Skip to contentSkip to conversation

efi: fdt: avoid FDT manipulation after ExitBootServices()

PATCH New Ard Biesheuvel 4 participants 5 messages Feb 1, 2017 EFI

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 efi: fdt: avoid FDT manipulation after ExitBootServices()

Some AArch64 UEFI implementations disable the MMU in ExitBootServices(), after which unaligned accesses to RAM are no longer supported.

Commit abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel") fixed an issue in the memory map handling of the stub FDT code, but inadvertently created an issue with such firmwares, by moving some of the FDT manipulation to after the invocation of ExitBootServices(). However, the stub's libfdt implementation uses the ordinary, accelerated string functions, which rely on hardware handling of unaligned accesses, resulting in alignment faults when executed with the MMU off.

Whether disabling the MMU in ExitBootServices() complies with the UEFI spec is unclear, but it is a reality we have to deal with, given that it wasn't a problem before commit abfb7b686a3e was applied.

So fix the situation by moving the update_fdt_memmap() into the callback invoked by efi_exit_boot_services() right before it hits the button.

Fixes abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel")
---
 drivers/firmware/efi/libstub/fdt.c | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)
drivers/firmware/efi/libstub/fdt.c +3 -11
index 921dfa047202..260c4b4b492e 100644
@@ -187,6 +187,7 @@ static efi_status_t update_fdt_memmap(void *fdt, struct efi_boot_memmap *map)
 struct exit_boot_struct {
 	efi_memory_desc_t *runtime_map;
 	int *runtime_entry_count;
+	void *new_fdt_addr;
 };
 
 static efi_status_t exit_boot_func(efi_system_table_t *sys_table_arg,
@@ -202,7 +203,7 @@ static efi_status_t exit_boot_func(efi_system_table_t *sys_table_arg,
 	efi_get_virtmap(*map->map, *map->map_size, *map->desc_size,
 			p->runtime_map, p->runtime_entry_count);
 
-	return EFI_SUCCESS;
+	return update_fdt_memmap(p->new_fdt_addr, map);
 }
 
 /*
@@ -300,22 +301,13 @@ efi_status_t allocate_new_fdt_and_exit_boot(efi_system_table_t *sys_table,
 
 	priv.runtime_map = runtime_map;
 	priv.runtime_entry_count = &runtime_entry_count;
+	priv.new_fdt_addr = (void *)*new_fdt_addr;
 	status = efi_exit_boot_services(sys_table, handle, &map, &priv,
 					exit_boot_func);
 
 	if (status == EFI_SUCCESS) {
 		efi_set_virtual_address_map_t *svam;
 
-		status = update_fdt_memmap((void *)*new_fdt_addr, &map);
-		if (status != EFI_SUCCESS) {
-			/*
-			 * The kernel won't get far without the memory map, but
-			 * may still be able to print something meaningful so
-			 * return success here.
-			 */
-			return EFI_SUCCESS;
-		}
-
 		/* Install the new virtual address map */
 		svam = sys_table->runtime->set_virtual_address_map;
 		status = svam(runtime_entry_count * desc_size, desc_size,
Signature · 1 line
2.7.4
#1 lore raw

PATCH 1/1 efi: fdt: avoid FDT manipulation after ExitBootServices()

Quoted 11 lines · Ard Biesheuvel

Some AArch64 UEFI implementations disable the MMU in ExitBootServices(), after which unaligned accesses to RAM are no longer supported.

Commit abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel") fixed an issue in the memory map handling of the stub FDT code, but inadvertently created an issue with such firmwares, by moving some of the FDT manipulation to after the invocation of ExitBootServices(). However, the stub's libfdt implementation uses the ordinary, accelerated string functions, which rely on hardware handling of unaligned accesses, resulting in alignment faults when executed with the MMU off.

Quoted 3 lines

Whether disabling the MMU in ExitBootServices() complies with the UEFI spec is unclear, but it is a reality we have to deal with, given that it wasn't a problem before commit abfb7b686a3e was applied.

I'm not sure it's unclear, but someone not reading the specification and instead looking
at what the unfortunate LinuxLoader did might incorrectly deduce that such is the case.

Quoted 4 lines

So fix the situation by moving the update_fdt_memmap() into the callback invoked by efi_exit_boot_services() right before it hits the button.

