efi/libstub: arm: omit sorting of the UEFI memory map

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Signed-off-by Ard Biesheuvel <[email protected]>

---
  drivers/firmware/efi/libstub/Makefile   | 6 +++---
  drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
  2 files changed, 8 insertions(+), 5 deletions(-)

drivers/firmware/efi/libstub/Makefile +3 -3

index dedf9bde44db..f3e8431565ea 100644
@@ -33,13 +33,14 @@ lib-y				:= efi-stub-helper.o gop.o secureboot.o
  lib-$(CONFIG_RESET_ATTACK_MITIGATION) += tpm.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
+arm-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c
+arm-deps-$(CONFIG_ARM64) += 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 random.o \
-				   $(patsubst %.c,lib-%.o,$(arm-deps))
+				   $(patsubst %.c,lib-%.o,$(arm-deps-y))
 
  lib-$(CONFIG_ARM)		+= arm32-stub.o
  lib-$(CONFIG_ARM64)		+= arm64-stub.o

@@ -90,5 +91,4 @@ quiet_cmd_stubcopy = STUBCPY $@
  # explicitly by the decompressor linker script.
  #
  STUBCOPY_FLAGS-$(CONFIG_ARM)	+= --rename-section .data=.data.efistub
-STUBCOPY_RM-$(CONFIG_ARM)	+= -R ___ksymtab+sort -R ___kcrctab+sort
  STUBCOPY_RELOC-$(CONFIG_ARM)	:= R_ARM_ABS

drivers/firmware/efi/libstub/arm-stub.c +5 -2

index 1cb2d1c070c3..3061e4057483 100644
@@ -349,7 +349,9 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
  	 * The easiest way to find adjacent regions is to sort the memory map
  	 * before traversing it.
  	 */
-	sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc, NULL);
+	if (IS_ENABLED(CONFIG_ARM64))
+		sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc,
+		     NULL);
 
  	for (l = 0; l < map_size; l += desc_size, prev = in) {
  		u64 paddr, size;

@@ -366,7 +368,8 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
  		 * a 4k page size kernel to kexec a 64k page size kernel and
  		 * vice versa.
  		 */
-		if (!regions_are_adjacent(prev, in) ||
+		if ((IS_ENABLED(CONFIG_ARM64) &&
+		     !regions_are_adjacent(prev, in)) ||
  		    !regions_have_compatible_memory_type_attrs(prev, in)) {
 
  			paddr = round_down(in->phys_addr, SZ_64K);

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Signed-off-by Ard Biesheuvel <[email protected]>

---

 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

drivers/firmware/efi/libstub/Makefile +3 -3

index dedf9bde44db..f3e8431565ea 100644
@@ -33,13 +33,14 @@ lib-y				:= efi-stub-helper.o gop.o secureboot.o
 lib-$(CONFIG_RESET_ATTACK_MITIGATION) += tpm.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
+arm-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c
+arm-deps-$(CONFIG_ARM64) += 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 random.o \
-				   $(patsubst %.c,lib-%.o,$(arm-deps))
+				   $(patsubst %.c,lib-%.o,$(arm-deps-y))
 
 lib-$(CONFIG_ARM)		+= arm32-stub.o
 lib-$(CONFIG_ARM64)		+= arm64-stub.o

@@ -90,5 +91,4 @@ quiet_cmd_stubcopy = STUBCPY $@
 # explicitly by the decompressor linker script.
 #
 STUBCOPY_FLAGS-$(CONFIG_ARM)	+= --rename-section .data=.data.efistub
-STUBCOPY_RM-$(CONFIG_ARM)	+= -R ___ksymtab+sort -R ___kcrctab+sort
 STUBCOPY_RELOC-$(CONFIG_ARM)	:= R_ARM_ABS

drivers/firmware/efi/libstub/arm-stub.c +5 -2

index 1cb2d1c070c3..3061e4057483 100644
@@ -349,7 +349,9 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 	 * The easiest way to find adjacent regions is to sort the memory map
 	 * before traversing it.
 	 */
-	sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc, NULL);
+	if (IS_ENABLED(CONFIG_ARM64))
+		sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc,
+		     NULL);
 
 	for (l = 0; l < map_size; l += desc_size, prev = in) {
 		u64 paddr, size;

