arm-trusted-firmware/common/bl_common.c

588 lines
18 KiB
C
Raw Normal View History

2013-10-25 09:08:21 +01:00
/*
Introduce unified API to zero memory Introduce zeromem_dczva function on AArch64 that can handle unaligned addresses and make use of DC ZVA instruction to zero a whole block at a time. This zeroing takes place directly in the cache to speed it up without doing external memory access. Remove the zeromem16 function on AArch64 and replace it with an alias to zeromem. This zeromem16 function is now deprecated. Remove the 16-bytes alignment constraint on __BSS_START__ in firmware-design.md as it is now not mandatory anymore (it used to comply with zeromem16 requirements). Change the 16-bytes alignment constraints in SP min's linker script to a 8-bytes alignment constraint as the AArch32 zeromem implementation is now more efficient on 8-bytes aligned addresses. Introduce zero_normalmem and zeromem helpers in platform agnostic header that are implemented this way: * AArch32: * zero_normalmem: zero using usual data access * zeromem: alias for zero_normalmem * AArch64: * zero_normalmem: zero normal memory using DC ZVA instruction (needs MMU enabled) * zeromem: zero using usual data access Usage guidelines: in most cases, zero_normalmem should be preferred. There are 2 scenarios where zeromem (or memset) must be used instead: * Code that must run with MMU disabled (which means all memory is considered device memory for data accesses). * Code that fills device memory with null bytes. Optionally, the following rule can be applied if performance is important: * Code zeroing small areas (few bytes) that are not secrets should use memset to take advantage of compiler optimizations. Note: Code zeroing security-related critical information should use zero_normalmem/zeromem instead of memset to avoid removal by compilers' optimizations in some cases or misbehaving versions of GCC. Fixes ARM-software/tf-issues#408 Change-Id: Iafd9663fc1070413c3e1904e54091cf60effaa82 Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
2016-12-02 13:51:54 +00:00
* Copyright (c) 2013-2017, ARM Limited and Contributors. All rights reserved.
2013-10-25 09:08:21 +01:00
*
* SPDX-License-Identifier: BSD-3-Clause
2013-10-25 09:08:21 +01:00
*/
#include <arch.h>
2013-10-25 09:08:21 +01:00
#include <arch_helpers.h>
#include <assert.h>
#include <auth_mod.h>
2013-10-25 09:08:21 +01:00
#include <bl_common.h>
#include <debug.h>
#include <errno.h>
#include <io_storage.h>
#include <platform.h>
#include <string.h>
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
#include <utils.h>
#include <xlat_tables_defs.h>
2013-10-25 09:08:21 +01:00
uintptr_t page_align(uintptr_t value, unsigned dir)
2013-10-25 09:08:21 +01:00
{
/* Round up the limit to the next page boundary */
if (value & (PAGE_SIZE - 1)) {
value &= ~(PAGE_SIZE - 1);
2013-10-25 09:08:21 +01:00
if (dir == UP)
value += PAGE_SIZE;
2013-10-25 09:08:21 +01:00
}
return value;
}
/******************************************************************************
* Determine whether the memory region delimited by 'addr' and 'size' is free,
* given the extents of free memory.
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
* Return 1 if it is free, 0 if it is not free or if the input values are
* invalid.
*****************************************************************************/
int is_mem_free(uintptr_t free_base, size_t free_size,
uintptr_t addr, size_t size)
{
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
uintptr_t free_end, requested_end;
/*
* Handle corner cases first.
*
* The order of the 2 tests is important, because if there's no space
* left (i.e. free_size == 0) but we don't ask for any memory
* (i.e. size == 0) then we should report that the memory is free.
*/
if (size == 0)
return 1; /* A zero-byte region is always free */
if (free_size == 0)
return 0;
/*
* Check that the end addresses don't overflow.
* If they do, consider that this memory region is not free, as this
* is an invalid scenario.
*/
if (check_uptr_overflow(free_base, free_size - 1))
return 0;
free_end = free_base + (free_size - 1);
if (check_uptr_overflow(addr, size - 1))
return 0;
requested_end = addr + (size - 1);
/*
* Finally, check that the requested memory region lies within the free
* region.
