOPYFLAGS := --prefix-symbols=__efistub_
>  $(obj)/%.stub.o: $(obj)/%.o FORCE
> @@ -42,7 +37,7 @@ arm64-obj-$(CONFIG_CPU_PM)		+= sleep.o suspend.o
>  arm64-obj-$(CONFIG_CPU_IDLE)		+= cpuidle.o
>  arm64-obj-$(CONFIG_JUMP_LABEL)		+= jump_label.o
>  arm64-obj-$(CONFIG_KGDB)		+= kgdb.o
> -arm64-obj-$(CONFIG_EFI)			+= efi.o $(stub-obj)
> +arm64-obj-$(CONFIG_EFI)			+= efi.o efi-entry.stub.o
>  arm64-obj-$(CONFIG_PCI)			+= pci.o
>  arm64-obj-$(CONFIG_ARMV8_DEPRECATED)	+= armv8_deprecated.o
>  arm64-obj-$(CONFIG_ACPI)		+= acpi.o
> diff --git a/drivers/firmware/efi/libstub/Makefile b/drivers/firmware/efi/libstub/Makefile
> index bca9a76cbd33..92ae557abbbc 100644
> --- a/drivers/firmware/efi/libstub/Makefile
> +++ b/drivers/firmware/efi/libstub/Makefile
> @@ -34,6 +34,9 @@ $(obj)/lib-%.o: $(srctree)/lib/%.c FORCE
>  lib-$(CONFIG_EFI_ARMSTUB)	+= arm-stub.o fdt.o string.o \
>  				   $(patsubst %.c,lib-%.o,$(arm-deps))
>  
> +lib-$(CONFIG_ARM64)		+= arm64-stub.o
> +CFLAGS_arm64-stub.o 		:= -DTEXT_OFFSET=$(TEXT_OFFSET)
> +
>  #
>  # arm64 puts the stub in the kernel proper, which will unnecessarily retain all
>  # code indefinitely unless it is annotated as __init/__initdata/__initconst etc.
> diff --git a/arch/arm64/kernel/efi-stub.c b/drivers/firmware/efi/libstub/arm64-stub.c
> similarity index 100%
> rename from arch/arm64/kernel/efi-stub.c
> rename to drivers/firmware/efi/libstub/arm64-stub.c
> -- 
> 2.1.4
> 

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Ard Biesheuvel]

On 27 October 2015 at 06:11, Matt Fleming <matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@public.gmane.org> wrote:
> On Fri, 23 Oct, at 04:48:14PM, Ard Biesheuvel wrote:
>> Now that we added special handling to the C files in libstub, move
>> the one remaining arm64 specific EFI stub C file to libstub as
>> well, so that it gets the same treatment. This should prevent future
>> changes from resulting in binaries that may execute incorrectly in
>> UEFI context.
>
> Sorry, I'm coming at this cache-cold: What special handling was added
> to libstub that you want to make use of here?
>

This is about the libstub patches that you reviewed ~2 weeks ago and
that have been queued in Catalin's arm64 tree for 4.4

>> With efi-entry.S the only remaining EFI stub source file under
>> arch/arm64, we can also simplify the Makefile logic somewhat.
>>
>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>> ---
>>
>> I would like to suggest that this be taken on top of the stuff that is
>> queued for 4.4 at the moment (if it is not too late already).
>
> That's upto the tip folks (Cc'd). Ingo, would you consider applying
> a minimal build cleanup patch like this for v4.4?
>

Actually, the suggestion was aimed at Catalin, who has all the libstub
and KASAN patches queued that this patch depends on.
Apologies for the confusion.

Thanks,
Ard.


