[PATCH v2 0/2] arm64: EFI stub isolation

[PATCH v2 0/2] arm64: EFI stub isolation

[Ard Biesheuvel]

We need to ensure that the EFI stub only uses parts of the kernel proper
that are safe to use when the kernel virtual mapping is not active yet.

So move all C code dependencies to the libstub, and do a verification pass
to ensure that no absolute relocations are used.

On the arm64 side, annotate all the stub's dependencies as safe for PI
(position independent

Changes since v1:
- add .size and .type directives to correctly annotate the __pi_xxx aliases
  as functions of the correct size
- add '#ifndef __HAVE_ARCH_STRSTR' around strstr() definition
- added linux-efi/Matt to cc

Ard Biesheuvel (2):
  arm64: use ENDPIPROC() to annotate position independent assembler
    routines
  arm64/efi: isolate EFI stub from the kernel proper

 arch/arm64/include/asm/assembler.h    | 11 ++++
 arch/arm64/kernel/Makefile            | 12 ++++-
 arch/arm64/kernel/efi-entry.S         | 10 ++--
 arch/arm64/kernel/head.S              | 14 ++---
 arch/arm64/kernel/image.h             | 27 ++++++++++
 arch/arm64/lib/memchr.S               |  2 +-
 arch/arm64/lib/memcmp.S               |  2 +-
 arch/arm64/lib/memcpy.S               |  2 +-
 arch/arm64/lib/memmove.S              |  2 +-
 arch/arm64/lib/memset.S               |  2 +-
 arch/arm64/lib/strlen.S               |  2 +-
 arch/arm64/lib/strncmp.S              |  2 +-
 arch/arm64/mm/cache.S                 | 10 ++--
 drivers/firmware/efi/libstub/Makefile | 31 +++++++++--
 drivers/firmware/efi/libstub/string.c | 57 ++++++++++++++++++++
 15 files changed, 157 insertions(+), 29 deletions(-)
 create mode 100644 drivers/firmware/efi/libstub/string.c

-- 
1.9.1

[PATCH v2 1/2] arm64: use ENDPIPROC() to annotate position independent assembler routines

[Ard Biesheuvel]

For more control over which functions are called with the MMU off or
with the UEFI 1:1 mapping active, annotate some assembler routines as
position independent. This is done by introducing ENDPIPROC(), which
replaces the ENDPROC() declaration of those routines.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---
 arch/arm64/include/asm/assembler.h | 11 +++++++++++
 arch/arm64/lib/memchr.S            |  2 +-
 arch/arm64/lib/memcmp.S            |  2 +-
 arch/arm64/lib/memcpy.S            |  2 +-
 arch/arm64/lib/memmove.S           |  2 +-
 arch/arm64/lib/memset.S            |  2 +-
 arch/arm64/lib/strlen.S            |  2 +-
 arch/arm64/lib/strncmp.S           |  2 +-
 arch/arm64/mm/cache.S              | 10 +++++-----
 9 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/arch/arm64/include/asm/assembler.h b/arch/arm64/include/asm/assembler.h
index b51f2cc22ca9..12eff928ef8b 100644
--- a/arch/arm64/include/asm/assembler.h
+++ b/arch/arm64/include/asm/assembler.h
@@ -193,4 +193,15 @@ lr	.req	x30		// link register
 	str	\src, [\tmp, :lo12:\sym]
 	.endm
 