*/
return (addr >= free_base) && (requested_end <= free_end);
}
2013-10-25 09:08:21 +01:00
#if !LOAD_IMAGE_V2
/******************************************************************************
* Inside a given memory region, determine whether a sub-region of memory is
* closer from the top or the bottom of the encompassing region. Return the
* size of the smallest chunk of free memory surrounding the sub-region in
* 'small_chunk_size'.
*****************************************************************************/
static unsigned int choose_mem_pos(uintptr_t mem_start, uintptr_t mem_end,
uintptr_t submem_start, uintptr_t submem_end,
size_t *small_chunk_size)
2013-10-25 09:08:21 +01:00
{
size_t top_chunk_size, bottom_chunk_size;
2013-10-25 09:08:21 +01:00
assert(mem_start <= submem_start);
assert(submem_start <= submem_end);
assert(submem_end <= mem_end);
assert(small_chunk_size != NULL);
2013-10-25 09:08:21 +01:00
top_chunk_size = mem_end - submem_end;
bottom_chunk_size = submem_start - mem_start;
if (top_chunk_size < bottom_chunk_size) {
*small_chunk_size = top_chunk_size;
return TOP;
2013-10-25 09:08:21 +01:00
} else {
*small_chunk_size = bottom_chunk_size;
return BOTTOM;
2013-10-25 09:08:21 +01:00
}
}
2013-10-25 09:08:21 +01:00
/******************************************************************************
* Reserve the memory region delimited by 'addr' and 'size'. The extents of free
* memory are passed in 'free_base' and 'free_size' and they will be updated to
* reflect the memory usage.
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
* The caller must ensure the memory to reserve is free and that the addresses
* and sizes passed in arguments are sane.
*****************************************************************************/
void reserve_mem(uintptr_t *free_base, size_t *free_size,
uintptr_t addr, size_t size)
{
size_t discard_size;
size_t reserved_size;
unsigned int pos;
2013-10-25 09:08:21 +01:00
assert(free_base != NULL);
assert(free_size != NULL);
assert(is_mem_free(*free_base, *free_size, addr, size));
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
if (size == 0) {
WARN("Nothing to allocate, requested size is zero\n");
return;
}
pos = choose_mem_pos(*free_base, *free_base + (*free_size - 1),
addr, addr + (size - 1),
&discard_size);
reserved_size = size + discard_size;
*free_size -= reserved_size;
if (pos == BOTTOM)
*free_base = addr + size;
VERBOSE("Reserved 0x%zx bytes (discarded 0x%zx bytes %s)\n",
reserved_size, discard_size,
pos == TOP ? "above" : "below");
2013-10-25 09:08:21 +01:00
}
static void dump_load_info(uintptr_t image_load_addr,
size_t image_size,
const meminfo_t *mem_layout)
2013-10-25 09:08:21 +01:00
{
INFO("Trying to load image at address %p, size = 0x%zx\n",
(void *)image_load_addr, image_size);
INFO("Current memory layout:\n");
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
INFO(" total region = [base = %p, size = 0x%zx]\n",
(void *) mem_layout->total_base, mem_layout->total_size);
INFO(" free region = [base = %p, size = 0x%zx]\n",
(void *) mem_layout->free_base, mem_layout->free_size);
2013-10-25 09:08:21 +01:00
}
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
#endif /* LOAD_IMAGE_V2 */
2013-10-25 09:08:21 +01:00
/* Generic function to return the size of an image */
size_t image_size(unsigned int image_id)
{
uintptr_t dev_handle;
uintptr_t image_handle;
uintptr_t image_spec;
size_t image_size = 0;
int io_result;
/* Obtain a reference to the image by querying the platform layer */
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
io_result = plat_get_image_source(image_id, &dev_handle, &image_spec);
if (io_result != 0) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to obtain reference to image id=%u (%i)\n",
image_id, io_result);
return 0;
}
/* Attempt to access the image */
io_result = io_open(dev_handle, image_spec, &image_handle);
if (io_result != 0) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to access image id=%u (%i)\n",
image_id, io_result);
return 0;
}
/* Find the size of the image */
io_result = io_size(image_handle, &image_size);
if ((io_result != 0) || (image_size == 0)) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to determine the size of the image id=%u (%i)\n",
image_id, io_result);
}
io_result = io_close(image_handle);
/* Ignore improbable/unrecoverable error in 'close' */
/* TODO: Consider maintaining open device connection from this
* bootloader stage
*/
io_result = io_dev_close(dev_handle);
/* Ignore improbable/unrecoverable error in 'dev_close' */
return image_size;
}
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
#if LOAD_IMAGE_V2
/*******************************************************************************
* Internal function to load an image at a specific address given
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
* an image ID and extents of free memory.