Fixes abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel")

I would sort argue against the Fixes: tag, given that I don't consider that code to be
incorrect.

I agree that this is a good change regardless.

/
Leif

Quoted 53 lines
---
 drivers/firmware/efi/libstub/fdt.c | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)
drivers/firmware/efi/libstub/fdt.c +3 -12
index 921dfa047202..260c4b4b492e 100644
@@ -187,6 +187,7 @@ static efi_status_t update_fdt_memmap(void *fdt, struct efi_boot_memmap *map)
 struct exit_boot_struct {
        efi_memory_desc_t *runtime_map;
        int *runtime_entry_count;
+       void *new_fdt_addr;
 };
 
 static efi_status_t exit_boot_func(efi_system_table_t *sys_table_arg,
@@ -202,7 +203,7 @@ static efi_status_t exit_boot_func(efi_system_table_t *sys_table_arg,
        efi_get_virtmap(*map->map, *map->map_size, *map->desc_size,
                        p->runtime_map, p->runtime_entry_count);
 
-       return EFI_SUCCESS;
+       return update_fdt_memmap(p->new_fdt_addr, map);
 }
 
 /*
@@ -300,22 +301,13 @@ efi_status_t allocate_new_fdt_and_exit_boot(efi_system_table_t *sys_table,
 
        priv.runtime_map = runtime_map;
        priv.runtime_entry_count = &runtime_entry_count;
+       priv.new_fdt_addr = (void *)*new_fdt_addr;
        status = efi_exit_boot_services(sys_table, handle, &map, &priv,
                                        exit_boot_func);
 
        if (status == EFI_SUCCESS) {
                efi_set_virtual_address_map_t *svam;
 
-               status = update_fdt_memmap((void *)*new_fdt_addr, &map);
-               if (status != EFI_SUCCESS) {
-                       /*
-                        * The kernel won't get far without the memory map, but
-                        * may still be able to print something meaningful so
-                        * return success here.
-                        */
-                       return EFI_SUCCESS;
-               }
-
                /* Install the new virtual address map */
                svam = sys_table->runtime->set_virtual_address_map;
                status = svam(runtime_entry_count * desc_size, desc_size,
--

2.7.4

#2 lore raw

PATCH 1/1 efi: fdt: avoid FDT manipulation after ExitBootServices()

Quoted 21 lines · Leif Lindholm
Quoted 11 lines · Ard Biesheuvel

Some AArch64 UEFI implementations disable the MMU in ExitBootServices(), after which unaligned accesses to RAM are no longer supported.

Commit abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel") fixed an issue in the memory map handling of the stub FDT code, but inadvertently created an issue with such firmwares, by moving some of the FDT manipulation to after the invocation of ExitBootServices(). However, the stub's libfdt implementation uses the ordinary, accelerated string functions, which rely on hardware handling of unaligned accesses, resulting in alignment faults when executed with the MMU off.

Quoted 3 lines

Whether disabling the MMU in ExitBootServices() complies with the UEFI spec is unclear, but it is a reality we have to deal with, given that it wasn't a problem before commit abfb7b686a3e was applied.

I'm not sure it's unclear, but someone not reading the specification and instead looking
at what the unfortunate LinuxLoader did might incorrectly deduce that such is the case.

Quoted 9 lines
Quoted 4 lines

So fix the situation by moving the update_fdt_memmap() into the callback invoked by efi_exit_boot_services() right before it hits the button.

Fixes abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel")

I would sort argue against the Fixes: tag, given that I don't consider that code to be
incorrect.

Well, that patch ended up in -stable, and is breaking actual systems that worked 'fine' before.

Quoted 2 lines

I agree that this is a good change regardless.

Cheers.

#3 lore raw

PATCH 1/1 efi: fdt: avoid FDT manipulation after ExitBootServices()

In reply to Ard Biesheuvel
Quoted 2 lines · Ard Biesheuvel

Some AArch64 UEFI implementations disable the MMU in ExitBootServices(), after which unaligned accesses to RAM are no longer supported.

Urrgh. That's a pretty horrible bug. :(

Quoted 11 lines

Commit abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel") fixed an issue in the memory map handling of the stub FDT code, but inadvertently created an issue with such firmwares, by moving some of the FDT manipulation to after the invocation of ExitBootServices(). However, the stub's libfdt implementation uses the ordinary, accelerated string functions, which rely on hardware handling of unaligned accesses, resulting in alignment faults when executed with the MMU off.