+/*
+ * Annotate a function as position independent, i.e., safe to be called before
+ * the kernel virtual mapping is activated.
+ */
+#define ENDPIPROC(x)			\
+	.globl	__pi_##x;		\
+	.type 	__pi_##x, %function;	\
+	.set	__pi_##x, x;		\
+	.size	__pi_##x, . - x;	\
+	ENDPROC(x)
+
 #endif	/* __ASM_ASSEMBLER_H */
diff --git a/arch/arm64/lib/memchr.S b/arch/arm64/lib/memchr.S
index 8636b7549163..4444c1d25f4b 100644
--- a/arch/arm64/lib/memchr.S
+++ b/arch/arm64/lib/memchr.S
@@ -41,4 +41,4 @@ ENTRY(memchr)
 	ret
 2:	mov	x0, #0
 	ret
-ENDPROC(memchr)
+ENDPIPROC(memchr)
diff --git a/arch/arm64/lib/memcmp.S b/arch/arm64/lib/memcmp.S
index 6ea0776ba6de..ffbdec00327d 100644
--- a/arch/arm64/lib/memcmp.S
+++ b/arch/arm64/lib/memcmp.S
@@ -255,4 +255,4 @@ CPU_LE( rev	data2, data2 )
 .Lret0:
 	mov	result, #0
 	ret
-ENDPROC(memcmp)
+ENDPIPROC(memcmp)
diff --git a/arch/arm64/lib/memcpy.S b/arch/arm64/lib/memcpy.S
index 8a9a96d3ddae..5d6ddb74454d 100644
--- a/arch/arm64/lib/memcpy.S
+++ b/arch/arm64/lib/memcpy.S
@@ -198,4 +198,4 @@ ENTRY(memcpy)
 	tst	count, #0x3f
 	b.ne	.Ltail63
 	ret
-ENDPROC(memcpy)
+ENDPIPROC(memcpy)
diff --git a/arch/arm64/lib/memmove.S b/arch/arm64/lib/memmove.S
index 57b19ea2dad4..68e2f2035e23 100644
--- a/arch/arm64/lib/memmove.S
+++ b/arch/arm64/lib/memmove.S
@@ -194,4 +194,4 @@ ENTRY(memmove)
 	tst	count, #0x3f
 	b.ne	.Ltail63
 	ret
-ENDPROC(memmove)
+ENDPIPROC(memmove)
diff --git a/arch/arm64/lib/memset.S b/arch/arm64/lib/memset.S
index 7c72dfd36b63..29f405f08792 100644
--- a/arch/arm64/lib/memset.S
+++ b/arch/arm64/lib/memset.S
@@ -213,4 +213,4 @@ ENTRY(memset)
 	ands	count, count, zva_bits_x
 	b.ne	.Ltail_maybe_long
 	ret
-ENDPROC(memset)
+ENDPIPROC(memset)
diff --git a/arch/arm64/lib/strlen.S b/arch/arm64/lib/strlen.S
index 987b68b9ce44..55ccc8e24c08 100644
--- a/arch/arm64/lib/strlen.S
+++ b/arch/arm64/lib/strlen.S
@@ -123,4 +123,4 @@ CPU_LE( lsr	tmp2, tmp2, tmp1 )	/* Shift (tmp1 & 63).  */
 	csinv	data1, data1, xzr, le
 	csel	data2, data2, data2a, le
 	b	.Lrealigned
-ENDPROC(strlen)
+ENDPIPROC(strlen)
diff --git a/arch/arm64/lib/strncmp.S b/arch/arm64/lib/strncmp.S
index 0224cf5a5533..e267044761c6 100644
--- a/arch/arm64/lib/strncmp.S
+++ b/arch/arm64/lib/strncmp.S
@@ -307,4 +307,4 @@ CPU_BE( orr	syndrome, diff, has_nul )
 .Lret0:
 	mov	result, #0
 	ret
-ENDPROC(strncmp)
+ENDPIPROC(strncmp)
diff --git a/arch/arm64/mm/cache.S b/arch/arm64/mm/cache.S
index eb48d5df4a0f..cfa44a6adc0a 100644
--- a/arch/arm64/mm/cache.S
+++ b/arch/arm64/mm/cache.S
@@ -98,7 +98,7 @@ ENTRY(__flush_dcache_area)
 	b.lo	1b
 	dsb	sy
 	ret
-ENDPROC(__flush_dcache_area)
+ENDPIPROC(__flush_dcache_area)
 
 /*
  *	__inval_cache_range(start, end)
@@ -131,7 +131,7 @@ __dma_inv_range:
 	b.lo	2b
 	dsb	sy
 	ret
-ENDPROC(__inval_cache_range)
+ENDPIPROC(__inval_cache_range)
 ENDPROC(__dma_inv_range)
 
 /*
@@ -171,7 +171,7 @@ ENTRY(__dma_flush_range)
 	b.lo	1b
 	dsb	sy
 	ret
-ENDPROC(__dma_flush_range)
+ENDPIPROC(__dma_flush_range)
 
 /*
  *	__dma_map_area(start, size, dir)
@@ -184,7 +184,7 @@ ENTRY(__dma_map_area)
 	cmp	w2, #DMA_FROM_DEVICE
 	b.eq	__dma_inv_range
 	b	__dma_clean_range
-ENDPROC(__dma_map_area)
+ENDPIPROC(__dma_map_area)
 