*
* If the load is successful then the image information is updated.
*
* Returns 0 on success, a negative error code otherwise.
******************************************************************************/
static int load_image(unsigned int image_id, image_info_t *image_data)
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
{
uintptr_t dev_handle;
uintptr_t image_handle;
uintptr_t image_spec;
uintptr_t image_base;
size_t image_size;
size_t bytes_read;
int io_result;
assert(image_data != NULL);
assert(image_data->h.version >= VERSION_2);
image_base = image_data->image_base;
/* Obtain a reference to the image by querying the platform layer */
io_result = plat_get_image_source(image_id, &dev_handle, &image_spec);
if (io_result != 0) {
WARN("Failed to obtain reference to image id=%u (%i)\n",
image_id, io_result);
return io_result;
}
/* Attempt to access the image */
io_result = io_open(dev_handle, image_spec, &image_handle);
if (io_result != 0) {
WARN("Failed to access image id=%u (%i)\n",
image_id, io_result);
return io_result;
}
INFO("Loading image id=%u at address %p\n", image_id,
(void *) image_base);
/* Find the size of the image */
io_result = io_size(image_handle, &image_size);
if ((io_result != 0) || (image_size == 0)) {
WARN("Failed to determine the size of the image id=%u (%i)\n",
image_id, io_result);
goto exit;
}
/* Check that the image size to load is within limit */
if (image_size > image_data->image_max_size) {
WARN("Image id=%u size out of bounds\n", image_id);
io_result = -EFBIG;
goto exit;
}
image_data->image_size = image_size;
/* We have enough space so load the image now */
/* TODO: Consider whether to try to recover/retry a partially successful read */
io_result = io_read(image_handle, image_base, image_size, &bytes_read);
if ((io_result != 0) || (bytes_read < image_size)) {
WARN("Failed to load image id=%u (%i)\n", image_id, io_result);
goto exit;
}
INFO("Image id=%u loaded: %p - %p\n", image_id, (void *) image_base,
(void *) (image_base + image_size));
exit:
io_close(image_handle);
/* Ignore improbable/unrecoverable error in 'close' */
/* TODO: Consider maintaining open device connection from this bootloader stage */
io_dev_close(dev_handle);
/* Ignore improbable/unrecoverable error in 'dev_close' */
return io_result;
}
static int load_auth_image_internal(unsigned int image_id,
image_info_t *image_data,
int is_parent_image)
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
{
int rc;
#if TRUSTED_BOARD_BOOT
unsigned int parent_id;
/* Use recursion to authenticate parent images */
rc = auth_mod_get_parent_id(image_id, &parent_id);
if (rc == 0) {
rc = load_auth_image_internal(parent_id, image_data, 1);
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
if (rc != 0) {
return rc;
}
}
#endif /* TRUSTED_BOARD_BOOT */
/* Load the image */
rc = load_image(image_id, image_data);
if (rc != 0) {
return rc;
}
#if TRUSTED_BOARD_BOOT
/* Authenticate it */
rc = auth_mod_verify_img(image_id,
(void *)image_data->image_base,
image_data->image_size);
if (rc != 0) {
/* Authentication error, zero memory and flush it right away. */
Introduce unified API to zero memory Introduce zeromem_dczva function on AArch64 that can handle unaligned addresses and make use of DC ZVA instruction to zero a whole block at a time. This zeroing takes place directly in the cache to speed it up without doing external memory access. Remove the zeromem16 function on AArch64 and replace it with an alias to zeromem. This zeromem16 function is now deprecated. Remove the 16-bytes alignment constraint on __BSS_START__ in firmware-design.md as it is now not mandatory anymore (it used to comply with zeromem16 requirements). Change the 16-bytes alignment constraints in SP min's linker