@@ -366,7 +368,8 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 		 * a 4k page size kernel to kexec a 64k page size kernel and
 		 * vice versa.
 		 */
-		if (!regions_are_adjacent(prev, in) ||
+		if ((IS_ENABLED(CONFIG_ARM64) &&
+		     !regions_are_adjacent(prev, in)) ||
 		    !regions_have_compatible_memory_type_attrs(prev, in)) {
 
 			paddr = round_down(in->phys_addr, SZ_64K);

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Signed-off-by Ard Biesheuvel <[email protected]>

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

drivers/firmware/efi/libstub/Makefile +3 -3

index dedf9bde44db..f3e8431565ea 100644
@@ -33,13 +33,14 @@ lib-y				:= efi-stub-helper.o gop.o secureboot.o
 lib-$(CONFIG_RESET_ATTACK_MITIGATION) += tpm.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
+arm-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c
+arm-deps-$(CONFIG_ARM64) += 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 random.o \
-				   $(patsubst %.c,lib-%.o,$(arm-deps))
+				   $(patsubst %.c,lib-%.o,$(arm-deps-y))
 
 lib-$(CONFIG_ARM)		+= arm32-stub.o
 lib-$(CONFIG_ARM64)		+= arm64-stub.o

@@ -90,5 +91,4 @@ quiet_cmd_stubcopy = STUBCPY $@
 # explicitly by the decompressor linker script.
 #
 STUBCOPY_FLAGS-$(CONFIG_ARM)	+= --rename-section .data=.data.efistub
-STUBCOPY_RM-$(CONFIG_ARM)	+= -R ___ksymtab+sort -R ___kcrctab+sort
 STUBCOPY_RELOC-$(CONFIG_ARM)	:= R_ARM_ABS

drivers/firmware/efi/libstub/arm-stub.c +5 -2

index 1cb2d1c070c3..3061e4057483 100644
@@ -349,7 +349,9 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 	 * The easiest way to find adjacent regions is to sort the memory map
 	 * before traversing it.
 	 */
-	sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc, NULL);
+	if (IS_ENABLED(CONFIG_ARM64))
+		sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc,
+		     NULL);
 
 	for (l = 0; l < map_size; l += desc_size, prev = in) {
 		u64 paddr, size;

@@ -366,7 +368,8 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 		 * a 4k page size kernel to kexec a 64k page size kernel and
 		 * vice versa.
 		 */
-		if (!regions_are_adjacent(prev, in) ||
+		if ((IS_ENABLED(CONFIG_ARM64) &&
+		     !regions_are_adjacent(prev, in)) ||
 		    !regions_have_compatible_memory_type_attrs(prev, in)) {
 
 			paddr = round_down(in->phys_addr, SZ_64K);

Signature · 1 line

2.11.0

Quoted 26 lines · Ard Biesheuvel

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Quoted 1 line

Signed-off-by Ard Biesheuvel <[email protected]>

I guess as this is shared between ARM and ARM64, ARM64 folk should ack this before I merge it - can I have an ack from that side please?

Quoted 58 lines

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

drivers/firmware/efi/libstub/Makefile +3 -3

index dedf9bde44db..f3e8431565ea 100644
@@ -33,13 +33,14 @@ lib-y                             := efi-stub-helper.o gop.o secureboot.o
 lib-$(CONFIG_RESET_ATTACK_MITIGATION) += tpm.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
+arm-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c
+arm-deps-$(CONFIG_ARM64) += 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 random.o \
-                                $(patsubst %.c,lib-%.o,$(arm-deps))
+                                $(patsubst %.c,lib-%.o,$(arm-deps-y))
 
 lib-$(CONFIG_ARM)            += arm32-stub.o
 lib-$(CONFIG_ARM64)          += arm64-stub.o

@@ -90,5 +91,4 @@ quiet_cmd_stubcopy = STUBCPY $@
 # explicitly by the decompressor linker script.
 #
 STUBCOPY_FLAGS-$(CONFIG_ARM) += --rename-section .data=.data.efistub
-STUBCOPY_RM-$(CONFIG_ARM)    += -R ___ksymtab+sort -R ___kcrctab+sort
 STUBCOPY_RELOC-$(CONFIG_ARM) := R_ARM_ABS

drivers/firmware/efi/libstub/arm-stub.c +5 -3

index 1cb2d1c070c3..3061e4057483 100644
@@ -349,7 +349,9 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
       * The easiest way to find adjacent regions is to sort the memory map
       * before traversing it.
       */
-     sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc, NULL);
+     if (IS_ENABLED(CONFIG_ARM64))
+             sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc,
+                  NULL);
 
      for (l = 0; l < map_size; l += desc_size, prev = in) {
              u64 paddr, size;