>>  arch/arm64/kernel/Makefile                                                | 9 ++-------
>>  drivers/firmware/efi/libstub/Makefile                                     | 3 +++
>>  arch/arm64/kernel/efi-stub.c => drivers/firmware/efi/libstub/arm64-stub.c | 0
>>  3 files changed, 5 insertions(+), 7 deletions(-)
>>
>> diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
>> index 1b6bda2ff102..474691f8b13a 100644
>> --- a/arch/arm64/kernel/Makefile
>> +++ b/arch/arm64/kernel/Makefile
>> @@ -4,11 +4,8 @@
>>
>>  CPPFLAGS_vmlinux.lds := -DTEXT_OFFSET=$(TEXT_OFFSET)
>>  AFLAGS_head.o                := -DTEXT_OFFSET=$(TEXT_OFFSET)
>> -CFLAGS_efi-stub.o    := -DTEXT_OFFSET=$(TEXT_OFFSET)
>>  CFLAGS_armv8_deprecated.o := -I$(src)
>>
>> -KASAN_SANITIZE_efi-stub.o    := n
>> -
>>  CFLAGS_REMOVE_ftrace.o = -pg
>>  CFLAGS_REMOVE_insn.o = -pg
>>  CFLAGS_REMOVE_return_address.o = -pg
>> @@ -22,9 +19,7 @@ arm64-obj-y         := debug-monitors.o entry.o irq.o fpsimd.o              \
>>                          cpufeature.o alternative.o cacheinfo.o               \
>>                          smp.o smp_spin_table.o topology.o
>>
>> -stub-obj                             := efi-stub.o efi-entry.o
>> -extra-y                                      := $(stub-obj)
>> -stub-obj                             := $(patsubst %.o,%.stub.o,$(stub-obj))
>> +extra-$(CONFIG_EFI)                  := efi-entry.o
>>
>>  OBJCOPYFLAGS := --prefix-symbols=__efistub_
>>  $(obj)/%.stub.o: $(obj)/%.o FORCE
>> @@ -42,7 +37,7 @@ arm64-obj-$(CONFIG_CPU_PM)          += sleep.o suspend.o
>>  arm64-obj-$(CONFIG_CPU_IDLE)         += cpuidle.o
>>  arm64-obj-$(CONFIG_JUMP_LABEL)               += jump_label.o
>>  arm64-obj-$(CONFIG_KGDB)             += kgdb.o
>> -arm64-obj-$(CONFIG_EFI)                      += efi.o $(stub-obj)
>> +arm64-obj-$(CONFIG_EFI)                      += efi.o efi-entry.stub.o
>>  arm64-obj-$(CONFIG_PCI)                      += pci.o
>>  arm64-obj-$(CONFIG_ARMV8_DEPRECATED) += armv8_deprecated.o
>>  arm64-obj-$(CONFIG_ACPI)             += acpi.o
>> diff --git a/drivers/firmware/efi/libstub/Makefile b/drivers/firmware/efi/libstub/Makefile
>> index bca9a76cbd33..92ae557abbbc 100644
>> --- a/drivers/firmware/efi/libstub/Makefile
>> +++ b/drivers/firmware/efi/libstub/Makefile
>> @@ -34,6 +34,9 @@ $(obj)/lib-%.o: $(srctree)/lib/%.c FORCE
>>  lib-$(CONFIG_EFI_ARMSTUB)    += arm-stub.o fdt.o string.o \
>>                                  $(patsubst %.c,lib-%.o,$(arm-deps))
>>
>> +lib-$(CONFIG_ARM64)          += arm64-stub.o
>> +CFLAGS_arm64-stub.o          := -DTEXT_OFFSET=$(TEXT_OFFSET)
>> +
>>  #
>>  # arm64 puts the stub in the kernel proper, which will unnecessarily retain all
>>  # code indefinitely unless it is annotated as __init/__initdata/__initconst etc.
>> diff --git a/arch/arm64/kernel/efi-stub.c b/drivers/firmware/efi/libstub/arm64-stub.c
>> similarity index 100%
>> rename from arch/arm64/kernel/efi-stub.c
>> rename to drivers/firmware/efi/libstub/arm64-stub.c
>> --
>> 2.1.4
>>

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Matt Fleming]

On Tue, 27 Oct, at 11:10:43AM, Ard Biesheuvel wrote:
> On 27 October 2015 at 06:11, Matt Fleming <matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@public.gmane.org> wrote:
> > On Fri, 23 Oct, at 04:48:14PM, Ard Biesheuvel wrote:
> >> Now that we added special handling to the C files in libstub, move
> >> the one remaining arm64 specific EFI stub C file to libstub as
> >> well, so that it gets the same treatment. This should prevent future
> >> changes from resulting in binaries that may execute incorrectly in
> >> UEFI context.
> >
> > Sorry, I'm coming at this cache-cold: What special handling was added
> > to libstub that you want to make use of here?
> >
> 
> This is about the libstub patches that you reviewed ~2 weeks ago and
> that have been queued in Catalin's arm64 tree for 4.4
> 
> >> With efi-entry.S the only remaining EFI stub source file under
> >> arch/arm64, we can also simplify the Makefile logic somewhat.
> >>
> >> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> >> ---
> >>
> >> I would like to suggest that this be taken on top of the stuff that is
> >> queued for 4.4 at the moment (if it is not too late already).
> >
> > That's upto the tip folks (Cc'd). Ingo, would you consider applying
> > a minimal build cleanup patch like this for v4.4?
> >
> 
> Actually, the suggestion was aimed at Catalin, who has all the libstub
> and KASAN patches queued that this patch depends on.
> Apologies for the confusion.