script to a 8-bytes alignment constraint as the AArch32 zeromem implementation is now more efficient on 8-bytes aligned addresses. Introduce zero_normalmem and zeromem helpers in platform agnostic header that are implemented this way: * AArch32: * zero_normalmem: zero using usual data access * zeromem: alias for zero_normalmem * AArch64: * zero_normalmem: zero normal memory using DC ZVA instruction (needs MMU enabled) * zeromem: zero using usual data access Usage guidelines: in most cases, zero_normalmem should be preferred. There are 2 scenarios where zeromem (or memset) must be used instead: * Code that must run with MMU disabled (which means all memory is considered device memory for data accesses). * Code that fills device memory with null bytes. Optionally, the following rule can be applied if performance is important: * Code zeroing small areas (few bytes) that are not secrets should use memset to take advantage of compiler optimizations. Note: Code zeroing security-related critical information should use zero_normalmem/zeromem instead of memset to avoid removal by compilers' optimizations in some cases or misbehaving versions of GCC. Fixes ARM-software/tf-issues#408 Change-Id: Iafd9663fc1070413c3e1904e54091cf60effaa82 Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
2016-12-02 13:51:54 +00:00
zero_normalmem((void *)image_data->image_base,
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
image_data->image_size);
flush_dcache_range(image_data->image_base,
image_data->image_size);
return -EAUTH;
}
#endif /* TRUSTED_BOARD_BOOT */
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
/*
* Flush the image to main memory so that it can be executed later by
* any CPU, regardless of cache and MMU state. If TBB is enabled, then
* the file has been successfully loaded and authenticated and flush
* only for child images, not for the parents (certificates).
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
*/
if (!is_parent_image) {
flush_dcache_range(image_data->image_base,
image_data->image_size);
}
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
return 0;
}
/*******************************************************************************
* Generic function to load and authenticate an image. The image is actually
* loaded by calling the 'load_image()' function. Therefore, it returns the
* same error codes if the loading operation failed, or -EAUTH if the
* authentication failed. In addition, this function uses recursion to
* authenticate the parent images up to the root of trust.
******************************************************************************/
int load_auth_image(unsigned int image_id, image_info_t *image_data)
{
int err;
do {
err = load_auth_image_internal(image_id, image_data, 0);
} while (err != 0 && plat_try_next_boot_source());
return err;
}
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
#else /* LOAD_IMAGE_V2 */
2013-10-25 09:08:21 +01:00
/*******************************************************************************
* Generic function to load an image at a specific address given an image ID and
* extents of free memory.
*
* If the load is successful then the image information is updated.
*
* If the entry_point_info argument is not NULL then this function also updates:
* - the memory layout to mark the memory as reserved;
* - the entry point information.
*
* The caller might pass a NULL pointer for the entry point if they are not
* interested in this information. This is typically the case for non-executable
* images (e.g. certificates) and executable images that won't ever be executed
* on the application processor (e.g. additional microcontroller firmware).
*
* Returns 0 on success, a negative error code otherwise.