@@ -366,7 +368,8 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
               * a 4k page size kernel to kexec a 64k page size kernel and
               * vice versa.
               */
-             if (!regions_are_adjacent(prev, in) ||
+             if ((IS_ENABLED(CONFIG_ARM64) &&
+                  !regions_are_adjacent(prev, in)) ||
                  !regions_have_compatible_memory_type_attrs(prev, in)) {
 
                      paddr = round_down(in->phys_addr, SZ_64K);
--

Quoted 2 lines

2.11.0

--
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 8.8Mbps down 630kbps up According to speedtest.net: 8.21Mbps down 510kbps up

Quoted 29 lines · Russell King - ARM Linux

Quoted 27 lines · Ard Biesheuvel

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Signed-off-by Ard Biesheuvel <[email protected]>

Quoted 3 lines

I guess as this is shared between ARM and ARM64, ARM64 folk should ack this before I merge it - can I have an ack from that side please?

Note that the patch does not touch arch/arm64, nor does it affect arm64, given that the change is a no-op if CONFIG_ARM64=y. That said, I welcome any acks from the arm64 maintainers, but I don't think they are actually required here.

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Signed-off-by Ard Biesheuvel <[email protected]>

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

drivers/firmware/efi/libstub/Makefile +3 -3

index dedf9bde44db..f3e8431565ea 100644
@@ -33,13 +33,14 @@ lib-y				:= efi-stub-helper.o gop.o secureboot.o
 lib-$(CONFIG_RESET_ATTACK_MITIGATION) += tpm.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
+arm-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c
+arm-deps-$(CONFIG_ARM64) += 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 random.o \
-				   $(patsubst %.c,lib-%.o,$(arm-deps))
+				   $(patsubst %.c,lib-%.o,$(arm-deps-y))
 
 lib-$(CONFIG_ARM)		+= arm32-stub.o
 lib-$(CONFIG_ARM64)		+= arm64-stub.o

@@ -90,5 +91,4 @@ quiet_cmd_stubcopy = STUBCPY $@
 # explicitly by the decompressor linker script.
 #
 STUBCOPY_FLAGS-$(CONFIG_ARM)	+= --rename-section .data=.data.efistub
-STUBCOPY_RM-$(CONFIG_ARM)	+= -R ___ksymtab+sort -R ___kcrctab+sort
 STUBCOPY_RELOC-$(CONFIG_ARM)	:= R_ARM_ABS

drivers/firmware/efi/libstub/arm-stub.c +5 -2

index 1cb2d1c070c3..3061e4057483 100644
@@ -349,7 +349,9 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 	 * The easiest way to find adjacent regions is to sort the memory map
 	 * before traversing it.
 	 */
-	sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc, NULL);
+	if (IS_ENABLED(CONFIG_ARM64))
+		sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc,
+		     NULL);
 
 	for (l = 0; l < map_size; l += desc_size, prev = in) {
 		u64 paddr, size;

@@ -366,7 +368,8 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 		 * a 4k page size kernel to kexec a 64k page size kernel and
 		 * vice versa.
 		 */
-		if (!regions_are_adjacent(prev, in) ||
+		if ((IS_ENABLED(CONFIG_ARM64) &&
+		     !regions_are_adjacent(prev, in)) ||
 		    !regions_have_compatible_memory_type_attrs(prev, in)) {
 
 			paddr = round_down(in->phys_addr, SZ_64K);

Signature · 1 line

2.11.0

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Signed-off-by Ard Biesheuvel <[email protected]>

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

drivers/firmware/efi/libstub/Makefile +3 -3

index dedf9bde44db..f3e8431565ea 100644
@@ -33,13 +33,14 @@ lib-y				:= efi-stub-helper.o gop.o secureboot.o
 lib-$(CONFIG_RESET_ATTACK_MITIGATION) += tpm.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
+arm-deps-y := fdt_rw.c fdt_ro.c fdt_wip.c fdt.c fdt_empty_tree.c fdt_sw.c
+arm-deps-$(CONFIG_ARM64) += 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 random.o \
-				   $(patsubst %.c,lib-%.o,$(arm-deps))
+				   $(patsubst %.c,lib-%.o,$(arm-deps-y))
 