 /*
  *	__dma_unmap_area(start, size, dir)
@@ -197,4 +197,4 @@ ENTRY(__dma_unmap_area)
 	cmp	w2, #DMA_TO_DEVICE
 	b.ne	__dma_inv_range
 	ret
-ENDPROC(__dma_unmap_area)
+ENDPIPROC(__dma_unmap_area)
-- 
1.9.1

[PATCH v2 2/2] arm64/efi: isolate EFI stub from the kernel proper

[Ard Biesheuvel]

Since arm64 does not use a builtin decompressor, the EFI stub is built
into the kernel proper. So far, this has been working fine, but actually,
since the stub is in fact a PE/COFF relocatable binary that is executed
at an unknown offset in the 1:1 mapping provided by the UEFI firmware, we
should not be seamlessly sharing code with the kernel proper, which is a
position dependent executable linked at a high virtual offset.

So instead, separate the contents of libstub and its dependencies, by
putting them into their own namespace by prefixing all of its symbols
with __efistub. This way, we have tight control over what parts of the
kernel proper are referenced by the stub.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---
 arch/arm64/kernel/Makefile            | 12 ++++-
 arch/arm64/kernel/efi-entry.S         | 10 ++--
 arch/arm64/kernel/head.S              | 14 ++---
 arch/arm64/kernel/image.h             | 27 ++++++++++
 drivers/firmware/efi/libstub/Makefile | 31 +++++++++--
 drivers/firmware/efi/libstub/string.c | 57 ++++++++++++++++++++
 6 files changed, 134 insertions(+), 17 deletions(-)

diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index 22dc9bc781be..7b17f6245f1e 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -20,6 +20,14 @@ 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))
+
+OBJCOPYFLAGS := --prefix-symbols=__efistub_
+$(obj)/%.stub.o: $(obj)/%.o FORCE
+	$(call if_changed,objcopy)
+
 arm64-obj-$(CONFIG_COMPAT)		+= sys32.o kuser32.o signal32.o 	\
 					   sys_compat.o entry32.o		\
 					   ../../arm/kernel/opcodes.o
@@ -32,7 +40,7 @@ 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-$(CONFIG_EFI)			+= efi.o efi-stub.o efi-entry.o
+arm64-obj-$(CONFIG_EFI)			+= efi.o $(stub-obj)
 arm64-obj-$(CONFIG_PCI)			+= pci.o
 arm64-obj-$(CONFIG_ARMV8_DEPRECATED)	+= armv8_deprecated.o
 arm64-obj-$(CONFIG_ACPI)		+= acpi.o
@@ -40,7 +48,7 @@ arm64-obj-$(CONFIG_ACPI)		+= acpi.o
 obj-y					+= $(arm64-obj-y) vdso/
 obj-m					+= $(arm64-obj-m)
 head-y					:= head.o
-extra-y					:= $(head-y) vmlinux.lds
+extra-y					+= $(head-y) vmlinux.lds
 
 # vDSO - this must be built first to generate the symbol offsets
 $(call objectify,$(arm64-obj-y)): $(obj)/vdso/vdso-offsets.h
diff --git a/arch/arm64/kernel/efi-entry.S b/arch/arm64/kernel/efi-entry.S
index 8ce9b0577442..a773db92908b 100644
--- a/arch/arm64/kernel/efi-entry.S
+++ b/arch/arm64/kernel/efi-entry.S
@@ -29,7 +29,7 @@
 	 * we want to be. The kernel image wants to be placed at TEXT_OFFSET
 	 * from start of RAM.
 	 */
-ENTRY(efi_stub_entry)
+ENTRY(entry)
 	/*
 	 * Create a stack frame to save FP/LR with extra space
 	 * for image_addr variable passed to efi_entry().
@@ -86,8 +86,8 @@ ENTRY(efi_stub_entry)
 	 * entries for the VA range of the current image, so no maintenance is
 	 * necessary.
 	 */
-	adr	x0, efi_stub_entry
-	adr	x1, efi_stub_entry_end
+	adr	x0, entry
+	adr	x1, entry_end
 	sub	x1, x1, x0
 	bl	__flush_dcache_area
 
@@ -120,5 +120,5 @@ efi_load_fail:
 	ldp	x29, x30, [sp], #32
 	ret
 
-efi_stub_entry_end:
-ENDPROC(efi_stub_entry)
+entry_end:
+ENDPROC(entry)
diff --git a/arch/arm64/kernel/head.S b/arch/arm64/kernel/head.S
index a055be6125cf..e8f52943ac65 100644
--- a/arch/arm64/kernel/head.S
+++ b/arch/arm64/kernel/head.S
@@ -120,8 +120,8 @@ efi_head:
 #endif
 