2013-10-25 09:08:21 +01:00
******************************************************************************/
int load_image(meminfo_t *mem_layout,
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
unsigned int image_id,
uintptr_t image_base,
image_info_t *image_data,
entry_point_info_t *entry_point_info)
2013-10-25 09:08:21 +01:00
{
uintptr_t dev_handle;
uintptr_t image_handle;
uintptr_t image_spec;
size_t image_size;
size_t bytes_read;
int io_result;
assert(mem_layout != NULL);
assert(image_data != NULL);
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
assert(image_data->h.version == VERSION_1);
/* Obtain a reference to the image by querying the platform layer */
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
io_result = plat_get_image_source(image_id, &dev_handle, &image_spec);
if (io_result != 0) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to obtain reference to image id=%u (%i)\n",
image_id, io_result);
return io_result;
}
2013-10-25 09:08:21 +01:00
/* Attempt to access the image */
io_result = io_open(dev_handle, image_spec, &image_handle);
if (io_result != 0) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to access image id=%u (%i)\n",
image_id, io_result);
return io_result;
2013-10-25 09:08:21 +01:00
}
INFO("Loading image id=%u at address %p\n", image_id,
(void *) image_base);
/* Find the size of the image */
io_result = io_size(image_handle, &image_size);
if ((io_result != 0) || (image_size == 0)) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to determine the size of the image id=%u (%i)\n",
image_id, io_result);
goto exit;
}
/* Check that the memory where the image will be loaded is free */
if (!is_mem_free(mem_layout->free_base, mem_layout->free_size,
image_base, image_size)) {
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
WARN("Failed to reserve region [base = %p, size = 0x%zx]\n",
(void *) image_base, image_size);
dump_load_info(image_base, image_size, mem_layout);
io_result = -ENOMEM;
goto exit;
2013-10-25 09:08:21 +01:00
}
/* We have enough space so load the image now */
/* TODO: Consider whether to try to recover/retry a partially successful read */
io_result = io_read(image_handle, image_base, image_size, &bytes_read);
if ((io_result != 0) || (bytes_read < image_size)) {
Use numbers to identify images instead of names The Trusted firmware code identifies BL images by name. The platform port defines a name for each image e.g. the IO framework uses this mechanism in the platform function plat_get_image_source(). For a given image name, it returns the handle to the image file which involves comparing images names. In addition, if the image is packaged in a FIP, a name comparison is required to find the UUID for the image. This method is not optimal. This patch changes the interface between the generic and platform code with regard to identifying images. The platform port must now allocate a unique number (ID) for every image. The generic code will use the image ID instead of the name to access its attributes. As a result, the plat_get_image_source() function now takes an image ID as an input parameter. The organisation of data structures within the IO framework has been rationalised to use an image ID as an index into an array which contains attributes of the image such as UUID and name. This prevents the name comparisons. A new type 'io_uuid_spec_t' has been introduced in the IO framework to specify images identified by UUID (i.e. when the image is contained in a FIP file). There is no longer need to maintain a look-up table [iname_name --> uuid] in the io_fip driver code. Because image names are no longer mandatory in the platform port, the debug messages in the generic code will show the image identifier instead of the file name. The platforms that support semihosting to load images (i.e. FVP) must provide the file names as definitions private to the platform. The ARM platform ports and documentation have been updated accordingly. All ARM platforms reuse the image IDs defined in the platform common code. These IDs will be used to access other attributes of an image in subsequent patches. IMPORTANT: applying this patch breaks compatibility for platforms that use TF BL1 or BL2 images or the image loading code. The platform port must be updated to match the new interface. Change-Id: I9c1b04cb1a0684c6ee65dee66146dd6731751ea5
2015-04-13 17:36:19 +01:00
WARN("Failed to load image id=%u (%i)\n", image_id, io_result);
goto exit;
2013-10-25 09:08:21 +01:00
}
image_data->image_base = image_base;
image_data->image_size = image_size;
/*
* Update the memory usage info.
* This is done after the actual loading so that it is not updated when
* the load is unsuccessful.
* If the caller does not provide an entry point, bypass the memory
* reservation.
*/
if (entry_point_info != NULL) {
reserve_mem(&mem_layout->free_base, &mem_layout->free_size,
image_base, image_size);
entry_point_info->pc = image_base;
} else {
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
INFO("Skip reserving region [base = %p, size = 0x%zx]\n",
(void *) image_base, image_size);
}
#if !TRUSTED_BOARD_BOOT
2013-10-25 09:08:21 +01:00
/*
* File has been successfully loaded.
* Flush the image to main memory so that it can be executed later by
* any CPU, regardless of cache and MMU state.
* When TBB is enabled the image is flushed later, after image
* authentication.