Fair enough. In that case (for the drivers/firmware bits),

Reviewed-by: Matt Fleming <matt-mF/unelCI9GS6iBeEJttW/XRex20P6io@public.gmane.org>

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Catalin Marinas]

On Fri, Oct 23, 2015 at 04:48:14PM +0200, Ard Biesheuvel wrote:
> Now that we added special handling to the C files in libstub, move
> the one remaining arm64 specific EFI stub C file to libstub as
> well, so that it gets the same treatment. This should prevent future
> changes from resulting in binaries that may execute incorrectly in
> UEFI context.
> 
> With efi-entry.S the only remaining EFI stub source file under
> arch/arm64, we can also simplify the Makefile logic somewhat.
> 
> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> ---
> 
> I would like to suggest that this be taken on top of the stuff that is
> queued for 4.4 at the moment (if it is not too late already).

Applied. Thanks.

-- 
Catalin

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Ard Biesheuvel]

On 29 October 2015 at 19:39, Catalin Marinas <catalin.marinas-5wv7dgnIgG8@public.gmane.org> wrote:
> On Fri, Oct 23, 2015 at 04:48:14PM +0200, Ard Biesheuvel wrote:
>> Now that we added special handling to the C files in libstub, move
>> the one remaining arm64 specific EFI stub C file to libstub as
>> well, so that it gets the same treatment. This should prevent future
>> changes from resulting in binaries that may execute incorrectly in
>> UEFI context.
>>
>> With efi-entry.S the only remaining EFI stub source file under
>> arch/arm64, we can also simplify the Makefile logic somewhat.
>>
>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>> ---
>>
>> I would like to suggest that this be taken on top of the stuff that is
>> queued for 4.4 at the moment (if it is not too late already).
>
> Applied. Thanks.
>

Please note (before you push it out), that Will just pulled a 4.3-rc
fix that changes arch/arm64/kernel/efi-stub.c

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Catalin Marinas]

On Fri, Oct 30, 2015 at 12:57:48PM +0100, Ard Biesheuvel wrote:
> On 29 October 2015 at 19:39, Catalin Marinas <catalin.marinas-5wv7dgnIgG8@public.gmane.org> wrote:
> > On Fri, Oct 23, 2015 at 04:48:14PM +0200, Ard Biesheuvel wrote:
> >> Now that we added special handling to the C files in libstub, move
> >> the one remaining arm64 specific EFI stub C file to libstub as
> >> well, so that it gets the same treatment. This should prevent future
> >> changes from resulting in binaries that may execute incorrectly in
> >> UEFI context.
> >>
> >> With efi-entry.S the only remaining EFI stub source file under
> >> arch/arm64, we can also simplify the Makefile logic somewhat.
> >>
> >> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> >> ---
> >>
> >> I would like to suggest that this be taken on top of the stuff that is
> >> queued for 4.4 at the moment (if it is not too late already).
> >
> > Applied. Thanks.
> 
> Please note (before you push it out), that Will just pulled a 4.3-rc
> fix that changes arch/arm64/kernel/efi-stub.c

Thanks for the heads up. Apparently git is smart enough to move the
patched file if I merge my upstream branch into Will's without even
complaining.

-- 
Catalin

Re: [PATCH] arm64/efi: move arm64 specific stub C code to libstub

[Jeremy Linton]

On 10/23/2015 09:48 AM, Ard Biesheuvel wrote:
> Now that we added special handling to the C files in libstub, move
> the one remaining arm64 specific EFI stub C file to libstub as
> well, so that it gets the same treatment. This should prevent future
> changes from resulting in binaries that may execute incorrectly in
> UEFI context.
>
> With efi-entry.S the only remaining EFI stub source file under
> arch/arm64, we can also simplify the Makefile logic somewhat.
>
> Signed-off-by: Ard Biesheuvel <ard.biesheuvel-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>

I'm a little late here, but I've been running this for a few days in an 
EFI/ACPI environment without any issues.

So if there are any further merges..

Tested-by: Jeremy Linton <jeremy.linton-5wv7dgnIgG8@public.gmane.org>

Thanks,