 #ifdef CONFIG_EFI
-	.globl	stext_offset
-	.set	stext_offset, stext - efi_head
+	.globl	__efistub_stext_offset
+	.set	__efistub_stext_offset, stext - efi_head
 	.align 3
 pe_header:
 	.ascii	"PE"
@@ -144,8 +144,8 @@ optional_header:
 	.long	_end - stext			// SizeOfCode
 	.long	0				// SizeOfInitializedData
 	.long	0				// SizeOfUninitializedData
-	.long	efi_stub_entry - efi_head	// AddressOfEntryPoint
-	.long	stext_offset			// BaseOfCode
+	.long	__efistub_entry - efi_head	// AddressOfEntryPoint
+	.long	__efistub_stext_offset		// BaseOfCode
 
 extra_header_fields:
 	.quad	0				// ImageBase
@@ -162,7 +162,7 @@ extra_header_fields:
 	.long	_end - efi_head			// SizeOfImage
 
 	// Everything before the kernel image is considered part of the header
-	.long	stext_offset			// SizeOfHeaders
+	.long	__efistub_stext_offset		// SizeOfHeaders
 	.long	0				// CheckSum
 	.short	0xa				// Subsystem (EFI application)
 	.short	0				// DllCharacteristics
@@ -207,9 +207,9 @@ section_table:
 	.byte	0
 	.byte	0        		// end of 0 padding of section name
 	.long	_end - stext		// VirtualSize
-	.long	stext_offset		// VirtualAddress
+	.long	__efistub_stext_offset	// VirtualAddress
 	.long	_edata - stext		// SizeOfRawData
-	.long	stext_offset		// PointerToRawData
+	.long	__efistub_stext_offset	// PointerToRawData
 
 	.long	0		// PointerToRelocations (0 for executables)
 	.long	0		// PointerToLineNumbers (0 for executables)
diff --git a/arch/arm64/kernel/image.h b/arch/arm64/kernel/image.h
index 8fae0756e175..2f6a426de864 100644
--- a/arch/arm64/kernel/image.h
+++ b/arch/arm64/kernel/image.h
@@ -59,4 +59,31 @@
 	_kernel_offset_le	= DATA_LE64(TEXT_OFFSET);	\
 	_kernel_flags_le	= DATA_LE64(__HEAD_FLAGS);
 
+#ifdef CONFIG_EFI
+
+/*
+ * The EFI stub has its own symbol namespace prefixed by __efistub_, to
+ * isolate it from the kernel proper. The following symbols are legally
+ * accessed by the stub, so provide some aliases to make them accessible.
+ * Only include data symbols here, or text symbols of functions that are
+ * guaranteed to be safe when executed at another offset than they were
+ * linked at. The routines below are all implemented in assembler in a
+ * position independent manner
+ */
+__efistub_memcmp		= __pi_memcmp;
+__efistub_memchr		= __pi_memchr;
+__efistub_memcpy		= __pi_memcpy;
+__efistub_memmove		= __pi_memmove;
+__efistub_memset		= __pi_memset;
+__efistub_strlen		= __pi_strlen;
+__efistub_strncmp		= __pi_strncmp;
+__efistub___flush_dcache_area	= __pi___flush_dcache_area;
+
+__efistub_linux_banner		= linux_banner;
+__efistub__text			= _text;
+__efistub__end			= _end;
+__efistub__edata		= _edata;
+
+#endif
+
 #endif /* __ASM_IMAGE_H */
diff --git a/drivers/firmware/efi/libstub/Makefile b/drivers/firmware/efi/libstub/Makefile
index 816dbe9f4b82..50557f4d12db 100644
--- a/drivers/firmware/efi/libstub/Makefile
+++ b/drivers/firmware/efi/libstub/Makefile
@@ -14,6 +14,8 @@ cflags-$(CONFIG_ARM64)		:= $(subst -pg,,$(KBUILD_CFLAGS))
 cflags-$(CONFIG_ARM)		:= $(subst -pg,,$(KBUILD_CFLAGS)) \
 				   -fno-builtin -fpic -mno-single-pic-base
 