 lib-$(CONFIG_ARM)		+= arm32-stub.o
 lib-$(CONFIG_ARM64)		+= arm64-stub.o

@@ -90,5 +91,4 @@ quiet_cmd_stubcopy = STUBCPY $@
 # explicitly by the decompressor linker script.
 #
 STUBCOPY_FLAGS-$(CONFIG_ARM)	+= --rename-section .data=.data.efistub
-STUBCOPY_RM-$(CONFIG_ARM)	+= -R ___ksymtab+sort -R ___kcrctab+sort
 STUBCOPY_RELOC-$(CONFIG_ARM)	:= R_ARM_ABS

drivers/firmware/efi/libstub/arm-stub.c +5 -2

index 1cb2d1c070c3..3061e4057483 100644
@@ -349,7 +349,9 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 	 * The easiest way to find adjacent regions is to sort the memory map
 	 * before traversing it.
 	 */
-	sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc, NULL);
+	if (IS_ENABLED(CONFIG_ARM64))
+		sort(memory_map, map_size / desc_size, desc_size, cmp_mem_desc,
+		     NULL);
 
 	for (l = 0; l < map_size; l += desc_size, prev = in) {
 		u64 paddr, size;

@@ -366,7 +368,8 @@ void efi_get_virtmap(efi_memory_desc_t *memory_map, unsigned long map_size,
 		 * a 4k page size kernel to kexec a 64k page size kernel and
 		 * vice versa.
 		 */
-		if (!regions_are_adjacent(prev, in) ||
+		if ((IS_ENABLED(CONFIG_ARM64) &&
+		     !regions_are_adjacent(prev, in)) ||
 		    !regions_have_compatible_memory_type_attrs(prev, in)) {
 
 			paddr = round_down(in->phys_addr, SZ_64K);

Hi Ard,

Quoted 26 lines · Ard Biesheuvel

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Quoted 6 lines

Signed-off-by Ard Biesheuvel <[email protected]>

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

I thought I already provided my tag, but just in case:
This fixes the boot problems I had on the bananapi-r2.

Tested-by Matthias Brugger <[email protected]>

Quoted 39 lines · Matthias Brugger

Hi Ard,

Quoted 26 lines · Ard Biesheuvel

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Quoted 6 lines

Signed-off-by Ard Biesheuvel <[email protected]>

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

I thought I already provided my tag, but just in case:
This fixes the boot problems I had on the bananapi-r2.

Quoted 1 line

Tested-by Matthias Brugger <[email protected]>

Please test my "fixes" branch - that has all three patches merged.

We don't have linux-next running at the moment, so folk need to be manually fetching the appropriate git trees for testing.

Quoted 39 lines · Matthias Brugger

Hi Ard,

Quoted 26 lines · Ard Biesheuvel

ARM shares its EFI stub implementation with arm64, which has some special handling in the virtual remapping code to
a) make sure that we can map everything even if the OS executes with 64k page size, and
b) make sure that adjacent regions with the same attributes are not reordered or moved apart in memory.

The latter is a workaround for a 'feature' that was shortly recommended by UEFI spec v2.5, but deprecated shortly after, due to the fact that it broke many OS installers, including non-Linux ones, and it was never widely implemented for ARM systems. Before implementing b), the arm64 code simply rounded up all regions to 64 KB granularity, but given that that results in moving adjacent regions apart, it had to be refined when
b) was implemented.

The adjacency check requires a sort() pass, due to the fact that the UEFI spec does not mandate any ordering, and the inclusion of the lib/sort.c code into the ARM EFI stub is causing some trouble with the decompressor build due to the fact that its EXPORT_SYMBOL() call triggers the creation of ksymtab/kcrctab sections.

So let's simply do away with the adjacency check for ARM, and simply put all UEFI runtime regions together if they have the same memory attributes. This is guaranteed to work, given that ARM only supports 4 KB pages, and allows us to remove the sort() call entirely.

Quoted 6 lines

Signed-off-by Ard Biesheuvel <[email protected]>

---
 drivers/firmware/efi/libstub/Makefile   | 6 +++---
 drivers/firmware/efi/libstub/arm-stub.c | 7 +++++--
 2 files changed, 8 insertions(+), 5 deletions(-)

I thought I already provided my tag, but just in case:
This fixes the boot problems I had on the bananapi-r2.

Quoted 1 line

Tested-by Matthias Brugger <[email protected]>

Please test my "fixes" branch - that has all three patches merged.

We don't have linux-next running at the moment, so folk need to be manually fetching the appropriate git trees for testing.