Whether disabling the MMU in ExitBootServices() complies with the UEFI spec is unclear, but it is a reality we have to deal with, given that it wasn't a problem before commit abfb7b686a3e was applied.

I'd argue that this quite clearly violates the calling convention requiremments in 2.3.6 (e.g. SCTLR_EL1 configuration, which permits unaligned accesses). If ExitBootServices() changes the SCTLR_EL1 value, it clearly violates the calling convention.

This would also mean we're calling SetVirtualAddressMap() in violation of said convention.

For the timebeing, can we drop this statement regarding the spec from the commit message?

Regardless, I agree that we need to handle this somehow.

Quoted 2 lines

So fix the situation by moving the update_fdt_memmap() into the callback invoked by efi_exit_boot_services() right before it hits the button.

I think this is a sensible change, but I do suspect that we're likely to have more fun with this broken platform in future as the stub changes (or compilers decide to generate unaligned accesses elsewhere).

Thanks,
Mark.

#4 lore raw

PATCH 1/1 efi: fdt: avoid FDT manipulation after ExitBootServices()

In reply to Ard Biesheuvel
Quoted 15 lines · Ard Biesheuvel

Some AArch64 UEFI implementations disable the MMU in ExitBootServices(), after which unaligned accesses to RAM are no longer supported.

Commit abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel") fixed an issue in the memory map handling of the stub FDT code, but inadvertently created an issue with such firmwares, by moving some of the FDT manipulation to after the invocation of ExitBootServices(). However, the stub's libfdt implementation uses the ordinary, accelerated string functions, which rely on hardware handling of unaligned accesses, resulting in alignment faults when executed with the MMU off.

Whether disabling the MMU in ExitBootServices() complies with the UEFI spec is unclear, but it is a reality we have to deal with, given that it wasn't a problem before commit abfb7b686a3e was applied.

Quoted 5 lines

So fix the situation by moving the update_fdt_memmap() into the callback invoked by efi_exit_boot_services() right before it hits the button.

Fixes abfb7b686a3e ("efi/libstub/arm*: Pass latest memory map to the kernel")
Quoted 52 lines
---
 drivers/firmware/efi/libstub/fdt.c | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)
drivers/firmware/efi/libstub/fdt.c +3 -12
index 921dfa047202..260c4b4b492e 100644
@@ -187,6 +187,7 @@ static efi_status_t update_fdt_memmap(void *fdt, struct efi_boot_memmap *map)
 struct exit_boot_struct {
        efi_memory_desc_t *runtime_map;
        int *runtime_entry_count;
+       void *new_fdt_addr;
 };
 
 static efi_status_t exit_boot_func(efi_system_table_t *sys_table_arg,
@@ -202,7 +203,7 @@ static efi_status_t exit_boot_func(efi_system_table_t *sys_table_arg,
        efi_get_virtmap(*map->map, *map->map_size, *map->desc_size,
                        p->runtime_map, p->runtime_entry_count);
 
-       return EFI_SUCCESS;
+       return update_fdt_memmap(p->new_fdt_addr, map);
 }
 
 /*
@@ -300,22 +301,13 @@ efi_status_t allocate_new_fdt_and_exit_boot(efi_system_table_t *sys_table,
 
        priv.runtime_map = runtime_map;
        priv.runtime_entry_count = &runtime_entry_count;
+       priv.new_fdt_addr = (void *)*new_fdt_addr;
        status = efi_exit_boot_services(sys_table, handle, &map, &priv,
                                        exit_boot_func);
 
        if (status == EFI_SUCCESS) {
                efi_set_virtual_address_map_t *svam;
 
-               status = update_fdt_memmap((void *)*new_fdt_addr, &map);
-               if (status != EFI_SUCCESS) {
-                       /*
-                        * The kernel won't get far without the memory map, but
-                        * may still be able to print something meaningful so
-                        * return success here.
-                        */
-                       return EFI_SUCCESS;
-               }
-
                /* Install the new virtual address map */
                svam = sys_table->runtime->set_virtual_address_map;
                status = svam(runtime_entry_count * desc_size, desc_size,
--

2.7.4