2013-10-25 09:08:21 +01:00
*/
flush_dcache_range(image_base, image_size);
#endif /* TRUSTED_BOARD_BOOT */
2013-10-25 09:08:21 +01:00
Ensure addresses in is_mem_free() don't overflow This patch adds some runtime checks to prevent some potential pointer overflow issues in the is_mem_free() function. The overflow could happen in the case where the end addresses, computed as the sum of a base address and a size, results in a value large enough to wrap around. This, in turn, could lead to unpredictable behaviour. If such an overflow is detected, the is_mem_free() function will now declare the memory region as not free. The overflow is detected using a new macro, called check_uptr_overflow(). This patch also modifies all other places in the 'bl_common.c' file where an end address was computed as the sum of a base address and a size and instead keeps the two values separate. This avoids the need to handle pointer overflows everywhere. The code doesn't actually need to compute any end address before the is_mem_free() function is called other than to print information message to the serial output. This patch also introduces 2 slight changes to the reserve_mem() function: - It fixes the end addresses passed to choose_mem_pos(). It was incorrectly passing (base + size) instead of (base + size - 1). - When the requested allocation size is 0, the function now exits straight away and says so using a warning message. Previously, it used to actually reserve some memory. A zero-byte allocation was not considered as a special case so the function was using the same top/bottom allocation mechanism as for any other allocation. As a result, the smallest area of memory starting from the requested base address within the free region was reserved. Change-Id: I0e695f961e24e56ffe000718014e0496dc6e1ec6
2016-07-12 09:12:24 +01:00
INFO("Image id=%u loaded at address %p, size = 0x%zx\n", image_id,
(void *) image_base, image_size);
exit:
io_close(image_handle);
/* Ignore improbable/unrecoverable error in 'close' */
/* TODO: Consider maintaining open device connection from this bootloader stage */
io_dev_close(dev_handle);
/* Ignore improbable/unrecoverable error in 'dev_close' */
2013-10-25 09:08:21 +01:00
return io_result;
2013-10-25 09:08:21 +01:00
}
static int load_auth_image_internal(meminfo_t *mem_layout,
unsigned int image_id,
uintptr_t image_base,
image_info_t *image_data,
entry_point_info_t *entry_point_info,
int is_parent_image)
{
int rc;
#if TRUSTED_BOARD_BOOT
unsigned int parent_id;
/* Use recursion to authenticate parent images */
rc = auth_mod_get_parent_id(image_id, &parent_id);
if (rc == 0) {
rc = load_auth_image_internal(mem_layout, parent_id, image_base,
image_data, NULL, 1);
if (rc != 0) {
return rc;
}
}
#endif /* TRUSTED_BOARD_BOOT */
/* Load the image */
rc = load_image(mem_layout, image_id, image_base, image_data,
entry_point_info);
if (rc != 0) {
return rc;
}
#if TRUSTED_BOARD_BOOT
/* Authenticate it */
rc = auth_mod_verify_img(image_id,
(void *)image_data->image_base,
image_data->image_size);
if (rc != 0) {
/* Authentication error, zero memory and flush it right away. */
Introduce unified API to zero memory Introduce zeromem_dczva function on AArch64 that can handle unaligned addresses and make use of DC ZVA instruction to zero a whole block at a time. This zeroing takes place directly in the cache to speed it up without doing external memory access. Remove the zeromem16 function on AArch64 and replace it with an alias to zeromem. This zeromem16 function is now deprecated. Remove the 16-bytes alignment constraint on __BSS_START__ in firmware-design.md as it is now not mandatory anymore (it used to comply with zeromem16 requirements). Change the 16-bytes alignment constraints in SP min's linker script to a 8-bytes alignment constraint as the AArch32 zeromem implementation is now more efficient on 8-bytes aligned addresses. Introduce zero_normalmem and zeromem helpers in platform agnostic header that are implemented this way: * AArch32: * zero_normalmem: zero using usual data access * zeromem: alias for zero_normalmem * AArch64: * zero_normalmem: zero normal memory using DC ZVA instruction (needs MMU enabled) * zeromem: zero using usual data access Usage guidelines: in most cases, zero_normalmem should be preferred. There are 2 scenarios where zeromem (or memset) must be used instead: * Code that must run with MMU disabled (which means all memory is considered device memory for data accesses). * Code that fills device memory with null bytes. Optionally, the following rule can be applied if performance is important: * Code zeroing small areas (few bytes) that are not secrets should use memset to take advantage of compiler optimizations. Note: Code zeroing security-related critical information should use zero_normalmem/zeromem instead of memset to avoid removal by compilers' optimizations in some cases or misbehaving versions of GCC. Fixes ARM-software/tf-issues#408 Change-Id: Iafd9663fc1070413c3e1904e54091cf60effaa82 Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
2016-12-02 13:51:54 +00:00
zero_normalmem((void *)image_data->image_base,
image_data->image_size);
flush_dcache_range(image_data->image_base,
image_data->image_size);
return -EAUTH;
}
/*
* File has been successfully loaded and authenticated.