+cflags-$(CONFIG_EFI_ARMSTUB)	+= -I$(srctree)/scripts/dtc/libfdt
+
 KBUILD_CFLAGS			:= $(cflags-y) \
 				   $(call cc-option,-ffreestanding) \
 				   $(call cc-option,-fno-stack-protector)
@@ -22,7 +24,15 @@ GCOV_PROFILE			:= n
 KASAN_SANITIZE			:= n
 
 lib-y				:= efi-stub-helper.o
-lib-$(CONFIG_EFI_ARMSTUB)	+= arm-stub.o fdt.o
+
+# include the stub's generic dependencies from lib/ when building for ARM/arm64
+arm-deps := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c sort.c
+
+$(obj)/lib-%.o: $(srctree)/lib/%.c FORCE
+	$(call if_changed_rule,cc_o_c)
+
+lib-$(CONFIG_EFI_ARMSTUB)	+= arm-stub.o fdt.o string.o \
+				   $(patsubst %.c,lib-%.o,$(arm-deps))
 
 #
 # arm64 puts the stub in the kernel proper, which will unnecessarily retain all
@@ -30,10 +40,25 @@ lib-$(CONFIG_EFI_ARMSTUB)	+= arm-stub.o fdt.o
 # So let's apply the __init annotations at the section level, by prefixing
 # the section names directly. This will ensure that even all the inline string
 # literals are covered.
+# The fact that the stub and the kernel proper are essentially the same binary
+# also means that we need to be extra careful to make sure that the stub does
+# not rely on any absolute symbol references, considering that the virtual
+# kernel mapping that the linker uses is not active yet when the stub is
+# executing. So build all C dependencies of the EFI stub into libstub, and do
+# a verification pass to see if any absolute relocations exist in any of the
+# object files.
 #
 extra-$(CONFIG_ARM64)		:= $(lib-y)
 lib-$(CONFIG_ARM64)		:= $(patsubst %.o,%.init.o,$(lib-y))
 
-OBJCOPYFLAGS := --prefix-alloc-sections=.init
+OBJCOPYFLAGS := --prefix-alloc-sections=.init --prefix-symbols=__efistub_ \
+		-R .debug* -R *ksymtab*
+
 $(obj)/%.init.o: $(obj)/%.o FORCE
-	$(call if_changed,objcopy)
+	$(call if_changed,stubcopy)
+
+quiet_cmd_stubcopy = STUBCPY $@
+      cmd_stubcopy = if $(OBJCOPY) $(OBJCOPYFLAGS) $< $@; then		\
+		     $(OBJDUMP) -r $@ | grep R_AARCH64_ABS 		\
+		     && (echo >&2 "$@: absolute symbol references not allowed in the EFI stub"; \
+			 rm -f $@; /bin/false); else /bin/false; fi
diff --git a/drivers/firmware/efi/libstub/string.c b/drivers/firmware/efi/libstub/string.c
new file mode 100644
index 000000000000..09d5a0894343
--- /dev/null
+++ b/drivers/firmware/efi/libstub/string.c
@@ -0,0 +1,57 @@
+/*
+ * Taken from:
+ *  linux/lib/string.c
+ *
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ */
+
+#include <linux/types.h>
+#include <linux/string.h>
+
+#ifndef __HAVE_ARCH_STRSTR
+/**
+ * strstr - Find the first substring in a %NUL terminated string
+ * @s1: The string to be searched
+ * @s2: The string to search for
+ */
+char *strstr(const char *s1, const char *s2)
+{
+	size_t l1, l2;
+
+	l2 = strlen(s2);
+	if (!l2)
+		return (char *)s1;
+	l1 = strlen(s1);
+	while (l1 >= l2) {
+		l1--;
+		if (!memcmp(s1, s2, l2))
+			return (char *)s1;
+		s1++;
+	}
+	return NULL;
+}
+#endif
+
+#ifndef __HAVE_ARCH_STRNCMP
+/**
+ * strncmp - Compare two length-limited strings
+ * @cs: One string
+ * @ct: Another string
+ * @count: The maximum number of bytes to compare
+ */
+int strncmp(const char *cs, const char *ct, size_t count)
+{
+	unsigned char c1, c2;
+
+	while (count) {
+		c1 = *cs++;
+		c2 = *ct++;
+		if (c1 != c2)
+			return c1 < c2 ? -1 : 1;
+		if (!c1)
+			break;
+		count--;
+	}
+	return 0;
+}
+#endif
-- 
1.9.1