* Flush the image to main memory so that it can be executed later by
* any CPU, regardless of cache and MMU state.
* Do it only for child images, not for the parents (certificates).
*/
if (!is_parent_image) {
flush_dcache_range(image_data->image_base,
image_data->image_size);
}
#endif /* TRUSTED_BOARD_BOOT */
return 0;
}
/*******************************************************************************
* Generic function to load and authenticate an image. The image is actually
* loaded by calling the 'load_image()' function. Therefore, it returns the
* same error codes if the loading operation failed, or -EAUTH if the
* authentication failed. In addition, this function uses recursion to
* authenticate the parent images up to the root of trust.
******************************************************************************/
int load_auth_image(meminfo_t *mem_layout,
unsigned int image_id,
uintptr_t image_base,
image_info_t *image_data,
entry_point_info_t *entry_point_info)
{
int err;
do {
err = load_auth_image_internal(mem_layout, image_id, image_base,
image_data, entry_point_info, 0);
} while (err != 0 && plat_try_next_boot_source());
return err;
}
Add new version of image loading. This patch adds capability to load BL images based on image descriptors instead of hard coded way of loading BL images. This framework is designed such that it can be readily adapted by any BL stage that needs to load images. In order to provide the above capability the following new platform functions are introduced: bl_load_info_t *plat_get_bl_image_load_info(void); This function returns pointer to the list of images that the platform has populated to load. bl_params_t *plat_get_next_bl_params(void); This function returns a pointer to the shared memory that the platform has kept aside to pass trusted firmware related information that next BL image needs. void plat_flush_next_bl_params(void); This function flushes to main memory all the params that are passed to next image. int bl2_plat_handle_post_image_load(unsigned int image_id) This function can be used by the platforms to update/use image information for given `image_id`. `desc_image_load.c` contains utility functions which can be used by the platforms to generate, load and executable, image list based on the registered image descriptors. This patch also adds new version of `load_image/load_auth_image` functions in-order to achieve the above capability. Following are the changes for the new version as compared to old: - Refactor the signature and only keep image_id and image_info_t arguments. Removed image_base argument as it is already passed through image_info_t. Given that the BL image base addresses and limit/size are already provided by the platforms, the meminfo_t and entry_point_info arguments are not needed to provide/reserve the extent of free memory for the given BL image. - Added check for the image size against the defined max size. This is needed because the image size could come from an unauthenticated source (e.g. the FIP header). To make this check, new member is added to the image_info_t struct for identifying the image maximum size. New flag `LOAD_IMAGE_V2` is added in the Makefile. Default value is 0. NOTE: `TRUSTED_BOARD_BOOT` is currently not supported when `LOAD_IMAGE_V2` is enabled. Change-Id: Ia7b643f4817a170d5a2fbf479b9bc12e63112e79
2016-09-12 16:08:41 +01:00
#endif /* LOAD_IMAGE_V2 */
/*******************************************************************************
* Print the content of an entry_point_info_t structure.
******************************************************************************/
void print_entry_point_info(const entry_point_info_t *ep_info)
{
INFO("Entry point address = %p\n", (void *)ep_info->pc);
INFO("SPSR = 0x%x\n", ep_info->spsr);
#define PRINT_IMAGE_ARG(n) \
VERBOSE("Argument #" #n " = 0x%llx\n", \
(unsigned long long) ep_info->args.arg##n)
PRINT_IMAGE_ARG(0);
PRINT_IMAGE_ARG(1);
PRINT_IMAGE_ARG(2);
PRINT_IMAGE_ARG(3);
#ifndef AARCH32
PRINT_IMAGE_ARG(4);
PRINT_IMAGE_ARG(5);
PRINT_IMAGE_ARG(6);
PRINT_IMAGE_ARG(7);
#endif
#undef PRINT_IMAGE_ARG
}